qemu_process.c 239.4 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 36
#include <sys/utsname.h>

37
#include "qemu_process.h"
38
#include "qemu_processpriv.h"
39
#include "qemu_alias.h"
40
#include "qemu_block.h"
41
#include "qemu_domain.h"
42
#include "qemu_domain_address.h"
43 44 45 46 47 48
#include "qemu_cgroup.h"
#include "qemu_capabilities.h"
#include "qemu_monitor.h"
#include "qemu_command.h"
#include "qemu_hostdev.h"
#include "qemu_hotplug.h"
49
#include "qemu_migration.h"
50
#include "qemu_migration_params.h"
51
#include "qemu_interface.h"
52
#include "qemu_security.h"
53
#include "qemu_extdevice.h"
54

55
#include "cpu/cpu.h"
56
#include "datatypes.h"
57
#include "virlog.h"
58
#include "virerror.h"
59
#include "viralloc.h"
60
#include "virhook.h"
E
Eric Blake 已提交
61
#include "virfile.h"
62
#include "virpidfile.h"
63
#include "virhostcpu.h"
64
#include "domain_audit.h"
65
#include "domain_nwfilter.h"
66
#include "locking/domain_lock.h"
67
#include "viruuid.h"
68
#include "virprocess.h"
69
#include "virtime.h"
A
Ansis Atteka 已提交
70
#include "virnetdevtap.h"
71
#include "virnetdevopenvswitch.h"
72
#include "virnetdevmidonet.h"
73
#include "virbitmap.h"
74
#include "viratomic.h"
75
#include "virnuma.h"
76
#include "virstring.h"
77
#include "virhostdev.h"
J
John Ferlan 已提交
78
#include "secret_util.h"
79
#include "configmake.h"
80
#include "nwfilter_conf.h"
81
#include "netdev_bandwidth_conf.h"
82
#include "virresctrl.h"
83
#include "virvsock.h"
84 85 86

#define VIR_FROM_THIS VIR_FROM_QEMU

87 88
VIR_LOG_INIT("qemu.qemu_process");

89
/**
90
 * qemuProcessRemoveDomainStatus
91 92 93 94 95 96
 *
 * remove all state files of a domain from statedir
 *
 * Returns 0 on success
 */
static int
97
qemuProcessRemoveDomainStatus(virQEMUDriverPtr driver,
98 99 100 101
                              virDomainObjPtr vm)
{
    char ebuf[1024];
    char *file = NULL;
102
    qemuDomainObjPrivatePtr priv = vm->privateData;
103 104
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    int ret = -1;
105

106
    if (virAsprintf(&file, "%s/%s.xml", cfg->stateDir, vm->def->name) < 0)
107
        goto cleanup;
108 109 110 111 112 113

    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);

114 115 116
    if (priv->pidfile &&
        unlink(priv->pidfile) < 0 &&
        errno != ENOENT)
117 118 119
        VIR_WARN("Failed to remove PID file for %s: %s",
                 vm->def->name, virStrerror(errno, ebuf, sizeof(ebuf)));

120
    ret = 0;
121
 cleanup:
122 123
    virObjectUnref(cfg);
    return ret;
124 125 126
}


D
Daniel P. Berrange 已提交
127 128 129 130 131 132 133
/*
 * 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
134
qemuProcessHandleAgentEOF(qemuAgentPtr agent,
D
Daniel P. Berrange 已提交
135 136 137 138 139 140
                          virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv;

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

141
    virObjectLock(vm);
D
Daniel P. Berrange 已提交
142 143

    priv = vm->privateData;
144 145 146 147 148 149 150 151 152 153 154

    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;
    }

155
    qemuAgentClose(agent);
156
    priv->agent = NULL;
157
    priv->agentError = false;
D
Daniel P. Berrange 已提交
158

159
    virObjectUnlock(vm);
160 161
    return;

162
 unlock:
163 164
    virObjectUnlock(vm);
    return;
D
Daniel P. Berrange 已提交
165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181
}


/*
 * 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);

182
    virObjectLock(vm);
D
Daniel P. Berrange 已提交
183 184 185 186 187

    priv = vm->privateData;

    priv->agentError = true;

188
    virObjectUnlock(vm);
D
Daniel P. Berrange 已提交
189 190 191 192 193
}

static void qemuProcessHandleAgentDestroy(qemuAgentPtr agent,
                                          virDomainObjPtr vm)
{
194 195
    VIR_DEBUG("Received destroy agent=%p vm=%p", agent, vm);

196
    virObjectUnref(vm);
D
Daniel P. Berrange 已提交
197 198 199 200 201 202 203 204 205 206
}


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


207
int
208
qemuConnectAgent(virQEMUDriverPtr driver, virDomainObjPtr vm)
D
Daniel P. Berrange 已提交
209 210 211
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    qemuAgentPtr agent = NULL;
212
    virDomainChrDefPtr config = qemuFindAgentConfig(vm->def);
D
Daniel P. Berrange 已提交
213 214 215 216

    if (!config)
        return 0;

217 218 219 220 221 222 223 224 225
    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;
    }

226
    if (qemuSecuritySetDaemonSocketLabel(driver->securityManager, vm->def) < 0) {
D
Daniel P. Berrange 已提交
227 228 229 230 231 232 233
        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 */
234
    virObjectRef(vm);
D
Daniel P. Berrange 已提交
235

236
    virObjectUnlock(vm);
D
Daniel P. Berrange 已提交
237 238

    agent = qemuAgentOpen(vm,
239
                          config->source,
D
Daniel P. Berrange 已提交
240 241
                          &agentCallbacks);

242
    virObjectLock(vm);
D
Daniel P. Berrange 已提交
243

244 245 246 247 248 249 250
    if (agent == NULL)
        virObjectUnref(vm);

    if (!virDomainObjIsActive(vm)) {
        qemuAgentClose(agent);
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("guest crashed while connecting to the guest agent"));
251
        return -1;
252 253
    }

254
    if (qemuSecurityClearSocketLabel(driver->securityManager, vm->def) < 0) {
D
Daniel P. Berrange 已提交
255 256
        VIR_ERROR(_("Failed to clear security context for agent for %s"),
                  vm->def->name);
257
        qemuAgentClose(agent);
D
Daniel P. Berrange 已提交
258 259 260 261
        goto cleanup;
    }

    priv->agent = agent;
262
    if (!priv->agent)
D
Daniel P. Berrange 已提交
263 264
        VIR_INFO("Failed to connect agent for %s", vm->def->name);

265
 cleanup:
266 267 268 269 270 271 272
    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
273 274 275
}


276
/*
277
 * This is a callback registered with a qemuMonitorPtr instance,
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
283
qemuProcessHandleMonitorEOF(qemuMonitorPtr mon,
284 285
                            virDomainObjPtr vm,
                            void *opaque)
286
{
287
    virQEMUDriverPtr driver = opaque;
288
    qemuDomainObjPrivatePtr priv;
289
    struct qemuProcessEvent *processEvent;
290

291
    virObjectLock(vm);
292

293
    VIR_DEBUG("Received EOF on %p '%s'", vm, vm->def->name);
294

295
    priv = vm->privateData;
296 297
    if (priv->beingDestroyed) {
        VIR_DEBUG("Domain is being destroyed, EOF is expected");
298
        goto cleanup;
299 300
    }

301
    if (VIR_ALLOC(processEvent) < 0)
302
        goto cleanup;
303

304
    processEvent->eventType = QEMU_PROCESS_EVENT_MONITOR_EOF;
305
    processEvent->vm = virObjectRef(vm);
306

307 308
    if (virThreadPoolSendJob(driver->workerPool, 0, processEvent) < 0) {
        ignore_value(virObjectUnref(vm));
309
        qemuProcessEventFree(processEvent);
310
        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

318 319 320 321
    /* We don't want any cleanup from EOF handler (or any other
     * thread) to enter qemu namespace. */
    qemuDomainDestroyNamespace(driver, vm);

322
 cleanup:
323
    virObjectUnlock(vm);
324 325 326 327 328 329 330 331 332 333 334
}


/*
 * 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,
335 336
                              virDomainObjPtr vm,
                              void *opaque)
337
{
338
    virQEMUDriverPtr driver = opaque;
339
    virObjectEventPtr event = NULL;
340 341 342

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

343
    virObjectLock(vm);
344

345
    ((qemuDomainObjPrivatePtr) vm->privateData)->monError = true;
346
    event = virDomainEventControlErrorNewFromObj(vm);
347
    virObjectEventStateQueue(driver->domainEventState, event);
348

349
    virObjectUnlock(vm);
350 351 352
}


353
virDomainDiskDefPtr
354 355 356
qemuProcessFindDomainDiskByAlias(virDomainObjPtr vm,
                                 const char *alias)
{
357
    size_t i;
358

359
    alias = qemuAliasDiskDriveSkipPrefix(alias);
360 361 362 363 364 365 366 367 368

    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;
    }

369 370 371
    virReportError(VIR_ERR_INTERNAL_ERROR,
                   _("no disk found with alias %s"),
                   alias);
372 373 374 375 376
    return NULL;
}

static int
qemuProcessHandleReset(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
377 378
                       virDomainObjPtr vm,
                       void *opaque)
379
{
380
    virQEMUDriverPtr driver = opaque;
381
    virObjectEventPtr event;
382
    qemuDomainObjPrivatePtr priv;
383
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
M
Michal Privoznik 已提交
384
    int ret = -1;
385

386
    virObjectLock(vm);
387

388
    event = virDomainEventRebootNewFromObj(vm);
389 390 391
    priv = vm->privateData;
    if (priv->agent)
        qemuAgentNotifyEvent(priv->agent, QEMU_AGENT_EVENT_RESET);
392

393
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0)
394
        VIR_WARN("Failed to save status on vm %s", vm->def->name);
395

396 397
    if (vm->def->onReboot == VIR_DOMAIN_LIFECYCLE_ACTION_DESTROY ||
        vm->def->onReboot == VIR_DOMAIN_LIFECYCLE_ACTION_PRESERVE) {
398

M
Michal Privoznik 已提交
399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414
        if (qemuDomainObjBeginJob(driver, vm, QEMU_JOB_MODIFY) < 0)
            goto cleanup;

        if (!virDomainObjIsActive(vm)) {
            VIR_DEBUG("Ignoring RESET event from inactive domain %s",
                      vm->def->name);
            goto endjob;
        }

        qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_DESTROYED,
                        QEMU_ASYNC_JOB_NONE, 0);
        virDomainAuditStop(vm, "destroyed");
        qemuDomainRemoveInactive(driver, vm);
     endjob:
        qemuDomainObjEndJob(driver, vm);
    }
415

M
Michal Privoznik 已提交
416 417 418
    ret = 0;
 cleanup:
    virObjectUnlock(vm);
419
    virObjectEventStateQueue(driver->domainEventState, event);
420
    virObjectUnref(cfg);
M
Michal Privoznik 已提交
421
    return ret;
422 423 424
}


425 426 427 428 429 430 431 432 433 434 435 436 437
/*
 * 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)
{
    virDomainObjPtr vm = opaque;
    qemuDomainObjPrivatePtr priv = vm->privateData;
438
    virQEMUDriverPtr driver = priv->driver;
439
    virObjectEventPtr event = NULL;
440
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
441
    virDomainRunningReason reason = VIR_DOMAIN_RUNNING_BOOTED;
442
    int ret = -1, rc;
443

444
    VIR_DEBUG("vm=%p", vm);
445
    virObjectLock(vm);
446
    if (qemuDomainObjBeginJob(driver, vm, QEMU_JOB_MODIFY) < 0)
447 448 449
        goto cleanup;

    if (!virDomainObjIsActive(vm)) {
450 451
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("guest unexpectedly quit"));
452 453 454
        goto endjob;
    }

455
    qemuDomainObjEnterMonitor(driver, vm);
456 457 458
    rc = qemuMonitorSystemReset(priv->mon);

    if (qemuDomainObjExitMonitor(driver, vm) < 0)
459 460
        goto endjob;

461
    if (rc < 0)
462 463
        goto endjob;

464 465 466
    if (virDomainObjGetState(vm, NULL) == VIR_DOMAIN_CRASHED)
        reason = VIR_DOMAIN_RUNNING_CRASHED;

467
    if (qemuProcessStartCPUs(driver, vm,
468
                             reason,
469
                             QEMU_ASYNC_JOB_NONE) < 0) {
470
        if (virGetLastErrorCode() == VIR_ERR_OK)
471 472
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("resume operation failed"));
473 474
        goto endjob;
    }
475
    priv->gotShutdown = false;
476
    event = virDomainEventLifecycleNewFromObj(vm,
477 478 479
                                     VIR_DOMAIN_EVENT_RESUMED,
                                     VIR_DOMAIN_EVENT_RESUMED_UNPAUSED);

480
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0) {
481 482 483 484
        VIR_WARN("Unable to save status on vm %s after state change",
                 vm->def->name);
    }

485 486
    ret = 0;

487
 endjob:
488
    qemuDomainObjEndJob(driver, vm);
489

490
 cleanup:
491 492
    if (ret == -1)
        ignore_value(qemuProcessKill(vm, VIR_QEMU_PROCESS_KILL_FORCE));
M
Michal Privoznik 已提交
493
    virDomainObjEndAPI(&vm);
494
    virObjectEventStateQueue(driver->domainEventState, event);
495
    virObjectUnref(cfg);
496 497 498
}


499
void
500
qemuProcessShutdownOrReboot(virQEMUDriverPtr driver,
501
                            virDomainObjPtr vm)
502
{
503 504 505
    qemuDomainObjPrivatePtr priv = vm->privateData;

    if (priv->fakeReboot) {
506
        qemuDomainSetFakeReboot(driver, vm, false);
507
        virObjectRef(vm);
508 509 510 511 512
        virThread th;
        if (virThreadCreate(&th,
                            false,
                            qemuProcessFakeReboot,
                            vm) < 0) {
513
            VIR_ERROR(_("Failed to create reboot thread, killing domain"));
514
            ignore_value(qemuProcessKill(vm, VIR_QEMU_PROCESS_KILL_NOWAIT));
515
            virObjectUnref(vm);
516 517
        }
    } else {
518
        ignore_value(qemuProcessKill(vm, VIR_QEMU_PROCESS_KILL_NOWAIT));
519
    }
520
}
521

522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542

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);
543
    virObjectEventStateQueue(driver->domainEventState, event);
544 545 546 547 548

    return 0;
}


549 550
static int
qemuProcessHandleShutdown(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
551
                          virDomainObjPtr vm,
552
                          virTristateBool guest_initiated,
553
                          void *opaque)
554
{
555
    virQEMUDriverPtr driver = opaque;
556
    qemuDomainObjPrivatePtr priv;
557
    virObjectEventPtr event = NULL;
558
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
559
    int detail = 0;
560

561 562
    VIR_DEBUG("vm=%p", vm);

563
    virObjectLock(vm);
564 565 566 567 568 569

    priv = vm->privateData;
    if (priv->gotShutdown) {
        VIR_DEBUG("Ignoring repeated SHUTDOWN event from domain %s",
                  vm->def->name);
        goto unlock;
570 571 572 573
    } else if (!virDomainObjIsActive(vm)) {
        VIR_DEBUG("Ignoring SHUTDOWN event from inactive domain %s",
                  vm->def->name);
        goto unlock;
574 575 576 577 578 579 580 581
    }
    priv->gotShutdown = true;

    VIR_DEBUG("Transitioned guest %s to shutdown state",
              vm->def->name);
    virDomainObjSetState(vm,
                         VIR_DOMAIN_SHUTDOWN,
                         VIR_DOMAIN_SHUTDOWN_UNKNOWN);
582 583 584 585 586 587 588 589 590 591

    switch (guest_initiated) {
    case VIR_TRISTATE_BOOL_YES:
        detail = VIR_DOMAIN_EVENT_SHUTDOWN_GUEST;
        break;

    case VIR_TRISTATE_BOOL_NO:
        detail = VIR_DOMAIN_EVENT_SHUTDOWN_HOST;
        break;

592 593
    case VIR_TRISTATE_BOOL_ABSENT:
    case VIR_TRISTATE_BOOL_LAST:
594 595 596 597 598
    default:
        detail = VIR_DOMAIN_EVENT_SHUTDOWN_FINISHED;
        break;
    }

599
    event = virDomainEventLifecycleNewFromObj(vm,
600 601
                                              VIR_DOMAIN_EVENT_SHUTDOWN,
                                              detail);
602

603
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0) {
604 605 606 607
        VIR_WARN("Unable to save status on vm %s after state change",
                 vm->def->name);
    }

608 609 610
    if (priv->agent)
        qemuAgentNotifyEvent(priv->agent, QEMU_AGENT_EVENT_SHUTDOWN);

611 612
    qemuProcessShutdownOrReboot(driver, vm);

613
 unlock:
614
    virObjectUnlock(vm);
615
    virObjectEventStateQueue(driver->domainEventState, event);
616
    virObjectUnref(cfg);
617

618 619 620 621 622 623
    return 0;
}


static int
qemuProcessHandleStop(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
624 625
                      virDomainObjPtr vm,
                      void *opaque)
626
{
627
    virQEMUDriverPtr driver = opaque;
628
    virObjectEventPtr event = NULL;
629 630
    virDomainPausedReason reason = VIR_DOMAIN_PAUSED_UNKNOWN;
    virDomainEventSuspendedDetailType detail = VIR_DOMAIN_EVENT_SUSPENDED_PAUSED;
631
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
632

633
    virObjectLock(vm);
J
Jiri Denemark 已提交
634
    if (virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING) {
635
        qemuDomainObjPrivatePtr priv = vm->privateData;
636

637
        if (priv->gotShutdown) {
638 639
            VIR_DEBUG("Ignoring STOP event after SHUTDOWN");
            goto unlock;
640 641
        }

642
        if (priv->job.asyncJob == QEMU_ASYNC_JOB_MIGRATION_OUT) {
643 644
            if (priv->job.current->status ==
                        QEMU_DOMAIN_JOB_STATUS_POSTCOPY) {
645 646 647 648 649 650
                reason = VIR_DOMAIN_PAUSED_POSTCOPY;
                detail = VIR_DOMAIN_EVENT_SUSPENDED_POSTCOPY;
            } else {
                reason = VIR_DOMAIN_PAUSED_MIGRATION;
                detail = VIR_DOMAIN_EVENT_SUSPENDED_MIGRATED;
            }
651 652 653 654
        }

        VIR_DEBUG("Transitioned guest %s to paused state, reason %s",
                  vm->def->name, virDomainPausedReasonTypeToString(reason));
655

656 657 658
        if (priv->job.current)
            ignore_value(virTimeMillisNow(&priv->job.current->stopped));

659 660 661 662
        if (priv->signalStop)
            virDomainObjBroadcast(vm);

        virDomainObjSetState(vm, VIR_DOMAIN_PAUSED, reason);
663
        event = virDomainEventLifecycleNewFromObj(vm,
664 665
                                                  VIR_DOMAIN_EVENT_SUSPENDED,
                                                  detail);
666

667 668 669 670 671
        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));

672
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0) {
673 674 675
            VIR_WARN("Unable to save status on vm %s after state change",
                     vm->def->name);
        }
676
    }
677

678
 unlock:
679
    virObjectUnlock(vm);
680
    virObjectEventStateQueue(driver->domainEventState, event);
681
    virObjectUnref(cfg);
682 683 684 685 686

    return 0;
}


687 688
static int
qemuProcessHandleResume(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
689 690
                        virDomainObjPtr vm,
                        void *opaque)
691
{
692
    virQEMUDriverPtr driver = opaque;
693
    virObjectEventPtr event = NULL;
694
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
695

696
    virObjectLock(vm);
697 698 699 700 701 702 703 704 705 706 707 708 709
    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);
710
        event = virDomainEventLifecycleNewFromObj(vm,
711 712 713
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_UNPAUSED);

714
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0) {
715 716 717 718 719
            VIR_WARN("Unable to save status on vm %s after state change",
                     vm->def->name);
        }
    }

720
 unlock:
721
    virObjectUnlock(vm);
722
    virObjectEventStateQueue(driver->domainEventState, event);
723
    virObjectUnref(cfg);
724 725 726
    return 0;
}

727 728 729
static int
qemuProcessHandleRTCChange(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                           virDomainObjPtr vm,
730 731
                           long long offset,
                           void *opaque)
732
{
733
    virQEMUDriverPtr driver = opaque;
734
    virObjectEventPtr event = NULL;
735
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
736

737
    virObjectLock(vm);
738

739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755
    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;
756
        vm->def->clock.data.variable.adjustment = offset;
757

758
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0)
759 760 761 762
           VIR_WARN("unable to save domain status with RTC change");
    }

    event = virDomainEventRTCChangeNewFromObj(vm, offset);
763

764
    virObjectUnlock(vm);
765

766
    virObjectEventStateQueue(driver->domainEventState, event);
767
    virObjectUnref(cfg);
768 769 770 771 772 773 774
    return 0;
}


static int
qemuProcessHandleWatchdog(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                          virDomainObjPtr vm,
775 776
                          int action,
                          void *opaque)
777
{
778
    virQEMUDriverPtr driver = opaque;
779 780
    virObjectEventPtr watchdogEvent = NULL;
    virObjectEventPtr lifecycleEvent = NULL;
781
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
782

783
    virObjectLock(vm);
784 785 786
    watchdogEvent = virDomainEventWatchdogNewFromObj(vm, action);

    if (action == VIR_DOMAIN_EVENT_WATCHDOG_PAUSE &&
J
Jiri Denemark 已提交
787
        virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING) {
788
        qemuDomainObjPrivatePtr priv = vm->privateData;
789 790
        VIR_DEBUG("Transitioned guest %s to paused state due to watchdog", vm->def->name);

J
Jiri Denemark 已提交
791
        virDomainObjSetState(vm, VIR_DOMAIN_PAUSED, VIR_DOMAIN_PAUSED_WATCHDOG);
792
        lifecycleEvent = virDomainEventLifecycleNewFromObj(vm,
793 794 795
                                                  VIR_DOMAIN_EVENT_SUSPENDED,
                                                  VIR_DOMAIN_EVENT_SUSPENDED_WATCHDOG);

796 797 798 799 800
        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));

801
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0) {
802 803 804
            VIR_WARN("Unable to save status on vm %s after watchdog event",
                     vm->def->name);
        }
805 806 807
    }

    if (vm->def->watchdog->action == VIR_DOMAIN_WATCHDOG_ACTION_DUMP) {
C
Chen Fan 已提交
808 809 810 811
        struct qemuProcessEvent *processEvent;
        if (VIR_ALLOC(processEvent) == 0) {
            processEvent->eventType = QEMU_PROCESS_EVENT_WATCHDOG;
            processEvent->action = VIR_DOMAIN_WATCHDOG_ACTION_DUMP;
W
Wen Congyang 已提交
812 813 814
            /* Hold an extra reference because we can't allow 'vm' to be
             * deleted before handling watchdog event is finished.
             */
815
            processEvent->vm = virObjectRef(vm);
C
Chen Fan 已提交
816
            if (virThreadPoolSendJob(driver->workerPool, 0, processEvent) < 0) {
817
                if (!virObjectUnref(vm))
818
                    vm = NULL;
819
                qemuProcessEventFree(processEvent);
W
Wen Congyang 已提交
820
            }
821
        }
822 823
    }

824
    if (vm)
825
        virObjectUnlock(vm);
826 827
    virObjectEventStateQueue(driver->domainEventState, watchdogEvent);
    virObjectEventStateQueue(driver->domainEventState, lifecycleEvent);
828

829
    virObjectUnref(cfg);
830 831 832 833 834 835 836 837 838
    return 0;
}


static int
qemuProcessHandleIOError(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                         virDomainObjPtr vm,
                         const char *diskAlias,
                         int action,
839 840
                         const char *reason,
                         void *opaque)
841
{
842
    virQEMUDriverPtr driver = opaque;
843 844 845
    virObjectEventPtr ioErrorEvent = NULL;
    virObjectEventPtr ioErrorEvent2 = NULL;
    virObjectEventPtr lifecycleEvent = NULL;
846 847 848
    const char *srcPath;
    const char *devAlias;
    virDomainDiskDefPtr disk;
849
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
850

851
    virObjectLock(vm);
852 853 854
    disk = qemuProcessFindDomainDiskByAlias(vm, diskAlias);

    if (disk) {
855
        srcPath = virDomainDiskGetSource(disk);
856 857 858 859 860 861 862 863 864 865
        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
866
        virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING) {
867
        qemuDomainObjPrivatePtr priv = vm->privateData;
868 869
        VIR_DEBUG("Transitioned guest %s to paused state due to IO error", vm->def->name);

870 871 872
        if (priv->signalIOError)
            virDomainObjBroadcast(vm);

J
Jiri Denemark 已提交
873
        virDomainObjSetState(vm, VIR_DOMAIN_PAUSED, VIR_DOMAIN_PAUSED_IOERROR);
874
        lifecycleEvent = virDomainEventLifecycleNewFromObj(vm,
875 876 877
                                                  VIR_DOMAIN_EVENT_SUSPENDED,
                                                  VIR_DOMAIN_EVENT_SUSPENDED_IOERROR);

878 879 880 881 882
        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));

883
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0)
884 885
            VIR_WARN("Unable to save status on vm %s after IO error", vm->def->name);
    }
886
    virObjectUnlock(vm);
887

888 889 890
    virObjectEventStateQueue(driver->domainEventState, ioErrorEvent);
    virObjectEventStateQueue(driver->domainEventState, ioErrorEvent2);
    virObjectEventStateQueue(driver->domainEventState, lifecycleEvent);
891
    virObjectUnref(cfg);
892 893 894
    return 0;
}

895 896 897 898 899
static int
qemuProcessHandleBlockJob(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                          virDomainObjPtr vm,
                          const char *diskAlias,
                          int type,
900
                          int status,
901
                          const char *error,
902
                          void *opaque)
903
{
904
    virQEMUDriverPtr driver = opaque;
905
    struct qemuProcessEvent *processEvent = NULL;
906
    virDomainDiskDefPtr disk;
907
    qemuDomainDiskPrivatePtr diskPriv;
908
    char *data = NULL;
909

910
    virObjectLock(vm);
911

912 913
    VIR_DEBUG("Block job for device %s (domain: %p,%s) type %d status %d",
              diskAlias, vm, vm->def->name, type, status);
914

915
    if (!(disk = qemuProcessFindDomainDiskByAlias(vm, diskAlias)))
916
        goto error;
917
    diskPriv = QEMU_DOMAIN_DISK_PRIVATE(disk);
918

919
    if (diskPriv->blockJobSync) {
920
        /* We have a SYNC API waiting for this event, dispatch it back */
921 922
        diskPriv->blockJobType = type;
        diskPriv->blockJobStatus = status;
923 924
        VIR_FREE(diskPriv->blockJobError);
        ignore_value(VIR_STRDUP_QUIET(diskPriv->blockJobError, error));
925
        virDomainObjBroadcast(vm);
926 927 928 929 930 931 932 933 934
    } 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;
935
        processEvent->vm = virObjectRef(vm);
936 937
        processEvent->action = type;
        processEvent->status = status;
938

939 940 941 942
        if (virThreadPoolSendJob(driver->workerPool, 0, processEvent) < 0) {
            ignore_value(virObjectUnref(vm));
            goto error;
        }
943 944
    }

945
 cleanup:
946
    virObjectUnlock(vm);
947
    return 0;
948
 error:
949
    qemuProcessEventFree(processEvent);
950
    goto cleanup;
951
}
952

953

954 955 956 957 958 959 960 961 962 963 964 965
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,
966 967
                          const char *saslUsername,
                          void *opaque)
968
{
969
    virQEMUDriverPtr driver = opaque;
970
    virObjectEventPtr event;
971 972 973
    virDomainEventGraphicsAddressPtr localAddr = NULL;
    virDomainEventGraphicsAddressPtr remoteAddr = NULL;
    virDomainEventGraphicsSubjectPtr subject = NULL;
974
    size_t i;
975 976

    if (VIR_ALLOC(localAddr) < 0)
977
        goto error;
978
    localAddr->family = localFamily;
979 980 981
    if (VIR_STRDUP(localAddr->service, localService) < 0 ||
        VIR_STRDUP(localAddr->node, localNode) < 0)
        goto error;
982 983

    if (VIR_ALLOC(remoteAddr) < 0)
984
        goto error;
985
    remoteAddr->family = remoteFamily;
986 987 988
    if (VIR_STRDUP(remoteAddr->service, remoteService) < 0 ||
        VIR_STRDUP(remoteAddr->node, remoteNode) < 0)
        goto error;
989 990

    if (VIR_ALLOC(subject) < 0)
991
        goto error;
992 993
    if (x509dname) {
        if (VIR_REALLOC_N(subject->identities, subject->nidentity+1) < 0)
994
            goto error;
995
        subject->nidentity++;
996 997 998
        if (VIR_STRDUP(subject->identities[subject->nidentity-1].type, "x509dname") < 0 ||
            VIR_STRDUP(subject->identities[subject->nidentity-1].name, x509dname) < 0)
            goto error;
999 1000 1001
    }
    if (saslUsername) {
        if (VIR_REALLOC_N(subject->identities, subject->nidentity+1) < 0)
1002
            goto error;
1003
        subject->nidentity++;
1004 1005 1006
        if (VIR_STRDUP(subject->identities[subject->nidentity-1].type, "saslUsername") < 0 ||
            VIR_STRDUP(subject->identities[subject->nidentity-1].name, saslUsername) < 0)
            goto error;
1007 1008
    }

1009
    virObjectLock(vm);
1010
    event = virDomainEventGraphicsNewFromObj(vm, phase, localAddr, remoteAddr, authScheme, subject);
1011
    virObjectUnlock(vm);
1012

1013
    virObjectEventStateQueue(driver->domainEventState, event);
1014 1015 1016

    return 0;

1017
 error:
1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028
    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) {
1029
        for (i = 0; i < subject->nidentity; i++) {
1030 1031 1032 1033 1034 1035 1036 1037 1038 1039
            VIR_FREE(subject->identities[i].type);
            VIR_FREE(subject->identities[i].name);
        }
        VIR_FREE(subject->identities);
        VIR_FREE(subject);
    }

    return -1;
}

1040 1041 1042 1043
static int
qemuProcessHandleTrayChange(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                            virDomainObjPtr vm,
                            const char *devAlias,
1044 1045
                            int reason,
                            void *opaque)
1046
{
1047
    virQEMUDriverPtr driver = opaque;
1048
    virObjectEventPtr event = NULL;
1049
    virDomainDiskDefPtr disk;
1050
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
1051

1052
    virObjectLock(vm);
1053 1054 1055 1056 1057 1058
    disk = qemuProcessFindDomainDiskByAlias(vm, devAlias);

    if (disk) {
        event = virDomainEventTrayChangeNewFromObj(vm,
                                                   devAlias,
                                                   reason);
1059 1060 1061 1062 1063 1064
        /* 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;

1065
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0) {
1066 1067 1068
            VIR_WARN("Unable to save status on vm %s after tray moved event",
                     vm->def->name);
        }
1069 1070

        virDomainObjBroadcast(vm);
1071 1072
    }

1073
    virObjectUnlock(vm);
1074
    virObjectEventStateQueue(driver->domainEventState, event);
1075
    virObjectUnref(cfg);
1076 1077 1078
    return 0;
}

O
Osier Yang 已提交
1079 1080
static int
qemuProcessHandlePMWakeup(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
1081 1082
                          virDomainObjPtr vm,
                          void *opaque)
O
Osier Yang 已提交
1083
{
1084
    virQEMUDriverPtr driver = opaque;
1085 1086
    virObjectEventPtr event = NULL;
    virObjectEventPtr lifecycleEvent = NULL;
1087
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
O
Osier Yang 已提交
1088

1089
    virObjectLock(vm);
O
Osier Yang 已提交
1090 1091
    event = virDomainEventPMWakeupNewFromObj(vm);

1092 1093 1094 1095 1096 1097 1098 1099 1100
    /* 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);
1101
        lifecycleEvent = virDomainEventLifecycleNewFromObj(vm,
1102 1103 1104
                                                  VIR_DOMAIN_EVENT_STARTED,
                                                  VIR_DOMAIN_EVENT_STARTED_WAKEUP);

1105
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0) {
1106 1107 1108 1109 1110
            VIR_WARN("Unable to save status on vm %s after wakeup event",
                     vm->def->name);
        }
    }

1111
    virObjectUnlock(vm);
1112 1113
    virObjectEventStateQueue(driver->domainEventState, event);
    virObjectEventStateQueue(driver->domainEventState, lifecycleEvent);
1114
    virObjectUnref(cfg);
O
Osier Yang 已提交
1115 1116
    return 0;
}
1117

O
Osier Yang 已提交
1118 1119
static int
qemuProcessHandlePMSuspend(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
1120 1121
                           virDomainObjPtr vm,
                           void *opaque)
O
Osier Yang 已提交
1122
{
1123
    virQEMUDriverPtr driver = opaque;
1124 1125
    virObjectEventPtr event = NULL;
    virObjectEventPtr lifecycleEvent = NULL;
1126
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
O
Osier Yang 已提交
1127

1128
    virObjectLock(vm);
O
Osier Yang 已提交
1129 1130
    event = virDomainEventPMSuspendNewFromObj(vm);

1131
    if (virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING) {
1132
        qemuDomainObjPrivatePtr priv = vm->privateData;
1133 1134 1135 1136 1137
        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
1138
        lifecycleEvent =
1139
            virDomainEventLifecycleNewFromObj(vm,
J
Jiri Denemark 已提交
1140 1141
                                     VIR_DOMAIN_EVENT_PMSUSPENDED,
                                     VIR_DOMAIN_EVENT_PMSUSPENDED_MEMORY);
1142

1143
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0) {
1144 1145 1146
            VIR_WARN("Unable to save status on vm %s after suspend event",
                     vm->def->name);
        }
1147 1148 1149

        if (priv->agent)
            qemuAgentNotifyEvent(priv->agent, QEMU_AGENT_EVENT_SUSPEND);
1150 1151
    }

1152
    virObjectUnlock(vm);
O
Osier Yang 已提交
1153

1154 1155
    virObjectEventStateQueue(driver->domainEventState, event);
    virObjectEventStateQueue(driver->domainEventState, lifecycleEvent);
1156
    virObjectUnref(cfg);
O
Osier Yang 已提交
1157 1158 1159
    return 0;
}

1160 1161 1162
static int
qemuProcessHandleBalloonChange(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                               virDomainObjPtr vm,
1163 1164
                               unsigned long long actual,
                               void *opaque)
1165
{
1166
    virQEMUDriverPtr driver = opaque;
1167
    virObjectEventPtr event = NULL;
1168
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
1169

1170
    virObjectLock(vm);
1171 1172 1173 1174 1175 1176
    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;

1177
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0)
1178 1179
        VIR_WARN("unable to save domain status with balloon change");

1180
    virObjectUnlock(vm);
1181

1182
    virObjectEventStateQueue(driver->domainEventState, event);
1183
    virObjectUnref(cfg);
1184 1185 1186
    return 0;
}

1187 1188
static int
qemuProcessHandlePMSuspendDisk(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
1189 1190
                               virDomainObjPtr vm,
                               void *opaque)
1191
{
1192
    virQEMUDriverPtr driver = opaque;
1193 1194
    virObjectEventPtr event = NULL;
    virObjectEventPtr lifecycleEvent = NULL;
1195
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
1196

1197
    virObjectLock(vm);
1198 1199 1200 1201 1202 1203 1204 1205 1206 1207
    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 =
1208
            virDomainEventLifecycleNewFromObj(vm,
1209 1210 1211
                                     VIR_DOMAIN_EVENT_PMSUSPENDED,
                                     VIR_DOMAIN_EVENT_PMSUSPENDED_DISK);

1212
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0) {
1213 1214 1215 1216 1217 1218 1219 1220
            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);
    }

1221
    virObjectUnlock(vm);
1222

1223 1224
    virObjectEventStateQueue(driver->domainEventState, event);
    virObjectEventStateQueue(driver->domainEventState, lifecycleEvent);
1225 1226
    virObjectUnref(cfg);

1227 1228 1229
    return 0;
}

1230

1231 1232
static int
qemuProcessHandleGuestPanic(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
1233
                            virDomainObjPtr vm,
1234
                            qemuMonitorEventPanicInfoPtr info,
1235
                            void *opaque)
1236
{
1237
    virQEMUDriverPtr driver = opaque;
1238 1239 1240
    struct qemuProcessEvent *processEvent;

    virObjectLock(vm);
1241
    if (VIR_ALLOC(processEvent) < 0)
1242 1243 1244 1245
        goto cleanup;

    processEvent->eventType = QEMU_PROCESS_EVENT_GUESTPANIC;
    processEvent->action = vm->def->onCrash;
1246
    processEvent->data = info;
1247 1248 1249
    /* Hold an extra reference because we can't allow 'vm' to be
     * deleted before handling guest panic event is finished.
     */
1250 1251
    processEvent->vm = virObjectRef(vm);

1252 1253 1254
    if (virThreadPoolSendJob(driver->workerPool, 0, processEvent) < 0) {
        if (!virObjectUnref(vm))
            vm = NULL;
1255
        qemuProcessEventFree(processEvent);
1256 1257
    }

1258
 cleanup:
1259
    if (vm)
1260
        virObjectUnlock(vm);
1261 1262 1263 1264 1265

    return 0;
}


1266
int
1267 1268
qemuProcessHandleDeviceDeleted(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                               virDomainObjPtr vm,
1269 1270
                               const char *devAlias,
                               void *opaque)
1271
{
1272
    virQEMUDriverPtr driver = opaque;
1273 1274
    struct qemuProcessEvent *processEvent = NULL;
    char *data;
1275 1276 1277 1278 1279 1280

    virObjectLock(vm);

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

1281 1282
    if (qemuDomainSignalDeviceRemoval(vm, devAlias,
                                      QEMU_DOMAIN_UNPLUGGING_DEVICE_STATUS_OK))
1283
        goto cleanup;
1284

1285 1286
    if (VIR_ALLOC(processEvent) < 0)
        goto error;
1287

1288 1289 1290 1291
    processEvent->eventType = QEMU_PROCESS_EVENT_DEVICE_DELETED;
    if (VIR_STRDUP(data, devAlias) < 0)
        goto error;
    processEvent->data = data;
1292
    processEvent->vm = virObjectRef(vm);
1293

1294 1295 1296 1297
    if (virThreadPoolSendJob(driver->workerPool, 0, processEvent) < 0) {
        ignore_value(virObjectUnref(vm));
        goto error;
    }
1298

1299
 cleanup:
1300 1301
    virObjectUnlock(vm);
    return 0;
1302
 error:
1303
    qemuProcessEventFree(processEvent);
1304
    goto cleanup;
1305 1306 1307
}


1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369
/**
 *
 * 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);
1370
    virObjectEventStateQueue(driver->domainEventState, event);
1371 1372 1373 1374 1375

    return 0;
}


1376 1377 1378 1379 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
static int
qemuProcessHandleBlockThreshold(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                                virDomainObjPtr vm,
                                const char *nodename,
                                unsigned long long threshold,
                                unsigned long long excess,
                                void *opaque)
{
    virQEMUDriverPtr driver = opaque;
    virObjectEventPtr event = NULL;
    virDomainDiskDefPtr disk;
    virStorageSourcePtr src;
    unsigned int idx;
    char *dev = NULL;
    const char *path = NULL;

    virObjectLock(vm);

    VIR_DEBUG("BLOCK_WRITE_THRESHOLD event for block node '%s' in domain %p %s:"
              "threshold '%llu' exceeded by '%llu'",
              nodename, vm, vm->def->name, threshold, excess);

    if ((disk = qemuDomainDiskLookupByNodename(vm->def, nodename, &src, &idx))) {
        if (virStorageSourceIsLocalStorage(src))
            path = src->path;

        if ((dev = qemuDomainDiskBackingStoreGetName(disk, src, idx))) {
            event = virDomainEventBlockThresholdNewFromObj(vm, dev, path,
                                                           threshold, excess);
            VIR_FREE(dev);
        }
    }

    virObjectUnlock(vm);
1410
    virObjectEventStateQueue(driver->domainEventState, event);
1411 1412 1413 1414 1415

    return 0;
}


1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437
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;
1438
    processEvent->vm = virObjectRef(vm);
1439 1440 1441 1442 1443 1444 1445 1446 1447 1448

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

 cleanup:
    virObjectUnlock(vm);
    return 0;
 error:
1449
    qemuProcessEventFree(processEvent);
1450 1451 1452 1453
    goto cleanup;
}


1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477
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;
1478
    processEvent->vm = virObjectRef(vm);
1479 1480 1481 1482 1483 1484 1485 1486 1487 1488

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

 cleanup:
    virObjectUnlock(vm);
    return 0;
 error:
1489
    qemuProcessEventFree(processEvent);
1490 1491 1492 1493
    goto cleanup;
}


1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512
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;
1513
    virDomainObjBroadcast(vm);
1514 1515 1516 1517 1518 1519 1520

 cleanup:
    virObjectUnlock(vm);
    return 0;
}


1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535
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;
1536
    if (priv->job.asyncJob == QEMU_ASYNC_JOB_NONE) {
1537 1538 1539 1540
        VIR_DEBUG("got MIGRATION event without a migration job");
        goto cleanup;
    }

1541
    priv->job.current->stats.mig.status = status;
1542 1543 1544 1545 1546 1547 1548 1549
    virDomainObjBroadcast(vm);

 cleanup:
    virObjectUnlock(vm);
    return 0;
}


1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569
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;
    }

1570
    virObjectEventStateQueue(driver->domainEventState,
1571 1572 1573 1574 1575 1576 1577 1578
                         virDomainEventMigrationIterationNewFromObj(vm, pass));

 cleanup:
    virObjectUnlock(vm);
    return 0;
}


1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 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
static int
qemuProcessHandleDumpCompleted(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                               virDomainObjPtr vm,
                               int status,
                               qemuMonitorDumpStatsPtr stats,
                               const char *error,
                               void *opaque ATTRIBUTE_UNUSED)
{
    qemuDomainObjPrivatePtr priv;

    virObjectLock(vm);

    VIR_DEBUG("Dump completed for domain %p %s with stats=%p error='%s'",
              vm, vm->def->name, stats, NULLSTR(error));

    priv = vm->privateData;
    if (priv->job.asyncJob == QEMU_ASYNC_JOB_NONE) {
        VIR_DEBUG("got DUMP_COMPLETED event without a dump_completed job");
        goto cleanup;
    }
    priv->job.dumpCompleted = true;
    priv->job.current->stats.dump = *stats;
    ignore_value(VIR_STRDUP_QUIET(priv->job.error, error));

    /* Force error if extracting the DUMP_COMPLETED status failed */
    if (!error && status < 0) {
        ignore_value(VIR_STRDUP_QUIET(priv->job.error, virGetLastErrorMessage()));
        priv->job.current->stats.dump.status = QEMU_MONITOR_DUMP_STATUS_FAILED;
    }

    virDomainObjBroadcast(vm);

 cleanup:
    virResetLastError();
    virObjectUnlock(vm);
    return 0;
}


1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671
static int
qemuProcessHandlePRManagerStatusChanged(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                                        virDomainObjPtr vm,
                                        const char *prManager,
                                        bool connected,
                                        void *opaque)
{
    virQEMUDriverPtr driver = opaque;
    qemuDomainObjPrivatePtr priv;
    struct qemuProcessEvent *processEvent = NULL;
    const char *managedAlias = qemuDomainGetManagedPRAlias();
    int ret = -1;

    virObjectLock(vm);

    VIR_DEBUG("pr-manager %s status changed for domain %p %s connected=%d",
              prManager, vm, vm->def->name, connected);

    if (connected) {
        /* Connect events are boring. */
        ret = 0;
        goto cleanup;
    }
    /* Disconnect events are more interesting. */

    if (STRNEQ(prManager, managedAlias)) {
        VIR_DEBUG("pr-manager %s not managed, ignoring event",
                  prManager);
        ret = 0;
        goto cleanup;
    }

    priv = vm->privateData;
    priv->prDaemonRunning = false;

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

    processEvent->eventType = QEMU_PROCESS_EVENT_PR_DISCONNECT;
    processEvent->vm = virObjectRef(vm);

    if (virThreadPoolSendJob(driver->workerPool, 0, processEvent) < 0) {
        qemuProcessEventFree(processEvent);
        virObjectUnref(vm);
        goto cleanup;
    }

    ret = 0;
 cleanup:
    virObjectUnlock(vm);
    return ret;
}


1672 1673
static qemuMonitorCallbacks monitorCallbacks = {
    .eofNotify = qemuProcessHandleMonitorEOF,
1674
    .errorNotify = qemuProcessHandleMonitorError,
1675
    .domainEvent = qemuProcessHandleEvent,
1676 1677
    .domainShutdown = qemuProcessHandleShutdown,
    .domainStop = qemuProcessHandleStop,
1678
    .domainResume = qemuProcessHandleResume,
1679 1680 1681 1682 1683
    .domainReset = qemuProcessHandleReset,
    .domainRTCChange = qemuProcessHandleRTCChange,
    .domainWatchdog = qemuProcessHandleWatchdog,
    .domainIOError = qemuProcessHandleIOError,
    .domainGraphics = qemuProcessHandleGraphics,
1684
    .domainBlockJob = qemuProcessHandleBlockJob,
1685
    .domainTrayChange = qemuProcessHandleTrayChange,
O
Osier Yang 已提交
1686
    .domainPMWakeup = qemuProcessHandlePMWakeup,
O
Osier Yang 已提交
1687
    .domainPMSuspend = qemuProcessHandlePMSuspend,
1688
    .domainBalloonChange = qemuProcessHandleBalloonChange,
1689
    .domainPMSuspendDisk = qemuProcessHandlePMSuspendDisk,
1690
    .domainGuestPanic = qemuProcessHandleGuestPanic,
1691
    .domainDeviceDeleted = qemuProcessHandleDeviceDeleted,
1692
    .domainNicRxFilterChanged = qemuProcessHandleNicRxFilterChanged,
1693
    .domainSerialChange = qemuProcessHandleSerialChanged,
1694
    .domainSpiceMigrated = qemuProcessHandleSpiceMigrated,
1695
    .domainMigrationStatus = qemuProcessHandleMigrationStatus,
1696
    .domainMigrationPass = qemuProcessHandleMigrationPass,
1697
    .domainAcpiOstInfo = qemuProcessHandleAcpiOstInfo,
1698
    .domainBlockThreshold = qemuProcessHandleBlockThreshold,
1699
    .domainDumpCompleted = qemuProcessHandleDumpCompleted,
1700
    .domainPRManagerStatusChanged = qemuProcessHandlePRManagerStatusChanged,
1701 1702
};

1703 1704 1705 1706 1707 1708 1709 1710 1711 1712
static void
qemuProcessMonitorReportLogError(qemuMonitorPtr mon,
                                 const char *msg,
                                 void *opaque);


static void
qemuProcessMonitorLogFree(void *opaque)
{
    qemuDomainLogContextPtr logCtxt = opaque;
1713
    virObjectUnref(logCtxt);
1714 1715
}

1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735

static int
qemuProcessInitMonitor(virQEMUDriverPtr driver,
                       virDomainObjPtr vm,
                       qemuDomainAsyncJob asyncJob)
{
    int ret;

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

    ret = qemuMonitorSetCapabilities(QEMU_DOMAIN_PRIVATE(vm)->mon);

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

    return ret;
}


1736
static int
1737
qemuConnectMonitor(virQEMUDriverPtr driver, virDomainObjPtr vm, int asyncJob,
1738
                   bool retry, qemuDomainLogContextPtr logCtxt)
1739 1740
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
1741
    qemuMonitorPtr mon = NULL;
1742
    unsigned long long timeout = 0;
1743
    virDomainChrSourceDefPtr monConfig;
1744

1745
    if (qemuSecuritySetDaemonSocketLabel(driver->securityManager, vm->def) < 0) {
1746 1747
        VIR_ERROR(_("Failed to set security context for monitor for %s"),
                  vm->def->name);
1748
        return -1;
1749 1750
    }

1751 1752 1753 1754 1755 1756
    /* When using hugepages, kernel zeroes them out before
     * handing them over to qemu. This can be very time
     * consuming. Therefore, add a second to timeout for each
     * 1GiB of guest RAM. */
    timeout = vm->def->mem.total_memory / (1024 * 1024);

1757
    /* Hold an extra reference because we can't allow 'vm' to be
C
Chen Hanxiao 已提交
1758
     * deleted until the monitor gets its own reference. */
1759
    virObjectRef(vm);
1760

1761
    ignore_value(virTimeMillisNow(&priv->monStart));
1762 1763
    monConfig = priv->monConfig;
    virObjectRef(monConfig);
1764
    virObjectUnlock(vm);
1765 1766

    mon = qemuMonitorOpen(vm,
1767
                          monConfig,
1768
                          priv->monJSON,
1769
                          retry,
1770
                          timeout,
1771 1772
                          &monitorCallbacks,
                          driver);
1773

1774
    if (mon && logCtxt) {
1775
        virObjectRef(logCtxt);
1776 1777 1778 1779 1780
        qemuMonitorSetDomainLog(mon,
                                qemuProcessMonitorReportLogError,
                                logCtxt,
                                qemuProcessMonitorLogFree);
    }
1781

1782
    virObjectLock(vm);
1783
    virObjectUnref(monConfig);
M
Michal Privoznik 已提交
1784
    virObjectUnref(vm);
1785
    priv->monStart = 0;
1786

M
Michal Privoznik 已提交
1787
    if (!virDomainObjIsActive(vm)) {
1788
        qemuMonitorClose(mon);
1789
        mon = NULL;
1790 1791 1792
    }
    priv->mon = mon;

1793
    if (qemuSecurityClearSocketLabel(driver->securityManager, vm->def) < 0) {
1794 1795
        VIR_ERROR(_("Failed to clear security context for monitor for %s"),
                  vm->def->name);
1796
        return -1;
1797 1798 1799 1800
    }

    if (priv->mon == NULL) {
        VIR_INFO("Failed to connect monitor for %s", vm->def->name);
1801
        return -1;
1802 1803
    }

1804 1805
    if (qemuProcessInitMonitor(driver, vm, asyncJob) < 0)
        return -1;
1806

1807
    if (qemuMigrationCapsCheck(driver, vm, asyncJob) < 0)
1808 1809 1810
        return -1;

    return 0;
1811 1812
}

1813 1814 1815

/**
 * qemuProcessReadLog: Read log file of a qemu VM
1816
 * @logCtxt: the domain log context
1817
 * @msg: pointer to buffer to store the read messages in
1818
 * @max: maximum length of the message returned in @msg
1819 1820
 *
 * Reads log of a qemu VM. Skips messages not produced by qemu or irrelevant
1821 1822 1823 1824
 * messages. If @max is not zero, @msg will contain at most @max characters
 * from the end of the log and @msg will start after a new line if possible.
 *
 * Returns 0 on success or -1 on error
1825
 */
1826
static int
1827 1828 1829
qemuProcessReadLog(qemuDomainLogContextPtr logCtxt,
                   char **msg,
                   size_t max)
1830
{
1831 1832
    char *buf;
    ssize_t got;
1833
    char *eol;
1834
    char *filter_next;
1835
    size_t skip;
1836

1837
    if ((got = qemuDomainLogContextRead(logCtxt, &buf)) < 0)
1838
        return -1;
1839

1840 1841 1842 1843 1844
    /* Filter out debug messages from intermediate libvirt process */
    filter_next = buf;
    while ((eol = strchr(filter_next, '\n'))) {
        *eol = '\0';
        if (virLogProbablyLogMessage(filter_next) ||
1845
            strstr(filter_next, "char device redirected to")) {
1846
            skip = (eol + 1) - filter_next;
1847
            memmove(filter_next, eol + 1, buf + got - eol);
1848 1849 1850 1851
            got -= skip;
        } else {
            filter_next = eol + 1;
            *eol = '\n';
1852 1853
        }
    }
1854
    filter_next = NULL; /* silence false coverity warning */
1855

1856 1857
    if (got > 0 &&
        buf[got - 1] == '\n') {
1858 1859
        buf[got - 1] = '\0';
        got--;
1860
    }
1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873

    if (max > 0 && got > max) {
        skip = got - max;

        if (buf[skip - 1] != '\n' &&
            (eol = strchr(buf + skip, '\n')) &&
            !virStringIsEmpty(eol + 1))
            skip = eol + 1 - buf;

        memmove(buf, buf + skip, got - skip + 1);
        got -= skip;
    }

1874
    ignore_value(VIR_REALLOC_N_QUIET(buf, got + 1));
1875 1876 1877
    *msg = buf;
    return 0;
}
1878 1879


1880 1881
static int
qemuProcessReportLogError(qemuDomainLogContextPtr logCtxt,
1882 1883 1884
                          const char *msgprefix)
{
    char *logmsg = NULL;
1885 1886 1887 1888 1889 1890
    size_t max;

    max = VIR_ERROR_MAX_LENGTH - 1;
    max -= strlen(msgprefix);
    /* The length of the formatting string minus two '%s' */
    max -= strlen(_("%s: %s")) - 4;
1891

1892
    if (qemuProcessReadLog(logCtxt, &logmsg, max) < 0)
1893 1894 1895
        return -1;

    virResetLastError();
1896 1897 1898 1899 1900
    if (virStringIsEmpty(logmsg))
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s", msgprefix);
    else
        virReportError(VIR_ERR_INTERNAL_ERROR, _("%s: %s"), msgprefix, logmsg);

1901 1902
    VIR_FREE(logmsg);
    return 0;
1903 1904 1905
}


1906 1907 1908 1909 1910 1911 1912 1913 1914 1915
static void
qemuProcessMonitorReportLogError(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                                 const char *msg,
                                 void *opaque)
{
    qemuDomainLogContextPtr logCtxt = opaque;
    qemuProcessReportLogError(logCtxt, msg);
}


1916
static int
1917
qemuProcessLookupPTYs(virDomainChrDefPtr *devices,
1918
                      int count,
1919
                      virHashTablePtr info)
1920
{
1921
    char *id = NULL;
1922
    size_t i;
1923
    int ret = -1;
1924

1925
    for (i = 0; i < count; i++) {
1926
        virDomainChrDefPtr chr = devices[i];
1927
        if (chr->source->type == VIR_DOMAIN_CHR_TYPE_PTY) {
1928
            qemuMonitorChardevInfoPtr entry;
1929

1930 1931
            VIR_FREE(id);
            if (virAsprintf(&id, "char%s", chr->info.alias) < 0)
1932 1933
                return -1;

1934 1935
            entry = virHashLookup(info, id);
            if (!entry || !entry->ptyPath) {
1936
                if (chr->source->data.file.path == NULL) {
1937 1938 1939
                    /* neither the log output nor 'info chardev' had a
                     * pty path for this chardev, report an error
                     */
1940 1941
                    virReportError(VIR_ERR_INTERNAL_ERROR,
                                   _("no assigned pty for device %s"), id);
1942
                    goto cleanup;
1943 1944 1945 1946 1947 1948 1949 1950
                } else {
                    /* 'info chardev' had no pty path for this chardev,
                     * but the log output had, so we're fine
                     */
                    continue;
                }
            }

1951 1952
            VIR_FREE(chr->source->data.file.path);
            if (VIR_STRDUP(chr->source->data.file.path, entry->ptyPath) < 0)
1953
                goto cleanup;
1954 1955 1956
        }
    }

1957 1958 1959 1960
    ret = 0;
 cleanup:
    VIR_FREE(id);
    return ret;
1961 1962
}

1963 1964
static int
qemuProcessFindCharDevicePTYsMonitor(virDomainObjPtr vm,
1965
                                     virHashTablePtr info)
1966
{
1967
    size_t i = 0;
C
Cole Robinson 已提交
1968

1969
    if (qemuProcessLookupPTYs(vm->def->serials, vm->def->nserials, info) < 0)
1970 1971
        return -1;

1972
    if (qemuProcessLookupPTYs(vm->def->parallels, vm->def->nparallels,
1973
                              info) < 0)
1974
        return -1;
1975

1976
    if (qemuProcessLookupPTYs(vm->def->channels, vm->def->nchannels, info) < 0)
1977
        return -1;
1978 1979 1980 1981
    /* 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];
1982

1983 1984 1985 1986 1987
        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;
1988 1989
            if (virDomainChrSourceDefCopy(chr->source,
                                          ((vm->def->serials[0])->source)) < 0)
1990 1991 1992 1993
                return -1;
        }
    }

1994
    if (qemuProcessLookupPTYs(vm->def->consoles + i, vm->def->nconsoles - i,
1995
                              info) < 0)
1996
        return -1;
1997 1998 1999 2000 2001

    return 0;
}


2002
static int
2003 2004 2005 2006
qemuProcessRefreshChannelVirtioState(virQEMUDriverPtr driver,
                                     virDomainObjPtr vm,
                                     virHashTablePtr info,
                                     int booted)
2007 2008
{
    size_t i;
2009
    int agentReason = VIR_CONNECT_DOMAIN_EVENT_AGENT_LIFECYCLE_REASON_CHANNEL;
2010
    qemuMonitorChardevInfoPtr entry;
2011
    virObjectEventPtr event = NULL;
2012 2013
    char *id = NULL;
    int ret = -1;
2014

2015 2016 2017
    if (booted)
        agentReason = VIR_CONNECT_DOMAIN_EVENT_AGENT_LIFECYCLE_REASON_DOMAIN_STARTED;

2018 2019 2020
    for (i = 0; i < vm->def->nchannels; i++) {
        virDomainChrDefPtr chr = vm->def->channels[i];
        if (chr->targetType == VIR_DOMAIN_CHR_CHANNEL_TARGET_TYPE_VIRTIO) {
2021 2022 2023 2024

            VIR_FREE(id);
            if (virAsprintf(&id, "char%s", chr->info.alias) < 0)
                goto cleanup;
2025 2026 2027 2028 2029 2030

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

2031 2032 2033 2034
            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)))
2035
                virObjectEventStateQueue(driver->domainEventState, event);
2036

2037 2038 2039 2040
            chr->state = entry->state;
        }
    }

2041 2042 2043 2044
    ret = 0;
 cleanup:
    VIR_FREE(id);
    return ret;
2045 2046 2047
}


2048 2049
int
qemuRefreshVirtioChannelState(virQEMUDriverPtr driver,
2050 2051
                              virDomainObjPtr vm,
                              qemuDomainAsyncJob asyncJob)
2052 2053 2054 2055 2056
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virHashTablePtr info = NULL;
    int ret = -1;

2057 2058 2059
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        goto cleanup;

2060
    ret = qemuMonitorGetChardevInfo(priv->mon, &info);
2061 2062
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        ret = -1;
2063 2064 2065 2066

    if (ret < 0)
        goto cleanup;

2067
    ret = qemuProcessRefreshChannelVirtioState(driver, vm, info, false);
2068 2069 2070 2071 2072 2073

 cleanup:
    virHashFree(info);
    return ret;
}

2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130

static int
qemuProcessRefreshPRManagerState(virDomainObjPtr vm,
                                 virHashTablePtr info)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    qemuMonitorPRManagerInfoPtr prManagerInfo;
    const char *managedAlias = qemuDomainGetManagedPRAlias();
    int ret = -1;

    if (!(prManagerInfo = virHashLookup(info, managedAlias))) {
        virReportError(VIR_ERR_OPERATION_FAILED,
                       _("missing info on pr-manager %s"),
                       managedAlias);
        goto cleanup;
    }

    priv->prDaemonRunning = prManagerInfo->connected;

    if (!priv->prDaemonRunning &&
        qemuProcessStartManagedPRDaemon(vm) < 0)
        goto cleanup;

    ret = 0;
 cleanup:
    return ret;
}


static int
qemuRefreshPRManagerState(virQEMUDriverPtr driver,
                          virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virHashTablePtr info = NULL;
    int ret = -1;

    if (!virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_PR_MANAGER_HELPER) ||
        !virDomainDefHasManagedPR(vm->def))
        return 0;

    qemuDomainObjEnterMonitor(driver, vm);
    ret = qemuMonitorGetPRManagerInfo(priv->mon, &info);
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        ret = -1;

    if (ret < 0)
        goto cleanup;

    ret = qemuProcessRefreshPRManagerState(vm, info);

 cleanup:
    virHashFree(info);
    return ret;
}


2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154
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;
2155
    if ((then = mktime(&thenbits)) == (time_t)-1) {
2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166
        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;
}
2167

2168
int
2169 2170 2171 2172 2173 2174 2175 2176 2177
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 */
2178
    if (!virDomainDefHasMemballoon(vm->def)) {
2179
        vm->def->mem.cur_balloon = virDomainDefGetMemoryTotal(vm->def);
2180 2181 2182 2183 2184 2185 2186
        return 0;
    }

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

    rc = qemuMonitorGetBalloonInfo(qemuDomainGetMonitor(vm), &balloon);
2187
    if (qemuDomainObjExitMonitor(driver, vm) < 0 || rc < 0)
2188 2189 2190 2191 2192 2193 2194 2195
        return -1;

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

    return 0;
}


2196
static int
2197
qemuProcessWaitForMonitor(virQEMUDriverPtr driver,
C
Cole Robinson 已提交
2198
                          virDomainObjPtr vm,
2199
                          int asyncJob,
2200
                          qemuDomainLogContextPtr logCtxt)
2201 2202
{
    int ret = -1;
2203
    virHashTablePtr info = NULL;
2204 2205 2206 2207 2208 2209
    qemuDomainObjPrivatePtr priv = vm->privateData;
    bool retry = true;

    if (priv->qemuCaps &&
        virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_CHARDEV_FD_PASS))
        retry = false;
2210

2211 2212 2213 2214
    VIR_DEBUG("Connect monitor to vm=%p name='%s' retry=%d",
              vm, vm->def->name, retry);

    if (qemuConnectMonitor(driver, vm, asyncJob, retry, logCtxt) < 0)
2215 2216 2217 2218 2219 2220
        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. */
2221 2222
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        goto cleanup;
2223 2224
    ret = qemuMonitorGetChardevInfo(priv->mon, &info);
    VIR_DEBUG("qemuMonitorGetChardevInfo returned %i", ret);
2225 2226 2227
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        ret = -1;

2228
    if (ret == 0) {
2229
        if ((ret = qemuProcessFindCharDevicePTYsMonitor(vm, info)) < 0)
2230 2231
            goto cleanup;

2232 2233
        if ((ret = qemuProcessRefreshChannelVirtioState(driver, vm, info,
                                                        true)) < 0)
2234 2235
            goto cleanup;
    }
2236

2237
 cleanup:
2238
    virHashFree(info);
2239

2240 2241
    if (logCtxt && kill(vm->pid, 0) == -1 && errno == ESRCH) {
        qemuProcessReportLogError(logCtxt,
2242
                                  _("process exited while connecting to monitor"));
2243 2244 2245 2246 2247 2248
        ret = -1;
    }

    return ret;
}

2249

2250 2251 2252 2253 2254 2255
static int
qemuProcessDetectIOThreadPIDs(virQEMUDriverPtr driver,
                              virDomainObjPtr vm,
                              int asyncJob)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
2256
    qemuMonitorIOThreadInfoPtr *iothreads = NULL;
2257 2258 2259 2260
    int niothreads = 0;
    int ret = -1;
    size_t i;

2261 2262 2263 2264 2265
    if (!virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_OBJECT_IOTHREAD)) {
        ret = 0;
        goto cleanup;
    }

2266 2267 2268 2269
    /* Get the list of IOThreads from qemu */
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        goto cleanup;
    niothreads = qemuMonitorGetIOThreads(priv->mon, &iothreads);
2270 2271
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        goto cleanup;
J
John Ferlan 已提交
2272
    if (niothreads < 0)
2273 2274
        goto cleanup;

2275
    if (niothreads != vm->def->niothreadids) {
2276 2277
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("got wrong number of IOThread pids from QEMU monitor. "
2278 2279
                         "got %d, wanted %zu"),
                       niothreads, vm->def->niothreadids);
2280 2281 2282
        goto cleanup;
    }

2283 2284 2285 2286 2287 2288
    /* Nothing to do */
    if (niothreads == 0) {
        ret = 0;
        goto cleanup;
    }

2289 2290 2291
    for (i = 0; i < niothreads; i++) {
        virDomainIOThreadIDDefPtr iothrid;

2292 2293
        if (!(iothrid = virDomainIOThreadIDFind(vm->def,
                                                iothreads[i]->iothread_id))) {
2294
            virReportError(VIR_ERR_INTERNAL_ERROR,
2295 2296
                           _("iothread %d not found"),
                           iothreads[i]->iothread_id);
2297 2298 2299 2300
            goto cleanup;
        }
        iothrid->thread_id = iothreads[i]->thread_id;
    }
2301 2302 2303 2304 2305 2306

    ret = 0;

 cleanup:
    if (iothreads) {
        for (i = 0; i < niothreads; i++)
2307
            VIR_FREE(iothreads[i]);
2308 2309 2310 2311 2312
        VIR_FREE(iothreads);
    }
    return ret;
}

2313 2314 2315 2316 2317

/*
 * To be run between fork/exec of QEMU only
 */
static int
2318
qemuProcessInitCpuAffinity(virDomainObjPtr vm)
2319 2320 2321 2322
{
    int ret = -1;
    virBitmapPtr cpumap = NULL;
    virBitmapPtr cpumapToSet = NULL;
2323
    virBitmapPtr hostcpumap = NULL;
2324
    qemuDomainObjPrivatePtr priv = vm->privateData;
2325

2326 2327 2328 2329 2330 2331
    if (!vm->pid) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Cannot setup CPU affinity until process is started"));
        return -1;
    }

2332 2333
    if (vm->def->placement_mode == VIR_DOMAIN_CPU_PLACEMENT_MODE_AUTO) {
        VIR_DEBUG("Set CPU affinity with advisory nodeset from numad");
2334
        cpumapToSet = priv->autoCpuset;
2335
    } else {
2336
        VIR_DEBUG("Set CPU affinity with specified cpuset");
O
Osier Yang 已提交
2337
        if (vm->def->cpumask) {
H
Hu Tao 已提交
2338
            cpumapToSet = vm->def->cpumask;
O
Osier Yang 已提交
2339 2340 2341 2342 2343
        } 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 */
2344 2345
            int hostcpus;

2346 2347 2348 2349 2350 2351 2352 2353 2354 2355
            if (virHostCPUHasBitmap()) {
                hostcpumap = virHostCPUGetOnlineBitmap();
                cpumap = virProcessGetAffinity(vm->pid);
            }

            if (hostcpumap && cpumap && virBitmapEqual(hostcpumap, cpumap)) {
                /* we're using all available CPUs, no reason to set
                 * mask. If libvirtd is running without explicit
                 * affinity, we can use hotplugged CPUs for this VM */
                ret = 0;
2356
                goto cleanup;
2357 2358 2359 2360 2361
            } else {
                /* setaffinity fails if you set bits for CPUs which
                 * aren't present, so we have to limit ourselves */
                if ((hostcpus = virHostCPUGetCount()) < 0)
                    goto cleanup;
2362

2363 2364
                if (hostcpus > QEMUD_CPUMASK_LEN)
                    hostcpus = QEMUD_CPUMASK_LEN;
2365

2366 2367 2368
                virBitmapFree(cpumap);
                if (!(cpumap = virBitmapNew(hostcpus)))
                    goto cleanup;
2369

2370
                virBitmapSetAll(cpumap);
2371

2372 2373
                cpumapToSet = cpumap;
            }
O
Osier Yang 已提交
2374
        }
2375 2376
    }

2377
    if (virProcessSetAffinity(vm->pid, cpumapToSet) < 0)
2378
        goto cleanup;
2379

2380 2381
    ret = 0;

2382
 cleanup:
2383
    virBitmapFree(cpumap);
2384
    virBitmapFree(hostcpumap);
2385
    return ret;
2386 2387
}

2388 2389
/* set link states to down on interfaces at qemu start */
static int
2390 2391 2392
qemuProcessSetLinkStates(virQEMUDriverPtr driver,
                         virDomainObjPtr vm,
                         qemuDomainAsyncJob asyncJob)
2393 2394 2395
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virDomainDefPtr def = vm->def;
2396
    size_t i;
2397 2398 2399 2400 2401
    int ret = -1;
    int rv;

    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;
2402 2403 2404

    for (i = 0; i < def->nnets; i++) {
        if (def->nets[i]->linkstate == VIR_DOMAIN_NET_INTERFACE_LINK_STATE_DOWN) {
2405 2406 2407
            if (!def->nets[i]->info.alias) {
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("missing alias for network device"));
2408
                goto cleanup;
2409 2410
            }

2411 2412
            VIR_DEBUG("Setting link state: %s", def->nets[i]->info.alias);

2413 2414 2415 2416
            rv = qemuMonitorSetLink(priv->mon,
                                    def->nets[i]->info.alias,
                                    VIR_DOMAIN_NET_INTERFACE_LINK_STATE_DOWN);
            if (rv < 0) {
2417
                virReportError(VIR_ERR_OPERATION_FAILED,
2418 2419 2420
                               _("Couldn't set link state on interface: %s"),
                               def->nets[i]->info.alias);
                goto cleanup;
2421 2422 2423 2424
            }
        }
    }

2425 2426 2427 2428 2429
    ret = 0;

 cleanup:
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        ret = -1;
2430 2431 2432
    return ret;
}

2433

2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541
/**
 * 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;
}


2542
static int
2543
qemuProcessSetupEmulator(virDomainObjPtr vm)
2544
{
2545 2546 2547 2548 2549
    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);
2550 2551
}

2552

2553 2554 2555 2556 2557 2558
static int
qemuProcessResctrlCreate(virQEMUDriverPtr driver,
                         virDomainObjPtr vm)
{
    int ret = -1;
    size_t i = 0;
2559
    virCapsPtr caps = NULL;
2560 2561
    qemuDomainObjPrivatePtr priv = vm->privateData;

2562 2563 2564 2565 2566 2567
    if (!vm->def->ncachetunes)
        return 0;

    /* Force capability refresh since resctrl info can change
     * XXX: move cache info into virresctrl so caps are not needed */
    caps = virQEMUDriverGetCapabilities(driver, true);
2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584
    if (!caps)
        return -1;

    for (i = 0; i < vm->def->ncachetunes; i++) {
        if (virResctrlAllocCreate(caps->host.resctrl,
                                  vm->def->cachetunes[i]->alloc,
                                  priv->machineName) < 0)
            goto cleanup;
    }

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


2585 2586 2587 2588 2589 2590 2591 2592 2593 2594
static char *
qemuProcessBuildPRHelperPidfilePath(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    const char *prdAlias = qemuDomainGetManagedPRAlias();

    return virPidFileBuildPath(priv->libDir, prdAlias);
}


2595
void
2596
qemuProcessKillManagedPRDaemon(virDomainObjPtr vm)
2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virErrorPtr orig_err;
    char *pidfile;

    if (!priv->prDaemonRunning)
        return;

    if (!(pidfile = qemuProcessBuildPRHelperPidfilePath(vm))) {
        VIR_WARN("Unable to construct pr-helper pidfile path");
        return;
    }

    virErrorPreserveLast(&orig_err);
    if (virPidFileForceCleanupPath(pidfile) < 0) {
        VIR_WARN("Unable to kill pr-helper process");
    } else {
        if (unlink(pidfile) < 0 &&
            errno != ENOENT) {
            virReportSystemError(errno,
                                 _("Unable to remove stale pidfile %s"),
                                 pidfile);
        } else {
            priv->prDaemonRunning = false;
        }
    }
    virErrorRestore(&orig_err);

    VIR_FREE(pidfile);
}


static int
qemuProcessStartPRDaemonHook(void *opaque)
{
    virDomainObjPtr vm = opaque;
    size_t i, nfds = 0;
    int *fds = NULL;
    int ret = -1;

2637 2638 2639
    if (qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT)) {
        if (virProcessGetNamespaces(vm->pid, &nfds, &fds) < 0)
            return ret;
2640

2641 2642 2643 2644
        if (nfds > 0 &&
            virProcessSetNamespaces(nfds, fds) < 0)
            goto cleanup;
    }
2645 2646 2647 2648 2649 2650 2651 2652 2653 2654

    ret = 0;
 cleanup:
    for (i = 0; i < nfds; i++)
        VIR_FORCE_CLOSE(fds[i]);
    VIR_FREE(fds);
    return ret;
}


2655
int
2656
qemuProcessStartManagedPRDaemon(virDomainObjPtr vm)
2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virQEMUDriverPtr driver = priv->driver;
    virQEMUDriverConfigPtr cfg;
    int errfd = -1;
    char *pidfile = NULL;
    int pidfd = -1;
    char *socketPath = NULL;
    pid_t cpid = -1;
    virCommandPtr cmd = NULL;
    virTimeBackOffVar timebackoff;
    const unsigned long long timeout = 500000; /* ms */
    int ret = -1;

    cfg = virQEMUDriverGetConfig(driver);

    if (!virFileIsExecutable(cfg->prHelperName)) {
        virReportSystemError(errno, _("'%s' is not a suitable pr helper"),
                             cfg->prHelperName);
        goto cleanup;
    }

    if (!(pidfile = qemuProcessBuildPRHelperPidfilePath(vm)))
        goto cleanup;

    /* Just try to acquire. Dummy pid will be replaced later */
    if ((pidfd = virPidFileAcquirePath(pidfile, false, -1)) < 0)
        goto cleanup;

2686
    if (!(socketPath = qemuDomainGetManagedPRSocketPath(priv)))
2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760
        goto cleanup;

    /* Remove stale socket */
    if (unlink(socketPath) < 0 &&
        errno != ENOENT) {
        virReportSystemError(errno,
                             _("Unable to remove stale socket path: %s"),
                             socketPath);
        goto cleanup;
    }

    if (!(cmd = virCommandNewArgList(cfg->prHelperName,
                                     "-k", socketPath,
                                     "-f", pidfile,
                                     NULL)))
        goto cleanup;

    virCommandDaemonize(cmd);
    /* We want our virCommand to write child PID into the pidfile
     * so that we can read it even before exec(). */
    virCommandSetPidFile(cmd, pidfile);
    virCommandSetErrorFD(cmd, &errfd);

    /* Place the process into the same namespace and cgroup as
     * qemu (so that it shares the same view of the system). */
    virCommandSetPreExecHook(cmd, qemuProcessStartPRDaemonHook, vm);

    if (virCommandRun(cmd, NULL) < 0)
        goto cleanup;

    if (virPidFileReadPath(pidfile, &cpid) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("pr helper %s didn't show up"),
                       cfg->prHelperName);
        goto cleanup;
    }

    if (virTimeBackOffStart(&timebackoff, 1, timeout) < 0)
        goto cleanup;
    while (virTimeBackOffWait(&timebackoff)) {
        char errbuf[1024] = { 0 };

        if (virFileExists(socketPath))
            break;

        if (virProcessKill(cpid, 0) == 0)
            continue;

        if (saferead(errfd, errbuf, sizeof(errbuf) - 1) < 0) {
            virReportSystemError(errno,
                                 _("pr helper %s died unexpectedly"),
                                 cfg->prHelperName);
        } else {
            virReportError(VIR_ERR_OPERATION_FAILED,
                           _("pr helper died and reported: %s"), errbuf);
        }
        goto cleanup;
    }

    if (!virFileExists(socketPath)) {
        virReportError(VIR_ERR_OPERATION_TIMEOUT, "%s",
                       _("pr helper socked did not show up"));
        goto cleanup;
    }

    if (priv->cgroup &&
        virCgroupAddMachineTask(priv->cgroup, cpid) < 0)
        goto cleanup;

    if (qemuSecurityDomainSetPathLabel(driver->securityManager,
                                       vm->def, socketPath, true) < 0)
        goto cleanup;

    priv->prDaemonRunning = true;
2761
    ret = 0;
2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779
 cleanup:
    if (ret < 0) {
        virCommandAbort(cmd);
        if (cpid >= 0)
            virProcessKillPainfully(cpid, true);
        if (pidfile)
            unlink(pidfile);
    }
    virCommandFree(cmd);
    VIR_FREE(socketPath);
    VIR_FORCE_CLOSE(pidfd);
    VIR_FREE(pidfile);
    VIR_FORCE_CLOSE(errfd);
    virObjectUnref(cfg);
    return ret;
}


2780
static int
2781
qemuProcessInitPasswords(virQEMUDriverPtr driver,
2782 2783
                         virDomainObjPtr vm,
                         int asyncJob)
2784 2785
{
    int ret = 0;
2786
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
2787
    size_t i;
2788

2789
    for (i = 0; i < vm->def->ngraphics; ++i) {
2790 2791
        virDomainGraphicsDefPtr graphics = vm->def->graphics[i];
        if (graphics->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC) {
2792 2793
            ret = qemuDomainChangeGraphicsPasswords(driver, vm,
                                                    VIR_DOMAIN_GRAPHICS_TYPE_VNC,
2794
                                                    &graphics->data.vnc.auth,
2795 2796
                                                    cfg->vncPassword,
                                                    asyncJob);
2797
        } else if (graphics->type == VIR_DOMAIN_GRAPHICS_TYPE_SPICE) {
2798 2799
            ret = qemuDomainChangeGraphicsPasswords(driver, vm,
                                                    VIR_DOMAIN_GRAPHICS_TYPE_SPICE,
2800
                                                    &graphics->data.spice.auth,
2801 2802
                                                    cfg->spicePassword,
                                                    asyncJob);
2803 2804
        }

2805 2806 2807
        if (ret < 0)
            goto cleanup;
    }
2808

2809
 cleanup:
2810
    virObjectUnref(cfg);
2811 2812 2813 2814 2815 2816 2817 2818 2819 2820
    return ret;
}


static int
qemuProcessPrepareChardevDevice(virDomainDefPtr def ATTRIBUTE_UNUSED,
                                virDomainChrDefPtr dev,
                                void *opaque ATTRIBUTE_UNUSED)
{
    int fd;
2821
    if (dev->source->type != VIR_DOMAIN_CHR_TYPE_FILE)
2822 2823
        return 0;

2824
    if ((fd = open(dev->source->data.file.path,
2825 2826 2827
                   O_CREAT | O_APPEND, S_IRUSR|S_IWUSR)) < 0) {
        virReportSystemError(errno,
                             _("Unable to pre-create chardev file '%s'"),
2828
                             dev->source->data.file.path);
2829 2830 2831 2832 2833 2834 2835 2836 2837
        return -1;
    }

    VIR_FORCE_CLOSE(fd);

    return 0;
}


2838 2839 2840 2841 2842
static int
qemuProcessCleanupChardevDevice(virDomainDefPtr def ATTRIBUTE_UNUSED,
                                virDomainChrDefPtr dev,
                                void *opaque ATTRIBUTE_UNUSED)
{
2843 2844 2845 2846
    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);
2847 2848 2849 2850 2851

    return 0;
}


2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887
/**
 * 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) {
2888
                if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_QXL_VGAMEM) &&
2889 2890
                    qemuMonitorUpdateVideoMemorySize(priv->mon, video,
                                                     "qxl-vga") < 0)
2891
                        goto error;
2892

2893
                if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_QXL_VRAM64) &&
2894 2895 2896
                    qemuMonitorUpdateVideoVram64Size(priv->mon, video,
                                                     "qxl-vga") < 0)
                    goto error;
2897
            } else {
2898 2899 2900 2901 2902 2903 2904 2905
                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)
2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924
                        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;
        }

    }

2925 2926
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        return -1;
2927 2928

    cfg = virQEMUDriverGetConfig(driver);
2929
    ret = virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps);
2930 2931 2932 2933 2934
    virObjectUnref(cfg);

    return ret;

 error:
2935
    ignore_value(qemuDomainObjExitMonitor(driver, vm));
2936 2937 2938 2939
    return -1;
}


2940 2941
struct qemuProcessHookData {
    virDomainObjPtr vm;
2942
    virQEMUDriverPtr driver;
2943
    virQEMUDriverConfigPtr cfg;
2944 2945 2946 2947 2948
};

static int qemuProcessHook(void *data)
{
    struct qemuProcessHookData *h = data;
2949
    qemuDomainObjPrivatePtr priv = h->vm->privateData;
2950
    int ret = -1;
2951
    int fd;
2952 2953 2954
    virBitmapPtr nodeset = NULL;
    virDomainNumatuneMemMode mode;

2955 2956 2957 2958
    /* This method cannot use any mutexes, which are not
     * protected across fork()
     */

2959
    qemuSecurityPostFork(h->driver->securityManager);
2960 2961 2962 2963 2964

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

    VIR_DEBUG("Obtaining domain lock");
2965 2966 2967 2968 2969 2970 2971
    /*
     * 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.
     */
2972
    if (qemuSecuritySetSocketLabel(h->driver->securityManager, h->vm->def) < 0)
2973
        goto cleanup;
2974
    if (virDomainLockProcessStart(h->driver->lockManager,
2975
                                  h->cfg->uri,
2976
                                  h->vm,
J
Ján Tomko 已提交
2977
                                  /* QEMU is always paused initially */
2978 2979
                                  true,
                                  &fd) < 0)
2980
        goto cleanup;
2981
    if (qemuSecurityClearSocketLabel(h->driver->securityManager, h->vm->def) < 0)
2982
        goto cleanup;
2983

2984
    if (qemuDomainBuildNamespace(h->cfg, h->driver->securityManager, h->vm) < 0)
2985 2986
        goto cleanup;

2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997
    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);
2998

2999 3000 3001
            if (virNumaSetupMemoryPolicy(mode, nodeset) < 0)
                goto cleanup;
        }
3002
    }
3003

3004 3005
    ret = 0;

3006
 cleanup:
3007
    virObjectUnref(h->cfg);
3008 3009
    VIR_DEBUG("Hook complete ret=%d", ret);
    return ret;
3010 3011 3012
}

int
3013 3014
qemuProcessPrepareMonitorChr(virDomainChrSourceDefPtr monConfig,
                             const char *domainDir)
3015 3016 3017 3018
{
    monConfig->type = VIR_DOMAIN_CHR_TYPE_UNIX;
    monConfig->data.nix.listen = true;

3019 3020
    if (virAsprintf(&monConfig->data.nix.path, "%s/monitor.sock",
                    domainDir) < 0)
3021 3022
        return -1;
    return 0;
3023 3024 3025
}


3026
/*
3027 3028
 * Precondition: vm must be locked, and a job must be active.
 * This method will call {Enter,Exit}Monitor
3029
 */
E
Eric Blake 已提交
3030
int
3031
qemuProcessStartCPUs(virQEMUDriverPtr driver, virDomainObjPtr vm,
3032
                     virDomainRunningReason reason,
3033
                     qemuDomainAsyncJob asyncJob)
3034
{
3035
    int ret = -1;
3036
    qemuDomainObjPrivatePtr priv = vm->privateData;
3037
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
3038

3039
    /* Bring up netdevs before starting CPUs */
3040
    if (qemuInterfaceStartDevices(vm->def) < 0)
3041 3042
       goto cleanup;

3043
    VIR_DEBUG("Using lock state '%s'", NULLSTR(priv->lockState));
3044
    if (virDomainLockProcessResume(driver->lockManager, cfg->uri,
3045
                                   vm, priv->lockState) < 0) {
3046 3047 3048 3049
        /* 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
         */
3050
        goto cleanup;
3051 3052 3053
    }
    VIR_FREE(priv->lockState);

3054 3055
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        goto release;
J
Jiri Denemark 已提交
3056

3057
    ret = qemuMonitorStartCPUs(priv->mon);
3058 3059
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        ret = -1;
3060 3061 3062 3063 3064

    if (ret < 0)
        goto release;

    virDomainObjSetState(vm, VIR_DOMAIN_RUNNING, reason);
3065

3066
 cleanup:
3067
    virObjectUnref(cfg);
3068
    return ret;
3069 3070 3071 3072 3073 3074

 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;
3075 3076 3077
}


3078 3079
int qemuProcessStopCPUs(virQEMUDriverPtr driver,
                        virDomainObjPtr vm,
3080
                        virDomainPausedReason reason,
3081
                        qemuDomainAsyncJob asyncJob)
3082
{
3083
    int ret = -1;
3084 3085
    qemuDomainObjPrivatePtr priv = vm->privateData;

3086
    VIR_FREE(priv->lockState);
J
Jiri Denemark 已提交
3087

3088 3089
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        goto cleanup;
J
Jiri Denemark 已提交
3090

3091
    ret = qemuMonitorStopCPUs(priv->mon);
3092 3093
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        ret = -1;
3094 3095 3096 3097

    if (ret < 0)
        goto cleanup;

3098 3099 3100
    /* de-activate netdevs after stopping CPUs */
    ignore_value(qemuInterfaceStopDevices(vm->def));

3101 3102 3103
    if (priv->job.current)
        ignore_value(virTimeMillisNow(&priv->job.current->stopped));

3104 3105 3106 3107
    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
3108

3109
 cleanup:
3110 3111 3112 3113 3114
    return ret;
}



3115
static void
3116 3117
qemuProcessNotifyNets(virDomainDefPtr def)
{
3118
    size_t i;
3119

3120 3121
    for (i = 0; i < def->nnets; i++) {
        virDomainNetDefPtr net = def->nets[i];
3122 3123 3124 3125 3126 3127 3128 3129
        /* 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));

3130
        virDomainNetNotifyActualDevice(def, net);
3131 3132 3133
    }
}

3134
static int
3135
qemuProcessFiltersInstantiate(virDomainDefPtr def, bool ignoreExists)
3136
{
3137
    size_t i;
3138

3139
    for (i = 0; i < def->nnets; i++) {
3140 3141
        virDomainNetDefPtr net = def->nets[i];
        if ((net->filter) && (net->ifname)) {
3142
            if (virDomainConfNWFilterInstantiate(def->name, def->uuid, net, ignoreExists) < 0)
J
Ján Tomko 已提交
3143
                return 1;
3144 3145 3146
        }
    }

J
Ján Tomko 已提交
3147
    return 0;
3148 3149
}

3150
static int
3151
qemuProcessUpdateState(virQEMUDriverPtr driver, virDomainObjPtr vm)
3152 3153 3154
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virDomainState state;
3155
    virDomainPausedReason reason;
3156
    virDomainState newState = VIR_DOMAIN_NOSTATE;
3157
    int oldReason;
3158
    int newReason;
3159
    bool running;
3160
    char *msg = NULL;
3161 3162
    int ret;

3163
    qemuDomainObjEnterMonitor(driver, vm);
3164
    ret = qemuMonitorGetStatus(priv->mon, &running, &reason);
3165 3166
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        return -1;
3167

3168
    if (ret < 0)
3169 3170
        return -1;

3171
    state = virDomainObjGetState(vm, &oldReason);
3172

3173 3174 3175 3176 3177 3178 3179 3180
    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) {
3181 3182
        newState = VIR_DOMAIN_RUNNING;
        newReason = VIR_DOMAIN_RUNNING_UNPAUSED;
3183
        ignore_value(VIR_STRDUP_QUIET(msg, "was unpaused"));
3184
    } else if (state == VIR_DOMAIN_RUNNING && !running) {
3185 3186 3187
        if (reason == VIR_DOMAIN_PAUSED_SHUTTING_DOWN) {
            newState = VIR_DOMAIN_SHUTDOWN;
            newReason = VIR_DOMAIN_SHUTDOWN_UNKNOWN;
3188
            ignore_value(VIR_STRDUP_QUIET(msg, "shutdown"));
3189
        } else if (reason == VIR_DOMAIN_PAUSED_CRASHED) {
3190 3191 3192
            newState = VIR_DOMAIN_CRASHED;
            newReason = VIR_DOMAIN_CRASHED_PANICKED;
            ignore_value(VIR_STRDUP_QUIET(msg, "crashed"));
3193 3194 3195
        } else {
            newState = VIR_DOMAIN_PAUSED;
            newReason = reason;
S
Stefan Berger 已提交
3196 3197
            ignore_value(virAsprintf(&msg, "was paused (%s)",
                                 virDomainPausedReasonTypeToString(reason)));
3198 3199 3200 3201 3202 3203 3204
        }
    }

    if (newState != VIR_DOMAIN_NOSTATE) {
        VIR_DEBUG("Domain %s %s while its monitor was disconnected;"
                  " changing state to %s (%s)",
                  vm->def->name,
3205
                  NULLSTR(msg),
3206 3207 3208 3209
                  virDomainStateTypeToString(newState),
                  virDomainStateReasonToString(newState, newReason));
        VIR_FREE(msg);
        virDomainObjSetState(vm, newState, newReason);
3210 3211 3212 3213 3214
    }

    return 0;
}

3215
static int
3216 3217
qemuProcessRecoverMigrationIn(virQEMUDriverPtr driver,
                              virDomainObjPtr vm,
3218
                              const qemuDomainJobObj *job,
3219
                              virDomainState state,
3220
                              int reason)
3221
{
3222 3223 3224 3225 3226
    bool postcopy = (state == VIR_DOMAIN_PAUSED &&
                     reason == VIR_DOMAIN_PAUSED_POSTCOPY_FAILED) ||
                    (state == VIR_DOMAIN_RUNNING &&
                     reason == VIR_DOMAIN_RUNNING_POSTCOPY);

3227
    switch ((qemuMigrationJobPhase) job->phase) {
3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242
    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;
3243

3244 3245 3246 3247 3248
    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);
3249
        if (qemuProcessStartCPUs(driver, vm,
3250 3251 3252 3253 3254 3255 3256 3257 3258
                                 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
3259 3260
         * we already started guest CPUs or we were in post-copy mode */
        if (postcopy) {
3261
            qemuMigrationAnyPostcopyFailed(driver, vm);
3262
        } else if (state != VIR_DOMAIN_RUNNING) {
3263
            VIR_DEBUG("Killing migrated domain %s", vm->def->name);
3264
            return -1;
3265 3266 3267
        }
        break;
    }
3268

3269 3270
    qemuMigrationParamsReset(driver, vm, QEMU_ASYNC_JOB_NONE,
                             job->migParams, job->apiFlags);
3271 3272
    return 0;
}
3273

3274 3275 3276
static int
qemuProcessRecoverMigrationOut(virQEMUDriverPtr driver,
                               virDomainObjPtr vm,
3277
                               const qemuDomainJobObj *job,
3278
                               virDomainState state,
3279 3280
                               int reason,
                               unsigned int *stopFlags)
3281
{
3282 3283 3284
    bool postcopy = state == VIR_DOMAIN_PAUSED &&
                    (reason == VIR_DOMAIN_PAUSED_POSTCOPY ||
                     reason == VIR_DOMAIN_PAUSED_POSTCOPY_FAILED);
3285
    bool resume = false;
3286

3287
    switch ((qemuMigrationJobPhase) job->phase) {
3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303
    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
3304 3305 3306 3307
         * domain; however we can only do that before migration enters
         * post-copy mode
         */
        if (postcopy) {
3308
            qemuMigrationAnyPostcopyFailed(driver, vm);
3309 3310 3311
        } else {
            VIR_DEBUG("Cancelling unfinished migration of domain %s",
                      vm->def->name);
3312
            if (qemuMigrationSrcCancel(driver, vm) < 0) {
3313 3314 3315
                VIR_WARN("Could not cancel ongoing migration of domain %s",
                         vm->def->name);
            }
3316
            resume = true;
3317
        }
3318
        break;
3319

3320 3321
    case QEMU_MIGRATION_PHASE_PERFORM3_DONE:
        /* migration finished but we didn't have a chance to get the result
3322 3323
         * 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
3324
         */
3325
        if (postcopy)
3326
            qemuMigrationAnyPostcopyFailed(driver, vm);
3327
        break;
3328

3329
    case QEMU_MIGRATION_PHASE_CONFIRM3_CANCELLED:
3330 3331 3332 3333 3334
        /* 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) {
3335
            qemuMigrationAnyPostcopyFailed(driver, vm);
3336 3337 3338
        } else {
            VIR_DEBUG("Resuming domain %s after failed migration",
                      vm->def->name);
3339
            resume = true;
3340 3341
        }
        break;
3342

3343 3344
    case QEMU_MIGRATION_PHASE_CONFIRM3:
        /* migration completed, we need to kill the domain here */
3345
        *stopFlags |= VIR_QEMU_PROCESS_STOP_MIGRATED;
3346 3347
        return -1;
    }
3348

3349 3350 3351 3352 3353 3354 3355
    if (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)) {
3356
            if (qemuProcessStartCPUs(driver, vm,
3357 3358 3359 3360
                                     VIR_DOMAIN_RUNNING_UNPAUSED,
                                     QEMU_ASYNC_JOB_NONE) < 0) {
                VIR_WARN("Could not resume domain %s", vm->def->name);
            }
3361 3362
        }
    }
3363

3364 3365
    qemuMigrationParamsReset(driver, vm, QEMU_ASYNC_JOB_NONE,
                             job->migParams, job->apiFlags);
3366 3367 3368
    return 0;
}

3369
static int
3370
qemuProcessRecoverJob(virQEMUDriverPtr driver,
3371
                      virDomainObjPtr vm,
3372
                      const qemuDomainJobObj *job,
3373
                      unsigned int *stopFlags)
3374
{
3375
    qemuDomainObjPrivatePtr priv = vm->privateData;
3376 3377 3378 3379 3380 3381 3382
    virDomainState state;
    int reason;

    state = virDomainObjGetState(vm, &reason);

    switch (job->asyncJob) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
3383
        if (qemuProcessRecoverMigrationOut(driver, vm, job,
3384
                                           state, reason, stopFlags) < 0)
3385 3386 3387
            return -1;
        break;

3388
    case QEMU_ASYNC_JOB_MIGRATION_IN:
3389
        if (qemuProcessRecoverMigrationIn(driver, vm, job,
3390
                                          state, reason) < 0)
3391
            return -1;
3392 3393 3394 3395
        break;

    case QEMU_ASYNC_JOB_SAVE:
    case QEMU_ASYNC_JOB_DUMP:
3396
    case QEMU_ASYNC_JOB_SNAPSHOT:
3397
        qemuDomainObjEnterMonitor(driver, vm);
3398
        ignore_value(qemuMonitorMigrateCancel(priv->mon));
3399 3400
        if (qemuDomainObjExitMonitor(driver, vm) < 0)
            return -1;
3401
        /* resume the domain but only if it was paused as a result of
3402 3403 3404 3405 3406 3407 3408 3409 3410
         * 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 &&
3411 3412
               (reason == VIR_DOMAIN_PAUSED_SNAPSHOT ||
                reason == VIR_DOMAIN_PAUSED_MIGRATION)) ||
3413
              reason == VIR_DOMAIN_PAUSED_UNKNOWN)) {
3414
             if (qemuProcessStartCPUs(driver, vm,
3415 3416 3417 3418
                                      VIR_DOMAIN_RUNNING_UNPAUSED,
                                      QEMU_ASYNC_JOB_NONE) < 0) {
                 VIR_WARN("Could not resume domain '%s' after migration to file",
                          vm->def->name);
3419 3420 3421 3422
            }
        }
        break;

3423 3424 3425 3426
    case QEMU_ASYNC_JOB_START:
        /* Already handled in VIR_DOMAIN_PAUSED_STARTING_UP check. */
        break;

3427 3428 3429 3430 3431 3432 3433 3434
    case QEMU_ASYNC_JOB_NONE:
    case QEMU_ASYNC_JOB_LAST:
        break;
    }

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

3435 3436 3437 3438
    /* 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.
     */
3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458
    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;

3459
    case QEMU_JOB_MIGRATION_OP:
3460
    case QEMU_JOB_ABORT:
3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471
    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;
}

3472 3473 3474 3475 3476 3477
static int
qemuProcessUpdateDevices(virQEMUDriverPtr driver,
                         virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virDomainDeviceDef dev;
3478
    const char **qemuDevices;
3479 3480 3481 3482 3483 3484 3485 3486 3487
    char **old;
    char **tmp;
    int ret = -1;

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

    old = priv->qemuDevices;
    priv->qemuDevices = NULL;
3488
    if (qemuDomainUpdateDeviceList(driver, vm, QEMU_ASYNC_JOB_NONE) < 0)
3489 3490
        goto cleanup;

3491
    qemuDevices = (const char **)priv->qemuDevices;
3492 3493
    if ((tmp = old)) {
        while (*tmp) {
3494
            if (!virStringListHasString(qemuDevices, *tmp) &&
3495 3496 3497 3498
                virDomainDefFindDevice(vm->def, *tmp, &dev, false) == 0 &&
                qemuDomainRemoveDevice(driver, vm, &dev) < 0) {
                goto cleanup;
            }
3499 3500 3501 3502 3503
            tmp++;
        }
    }
    ret = 0;

3504
 cleanup:
3505
    virStringListFree(old);
3506 3507 3508
    return ret;
}

3509 3510 3511 3512 3513 3514 3515
static int
qemuDomainPerfRestart(virDomainObjPtr vm)
{
    size_t i;
    virDomainDefPtr def = vm->def;
    qemuDomainObjPrivatePtr priv = vm->privateData;

3516
    if (!(priv->perf = virPerfNew()))
3517 3518 3519
        return -1;

    for (i = 0; i < VIR_PERF_EVENT_LAST; i++) {
3520 3521
        if (def->perf.events[i] &&
            def->perf.events[i] == VIR_TRISTATE_BOOL_YES) {
3522 3523 3524

            /* Failure to re-enable the perf event should not be fatal */
            if (virPerfEventEnable(priv->perf, i, vm->pid) < 0)
3525
                def->perf.events[i] = VIR_TRISTATE_BOOL_NO;
3526 3527 3528 3529 3530 3531
        }
    }

    return 0;
}

3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554

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;
        }
    }
}


3555
static bool
3556 3557
qemuProcessNeedHugepagesPath(virDomainDefPtr def,
                             virDomainMemoryDefPtr mem)
3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576
{
    const long system_pagesize = virGetSystemPageSizeKB();
    size_t i;

    if (def->mem.source == VIR_DOMAIN_MEMORY_SOURCE_FILE)
        return true;

    for (i = 0; i < def->mem.nhugepages; i++) {
        if (def->mem.hugepages[i].size != system_pagesize)
            return true;
    }

    for (i = 0; i < def->nmems; i++) {
        if (def->mems[i]->model == VIR_DOMAIN_MEMORY_MODEL_DIMM &&
            def->mems[i]->pagesize &&
            def->mems[i]->pagesize != system_pagesize)
            return true;
    }

3577 3578 3579 3580 3581 3582
    if (mem &&
        mem->model == VIR_DOMAIN_MEMORY_MODEL_DIMM &&
        mem->pagesize &&
        mem->pagesize != system_pagesize)
        return true;

3583 3584 3585 3586
    return false;
}


3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616
static bool
qemuProcessNeedMemoryBackingPath(virDomainDefPtr def,
                                 virDomainMemoryDefPtr mem)
{
    size_t i;
    size_t numaNodes;

    if (def->mem.source == VIR_DOMAIN_MEMORY_SOURCE_FILE ||
        def->mem.access != VIR_DOMAIN_MEMORY_ACCESS_DEFAULT)
        return true;

    numaNodes = virDomainNumaGetNodeCount(def->numa);
    for (i = 0; i < numaNodes; i++) {
        if (virDomainNumaGetNodeMemoryAccessMode(def->numa, i)
            != VIR_DOMAIN_MEMORY_ACCESS_DEFAULT)
            return true;
    }

    if (mem &&
        mem->model == VIR_DOMAIN_MEMORY_MODEL_DIMM &&
        (mem->access != VIR_DOMAIN_MEMORY_ACCESS_DEFAULT ||
         (mem->targetNode >= 0 &&
          virDomainNumaGetNodeMemoryAccessMode(def->numa, mem->targetNode)
          != VIR_DOMAIN_MEMORY_ACCESS_DEFAULT)))
        return true;

    return false;
}


3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634
static int
qemuProcessBuildDestroyMemoryPathsImpl(virQEMUDriverPtr driver,
                                       virDomainDefPtr def,
                                       const char *path,
                                       bool build)
{
    if (build) {
        if (virFileExists(path))
            return 0;

        if (virFileMakePathWithMode(path, 0700) < 0) {
            virReportSystemError(errno,
                                 _("Unable to create %s"),
                                 path);
            return -1;
        }

        if (qemuSecurityDomainSetPathLabel(driver->securityManager,
3635
                                           def, path, true) < 0) {
3636 3637 3638 3639 3640
            virReportError(VIR_ERR_INTERNAL_ERROR,
                            _("Unable to label %s"), path);
            return -1;
        }
    } else {
M
Michal Privoznik 已提交
3641 3642
        if (virFileDeleteTree(path) < 0)
            return -1;
3643 3644 3645 3646 3647 3648
    }

    return 0;
}


3649
int
3650 3651 3652 3653
qemuProcessBuildDestroyMemoryPaths(virQEMUDriverPtr driver,
                                   virDomainObjPtr vm,
                                   virDomainMemoryDefPtr mem,
                                   bool build)
3654 3655
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
3656
    char *path = NULL;
3657
    size_t i;
3658 3659
    bool shouldBuildHP = false;
    bool shouldBuildMB = false;
3660 3661
    int ret = -1;

3662 3663 3664 3665
    if (build) {
        shouldBuildHP = qemuProcessNeedHugepagesPath(vm->def, mem);
        shouldBuildMB = qemuProcessNeedMemoryBackingPath(vm->def, mem);
    }
3666

3667
    if (!build || shouldBuildHP) {
3668
        for (i = 0; i < cfg->nhugetlbfs; i++) {
3669
            path = qemuGetDomainHugepagePath(vm->def, &cfg->hugetlbfs[i]);
3670

3671
            if (!path)
3672 3673
                goto cleanup;

3674
            if (qemuProcessBuildDestroyMemoryPathsImpl(driver, vm->def,
3675
                                                       path, build) < 0)
3676
                goto cleanup;
3677

3678
            VIR_FREE(path);
3679 3680 3681
        }
    }

3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692
    if (!build || shouldBuildMB) {
        if (qemuGetMemoryBackingDomainPath(vm->def, cfg, &path) < 0)
            goto cleanup;

        if (qemuProcessBuildDestroyMemoryPathsImpl(driver, vm->def,
                                                   path, build) < 0)
            goto cleanup;

        VIR_FREE(path);
    }

3693 3694
    ret = 0;
 cleanup:
3695
    VIR_FREE(path);
3696 3697 3698 3699 3700
    virObjectUnref(cfg);
    return ret;
}


3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726
int
qemuProcessDestroyMemoryBackingPath(virQEMUDriverPtr driver,
                                    virDomainObjPtr vm,
                                    virDomainMemoryDefPtr mem)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    char *path = NULL;
    int ret = -1;

    if (qemuGetMemoryBackingPath(vm->def, cfg, mem->info.alias, &path) < 0)
        goto cleanup;

    if (unlink(path) < 0 &&
        errno != ENOENT) {
        virReportSystemError(errno, _("Unable to remove %s"), path);
        goto cleanup;
    }

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


3727 3728 3729 3730
static int
qemuProcessVNCAllocatePorts(virQEMUDriverPtr driver,
                            virDomainGraphicsDefPtr graphics,
                            bool allocate)
3731
{
3732
    unsigned short port;
3733

3734 3735 3736
    if (!allocate) {
        if (graphics->data.vnc.autoport)
            graphics->data.vnc.port = 5900;
3737

3738 3739
        return 0;
    }
3740

3741 3742 3743 3744 3745
    if (graphics->data.vnc.autoport) {
        if (virPortAllocatorAcquire(driver->remotePorts, &port) < 0)
            return -1;
        graphics->data.vnc.port = port;
    }
3746

3747 3748 3749 3750 3751 3752
    if (graphics->data.vnc.websocket == -1) {
        if (virPortAllocatorAcquire(driver->webSocketPorts, &port) < 0)
            return -1;
        graphics->data.vnc.websocket = port;
        graphics->data.vnc.websocketGenerated = true;
    }
3753

3754 3755
    return 0;
}
J
John Ferlan 已提交
3756

3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767
static int
qemuProcessSPICEAllocatePorts(virQEMUDriverPtr driver,
                              virDomainGraphicsDefPtr graphics,
                              bool allocate)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    unsigned short port = 0;
    unsigned short tlsPort;
    size_t i;
    int defaultMode = graphics->data.spice.defaultMode;
    int ret = -1;
3768

3769 3770
    bool needTLSPort = false;
    bool needPort = false;
3771

3772 3773 3774 3775 3776 3777 3778
    if (graphics->data.spice.autoport) {
        /* check if tlsPort or port need allocation */
        for (i = 0; i < VIR_DOMAIN_GRAPHICS_SPICE_CHANNEL_LAST; i++) {
            switch (graphics->data.spice.channels[i]) {
            case VIR_DOMAIN_GRAPHICS_SPICE_CHANNEL_MODE_SECURE:
                needTLSPort = true;
                break;
3779

3780 3781 3782
            case VIR_DOMAIN_GRAPHICS_SPICE_CHANNEL_MODE_INSECURE:
                needPort = true;
                break;
3783

3784 3785 3786 3787 3788 3789 3790 3791 3792
            case VIR_DOMAIN_GRAPHICS_SPICE_CHANNEL_MODE_ANY:
                /* default mode will be used */
                break;
            }
        }
        switch (defaultMode) {
        case VIR_DOMAIN_GRAPHICS_SPICE_CHANNEL_MODE_SECURE:
            needTLSPort = true;
            break;
3793

3794 3795 3796
        case VIR_DOMAIN_GRAPHICS_SPICE_CHANNEL_MODE_INSECURE:
            needPort = true;
            break;
3797

3798 3799 3800 3801 3802 3803 3804
        case VIR_DOMAIN_GRAPHICS_SPICE_CHANNEL_MODE_ANY:
            if (cfg->spiceTLS)
                needTLSPort = true;
            needPort = true;
            break;
        }
    }
3805

3806 3807 3808
    if (!allocate) {
        if (needPort || graphics->data.spice.port == -1)
            graphics->data.spice.port = 5901;
3809

3810 3811
        if (needTLSPort || graphics->data.spice.tlsPort == -1)
            graphics->data.spice.tlsPort = 5902;
3812

3813 3814
        ret = 0;
        goto cleanup;
3815 3816
    }

3817 3818 3819
    if (needPort || graphics->data.spice.port == -1) {
        if (virPortAllocatorAcquire(driver->remotePorts, &port) < 0)
            goto cleanup;
3820

3821 3822 3823 3824
        graphics->data.spice.port = port;

        if (!graphics->data.spice.autoport)
            graphics->data.spice.portReserved = true;
3825 3826
    }

3827 3828 3829 3830 3831 3832 3833
    if (needTLSPort || graphics->data.spice.tlsPort == -1) {
        if (!cfg->spiceTLS) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("Auto allocation of spice TLS port requested "
                             "but spice TLS is disabled in qemu.conf"));
            goto cleanup;
        }
3834

3835 3836
        if (virPortAllocatorAcquire(driver->remotePorts, &tlsPort) < 0)
            goto cleanup;
3837

3838
        graphics->data.spice.tlsPort = tlsPort;
3839

3840 3841
        if (!graphics->data.spice.autoport)
            graphics->data.spice.tlsPortReserved = true;
3842
    }
3843

3844
    ret = 0;
3845

3846 3847 3848 3849
 cleanup:
    virObjectUnref(cfg);
    return ret;
}
3850

3851

3852 3853 3854 3855 3856
static int
qemuValidateCpuCount(virDomainDefPtr def,
                     virQEMUCapsPtr qemuCaps)
{
    unsigned int maxCpus = virQEMUCapsGetMachineMaxCpus(qemuCaps, def->os.machine);
3857

3858 3859 3860 3861 3862
    if (virDomainDefGetVcpus(def) == 0) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("Domain requires at least 1 vCPU"));
        return -1;
    }
3863

3864 3865 3866 3867 3868
    if (maxCpus > 0 && virDomainDefGetVcpusMax(def) > maxCpus) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("Maximum CPUs greater than specified machine type limit"));
        return -1;
    }
3869

3870 3871
    return 0;
}
3872

3873

3874 3875 3876 3877 3878 3879 3880
static int
qemuProcessVerifyHypervFeatures(virDomainDefPtr def,
                                virCPUDataPtr cpu)
{
    char *cpuFeature;
    size_t i;
    int rc;
3881

3882 3883 3884 3885
    for (i = 0; i < VIR_DOMAIN_HYPERV_LAST; i++) {
        /* always supported string property */
        if (i == VIR_DOMAIN_HYPERV_VENDOR_ID)
            continue;
3886

3887 3888
        if (def->hyperv_features[i] != VIR_TRISTATE_SWITCH_ON)
            continue;
3889

3890 3891 3892
        if (virAsprintf(&cpuFeature, "__kvm_hv_%s",
                        virDomainHypervTypeToString(i)) < 0)
            return -1;
3893

3894 3895
        rc = virCPUDataCheckFeature(cpu, cpuFeature);
        VIR_FREE(cpuFeature);
3896

3897 3898 3899 3900
        if (rc < 0)
            return -1;
        else if (rc == 1)
            continue;
3901

3902 3903 3904 3905 3906 3907 3908
        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;
3909

3910 3911 3912 3913 3914 3915 3916 3917 3918
        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:
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("host doesn't support hyperv '%s' feature"),
                           virDomainHypervTypeToString(i));
            return -1;
3919

3920 3921 3922 3923
        /* coverity[dead_error_begin] */
        case VIR_DOMAIN_HYPERV_VENDOR_ID:
        case VIR_DOMAIN_HYPERV_LAST:
            break;
3924
        }
3925
    }
3926 3927

    return 0;
3928 3929
}

3930

3931
static int
3932 3933
qemuProcessVerifyKVMFeatures(virDomainDefPtr def,
                             virCPUDataPtr cpu)
3934
{
3935
    int rc = 0;
3936

3937
    if (def->features[VIR_DOMAIN_FEATURE_PVSPINLOCK] != VIR_TRISTATE_SWITCH_ON)
3938 3939
        return 0;

3940 3941 3942 3943 3944 3945
    rc = virCPUDataCheckFeature(cpu, VIR_CPU_x86_KVM_PV_UNHALT);

    if (rc <= 0) {
        if (rc == 0)
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("host doesn't support paravirtual spinlocks"));
3946
        return -1;
3947
    }
3948

3949 3950
    return 0;
}
3951 3952


3953 3954 3955 3956 3957
static int
qemuProcessVerifyCPUFeatures(virDomainDefPtr def,
                             virCPUDataPtr cpu)
{
    int rc;
S
Stefan Berger 已提交
3958

3959
    rc = virCPUCheckFeature(def->os.arch, def->cpu, "invtsc");
3960

3961
    if (rc < 0) {
3962
        return -1;
3963 3964 3965 3966 3967 3968 3969 3970 3971
    } else if (rc == 1) {
        rc = virCPUDataCheckFeature(cpu, "invtsc");
        if (rc <= 0) {
            if (rc == 0) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("host doesn't support invariant TSC"));
            }
            return -1;
        }
3972
    }
3973

3974
    return 0;
3975 3976 3977
}


3978
static int
3979 3980 3981 3982 3983
qemuProcessFetchGuestCPU(virQEMUDriverPtr driver,
                         virDomainObjPtr vm,
                         qemuDomainAsyncJob asyncJob,
                         virCPUDataPtr *enabled,
                         virCPUDataPtr *disabled)
3984
{
3985 3986 3987 3988
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virCPUDataPtr dataEnabled = NULL;
    virCPUDataPtr dataDisabled = NULL;
    int rc;
3989

3990 3991
    *enabled = NULL;
    *disabled = NULL;
3992

3993
    if (!ARCH_IS_X86(vm->def->os.arch))
3994 3995
        return 0;

3996 3997
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        goto error;
3998

3999 4000
    rc = qemuMonitorGetGuestCPU(priv->mon, vm->def->os.arch,
                                &dataEnabled, &dataDisabled);
4001

4002 4003 4004 4005 4006 4007 4008 4009
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        goto error;

    if (rc == -1)
        goto error;

    *enabled = dataEnabled;
    *disabled = dataDisabled;
4010
    return 0;
4011 4012 4013 4014 4015

 error:
    virCPUDataFree(dataEnabled);
    virCPUDataFree(dataDisabled);
    return -1;
4016
}
4017

4018

4019
static int
4020 4021
qemuProcessVerifyCPU(virDomainObjPtr vm,
                     virCPUDataPtr cpu)
4022
{
4023
    virDomainDefPtr def = vm->def;
4024

4025 4026
    if (!cpu)
        return 0;
J
Jiri Denemark 已提交
4027 4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043

    if (qemuProcessVerifyKVMFeatures(def, cpu) < 0 ||
        qemuProcessVerifyHypervFeatures(def, cpu) < 0)
        return -1;

    if (!def->cpu ||
        (def->cpu->mode == VIR_CPU_MODE_CUSTOM &&
         !def->cpu->model))
        return 0;

    if (qemuProcessVerifyCPUFeatures(def, cpu) < 0)
        return -1;

    return 0;
}


4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064 4065 4066 4067 4068 4069 4070 4071 4072 4073 4074 4075 4076 4077 4078 4079 4080 4081 4082 4083 4084 4085
static int
qemuProcessUpdateLiveGuestCPU(virDomainObjPtr vm,
                              virCPUDataPtr enabled,
                              virCPUDataPtr disabled)
{
    virDomainDefPtr def = vm->def;
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virCPUDefPtr orig = NULL;
    int rc;
    int ret = -1;

    if (!enabled)
        return 0;

    if (!def->cpu ||
        (def->cpu->mode == VIR_CPU_MODE_CUSTOM &&
         !def->cpu->model))
        return 0;

    if (!(orig = virCPUDefCopy(def->cpu)))
        goto cleanup;

    if ((rc = virCPUUpdateLive(def->os.arch, def->cpu, enabled, disabled)) < 0) {
        goto cleanup;
    } else if (rc == 0) {
        /* Store the original CPU in priv if QEMU changed it and we didn't
         * get the original CPU via migration, restore, or snapshot revert.
         */
        if (!priv->origCPU && !virCPUDefIsEqual(def->cpu, orig, false))
            VIR_STEAL_PTR(priv->origCPU, orig);

        def->cpu->check = VIR_CPU_CHECK_FULL;
    }

    ret = 0;

 cleanup:
    virCPUDefFree(orig);
    return ret;
}


4086
static int
4087
qemuProcessUpdateAndVerifyCPU(virQEMUDriverPtr driver,
4088 4089
                              virDomainObjPtr vm,
                              qemuDomainAsyncJob asyncJob)
4090
{
4091
    virCPUDataPtr cpu = NULL;
4092
    virCPUDataPtr disabled = NULL;
4093
    int ret = -1;
4094

4095 4096
    if (qemuProcessFetchGuestCPU(driver, vm, asyncJob, &cpu, &disabled) < 0)
        goto cleanup;
4097

J
Jiri Denemark 已提交
4098 4099
    if (qemuProcessVerifyCPU(vm, cpu) < 0)
        goto cleanup;
J
Ján Tomko 已提交
4100

4101 4102
    if (qemuProcessUpdateLiveGuestCPU(vm, cpu, disabled) < 0)
        goto cleanup;
4103

4104
    ret = 0;
4105

4106
 cleanup:
4107
    virCPUDataFree(cpu);
4108
    virCPUDataFree(disabled);
4109 4110 4111 4112
    return ret;
}


4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 4127 4128 4129 4130 4131 4132 4133 4134 4135 4136
static virDomainCapsCPUModelsPtr
qemuProcessFetchCPUDefinitions(virQEMUDriverPtr driver,
                               virDomainObjPtr vm,
                               qemuDomainAsyncJob asyncJob)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virDomainCapsCPUModelsPtr models = NULL;

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

    models = virQEMUCapsFetchCPUDefinitions(priv->mon);

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

    return models;

 error:
    virObjectUnref(models);
    return NULL;
}


4137 4138 4139 4140 4141 4142 4143
static int
qemuProcessUpdateCPU(virQEMUDriverPtr driver,
                     virDomainObjPtr vm,
                     qemuDomainAsyncJob asyncJob)
{
    virCPUDataPtr cpu = NULL;
    virCPUDataPtr disabled = NULL;
4144
    virDomainCapsCPUModelsPtr models = NULL;
4145 4146
    int ret = -1;

4147 4148 4149 4150 4151
    /* The host CPU model comes from host caps rather than QEMU caps so
     * fallback must be allowed no matter what the user specified in the XML.
     */
    vm->def->cpu->fallback = VIR_CPU_FALLBACK_ALLOW;

4152 4153 4154 4155 4156 4157
    if (qemuProcessFetchGuestCPU(driver, vm, asyncJob, &cpu, &disabled) < 0)
        goto cleanup;

    if (qemuProcessUpdateLiveGuestCPU(vm, cpu, disabled) < 0)
        goto cleanup;

4158 4159 4160 4161
    if (!(models = qemuProcessFetchCPUDefinitions(driver, vm, asyncJob)) ||
        virCPUTranslate(vm->def->os.arch, vm->def->cpu, models) < 0)
        goto cleanup;

4162 4163 4164 4165 4166
    ret = 0;

 cleanup:
    virCPUDataFree(cpu);
    virCPUDataFree(disabled);
4167
    virObjectUnref(models);
4168 4169 4170 4171
    return ret;
}


4172 4173
static int
qemuPrepareNVRAM(virQEMUDriverConfigPtr cfg,
4174
                 virDomainObjPtr vm)
4175 4176 4177 4178
{
    int ret = -1;
    int srcFD = -1;
    int dstFD = -1;
4179
    virDomainLoaderDefPtr loader = vm->def->os.loader;
4180
    bool created = false;
4181 4182
    const char *master_nvram_path;
    ssize_t r;
4183

4184
    if (!loader || !loader->nvram || virFileExists(loader->nvram))
4185 4186
        return 0;

4187 4188 4189
    master_nvram_path = loader->templt;
    if (!loader->templt) {
        size_t i;
4190 4191 4192
        for (i = 0; i < cfg->nfirmwares; i++) {
            if (STREQ(cfg->firmwares[i]->name, loader->path)) {
                master_nvram_path = cfg->firmwares[i]->nvram;
4193
                break;
4194 4195
            }
        }
4196
    }
4197

4198 4199 4200 4201 4202 4203
    if (!master_nvram_path) {
        virReportError(VIR_ERR_OPERATION_FAILED,
                       _("unable to find any master var store for "
                         "loader: %s"), loader->path);
        goto cleanup;
    }
4204

4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218 4219 4220 4221
    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;
4222

4223 4224
    do {
        char buf[1024];
4225

4226
        if ((r = saferead(srcFD, buf, sizeof(buf))) < 0) {
4227
            virReportSystemError(errno,
4228
                                 _("Unable to read from file '%s'"),
4229 4230 4231
                                 master_nvram_path);
            goto cleanup;
        }
4232 4233

        if (safewrite(dstFD, buf, r) < 0) {
4234
            virReportSystemError(errno,
4235
                                 _("Unable to write to file '%s'"),
4236 4237 4238
                                 loader->nvram);
            goto cleanup;
        }
4239 4240 4241 4242 4243 4244 4245 4246 4247 4248 4249 4250 4251
    } 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;
4252 4253 4254 4255 4256 4257 4258 4259 4260 4261 4262 4263 4264 4265 4266 4267 4268
    }

    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;
}


4269 4270 4271
static void
qemuLogOperation(virDomainObjPtr vm,
                 const char *msg,
4272 4273
                 virCommandPtr cmd,
                 qemuDomainLogContextPtr logCtxt)
4274 4275 4276 4277 4278
{
    char *timestamp;
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int qemuVersion = virQEMUCapsGetVersion(priv->qemuCaps);
    const char *package = virQEMUCapsGetPackage(priv->qemuCaps);
4279
    char *hostname = virGetHostname();
4280 4281 4282
    struct utsname uts;

    uname(&uts);
4283 4284

    if ((timestamp = virTimeStringNow()) == NULL)
4285
        goto cleanup;
4286

4287
    if (qemuDomainLogContextWrite(logCtxt,
4288
                                  "%s: %s %s, qemu version: %d.%d.%d%s, kernel: %s, hostname: %s\n",
4289 4290 4291 4292
                                  timestamp, msg, VIR_LOG_VERSION_STRING,
                                  (qemuVersion / 1000000) % 1000,
                                  (qemuVersion / 1000) % 1000,
                                  qemuVersion % 1000,
4293
                                  package ? package : "",
4294
                                  uts.release,
4295
                                  hostname ? hostname : "") < 0)
4296
        goto cleanup;
4297

4298 4299 4300 4301 4302
    if (cmd) {
        char *args = virCommandToString(cmd);
        qemuDomainLogContextWrite(logCtxt, "%s\n", args);
        VIR_FREE(args);
    }
4303 4304

 cleanup:
4305
    VIR_FREE(hostname);
4306 4307 4308
    VIR_FREE(timestamp);
}

4309 4310 4311 4312 4313 4314 4315

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

4316
    VIR_FREE(inc->address);
4317
    VIR_FREE(inc->launchURI);
4318
    VIR_FREE(inc->deferredURI);
4319 4320 4321 4322 4323 4324 4325 4326
    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.
4327 4328 4329
 *
 * The caller is responsible for closing @fd, calling
 * qemuProcessIncomingDefFree will NOT close it.
4330 4331 4332
 */
qemuProcessIncomingDefPtr
qemuProcessIncomingDefNew(virQEMUCapsPtr qemuCaps,
4333
                          const char *listenAddress,
4334 4335 4336 4337 4338 4339
                          const char *migrateFrom,
                          int fd,
                          const char *path)
{
    qemuProcessIncomingDefPtr inc = NULL;

4340
    if (qemuMigrationDstCheckProtocol(qemuCaps, migrateFrom) < 0)
4341 4342 4343 4344 4345
        return NULL;

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

4346 4347 4348
    if (VIR_STRDUP(inc->address, listenAddress) < 0)
        goto error;

4349
    inc->launchURI = qemuMigrationDstGetURI(migrateFrom, fd);
4350 4351 4352
    if (!inc->launchURI)
        goto error;

4353 4354 4355 4356 4357 4358
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_INCOMING_DEFER)) {
        inc->deferredURI = inc->launchURI;
        if (VIR_STRDUP(inc->launchURI, "defer") < 0)
            goto error;
    }

4359 4360 4361 4362 4363 4364 4365 4366 4367 4368 4369
    inc->fd = fd;
    inc->path = path;

    return inc;

 error:
    qemuProcessIncomingDefFree(inc);
    return NULL;
}


4370 4371 4372 4373 4374 4375 4376 4377
/*
 * 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,
4378
                    virDomainObjPtr vm,
4379 4380
                    virDomainJobOperation operation,
                    unsigned long apiFlags)
4381
{
4382
    if (qemuDomainObjBeginAsyncJob(driver, vm, QEMU_ASYNC_JOB_START,
4383
                                   operation, apiFlags) < 0)
4384 4385 4386 4387 4388 4389 4390 4391 4392 4393 4394 4395 4396 4397 4398
        return -1;

    qemuDomainObjSetAsyncJobMask(vm, QEMU_JOB_NONE);
    return 0;
}


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


4399 4400 4401 4402 4403 4404 4405 4406 4407 4408 4409 4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420 4421
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;
}


4422
static int
4423
qemuProcessGraphicsReservePorts(virDomainGraphicsDefPtr graphics,
4424
                                bool reconnect)
4425
{
4426 4427 4428 4429 4430 4431 4432 4433 4434 4435 4436
    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;

4437 4438
    switch (graphics->type) {
    case VIR_DOMAIN_GRAPHICS_TYPE_VNC:
4439 4440
        if (!graphics->data.vnc.autoport ||
            reconnect) {
4441
            if (virPortAllocatorSetUsed(graphics->data.vnc.port) < 0)
4442 4443 4444
                return -1;
            graphics->data.vnc.portReserved = true;
        }
4445
        if (graphics->data.vnc.websocket > 0 &&
4446
            virPortAllocatorSetUsed(graphics->data.vnc.websocket) < 0)
4447
            return -1;
4448 4449 4450
        break;

    case VIR_DOMAIN_GRAPHICS_TYPE_SPICE:
4451
        if (graphics->data.spice.autoport && !reconnect)
4452
            return 0;
4453

4454
        if (graphics->data.spice.port > 0) {
4455
            if (virPortAllocatorSetUsed(graphics->data.spice.port) < 0)
4456
                return -1;
4457 4458
            graphics->data.spice.portReserved = true;
        }
4459

4460
        if (graphics->data.spice.tlsPort > 0) {
4461
            if (virPortAllocatorSetUsed(graphics->data.spice.tlsPort) < 0)
4462 4463
                return -1;
            graphics->data.spice.tlsPortReserved = true;
4464
        }
4465 4466 4467 4468 4469
        break;

    case VIR_DOMAIN_GRAPHICS_TYPE_SDL:
    case VIR_DOMAIN_GRAPHICS_TYPE_RDP:
    case VIR_DOMAIN_GRAPHICS_TYPE_DESKTOP:
4470
    case VIR_DOMAIN_GRAPHICS_TYPE_EGL_HEADLESS:
4471 4472
    case VIR_DOMAIN_GRAPHICS_TYPE_LAST:
        break;
4473 4474
    }

4475 4476 4477 4478
    return 0;
}


4479 4480 4481 4482 4483 4484 4485 4486 4487 4488 4489 4490 4491 4492 4493 4494 4495 4496 4497 4498 4499 4500 4501 4502 4503 4504 4505 4506 4507 4508
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:
4509
    case VIR_DOMAIN_GRAPHICS_TYPE_EGL_HEADLESS:
4510 4511 4512 4513 4514 4515 4516
    case VIR_DOMAIN_GRAPHICS_TYPE_LAST:
        break;
    }

    return 0;
}

4517
static int
4518
qemuProcessGetNetworkAddress(const char *netname,
4519 4520
                             char **netaddr)
{
4521
    virConnectPtr conn = NULL;
4522 4523 4524 4525 4526 4527 4528 4529 4530 4531
    int ret = -1;
    virNetworkPtr net;
    virNetworkDefPtr netdef = NULL;
    virNetworkIPDefPtr ipdef;
    virSocketAddr addr;
    virSocketAddrPtr addrptr = NULL;
    char *dev_name = NULL;
    char *xml = NULL;

    *netaddr = NULL;
4532 4533 4534 4535

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

4536 4537 4538 4539 4540 4541 4542 4543 4544 4545 4546 4547
    net = virNetworkLookupByName(conn, netname);
    if (!net)
        goto cleanup;

    xml = virNetworkGetXMLDesc(net, 0);
    if (!xml)
        goto cleanup;

    netdef = virNetworkDefParseString(xml);
    if (!netdef)
        goto cleanup;

4548
    switch ((virNetworkForwardType) netdef->forward.type) {
4549 4550 4551 4552 4553 4554 4555 4556 4557 4558 4559 4560 4561 4562 4563 4564 4565 4566 4567 4568 4569 4570 4571 4572 4573 4574 4575 4576 4577 4578 4579 4580 4581 4582 4583 4584
    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
    case VIR_NETWORK_FORWARD_OPEN:
        ipdef = virNetworkDefGetIPByIndex(netdef, AF_UNSPEC, 0);
        if (!ipdef) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' doesn't have an IP address"),
                           netdef->name);
            goto cleanup;
        }
        addrptr = &ipdef->address;
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
        if ((dev_name = netdef->bridge))
            break;
        /*
         * fall through if netdef->bridge wasn't set, since that is
         * macvtap bridge mode network.
         */
        ATTRIBUTE_FALLTHROUGH;

    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
        if ((netdef->forward.nifs > 0) && netdef->forward.ifs)
            dev_name = netdef->forward.ifs[0].device.dev;

        if (!dev_name) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' has no associated interface or bridge"),
                           netdef->name);
            goto cleanup;
        }
        break;
4585 4586 4587 4588 4589 4590 4591 4592

    case VIR_NETWORK_FORWARD_HOSTDEV:
        break;

    case VIR_NETWORK_FORWARD_LAST:
    default:
        virReportEnumRangeError(virNetworkForwardType, netdef->forward.type);
        goto cleanup;
4593 4594 4595 4596 4597 4598 4599 4600 4601 4602 4603 4604 4605 4606 4607 4608 4609
    }

    if (dev_name) {
        if (virNetDevIPAddrGet(dev_name, &addr) < 0)
            goto cleanup;
        addrptr = &addr;
    }

    if (!(addrptr &&
          (*netaddr = virSocketAddrFormat(addrptr)))) {
        goto cleanup;
    }

    ret = 0;
 cleanup:
    virNetworkDefFree(netdef);
    virObjectUnref(net);
4610
    virObjectUnref(conn);
4611 4612 4613 4614
    VIR_FREE(xml);
    return ret;
}

4615

4616
static int
4617
qemuProcessGraphicsSetupNetworkAddress(virDomainGraphicsListenDefPtr glisten,
4618 4619 4620 4621 4622 4623 4624 4625 4626 4627 4628
                                       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;
    }

4629
    rc = qemuProcessGetNetworkAddress(glisten->network, &glisten->address);
4630 4631 4632 4633 4634 4635 4636 4637 4638 4639 4640 4641 4642
    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;
}


4643
static int
4644
qemuProcessGraphicsSetupListen(virQEMUDriverPtr driver,
4645 4646
                               virDomainGraphicsDefPtr graphics,
                               virDomainObjPtr vm)
4647
{
4648
    qemuDomainObjPrivatePtr priv = vm->privateData;
4649
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
4650
    const char *type = virDomainGraphicsTypeToString(graphics->type);
4651
    char *listenAddr = NULL;
4652
    bool useSocket = false;
4653
    size_t i;
4654
    int ret = -1;
4655 4656 4657

    switch (graphics->type) {
    case VIR_DOMAIN_GRAPHICS_TYPE_VNC:
4658
        useSocket = cfg->vncAutoUnixSocket;
4659 4660 4661 4662
        listenAddr = cfg->vncListen;
        break;

    case VIR_DOMAIN_GRAPHICS_TYPE_SPICE:
4663
        useSocket = cfg->spiceAutoUnixSocket;
4664 4665 4666 4667 4668 4669
        listenAddr = cfg->spiceListen;
        break;

    case VIR_DOMAIN_GRAPHICS_TYPE_SDL:
    case VIR_DOMAIN_GRAPHICS_TYPE_RDP:
    case VIR_DOMAIN_GRAPHICS_TYPE_DESKTOP:
4670
    case VIR_DOMAIN_GRAPHICS_TYPE_EGL_HEADLESS:
4671 4672 4673 4674 4675 4676 4677 4678 4679
    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:
4680 4681 4682 4683 4684
            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) {
4685 4686
                    memset(glisten, 0, sizeof(virDomainGraphicsListenDef));
                    if (virAsprintf(&glisten->socket, "%s/%s.sock",
4687
                                    priv->libDir, type) < 0)
4688
                        goto cleanup;
4689 4690
                    glisten->fromConfig = true;
                    glisten->type = VIR_DOMAIN_GRAPHICS_LISTEN_TYPE_SOCKET;
4691 4692
                } else if (listenAddr) {
                    if (VIR_STRDUP(glisten->address, listenAddr) < 0)
4693
                        goto cleanup;
4694 4695 4696
                    glisten->fromConfig = true;
                }
            }
4697 4698 4699 4700 4701 4702
            break;

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

4703
            if (qemuProcessGraphicsSetupNetworkAddress(glisten,
4704
                                                       listenAddr) < 0)
4705
                goto cleanup;
4706 4707
            break;

4708 4709 4710 4711
        case VIR_DOMAIN_GRAPHICS_LISTEN_TYPE_SOCKET:
            if (!glisten->socket) {
                if (virAsprintf(&glisten->socket, "%s/%s.sock",
                                priv->libDir, type) < 0)
4712
                    goto cleanup;
4713 4714 4715 4716
                glisten->autoGenerated = true;
            }
            break;

4717 4718 4719 4720 4721 4722
        case VIR_DOMAIN_GRAPHICS_LISTEN_TYPE_NONE:
        case VIR_DOMAIN_GRAPHICS_LISTEN_TYPE_LAST:
            break;
        }
    }

4723 4724 4725 4726 4727
    ret = 0;

 cleanup:
    virObjectUnref(cfg);
    return ret;
4728 4729 4730
}


4731
static int
4732
qemuProcessSetupGraphics(virQEMUDriverPtr driver,
4733 4734
                         virDomainObjPtr vm,
                         unsigned int flags)
4735
{
4736
    virDomainGraphicsDefPtr graphics;
4737
    bool allocate = !(flags & VIR_QEMU_PROCESS_START_PRETEND);
4738
    size_t i;
4739 4740
    int ret = -1;

4741 4742 4743
    for (i = 0; i < vm->def->ngraphics; i++) {
        graphics = vm->def->graphics[i];

4744
        if (qemuProcessGraphicsSetupListen(driver, graphics, vm) < 0)
4745 4746 4747
            goto cleanup;
    }

4748 4749 4750 4751
    if (allocate) {
        for (i = 0; i < vm->def->ngraphics; i++) {
            graphics = vm->def->graphics[i];

4752
            if (qemuProcessGraphicsReservePorts(graphics, false) < 0)
4753 4754 4755
                goto cleanup;
        }
    }
4756

4757
    for (i = 0; i < vm->def->ngraphics; ++i) {
4758
        graphics = vm->def->graphics[i];
4759

4760 4761
        if (qemuProcessGraphicsAllocatePorts(driver, graphics, allocate) < 0)
            goto cleanup;
4762 4763 4764 4765 4766 4767 4768 4769 4770
    }

    ret = 0;

 cleanup:
    return ret;
}


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
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++) {
4804
            if (!virHostdevIsSCSIDevice(vm->def->hostdevs[i]))
4805 4806
                continue;

4807 4808 4809 4810 4811 4812 4813 4814 4815 4816 4817 4818 4819 4820 4821 4822 4823 4824 4825 4826 4827 4828 4829 4830 4831 4832
            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;
}


4833 4834 4835 4836 4837 4838 4839 4840 4841
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;

4842
    if (!virDomainDefHasMemballoon(vm->def))
4843 4844 4845
        return 0;

    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
4846
        return -1;
4847

4848 4849 4850
    if (vm->def->memballoon->period)
        qemuMonitorSetMemoryStatsPeriod(priv->mon, vm->def->memballoon,
                                        vm->def->memballoon->period);
4851 4852 4853 4854 4855 4856 4857 4858 4859 4860 4861 4862
    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
4863 4864 4865
static int
qemuProcessMakeDir(virQEMUDriverPtr driver,
                   virDomainObjPtr vm,
4866
                   const char *path)
J
Jiri Denemark 已提交
4867 4868 4869 4870 4871 4872 4873 4874
{
    int ret = -1;

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

4875
    if (qemuSecurityDomainSetPathLabel(driver->securityManager,
4876
                                       vm->def, path, true) < 0)
J
Jiri Denemark 已提交
4877 4878 4879 4880 4881 4882 4883 4884 4885
        goto cleanup;

    ret = 0;

 cleanup:
    return ret;
}


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
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) ==
4922
                VIR_DOMAIN_MEMORY_ACCESS_SHARED) {
4923 4924 4925 4926 4927 4928 4929 4930 4931 4932 4933 4934
                shmem = true;
                break;
            }
        }
    }

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

4935 4936 4937 4938 4939 4940 4941 4942 4943 4944 4945 4946 4947 4948 4949 4950 4951 4952 4953 4954 4955 4956 4957

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:
4958
        case VIR_DOMAIN_GRAPHICS_TYPE_EGL_HEADLESS:
4959 4960 4961 4962 4963 4964 4965 4966 4967
        case VIR_DOMAIN_GRAPHICS_TYPE_LAST:
            break;
        }
    }

    return 0;
}


4968 4969 4970 4971 4972 4973 4974 4975 4976 4977
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];

4978 4979 4980 4981 4982 4983 4984 4985 4986
        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 &&
4987 4988 4989 4990
             !virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VIRTIO_GPU)) ||
            (video->type == VIR_DOMAIN_VIDEO_TYPE_VIRTIO &&
             video->info.type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_CCW &&
             !virQEMUCapsGet(qemuCaps, QEMU_CAPS_DEVICE_VIRTIO_GPU_CCW))) {
4991 4992 4993 4994 4995 4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009 5010 5011 5012
            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;
}


5013 5014 5015 5016 5017 5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 5037 5038 5039 5040 5041 5042 5043
static int
qemuProcessStartValidateIOThreads(virDomainObjPtr vm,
                                  virQEMUCapsPtr qemuCaps)
{
    size_t i;

    if (vm->def->niothreadids > 0 &&
        !virQEMUCapsGet(qemuCaps, QEMU_CAPS_OBJECT_IOTHREAD)) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("IOThreads not supported for this QEMU"));
        return -1;
    }

    for (i = 0; i < vm->def->ncontrollers; i++) {
        virDomainControllerDefPtr cont = vm->def->controllers[i];

        if (cont->type == VIR_DOMAIN_CONTROLLER_TYPE_SCSI &&
            cont->model == VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_SCSI &&
            cont->iothread > 0 &&
            !virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_SCSI_IOTHREAD)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("IOThreads for virtio-scsi not supported for "
                             "this QEMU"));
            return -1;
        }
    }

    return 0;
}


5044
static int
5045 5046 5047 5048 5049 5050 5051 5052 5053 5054 5055 5056 5057 5058 5059 5060 5061 5062 5063
qemuProcessStartValidateShmem(virDomainObjPtr vm)
{
    size_t i;

    for (i = 0; i < vm->def->nshmems; i++) {
        virDomainShmemDefPtr shmem = vm->def->shmems[i];

        if (strchr(shmem->name, '/')) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("shmem name '%s' must not contain '/'"),
                           shmem->name);
            return -1;
        }
    }

    return 0;
}


5064 5065 5066 5067 5068 5069 5070
static int
qemuProcessStartValidateDisks(virDomainObjPtr vm,
                              virQEMUCapsPtr qemuCaps)
{
    size_t i;

    for (i = 0; i < vm->def->ndisks; i++) {
5071 5072
        virDomainDiskDefPtr disk = vm->def->disks[i];
        virStorageSourcePtr src = disk->src;
5073 5074 5075 5076 5077 5078 5079 5080 5081 5082 5083 5084

        /* This is a best effort check as we can only check if the command
         * option exists, but we cannot determine whether the running QEMU
         * was build with '--enable-vxhs'. */
        if (src->type == VIR_STORAGE_TYPE_NETWORK &&
            src->protocol == VIR_STORAGE_NET_PROTOCOL_VXHS &&
            !virQEMUCapsGet(qemuCaps, QEMU_CAPS_VXHS)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("VxHS protocol is not supported with this "
                             "QEMU binary"));
            return -1;
        }
5085 5086 5087 5088 5089 5090 5091 5092

        /* PowerPC pseries based VMs do not support floppy device */
        if (disk->device == VIR_DOMAIN_DISK_DEVICE_FLOPPY &&
            qemuDomainIsPSeries(vm->def)) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("PowerPC pseries machines do not support floppy device"));
            return -1;
        }
5093 5094 5095 5096 5097 5098
    }

    return 0;
}


5099
static int
5100 5101
qemuProcessStartValidateXML(virQEMUDriverPtr driver,
                            virDomainObjPtr vm,
5102
                            virQEMUCapsPtr qemuCaps,
5103
                            virCapsPtr caps,
5104
                            unsigned int flags)
5105 5106 5107 5108 5109 5110 5111 5112 5113 5114 5115 5116
{
    /* 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;

5117 5118 5119
    /* 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 */
5120 5121 5122
    if ((flags & VIR_QEMU_PROCESS_START_NEW) &&
        virDomainDefValidate(vm->def, caps, 0, driver->xmlopt) < 0)
        return -1;
5123 5124 5125

    return 0;
}
5126

5127

5128 5129 5130 5131 5132 5133
/**
 * qemuProcessStartValidate:
 * @vm: domain object
 * @qemuCaps: emulator capabilities
 * @migration: restoration of existing state
 *
5134 5135 5136 5137 5138
 * 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.
5139
 */
5140
static int
5141 5142
qemuProcessStartValidate(virQEMUDriverPtr driver,
                         virDomainObjPtr vm,
5143
                         virQEMUCapsPtr qemuCaps,
5144
                         virCapsPtr caps,
5145
                         unsigned int flags)
5146
{
5147 5148 5149 5150 5151 5152 5153 5154 5155 5156 5157 5158 5159 5160
    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");
5161
        if (qemuSecurityCheckAllLabel(driver->securityManager, vm->def) < 0)
5162 5163 5164 5165
            return -1;

    }

5166
    if (qemuProcessStartValidateXML(driver, vm, qemuCaps, caps, flags) < 0)
5167 5168
        return -1;

5169 5170 5171
    if (qemuProcessStartValidateGraphics(vm) < 0)
        return -1;

5172 5173 5174
    if (qemuProcessStartValidateVideo(vm, qemuCaps) < 0)
        return -1;

5175 5176 5177
    if (qemuProcessStartValidateIOThreads(vm, qemuCaps) < 0)
        return -1;

5178 5179 5180
    if (qemuProcessStartValidateShmem(vm) < 0)
        return -1;

5181 5182 5183 5184
    if (vm->def->cpu &&
        virCPUValidateFeatures(vm->def->os.arch, vm->def->cpu) < 0)
        return -1;

5185 5186 5187
    if (qemuProcessStartValidateDisks(vm, qemuCaps) < 0)
        return -1;

5188 5189
    VIR_DEBUG("Checking for any possible (non-fatal) issues");

5190
    qemuProcessStartWarnShmem(vm);
5191

5192
    return 0;
5193 5194 5195
}


J
Jiri Denemark 已提交
5196 5197 5198 5199 5200 5201
/**
 * qemuProcessInit:
 *
 * Prepares the domain up to the point when priv->qemuCaps is initialized. The
 * function calls qemuProcessStop when needed.
 *
5202 5203 5204 5205
 * 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
5206 5207 5208 5209 5210
 * Returns 0 on success, -1 on error.
 */
int
qemuProcessInit(virQEMUDriverPtr driver,
                virDomainObjPtr vm,
5211
                virCPUDefPtr updatedCPU,
5212
                qemuDomainAsyncJob asyncJob,
5213
                bool migration,
5214
                unsigned int flags)
J
Jiri Denemark 已提交
5215 5216 5217 5218 5219
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    virCapsPtr caps = NULL;
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int stopFlags;
5220
    virCPUDefPtr origCPU = NULL;
J
Jiri Denemark 已提交
5221 5222 5223 5224 5225 5226 5227 5228 5229 5230 5231 5232 5233 5234
    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)))
5235 5236
        goto cleanup;

5237 5238 5239 5240 5241 5242 5243 5244 5245
    /* in case when the post parse callback failed we need to re-run it on the
     * old config prior we start the VM */
    if (vm->def->postParseFailed) {
        VIR_DEBUG("re-running the post parse callback");

        if (virDomainDefPostParse(vm->def, caps, 0, driver->xmlopt, NULL) < 0)
            goto cleanup;
    }

5246 5247
    VIR_DEBUG("Determining emulator version");
    virObjectUnref(priv->qemuCaps);
5248
    if (!(priv->qemuCaps = virQEMUCapsCacheLookupCopy(driver->qemuCapsCache,
5249 5250 5251
                                                      vm->def->emulator,
                                                      vm->def->os.machine)))
        goto cleanup;
J
Jiri Denemark 已提交
5252

5253 5254 5255
    if (flags & VIR_QEMU_PROCESS_START_STANDALONE)
        virQEMUCapsClear(priv->qemuCaps, QEMU_CAPS_CHARDEV_FD_PASS);

5256 5257 5258
    if (qemuDomainUpdateCPU(vm, updatedCPU, &origCPU) < 0)
        goto cleanup;

5259
    if (qemuProcessStartValidate(driver, vm, priv->qemuCaps, caps, flags) < 0)
5260 5261
        goto cleanup;

J
Jiri Denemark 已提交
5262 5263 5264 5265 5266
    /* 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");
5267
    if (virDomainObjSetDefTransient(caps, driver->xmlopt, vm) < 0)
5268
        goto cleanup;
J
Jiri Denemark 已提交
5269

5270
    if (flags & VIR_QEMU_PROCESS_START_PRETEND) {
5271 5272 5273 5274
        if (qemuDomainSetPrivatePaths(driver, vm) < 0) {
            virDomainObjRemoveTransientDef(vm);
            goto cleanup;
        }
5275
    } else {
5276 5277 5278
        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
5279

5280 5281
        if (virAtomicIntInc(&driver->nactive) == 1 && driver->inhibitCallback)
            driver->inhibitCallback(true, driver->inhibitOpaque);
J
Jiri Denemark 已提交
5282

5283 5284 5285 5286 5287
        /* 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
5288

5289 5290
        if (qemuDomainSetPrivatePaths(driver, vm) < 0)
            goto stop;
5291 5292

        VIR_STEAL_PTR(priv->origCPU, origCPU);
5293
    }
5294

J
Jiri Denemark 已提交
5295 5296 5297
    ret = 0;

 cleanup:
5298
    virCPUDefFree(origCPU);
J
Jiri Denemark 已提交
5299 5300 5301 5302 5303 5304 5305 5306
    virObjectUnref(cfg);
    virObjectUnref(caps);
    return ret;

 stop:
    stopFlags = VIR_QEMU_PROCESS_STOP_NO_RELABEL;
    if (migration)
        stopFlags |= VIR_QEMU_PROCESS_STOP_MIGRATED;
5307
    qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_FAILED, asyncJob, stopFlags);
J
Jiri Denemark 已提交
5308 5309 5310 5311
    goto cleanup;
}


5312 5313 5314 5315 5316 5317 5318 5319 5320 5321 5322
/**
 * qemuProcessNetworkPrepareDevices
 */
static int
qemuProcessNetworkPrepareDevices(virDomainDefPtr def)
{
    int ret = -1;
    size_t i;

    for (i = 0; i < def->nnets; i++) {
        virDomainNetDefPtr net = def->nets[i];
5323
        virDomainNetType actualType;
5324 5325 5326 5327 5328

        /* 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.
         */
5329
        if (virDomainNetAllocateActualDevice(def, net) < 0)
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 5363 5364
            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;
}


5365 5366 5367 5368 5369 5370 5371 5372 5373 5374 5375 5376 5377 5378 5379 5380
/**
 * 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);
5381
    virDomainVcpuDefPtr vcpu = virDomainDefGetVcpu(vm->def, vcpuid);
5382 5383 5384 5385 5386 5387 5388 5389 5390 5391 5392 5393 5394 5395 5396 5397 5398 5399
    size_t i = 0;

    if (qemuProcessSetupPid(vm, vcpupid, VIR_CGROUP_THREAD_VCPU,
                            vcpuid, vcpu->cpumask,
                            vm->def->cputune.period,
                            vm->def->cputune.quota,
                            &vcpu->sched) < 0)
        return -1;

    for (i = 0; i < vm->def->ncachetunes; i++) {
        virDomainCachetuneDefPtr ct = vm->def->cachetunes[i];

        if (virBitmapIsBitSet(ct->vcpus, vcpuid)) {
            if (virResctrlAllocAddPID(ct->alloc, vcpupid) < 0)
                return -1;
            break;
        }
    }
5400

5401
    return 0;
5402 5403 5404 5405 5406 5407
}


static int
qemuProcessSetupVcpus(virDomainObjPtr vm)
{
5408
    virDomainVcpuDefPtr vcpu;
5409 5410 5411 5412 5413 5414 5415 5416 5417 5418 5419 5420 5421 5422 5423 5424 5425 5426 5427 5428 5429 5430 5431 5432 5433 5434 5435 5436 5437 5438 5439 5440 5441 5442 5443 5444 5445 5446 5447 5448 5449 5450 5451 5452 5453
    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;
}


5454 5455 5456 5457 5458
int
qemuProcessSetupIOThread(virDomainObjPtr vm,
                         virDomainIOThreadIDDefPtr iothread)
{

5459 5460
    return qemuProcessSetupPid(vm, iothread->thread_id,
                               VIR_CGROUP_THREAD_IOTHREAD,
5461
                               iothread->iothread_id,
5462
                               iothread->cpumask,
5463 5464
                               vm->def->cputune.iothread_period,
                               vm->def->cputune.iothread_quota,
5465
                               &iothread->sched);
5466 5467 5468 5469 5470 5471 5472 5473 5474 5475 5476 5477 5478 5479 5480 5481 5482 5483 5484
}


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;
}


5485 5486 5487 5488 5489 5490 5491 5492 5493 5494 5495 5496
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;

5497
    if (!(ordermap = virBitmapNew(maxvcpus + 1)))
5498 5499 5500 5501 5502 5503 5504 5505 5506 5507 5508 5509 5510 5511 5512 5513
        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) {
5514
            if (virBitmapIsBitSet(ordermap, vcpu->order)) {
5515
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
5516
                               _("duplicate vcpu order '%u'"), vcpu->order);
5517 5518 5519
                goto cleanup;
            }

5520 5521 5522 5523 5524 5525
            if (virBitmapSetBit(ordermap, vcpu->order)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("vcpu order '%u' exceeds vcpu count"),
                               vcpu->order);
                goto cleanup;
            }
5526 5527 5528 5529 5530 5531 5532 5533 5534 5535 5536 5537 5538 5539 5540 5541
        }

        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 &&
5542
                 vcpupriv->thread_id == -1 && vcpupriv->node_id == -1) ||
5543 5544 5545 5546 5547 5548 5549 5550 5551 5552 5553 5554 5555 5556 5557 5558 5559 5560 5561 5562 5563 5564 5565 5566 5567 5568 5569 5570 5571 5572 5573 5574 5575 5576 5577 5578 5579 5580 5581 5582 5583 5584 5585 5586 5587 5588 5589 5590 5591 5592
                !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);
5593 5594
    qemuDomainObjPrivatePtr priv = vm->privateData;
    qemuCgroupEmulatorAllNodesDataPtr emulatorCgroup = NULL;
5595 5596 5597 5598 5599 5600 5601 5602 5603 5604 5605 5606 5607 5608 5609 5610 5611 5612 5613 5614 5615 5616 5617 5618 5619 5620 5621 5622 5623 5624 5625 5626
    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);

5627 5628 5629
    if (qemuCgroupEmulatorAllNodesAllow(priv->cgroup, &emulatorCgroup) < 0)
        goto cleanup;

5630 5631 5632 5633 5634 5635 5636 5637 5638 5639 5640 5641 5642 5643 5644 5645 5646 5647 5648 5649 5650 5651 5652 5653
    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:
5654
    qemuCgroupEmulatorAllNodesRestore(emulatorCgroup);
5655 5656 5657 5658 5659 5660
    VIR_FREE(bootHotplug);
    virJSONValueFree(vcpuprops);
    return ret;
}


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 5690 5691 5692 5693 5694 5695 5696 5697 5698 5699
static int
qemuProcessUpdateGuestCPU(virDomainDefPtr def,
                          virQEMUCapsPtr qemuCaps,
                          virCapsPtr caps,
                          unsigned int flags)
{
    int ret = -1;

    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;
    }

5700 5701 5702
    if (virCPUConvertLegacy(caps->host.arch, def->cpu) < 0)
        return -1;

5703 5704 5705 5706
    /* nothing to update for host-passthrough */
    if (def->cpu->mode == VIR_CPU_MODE_HOST_PASSTHROUGH)
        return 0;

5707 5708
    if (def->cpu->check == VIR_CPU_CHECK_PARTIAL &&
        virCPUCompare(caps->host.arch,
5709
                      virQEMUCapsGetHostModel(qemuCaps, def->virtType,
5710
                                              VIR_QEMU_CAPS_HOST_CPU_FULL),
5711 5712
                      def->cpu, true) < 0)
        return -1;
5713 5714

    if (virCPUUpdate(def->os.arch, def->cpu,
5715
                     virQEMUCapsGetHostModel(qemuCaps, def->virtType,
5716
                                             VIR_QEMU_CAPS_HOST_CPU_MIGRATABLE)) < 0)
5717 5718
        goto cleanup;

5719 5720
    if (virCPUTranslate(def->os.arch, def->cpu,
                        virQEMUCapsGetCPUDefinitions(qemuCaps, def->virtType)) < 0)
5721 5722 5723 5724 5725 5726 5727 5728 5729 5730
        goto cleanup;

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

 cleanup:
    return ret;
}


5731 5732 5733 5734 5735 5736
static int
qemuProcessPrepareDomainNUMAPlacement(virDomainObjPtr vm,
                                      virCapsPtr caps)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    char *nodeset = NULL;
5737 5738
    virBitmapPtr numadNodeset = NULL;
    virBitmapPtr hostMemoryNodeset = NULL;
5739 5740 5741 5742 5743 5744 5745 5746 5747 5748 5749 5750 5751 5752
    int ret = -1;

    /* Get the advisory nodeset from numad if 'placement' of
     * either <vcpu> or <numatune> is 'auto'.
     */
    if (!virDomainDefNeedsPlacementAdvice(vm->def))
        return 0;

    nodeset = virNumaGetAutoPlacementAdvice(virDomainDefGetVcpus(vm->def),
                                            virDomainDefGetMemoryTotal(vm->def));

    if (!nodeset)
        goto cleanup;

5753 5754 5755
    if (!(hostMemoryNodeset = virNumaGetHostMemoryNodeset()))
        goto cleanup;

5756 5757
    VIR_DEBUG("Nodeset returned from numad: %s", nodeset);

5758
    if (virBitmapParse(nodeset, &numadNodeset, VIR_DOMAIN_CPUMASK_LEN) < 0)
5759 5760
        goto cleanup;

5761 5762 5763 5764
    /* numad may return a nodeset that only contains cpus but cgroups don't play
     * well with that. Set the autoCpuset from all cpus from that nodeset, but
     * assign autoNodeset only with nodes containing memory. */
    if (!(priv->autoCpuset = virCapabilitiesGetCpusForNodemask(caps, numadNodeset)))
5765 5766
        goto cleanup;

5767 5768 5769 5770
    virBitmapIntersect(numadNodeset, hostMemoryNodeset);

    VIR_STEAL_PTR(priv->autoNodeset, numadNodeset);

5771 5772 5773 5774
    ret = 0;

 cleanup:
    VIR_FREE(nodeset);
5775 5776
    virBitmapFree(numadNodeset);
    virBitmapFree(hostMemoryNodeset);
5777 5778 5779 5780
    return ret;
}


5781
static int
5782
qemuProcessPrepareDomainStorage(virQEMUDriverPtr driver,
5783
                                virDomainObjPtr vm,
5784
                                qemuDomainObjPrivatePtr priv,
5785
                                virQEMUDriverConfigPtr cfg,
5786 5787 5788 5789 5790 5791 5792 5793 5794
                                unsigned int flags)
{
    size_t i;
    bool cold_boot = flags & VIR_QEMU_PROCESS_START_COLD;

    for (i = vm->def->ndisks; i > 0; i--) {
        size_t idx = i - 1;
        virDomainDiskDefPtr disk = vm->def->disks[idx];

5795
        if (virDomainDiskTranslateSourcePool(disk) < 0) {
5796 5797 5798 5799 5800 5801
            if (qemuDomainCheckDiskStartupPolicy(driver, vm, idx, cold_boot) < 0)
                return -1;

            /* disk source was dropped */
            continue;
        }
5802

5803
        if (qemuDomainPrepareDiskSource(disk, priv, cfg) < 0)
5804
            return -1;
5805 5806 5807 5808 5809 5810
    }

    return 0;
}


5811 5812 5813 5814 5815 5816 5817 5818 5819 5820 5821 5822 5823 5824 5825 5826 5827 5828 5829 5830
static void
qemuProcessPrepareAllowReboot(virDomainObjPtr vm)
{
    virDomainDefPtr def = vm->def;
    qemuDomainObjPrivatePtr priv = vm->privateData;

    if (priv->allowReboot != VIR_TRISTATE_BOOL_ABSENT)
        return;

    if (def->onReboot == VIR_DOMAIN_LIFECYCLE_ACTION_DESTROY &&
        def->onPoweroff == VIR_DOMAIN_LIFECYCLE_ACTION_DESTROY &&
        (def->onCrash == VIR_DOMAIN_LIFECYCLE_ACTION_DESTROY ||
         def->onCrash == VIR_DOMAIN_LIFECYCLE_ACTION_COREDUMP_DESTROY)) {
        priv->allowReboot = VIR_TRISTATE_BOOL_NO;
    } else {
        priv->allowReboot = VIR_TRISTATE_BOOL_YES;
    }
}


5831
/**
5832 5833 5834 5835
 * qemuProcessPrepareDomain:
 * @driver: qemu driver
 * @vm: domain object
 * @flags: qemuProcessStartFlags
5836 5837 5838 5839 5840 5841 5842 5843 5844 5845 5846
 *
 * 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
5847
qemuProcessPrepareDomain(virQEMUDriverPtr driver,
5848 5849 5850 5851 5852 5853
                         virDomainObjPtr vm,
                         unsigned int flags)
{
    int ret = -1;
    size_t i;
    qemuDomainObjPrivatePtr priv = vm->privateData;
5854
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
5855 5856 5857 5858 5859
    virCapsPtr caps;

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

5860 5861 5862 5863
    priv->machineName = qemuDomainGetMachineName(vm);
    if (!priv->machineName)
        goto cleanup;

5864 5865 5866 5867
    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)");
5868
        if (qemuSecurityGenLabel(driver->securityManager, vm->def) < 0) {
5869 5870 5871 5872 5873
            virDomainAuditSecurityLabel(vm, false);
            goto cleanup;
        }
        virDomainAuditSecurityLabel(vm, true);

5874 5875
        if (qemuProcessPrepareDomainNUMAPlacement(vm, caps) < 0)
            goto cleanup;
5876 5877
    }

5878 5879 5880 5881 5882 5883 5884
    /* Whether we should use virtlogd as stdio handler for character
     * devices source backend. */
    if (cfg->stdioLogD &&
        virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_CHARDEV_FILE_APPEND)) {
        priv->chardevStdioLogd = true;
    }

5885 5886
    qemuProcessPrepareAllowReboot(vm);

5887 5888 5889 5890 5891 5892 5893
    /*
     * 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
     */
5894
    VIR_DEBUG("Assigning domain PCI addresses");
5895 5896
    if ((qemuDomainAssignAddresses(vm->def, priv->qemuCaps, driver, vm,
                                   !!(flags & VIR_QEMU_PROCESS_START_NEW))) < 0) {
5897
        goto cleanup;
5898
    }
5899

5900 5901 5902
    if (qemuAssignDeviceAliases(vm->def, priv->qemuCaps) < 0)
        goto cleanup;

5903
    VIR_DEBUG("Setting graphics devices");
5904
    if (qemuProcessSetupGraphics(driver, vm, flags) < 0)
5905 5906
        goto cleanup;

5907 5908 5909 5910
    VIR_DEBUG("Create domain masterKey");
    if (qemuDomainMasterKeyCreate(vm) < 0)
        goto cleanup;

5911
    VIR_DEBUG("Setting up storage");
5912
    if (qemuProcessPrepareDomainStorage(driver, vm, priv, cfg, flags) < 0)
5913 5914
        goto cleanup;

5915
    VIR_DEBUG("Prepare chardev source backends for TLS");
5916
    qemuDomainPrepareChardevSource(vm->def, cfg);
5917

5918
    VIR_DEBUG("Add secrets to hostdevs and chardevs");
5919
    if (qemuDomainSecretPrepare(driver, vm) < 0)
5920 5921
        goto cleanup;

5922 5923 5924 5925 5926 5927
    for (i = 0; i < vm->def->nchannels; i++) {
        if (qemuDomainPrepareChannel(vm->def->channels[i],
                                     priv->channelTargetDir) < 0)
            goto cleanup;
    }

5928
    if (!(priv->monConfig = virDomainChrSourceDefNew(NULL)))
5929 5930 5931 5932 5933 5934
        goto cleanup;

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

5935
    priv->monJSON = true;
5936 5937 5938 5939
    priv->monError = false;
    priv->monStart = 0;
    priv->gotShutdown = false;

5940 5941 5942 5943
    VIR_DEBUG("Updating guest CPU definition");
    if (qemuProcessUpdateGuestCPU(vm->def, priv->qemuCaps, caps, flags) < 0)
        goto cleanup;

5944 5945 5946 5947 5948
    for (i = 0; i < vm->def->nshmems; i++) {
        if (qemuDomainPrepareShmemChardev(vm->def->shmems[i]) < 0)
            goto cleanup;
    }

5949 5950 5951
    ret = 0;
 cleanup:
    virObjectUnref(caps);
5952
    virObjectUnref(cfg);
5953 5954 5955 5956
    return ret;
}


5957
static int
5958 5959 5960
qemuProcessSEVCreateFile(const char *configDir,
                         const char *name,
                         const char *data)
5961 5962
{
    char *configFile;
5963
    int ret = -1;
5964 5965 5966 5967 5968 5969 5970

    if (!(configFile = virFileBuildPath(configDir, name, ".base64")))
        return -1;

    if (virFileRewriteStr(configFile, S_IRUSR | S_IWUSR, data) < 0) {
        virReportSystemError(errno, _("failed to write data to config '%s'"),
                             configFile);
5971
        goto cleanup;
5972 5973
    }

5974
    ret = 0;
5975
 cleanup:
5976
    VIR_FREE(configFile);
5977
    return ret;
5978 5979 5980 5981
}


static int
J
Ján Tomko 已提交
5982
qemuProcessPrepareSEVGuestInput(virDomainObjPtr vm)
5983 5984 5985 5986
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virDomainDefPtr def = vm->def;
    virQEMUCapsPtr qemuCaps = priv->qemuCaps;
5987
    virDomainSEVDefPtr sev = def->sev;
5988 5989 5990 5991

    if (!sev)
        return 0;

5992
    VIR_DEBUG("Preparing SEV guest");
5993 5994 5995 5996 5997 5998 5999 6000 6001

    if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_SEV_GUEST)) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Domain %s asked for 'sev' launch but this "
                          "QEMU does not support SEV feature"), vm->def->name);
        return -1;
    }

    if (sev->dh_cert) {
6002
        if (qemuProcessSEVCreateFile(priv->libDir, "dh_cert", sev->dh_cert) < 0)
6003 6004 6005 6006
            return -1;
    }

    if (sev->session) {
6007
        if (qemuProcessSEVCreateFile(priv->libDir, "session", sev->session) < 0)
6008 6009 6010 6011 6012 6013 6014
            return -1;
    }

    return 0;
}


6015 6016 6017 6018 6019 6020 6021 6022 6023 6024 6025 6026 6027 6028 6029
static int
qemuProcessPrepareHostStorage(virQEMUDriverPtr driver,
                              virDomainObjPtr vm,
                              unsigned int flags)
{
    size_t i;
    bool cold_boot = flags & VIR_QEMU_PROCESS_START_COLD;

    for (i = vm->def->ndisks; i > 0; i--) {
        size_t idx = i - 1;
        virDomainDiskDefPtr disk = vm->def->disks[idx];

        if (virStorageSourceIsEmpty(disk->src))
            continue;

6030 6031 6032
        virStorageSourceBackingStoreClear(disk->src);

        if (qemuDomainDetermineDiskChain(driver, vm, disk, true) >= 0)
6033 6034 6035 6036 6037 6038 6039 6040 6041 6042 6043 6044
            continue;

        if (qemuDomainCheckDiskStartupPolicy(driver, vm, idx, cold_boot) >= 0)
            continue;

        return -1;
    }

    return 0;
}


J
Ján Tomko 已提交
6045
int
6046 6047 6048 6049 6050 6051 6052 6053 6054 6055 6056 6057 6058 6059 6060 6061 6062 6063 6064 6065 6066 6067 6068 6069 6070 6071 6072 6073 6074
qemuProcessOpenVhostVsock(virDomainVsockDefPtr vsock)
{
    qemuDomainVsockPrivatePtr priv = (qemuDomainVsockPrivatePtr)vsock->privateData;
    const char *vsock_path = "/dev/vhost-vsock";
    int fd;

    if ((fd = open(vsock_path, O_RDWR)) < 0) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       "%s", _("unable to open vhost-vsock device"));
        return -1;
    }

    if (vsock->auto_cid == VIR_TRISTATE_BOOL_YES) {
        if (virVsockAcquireGuestCid(fd, &vsock->guest_cid) < 0)
            goto error;
    } else {
        if (virVsockSetGuestCid(fd, vsock->guest_cid) < 0)
            goto error;
    }

    priv->vhostfd = fd;
    return 0;

 error:
    VIR_FORCE_CLOSE(fd);
    return -1;
}


J
Jiri Denemark 已提交
6075
/**
6076 6077 6078
 * qemuProcessPrepareHost:
 * @driver: qemu driver
 * @vm: domain object
6079
 * @flags: qemuProcessStartFlags
J
Jiri Denemark 已提交
6080
 *
6081 6082 6083
 * 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
6084
 *
6085
 * TODO: move all host modification from qemuBuildCommandLine into this function
J
Jiri Denemark 已提交
6086 6087
 */
int
6088 6089
qemuProcessPrepareHost(virQEMUDriverPtr driver,
                       virDomainObjPtr vm,
6090
                       unsigned int flags)
6091
{
6092
    int ret = -1;
6093
    unsigned int hostdev_flags = 0;
6094 6095
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
6096

6097
    if (qemuPrepareNVRAM(cfg, vm) < 0)
J
Jiri Denemark 已提交
6098
        goto cleanup;
6099

6100 6101 6102 6103
    if (vm->def->vsock) {
        if (qemuProcessOpenVhostVsock(vm->def->vsock) < 0)
            goto cleanup;
    }
6104 6105 6106 6107 6108
    /* 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");
6109
    if (qemuProcessNetworkPrepareDevices(vm->def) < 0)
J
Jiri Denemark 已提交
6110
        goto cleanup;
6111

6112
    /* Must be run before security labelling */
6113
    VIR_DEBUG("Preparing host devices");
6114 6115
    if (!cfg->relaxedACS)
        hostdev_flags |= VIR_HOSTDEV_STRICT_ACS_CHECK;
6116
    if (flags & VIR_QEMU_PROCESS_START_NEW)
6117
        hostdev_flags |= VIR_HOSTDEV_COLD_BOOT;
6118 6119
    if (qemuHostdevPrepareDomainDevices(driver, vm->def, priv->qemuCaps,
                                        hostdev_flags) < 0)
J
Jiri Denemark 已提交
6120
        goto cleanup;
6121

6122
    VIR_DEBUG("Preparing chr devices");
6123 6124 6125 6126
    if (virDomainChrDefForeach(vm->def,
                               true,
                               qemuProcessPrepareChardevDevice,
                               NULL) < 0)
J
Jiri Denemark 已提交
6127
        goto cleanup;
6128

6129
    if (qemuProcessBuildDestroyMemoryPaths(driver, vm, NULL, true) < 0)
6130
        goto cleanup;
6131

6132 6133
    /* Ensure no historical cgroup for this VM is lying around bogus
     * settings */
6134
    VIR_DEBUG("Ensuring no historical cgroup is lying around");
6135
    qemuRemoveCgroup(vm);
6136

6137
    if (virFileMakePath(cfg->logDir) < 0) {
6138 6139
        virReportSystemError(errno,
                             _("cannot create log directory %s"),
6140
                             cfg->logDir);
J
Jiri Denemark 已提交
6141
        goto cleanup;
6142 6143
    }

6144 6145 6146 6147 6148 6149 6150 6151 6152 6153 6154 6155 6156 6157 6158 6159 6160 6161 6162 6163 6164 6165 6166
    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;

6167 6168
    VIR_DEBUG("Write domain masterKey");
    if (qemuDomainWriteMasterKeyFile(driver, vm) < 0)
J
John Ferlan 已提交
6169 6170
        goto cleanup;

6171 6172 6173 6174
    VIR_DEBUG("Preparing disks (host)");
    if (qemuProcessPrepareHostStorage(driver, vm, flags) < 0)
        goto cleanup;

6175 6176 6177 6178
    VIR_DEBUG("Preparing external devices");
    if (qemuExtDevicesPrepareHost(driver, vm->def) < 0)
        goto cleanup;

J
Ján Tomko 已提交
6179
    if (qemuProcessPrepareSEVGuestInput(vm) < 0)
6180 6181
        goto cleanup;

6182 6183 6184 6185 6186 6187 6188
    ret = 0;
 cleanup:
    virObjectUnref(cfg);
    return ret;
}


6189 6190 6191 6192 6193 6194 6195 6196 6197 6198 6199 6200 6201 6202 6203 6204 6205 6206 6207 6208 6209 6210 6211 6212 6213 6214 6215 6216 6217 6218 6219 6220 6221
/**
 * qemuProcessGenID:
 * @vm: Pointer to domain object
 * @flags: qemuProcessStartFlags
 *
 * If this domain is requesting to use genid, then update the GUID
 * value if the VIR_QEMU_PROCESS_START_GEN_VMID flag is set. This
 * flag is set on specific paths during domain start processing when
 * there is the possibility that the VM is potentially re-executing
 * something that has already been executed before.
 */
static int
qemuProcessGenID(virDomainObjPtr vm,
                 unsigned int flags)
{
    if (!vm->def->genidRequested)
        return 0;

    /* If we are coming from a path where we must provide a new gen id
     * value regardless of whether it was previously generated or provided,
     * then generate a new GUID value before we build the command line. */
    if (flags & VIR_QEMU_PROCESS_START_GEN_VMID) {
        if (virUUIDGenerate(vm->def->genid) < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("failed to regenerate genid"));
            return -1;
        }
    }

    return 0;
}


6222 6223 6224 6225 6226 6227 6228 6229 6230 6231 6232 6233 6234 6235 6236 6237 6238 6239 6240 6241 6242 6243 6244 6245 6246 6247 6248 6249 6250 6251 6252 6253 6254 6255
/**
 * 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
6256
    size_t i;
6257

6258
    VIR_DEBUG("conn=%p driver=%p vm=%p name=%s if=%d asyncJob=%d "
6259 6260 6261
              "incoming.launchURI=%s incoming.deferredURI=%s "
              "incoming.fd=%d incoming.path=%s "
              "snapshot=%p vmop=%d flags=0x%x",
6262
              conn, driver, vm, vm->def->name, vm->def->id, asyncJob,
6263 6264 6265 6266 6267 6268 6269 6270 6271 6272
              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 |
6273
                  VIR_QEMU_PROCESS_START_AUTODESTROY |
6274 6275
                  VIR_QEMU_PROCESS_START_NEW |
                  VIR_QEMU_PROCESS_START_GEN_VMID, -1);
6276 6277 6278

    cfg = virQEMUDriverGetConfig(driver);

6279 6280 6281 6282 6283 6284
    if ((flags & VIR_QEMU_PROCESS_START_AUTODESTROY) && !conn) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Domain autodestroy requires a connection handle"));
        return -1;
    }

6285 6286 6287 6288 6289 6290 6291 6292
    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;

6293
    VIR_DEBUG("Creating domain log file");
6294 6295
    if (!(logCtxt = qemuDomainLogContextNew(driver, vm,
                                            QEMU_DOMAIN_LOG_CONTEXT_MODE_START)))
J
Jiri Denemark 已提交
6296
        goto cleanup;
6297
    logfile = qemuDomainLogContextGetWriteFD(logCtxt);
6298

6299 6300 6301
    if (qemuProcessGenID(vm, flags) < 0)
        goto cleanup;

6302 6303 6304
    if (qemuExtDevicesStart(driver, vm->def, logCtxt) < 0)
        goto cleanup;

6305
    VIR_DEBUG("Building emulator command line");
6306
    if (!(cmd = qemuBuildCommandLine(driver,
6307
                                     qemuDomainLogContextGetManager(logCtxt),
6308
                                     driver->securityManager,
6309
                                     vm,
6310 6311
                                     incoming ? incoming->launchURI : NULL,
                                     snapshot, vmop,
J
Ján Tomko 已提交
6312
                                     false,
6313
                                     qemuCheckFips(),
6314
                                     &nnicindexes, &nicindexes)))
J
Jiri Denemark 已提交
6315
        goto cleanup;
6316

6317 6318
    if (incoming && incoming->fd != -1)
        virCommandPassFD(cmd, incoming->fd, 0);
6319

6320
    /* now that we know it is about to start call the hook if present */
6321 6322 6323
    if (qemuProcessStartHook(driver, vm,
                             VIR_HOOK_QEMU_OP_START,
                             VIR_HOOK_SUBOP_BEGIN) < 0)
J
Jiri Denemark 已提交
6324
        goto cleanup;
6325

6326
    qemuLogOperation(vm, "starting up", cmd, logCtxt);
6327

6328
    qemuDomainObjCheckTaint(driver, vm, logCtxt);
6329

6330
    qemuDomainLogContextMarkPosition(logCtxt);
6331

6332 6333 6334 6335 6336
    VIR_DEBUG("Building mount namespace");

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

6337
    VIR_DEBUG("Clear emulator capabilities: %d",
6338 6339
              cfg->clearEmulatorCapabilities);
    if (cfg->clearEmulatorCapabilities)
6340 6341
        virCommandClearCaps(cmd);

6342 6343
    VIR_DEBUG("Setting up raw IO");
    if (qemuProcessSetupRawIO(driver, vm, cmd) < 0)
J
Jiri Denemark 已提交
6344
        goto cleanup;
6345

6346
    virCommandSetPreExecHook(cmd, qemuProcessHook, &hookData);
6347 6348
    virCommandSetMaxProcesses(cmd, cfg->maxProcesses);
    virCommandSetMaxFiles(cmd, cfg->maxFiles);
6349
    virCommandSetMaxCoreSize(cmd, cfg->maxCore);
6350
    virCommandSetUmask(cmd, 0x002);
6351

6352
    VIR_DEBUG("Setting up security labelling");
6353 6354
    if (qemuSecuritySetChildProcessLabel(driver->securityManager,
                                         vm->def, cmd) < 0)
J
Jiri Denemark 已提交
6355
        goto cleanup;
6356

6357 6358 6359
    virCommandSetOutputFD(cmd, &logfile);
    virCommandSetErrorFD(cmd, &logfile);
    virCommandNonblockingFDs(cmd);
6360
    virCommandSetPidFile(cmd, priv->pidfile);
6361
    virCommandDaemonize(cmd);
6362
    virCommandRequireHandshake(cmd);
6363

6364
    if (qemuSecurityPreFork(driver->securityManager) < 0)
J
Jiri Denemark 已提交
6365
        goto cleanup;
6366
    rv = virCommandRun(cmd, NULL);
6367
    qemuSecurityPostFork(driver->securityManager);
6368

E
Eric Blake 已提交
6369
    /* wait for qemu process to show up */
6370
    if (rv == 0) {
6371
        if (virPidFileReadPath(priv->pidfile, &vm->pid) < 0) {
6372 6373
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Domain %s didn't show up"), vm->def->name);
6374
            rv = -1;
6375
        }
M
Michal Privoznik 已提交
6376
        VIR_DEBUG("QEMU vm=%p name=%s running with pid=%lld",
6377
                  vm, vm->def->name, (long long)vm->pid);
6378 6379 6380
    } else {
        VIR_DEBUG("QEMU vm=%p name=%s failed to spawn",
                  vm, vm->def->name);
6381 6382
    }

6383
    VIR_DEBUG("Writing early domain status to disk");
6384
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0)
J
Jiri Denemark 已提交
6385
        goto cleanup;
6386

6387 6388
    VIR_DEBUG("Waiting for handshake from child");
    if (virCommandHandshakeWait(cmd) < 0) {
6389
        /* Read errors from child that occurred between fork and exec. */
6390 6391
        qemuProcessReportLogError(logCtxt,
                                  _("Process exited prior to exec"));
J
Jiri Denemark 已提交
6392
        goto cleanup;
6393 6394
    }

6395
    VIR_DEBUG("Setting up domain cgroup (if required)");
6396
    if (qemuSetupCgroup(vm, nnicindexes, nicindexes) < 0)
J
Jiri Denemark 已提交
6397
        goto cleanup;
6398

6399 6400 6401 6402
    if (!(priv->perf = virPerfNew()))
        goto cleanup;

    for (i = 0; i < VIR_PERF_EVENT_LAST; i++) {
6403
        if (vm->def->perf.events[i] == VIR_TRISTATE_BOOL_YES &&
6404 6405
            virPerfEventEnable(priv->perf, i, vm->pid) < 0)
            goto cleanup;
Q
Qiaowei Ren 已提交
6406
    }
6407

6408 6409 6410 6411 6412 6413
    /* This must be done after cgroup placement to avoid resetting CPU
     * affinity */
    if (!vm->def->cputune.emulatorpin &&
        qemuProcessInitCpuAffinity(vm) < 0)
        goto cleanup;

6414 6415 6416 6417
    VIR_DEBUG("Setting emulator tuning/settings");
    if (qemuProcessSetupEmulator(vm) < 0)
        goto cleanup;

6418 6419 6420 6421
    VIR_DEBUG("Setting cgroup for external devices (if required)");
    if (qemuSetupCgroupForExtDevices(vm, driver) < 0)
        goto cleanup;

6422
    VIR_DEBUG("Setting up resctrl");
6423 6424 6425
    if (qemuProcessResctrlCreate(driver, vm) < 0)
        goto cleanup;

6426
    VIR_DEBUG("Setting up managed PR daemon");
6427 6428
    if (virDomainDefHasManagedPR(vm->def) &&
        qemuProcessStartManagedPRDaemon(vm) < 0)
6429 6430
        goto cleanup;

6431
    VIR_DEBUG("Setting domain security labels");
6432 6433 6434 6435
    if (qemuSecuritySetAllLabel(driver,
                                vm,
                                incoming ? incoming->path : NULL) < 0)
        goto cleanup;
6436

6437
    /* Security manager labeled all devices, therefore
J
Jiri Denemark 已提交
6438 6439 6440 6441
     * if any operation from now on fails, we need to ask the caller to
     * restore labels.
     */
    ret = -2;
6442

J
Jiri Denemark 已提交
6443
    if (incoming && incoming->fd != -1) {
6444 6445 6446 6447 6448 6449 6450
        /* 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
6451
        if (fstat(incoming->fd, &stdin_sb) < 0) {
6452
            virReportSystemError(errno,
J
Jiri Denemark 已提交
6453 6454
                                 _("cannot stat fd %d"), incoming->fd);
            goto cleanup;
6455 6456
        }
        if (S_ISFIFO(stdin_sb.st_mode) &&
6457 6458
            qemuSecuritySetImageFDLabel(driver->securityManager,
                                        vm->def, incoming->fd) < 0)
J
Jiri Denemark 已提交
6459
            goto cleanup;
6460 6461 6462
    }

    VIR_DEBUG("Labelling done, completing handshake to child");
6463
    if (virCommandHandshakeNotify(cmd) < 0)
J
Jiri Denemark 已提交
6464
        goto cleanup;
6465 6466
    VIR_DEBUG("Handshake complete, child running");

6467
    if (rv == -1) /* The VM failed to start; tear filters before taps */
6468 6469
        virDomainConfVMNWFilterTeardown(vm);

6470
    if (rv == -1) /* The VM failed to start */
J
Jiri Denemark 已提交
6471
        goto cleanup;
6472

6473
    VIR_DEBUG("Waiting for monitor to show up");
6474
    if (qemuProcessWaitForMonitor(driver, vm, asyncJob, logCtxt) < 0)
J
Jiri Denemark 已提交
6475
        goto cleanup;
6476

6477 6478
    if (qemuConnectAgent(driver, vm) < 0)
        goto cleanup;
D
Daniel P. Berrange 已提交
6479

6480
    VIR_DEBUG("Verifying and updating provided guest CPU");
6481
    if (qemuProcessUpdateAndVerifyCPU(driver, vm, asyncJob) < 0)
J
Jiri Denemark 已提交
6482
        goto cleanup;
6483

6484
    VIR_DEBUG("Setting up post-init cgroup restrictions");
6485
    if (qemuSetupCpusetMems(vm) < 0)
J
Jiri Denemark 已提交
6486
        goto cleanup;
6487

6488 6489 6490 6491 6492 6493 6494 6495 6496 6497 6498 6499
    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;
    }

6500
    VIR_DEBUG("Refreshing VCPU info");
6501
    if (qemuDomainRefreshVcpuInfo(driver, vm, asyncJob, false) < 0)
J
Jiri Denemark 已提交
6502
        goto cleanup;
6503

6504 6505 6506
    if (qemuDomainValidateVcpuInfo(vm) < 0)
        goto cleanup;

6507 6508
    qemuDomainVcpuPersistOrder(vm->def);

6509 6510
    VIR_DEBUG("Detecting IOThread PIDs");
    if (qemuProcessDetectIOThreadPIDs(driver, vm, asyncJob) < 0)
J
Jiri Denemark 已提交
6511
        goto cleanup;
6512

6513 6514 6515 6516
    VIR_DEBUG("Setting global CPU cgroup (if required)");
    if (qemuSetupGlobalCpuCgroup(vm) < 0)
        goto cleanup;

6517 6518
    VIR_DEBUG("Setting vCPU tuning/settings");
    if (qemuProcessSetupVcpus(vm) < 0)
J
Jiri Denemark 已提交
6519
        goto cleanup;
6520

6521 6522
    VIR_DEBUG("Setting IOThread tuning/settings");
    if (qemuProcessSetupIOThreads(vm) < 0)
J
Jiri Denemark 已提交
6523
        goto cleanup;
6524

6525
    VIR_DEBUG("Setting any required VM passwords");
6526
    if (qemuProcessInitPasswords(driver, vm, asyncJob) < 0)
J
Jiri Denemark 已提交
6527
        goto cleanup;
6528

6529 6530 6531 6532
    /* set default link states */
    /* qemu doesn't support setting this on the command line, so
     * enter the monitor */
    VIR_DEBUG("Setting network link states");
6533
    if (qemuProcessSetLinkStates(driver, vm, asyncJob) < 0)
J
Jiri Denemark 已提交
6534
        goto cleanup;
6535

6536
    VIR_DEBUG("Setting initial memory amount");
6537
    if (qemuProcessSetupBalloon(driver, vm, asyncJob) < 0)
J
Jiri Denemark 已提交
6538
        goto cleanup;
6539

6540
    /* Since CPUs were not started yet, the balloon could not return the memory
6541 6542
     * 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 */
6543
    if (!incoming && !snapshot &&
6544
        qemuProcessRefreshBalloonState(driver, vm, asyncJob) < 0)
J
Jiri Denemark 已提交
6545
        goto cleanup;
6546

J
Jiri Denemark 已提交
6547 6548 6549 6550 6551 6552 6553
    if (flags & VIR_QEMU_PROCESS_START_AUTODESTROY &&
        qemuProcessAutoDestroyAdd(driver, vm, conn) < 0)
        goto cleanup;

    ret = 0;

 cleanup:
6554 6555
    if (ret < 0)
        qemuExtDevicesStop(driver, vm->def);
6556
    qemuDomainSecretDestroy(vm);
J
Jiri Denemark 已提交
6557
    virCommandFree(cmd);
6558
    virObjectUnref(logCtxt);
J
Jiri Denemark 已提交
6559 6560 6561 6562 6563 6564 6565
    virObjectUnref(cfg);
    virObjectUnref(caps);
    VIR_FREE(nicindexes);
    return ret;
}


6566 6567 6568 6569 6570 6571 6572 6573 6574 6575
/**
 * qemuProcessRefreshState:
 * @driver: qemu driver data
 * @vm: domain to refresh
 * @asyncJob: async job type
 *
 * This function gathers calls to refresh qemu state after startup. This
 * function is called after a deferred migration finishes so that we can update
 * state influenced by the migration stream.
 */
6576
int
6577 6578 6579 6580 6581 6582 6583 6584 6585 6586 6587 6588 6589 6590 6591 6592 6593 6594 6595 6596 6597 6598 6599 6600
qemuProcessRefreshState(virQEMUDriverPtr driver,
                        virDomainObjPtr vm,
                        qemuDomainAsyncJob asyncJob)
{
    VIR_DEBUG("Fetching list of active devices");
    if (qemuDomainUpdateDeviceList(driver, vm, asyncJob) < 0)
        return -1;

    VIR_DEBUG("Updating info of memory devices");
    if (qemuDomainUpdateMemoryDeviceInfo(driver, vm, asyncJob) < 0)
        return -1;

    VIR_DEBUG("Detecting actual memory size for video device");
    if (qemuProcessUpdateVideoRamSize(driver, vm, asyncJob) < 0)
        return -1;

    VIR_DEBUG("Updating disk data");
    if (qemuProcessRefreshDisks(driver, vm, asyncJob) < 0)
        return -1;

    return 0;
}


6601 6602 6603 6604 6605 6606
/**
 * qemuProcessFinishStartup:
 *
 * Finish starting a new domain.
 */
int
6607
qemuProcessFinishStartup(virQEMUDriverPtr driver,
6608 6609 6610 6611 6612 6613 6614 6615 6616 6617
                         virDomainObjPtr vm,
                         qemuDomainAsyncJob asyncJob,
                         bool startCPUs,
                         virDomainPausedReason pausedReason)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    int ret = -1;

    if (startCPUs) {
        VIR_DEBUG("Starting domain CPUs");
6618
        if (qemuProcessStartCPUs(driver, vm,
6619 6620
                                 VIR_DOMAIN_RUNNING_BOOTED,
                                 asyncJob) < 0) {
6621
            if (virGetLastErrorCode() == VIR_ERR_OK)
6622 6623 6624 6625 6626 6627 6628 6629 6630
                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");
6631
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0)
6632 6633 6634 6635 6636 6637 6638 6639 6640 6641 6642 6643 6644 6645 6646
        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
6647 6648 6649 6650
int
qemuProcessStart(virConnectPtr conn,
                 virQEMUDriverPtr driver,
                 virDomainObjPtr vm,
6651
                 virCPUDefPtr updatedCPU,
J
Jiri Denemark 已提交
6652 6653 6654 6655 6656 6657 6658 6659 6660 6661 6662 6663 6664 6665 6666 6667 6668 6669 6670 6671 6672 6673 6674 6675 6676
                 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 |
6677 6678
                      VIR_QEMU_PROCESS_START_AUTODESTROY |
                      VIR_QEMU_PROCESS_START_GEN_VMID, cleanup);
J
Jiri Denemark 已提交
6679

6680 6681 6682
    if (!migrateFrom && !snapshot)
        flags |= VIR_QEMU_PROCESS_START_NEW;

6683 6684
    if (qemuProcessInit(driver, vm, updatedCPU,
                        asyncJob, !!migrateFrom, flags) < 0)
J
Jiri Denemark 已提交
6685 6686 6687
        goto cleanup;

    if (migrateFrom) {
6688
        incoming = qemuProcessIncomingDefNew(priv->qemuCaps, NULL, migrateFrom,
J
Jiri Denemark 已提交
6689 6690 6691 6692 6693
                                             migrateFd, migratePath);
        if (!incoming)
            goto stop;
    }

6694
    if (qemuProcessPrepareDomain(driver, vm, flags) < 0)
6695 6696
        goto stop;

6697
    if (qemuProcessPrepareHost(driver, vm, flags) < 0)
6698 6699
        goto stop;

J
Jiri Denemark 已提交
6700 6701
    if ((rv = qemuProcessLaunch(conn, driver, vm, asyncJob, incoming,
                                snapshot, vmop, flags)) < 0) {
6702
        if (rv == -2)
J
Jiri Denemark 已提交
6703 6704 6705 6706
            relabel = true;
        goto stop;
    }
    relabel = true;
6707

6708 6709
    if (incoming &&
        incoming->deferredURI &&
6710
        qemuMigrationDstRun(driver, vm, incoming->deferredURI, asyncJob) < 0)
J
Jiri Denemark 已提交
6711
        goto stop;
6712

6713
    if (qemuProcessFinishStartup(driver, vm, asyncJob,
6714 6715 6716 6717
                                 !(flags & VIR_QEMU_PROCESS_START_PAUSED),
                                 incoming ?
                                 VIR_DOMAIN_PAUSED_MIGRATION :
                                 VIR_DOMAIN_PAUSED_USER) < 0)
J
Jiri Denemark 已提交
6718
        goto stop;
6719

6720 6721 6722
    if (!incoming) {
        /* Keep watching qemu log for errors during incoming migration, otherwise
         * unset reporting errors from qemu log. */
6723
        qemuMonitorSetDomainLog(priv->mon, NULL, NULL, NULL);
6724

6725 6726 6727 6728 6729 6730
        /* Refresh state of devices from qemu. During migration this needs to
         * happen after the state information is fully transferred. */
        if (qemuProcessRefreshState(driver, vm, asyncJob) < 0)
            goto stop;
    }

6731 6732 6733
    ret = 0;

 cleanup:
6734
    qemuProcessIncomingDefFree(incoming);
6735
    return ret;
6736

J
Jiri Denemark 已提交
6737 6738 6739 6740 6741 6742
 stop:
    stopFlags = 0;
    if (!relabel)
        stopFlags |= VIR_QEMU_PROCESS_STOP_NO_RELABEL;
    if (migrateFrom)
        stopFlags |= VIR_QEMU_PROCESS_STOP_MIGRATED;
6743
    if (priv->mon)
6744
        qemuMonitorSetDomainLog(priv->mon, NULL, NULL, NULL);
6745
    qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_FAILED, asyncJob, stopFlags);
6746
    goto cleanup;
6747 6748 6749
}


6750
virCommandPtr
6751
qemuProcessCreatePretendCmd(virQEMUDriverPtr driver,
6752 6753
                            virDomainObjPtr vm,
                            const char *migrateURI,
6754
                            bool enableFips,
6755 6756 6757 6758 6759 6760 6761 6762 6763 6764
                            bool standalone,
                            unsigned int flags)
{
    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;
6765
    flags |= VIR_QEMU_PROCESS_START_NEW;
6766 6767
    if (standalone)
        flags |= VIR_QEMU_PROCESS_START_STANDALONE;
6768

6769 6770
    if (qemuProcessInit(driver, vm, NULL, QEMU_ASYNC_JOB_NONE,
                        !!migrateURI, flags) < 0)
6771 6772
        goto cleanup;

6773
    if (qemuProcessPrepareDomain(driver, vm, flags) < 0)
6774 6775 6776
        goto cleanup;

    VIR_DEBUG("Building emulator command line");
6777
    cmd = qemuBuildCommandLine(driver,
6778
                               NULL,
6779
                               driver->securityManager,
6780
                               vm,
6781 6782 6783 6784
                               migrateURI,
                               NULL,
                               VIR_NETDEV_VPORT_PROFILE_OP_NO_OP,
                               standalone,
6785
                               enableFips,
6786
                               NULL,
6787
                               NULL);
6788 6789 6790 6791 6792 6793

 cleanup:
    return cmd;
}


6794
int
6795
qemuProcessKill(virDomainObjPtr vm, unsigned int flags)
6796
{
6797
    int ret;
6798

6799
    VIR_DEBUG("vm=%p name=%s pid=%lld flags=0x%x",
6800
              vm, vm->def->name,
6801
              (long long)vm->pid, flags);
6802

6803 6804 6805 6806 6807
    if (!(flags & VIR_QEMU_PROCESS_KILL_NOCHECK)) {
        if (!virDomainObjIsActive(vm)) {
            VIR_DEBUG("VM '%s' not active", vm->def->name);
            return 0;
        }
6808 6809
    }

6810
    if (flags & VIR_QEMU_PROCESS_KILL_NOWAIT) {
6811 6812 6813 6814 6815
        virProcessKill(vm->pid,
                       (flags & VIR_QEMU_PROCESS_KILL_FORCE) ?
                       SIGKILL : SIGTERM);
        return 0;
    }
6816

6817 6818
    ret = virProcessKillPainfully(vm->pid,
                                  !!(flags & VIR_QEMU_PROCESS_KILL_FORCE));
6819

6820
    return ret;
6821 6822 6823
}


6824 6825 6826 6827 6828 6829 6830 6831 6832 6833 6834 6835 6836 6837 6838 6839 6840 6841 6842 6843 6844 6845 6846 6847 6848 6849 6850 6851 6852 6853 6854 6855 6856 6857 6858 6859 6860 6861 6862
/**
 * 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;
}


6863
void qemuProcessStop(virQEMUDriverPtr driver,
6864
                     virDomainObjPtr vm,
6865
                     virDomainShutoffReason reason,
6866
                     qemuDomainAsyncJob asyncJob,
6867
                     unsigned int flags)
6868 6869 6870 6871 6872 6873
{
    int ret;
    int retries = 0;
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virErrorPtr orig_err;
    virDomainDefPtr def;
A
Ansis Atteka 已提交
6874
    virNetDevVPortProfilePtr vport = NULL;
6875
    size_t i;
6876
    char *timestamp;
6877
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
6878

M
Michal Privoznik 已提交
6879
    VIR_DEBUG("Shutting down vm=%p name=%s id=%d pid=%lld, "
6880
              "reason=%s, asyncJob=%s, flags=0x%x",
6881
              vm, vm->def->name, vm->def->id,
6882
              (long long)vm->pid,
6883 6884 6885
              virDomainShutoffReasonTypeToString(reason),
              qemuDomainAsyncJobTypeToString(asyncJob),
              flags);
6886

6887 6888 6889 6890
    /* This method is routinely used in clean up paths. Disable error
     * reporting so we don't squash a legit error. */
    orig_err = virSaveLastError();

6891 6892 6893 6894 6895 6896 6897 6898 6899 6900 6901 6902 6903 6904 6905
    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;
    }

6906
    qemuProcessBuildDestroyMemoryPaths(driver, vm, NULL, false);
6907

6908 6909
    vm->def->id = -1;

6910
    if (virAtomicIntDecAndTest(&driver->nactive) && driver->inhibitCallback)
6911 6912
        driver->inhibitCallback(false, driver->inhibitOpaque);

6913 6914
    /* Wake up anything waiting on domain condition */
    virDomainObjBroadcast(vm);
6915

6916
    if ((timestamp = virTimeStringNow()) != NULL) {
6917 6918 6919
        qemuDomainLogAppendMessage(driver, vm, "%s: shutting down, reason=%s\n",
                                   timestamp,
                                   virDomainShutoffReasonTypeToString(reason));
6920
        VIR_FREE(timestamp);
6921 6922
    }

6923 6924 6925
    /* Clear network bandwidth */
    virDomainClearNetBandwidth(vm);

6926 6927
    virDomainConfVMNWFilterTeardown(vm);

6928
    if (cfg->macFilter) {
6929
        def = vm->def;
6930
        for (i = 0; i < def->nnets; i++) {
6931 6932 6933
            virDomainNetDefPtr net = def->nets[i];
            if (net->ifname == NULL)
                continue;
6934 6935 6936
            ignore_value(ebtablesRemoveForwardAllowIn(driver->ebtables,
                                                      net->ifname,
                                                      &net->mac));
6937 6938 6939
        }
    }

6940
    virPortAllocatorRelease(priv->nbdPort);
6941
    priv->nbdPort = 0;
6942

D
Daniel P. Berrange 已提交
6943 6944 6945 6946
    if (priv->agent) {
        qemuAgentClose(priv->agent);
        priv->agent = NULL;
    }
6947
    priv->agentError = false;
D
Daniel P. Berrange 已提交
6948

6949
    if (priv->mon) {
6950
        qemuMonitorClose(priv->mon);
6951 6952
        priv->mon = NULL;
    }
6953 6954 6955 6956 6957 6958 6959 6960

    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
6961 6962 6963
    /* Remove the master key */
    qemuDomainMasterKeyRemove(priv);

6964
    /* Do this before we delete the tree and remove pidfile. */
6965
    qemuProcessKillManagedPRDaemon(vm);
6966

6967 6968
    virFileDeleteTree(priv->libDir);
    virFileDeleteTree(priv->channelTargetDir);
6969

6970 6971 6972 6973 6974 6975
    ignore_value(virDomainChrDefForeach(vm->def,
                                        false,
                                        qemuProcessCleanupChardevDevice,
                                        NULL));


6976
    /* shut it off for sure */
6977 6978 6979
    ignore_value(qemuProcessKill(vm,
                                 VIR_QEMU_PROCESS_KILL_FORCE|
                                 VIR_QEMU_PROCESS_KILL_NOCHECK));
6980

6981 6982
    qemuDomainCleanupRun(driver, vm);

6983 6984
    qemuExtDevicesStop(driver, vm->def);

6985
    /* Stop autodestroy in case guest is restarted */
6986
    qemuProcessAutoDestroyRemove(driver, vm);
6987

6988 6989
    /* now that we know it's stopped call the hook if present */
    if (virHookPresent(VIR_HOOK_DRIVER_QEMU)) {
6990
        char *xml = qemuDomainDefFormatXML(driver, vm->def, 0);
6991 6992

        /* we can't stop the operation even if the script raised an error */
6993 6994 6995
        ignore_value(virHookCall(VIR_HOOK_DRIVER_QEMU, vm->def->name,
                                 VIR_HOOK_QEMU_OP_STOPPED, VIR_HOOK_SUBOP_END,
                                 NULL, xml, NULL));
6996 6997 6998
        VIR_FREE(xml);
    }

6999 7000
    /* Reset Security Labels unless caller don't want us to */
    if (!(flags & VIR_QEMU_PROCESS_STOP_NO_RELABEL))
7001 7002 7003
        qemuSecurityRestoreAllLabel(driver, vm,
                                    !!(flags & VIR_QEMU_PROCESS_STOP_MIGRATED));

7004
    qemuSecurityReleaseLabel(driver->securityManager, vm->def);
7005

7006
    for (i = 0; i < vm->def->ndisks; i++) {
7007
        virDomainDeviceDef dev;
7008
        virDomainDiskDefPtr disk = vm->def->disks[i];
7009 7010 7011 7012

        dev.type = VIR_DOMAIN_DEVICE_DISK;
        dev.data.disk = disk;
        ignore_value(qemuRemoveSharedDevice(driver, &dev, vm->def->name));
7013 7014
    }

7015
    /* Clear out dynamically assigned labels */
7016
    for (i = 0; i < vm->def->nseclabels; i++) {
7017
        if (vm->def->seclabels[i]->type == VIR_DOMAIN_SECLABEL_DYNAMIC)
7018 7019
            VIR_FREE(vm->def->seclabels[i]->label);
        VIR_FREE(vm->def->seclabels[i]->imagelabel);
7020 7021
    }

7022
    qemuHostdevReAttachDomainDevices(driver, vm->def);
7023 7024 7025 7026

    def = vm->def;
    for (i = 0; i < def->nnets; i++) {
        virDomainNetDefPtr net = def->nets[i];
7027 7028 7029 7030
        vport = virDomainNetGetActualVirtPortProfile(net);

        switch (virDomainNetGetActualType(net)) {
        case VIR_DOMAIN_NET_TYPE_DIRECT:
7031
            ignore_value(virNetDevMacVLanDeleteWithVPortProfile(
7032
                             net->ifname, &net->mac,
7033 7034
                             virDomainNetGetActualDirectDev(net),
                             virDomainNetGetActualDirectMode(net),
7035
                             virDomainNetGetActualVirtPortProfile(net),
7036
                             cfg->stateDir));
7037
            break;
7038 7039 7040 7041 7042 7043
        case VIR_DOMAIN_NET_TYPE_ETHERNET:
            if (net->ifname) {
                ignore_value(virNetDevTapDelete(net->ifname, net->backend.tap));
                VIR_FREE(net->ifname);
            }
            break;
7044 7045 7046 7047
        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))
7048
                ignore_value(virNetDevTapDelete(net->ifname, net->backend.tap));
7049 7050
#endif
            break;
7051 7052 7053 7054 7055 7056 7057 7058 7059 7060 7061
        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;
7062
        }
7063 7064 7065
        /* release the physical device (or any other resources used by
         * this interface in the network driver
         */
7066 7067 7068 7069 7070 7071 7072 7073 7074
        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
7075

7076 7077
        /* kick the device out of the hostdev list too */
        virDomainNetRemoveHostdev(def, net);
7078
        virDomainNetReleaseActualDevice(vm->def, net);
7079
    }
7080

7081
 retry:
7082
    if ((ret = qemuRemoveCgroup(vm)) < 0) {
7083 7084 7085 7086 7087 7088 7089
        if (ret == -EBUSY && (retries++ < 5)) {
            usleep(200*1000);
            goto retry;
        }
        VIR_WARN("Failed to remove cgroup for %s",
                 vm->def->name);
    }
7090

7091 7092 7093 7094 7095 7096
    /* Remove resctrl allocation after cgroups are cleaned up which makes it
     * kind of safer (although removing the allocation should work even with
     * pids in tasks file */
    for (i = 0; i < vm->def->ncachetunes; i++)
        virResctrlAllocRemove(vm->def->cachetunes[i]->alloc);

7097 7098
    qemuProcessRemoveDomainStatus(driver, vm);

7099 7100
    /* Remove VNC and Spice ports from port reservation bitmap, but only if
       they were reserved by the driver (autoport=yes)
7101
    */
7102
    for (i = 0; i < vm->def->ngraphics; ++i) {
7103
        virDomainGraphicsDefPtr graphics = vm->def->graphics[i];
7104 7105
        if (graphics->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC) {
            if (graphics->data.vnc.autoport) {
7106
                virPortAllocatorRelease(graphics->data.vnc.port);
7107
            } else if (graphics->data.vnc.portReserved) {
7108
                virPortAllocatorRelease(graphics->data.spice.port);
7109 7110
                graphics->data.vnc.portReserved = false;
            }
7111
            if (graphics->data.vnc.websocketGenerated) {
7112
                virPortAllocatorRelease(graphics->data.vnc.websocket);
7113 7114 7115
                graphics->data.vnc.websocketGenerated = false;
                graphics->data.vnc.websocket = -1;
            } else if (graphics->data.vnc.websocket) {
7116
                virPortAllocatorRelease(graphics->data.vnc.websocket);
7117
            }
7118
        }
7119 7120
        if (graphics->type == VIR_DOMAIN_GRAPHICS_TYPE_SPICE) {
            if (graphics->data.spice.autoport) {
7121 7122
                virPortAllocatorRelease(graphics->data.spice.port);
                virPortAllocatorRelease(graphics->data.spice.tlsPort);
7123 7124
            } else {
                if (graphics->data.spice.portReserved) {
7125
                    virPortAllocatorRelease(graphics->data.spice.port);
7126 7127 7128 7129
                    graphics->data.spice.portReserved = false;
                }

                if (graphics->data.spice.tlsPortReserved) {
7130
                    virPortAllocatorRelease(graphics->data.spice.tlsPort);
7131 7132 7133
                    graphics->data.spice.tlsPortReserved = false;
                }
            }
7134
        }
7135 7136
    }

7137
    vm->taint = 0;
7138
    vm->pid = -1;
J
Jiri Denemark 已提交
7139
    virDomainObjSetState(vm, VIR_DOMAIN_SHUTOFF, reason);
7140 7141
    for (i = 0; i < vm->def->niothreadids; i++)
        vm->def->iothreadids[i]->thread_id = 0;
7142

7143 7144
    /* clear all private data entries which are no longer needed */
    qemuDomainObjPrivateDataClear(priv);
7145

7146
    /* The "release" hook cleans up additional resources */
7147
    if (virHookPresent(VIR_HOOK_DRIVER_QEMU)) {
7148
        char *xml = qemuDomainDefFormatXML(driver, vm->def, 0);
7149 7150 7151

        /* we can't stop the operation even if the script raised an error */
        virHookCall(VIR_HOOK_DRIVER_QEMU, vm->def->name,
7152 7153
                    VIR_HOOK_QEMU_OP_RELEASE, VIR_HOOK_SUBOP_END,
                    NULL, xml, NULL);
7154 7155 7156
        VIR_FREE(xml);
    }

7157
    virDomainObjRemoveTransientDef(vm);
7158

7159 7160 7161 7162 7163
 endjob:
    if (asyncJob != QEMU_ASYNC_JOB_NONE)
        qemuDomainObjEndJob(driver, vm);

 cleanup:
7164 7165 7166 7167
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
7168
    virObjectUnref(cfg);
7169
}
7170 7171


7172 7173 7174 7175 7176 7177 7178 7179 7180 7181 7182 7183 7184 7185 7186 7187 7188 7189 7190 7191 7192 7193 7194 7195 7196 7197 7198 7199 7200 7201 7202 7203 7204 7205 7206 7207 7208 7209 7210 7211 7212 7213 7214 7215 7216 7217 7218 7219 7220 7221 7222 7223 7224 7225 7226 7227 7228 7229 7230 7231 7232 7233 7234 7235 7236 7237 7238 7239 7240 7241 7242 7243 7244 7245 7246 7247 7248 7249 7250 7251 7252 7253 7254 7255 7256 7257 7258 7259 7260 7261 7262 7263 7264 7265 7266 7267 7268 7269 7270 7271 7272 7273 7274 7275 7276 7277 7278 7279 7280 7281 7282 7283
int qemuProcessAttach(virConnectPtr conn ATTRIBUTE_UNUSED,
                      virQEMUDriverPtr driver,
                      virDomainObjPtr vm,
                      pid_t pid,
                      const char *pidfile,
                      virDomainChrSourceDefPtr monConfig,
                      bool monJSON)
{
    size_t i;
    qemuDomainLogContextPtr logCtxt = NULL;
    char *timestamp;
    qemuDomainObjPrivatePtr priv = vm->privateData;
    bool running = true;
    virDomainPausedReason reason;
    virSecurityLabelPtr seclabel = NULL;
    virSecurityLabelDefPtr seclabeldef = NULL;
    bool seclabelgen = false;
    virSecurityManagerPtr* sec_managers = NULL;
    const char *model;
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    virCapsPtr caps = NULL;
    bool active = false;

    VIR_DEBUG("Beginning VM attach process");

    if (virDomainObjIsActive(vm)) {
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("VM is already active"));
        virObjectUnref(cfg);
        return -1;
    }

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

    /* 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");
    if (virDomainObjSetDefTransient(caps, driver->xmlopt, vm) < 0)
        goto error;

    vm->def->id = qemuDriverAllocateID(driver);

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

    if (virFileMakePath(cfg->logDir) < 0) {
        virReportSystemError(errno,
                             _("cannot create log directory %s"),
                             cfg->logDir);
        goto error;
    }

    VIR_FREE(priv->pidfile);
    if (VIR_STRDUP(priv->pidfile, pidfile) < 0)
        goto error;

    vm->pid = pid;

    VIR_DEBUG("Detect security driver config");
    sec_managers = qemuSecurityGetNested(driver->securityManager);
    if (sec_managers == NULL)
        goto error;

    for (i = 0; sec_managers[i]; i++) {
        seclabelgen = false;
        model = qemuSecurityGetModel(sec_managers[i]);
        seclabeldef = virDomainDefGetSecurityLabelDef(vm->def, model);
        if (seclabeldef == NULL) {
            if (!(seclabeldef = virSecurityLabelDefNew(model)))
                goto error;
            seclabelgen = true;
        }
        seclabeldef->type = VIR_DOMAIN_SECLABEL_STATIC;
        if (VIR_ALLOC(seclabel) < 0)
            goto error;
        if (qemuSecurityGetProcessLabel(sec_managers[i], vm->def,
                                        vm->pid, seclabel) < 0)
            goto error;

        if (VIR_STRDUP(seclabeldef->model, model) < 0)
            goto error;

        if (VIR_STRDUP(seclabeldef->label, seclabel->label) < 0)
            goto error;
        VIR_FREE(seclabel);

        if (seclabelgen) {
            if (VIR_APPEND_ELEMENT(vm->def->seclabels, vm->def->nseclabels, seclabeldef) < 0)
                goto error;
            seclabelgen = false;
        }
    }

    if (qemuSecurityCheckAllLabel(driver->securityManager, vm->def) < 0)
        goto error;
    if (qemuSecurityGenLabel(driver->securityManager, vm->def) < 0)
        goto error;

    if (qemuDomainPerfRestart(vm) < 0)
        goto error;

    VIR_DEBUG("Creating domain log file");
    if (!(logCtxt = qemuDomainLogContextNew(driver, vm,
                                            QEMU_DOMAIN_LOG_CONTEXT_MODE_ATTACH)))
        goto error;

    VIR_DEBUG("Determining emulator version");
    virObjectUnref(priv->qemuCaps);
7284
    if (!(priv->qemuCaps = virQEMUCapsCacheLookupCopy(driver->qemuCapsCache,
7285 7286 7287 7288 7289 7290 7291 7292 7293 7294 7295
                                                      vm->def->emulator,
                                                      vm->def->os.machine)))
        goto error;

    VIR_DEBUG("Preparing monitor state");
    priv->monConfig = monConfig;
    monConfig = NULL;
    priv->monJSON = monJSON;

    priv->gotShutdown = false;

7296 7297 7298 7299
    /* Attaching to running QEMU so we need to detect whether it was started
     * with -no-reboot. */
    qemuProcessPrepareAllowReboot(vm);

7300 7301 7302 7303 7304 7305 7306 7307 7308 7309 7310 7311 7312 7313 7314 7315 7316 7317 7318 7319 7320 7321 7322 7323 7324 7325 7326 7327 7328 7329 7330 7331 7332 7333 7334 7335 7336 7337 7338 7339 7340 7341 7342 7343 7344 7345 7346 7347 7348 7349 7350 7351 7352 7353 7354 7355 7356 7357 7358 7359 7360 7361 7362 7363 7364 7365 7366 7367 7368 7369 7370 7371 7372 7373 7374 7375 7376 7377 7378 7379 7380 7381 7382 7383 7384 7385 7386 7387 7388 7389 7390 7391 7392 7393 7394 7395 7396 7397 7398 7399 7400 7401 7402 7403 7404 7405 7406 7407 7408 7409 7410 7411 7412 7413 7414 7415 7416 7417 7418
    /*
     * 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
     */
    VIR_DEBUG("Assigning domain PCI addresses");
    if ((qemuDomainAssignAddresses(vm->def, priv->qemuCaps,
                                   driver, vm, false)) < 0) {
        goto error;
    }

    if ((timestamp = virTimeStringNow()) == NULL)
        goto error;

    qemuDomainLogContextWrite(logCtxt, "%s: attaching\n", timestamp);
    VIR_FREE(timestamp);

    qemuDomainObjTaint(driver, vm, VIR_DOMAIN_TAINT_EXTERNAL_LAUNCH, logCtxt);

    VIR_DEBUG("Waiting for monitor to show up");
    if (qemuProcessWaitForMonitor(driver, vm, QEMU_ASYNC_JOB_NONE, NULL) < 0)
        goto error;

    if (qemuConnectAgent(driver, vm) < 0)
        goto error;

    VIR_DEBUG("Detecting VCPU PIDs");
    if (qemuDomainRefreshVcpuInfo(driver, vm, QEMU_ASYNC_JOB_NONE, false) < 0)
        goto error;

    if (qemuDomainValidateVcpuInfo(vm) < 0)
        goto error;

    VIR_DEBUG("Detecting IOThread PIDs");
    if (qemuProcessDetectIOThreadPIDs(driver, vm, QEMU_ASYNC_JOB_NONE) < 0)
        goto error;

    VIR_DEBUG("Getting initial memory amount");
    qemuDomainObjEnterMonitor(driver, vm);
    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)
        goto error;

    if (running) {
        virDomainObjSetState(vm, VIR_DOMAIN_RUNNING,
                             VIR_DOMAIN_RUNNING_UNPAUSED);
        if (vm->def->memballoon &&
            vm->def->memballoon->model == VIR_DOMAIN_MEMBALLOON_MODEL_VIRTIO &&
            vm->def->memballoon->period) {
            qemuDomainObjEnterMonitor(driver, vm);
            qemuMonitorSetMemoryStatsPeriod(priv->mon, vm->def->memballoon,
                                            vm->def->memballoon->period);
            if (qemuDomainObjExitMonitor(driver, vm) < 0)
                goto error;
        }
    } else {
        virDomainObjSetState(vm, VIR_DOMAIN_PAUSED, reason);
    }

    VIR_DEBUG("Writing domain status to disk");
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0)
        goto error;

    /* Run an hook to allow admins to do some magic */
    if (virHookPresent(VIR_HOOK_DRIVER_QEMU)) {
        char *xml = qemuDomainDefFormatXML(driver, vm->def, 0);
        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)
            goto error;
    }

    virObjectUnref(logCtxt);
    VIR_FREE(seclabel);
    VIR_FREE(sec_managers);
    virObjectUnref(cfg);
    virObjectUnref(caps);

    return 0;

 exit_monitor:
    ignore_value(qemuDomainObjExitMonitor(driver, vm));
 error:
    /* 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);

    qemuMonitorClose(priv->mon);
    priv->mon = NULL;
    virObjectUnref(logCtxt);
    VIR_FREE(seclabel);
    VIR_FREE(sec_managers);
    if (seclabelgen)
        virSecurityLabelDefFree(seclabeldef);
    virDomainChrSourceDefFree(monConfig);
    virObjectUnref(cfg);
    virObjectUnref(caps);
    return -1;
}


7419
static void
7420 7421 7422 7423 7424 7425 7426 7427 7428 7429 7430 7431 7432 7433 7434 7435 7436 7437 7438 7439 7440 7441 7442
qemuProcessAutoDestroy(virDomainObjPtr dom,
                       virConnectPtr conn,
                       void *opaque)
{
    virQEMUDriverPtr driver = opaque;
    qemuDomainObjPrivatePtr priv = dom->privateData;
    virObjectEventPtr event = NULL;
    unsigned int stopFlags = 0;

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

    if (priv->job.asyncJob == QEMU_ASYNC_JOB_MIGRATION_IN)
        stopFlags |= VIR_QEMU_PROCESS_STOP_MIGRATED;

    if (priv->job.asyncJob) {
        VIR_DEBUG("vm=%s has long-term job active, cancelling",
                  dom->def->name);
        qemuDomainObjDiscardAsyncJob(driver, dom);
    }

    VIR_DEBUG("Killing domain");

    if (qemuProcessBeginStopJob(driver, dom, QEMU_JOB_DESTROY, true) < 0)
7443
        return;
7444 7445 7446 7447 7448 7449 7450 7451 7452 7453 7454

    qemuProcessStop(driver, dom, VIR_DOMAIN_SHUTOFF_DESTROYED,
                    QEMU_ASYNC_JOB_NONE, stopFlags);

    virDomainAuditStop(dom, "destroyed");
    event = virDomainEventLifecycleNewFromObj(dom,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_DESTROYED);

    qemuDomainRemoveInactive(driver, dom);

7455 7456
    qemuDomainObjEndJob(driver, dom);

7457
    virObjectEventStateQueue(driver->domainEventState, event);
7458 7459 7460 7461 7462 7463 7464 7465 7466 7467 7468 7469 7470 7471 7472 7473 7474 7475 7476 7477 7478 7479 7480 7481 7482 7483 7484 7485 7486 7487 7488 7489 7490 7491 7492 7493 7494 7495 7496 7497 7498 7499 7500 7501 7502 7503 7504 7505 7506 7507 7508 7509 7510 7511 7512 7513 7514 7515 7516 7517 7518 7519 7520 7521 7522 7523 7524 7525 7526 7527 7528 7529 7530 7531 7532 7533 7534 7535 7536 7537 7538 7539 7540
}

int qemuProcessAutoDestroyAdd(virQEMUDriverPtr driver,
                              virDomainObjPtr vm,
                              virConnectPtr conn)
{
    VIR_DEBUG("vm=%s, conn=%p", vm->def->name, conn);
    return virCloseCallbacksSet(driver->closeCallbacks, vm, conn,
                                qemuProcessAutoDestroy);
}

int qemuProcessAutoDestroyRemove(virQEMUDriverPtr driver,
                                 virDomainObjPtr vm)
{
    int ret;
    VIR_DEBUG("vm=%s", vm->def->name);
    ret = virCloseCallbacksUnset(driver->closeCallbacks, vm,
                                 qemuProcessAutoDestroy);
    return ret;
}

bool qemuProcessAutoDestroyActive(virQEMUDriverPtr driver,
                                  virDomainObjPtr vm)
{
    virCloseCallback cb;
    VIR_DEBUG("vm=%s", vm->def->name);
    cb = virCloseCallbacksGet(driver->closeCallbacks, vm, NULL);
    return cb == qemuProcessAutoDestroy;
}


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];
        qemuDomainDiskPrivatePtr diskpriv = QEMU_DOMAIN_DISK_PRIVATE(disk);
        struct qemuDomainDiskInfo *info;

        if (!(info = virHashLookup(table, disk->info.alias)))
            continue;

        if (info->removable) {
            if (info->empty)
                virDomainDiskEmptySource(disk);

            if (info->tray) {
                if (info->tray_open)
                    disk->tray_status = VIR_DOMAIN_DISK_TRAY_OPEN;
                else
                    disk->tray_status = VIR_DOMAIN_DISK_TRAY_CLOSED;
            }
        }

        /* fill in additional data */
        diskpriv->removable = info->removable;
        diskpriv->tray = info->tray;
    }

    ret = 0;

 cleanup:
    virHashFree(table);
    return ret;
}


7541 7542 7543 7544 7545
static int
qemuProcessRefreshCPU(virQEMUDriverPtr driver,
                      virDomainObjPtr vm)
{
    virCapsPtr caps = virQEMUDriverGetCapabilities(driver, false);
7546
    qemuDomainObjPrivatePtr priv = vm->privateData;
7547
    virCPUDefPtr host = NULL;
7548
    virCPUDefPtr cpu = NULL;
7549 7550 7551 7552 7553 7554 7555 7556 7557 7558 7559 7560 7561 7562 7563 7564 7565 7566 7567 7568 7569
    int ret = -1;

    if (!caps)
        return -1;

    if (!virQEMUCapsGuestIsNative(caps->host.arch, vm->def->os.arch) ||
        !caps->host.cpu ||
        !vm->def->cpu) {
        ret = 0;
        goto cleanup;
    }

    /* If the domain with a host-model CPU was started by an old libvirt
     * (< 2.3) which didn't replace the CPU with a custom one, let's do it now
     * since the rest of our code does not really expect a host-model CPU in a
     * running domain.
     */
    if (vm->def->cpu->mode == VIR_CPU_MODE_HOST_MODEL) {
        if (!(host = virCPUCopyMigratable(caps->host.cpu->arch, caps->host.cpu)))
            goto cleanup;

7570 7571 7572 7573 7574 7575 7576
        if (!(cpu = virCPUDefCopyWithoutModel(host)) ||
            virCPUDefCopyModelFilter(cpu, host, false,
                                     virQEMUCapsCPUFilterFeatures,
                                     &caps->host.cpu->arch) < 0)
            goto cleanup;

        if (virCPUUpdate(vm->def->os.arch, vm->def->cpu, cpu) < 0)
7577 7578 7579 7580
            goto cleanup;

        if (qemuProcessUpdateCPU(driver, vm, QEMU_ASYNC_JOB_NONE) < 0)
            goto cleanup;
7581 7582 7583 7584 7585 7586 7587 7588
    } else if (!virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_QUERY_CPU_MODEL_EXPANSION)) {
        /* We only try to fix CPUs when the libvirt/QEMU combo used to start
         * the domain did not know about query-cpu-model-expansion in which
         * case the host-model is known to not contain features which QEMU
         * doesn't know about.
         */
        if (qemuDomainFixupCPUs(vm, &priv->origCPU) < 0)
            goto cleanup;
7589 7590 7591 7592 7593
    }

    ret = 0;

 cleanup:
7594
    virCPUDefFree(cpu);
7595 7596 7597 7598 7599 7600
    virCPUDefFree(host);
    virObjectUnref(caps);
    return ret;
}


7601 7602 7603 7604 7605 7606 7607 7608 7609 7610 7611 7612 7613 7614 7615 7616 7617 7618 7619 7620 7621 7622 7623 7624
struct qemuProcessReconnectData {
    virQEMUDriverPtr driver;
    virDomainObjPtr obj;
};
/*
 * Open an existing VM's monitor, re-detect VCPU threads
 * and re-reserve the security labels in use
 *
 * This function also inherits a locked and ref'd domain object.
 *
 * 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.
 */
static void
qemuProcessReconnect(void *opaque)
7625
{
7626 7627 7628 7629
    struct qemuProcessReconnectData *data = opaque;
    virQEMUDriverPtr driver = data->driver;
    virDomainObjPtr obj = data->obj;
    qemuDomainObjPrivatePtr priv;
7630
    qemuDomainJobObj oldjob;
7631 7632 7633
    int state;
    int reason;
    virQEMUDriverConfigPtr cfg;
7634
    size_t i;
7635 7636
    unsigned int stopFlags = 0;
    bool jobStarted = false;
7637
    virCapsPtr caps = NULL;
7638
    bool retry = true;
7639

7640
    VIR_FREE(data);
7641

7642 7643 7644 7645 7646 7647
    qemuDomainObjRestoreJob(obj, &oldjob);
    if (oldjob.asyncJob == QEMU_ASYNC_JOB_MIGRATION_IN)
        stopFlags |= VIR_QEMU_PROCESS_STOP_MIGRATED;

    cfg = virQEMUDriverGetConfig(driver);
    priv = obj->privateData;
7648

7649
    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
7650
        goto error;
7651

7652
    if (qemuDomainObjBeginJob(driver, obj, QEMU_JOB_MODIFY) < 0)
7653
        goto error;
7654
    jobStarted = true;
7655

7656 7657 7658 7659
    /* 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)))
        goto error;
7660

7661 7662 7663
    /* Restore the masterKey */
    if (qemuDomainMasterKeyReadFile(priv) < 0)
        goto error;
7664

7665 7666 7667 7668
    /* If we are connecting to a guest started by old libvirt there is no
     * allowReboot in status XML and we need to initialize it. */
    qemuProcessPrepareAllowReboot(obj);

7669 7670 7671 7672 7673 7674
    if (priv->qemuCaps &&
        virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_CHARDEV_FD_PASS))
        retry = false;

    VIR_DEBUG("Reconnect monitor to def=%p name='%s' retry=%d",
              obj, obj->def->name, retry);
7675 7676

    /* XXX check PID liveliness & EXE path */
7677
    if (qemuConnectMonitor(driver, obj, QEMU_ASYNC_JOB_NONE, retry, NULL) < 0)
7678
        goto error;
7679

7680
    if (qemuHostdevUpdateActiveDomainDevices(driver, obj->def) < 0)
7681
        goto error;
7682

7683 7684 7685 7686
    priv->machineName = qemuDomainGetMachineName(obj);
    if (!priv->machineName)
        goto error;

7687
    if (qemuConnectCgroup(obj) < 0)
7688
        goto error;
7689

7690
    if (qemuDomainPerfRestart(obj) < 0)
7691
        goto error;
7692

7693 7694 7695 7696
    /* XXX: Need to change as long as lock is introduced for
     * qemu_driver->sharedDevices.
     */
    for (i = 0; i < obj->def->ndisks; i++) {
7697
        virDomainDiskDefPtr disk = obj->def->disks[i];
7698
        virDomainDeviceDef dev;
7699

7700
        if (virDomainDiskTranslateSourcePool(disk) < 0)
7701
            goto error;
7702

7703 7704 7705 7706 7707
        /* backing chains need to be refreshed only if they could change */
        if (priv->reconnectBlockjobs != VIR_TRISTATE_BOOL_NO) {
            /* This should be the only place that calls
             * qemuDomainDetermineDiskChain with @report_broken == false
             * to guarantee best-effort domain reconnect */
7708 7709
            virStorageSourceBackingStoreClear(disk->src);
            if (qemuDomainDetermineDiskChain(driver, obj, disk, false) < 0)
7710 7711 7712 7713
                goto error;
        } else {
            VIR_DEBUG("skipping backing chain detection for '%s'", disk->dst);
        }
7714

7715
        dev.type = VIR_DOMAIN_DEVICE_DISK;
7716
        dev.data.disk = disk;
7717
        if (qemuAddSharedDevice(driver, &dev, obj->def->name) < 0)
7718 7719 7720 7721
            goto error;
    }

    for (i = 0; i < obj->def->ngraphics; i++) {
7722
        if (qemuProcessGraphicsReservePorts(obj->def->graphics[i], true) < 0)
7723
            goto error;
7724
    }
7725

7726
    if (qemuProcessUpdateState(driver, obj) < 0)
7727 7728
        goto error;

7729 7730 7731 7732 7733 7734
    state = virDomainObjGetState(obj, &reason);
    if (state == VIR_DOMAIN_SHUTOFF ||
        (state == VIR_DOMAIN_PAUSED &&
         reason == VIR_DOMAIN_PAUSED_STARTING_UP)) {
        VIR_DEBUG("Domain '%s' wasn't fully started yet, killing it",
                  obj->def->name);
7735
        goto error;
7736
    }
7737

7738 7739 7740 7741
    /* If upgrading from old libvirtd we won't have found any
     * caps in the domain status, so re-query them
     */
    if (!priv->qemuCaps &&
7742
        !(priv->qemuCaps = virQEMUCapsCacheLookupCopy(driver->qemuCapsCache,
7743 7744
                                                      obj->def->emulator,
                                                      obj->def->os.machine)))
7745
        goto error;
7746

7747 7748 7749 7750 7751 7752 7753 7754 7755 7756 7757 7758
    /* In case the domain shutdown while we were not running,
     * we need to finish the shutdown process. And we need to do it after
     * we have virQEMUCaps filled in.
     */
    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);
        goto cleanup;
    }
7759

7760
    if (qemuProcessBuildDestroyMemoryPaths(driver, obj, NULL, true) < 0)
7761
        goto error;
7762

7763 7764
    if ((qemuDomainAssignAddresses(obj->def, priv->qemuCaps,
                                   driver, obj, false)) < 0) {
7765
        goto error;
7766
    }
7767

7768 7769 7770 7771 7772 7773
    /* if domain requests security driver we haven't loaded, report error, but
     * do not kill the domain
     */
    ignore_value(qemuSecurityCheckAllLabel(driver->securityManager,
                                           obj->def));

7774 7775
    if (qemuProcessRefreshCPU(driver, obj) < 0)
        goto error;
7776

7777
    if (qemuDomainRefreshVcpuInfo(driver, obj, QEMU_ASYNC_JOB_NONE, true) < 0)
7778
        goto error;
7779

7780
    qemuDomainVcpuPersistOrder(obj->def);
7781

7782 7783
    if (qemuSecurityReserveLabel(driver->securityManager, obj->def, obj->pid) < 0)
        goto error;
7784

7785 7786
    qemuProcessNotifyNets(obj->def);

7787
    if (qemuProcessFiltersInstantiate(obj->def, true))
7788
        goto error;
7789

7790
    if (qemuProcessRefreshDisks(driver, obj, QEMU_ASYNC_JOB_NONE) < 0)
7791
        goto error;
D
Daniel P. Berrange 已提交
7792

7793
    if (qemuBlockNodeNamesDetect(driver, obj, QEMU_ASYNC_JOB_NONE) < 0)
7794 7795
        goto error;

7796
    if (qemuRefreshVirtioChannelState(driver, obj, QEMU_ASYNC_JOB_NONE) < 0)
7797 7798
        goto error;

7799 7800 7801 7802
    /* If querying of guest's RTC failed, report error, but do not kill the domain. */
    qemuRefreshRTC(driver, obj);

    if (qemuProcessRefreshBalloonState(driver, obj, QEMU_ASYNC_JOB_NONE) < 0)
7803
        goto error;
7804

7805
    if (qemuProcessRecoverJob(driver, obj, &oldjob, &stopFlags) < 0)
7806
        goto error;
7807

7808 7809
    if (qemuProcessUpdateDevices(driver, obj) < 0)
        goto error;
7810

7811 7812 7813
    if (qemuRefreshPRManagerState(driver, obj) < 0)
        goto error;

7814 7815 7816 7817 7818
    qemuProcessReconnectCheckMemAliasOrderMismatch(obj);

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

7819 7820 7821 7822 7823 7824
    for (i = 0; i < obj->def->ncachetunes; i++) {
        if (virResctrlAllocDeterminePath(obj->def->cachetunes[i]->alloc,
                                         priv->machineName) < 0)
            goto error;
    }

7825 7826
    /* update domain state XML with possibly updated state in virDomainObj */
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, obj, driver->caps) < 0)
7827
        goto error;
7828

7829 7830
    /* Run an hook to allow admins to do some magic */
    if (virHookPresent(VIR_HOOK_DRIVER_QEMU)) {
7831
        char *xml = qemuDomainDefFormatXML(driver, obj->def, 0);
7832 7833
        int hookret;

7834 7835
        hookret = virHookCall(VIR_HOOK_DRIVER_QEMU, obj->def->name,
                              VIR_HOOK_QEMU_OP_RECONNECT, VIR_HOOK_SUBOP_BEGIN,
7836 7837 7838 7839 7840 7841 7842
                              NULL, xml, NULL);
        VIR_FREE(xml);

        /*
         * If the script raised an error abort the launch
         */
        if (hookret < 0)
7843
            goto error;
7844 7845
    }

7846 7847
    if (virAtomicIntInc(&driver->nactive) == 1 && driver->inhibitCallback)
        driver->inhibitCallback(true, driver->inhibitOpaque);
7848

7849
 cleanup:
7850 7851 7852
    if (jobStarted) {
        if (!virDomainObjIsActive(obj))
            qemuDomainRemoveInactive(driver, obj);
7853
        qemuDomainObjEndJob(driver, obj);
7854 7855 7856 7857
    } else {
        if (!virDomainObjIsActive(obj))
            qemuDomainRemoveInactiveJob(driver, obj);
    }
7858
    virDomainObjEndAPI(&obj);
7859
    virObjectUnref(cfg);
7860
    virObjectUnref(caps);
7861 7862
    virNWFilterUnlockFilterUpdates();
    return;
7863

7864 7865 7866 7867 7868
 error:
    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
7869 7870 7871 7872
         * If we couldn't get the monitor since QEMU supports
         * no-shutdown, we can safely say that the domain
         * crashed ... */
        state = VIR_DOMAIN_SHUTOFF_CRASHED;
7873 7874 7875 7876 7877
        /* 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);
7878
    }
7879
    goto cleanup;
7880
}
7881

7882 7883 7884
static int
qemuProcessReconnectHelper(virDomainObjPtr obj,
                           void *opaque)
7885
{
7886 7887 7888
    virThread thread;
    struct qemuProcessReconnectData *src = opaque;
    struct qemuProcessReconnectData *data;
7889

7890 7891 7892
    /* If the VM was inactive, we don't need to reconnect */
    if (!obj->pid)
        return 0;
7893

7894 7895
    if (VIR_ALLOC(data) < 0)
        return -1;
7896

7897 7898
    memcpy(data, src, sizeof(*data));
    data->obj = obj;
7899

7900 7901
    virNWFilterReadLockFilterUpdates();

7902 7903 7904 7905
    /* this lock and reference will be eventually transferred to the thread
     * that handles the reconnect */
    virObjectLock(obj);
    virObjectRef(obj);
7906

7907 7908 7909 7910 7911 7912 7913 7914 7915 7916 7917
    if (virThreadCreate(&thread, false, qemuProcessReconnect, data) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Could not create thread. QEMU initialization "
                         "might be incomplete"));
        /* 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);
7918
        qemuDomainRemoveInactiveJob(src->driver, obj);
7919

7920
        virDomainObjEndAPI(&obj);
7921
        virNWFilterUnlockFilterUpdates();
7922 7923
        VIR_FREE(data);
        return -1;
7924 7925
    }

7926 7927
    return 0;
}
7928

7929 7930 7931 7932 7933 7934 7935
/**
 * qemuProcessReconnectAll
 *
 * Try to re-open the resources for live VMs that we care
 * about.
 */
void
7936
qemuProcessReconnectAll(virQEMUDriverPtr driver)
7937
{
7938
    struct qemuProcessReconnectData data = {.driver = driver};
7939
    virDomainObjListForEach(driver->domains, qemuProcessReconnectHelper, &data);
7940
}