qemu_process.c 239.5 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
    disk = qemuProcessFindDomainDiskByAlias(vm, devAlias);

    if (disk) {
1056
        event = virDomainEventTrayChangeNewFromObj(vm, disk->info.alias, reason);
1057 1058 1059 1060 1061 1062
        /* 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;

1063
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0) {
1064 1065 1066
            VIR_WARN("Unable to save status on vm %s after tray moved event",
                     vm->def->name);
        }
1067 1068

        virDomainObjBroadcast(vm);
1069 1070
    }

1071
    virObjectUnlock(vm);
1072
    virObjectEventStateQueue(driver->domainEventState, event);
1073
    virObjectUnref(cfg);
1074 1075 1076
    return 0;
}

O
Osier Yang 已提交
1077 1078
static int
qemuProcessHandlePMWakeup(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
1079 1080
                          virDomainObjPtr vm,
                          void *opaque)
O
Osier Yang 已提交
1081
{
1082
    virQEMUDriverPtr driver = opaque;
1083 1084
    virObjectEventPtr event = NULL;
    virObjectEventPtr lifecycleEvent = NULL;
1085
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
O
Osier Yang 已提交
1086

1087
    virObjectLock(vm);
O
Osier Yang 已提交
1088 1089
    event = virDomainEventPMWakeupNewFromObj(vm);

1090 1091 1092 1093 1094 1095 1096 1097 1098
    /* 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);
1099
        lifecycleEvent = virDomainEventLifecycleNewFromObj(vm,
1100 1101 1102
                                                  VIR_DOMAIN_EVENT_STARTED,
                                                  VIR_DOMAIN_EVENT_STARTED_WAKEUP);

1103
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0) {
1104 1105 1106 1107 1108
            VIR_WARN("Unable to save status on vm %s after wakeup event",
                     vm->def->name);
        }
    }

1109
    virObjectUnlock(vm);
1110 1111
    virObjectEventStateQueue(driver->domainEventState, event);
    virObjectEventStateQueue(driver->domainEventState, lifecycleEvent);
1112
    virObjectUnref(cfg);
O
Osier Yang 已提交
1113 1114
    return 0;
}
1115

O
Osier Yang 已提交
1116 1117
static int
qemuProcessHandlePMSuspend(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
1118 1119
                           virDomainObjPtr vm,
                           void *opaque)
O
Osier Yang 已提交
1120
{
1121
    virQEMUDriverPtr driver = opaque;
1122 1123
    virObjectEventPtr event = NULL;
    virObjectEventPtr lifecycleEvent = NULL;
1124
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
O
Osier Yang 已提交
1125

1126
    virObjectLock(vm);
O
Osier Yang 已提交
1127 1128
    event = virDomainEventPMSuspendNewFromObj(vm);

1129
    if (virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING) {
1130
        qemuDomainObjPrivatePtr priv = vm->privateData;
1131 1132 1133 1134 1135
        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
1136
        lifecycleEvent =
1137
            virDomainEventLifecycleNewFromObj(vm,
J
Jiri Denemark 已提交
1138 1139
                                     VIR_DOMAIN_EVENT_PMSUSPENDED,
                                     VIR_DOMAIN_EVENT_PMSUSPENDED_MEMORY);
1140

1141
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0) {
1142 1143 1144
            VIR_WARN("Unable to save status on vm %s after suspend event",
                     vm->def->name);
        }
1145 1146 1147

        if (priv->agent)
            qemuAgentNotifyEvent(priv->agent, QEMU_AGENT_EVENT_SUSPEND);
1148 1149
    }

1150
    virObjectUnlock(vm);
O
Osier Yang 已提交
1151

1152 1153
    virObjectEventStateQueue(driver->domainEventState, event);
    virObjectEventStateQueue(driver->domainEventState, lifecycleEvent);
1154
    virObjectUnref(cfg);
O
Osier Yang 已提交
1155 1156 1157
    return 0;
}

1158 1159 1160
static int
qemuProcessHandleBalloonChange(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                               virDomainObjPtr vm,
1161 1162
                               unsigned long long actual,
                               void *opaque)
1163
{
1164
    virQEMUDriverPtr driver = opaque;
1165
    virObjectEventPtr event = NULL;
1166
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
1167

1168
    virObjectLock(vm);
1169 1170 1171 1172 1173 1174
    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;

1175
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0)
1176 1177
        VIR_WARN("unable to save domain status with balloon change");

1178
    virObjectUnlock(vm);
1179

1180
    virObjectEventStateQueue(driver->domainEventState, event);
1181
    virObjectUnref(cfg);
1182 1183 1184
    return 0;
}

1185 1186
static int
qemuProcessHandlePMSuspendDisk(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
1187 1188
                               virDomainObjPtr vm,
                               void *opaque)
1189
{
1190
    virQEMUDriverPtr driver = opaque;
1191 1192
    virObjectEventPtr event = NULL;
    virObjectEventPtr lifecycleEvent = NULL;
1193
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
1194

1195
    virObjectLock(vm);
1196 1197 1198 1199 1200 1201 1202 1203 1204 1205
    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 =
1206
            virDomainEventLifecycleNewFromObj(vm,
1207 1208 1209
                                     VIR_DOMAIN_EVENT_PMSUSPENDED,
                                     VIR_DOMAIN_EVENT_PMSUSPENDED_DISK);

1210
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0) {
1211 1212 1213 1214 1215 1216 1217 1218
            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);
    }

1219
    virObjectUnlock(vm);
1220

1221 1222
    virObjectEventStateQueue(driver->domainEventState, event);
    virObjectEventStateQueue(driver->domainEventState, lifecycleEvent);
1223 1224
    virObjectUnref(cfg);

1225 1226 1227
    return 0;
}

1228

1229 1230
static int
qemuProcessHandleGuestPanic(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
1231
                            virDomainObjPtr vm,
1232
                            qemuMonitorEventPanicInfoPtr info,
1233
                            void *opaque)
1234
{
1235
    virQEMUDriverPtr driver = opaque;
1236 1237 1238
    struct qemuProcessEvent *processEvent;

    virObjectLock(vm);
1239
    if (VIR_ALLOC(processEvent) < 0)
1240 1241 1242 1243
        goto cleanup;

    processEvent->eventType = QEMU_PROCESS_EVENT_GUESTPANIC;
    processEvent->action = vm->def->onCrash;
1244
    processEvent->data = info;
1245 1246 1247
    /* Hold an extra reference because we can't allow 'vm' to be
     * deleted before handling guest panic event is finished.
     */
1248 1249
    processEvent->vm = virObjectRef(vm);

1250 1251 1252
    if (virThreadPoolSendJob(driver->workerPool, 0, processEvent) < 0) {
        if (!virObjectUnref(vm))
            vm = NULL;
1253
        qemuProcessEventFree(processEvent);
1254 1255
    }

1256
 cleanup:
1257
    if (vm)
1258
        virObjectUnlock(vm);
1259 1260 1261 1262 1263

    return 0;
}


1264
int
1265 1266
qemuProcessHandleDeviceDeleted(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                               virDomainObjPtr vm,
1267 1268
                               const char *devAlias,
                               void *opaque)
1269
{
1270
    virQEMUDriverPtr driver = opaque;
1271 1272
    struct qemuProcessEvent *processEvent = NULL;
    char *data;
1273 1274 1275 1276 1277 1278

    virObjectLock(vm);

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

1279 1280
    if (qemuDomainSignalDeviceRemoval(vm, devAlias,
                                      QEMU_DOMAIN_UNPLUGGING_DEVICE_STATUS_OK))
1281
        goto cleanup;
1282

1283 1284
    if (VIR_ALLOC(processEvent) < 0)
        goto error;
1285

1286 1287 1288 1289
    processEvent->eventType = QEMU_PROCESS_EVENT_DEVICE_DELETED;
    if (VIR_STRDUP(data, devAlias) < 0)
        goto error;
    processEvent->data = data;
1290
    processEvent->vm = virObjectRef(vm);
1291

1292 1293 1294 1295
    if (virThreadPoolSendJob(driver->workerPool, 0, processEvent) < 0) {
        ignore_value(virObjectUnref(vm));
        goto error;
    }
1296

1297
 cleanup:
1298 1299
    virObjectUnlock(vm);
    return 0;
1300
 error:
1301
    qemuProcessEventFree(processEvent);
1302
    goto cleanup;
1303 1304 1305
}


1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367
/**
 *
 * 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);
1368
    virObjectEventStateQueue(driver->domainEventState, event);
1369 1370 1371 1372 1373

    return 0;
}


1374 1375 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
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);
1408
    virObjectEventStateQueue(driver->domainEventState, event);
1409 1410 1411 1412 1413

    return 0;
}


1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435
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;
1436
    processEvent->vm = virObjectRef(vm);
1437 1438 1439 1440 1441 1442 1443 1444 1445 1446

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

 cleanup:
    virObjectUnlock(vm);
    return 0;
 error:
1447
    qemuProcessEventFree(processEvent);
1448 1449 1450 1451
    goto cleanup;
}


1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475
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;
1476
    processEvent->vm = virObjectRef(vm);
1477 1478 1479 1480 1481 1482 1483 1484 1485 1486

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

 cleanup:
    virObjectUnlock(vm);
    return 0;
 error:
1487
    qemuProcessEventFree(processEvent);
1488 1489 1490 1491
    goto cleanup;
}


1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510
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;
1511
    virDomainObjBroadcast(vm);
1512 1513 1514 1515 1516 1517 1518

 cleanup:
    virObjectUnlock(vm);
    return 0;
}


1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533
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;
1534
    if (priv->job.asyncJob == QEMU_ASYNC_JOB_NONE) {
1535 1536 1537 1538
        VIR_DEBUG("got MIGRATION event without a migration job");
        goto cleanup;
    }

1539
    priv->job.current->stats.mig.status = status;
1540 1541 1542 1543 1544 1545 1546 1547
    virDomainObjBroadcast(vm);

 cleanup:
    virObjectUnlock(vm);
    return 0;
}


1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567
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;
    }

1568
    virObjectEventStateQueue(driver->domainEventState,
1569 1570 1571 1572 1573 1574 1575 1576
                         virDomainEventMigrationIterationNewFromObj(vm, pass));

 cleanup:
    virObjectUnlock(vm);
    return 0;
}


1577 1578 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
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;
}


1616 1617 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
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;
}


1670 1671
static qemuMonitorCallbacks monitorCallbacks = {
    .eofNotify = qemuProcessHandleMonitorEOF,
1672
    .errorNotify = qemuProcessHandleMonitorError,
1673
    .domainEvent = qemuProcessHandleEvent,
1674 1675
    .domainShutdown = qemuProcessHandleShutdown,
    .domainStop = qemuProcessHandleStop,
1676
    .domainResume = qemuProcessHandleResume,
1677 1678 1679 1680 1681
    .domainReset = qemuProcessHandleReset,
    .domainRTCChange = qemuProcessHandleRTCChange,
    .domainWatchdog = qemuProcessHandleWatchdog,
    .domainIOError = qemuProcessHandleIOError,
    .domainGraphics = qemuProcessHandleGraphics,
1682
    .domainBlockJob = qemuProcessHandleBlockJob,
1683
    .domainTrayChange = qemuProcessHandleTrayChange,
O
Osier Yang 已提交
1684
    .domainPMWakeup = qemuProcessHandlePMWakeup,
O
Osier Yang 已提交
1685
    .domainPMSuspend = qemuProcessHandlePMSuspend,
1686
    .domainBalloonChange = qemuProcessHandleBalloonChange,
1687
    .domainPMSuspendDisk = qemuProcessHandlePMSuspendDisk,
1688
    .domainGuestPanic = qemuProcessHandleGuestPanic,
1689
    .domainDeviceDeleted = qemuProcessHandleDeviceDeleted,
1690
    .domainNicRxFilterChanged = qemuProcessHandleNicRxFilterChanged,
1691
    .domainSerialChange = qemuProcessHandleSerialChanged,
1692
    .domainSpiceMigrated = qemuProcessHandleSpiceMigrated,
1693
    .domainMigrationStatus = qemuProcessHandleMigrationStatus,
1694
    .domainMigrationPass = qemuProcessHandleMigrationPass,
1695
    .domainAcpiOstInfo = qemuProcessHandleAcpiOstInfo,
1696
    .domainBlockThreshold = qemuProcessHandleBlockThreshold,
1697
    .domainDumpCompleted = qemuProcessHandleDumpCompleted,
1698
    .domainPRManagerStatusChanged = qemuProcessHandlePRManagerStatusChanged,
1699 1700
};

1701 1702 1703 1704 1705 1706 1707 1708 1709 1710
static void
qemuProcessMonitorReportLogError(qemuMonitorPtr mon,
                                 const char *msg,
                                 void *opaque);


static void
qemuProcessMonitorLogFree(void *opaque)
{
    qemuDomainLogContextPtr logCtxt = opaque;
1711
    virObjectUnref(logCtxt);
1712 1713
}

1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733

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;
}


1734
static int
1735
qemuConnectMonitor(virQEMUDriverPtr driver, virDomainObjPtr vm, int asyncJob,
1736
                   bool retry, qemuDomainLogContextPtr logCtxt)
1737 1738
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
1739
    qemuMonitorPtr mon = NULL;
1740
    unsigned long long timeout = 0;
1741
    virDomainChrSourceDefPtr monConfig;
1742

1743
    if (qemuSecuritySetDaemonSocketLabel(driver->securityManager, vm->def) < 0) {
1744 1745
        VIR_ERROR(_("Failed to set security context for monitor for %s"),
                  vm->def->name);
1746
        return -1;
1747 1748
    }

1749 1750 1751 1752 1753 1754
    /* 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);

1755
    /* Hold an extra reference because we can't allow 'vm' to be
C
Chen Hanxiao 已提交
1756
     * deleted until the monitor gets its own reference. */
1757
    virObjectRef(vm);
1758

1759
    ignore_value(virTimeMillisNow(&priv->monStart));
1760 1761
    monConfig = priv->monConfig;
    virObjectRef(monConfig);
1762
    virObjectUnlock(vm);
1763 1764

    mon = qemuMonitorOpen(vm,
1765
                          monConfig,
1766
                          priv->monJSON,
1767
                          retry,
1768
                          timeout,
1769 1770
                          &monitorCallbacks,
                          driver);
1771

1772
    if (mon && logCtxt) {
1773
        virObjectRef(logCtxt);
1774 1775 1776 1777 1778
        qemuMonitorSetDomainLog(mon,
                                qemuProcessMonitorReportLogError,
                                logCtxt,
                                qemuProcessMonitorLogFree);
    }
1779

1780
    virObjectLock(vm);
1781
    virObjectUnref(monConfig);
M
Michal Privoznik 已提交
1782
    virObjectUnref(vm);
1783
    priv->monStart = 0;
1784

M
Michal Privoznik 已提交
1785
    if (!virDomainObjIsActive(vm)) {
1786
        qemuMonitorClose(mon);
1787
        mon = NULL;
1788 1789 1790
    }
    priv->mon = mon;

1791
    if (qemuSecurityClearSocketLabel(driver->securityManager, vm->def) < 0) {
1792 1793
        VIR_ERROR(_("Failed to clear security context for monitor for %s"),
                  vm->def->name);
1794
        return -1;
1795 1796 1797 1798
    }

    if (priv->mon == NULL) {
        VIR_INFO("Failed to connect monitor for %s", vm->def->name);
1799
        return -1;
1800 1801
    }

1802 1803
    if (qemuProcessInitMonitor(driver, vm, asyncJob) < 0)
        return -1;
1804

1805
    if (qemuMigrationCapsCheck(driver, vm, asyncJob) < 0)
1806 1807 1808
        return -1;

    return 0;
1809 1810
}

1811 1812 1813

/**
 * qemuProcessReadLog: Read log file of a qemu VM
1814
 * @logCtxt: the domain log context
1815
 * @msg: pointer to buffer to store the read messages in
1816
 * @max: maximum length of the message returned in @msg
1817 1818
 *
 * Reads log of a qemu VM. Skips messages not produced by qemu or irrelevant
1819 1820 1821 1822
 * 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
1823
 */
1824
static int
1825 1826 1827
qemuProcessReadLog(qemuDomainLogContextPtr logCtxt,
                   char **msg,
                   size_t max)
1828
{
1829 1830
    char *buf;
    ssize_t got;
1831
    char *eol;
1832
    char *filter_next;
1833
    size_t skip;
1834

1835
    if ((got = qemuDomainLogContextRead(logCtxt, &buf)) < 0)
1836
        return -1;
1837

1838 1839 1840 1841 1842
    /* Filter out debug messages from intermediate libvirt process */
    filter_next = buf;
    while ((eol = strchr(filter_next, '\n'))) {
        *eol = '\0';
        if (virLogProbablyLogMessage(filter_next) ||
1843
            strstr(filter_next, "char device redirected to")) {
1844
            skip = (eol + 1) - filter_next;
1845
            memmove(filter_next, eol + 1, buf + got - eol);
1846 1847 1848 1849
            got -= skip;
        } else {
            filter_next = eol + 1;
            *eol = '\n';
1850 1851
        }
    }
1852
    filter_next = NULL; /* silence false coverity warning */
1853

1854 1855
    if (got > 0 &&
        buf[got - 1] == '\n') {
1856 1857
        buf[got - 1] = '\0';
        got--;
1858
    }
1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871

    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;
    }

1872
    ignore_value(VIR_REALLOC_N_QUIET(buf, got + 1));
1873 1874 1875
    *msg = buf;
    return 0;
}
1876 1877


1878 1879
static int
qemuProcessReportLogError(qemuDomainLogContextPtr logCtxt,
1880 1881 1882
                          const char *msgprefix)
{
    char *logmsg = NULL;
1883 1884 1885 1886 1887 1888
    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;
1889

1890
    if (qemuProcessReadLog(logCtxt, &logmsg, max) < 0)
1891 1892 1893
        return -1;

    virResetLastError();
1894 1895 1896 1897 1898
    if (virStringIsEmpty(logmsg))
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s", msgprefix);
    else
        virReportError(VIR_ERR_INTERNAL_ERROR, _("%s: %s"), msgprefix, logmsg);

1899 1900
    VIR_FREE(logmsg);
    return 0;
1901 1902 1903
}


1904 1905 1906 1907 1908 1909 1910 1911 1912 1913
static void
qemuProcessMonitorReportLogError(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                                 const char *msg,
                                 void *opaque)
{
    qemuDomainLogContextPtr logCtxt = opaque;
    qemuProcessReportLogError(logCtxt, msg);
}


1914
static int
1915
qemuProcessLookupPTYs(virDomainChrDefPtr *devices,
1916
                      int count,
1917
                      virHashTablePtr info)
1918
{
1919
    char *id = NULL;
1920
    size_t i;
1921
    int ret = -1;
1922

1923
    for (i = 0; i < count; i++) {
1924
        virDomainChrDefPtr chr = devices[i];
1925
        if (chr->source->type == VIR_DOMAIN_CHR_TYPE_PTY) {
1926
            qemuMonitorChardevInfoPtr entry;
1927

1928 1929
            VIR_FREE(id);
            if (virAsprintf(&id, "char%s", chr->info.alias) < 0)
1930 1931
                return -1;

1932 1933
            entry = virHashLookup(info, id);
            if (!entry || !entry->ptyPath) {
1934
                if (chr->source->data.file.path == NULL) {
1935 1936 1937
                    /* neither the log output nor 'info chardev' had a
                     * pty path for this chardev, report an error
                     */
1938 1939
                    virReportError(VIR_ERR_INTERNAL_ERROR,
                                   _("no assigned pty for device %s"), id);
1940
                    goto cleanup;
1941 1942 1943 1944 1945 1946 1947 1948
                } else {
                    /* 'info chardev' had no pty path for this chardev,
                     * but the log output had, so we're fine
                     */
                    continue;
                }
            }

1949 1950
            VIR_FREE(chr->source->data.file.path);
            if (VIR_STRDUP(chr->source->data.file.path, entry->ptyPath) < 0)
1951
                goto cleanup;
1952 1953 1954
        }
    }

1955 1956 1957 1958
    ret = 0;
 cleanup:
    VIR_FREE(id);
    return ret;
1959 1960
}

1961 1962
static int
qemuProcessFindCharDevicePTYsMonitor(virDomainObjPtr vm,
1963
                                     virHashTablePtr info)
1964
{
1965
    size_t i = 0;
C
Cole Robinson 已提交
1966

1967
    if (qemuProcessLookupPTYs(vm->def->serials, vm->def->nserials, info) < 0)
1968 1969
        return -1;

1970
    if (qemuProcessLookupPTYs(vm->def->parallels, vm->def->nparallels,
1971
                              info) < 0)
1972
        return -1;
1973

1974
    if (qemuProcessLookupPTYs(vm->def->channels, vm->def->nchannels, info) < 0)
1975
        return -1;
1976 1977 1978 1979
    /* 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];
1980

1981 1982 1983 1984 1985
        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;
1986 1987
            if (virDomainChrSourceDefCopy(chr->source,
                                          ((vm->def->serials[0])->source)) < 0)
1988 1989 1990 1991
                return -1;
        }
    }

1992
    if (qemuProcessLookupPTYs(vm->def->consoles + i, vm->def->nconsoles - i,
1993
                              info) < 0)
1994
        return -1;
1995 1996 1997 1998 1999

    return 0;
}


2000
static int
2001 2002 2003 2004
qemuProcessRefreshChannelVirtioState(virQEMUDriverPtr driver,
                                     virDomainObjPtr vm,
                                     virHashTablePtr info,
                                     int booted)
2005 2006
{
    size_t i;
2007
    int agentReason = VIR_CONNECT_DOMAIN_EVENT_AGENT_LIFECYCLE_REASON_CHANNEL;
2008
    qemuMonitorChardevInfoPtr entry;
2009
    virObjectEventPtr event = NULL;
2010 2011
    char *id = NULL;
    int ret = -1;
2012

2013 2014 2015
    if (booted)
        agentReason = VIR_CONNECT_DOMAIN_EVENT_AGENT_LIFECYCLE_REASON_DOMAIN_STARTED;

2016 2017 2018
    for (i = 0; i < vm->def->nchannels; i++) {
        virDomainChrDefPtr chr = vm->def->channels[i];
        if (chr->targetType == VIR_DOMAIN_CHR_CHANNEL_TARGET_TYPE_VIRTIO) {
2019 2020 2021 2022

            VIR_FREE(id);
            if (virAsprintf(&id, "char%s", chr->info.alias) < 0)
                goto cleanup;
2023 2024 2025 2026 2027 2028

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

2029 2030 2031 2032
            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)))
2033
                virObjectEventStateQueue(driver->domainEventState, event);
2034

2035 2036 2037 2038
            chr->state = entry->state;
        }
    }

2039 2040 2041 2042
    ret = 0;
 cleanup:
    VIR_FREE(id);
    return ret;
2043 2044 2045
}


2046 2047
int
qemuRefreshVirtioChannelState(virQEMUDriverPtr driver,
2048 2049
                              virDomainObjPtr vm,
                              qemuDomainAsyncJob asyncJob)
2050 2051 2052 2053 2054
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virHashTablePtr info = NULL;
    int ret = -1;

2055 2056 2057
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        goto cleanup;

2058
    ret = qemuMonitorGetChardevInfo(priv->mon, &info);
2059 2060
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        ret = -1;
2061 2062 2063 2064

    if (ret < 0)
        goto cleanup;

2065
    ret = qemuProcessRefreshChannelVirtioState(driver, vm, info, false);
2066 2067 2068 2069 2070 2071

 cleanup:
    virHashFree(info);
    return ret;
}

2072 2073 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

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;
}


2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152
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;
2153
    if ((then = mktime(&thenbits)) == (time_t)-1) {
2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164
        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;
}
2165

2166
int
2167 2168 2169 2170 2171 2172 2173 2174 2175
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 */
2176
    if (!virDomainDefHasMemballoon(vm->def)) {
2177
        vm->def->mem.cur_balloon = virDomainDefGetMemoryTotal(vm->def);
2178 2179 2180 2181 2182 2183 2184
        return 0;
    }

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

    rc = qemuMonitorGetBalloonInfo(qemuDomainGetMonitor(vm), &balloon);
2185
    if (qemuDomainObjExitMonitor(driver, vm) < 0 || rc < 0)
2186 2187 2188 2189 2190 2191 2192 2193
        return -1;

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

    return 0;
}


2194
static int
2195
qemuProcessWaitForMonitor(virQEMUDriverPtr driver,
C
Cole Robinson 已提交
2196
                          virDomainObjPtr vm,
2197
                          int asyncJob,
2198
                          qemuDomainLogContextPtr logCtxt)
2199 2200
{
    int ret = -1;
2201
    virHashTablePtr info = NULL;
2202 2203 2204 2205 2206 2207
    qemuDomainObjPrivatePtr priv = vm->privateData;
    bool retry = true;

    if (priv->qemuCaps &&
        virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_CHARDEV_FD_PASS))
        retry = false;
2208

2209 2210 2211 2212
    VIR_DEBUG("Connect monitor to vm=%p name='%s' retry=%d",
              vm, vm->def->name, retry);

    if (qemuConnectMonitor(driver, vm, asyncJob, retry, logCtxt) < 0)
2213 2214 2215 2216 2217 2218
        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. */
2219 2220
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        goto cleanup;
2221 2222
    ret = qemuMonitorGetChardevInfo(priv->mon, &info);
    VIR_DEBUG("qemuMonitorGetChardevInfo returned %i", ret);
2223 2224 2225
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        ret = -1;

2226
    if (ret == 0) {
2227
        if ((ret = qemuProcessFindCharDevicePTYsMonitor(vm, info)) < 0)
2228 2229
            goto cleanup;

2230 2231
        if ((ret = qemuProcessRefreshChannelVirtioState(driver, vm, info,
                                                        true)) < 0)
2232 2233
            goto cleanup;
    }
2234

2235
 cleanup:
2236
    virHashFree(info);
2237

2238 2239
    if (logCtxt && kill(vm->pid, 0) == -1 && errno == ESRCH) {
        qemuProcessReportLogError(logCtxt,
2240
                                  _("process exited while connecting to monitor"));
2241 2242 2243 2244 2245 2246
        ret = -1;
    }

    return ret;
}

2247

2248 2249 2250 2251 2252 2253
static int
qemuProcessDetectIOThreadPIDs(virQEMUDriverPtr driver,
                              virDomainObjPtr vm,
                              int asyncJob)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
2254
    qemuMonitorIOThreadInfoPtr *iothreads = NULL;
2255 2256 2257 2258
    int niothreads = 0;
    int ret = -1;
    size_t i;

2259 2260 2261 2262 2263
    if (!virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_OBJECT_IOTHREAD)) {
        ret = 0;
        goto cleanup;
    }

2264 2265 2266 2267
    /* Get the list of IOThreads from qemu */
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        goto cleanup;
    niothreads = qemuMonitorGetIOThreads(priv->mon, &iothreads);
2268 2269
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        goto cleanup;
J
John Ferlan 已提交
2270
    if (niothreads < 0)
2271 2272
        goto cleanup;

2273
    if (niothreads != vm->def->niothreadids) {
2274 2275
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("got wrong number of IOThread pids from QEMU monitor. "
2276 2277
                         "got %d, wanted %zu"),
                       niothreads, vm->def->niothreadids);
2278 2279 2280
        goto cleanup;
    }

2281 2282 2283 2284 2285 2286
    /* Nothing to do */
    if (niothreads == 0) {
        ret = 0;
        goto cleanup;
    }

2287 2288 2289
    for (i = 0; i < niothreads; i++) {
        virDomainIOThreadIDDefPtr iothrid;

2290 2291
        if (!(iothrid = virDomainIOThreadIDFind(vm->def,
                                                iothreads[i]->iothread_id))) {
2292
            virReportError(VIR_ERR_INTERNAL_ERROR,
2293 2294
                           _("iothread %d not found"),
                           iothreads[i]->iothread_id);
2295 2296 2297 2298
            goto cleanup;
        }
        iothrid->thread_id = iothreads[i]->thread_id;
    }
2299 2300 2301 2302 2303 2304

    ret = 0;

 cleanup:
    if (iothreads) {
        for (i = 0; i < niothreads; i++)
2305
            VIR_FREE(iothreads[i]);
2306 2307 2308 2309 2310
        VIR_FREE(iothreads);
    }
    return ret;
}

2311 2312 2313 2314 2315

/*
 * To be run between fork/exec of QEMU only
 */
static int
2316
qemuProcessInitCpuAffinity(virDomainObjPtr vm)
2317 2318 2319 2320
{
    int ret = -1;
    virBitmapPtr cpumap = NULL;
    virBitmapPtr cpumapToSet = NULL;
2321
    virBitmapPtr hostcpumap = NULL;
2322
    qemuDomainObjPrivatePtr priv = vm->privateData;
2323

2324 2325 2326 2327 2328 2329
    if (!vm->pid) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Cannot setup CPU affinity until process is started"));
        return -1;
    }

2330 2331
    if (vm->def->placement_mode == VIR_DOMAIN_CPU_PLACEMENT_MODE_AUTO) {
        VIR_DEBUG("Set CPU affinity with advisory nodeset from numad");
2332
        cpumapToSet = priv->autoCpuset;
2333
    } else {
2334
        VIR_DEBUG("Set CPU affinity with specified cpuset");
O
Osier Yang 已提交
2335
        if (vm->def->cpumask) {
H
Hu Tao 已提交
2336
            cpumapToSet = vm->def->cpumask;
O
Osier Yang 已提交
2337 2338 2339 2340 2341
        } 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 */
2342 2343
            int hostcpus;

2344 2345 2346 2347 2348 2349 2350 2351 2352 2353
            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;
2354
                goto cleanup;
2355 2356 2357 2358 2359
            } 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;
2360

2361 2362
                if (hostcpus > QEMUD_CPUMASK_LEN)
                    hostcpus = QEMUD_CPUMASK_LEN;
2363

2364 2365 2366
                virBitmapFree(cpumap);
                if (!(cpumap = virBitmapNew(hostcpus)))
                    goto cleanup;
2367

2368
                virBitmapSetAll(cpumap);
2369

2370 2371
                cpumapToSet = cpumap;
            }
O
Osier Yang 已提交
2372
        }
2373 2374
    }

2375
    if (virProcessSetAffinity(vm->pid, cpumapToSet) < 0)
2376
        goto cleanup;
2377

2378 2379
    ret = 0;

2380
 cleanup:
2381
    virBitmapFree(cpumap);
2382
    virBitmapFree(hostcpumap);
2383
    return ret;
2384 2385
}

2386 2387
/* set link states to down on interfaces at qemu start */
static int
2388 2389 2390
qemuProcessSetLinkStates(virQEMUDriverPtr driver,
                         virDomainObjPtr vm,
                         qemuDomainAsyncJob asyncJob)
2391 2392 2393
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virDomainDefPtr def = vm->def;
2394
    size_t i;
2395 2396 2397 2398 2399
    int ret = -1;
    int rv;

    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        return -1;
2400 2401 2402

    for (i = 0; i < def->nnets; i++) {
        if (def->nets[i]->linkstate == VIR_DOMAIN_NET_INTERFACE_LINK_STATE_DOWN) {
2403 2404 2405
            if (!def->nets[i]->info.alias) {
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("missing alias for network device"));
2406
                goto cleanup;
2407 2408
            }

2409 2410
            VIR_DEBUG("Setting link state: %s", def->nets[i]->info.alias);

2411 2412 2413 2414
            rv = qemuMonitorSetLink(priv->mon,
                                    def->nets[i]->info.alias,
                                    VIR_DOMAIN_NET_INTERFACE_LINK_STATE_DOWN);
            if (rv < 0) {
2415
                virReportError(VIR_ERR_OPERATION_FAILED,
2416 2417 2418
                               _("Couldn't set link state on interface: %s"),
                               def->nets[i]->info.alias);
                goto cleanup;
2419 2420 2421 2422
            }
        }
    }

2423 2424 2425 2426 2427
    ret = 0;

 cleanup:
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        ret = -1;
2428 2429 2430
    return ret;
}

2431

2432 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
/**
 * 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);
2533
        virCgroupFree(&cgroup);
2534 2535 2536 2537 2538 2539
    }

    return ret;
}


2540
static int
2541
qemuProcessSetupEmulator(virDomainObjPtr vm)
2542
{
2543 2544 2545 2546 2547
    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);
2548 2549
}

2550

2551 2552 2553 2554 2555 2556
static int
qemuProcessResctrlCreate(virQEMUDriverPtr driver,
                         virDomainObjPtr vm)
{
    int ret = -1;
    size_t i = 0;
2557
    virCapsPtr caps = NULL;
2558 2559
    qemuDomainObjPrivatePtr priv = vm->privateData;

B
Bing Niu 已提交
2560
    if (!vm->def->nresctrls)
2561 2562 2563 2564 2565
        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);
2566 2567 2568
    if (!caps)
        return -1;

B
Bing Niu 已提交
2569
    for (i = 0; i < vm->def->nresctrls; i++) {
2570
        if (virResctrlAllocCreate(caps->host.resctrl,
B
Bing Niu 已提交
2571
                                  vm->def->resctrls[i]->alloc,
2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582
                                  priv->machineName) < 0)
            goto cleanup;
    }

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


2583 2584 2585 2586 2587 2588 2589 2590 2591 2592
static char *
qemuProcessBuildPRHelperPidfilePath(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    const char *prdAlias = qemuDomainGetManagedPRAlias();

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


2593
void
2594
qemuProcessKillManagedPRDaemon(virDomainObjPtr vm)
2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634
{
    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;

2635 2636 2637
    if (qemuDomainNamespaceEnabled(vm, QEMU_DOMAIN_NS_MOUNT)) {
        if (virProcessGetNamespaces(vm->pid, &nfds, &fds) < 0)
            return ret;
2638

2639 2640 2641 2642
        if (nfds > 0 &&
            virProcessSetNamespaces(nfds, fds) < 0)
            goto cleanup;
    }
2643 2644 2645 2646 2647 2648 2649 2650 2651 2652

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


2653
int
2654
qemuProcessStartManagedPRDaemon(virDomainObjPtr vm)
2655 2656 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
{
    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;

2684
    if (!(socketPath = qemuDomainGetManagedPRSocketPath(priv)))
2685 2686 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
        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;
2759
    ret = 0;
2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777
 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;
}


2778
static int
2779
qemuProcessInitPasswords(virQEMUDriverPtr driver,
2780 2781
                         virDomainObjPtr vm,
                         int asyncJob)
2782 2783
{
    int ret = 0;
2784
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
2785
    size_t i;
2786

2787
    for (i = 0; i < vm->def->ngraphics; ++i) {
2788 2789
        virDomainGraphicsDefPtr graphics = vm->def->graphics[i];
        if (graphics->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC) {
2790 2791
            ret = qemuDomainChangeGraphicsPasswords(driver, vm,
                                                    VIR_DOMAIN_GRAPHICS_TYPE_VNC,
2792
                                                    &graphics->data.vnc.auth,
2793 2794
                                                    cfg->vncPassword,
                                                    asyncJob);
2795
        } else if (graphics->type == VIR_DOMAIN_GRAPHICS_TYPE_SPICE) {
2796 2797
            ret = qemuDomainChangeGraphicsPasswords(driver, vm,
                                                    VIR_DOMAIN_GRAPHICS_TYPE_SPICE,
2798
                                                    &graphics->data.spice.auth,
2799 2800
                                                    cfg->spicePassword,
                                                    asyncJob);
2801 2802
        }

2803 2804 2805
        if (ret < 0)
            goto cleanup;
    }
2806

2807
 cleanup:
2808
    virObjectUnref(cfg);
2809 2810 2811 2812 2813 2814 2815 2816 2817 2818
    return ret;
}


static int
qemuProcessPrepareChardevDevice(virDomainDefPtr def ATTRIBUTE_UNUSED,
                                virDomainChrDefPtr dev,
                                void *opaque ATTRIBUTE_UNUSED)
{
    int fd;
2819
    if (dev->source->type != VIR_DOMAIN_CHR_TYPE_FILE)
2820 2821
        return 0;

2822
    if ((fd = open(dev->source->data.file.path,
2823 2824 2825
                   O_CREAT | O_APPEND, S_IRUSR|S_IWUSR)) < 0) {
        virReportSystemError(errno,
                             _("Unable to pre-create chardev file '%s'"),
2826
                             dev->source->data.file.path);
2827 2828 2829 2830 2831 2832 2833 2834 2835
        return -1;
    }

    VIR_FORCE_CLOSE(fd);

    return 0;
}


2836 2837 2838 2839 2840
static int
qemuProcessCleanupChardevDevice(virDomainDefPtr def ATTRIBUTE_UNUSED,
                                virDomainChrDefPtr dev,
                                void *opaque ATTRIBUTE_UNUSED)
{
2841 2842 2843 2844
    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);
2845 2846 2847 2848 2849

    return 0;
}


2850 2851 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
/**
 * 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) {
2886
                if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_QXL_VGAMEM) &&
2887 2888
                    qemuMonitorUpdateVideoMemorySize(priv->mon, video,
                                                     "qxl-vga") < 0)
2889
                        goto error;
2890

2891
                if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_QXL_VRAM64) &&
2892 2893 2894
                    qemuMonitorUpdateVideoVram64Size(priv->mon, video,
                                                     "qxl-vga") < 0)
                    goto error;
2895
            } else {
2896 2897 2898 2899 2900 2901 2902 2903
                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)
2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922
                        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;
        }

    }

2923 2924
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        return -1;
2925 2926

    cfg = virQEMUDriverGetConfig(driver);
2927
    ret = virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps);
2928 2929 2930 2931 2932
    virObjectUnref(cfg);

    return ret;

 error:
2933
    ignore_value(qemuDomainObjExitMonitor(driver, vm));
2934 2935 2936 2937
    return -1;
}


2938 2939
struct qemuProcessHookData {
    virDomainObjPtr vm;
2940
    virQEMUDriverPtr driver;
2941
    virQEMUDriverConfigPtr cfg;
2942 2943 2944 2945 2946
};

static int qemuProcessHook(void *data)
{
    struct qemuProcessHookData *h = data;
2947
    qemuDomainObjPrivatePtr priv = h->vm->privateData;
2948
    int ret = -1;
2949
    int fd;
2950 2951 2952
    virBitmapPtr nodeset = NULL;
    virDomainNumatuneMemMode mode;

2953 2954 2955 2956
    /* This method cannot use any mutexes, which are not
     * protected across fork()
     */

2957
    qemuSecurityPostFork(h->driver->securityManager);
2958 2959 2960 2961 2962

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

    VIR_DEBUG("Obtaining domain lock");
2963 2964 2965 2966 2967 2968 2969
    /*
     * 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.
     */
2970
    if (qemuSecuritySetSocketLabel(h->driver->securityManager, h->vm->def) < 0)
2971
        goto cleanup;
2972
    if (virDomainLockProcessStart(h->driver->lockManager,
2973
                                  h->cfg->uri,
2974
                                  h->vm,
J
Ján Tomko 已提交
2975
                                  /* QEMU is always paused initially */
2976 2977
                                  true,
                                  &fd) < 0)
2978
        goto cleanup;
2979
    if (qemuSecurityClearSocketLabel(h->driver->securityManager, h->vm->def) < 0)
2980
        goto cleanup;
2981

2982
    if (qemuDomainBuildNamespace(h->cfg, h->driver->securityManager, h->vm) < 0)
2983 2984
        goto cleanup;

2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995
    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);
2996

2997 2998 2999
            if (virNumaSetupMemoryPolicy(mode, nodeset) < 0)
                goto cleanup;
        }
3000
    }
3001

3002 3003
    ret = 0;

3004
 cleanup:
3005
    virObjectUnref(h->cfg);
3006 3007
    VIR_DEBUG("Hook complete ret=%d", ret);
    return ret;
3008 3009 3010
}

int
3011 3012
qemuProcessPrepareMonitorChr(virDomainChrSourceDefPtr monConfig,
                             const char *domainDir)
3013 3014 3015 3016
{
    monConfig->type = VIR_DOMAIN_CHR_TYPE_UNIX;
    monConfig->data.nix.listen = true;

3017 3018
    if (virAsprintf(&monConfig->data.nix.path, "%s/monitor.sock",
                    domainDir) < 0)
3019 3020
        return -1;
    return 0;
3021 3022 3023
}


3024
/*
3025 3026
 * Precondition: vm must be locked, and a job must be active.
 * This method will call {Enter,Exit}Monitor
3027
 */
E
Eric Blake 已提交
3028
int
3029
qemuProcessStartCPUs(virQEMUDriverPtr driver, virDomainObjPtr vm,
3030
                     virDomainRunningReason reason,
3031
                     qemuDomainAsyncJob asyncJob)
3032
{
3033
    int ret = -1;
3034
    qemuDomainObjPrivatePtr priv = vm->privateData;
3035
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
3036

3037
    /* Bring up netdevs before starting CPUs */
3038
    if (qemuInterfaceStartDevices(vm->def) < 0)
3039 3040
       goto cleanup;

3041
    VIR_DEBUG("Using lock state '%s'", NULLSTR(priv->lockState));
3042
    if (virDomainLockProcessResume(driver->lockManager, cfg->uri,
3043
                                   vm, priv->lockState) < 0) {
3044 3045 3046 3047
        /* 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
         */
3048
        goto cleanup;
3049 3050 3051
    }
    VIR_FREE(priv->lockState);

3052 3053
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        goto release;
J
Jiri Denemark 已提交
3054

3055
    ret = qemuMonitorStartCPUs(priv->mon);
3056 3057
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        ret = -1;
3058 3059 3060 3061 3062

    if (ret < 0)
        goto release;

    virDomainObjSetState(vm, VIR_DOMAIN_RUNNING, reason);
3063

3064
 cleanup:
3065
    virObjectUnref(cfg);
3066
    return ret;
3067 3068 3069 3070 3071 3072

 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;
3073 3074 3075
}


3076 3077
int qemuProcessStopCPUs(virQEMUDriverPtr driver,
                        virDomainObjPtr vm,
3078
                        virDomainPausedReason reason,
3079
                        qemuDomainAsyncJob asyncJob)
3080
{
3081
    int ret = -1;
3082 3083
    qemuDomainObjPrivatePtr priv = vm->privateData;

3084
    VIR_FREE(priv->lockState);
J
Jiri Denemark 已提交
3085

3086 3087
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        goto cleanup;
J
Jiri Denemark 已提交
3088

3089
    ret = qemuMonitorStopCPUs(priv->mon);
3090 3091
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        ret = -1;
3092 3093 3094 3095

    if (ret < 0)
        goto cleanup;

3096 3097 3098
    /* de-activate netdevs after stopping CPUs */
    ignore_value(qemuInterfaceStopDevices(vm->def));

3099 3100 3101
    if (priv->job.current)
        ignore_value(virTimeMillisNow(&priv->job.current->stopped));

3102 3103 3104 3105
    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
3106

3107
 cleanup:
3108 3109 3110 3111 3112
    return ret;
}



3113
static void
3114 3115
qemuProcessNotifyNets(virDomainDefPtr def)
{
3116
    size_t i;
3117

3118 3119
    for (i = 0; i < def->nnets; i++) {
        virDomainNetDefPtr net = def->nets[i];
3120 3121 3122 3123 3124 3125 3126 3127
        /* 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));

3128
        virDomainNetNotifyActualDevice(def, net);
3129 3130 3131
    }
}

3132
static int
3133
qemuProcessFiltersInstantiate(virDomainDefPtr def, bool ignoreExists)
3134
{
3135
    size_t i;
3136

3137
    for (i = 0; i < def->nnets; i++) {
3138 3139
        virDomainNetDefPtr net = def->nets[i];
        if ((net->filter) && (net->ifname)) {
3140
            if (virDomainConfNWFilterInstantiate(def->name, def->uuid, net, ignoreExists) < 0)
J
Ján Tomko 已提交
3141
                return 1;
3142 3143 3144
        }
    }

J
Ján Tomko 已提交
3145
    return 0;
3146 3147
}

3148
static int
3149
qemuProcessUpdateState(virQEMUDriverPtr driver, virDomainObjPtr vm)
3150 3151 3152
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virDomainState state;
3153
    virDomainPausedReason reason;
3154
    virDomainState newState = VIR_DOMAIN_NOSTATE;
3155
    int oldReason;
3156
    int newReason;
3157
    bool running;
3158
    char *msg = NULL;
3159 3160
    int ret;

3161
    qemuDomainObjEnterMonitor(driver, vm);
3162
    ret = qemuMonitorGetStatus(priv->mon, &running, &reason);
3163 3164
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        return -1;
3165

3166
    if (ret < 0)
3167 3168
        return -1;

3169
    state = virDomainObjGetState(vm, &oldReason);
3170

3171 3172 3173 3174 3175 3176 3177 3178
    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) {
3179 3180
        newState = VIR_DOMAIN_RUNNING;
        newReason = VIR_DOMAIN_RUNNING_UNPAUSED;
3181
        ignore_value(VIR_STRDUP_QUIET(msg, "was unpaused"));
3182
    } else if (state == VIR_DOMAIN_RUNNING && !running) {
3183 3184 3185
        if (reason == VIR_DOMAIN_PAUSED_SHUTTING_DOWN) {
            newState = VIR_DOMAIN_SHUTDOWN;
            newReason = VIR_DOMAIN_SHUTDOWN_UNKNOWN;
3186
            ignore_value(VIR_STRDUP_QUIET(msg, "shutdown"));
3187
        } else if (reason == VIR_DOMAIN_PAUSED_CRASHED) {
3188 3189 3190
            newState = VIR_DOMAIN_CRASHED;
            newReason = VIR_DOMAIN_CRASHED_PANICKED;
            ignore_value(VIR_STRDUP_QUIET(msg, "crashed"));
3191 3192 3193
        } else {
            newState = VIR_DOMAIN_PAUSED;
            newReason = reason;
S
Stefan Berger 已提交
3194 3195
            ignore_value(virAsprintf(&msg, "was paused (%s)",
                                 virDomainPausedReasonTypeToString(reason)));
3196 3197 3198 3199 3200 3201 3202
        }
    }

    if (newState != VIR_DOMAIN_NOSTATE) {
        VIR_DEBUG("Domain %s %s while its monitor was disconnected;"
                  " changing state to %s (%s)",
                  vm->def->name,
3203
                  NULLSTR(msg),
3204 3205 3206 3207
                  virDomainStateTypeToString(newState),
                  virDomainStateReasonToString(newState, newReason));
        VIR_FREE(msg);
        virDomainObjSetState(vm, newState, newReason);
3208 3209 3210 3211 3212
    }

    return 0;
}

3213
static int
3214 3215
qemuProcessRecoverMigrationIn(virQEMUDriverPtr driver,
                              virDomainObjPtr vm,
3216
                              const qemuDomainJobObj *job,
3217
                              virDomainState state,
3218
                              int reason)
3219
{
3220 3221 3222 3223 3224
    bool postcopy = (state == VIR_DOMAIN_PAUSED &&
                     reason == VIR_DOMAIN_PAUSED_POSTCOPY_FAILED) ||
                    (state == VIR_DOMAIN_RUNNING &&
                     reason == VIR_DOMAIN_RUNNING_POSTCOPY);

3225
    switch ((qemuMigrationJobPhase) job->phase) {
3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240
    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;
3241

3242 3243 3244 3245 3246
    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);
3247
        if (qemuProcessStartCPUs(driver, vm,
3248 3249 3250 3251 3252 3253 3254 3255 3256
                                 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
3257 3258
         * we already started guest CPUs or we were in post-copy mode */
        if (postcopy) {
3259
            qemuMigrationAnyPostcopyFailed(driver, vm);
3260
        } else if (state != VIR_DOMAIN_RUNNING) {
3261
            VIR_DEBUG("Killing migrated domain %s", vm->def->name);
3262
            return -1;
3263 3264 3265
        }
        break;
    }
3266

3267 3268
    qemuMigrationParamsReset(driver, vm, QEMU_ASYNC_JOB_NONE,
                             job->migParams, job->apiFlags);
3269 3270
    return 0;
}
3271

3272 3273 3274
static int
qemuProcessRecoverMigrationOut(virQEMUDriverPtr driver,
                               virDomainObjPtr vm,
3275
                               const qemuDomainJobObj *job,
3276
                               virDomainState state,
3277 3278
                               int reason,
                               unsigned int *stopFlags)
3279
{
3280 3281 3282
    bool postcopy = state == VIR_DOMAIN_PAUSED &&
                    (reason == VIR_DOMAIN_PAUSED_POSTCOPY ||
                     reason == VIR_DOMAIN_PAUSED_POSTCOPY_FAILED);
3283
    bool resume = false;
3284

3285
    switch ((qemuMigrationJobPhase) job->phase) {
3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301
    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
3302 3303 3304 3305
         * domain; however we can only do that before migration enters
         * post-copy mode
         */
        if (postcopy) {
3306
            qemuMigrationAnyPostcopyFailed(driver, vm);
3307 3308 3309
        } else {
            VIR_DEBUG("Cancelling unfinished migration of domain %s",
                      vm->def->name);
3310
            if (qemuMigrationSrcCancel(driver, vm) < 0) {
3311 3312 3313
                VIR_WARN("Could not cancel ongoing migration of domain %s",
                         vm->def->name);
            }
3314
            resume = true;
3315
        }
3316
        break;
3317

3318 3319
    case QEMU_MIGRATION_PHASE_PERFORM3_DONE:
        /* migration finished but we didn't have a chance to get the result
3320 3321
         * 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
3322
         */
3323
        if (postcopy)
3324
            qemuMigrationAnyPostcopyFailed(driver, vm);
3325
        break;
3326

3327
    case QEMU_MIGRATION_PHASE_CONFIRM3_CANCELLED:
3328 3329 3330 3331 3332
        /* 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) {
3333
            qemuMigrationAnyPostcopyFailed(driver, vm);
3334 3335 3336
        } else {
            VIR_DEBUG("Resuming domain %s after failed migration",
                      vm->def->name);
3337
            resume = true;
3338 3339
        }
        break;
3340

3341 3342
    case QEMU_MIGRATION_PHASE_CONFIRM3:
        /* migration completed, we need to kill the domain here */
3343
        *stopFlags |= VIR_QEMU_PROCESS_STOP_MIGRATED;
3344 3345
        return -1;
    }
3346

3347 3348 3349 3350 3351 3352 3353
    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)) {
3354
            if (qemuProcessStartCPUs(driver, vm,
3355 3356 3357 3358
                                     VIR_DOMAIN_RUNNING_UNPAUSED,
                                     QEMU_ASYNC_JOB_NONE) < 0) {
                VIR_WARN("Could not resume domain %s", vm->def->name);
            }
3359 3360
        }
    }
3361

3362 3363
    qemuMigrationParamsReset(driver, vm, QEMU_ASYNC_JOB_NONE,
                             job->migParams, job->apiFlags);
3364 3365 3366
    return 0;
}

3367
static int
3368
qemuProcessRecoverJob(virQEMUDriverPtr driver,
3369
                      virDomainObjPtr vm,
3370
                      const qemuDomainJobObj *job,
3371
                      unsigned int *stopFlags)
3372
{
3373
    qemuDomainObjPrivatePtr priv = vm->privateData;
3374 3375 3376 3377 3378 3379 3380
    virDomainState state;
    int reason;

    state = virDomainObjGetState(vm, &reason);

    switch (job->asyncJob) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
3381
        if (qemuProcessRecoverMigrationOut(driver, vm, job,
3382
                                           state, reason, stopFlags) < 0)
3383 3384 3385
            return -1;
        break;

3386
    case QEMU_ASYNC_JOB_MIGRATION_IN:
3387
        if (qemuProcessRecoverMigrationIn(driver, vm, job,
3388
                                          state, reason) < 0)
3389
            return -1;
3390 3391 3392 3393
        break;

    case QEMU_ASYNC_JOB_SAVE:
    case QEMU_ASYNC_JOB_DUMP:
3394
    case QEMU_ASYNC_JOB_SNAPSHOT:
3395
        qemuDomainObjEnterMonitor(driver, vm);
3396
        ignore_value(qemuMonitorMigrateCancel(priv->mon));
3397 3398
        if (qemuDomainObjExitMonitor(driver, vm) < 0)
            return -1;
3399
        /* resume the domain but only if it was paused as a result of
3400 3401 3402 3403 3404 3405 3406 3407 3408
         * 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 &&
3409 3410
               (reason == VIR_DOMAIN_PAUSED_SNAPSHOT ||
                reason == VIR_DOMAIN_PAUSED_MIGRATION)) ||
3411
              reason == VIR_DOMAIN_PAUSED_UNKNOWN)) {
3412
             if (qemuProcessStartCPUs(driver, vm,
3413 3414 3415 3416
                                      VIR_DOMAIN_RUNNING_UNPAUSED,
                                      QEMU_ASYNC_JOB_NONE) < 0) {
                 VIR_WARN("Could not resume domain '%s' after migration to file",
                          vm->def->name);
3417 3418 3419 3420
            }
        }
        break;

3421 3422 3423 3424
    case QEMU_ASYNC_JOB_START:
        /* Already handled in VIR_DOMAIN_PAUSED_STARTING_UP check. */
        break;

3425 3426 3427 3428 3429 3430 3431 3432
    case QEMU_ASYNC_JOB_NONE:
    case QEMU_ASYNC_JOB_LAST:
        break;
    }

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

3433 3434 3435 3436
    /* 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.
     */
3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456
    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;

3457
    case QEMU_JOB_MIGRATION_OP:
3458
    case QEMU_JOB_ABORT:
3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469
    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;
}

3470 3471 3472 3473 3474 3475
static int
qemuProcessUpdateDevices(virQEMUDriverPtr driver,
                         virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virDomainDeviceDef dev;
3476
    const char **qemuDevices;
3477 3478 3479 3480 3481 3482 3483 3484 3485
    char **old;
    char **tmp;
    int ret = -1;

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

    old = priv->qemuDevices;
    priv->qemuDevices = NULL;
3486
    if (qemuDomainUpdateDeviceList(driver, vm, QEMU_ASYNC_JOB_NONE) < 0)
3487 3488
        goto cleanup;

3489
    qemuDevices = (const char **)priv->qemuDevices;
3490 3491
    if ((tmp = old)) {
        while (*tmp) {
3492
            if (!virStringListHasString(qemuDevices, *tmp) &&
3493 3494 3495 3496
                virDomainDefFindDevice(vm->def, *tmp, &dev, false) == 0 &&
                qemuDomainRemoveDevice(driver, vm, &dev) < 0) {
                goto cleanup;
            }
3497 3498 3499 3500 3501
            tmp++;
        }
    }
    ret = 0;

3502
 cleanup:
3503
    virStringListFree(old);
3504 3505 3506
    return ret;
}

3507 3508 3509 3510 3511 3512 3513
static int
qemuDomainPerfRestart(virDomainObjPtr vm)
{
    size_t i;
    virDomainDefPtr def = vm->def;
    qemuDomainObjPrivatePtr priv = vm->privateData;

3514
    if (!(priv->perf = virPerfNew()))
3515 3516 3517
        return -1;

    for (i = 0; i < VIR_PERF_EVENT_LAST; i++) {
3518 3519
        if (def->perf.events[i] &&
            def->perf.events[i] == VIR_TRISTATE_BOOL_YES) {
3520 3521 3522

            /* Failure to re-enable the perf event should not be fatal */
            if (virPerfEventEnable(priv->perf, i, vm->pid) < 0)
3523
                def->perf.events[i] = VIR_TRISTATE_BOOL_NO;
3524 3525 3526 3527 3528 3529
        }
    }

    return 0;
}

3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552

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;
        }
    }
}


3553
static bool
3554 3555
qemuProcessNeedHugepagesPath(virDomainDefPtr def,
                             virDomainMemoryDefPtr mem)
3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574
{
    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;
    }

3575 3576 3577 3578 3579 3580
    if (mem &&
        mem->model == VIR_DOMAIN_MEMORY_MODEL_DIMM &&
        mem->pagesize &&
        mem->pagesize != system_pagesize)
        return true;

3581 3582 3583 3584
    return false;
}


3585 3586 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
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;
}


3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632
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,
3633
                                           def, path, true) < 0) {
3634 3635 3636 3637 3638
            virReportError(VIR_ERR_INTERNAL_ERROR,
                            _("Unable to label %s"), path);
            return -1;
        }
    } else {
M
Michal Privoznik 已提交
3639 3640
        if (virFileDeleteTree(path) < 0)
            return -1;
3641 3642 3643 3644 3645 3646
    }

    return 0;
}


3647
int
3648 3649 3650 3651
qemuProcessBuildDestroyMemoryPaths(virQEMUDriverPtr driver,
                                   virDomainObjPtr vm,
                                   virDomainMemoryDefPtr mem,
                                   bool build)
3652 3653
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
3654
    char *path = NULL;
3655
    size_t i;
3656 3657
    bool shouldBuildHP = false;
    bool shouldBuildMB = false;
3658 3659
    int ret = -1;

3660 3661 3662 3663
    if (build) {
        shouldBuildHP = qemuProcessNeedHugepagesPath(vm->def, mem);
        shouldBuildMB = qemuProcessNeedMemoryBackingPath(vm->def, mem);
    }
3664

3665
    if (!build || shouldBuildHP) {
3666
        for (i = 0; i < cfg->nhugetlbfs; i++) {
3667
            path = qemuGetDomainHugepagePath(vm->def, &cfg->hugetlbfs[i]);
3668

3669
            if (!path)
3670 3671
                goto cleanup;

3672
            if (qemuProcessBuildDestroyMemoryPathsImpl(driver, vm->def,
3673
                                                       path, build) < 0)
3674
                goto cleanup;
3675

3676
            VIR_FREE(path);
3677 3678 3679
        }
    }

3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690
    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);
    }

3691 3692
    ret = 0;
 cleanup:
3693
    VIR_FREE(path);
3694 3695 3696 3697 3698
    virObjectUnref(cfg);
    return ret;
}


3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724
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;
}


3725 3726 3727 3728
static int
qemuProcessVNCAllocatePorts(virQEMUDriverPtr driver,
                            virDomainGraphicsDefPtr graphics,
                            bool allocate)
3729
{
3730
    unsigned short port;
3731

3732 3733 3734
    if (!allocate) {
        if (graphics->data.vnc.autoport)
            graphics->data.vnc.port = 5900;
3735

3736 3737
        return 0;
    }
3738

3739 3740 3741 3742 3743
    if (graphics->data.vnc.autoport) {
        if (virPortAllocatorAcquire(driver->remotePorts, &port) < 0)
            return -1;
        graphics->data.vnc.port = port;
    }
3744

3745 3746 3747 3748 3749 3750
    if (graphics->data.vnc.websocket == -1) {
        if (virPortAllocatorAcquire(driver->webSocketPorts, &port) < 0)
            return -1;
        graphics->data.vnc.websocket = port;
        graphics->data.vnc.websocketGenerated = true;
    }
3751

3752 3753
    return 0;
}
J
John Ferlan 已提交
3754

3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765
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;
3766

3767 3768
    bool needTLSPort = false;
    bool needPort = false;
3769

3770 3771 3772 3773 3774 3775 3776
    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;
3777

3778 3779 3780
            case VIR_DOMAIN_GRAPHICS_SPICE_CHANNEL_MODE_INSECURE:
                needPort = true;
                break;
3781

3782 3783 3784 3785 3786 3787 3788 3789 3790
            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;
3791

3792 3793 3794
        case VIR_DOMAIN_GRAPHICS_SPICE_CHANNEL_MODE_INSECURE:
            needPort = true;
            break;
3795

3796 3797 3798 3799 3800 3801 3802
        case VIR_DOMAIN_GRAPHICS_SPICE_CHANNEL_MODE_ANY:
            if (cfg->spiceTLS)
                needTLSPort = true;
            needPort = true;
            break;
        }
    }
3803

3804 3805 3806
    if (!allocate) {
        if (needPort || graphics->data.spice.port == -1)
            graphics->data.spice.port = 5901;
3807

3808 3809
        if (needTLSPort || graphics->data.spice.tlsPort == -1)
            graphics->data.spice.tlsPort = 5902;
3810

3811 3812
        ret = 0;
        goto cleanup;
3813 3814
    }

3815 3816 3817
    if (needPort || graphics->data.spice.port == -1) {
        if (virPortAllocatorAcquire(driver->remotePorts, &port) < 0)
            goto cleanup;
3818

3819 3820 3821 3822
        graphics->data.spice.port = port;

        if (!graphics->data.spice.autoport)
            graphics->data.spice.portReserved = true;
3823 3824
    }

3825 3826 3827 3828 3829 3830 3831
    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;
        }
3832

3833 3834
        if (virPortAllocatorAcquire(driver->remotePorts, &tlsPort) < 0)
            goto cleanup;
3835

3836
        graphics->data.spice.tlsPort = tlsPort;
3837

3838 3839
        if (!graphics->data.spice.autoport)
            graphics->data.spice.tlsPortReserved = true;
3840
    }
3841

3842
    ret = 0;
3843

3844 3845 3846 3847
 cleanup:
    virObjectUnref(cfg);
    return ret;
}
3848

3849

3850 3851 3852 3853 3854
static int
qemuValidateCpuCount(virDomainDefPtr def,
                     virQEMUCapsPtr qemuCaps)
{
    unsigned int maxCpus = virQEMUCapsGetMachineMaxCpus(qemuCaps, def->os.machine);
3855

3856 3857 3858 3859 3860
    if (virDomainDefGetVcpus(def) == 0) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("Domain requires at least 1 vCPU"));
        return -1;
    }
3861

3862 3863 3864 3865 3866
    if (maxCpus > 0 && virDomainDefGetVcpusMax(def) > maxCpus) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                       _("Maximum CPUs greater than specified machine type limit"));
        return -1;
    }
3867

3868 3869
    return 0;
}
3870

3871

3872 3873 3874 3875 3876 3877 3878
static int
qemuProcessVerifyHypervFeatures(virDomainDefPtr def,
                                virCPUDataPtr cpu)
{
    char *cpuFeature;
    size_t i;
    int rc;
3879

3880 3881 3882 3883
    for (i = 0; i < VIR_DOMAIN_HYPERV_LAST; i++) {
        /* always supported string property */
        if (i == VIR_DOMAIN_HYPERV_VENDOR_ID)
            continue;
3884

3885 3886
        if (def->hyperv_features[i] != VIR_TRISTATE_SWITCH_ON)
            continue;
3887

3888 3889 3890
        if (virAsprintf(&cpuFeature, "__kvm_hv_%s",
                        virDomainHypervTypeToString(i)) < 0)
            return -1;
3891

3892 3893
        rc = virCPUDataCheckFeature(cpu, cpuFeature);
        VIR_FREE(cpuFeature);
3894

3895 3896 3897 3898
        if (rc < 0)
            return -1;
        else if (rc == 1)
            continue;
3899

3900 3901 3902 3903 3904 3905 3906
        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;
3907

3908 3909 3910 3911 3912
        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:
3913
        case VIR_DOMAIN_HYPERV_FREQUENCIES:
3914
        case VIR_DOMAIN_HYPERV_REENLIGHTENMENT:
3915
        case VIR_DOMAIN_HYPERV_TLBFLUSH:
3916 3917 3918 3919
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("host doesn't support hyperv '%s' feature"),
                           virDomainHypervTypeToString(i));
            return -1;
3920

3921 3922 3923 3924
        /* coverity[dead_error_begin] */
        case VIR_DOMAIN_HYPERV_VENDOR_ID:
        case VIR_DOMAIN_HYPERV_LAST:
            break;
3925
        }
3926
    }
3927 3928

    return 0;
3929 3930
}

3931

3932
static int
3933 3934
qemuProcessVerifyKVMFeatures(virDomainDefPtr def,
                             virCPUDataPtr cpu)
3935
{
3936
    int rc = 0;
3937

3938
    if (def->features[VIR_DOMAIN_FEATURE_PVSPINLOCK] != VIR_TRISTATE_SWITCH_ON)
3939 3940
        return 0;

3941 3942 3943 3944 3945 3946
    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"));
3947
        return -1;
3948
    }
3949

3950 3951
    return 0;
}
3952 3953


3954 3955 3956 3957 3958
static int
qemuProcessVerifyCPUFeatures(virDomainDefPtr def,
                             virCPUDataPtr cpu)
{
    int rc;
S
Stefan Berger 已提交
3959

3960
    rc = virCPUCheckFeature(def->os.arch, def->cpu, "invtsc");
3961

3962
    if (rc < 0) {
3963
        return -1;
3964 3965 3966 3967 3968 3969 3970 3971 3972
    } 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;
        }
3973
    }
3974

3975
    return 0;
3976 3977 3978
}


3979
static int
3980 3981 3982 3983 3984
qemuProcessFetchGuestCPU(virQEMUDriverPtr driver,
                         virDomainObjPtr vm,
                         qemuDomainAsyncJob asyncJob,
                         virCPUDataPtr *enabled,
                         virCPUDataPtr *disabled)
3985
{
3986 3987 3988 3989
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virCPUDataPtr dataEnabled = NULL;
    virCPUDataPtr dataDisabled = NULL;
    int rc;
3990

3991 3992
    *enabled = NULL;
    *disabled = NULL;
3993

3994
    if (!ARCH_IS_X86(vm->def->os.arch))
3995 3996
        return 0;

3997 3998
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
        goto error;
3999

4000 4001
    rc = qemuMonitorGetGuestCPU(priv->mon, vm->def->os.arch,
                                &dataEnabled, &dataDisabled);
4002

4003 4004 4005 4006 4007 4008 4009 4010
    if (qemuDomainObjExitMonitor(driver, vm) < 0)
        goto error;

    if (rc == -1)
        goto error;

    *enabled = dataEnabled;
    *disabled = dataDisabled;
4011
    return 0;
4012 4013 4014 4015 4016

 error:
    virCPUDataFree(dataEnabled);
    virCPUDataFree(dataDisabled);
    return -1;
4017
}
4018

4019

4020
static int
4021 4022
qemuProcessVerifyCPU(virDomainObjPtr vm,
                     virCPUDataPtr cpu)
4023
{
4024
    virDomainDefPtr def = vm->def;
4025

4026 4027
    if (!cpu)
        return 0;
J
Jiri Denemark 已提交
4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044

    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;
}


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 4086
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;
}


4087
static int
4088
qemuProcessUpdateAndVerifyCPU(virQEMUDriverPtr driver,
4089 4090
                              virDomainObjPtr vm,
                              qemuDomainAsyncJob asyncJob)
4091
{
4092
    virCPUDataPtr cpu = NULL;
4093
    virCPUDataPtr disabled = NULL;
4094
    int ret = -1;
4095

4096 4097
    if (qemuProcessFetchGuestCPU(driver, vm, asyncJob, &cpu, &disabled) < 0)
        goto cleanup;
4098

J
Jiri Denemark 已提交
4099 4100
    if (qemuProcessVerifyCPU(vm, cpu) < 0)
        goto cleanup;
J
Ján Tomko 已提交
4101

4102 4103
    if (qemuProcessUpdateLiveGuestCPU(vm, cpu, disabled) < 0)
        goto cleanup;
4104

4105
    ret = 0;
4106

4107
 cleanup:
4108
    virCPUDataFree(cpu);
4109
    virCPUDataFree(disabled);
4110 4111 4112 4113
    return ret;
}


4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 4127 4128 4129 4130 4131 4132 4133 4134 4135 4136 4137
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;
}


4138 4139 4140 4141 4142 4143 4144
static int
qemuProcessUpdateCPU(virQEMUDriverPtr driver,
                     virDomainObjPtr vm,
                     qemuDomainAsyncJob asyncJob)
{
    virCPUDataPtr cpu = NULL;
    virCPUDataPtr disabled = NULL;
4145
    virDomainCapsCPUModelsPtr models = NULL;
4146 4147
    int ret = -1;

4148 4149 4150 4151 4152
    /* 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;

4153 4154 4155 4156 4157 4158
    if (qemuProcessFetchGuestCPU(driver, vm, asyncJob, &cpu, &disabled) < 0)
        goto cleanup;

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

4159 4160 4161 4162
    if (!(models = qemuProcessFetchCPUDefinitions(driver, vm, asyncJob)) ||
        virCPUTranslate(vm->def->os.arch, vm->def->cpu, models) < 0)
        goto cleanup;

4163 4164 4165 4166 4167
    ret = 0;

 cleanup:
    virCPUDataFree(cpu);
    virCPUDataFree(disabled);
4168
    virObjectUnref(models);
4169 4170 4171 4172
    return ret;
}


4173 4174
static int
qemuPrepareNVRAM(virQEMUDriverConfigPtr cfg,
4175
                 virDomainObjPtr vm)
4176 4177 4178 4179
{
    int ret = -1;
    int srcFD = -1;
    int dstFD = -1;
4180
    virDomainLoaderDefPtr loader = vm->def->os.loader;
4181
    bool created = false;
4182 4183
    const char *master_nvram_path;
    ssize_t r;
4184

4185
    if (!loader || !loader->nvram || virFileExists(loader->nvram))
4186 4187
        return 0;

4188 4189 4190
    master_nvram_path = loader->templt;
    if (!loader->templt) {
        size_t i;
4191 4192 4193
        for (i = 0; i < cfg->nfirmwares; i++) {
            if (STREQ(cfg->firmwares[i]->name, loader->path)) {
                master_nvram_path = cfg->firmwares[i]->nvram;
4194
                break;
4195 4196
            }
        }
4197
    }
4198

4199 4200 4201 4202 4203 4204
    if (!master_nvram_path) {
        virReportError(VIR_ERR_OPERATION_FAILED,
                       _("unable to find any master var store for "
                         "loader: %s"), loader->path);
        goto cleanup;
    }
4205

4206 4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218 4219 4220 4221 4222
    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;
4223

4224 4225
    do {
        char buf[1024];
4226

4227
        if ((r = saferead(srcFD, buf, sizeof(buf))) < 0) {
4228
            virReportSystemError(errno,
4229
                                 _("Unable to read from file '%s'"),
4230 4231 4232
                                 master_nvram_path);
            goto cleanup;
        }
4233 4234

        if (safewrite(dstFD, buf, r) < 0) {
4235
            virReportSystemError(errno,
4236
                                 _("Unable to write to file '%s'"),
4237 4238 4239
                                 loader->nvram);
            goto cleanup;
        }
4240 4241 4242 4243 4244 4245 4246 4247 4248 4249 4250 4251 4252
    } 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;
4253 4254 4255 4256 4257 4258 4259 4260 4261 4262 4263 4264 4265 4266 4267 4268 4269
    }

    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;
}


4270 4271 4272
static void
qemuLogOperation(virDomainObjPtr vm,
                 const char *msg,
4273 4274
                 virCommandPtr cmd,
                 qemuDomainLogContextPtr logCtxt)
4275 4276 4277 4278 4279
{
    char *timestamp;
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int qemuVersion = virQEMUCapsGetVersion(priv->qemuCaps);
    const char *package = virQEMUCapsGetPackage(priv->qemuCaps);
4280
    char *hostname = virGetHostname();
4281 4282 4283
    struct utsname uts;

    uname(&uts);
4284 4285

    if ((timestamp = virTimeStringNow()) == NULL)
4286
        goto cleanup;
4287

4288
    if (qemuDomainLogContextWrite(logCtxt,
4289
                                  "%s: %s %s, qemu version: %d.%d.%d%s, kernel: %s, hostname: %s\n",
4290 4291 4292 4293
                                  timestamp, msg, VIR_LOG_VERSION_STRING,
                                  (qemuVersion / 1000000) % 1000,
                                  (qemuVersion / 1000) % 1000,
                                  qemuVersion % 1000,
4294
                                  package ? package : "",
4295
                                  uts.release,
4296
                                  hostname ? hostname : "") < 0)
4297
        goto cleanup;
4298

4299 4300 4301 4302 4303
    if (cmd) {
        char *args = virCommandToString(cmd);
        qemuDomainLogContextWrite(logCtxt, "%s\n", args);
        VIR_FREE(args);
    }
4304 4305

 cleanup:
4306
    VIR_FREE(hostname);
4307 4308 4309
    VIR_FREE(timestamp);
}

4310 4311 4312 4313 4314 4315 4316

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

4317
    VIR_FREE(inc->address);
4318
    VIR_FREE(inc->launchURI);
4319
    VIR_FREE(inc->deferredURI);
4320 4321 4322 4323 4324 4325 4326 4327
    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.
4328 4329 4330
 *
 * The caller is responsible for closing @fd, calling
 * qemuProcessIncomingDefFree will NOT close it.
4331 4332 4333
 */
qemuProcessIncomingDefPtr
qemuProcessIncomingDefNew(virQEMUCapsPtr qemuCaps,
4334
                          const char *listenAddress,
4335 4336 4337 4338 4339 4340
                          const char *migrateFrom,
                          int fd,
                          const char *path)
{
    qemuProcessIncomingDefPtr inc = NULL;

4341
    if (qemuMigrationDstCheckProtocol(qemuCaps, migrateFrom) < 0)
4342 4343 4344 4345 4346
        return NULL;

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

4347 4348 4349
    if (VIR_STRDUP(inc->address, listenAddress) < 0)
        goto error;

4350
    inc->launchURI = qemuMigrationDstGetURI(migrateFrom, fd);
4351 4352 4353
    if (!inc->launchURI)
        goto error;

4354 4355 4356 4357 4358 4359
    if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_INCOMING_DEFER)) {
        inc->deferredURI = inc->launchURI;
        if (VIR_STRDUP(inc->launchURI, "defer") < 0)
            goto error;
    }

4360 4361 4362 4363 4364 4365 4366 4367 4368 4369 4370
    inc->fd = fd;
    inc->path = path;

    return inc;

 error:
    qemuProcessIncomingDefFree(inc);
    return NULL;
}


4371 4372 4373 4374 4375 4376 4377 4378
/*
 * 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,
4379
                    virDomainObjPtr vm,
4380 4381
                    virDomainJobOperation operation,
                    unsigned long apiFlags)
4382
{
4383
    if (qemuDomainObjBeginAsyncJob(driver, vm, QEMU_ASYNC_JOB_START,
4384
                                   operation, apiFlags) < 0)
4385 4386 4387 4388 4389 4390 4391 4392 4393 4394 4395 4396 4397 4398 4399
        return -1;

    qemuDomainObjSetAsyncJobMask(vm, QEMU_JOB_NONE);
    return 0;
}


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


4400 4401 4402 4403 4404 4405 4406 4407 4408 4409 4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420 4421 4422
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;
}


4423
static int
4424
qemuProcessGraphicsReservePorts(virDomainGraphicsDefPtr graphics,
4425
                                bool reconnect)
4426
{
4427 4428 4429 4430 4431 4432 4433 4434 4435 4436 4437
    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;

4438 4439
    switch (graphics->type) {
    case VIR_DOMAIN_GRAPHICS_TYPE_VNC:
4440 4441
        if (!graphics->data.vnc.autoport ||
            reconnect) {
4442
            if (virPortAllocatorSetUsed(graphics->data.vnc.port) < 0)
4443 4444 4445
                return -1;
            graphics->data.vnc.portReserved = true;
        }
4446
        if (graphics->data.vnc.websocket > 0 &&
4447
            virPortAllocatorSetUsed(graphics->data.vnc.websocket) < 0)
4448
            return -1;
4449 4450 4451
        break;

    case VIR_DOMAIN_GRAPHICS_TYPE_SPICE:
4452
        if (graphics->data.spice.autoport && !reconnect)
4453
            return 0;
4454

4455
        if (graphics->data.spice.port > 0) {
4456
            if (virPortAllocatorSetUsed(graphics->data.spice.port) < 0)
4457
                return -1;
4458 4459
            graphics->data.spice.portReserved = true;
        }
4460

4461
        if (graphics->data.spice.tlsPort > 0) {
4462
            if (virPortAllocatorSetUsed(graphics->data.spice.tlsPort) < 0)
4463 4464
                return -1;
            graphics->data.spice.tlsPortReserved = true;
4465
        }
4466 4467 4468 4469 4470
        break;

    case VIR_DOMAIN_GRAPHICS_TYPE_SDL:
    case VIR_DOMAIN_GRAPHICS_TYPE_RDP:
    case VIR_DOMAIN_GRAPHICS_TYPE_DESKTOP:
4471
    case VIR_DOMAIN_GRAPHICS_TYPE_EGL_HEADLESS:
4472 4473
    case VIR_DOMAIN_GRAPHICS_TYPE_LAST:
        break;
4474 4475
    }

4476 4477 4478 4479
    return 0;
}


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 4509
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:
4510
    case VIR_DOMAIN_GRAPHICS_TYPE_EGL_HEADLESS:
4511 4512 4513 4514 4515 4516 4517
    case VIR_DOMAIN_GRAPHICS_TYPE_LAST:
        break;
    }

    return 0;
}

4518
static int
4519
qemuProcessGetNetworkAddress(const char *netname,
4520 4521
                             char **netaddr)
{
4522
    virConnectPtr conn = NULL;
4523 4524 4525 4526 4527 4528 4529 4530 4531 4532
    int ret = -1;
    virNetworkPtr net;
    virNetworkDefPtr netdef = NULL;
    virNetworkIPDefPtr ipdef;
    virSocketAddr addr;
    virSocketAddrPtr addrptr = NULL;
    char *dev_name = NULL;
    char *xml = NULL;

    *netaddr = NULL;
4533 4534 4535 4536

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

4537 4538 4539 4540 4541 4542 4543 4544 4545 4546 4547 4548
    net = virNetworkLookupByName(conn, netname);
    if (!net)
        goto cleanup;

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

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

4549
    switch ((virNetworkForwardType) netdef->forward.type) {
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 4585
    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;
4586 4587 4588 4589 4590 4591 4592 4593

    case VIR_NETWORK_FORWARD_HOSTDEV:
        break;

    case VIR_NETWORK_FORWARD_LAST:
    default:
        virReportEnumRangeError(virNetworkForwardType, netdef->forward.type);
        goto cleanup;
4594 4595 4596 4597 4598 4599 4600 4601 4602 4603 4604 4605 4606 4607 4608 4609 4610
    }

    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);
4611
    virObjectUnref(conn);
4612 4613 4614 4615
    VIR_FREE(xml);
    return ret;
}

4616

4617
static int
4618
qemuProcessGraphicsSetupNetworkAddress(virDomainGraphicsListenDefPtr glisten,
4619 4620 4621 4622 4623 4624 4625 4626 4627 4628 4629
                                       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;
    }

4630
    rc = qemuProcessGetNetworkAddress(glisten->network, &glisten->address);
4631 4632 4633 4634 4635 4636 4637 4638 4639 4640 4641 4642 4643
    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;
}


4644
static int
4645
qemuProcessGraphicsSetupListen(virQEMUDriverPtr driver,
4646 4647
                               virDomainGraphicsDefPtr graphics,
                               virDomainObjPtr vm)
4648
{
4649
    qemuDomainObjPrivatePtr priv = vm->privateData;
4650
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
4651
    const char *type = virDomainGraphicsTypeToString(graphics->type);
4652
    char *listenAddr = NULL;
4653
    bool useSocket = false;
4654
    size_t i;
4655
    int ret = -1;
4656 4657 4658

    switch (graphics->type) {
    case VIR_DOMAIN_GRAPHICS_TYPE_VNC:
4659
        useSocket = cfg->vncAutoUnixSocket;
4660 4661 4662 4663
        listenAddr = cfg->vncListen;
        break;

    case VIR_DOMAIN_GRAPHICS_TYPE_SPICE:
4664
        useSocket = cfg->spiceAutoUnixSocket;
4665 4666 4667 4668 4669 4670
        listenAddr = cfg->spiceListen;
        break;

    case VIR_DOMAIN_GRAPHICS_TYPE_SDL:
    case VIR_DOMAIN_GRAPHICS_TYPE_RDP:
    case VIR_DOMAIN_GRAPHICS_TYPE_DESKTOP:
4671
    case VIR_DOMAIN_GRAPHICS_TYPE_EGL_HEADLESS:
4672 4673 4674 4675 4676 4677 4678 4679 4680
    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:
4681 4682 4683 4684 4685
            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) {
4686 4687
                    memset(glisten, 0, sizeof(virDomainGraphicsListenDef));
                    if (virAsprintf(&glisten->socket, "%s/%s.sock",
4688
                                    priv->libDir, type) < 0)
4689
                        goto cleanup;
4690 4691
                    glisten->fromConfig = true;
                    glisten->type = VIR_DOMAIN_GRAPHICS_LISTEN_TYPE_SOCKET;
4692 4693
                } else if (listenAddr) {
                    if (VIR_STRDUP(glisten->address, listenAddr) < 0)
4694
                        goto cleanup;
4695 4696 4697
                    glisten->fromConfig = true;
                }
            }
4698 4699 4700 4701 4702 4703
            break;

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

4704
            if (qemuProcessGraphicsSetupNetworkAddress(glisten,
4705
                                                       listenAddr) < 0)
4706
                goto cleanup;
4707 4708
            break;

4709 4710 4711 4712
        case VIR_DOMAIN_GRAPHICS_LISTEN_TYPE_SOCKET:
            if (!glisten->socket) {
                if (virAsprintf(&glisten->socket, "%s/%s.sock",
                                priv->libDir, type) < 0)
4713
                    goto cleanup;
4714 4715 4716 4717
                glisten->autoGenerated = true;
            }
            break;

4718 4719 4720 4721 4722 4723
        case VIR_DOMAIN_GRAPHICS_LISTEN_TYPE_NONE:
        case VIR_DOMAIN_GRAPHICS_LISTEN_TYPE_LAST:
            break;
        }
    }

4724 4725 4726 4727 4728
    ret = 0;

 cleanup:
    virObjectUnref(cfg);
    return ret;
4729 4730 4731
}


4732
static int
4733
qemuProcessSetupGraphics(virQEMUDriverPtr driver,
4734 4735
                         virDomainObjPtr vm,
                         unsigned int flags)
4736
{
4737
    virDomainGraphicsDefPtr graphics;
4738
    bool allocate = !(flags & VIR_QEMU_PROCESS_START_PRETEND);
4739
    size_t i;
4740 4741
    int ret = -1;

4742 4743 4744
    for (i = 0; i < vm->def->ngraphics; i++) {
        graphics = vm->def->graphics[i];

4745
        if (qemuProcessGraphicsSetupListen(driver, graphics, vm) < 0)
4746 4747 4748
            goto cleanup;
    }

4749 4750 4751 4752
    if (allocate) {
        for (i = 0; i < vm->def->ngraphics; i++) {
            graphics = vm->def->graphics[i];

4753
            if (qemuProcessGraphicsReservePorts(graphics, false) < 0)
4754 4755 4756
                goto cleanup;
        }
    }
4757

4758
    for (i = 0; i < vm->def->ngraphics; ++i) {
4759
        graphics = vm->def->graphics[i];
4760

4761 4762
        if (qemuProcessGraphicsAllocatePorts(driver, graphics, allocate) < 0)
            goto cleanup;
4763 4764 4765 4766 4767 4768 4769 4770 4771
    }

    ret = 0;

 cleanup:
    return ret;
}


4772 4773 4774 4775 4776 4777 4778 4779 4780 4781 4782 4783 4784 4785 4786 4787 4788 4789 4790 4791 4792 4793 4794 4795 4796 4797 4798 4799 4800 4801 4802 4803 4804
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++) {
4805
            if (!virHostdevIsSCSIDevice(vm->def->hostdevs[i]))
4806 4807
                continue;

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 4833
            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;
}


4834 4835 4836 4837 4838 4839 4840 4841 4842
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;

4843
    if (!virDomainDefHasMemballoon(vm->def))
4844 4845 4846
        return 0;

    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
4847
        return -1;
4848

4849 4850 4851
    if (vm->def->memballoon->period)
        qemuMonitorSetMemoryStatsPeriod(priv->mon, vm->def->memballoon,
                                        vm->def->memballoon->period);
4852 4853 4854 4855 4856 4857 4858 4859 4860 4861 4862 4863
    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
4864 4865 4866
static int
qemuProcessMakeDir(virQEMUDriverPtr driver,
                   virDomainObjPtr vm,
4867
                   const char *path)
J
Jiri Denemark 已提交
4868 4869 4870 4871 4872 4873 4874 4875
{
    int ret = -1;

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

4876
    if (qemuSecurityDomainSetPathLabel(driver->securityManager,
4877
                                       vm->def, path, true) < 0)
J
Jiri Denemark 已提交
4878 4879 4880 4881 4882 4883 4884 4885 4886
        goto cleanup;

    ret = 0;

 cleanup:
    return ret;
}


4887 4888 4889 4890 4891 4892 4893 4894 4895 4896 4897 4898 4899 4900 4901 4902 4903 4904 4905 4906 4907 4908 4909 4910 4911 4912 4913 4914 4915 4916 4917 4918 4919 4920 4921 4922
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) ==
4923
                VIR_DOMAIN_MEMORY_ACCESS_SHARED) {
4924 4925 4926 4927 4928 4929 4930 4931 4932 4933 4934 4935
                shmem = true;
                break;
            }
        }
    }

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

4936 4937 4938 4939 4940 4941 4942 4943 4944 4945 4946 4947 4948 4949 4950 4951 4952 4953 4954 4955 4956 4957 4958

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:
4959
        case VIR_DOMAIN_GRAPHICS_TYPE_EGL_HEADLESS:
4960 4961 4962 4963 4964 4965 4966 4967 4968
        case VIR_DOMAIN_GRAPHICS_TYPE_LAST:
            break;
        }
    }

    return 0;
}


4969 4970 4971 4972 4973 4974 4975 4976 4977 4978
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];

4979 4980 4981 4982 4983 4984 4985 4986 4987
        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 &&
4988 4989 4990 4991
             !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))) {
4992 4993 4994 4995 4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009 5010 5011 5012 5013
            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;
}


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 5044
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;
}


5045
static int
5046 5047 5048 5049 5050 5051 5052 5053 5054 5055 5056 5057 5058 5059 5060 5061 5062 5063 5064
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;
}


5065 5066 5067 5068 5069 5070 5071
static int
qemuProcessStartValidateDisks(virDomainObjPtr vm,
                              virQEMUCapsPtr qemuCaps)
{
    size_t i;

    for (i = 0; i < vm->def->ndisks; i++) {
5072 5073
        virDomainDiskDefPtr disk = vm->def->disks[i];
        virStorageSourcePtr src = disk->src;
5074 5075 5076 5077 5078 5079 5080 5081 5082 5083 5084 5085

        /* 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;
        }
5086 5087 5088 5089 5090 5091 5092 5093

        /* 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;
        }
5094 5095 5096 5097 5098 5099
    }

    return 0;
}


5100
static int
5101 5102
qemuProcessStartValidateXML(virQEMUDriverPtr driver,
                            virDomainObjPtr vm,
5103
                            virQEMUCapsPtr qemuCaps,
5104
                            virCapsPtr caps,
5105
                            unsigned int flags)
5106 5107 5108 5109 5110 5111 5112 5113 5114 5115 5116 5117
{
    /* 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;

5118 5119 5120
    /* 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 */
5121 5122 5123
    if ((flags & VIR_QEMU_PROCESS_START_NEW) &&
        virDomainDefValidate(vm->def, caps, 0, driver->xmlopt) < 0)
        return -1;
5124 5125 5126

    return 0;
}
5127

5128

5129 5130 5131 5132 5133 5134
/**
 * qemuProcessStartValidate:
 * @vm: domain object
 * @qemuCaps: emulator capabilities
 * @migration: restoration of existing state
 *
5135 5136 5137 5138 5139
 * 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.
5140
 */
5141
static int
5142 5143
qemuProcessStartValidate(virQEMUDriverPtr driver,
                         virDomainObjPtr vm,
5144
                         virQEMUCapsPtr qemuCaps,
5145
                         virCapsPtr caps,
5146
                         unsigned int flags)
5147
{
5148 5149 5150 5151 5152 5153 5154 5155 5156 5157 5158 5159 5160 5161
    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");
5162
        if (qemuSecurityCheckAllLabel(driver->securityManager, vm->def) < 0)
5163 5164 5165 5166
            return -1;

    }

5167
    if (qemuProcessStartValidateXML(driver, vm, qemuCaps, caps, flags) < 0)
5168 5169
        return -1;

5170 5171 5172
    if (qemuProcessStartValidateGraphics(vm) < 0)
        return -1;

5173 5174 5175
    if (qemuProcessStartValidateVideo(vm, qemuCaps) < 0)
        return -1;

5176 5177 5178
    if (qemuProcessStartValidateIOThreads(vm, qemuCaps) < 0)
        return -1;

5179 5180 5181
    if (qemuProcessStartValidateShmem(vm) < 0)
        return -1;

5182 5183 5184 5185
    if (vm->def->cpu &&
        virCPUValidateFeatures(vm->def->os.arch, vm->def->cpu) < 0)
        return -1;

5186 5187 5188
    if (qemuProcessStartValidateDisks(vm, qemuCaps) < 0)
        return -1;

5189 5190
    VIR_DEBUG("Checking for any possible (non-fatal) issues");

5191
    qemuProcessStartWarnShmem(vm);
5192

5193
    return 0;
5194 5195 5196
}


J
Jiri Denemark 已提交
5197 5198 5199 5200 5201 5202
/**
 * qemuProcessInit:
 *
 * Prepares the domain up to the point when priv->qemuCaps is initialized. The
 * function calls qemuProcessStop when needed.
 *
5203 5204 5205 5206
 * 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
5207 5208 5209 5210 5211
 * Returns 0 on success, -1 on error.
 */
int
qemuProcessInit(virQEMUDriverPtr driver,
                virDomainObjPtr vm,
5212
                virCPUDefPtr updatedCPU,
5213
                qemuDomainAsyncJob asyncJob,
5214
                bool migration,
5215
                unsigned int flags)
J
Jiri Denemark 已提交
5216 5217 5218 5219 5220
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    virCapsPtr caps = NULL;
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int stopFlags;
5221
    virCPUDefPtr origCPU = NULL;
J
Jiri Denemark 已提交
5222 5223 5224 5225 5226 5227 5228 5229 5230 5231 5232 5233 5234 5235
    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)))
5236 5237
        goto cleanup;

5238 5239 5240 5241 5242 5243 5244 5245 5246
    /* 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;
    }

5247 5248
    VIR_DEBUG("Determining emulator version");
    virObjectUnref(priv->qemuCaps);
5249
    if (!(priv->qemuCaps = virQEMUCapsCacheLookupCopy(driver->qemuCapsCache,
5250 5251 5252
                                                      vm->def->emulator,
                                                      vm->def->os.machine)))
        goto cleanup;
J
Jiri Denemark 已提交
5253

5254 5255 5256
    if (flags & VIR_QEMU_PROCESS_START_STANDALONE)
        virQEMUCapsClear(priv->qemuCaps, QEMU_CAPS_CHARDEV_FD_PASS);

5257 5258 5259
    if (qemuDomainUpdateCPU(vm, updatedCPU, &origCPU) < 0)
        goto cleanup;

5260
    if (qemuProcessStartValidate(driver, vm, priv->qemuCaps, caps, flags) < 0)
5261 5262
        goto cleanup;

J
Jiri Denemark 已提交
5263 5264 5265 5266 5267
    /* 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");
5268
    if (virDomainObjSetDefTransient(caps, driver->xmlopt, vm) < 0)
5269
        goto cleanup;
J
Jiri Denemark 已提交
5270

5271
    if (flags & VIR_QEMU_PROCESS_START_PRETEND) {
5272 5273 5274 5275
        if (qemuDomainSetPrivatePaths(driver, vm) < 0) {
            virDomainObjRemoveTransientDef(vm);
            goto cleanup;
        }
5276
    } else {
5277 5278 5279
        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
5280

5281 5282
        if (virAtomicIntInc(&driver->nactive) == 1 && driver->inhibitCallback)
            driver->inhibitCallback(true, driver->inhibitOpaque);
J
Jiri Denemark 已提交
5283

5284 5285 5286 5287 5288
        /* 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
5289

5290 5291
        if (qemuDomainSetPrivatePaths(driver, vm) < 0)
            goto stop;
5292 5293

        VIR_STEAL_PTR(priv->origCPU, origCPU);
5294
    }
5295

J
Jiri Denemark 已提交
5296 5297 5298
    ret = 0;

 cleanup:
5299
    virCPUDefFree(origCPU);
J
Jiri Denemark 已提交
5300 5301 5302 5303 5304 5305 5306 5307
    virObjectUnref(cfg);
    virObjectUnref(caps);
    return ret;

 stop:
    stopFlags = VIR_QEMU_PROCESS_STOP_NO_RELABEL;
    if (migration)
        stopFlags |= VIR_QEMU_PROCESS_STOP_MIGRATED;
5308
    qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_FAILED, asyncJob, stopFlags);
J
Jiri Denemark 已提交
5309 5310 5311 5312
    goto cleanup;
}


5313 5314 5315 5316 5317 5318 5319 5320 5321 5322 5323
/**
 * qemuProcessNetworkPrepareDevices
 */
static int
qemuProcessNetworkPrepareDevices(virDomainDefPtr def)
{
    int ret = -1;
    size_t i;

    for (i = 0; i < def->nnets; i++) {
        virDomainNetDefPtr net = def->nets[i];
5324
        virDomainNetType actualType;
5325 5326 5327 5328 5329

        /* 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.
         */
5330
        if (virDomainNetAllocateActualDevice(def, net) < 0)
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 5365
            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;
}


5366 5367 5368 5369 5370 5371 5372 5373 5374 5375 5376 5377 5378 5379 5380 5381
/**
 * 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);
5382
    virDomainVcpuDefPtr vcpu = virDomainDefGetVcpu(vm->def, vcpuid);
5383 5384 5385 5386 5387 5388 5389 5390 5391
    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;

B
Bing Niu 已提交
5392 5393
    for (i = 0; i < vm->def->nresctrls; i++) {
        virDomainResctrlDefPtr ct = vm->def->resctrls[i];
5394 5395 5396 5397 5398 5399 5400

        if (virBitmapIsBitSet(ct->vcpus, vcpuid)) {
            if (virResctrlAllocAddPID(ct->alloc, vcpupid) < 0)
                return -1;
            break;
        }
    }
5401

5402
    return 0;
5403 5404 5405 5406 5407 5408
}


static int
qemuProcessSetupVcpus(virDomainObjPtr vm)
{
5409
    virDomainVcpuDefPtr vcpu;
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 5454
    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;
}


5455 5456 5457 5458 5459
int
qemuProcessSetupIOThread(virDomainObjPtr vm,
                         virDomainIOThreadIDDefPtr iothread)
{

5460 5461
    return qemuProcessSetupPid(vm, iothread->thread_id,
                               VIR_CGROUP_THREAD_IOTHREAD,
5462
                               iothread->iothread_id,
5463
                               iothread->cpumask,
5464 5465
                               vm->def->cputune.iothread_period,
                               vm->def->cputune.iothread_quota,
5466
                               &iothread->sched);
5467 5468 5469 5470 5471 5472 5473 5474 5475 5476 5477 5478 5479 5480 5481 5482 5483 5484 5485
}


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;
}


5486 5487 5488 5489 5490 5491 5492 5493 5494 5495 5496 5497
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;

5498
    if (!(ordermap = virBitmapNew(maxvcpus + 1)))
5499 5500 5501 5502 5503 5504 5505 5506 5507 5508 5509 5510 5511 5512 5513 5514
        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) {
5515
            if (virBitmapIsBitSet(ordermap, vcpu->order)) {
5516
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
5517
                               _("duplicate vcpu order '%u'"), vcpu->order);
5518 5519 5520
                goto cleanup;
            }

5521 5522 5523 5524 5525 5526
            if (virBitmapSetBit(ordermap, vcpu->order)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("vcpu order '%u' exceeds vcpu count"),
                               vcpu->order);
                goto cleanup;
            }
5527 5528 5529 5530 5531 5532 5533 5534 5535 5536 5537 5538 5539 5540 5541 5542
        }

        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 &&
5543
                 vcpupriv->thread_id == -1 && vcpupriv->node_id == -1) ||
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 5593
                !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);
5594 5595
    qemuDomainObjPrivatePtr priv = vm->privateData;
    qemuCgroupEmulatorAllNodesDataPtr emulatorCgroup = NULL;
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 5627
    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);

5628 5629 5630
    if (qemuCgroupEmulatorAllNodesAllow(priv->cgroup, &emulatorCgroup) < 0)
        goto cleanup;

5631 5632 5633 5634 5635 5636 5637 5638 5639 5640 5641 5642 5643 5644 5645 5646 5647 5648 5649 5650 5651 5652 5653 5654
    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:
5655
    qemuCgroupEmulatorAllNodesRestore(emulatorCgroup);
5656 5657 5658 5659 5660 5661
    VIR_FREE(bootHotplug);
    virJSONValueFree(vcpuprops);
    return ret;
}


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 5700
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;
    }

5701 5702 5703
    if (virCPUConvertLegacy(caps->host.arch, def->cpu) < 0)
        return -1;

5704 5705 5706 5707
    /* nothing to update for host-passthrough */
    if (def->cpu->mode == VIR_CPU_MODE_HOST_PASSTHROUGH)
        return 0;

5708 5709
    if (def->cpu->check == VIR_CPU_CHECK_PARTIAL &&
        virCPUCompare(caps->host.arch,
5710
                      virQEMUCapsGetHostModel(qemuCaps, def->virtType,
5711
                                              VIR_QEMU_CAPS_HOST_CPU_FULL),
5712 5713
                      def->cpu, true) < 0)
        return -1;
5714 5715

    if (virCPUUpdate(def->os.arch, def->cpu,
5716
                     virQEMUCapsGetHostModel(qemuCaps, def->virtType,
5717
                                             VIR_QEMU_CAPS_HOST_CPU_MIGRATABLE)) < 0)
5718 5719
        goto cleanup;

5720 5721
    if (virCPUTranslate(def->os.arch, def->cpu,
                        virQEMUCapsGetCPUDefinitions(qemuCaps, def->virtType)) < 0)
5722 5723 5724 5725 5726 5727 5728 5729 5730 5731
        goto cleanup;

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

 cleanup:
    return ret;
}


5732 5733 5734 5735 5736 5737
static int
qemuProcessPrepareDomainNUMAPlacement(virDomainObjPtr vm,
                                      virCapsPtr caps)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    char *nodeset = NULL;
5738 5739
    virBitmapPtr numadNodeset = NULL;
    virBitmapPtr hostMemoryNodeset = NULL;
5740 5741 5742 5743 5744 5745 5746 5747 5748 5749 5750 5751 5752 5753
    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;

5754 5755 5756
    if (!(hostMemoryNodeset = virNumaGetHostMemoryNodeset()))
        goto cleanup;

5757 5758
    VIR_DEBUG("Nodeset returned from numad: %s", nodeset);

5759
    if (virBitmapParse(nodeset, &numadNodeset, VIR_DOMAIN_CPUMASK_LEN) < 0)
5760 5761
        goto cleanup;

5762 5763 5764 5765
    /* 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)))
5766 5767
        goto cleanup;

5768 5769 5770 5771
    virBitmapIntersect(numadNodeset, hostMemoryNodeset);

    VIR_STEAL_PTR(priv->autoNodeset, numadNodeset);

5772 5773 5774 5775
    ret = 0;

 cleanup:
    VIR_FREE(nodeset);
5776 5777
    virBitmapFree(numadNodeset);
    virBitmapFree(hostMemoryNodeset);
5778 5779 5780 5781
    return ret;
}


5782
static int
5783
qemuProcessPrepareDomainStorage(virQEMUDriverPtr driver,
5784
                                virDomainObjPtr vm,
5785
                                qemuDomainObjPrivatePtr priv,
5786
                                virQEMUDriverConfigPtr cfg,
5787 5788 5789 5790 5791 5792 5793 5794 5795
                                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];

5796
        if (virDomainDiskTranslateSourcePool(disk) < 0) {
5797 5798 5799 5800 5801 5802
            if (qemuDomainCheckDiskStartupPolicy(driver, vm, idx, cold_boot) < 0)
                return -1;

            /* disk source was dropped */
            continue;
        }
5803

5804
        if (qemuDomainPrepareDiskSource(disk, priv, cfg) < 0)
5805
            return -1;
5806 5807 5808 5809 5810 5811
    }

    return 0;
}


5812 5813 5814 5815 5816 5817 5818 5819 5820 5821 5822 5823 5824 5825 5826 5827 5828 5829 5830 5831
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;
    }
}


5832
/**
5833 5834 5835 5836
 * qemuProcessPrepareDomain:
 * @driver: qemu driver
 * @vm: domain object
 * @flags: qemuProcessStartFlags
5837 5838 5839 5840 5841 5842 5843 5844 5845 5846 5847
 *
 * 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
5848
qemuProcessPrepareDomain(virQEMUDriverPtr driver,
5849 5850 5851 5852 5853 5854
                         virDomainObjPtr vm,
                         unsigned int flags)
{
    int ret = -1;
    size_t i;
    qemuDomainObjPrivatePtr priv = vm->privateData;
5855
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
5856 5857 5858 5859 5860
    virCapsPtr caps;

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

5861 5862 5863 5864
    priv->machineName = qemuDomainGetMachineName(vm);
    if (!priv->machineName)
        goto cleanup;

5865 5866 5867 5868
    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)");
5869
        if (qemuSecurityGenLabel(driver->securityManager, vm->def) < 0) {
5870 5871 5872 5873 5874
            virDomainAuditSecurityLabel(vm, false);
            goto cleanup;
        }
        virDomainAuditSecurityLabel(vm, true);

5875 5876
        if (qemuProcessPrepareDomainNUMAPlacement(vm, caps) < 0)
            goto cleanup;
5877 5878
    }

5879 5880 5881 5882 5883 5884 5885
    /* 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;
    }

5886 5887
    qemuProcessPrepareAllowReboot(vm);

5888 5889 5890 5891 5892 5893 5894
    /*
     * 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
     */
5895
    VIR_DEBUG("Assigning domain PCI addresses");
5896 5897
    if ((qemuDomainAssignAddresses(vm->def, priv->qemuCaps, driver, vm,
                                   !!(flags & VIR_QEMU_PROCESS_START_NEW))) < 0) {
5898
        goto cleanup;
5899
    }
5900

5901 5902 5903
    if (qemuAssignDeviceAliases(vm->def, priv->qemuCaps) < 0)
        goto cleanup;

5904
    VIR_DEBUG("Setting graphics devices");
5905
    if (qemuProcessSetupGraphics(driver, vm, flags) < 0)
5906 5907
        goto cleanup;

5908 5909 5910 5911
    VIR_DEBUG("Create domain masterKey");
    if (qemuDomainMasterKeyCreate(vm) < 0)
        goto cleanup;

5912
    VIR_DEBUG("Setting up storage");
5913
    if (qemuProcessPrepareDomainStorage(driver, vm, priv, cfg, flags) < 0)
5914 5915
        goto cleanup;

5916
    VIR_DEBUG("Prepare chardev source backends for TLS");
5917
    qemuDomainPrepareChardevSource(vm->def, cfg);
5918

5919
    VIR_DEBUG("Add secrets to hostdevs and chardevs");
5920
    if (qemuDomainSecretPrepare(driver, vm) < 0)
5921 5922
        goto cleanup;

5923 5924 5925 5926 5927 5928
    for (i = 0; i < vm->def->nchannels; i++) {
        if (qemuDomainPrepareChannel(vm->def->channels[i],
                                     priv->channelTargetDir) < 0)
            goto cleanup;
    }

5929
    if (!(priv->monConfig = virDomainChrSourceDefNew(NULL)))
5930 5931 5932 5933 5934 5935
        goto cleanup;

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

5936
    priv->monJSON = true;
5937 5938 5939 5940
    priv->monError = false;
    priv->monStart = 0;
    priv->gotShutdown = false;

5941 5942 5943 5944
    VIR_DEBUG("Updating guest CPU definition");
    if (qemuProcessUpdateGuestCPU(vm->def, priv->qemuCaps, caps, flags) < 0)
        goto cleanup;

5945 5946 5947 5948 5949
    for (i = 0; i < vm->def->nshmems; i++) {
        if (qemuDomainPrepareShmemChardev(vm->def->shmems[i]) < 0)
            goto cleanup;
    }

5950 5951 5952
    ret = 0;
 cleanup:
    virObjectUnref(caps);
5953
    virObjectUnref(cfg);
5954 5955 5956 5957
    return ret;
}


5958
static int
5959 5960 5961
qemuProcessSEVCreateFile(const char *configDir,
                         const char *name,
                         const char *data)
5962 5963
{
    char *configFile;
5964
    int ret = -1;
5965 5966 5967 5968 5969 5970 5971

    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);
5972
        goto cleanup;
5973 5974
    }

5975
    ret = 0;
5976
 cleanup:
5977
    VIR_FREE(configFile);
5978
    return ret;
5979 5980 5981 5982
}


static int
J
Ján Tomko 已提交
5983
qemuProcessPrepareSEVGuestInput(virDomainObjPtr vm)
5984 5985 5986 5987
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virDomainDefPtr def = vm->def;
    virQEMUCapsPtr qemuCaps = priv->qemuCaps;
5988
    virDomainSEVDefPtr sev = def->sev;
5989 5990 5991 5992

    if (!sev)
        return 0;

5993
    VIR_DEBUG("Preparing SEV guest");
5994 5995 5996 5997 5998 5999 6000 6001 6002

    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) {
6003
        if (qemuProcessSEVCreateFile(priv->libDir, "dh_cert", sev->dh_cert) < 0)
6004 6005 6006 6007
            return -1;
    }

    if (sev->session) {
6008
        if (qemuProcessSEVCreateFile(priv->libDir, "session", sev->session) < 0)
6009 6010 6011 6012 6013 6014 6015
            return -1;
    }

    return 0;
}


6016 6017 6018 6019 6020 6021 6022 6023 6024 6025 6026 6027 6028 6029 6030
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;

6031 6032 6033
        virStorageSourceBackingStoreClear(disk->src);

        if (qemuDomainDetermineDiskChain(driver, vm, disk, true) >= 0)
6034 6035 6036 6037 6038 6039 6040 6041 6042 6043 6044 6045
            continue;

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

        return -1;
    }

    return 0;
}


J
Ján Tomko 已提交
6046
int
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 6075
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
6076
/**
6077 6078 6079
 * qemuProcessPrepareHost:
 * @driver: qemu driver
 * @vm: domain object
6080
 * @flags: qemuProcessStartFlags
J
Jiri Denemark 已提交
6081
 *
6082 6083 6084
 * 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
6085
 *
6086
 * TODO: move all host modification from qemuBuildCommandLine into this function
J
Jiri Denemark 已提交
6087 6088
 */
int
6089 6090
qemuProcessPrepareHost(virQEMUDriverPtr driver,
                       virDomainObjPtr vm,
6091
                       unsigned int flags)
6092
{
6093
    int ret = -1;
6094
    unsigned int hostdev_flags = 0;
6095 6096
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
6097

6098
    if (qemuPrepareNVRAM(cfg, vm) < 0)
J
Jiri Denemark 已提交
6099
        goto cleanup;
6100

6101 6102 6103 6104
    if (vm->def->vsock) {
        if (qemuProcessOpenVhostVsock(vm->def->vsock) < 0)
            goto cleanup;
    }
6105 6106 6107 6108 6109
    /* 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");
6110
    if (qemuProcessNetworkPrepareDevices(vm->def) < 0)
J
Jiri Denemark 已提交
6111
        goto cleanup;
6112

6113
    /* Must be run before security labelling */
6114
    VIR_DEBUG("Preparing host devices");
6115 6116
    if (!cfg->relaxedACS)
        hostdev_flags |= VIR_HOSTDEV_STRICT_ACS_CHECK;
6117
    if (flags & VIR_QEMU_PROCESS_START_NEW)
6118
        hostdev_flags |= VIR_HOSTDEV_COLD_BOOT;
6119 6120
    if (qemuHostdevPrepareDomainDevices(driver, vm->def, priv->qemuCaps,
                                        hostdev_flags) < 0)
J
Jiri Denemark 已提交
6121
        goto cleanup;
6122

6123
    VIR_DEBUG("Preparing chr devices");
6124 6125 6126 6127
    if (virDomainChrDefForeach(vm->def,
                               true,
                               qemuProcessPrepareChardevDevice,
                               NULL) < 0)
J
Jiri Denemark 已提交
6128
        goto cleanup;
6129

6130
    if (qemuProcessBuildDestroyMemoryPaths(driver, vm, NULL, true) < 0)
6131
        goto cleanup;
6132

6133 6134
    /* Ensure no historical cgroup for this VM is lying around bogus
     * settings */
6135
    VIR_DEBUG("Ensuring no historical cgroup is lying around");
6136
    qemuRemoveCgroup(vm);
6137

6138
    if (virFileMakePath(cfg->logDir) < 0) {
6139 6140
        virReportSystemError(errno,
                             _("cannot create log directory %s"),
6141
                             cfg->logDir);
J
Jiri Denemark 已提交
6142
        goto cleanup;
6143 6144
    }

6145 6146 6147 6148 6149 6150 6151 6152 6153 6154 6155 6156 6157 6158 6159 6160 6161 6162 6163 6164 6165 6166 6167
    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;

6168 6169
    VIR_DEBUG("Write domain masterKey");
    if (qemuDomainWriteMasterKeyFile(driver, vm) < 0)
J
John Ferlan 已提交
6170 6171
        goto cleanup;

6172 6173 6174 6175
    VIR_DEBUG("Preparing disks (host)");
    if (qemuProcessPrepareHostStorage(driver, vm, flags) < 0)
        goto cleanup;

6176 6177 6178 6179
    VIR_DEBUG("Preparing external devices");
    if (qemuExtDevicesPrepareHost(driver, vm->def) < 0)
        goto cleanup;

J
Ján Tomko 已提交
6180
    if (qemuProcessPrepareSEVGuestInput(vm) < 0)
6181 6182
        goto cleanup;

6183 6184 6185 6186 6187 6188 6189
    ret = 0;
 cleanup:
    virObjectUnref(cfg);
    return ret;
}


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 6222
/**
 * 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;
}


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 6256
/**
 * 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
6257
    size_t i;
6258

6259
    VIR_DEBUG("conn=%p driver=%p vm=%p name=%s if=%d asyncJob=%d "
6260 6261 6262
              "incoming.launchURI=%s incoming.deferredURI=%s "
              "incoming.fd=%d incoming.path=%s "
              "snapshot=%p vmop=%d flags=0x%x",
6263
              conn, driver, vm, vm->def->name, vm->def->id, asyncJob,
6264 6265 6266 6267 6268 6269 6270 6271 6272 6273
              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 |
6274
                  VIR_QEMU_PROCESS_START_AUTODESTROY |
6275 6276
                  VIR_QEMU_PROCESS_START_NEW |
                  VIR_QEMU_PROCESS_START_GEN_VMID, -1);
6277 6278 6279

    cfg = virQEMUDriverGetConfig(driver);

6280 6281 6282 6283 6284 6285
    if ((flags & VIR_QEMU_PROCESS_START_AUTODESTROY) && !conn) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Domain autodestroy requires a connection handle"));
        return -1;
    }

6286 6287 6288 6289 6290 6291 6292 6293
    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;

6294
    VIR_DEBUG("Creating domain log file");
6295 6296
    if (!(logCtxt = qemuDomainLogContextNew(driver, vm,
                                            QEMU_DOMAIN_LOG_CONTEXT_MODE_START)))
J
Jiri Denemark 已提交
6297
        goto cleanup;
6298
    logfile = qemuDomainLogContextGetWriteFD(logCtxt);
6299

6300 6301 6302
    if (qemuProcessGenID(vm, flags) < 0)
        goto cleanup;

6303 6304 6305
    if (qemuExtDevicesStart(driver, vm->def, logCtxt) < 0)
        goto cleanup;

6306
    VIR_DEBUG("Building emulator command line");
6307
    if (!(cmd = qemuBuildCommandLine(driver,
6308
                                     qemuDomainLogContextGetManager(logCtxt),
6309
                                     driver->securityManager,
6310
                                     vm,
6311 6312
                                     incoming ? incoming->launchURI : NULL,
                                     snapshot, vmop,
J
Ján Tomko 已提交
6313
                                     false,
6314
                                     qemuCheckFips(),
6315
                                     &nnicindexes, &nicindexes)))
J
Jiri Denemark 已提交
6316
        goto cleanup;
6317

6318 6319
    if (incoming && incoming->fd != -1)
        virCommandPassFD(cmd, incoming->fd, 0);
6320

6321
    /* now that we know it is about to start call the hook if present */
6322 6323 6324
    if (qemuProcessStartHook(driver, vm,
                             VIR_HOOK_QEMU_OP_START,
                             VIR_HOOK_SUBOP_BEGIN) < 0)
J
Jiri Denemark 已提交
6325
        goto cleanup;
6326

6327
    qemuLogOperation(vm, "starting up", cmd, logCtxt);
6328

6329
    qemuDomainObjCheckTaint(driver, vm, logCtxt);
6330

6331
    qemuDomainLogContextMarkPosition(logCtxt);
6332

6333 6334 6335 6336 6337
    VIR_DEBUG("Building mount namespace");

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

6338
    VIR_DEBUG("Clear emulator capabilities: %d",
6339 6340
              cfg->clearEmulatorCapabilities);
    if (cfg->clearEmulatorCapabilities)
6341 6342
        virCommandClearCaps(cmd);

6343 6344
    VIR_DEBUG("Setting up raw IO");
    if (qemuProcessSetupRawIO(driver, vm, cmd) < 0)
J
Jiri Denemark 已提交
6345
        goto cleanup;
6346

6347
    virCommandSetPreExecHook(cmd, qemuProcessHook, &hookData);
6348 6349
    virCommandSetMaxProcesses(cmd, cfg->maxProcesses);
    virCommandSetMaxFiles(cmd, cfg->maxFiles);
6350
    virCommandSetMaxCoreSize(cmd, cfg->maxCore);
6351
    virCommandSetUmask(cmd, 0x002);
6352

6353
    VIR_DEBUG("Setting up security labelling");
6354 6355
    if (qemuSecuritySetChildProcessLabel(driver->securityManager,
                                         vm->def, cmd) < 0)
J
Jiri Denemark 已提交
6356
        goto cleanup;
6357

6358 6359 6360
    virCommandSetOutputFD(cmd, &logfile);
    virCommandSetErrorFD(cmd, &logfile);
    virCommandNonblockingFDs(cmd);
6361
    virCommandSetPidFile(cmd, priv->pidfile);
6362
    virCommandDaemonize(cmd);
6363
    virCommandRequireHandshake(cmd);
6364

6365
    if (qemuSecurityPreFork(driver->securityManager) < 0)
J
Jiri Denemark 已提交
6366
        goto cleanup;
6367
    rv = virCommandRun(cmd, NULL);
6368
    qemuSecurityPostFork(driver->securityManager);
6369

E
Eric Blake 已提交
6370
    /* wait for qemu process to show up */
6371
    if (rv == 0) {
6372
        if (virPidFileReadPath(priv->pidfile, &vm->pid) < 0) {
6373 6374
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Domain %s didn't show up"), vm->def->name);
6375
            rv = -1;
6376
        }
M
Michal Privoznik 已提交
6377
        VIR_DEBUG("QEMU vm=%p name=%s running with pid=%lld",
6378
                  vm, vm->def->name, (long long)vm->pid);
6379 6380 6381
    } else {
        VIR_DEBUG("QEMU vm=%p name=%s failed to spawn",
                  vm, vm->def->name);
6382 6383
    }

6384
    VIR_DEBUG("Writing early domain status to disk");
6385
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0)
J
Jiri Denemark 已提交
6386
        goto cleanup;
6387

6388 6389
    VIR_DEBUG("Waiting for handshake from child");
    if (virCommandHandshakeWait(cmd) < 0) {
6390
        /* Read errors from child that occurred between fork and exec. */
6391 6392
        qemuProcessReportLogError(logCtxt,
                                  _("Process exited prior to exec"));
J
Jiri Denemark 已提交
6393
        goto cleanup;
6394 6395
    }

6396
    VIR_DEBUG("Setting up domain cgroup (if required)");
6397
    if (qemuSetupCgroup(vm, nnicindexes, nicindexes) < 0)
J
Jiri Denemark 已提交
6398
        goto cleanup;
6399

6400 6401 6402 6403
    if (!(priv->perf = virPerfNew()))
        goto cleanup;

    for (i = 0; i < VIR_PERF_EVENT_LAST; i++) {
6404
        if (vm->def->perf.events[i] == VIR_TRISTATE_BOOL_YES &&
6405 6406
            virPerfEventEnable(priv->perf, i, vm->pid) < 0)
            goto cleanup;
Q
Qiaowei Ren 已提交
6407
    }
6408

6409 6410 6411 6412 6413 6414
    /* This must be done after cgroup placement to avoid resetting CPU
     * affinity */
    if (!vm->def->cputune.emulatorpin &&
        qemuProcessInitCpuAffinity(vm) < 0)
        goto cleanup;

6415 6416 6417 6418
    VIR_DEBUG("Setting emulator tuning/settings");
    if (qemuProcessSetupEmulator(vm) < 0)
        goto cleanup;

6419 6420 6421 6422
    VIR_DEBUG("Setting cgroup for external devices (if required)");
    if (qemuSetupCgroupForExtDevices(vm, driver) < 0)
        goto cleanup;

6423
    VIR_DEBUG("Setting up resctrl");
6424 6425 6426
    if (qemuProcessResctrlCreate(driver, vm) < 0)
        goto cleanup;

6427
    VIR_DEBUG("Setting up managed PR daemon");
6428 6429
    if (virDomainDefHasManagedPR(vm->def) &&
        qemuProcessStartManagedPRDaemon(vm) < 0)
6430 6431
        goto cleanup;

6432
    VIR_DEBUG("Setting domain security labels");
6433 6434 6435 6436
    if (qemuSecuritySetAllLabel(driver,
                                vm,
                                incoming ? incoming->path : NULL) < 0)
        goto cleanup;
6437

6438
    /* Security manager labeled all devices, therefore
J
Jiri Denemark 已提交
6439 6440 6441 6442
     * if any operation from now on fails, we need to ask the caller to
     * restore labels.
     */
    ret = -2;
6443

J
Jiri Denemark 已提交
6444
    if (incoming && incoming->fd != -1) {
6445 6446 6447 6448 6449 6450 6451
        /* 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
6452
        if (fstat(incoming->fd, &stdin_sb) < 0) {
6453
            virReportSystemError(errno,
J
Jiri Denemark 已提交
6454 6455
                                 _("cannot stat fd %d"), incoming->fd);
            goto cleanup;
6456 6457
        }
        if (S_ISFIFO(stdin_sb.st_mode) &&
6458 6459
            qemuSecuritySetImageFDLabel(driver->securityManager,
                                        vm->def, incoming->fd) < 0)
J
Jiri Denemark 已提交
6460
            goto cleanup;
6461 6462 6463
    }

    VIR_DEBUG("Labelling done, completing handshake to child");
6464
    if (virCommandHandshakeNotify(cmd) < 0)
J
Jiri Denemark 已提交
6465
        goto cleanup;
6466 6467
    VIR_DEBUG("Handshake complete, child running");

6468
    if (rv == -1) /* The VM failed to start; tear filters before taps */
6469 6470
        virDomainConfVMNWFilterTeardown(vm);

6471
    if (rv == -1) /* The VM failed to start */
J
Jiri Denemark 已提交
6472
        goto cleanup;
6473

6474
    VIR_DEBUG("Waiting for monitor to show up");
6475
    if (qemuProcessWaitForMonitor(driver, vm, asyncJob, logCtxt) < 0)
J
Jiri Denemark 已提交
6476
        goto cleanup;
6477

6478 6479
    if (qemuConnectAgent(driver, vm) < 0)
        goto cleanup;
D
Daniel P. Berrange 已提交
6480

6481
    VIR_DEBUG("Verifying and updating provided guest CPU");
6482
    if (qemuProcessUpdateAndVerifyCPU(driver, vm, asyncJob) < 0)
J
Jiri Denemark 已提交
6483
        goto cleanup;
6484

6485
    VIR_DEBUG("Setting up post-init cgroup restrictions");
6486
    if (qemuSetupCpusetMems(vm) < 0)
J
Jiri Denemark 已提交
6487
        goto cleanup;
6488

6489 6490 6491 6492 6493 6494 6495 6496 6497 6498 6499 6500
    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;
    }

6501
    VIR_DEBUG("Refreshing VCPU info");
6502
    if (qemuDomainRefreshVcpuInfo(driver, vm, asyncJob, false) < 0)
J
Jiri Denemark 已提交
6503
        goto cleanup;
6504

6505 6506 6507
    if (qemuDomainValidateVcpuInfo(vm) < 0)
        goto cleanup;

6508 6509
    qemuDomainVcpuPersistOrder(vm->def);

6510 6511
    VIR_DEBUG("Detecting IOThread PIDs");
    if (qemuProcessDetectIOThreadPIDs(driver, vm, asyncJob) < 0)
J
Jiri Denemark 已提交
6512
        goto cleanup;
6513

6514 6515 6516 6517
    VIR_DEBUG("Setting global CPU cgroup (if required)");
    if (qemuSetupGlobalCpuCgroup(vm) < 0)
        goto cleanup;

6518 6519
    VIR_DEBUG("Setting vCPU tuning/settings");
    if (qemuProcessSetupVcpus(vm) < 0)
J
Jiri Denemark 已提交
6520
        goto cleanup;
6521

6522 6523
    VIR_DEBUG("Setting IOThread tuning/settings");
    if (qemuProcessSetupIOThreads(vm) < 0)
J
Jiri Denemark 已提交
6524
        goto cleanup;
6525

6526
    VIR_DEBUG("Setting any required VM passwords");
6527
    if (qemuProcessInitPasswords(driver, vm, asyncJob) < 0)
J
Jiri Denemark 已提交
6528
        goto cleanup;
6529

6530 6531 6532 6533
    /* set default link states */
    /* qemu doesn't support setting this on the command line, so
     * enter the monitor */
    VIR_DEBUG("Setting network link states");
6534
    if (qemuProcessSetLinkStates(driver, vm, asyncJob) < 0)
J
Jiri Denemark 已提交
6535
        goto cleanup;
6536

6537
    VIR_DEBUG("Setting initial memory amount");
6538
    if (qemuProcessSetupBalloon(driver, vm, asyncJob) < 0)
J
Jiri Denemark 已提交
6539
        goto cleanup;
6540

6541
    /* Since CPUs were not started yet, the balloon could not return the memory
6542 6543
     * 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 */
6544
    if (!incoming && !snapshot &&
6545
        qemuProcessRefreshBalloonState(driver, vm, asyncJob) < 0)
J
Jiri Denemark 已提交
6546
        goto cleanup;
6547

J
Jiri Denemark 已提交
6548 6549 6550 6551 6552 6553 6554
    if (flags & VIR_QEMU_PROCESS_START_AUTODESTROY &&
        qemuProcessAutoDestroyAdd(driver, vm, conn) < 0)
        goto cleanup;

    ret = 0;

 cleanup:
6555 6556
    if (ret < 0)
        qemuExtDevicesStop(driver, vm->def);
6557
    qemuDomainSecretDestroy(vm);
J
Jiri Denemark 已提交
6558
    virCommandFree(cmd);
6559
    virObjectUnref(logCtxt);
J
Jiri Denemark 已提交
6560 6561 6562 6563 6564 6565 6566
    virObjectUnref(cfg);
    virObjectUnref(caps);
    VIR_FREE(nicindexes);
    return ret;
}


6567 6568 6569 6570 6571 6572 6573 6574 6575 6576
/**
 * 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.
 */
6577
int
6578 6579 6580 6581 6582 6583 6584 6585 6586 6587 6588 6589 6590 6591 6592 6593 6594 6595 6596 6597 6598 6599 6600 6601
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;
}


6602 6603 6604 6605 6606 6607
/**
 * qemuProcessFinishStartup:
 *
 * Finish starting a new domain.
 */
int
6608
qemuProcessFinishStartup(virQEMUDriverPtr driver,
6609 6610 6611 6612 6613 6614 6615 6616 6617 6618
                         virDomainObjPtr vm,
                         qemuDomainAsyncJob asyncJob,
                         bool startCPUs,
                         virDomainPausedReason pausedReason)
{
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
    int ret = -1;

    if (startCPUs) {
        VIR_DEBUG("Starting domain CPUs");
6619
        if (qemuProcessStartCPUs(driver, vm,
6620 6621
                                 VIR_DOMAIN_RUNNING_BOOTED,
                                 asyncJob) < 0) {
6622
            if (virGetLastErrorCode() == VIR_ERR_OK)
6623 6624 6625 6626 6627 6628 6629 6630 6631
                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");
6632
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0)
6633 6634 6635 6636 6637 6638 6639 6640 6641 6642 6643 6644 6645 6646 6647
        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
6648 6649 6650 6651
int
qemuProcessStart(virConnectPtr conn,
                 virQEMUDriverPtr driver,
                 virDomainObjPtr vm,
6652
                 virCPUDefPtr updatedCPU,
J
Jiri Denemark 已提交
6653 6654 6655 6656 6657 6658 6659 6660 6661 6662 6663 6664 6665 6666 6667 6668 6669 6670 6671 6672 6673 6674 6675 6676 6677
                 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 |
6678 6679
                      VIR_QEMU_PROCESS_START_AUTODESTROY |
                      VIR_QEMU_PROCESS_START_GEN_VMID, cleanup);
J
Jiri Denemark 已提交
6680

6681 6682 6683
    if (!migrateFrom && !snapshot)
        flags |= VIR_QEMU_PROCESS_START_NEW;

6684 6685
    if (qemuProcessInit(driver, vm, updatedCPU,
                        asyncJob, !!migrateFrom, flags) < 0)
J
Jiri Denemark 已提交
6686 6687 6688
        goto cleanup;

    if (migrateFrom) {
6689
        incoming = qemuProcessIncomingDefNew(priv->qemuCaps, NULL, migrateFrom,
J
Jiri Denemark 已提交
6690 6691 6692 6693 6694
                                             migrateFd, migratePath);
        if (!incoming)
            goto stop;
    }

6695
    if (qemuProcessPrepareDomain(driver, vm, flags) < 0)
6696 6697
        goto stop;

6698
    if (qemuProcessPrepareHost(driver, vm, flags) < 0)
6699 6700
        goto stop;

J
Jiri Denemark 已提交
6701 6702
    if ((rv = qemuProcessLaunch(conn, driver, vm, asyncJob, incoming,
                                snapshot, vmop, flags)) < 0) {
6703
        if (rv == -2)
J
Jiri Denemark 已提交
6704 6705 6706 6707
            relabel = true;
        goto stop;
    }
    relabel = true;
6708

6709 6710
    if (incoming &&
        incoming->deferredURI &&
6711
        qemuMigrationDstRun(driver, vm, incoming->deferredURI, asyncJob) < 0)
J
Jiri Denemark 已提交
6712
        goto stop;
6713

6714
    if (qemuProcessFinishStartup(driver, vm, asyncJob,
6715 6716 6717 6718
                                 !(flags & VIR_QEMU_PROCESS_START_PAUSED),
                                 incoming ?
                                 VIR_DOMAIN_PAUSED_MIGRATION :
                                 VIR_DOMAIN_PAUSED_USER) < 0)
J
Jiri Denemark 已提交
6719
        goto stop;
6720

6721 6722 6723
    if (!incoming) {
        /* Keep watching qemu log for errors during incoming migration, otherwise
         * unset reporting errors from qemu log. */
6724
        qemuMonitorSetDomainLog(priv->mon, NULL, NULL, NULL);
6725

6726 6727 6728 6729 6730 6731
        /* 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;
    }

6732 6733 6734
    ret = 0;

 cleanup:
6735
    qemuProcessIncomingDefFree(incoming);
6736
    return ret;
6737

J
Jiri Denemark 已提交
6738 6739 6740 6741 6742 6743
 stop:
    stopFlags = 0;
    if (!relabel)
        stopFlags |= VIR_QEMU_PROCESS_STOP_NO_RELABEL;
    if (migrateFrom)
        stopFlags |= VIR_QEMU_PROCESS_STOP_MIGRATED;
6744
    if (priv->mon)
6745
        qemuMonitorSetDomainLog(priv->mon, NULL, NULL, NULL);
6746
    qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_FAILED, asyncJob, stopFlags);
6747
    goto cleanup;
6748 6749 6750
}


6751
virCommandPtr
6752
qemuProcessCreatePretendCmd(virQEMUDriverPtr driver,
6753 6754
                            virDomainObjPtr vm,
                            const char *migrateURI,
6755
                            bool enableFips,
6756 6757 6758 6759 6760 6761 6762 6763 6764 6765
                            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;
6766
    flags |= VIR_QEMU_PROCESS_START_NEW;
6767 6768
    if (standalone)
        flags |= VIR_QEMU_PROCESS_START_STANDALONE;
6769

6770 6771
    if (qemuProcessInit(driver, vm, NULL, QEMU_ASYNC_JOB_NONE,
                        !!migrateURI, flags) < 0)
6772 6773
        goto cleanup;

6774
    if (qemuProcessPrepareDomain(driver, vm, flags) < 0)
6775 6776 6777
        goto cleanup;

    VIR_DEBUG("Building emulator command line");
6778
    cmd = qemuBuildCommandLine(driver,
6779
                               NULL,
6780
                               driver->securityManager,
6781
                               vm,
6782 6783 6784 6785
                               migrateURI,
                               NULL,
                               VIR_NETDEV_VPORT_PROFILE_OP_NO_OP,
                               standalone,
6786
                               enableFips,
6787
                               NULL,
6788
                               NULL);
6789 6790 6791 6792 6793 6794

 cleanup:
    return cmd;
}


6795
int
6796
qemuProcessKill(virDomainObjPtr vm, unsigned int flags)
6797
{
6798
    int ret;
6799

6800
    VIR_DEBUG("vm=%p name=%s pid=%lld flags=0x%x",
6801
              vm, vm->def->name,
6802
              (long long)vm->pid, flags);
6803

6804 6805 6806 6807 6808
    if (!(flags & VIR_QEMU_PROCESS_KILL_NOCHECK)) {
        if (!virDomainObjIsActive(vm)) {
            VIR_DEBUG("VM '%s' not active", vm->def->name);
            return 0;
        }
6809 6810
    }

6811
    if (flags & VIR_QEMU_PROCESS_KILL_NOWAIT) {
6812 6813 6814 6815 6816
        virProcessKill(vm->pid,
                       (flags & VIR_QEMU_PROCESS_KILL_FORCE) ?
                       SIGKILL : SIGTERM);
        return 0;
    }
6817

6818 6819
    ret = virProcessKillPainfully(vm->pid,
                                  !!(flags & VIR_QEMU_PROCESS_KILL_FORCE));
6820

6821
    return ret;
6822 6823 6824
}


6825 6826 6827 6828 6829 6830 6831 6832 6833 6834 6835 6836 6837 6838 6839 6840 6841 6842 6843 6844 6845 6846 6847 6848 6849 6850 6851 6852 6853 6854 6855 6856 6857 6858 6859 6860 6861 6862 6863
/**
 * 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;
}


6864
void qemuProcessStop(virQEMUDriverPtr driver,
6865
                     virDomainObjPtr vm,
6866
                     virDomainShutoffReason reason,
6867
                     qemuDomainAsyncJob asyncJob,
6868
                     unsigned int flags)
6869 6870 6871 6872 6873 6874
{
    int ret;
    int retries = 0;
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virErrorPtr orig_err;
    virDomainDefPtr def;
A
Ansis Atteka 已提交
6875
    virNetDevVPortProfilePtr vport = NULL;
6876
    size_t i;
6877
    char *timestamp;
6878
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
6879

M
Michal Privoznik 已提交
6880
    VIR_DEBUG("Shutting down vm=%p name=%s id=%d pid=%lld, "
6881
              "reason=%s, asyncJob=%s, flags=0x%x",
6882
              vm, vm->def->name, vm->def->id,
6883
              (long long)vm->pid,
6884 6885 6886
              virDomainShutoffReasonTypeToString(reason),
              qemuDomainAsyncJobTypeToString(asyncJob),
              flags);
6887

6888 6889 6890 6891
    /* This method is routinely used in clean up paths. Disable error
     * reporting so we don't squash a legit error. */
    orig_err = virSaveLastError();

6892 6893 6894 6895 6896 6897 6898 6899 6900 6901 6902 6903 6904 6905 6906
    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;
    }

6907
    qemuProcessBuildDestroyMemoryPaths(driver, vm, NULL, false);
6908

6909 6910
    vm->def->id = -1;

6911
    if (virAtomicIntDecAndTest(&driver->nactive) && driver->inhibitCallback)
6912 6913
        driver->inhibitCallback(false, driver->inhibitOpaque);

6914 6915
    /* Wake up anything waiting on domain condition */
    virDomainObjBroadcast(vm);
6916

6917
    if ((timestamp = virTimeStringNow()) != NULL) {
6918 6919 6920
        qemuDomainLogAppendMessage(driver, vm, "%s: shutting down, reason=%s\n",
                                   timestamp,
                                   virDomainShutoffReasonTypeToString(reason));
6921
        VIR_FREE(timestamp);
6922 6923
    }

6924 6925 6926
    /* Clear network bandwidth */
    virDomainClearNetBandwidth(vm);

6927 6928
    virDomainConfVMNWFilterTeardown(vm);

6929
    if (cfg->macFilter) {
6930
        def = vm->def;
6931
        for (i = 0; i < def->nnets; i++) {
6932 6933 6934
            virDomainNetDefPtr net = def->nets[i];
            if (net->ifname == NULL)
                continue;
6935 6936 6937
            ignore_value(ebtablesRemoveForwardAllowIn(driver->ebtables,
                                                      net->ifname,
                                                      &net->mac));
6938 6939 6940
        }
    }

6941
    virPortAllocatorRelease(priv->nbdPort);
6942
    priv->nbdPort = 0;
6943

D
Daniel P. Berrange 已提交
6944 6945 6946 6947
    if (priv->agent) {
        qemuAgentClose(priv->agent);
        priv->agent = NULL;
    }
6948
    priv->agentError = false;
D
Daniel P. Berrange 已提交
6949

6950
    if (priv->mon) {
6951
        qemuMonitorClose(priv->mon);
6952 6953
        priv->mon = NULL;
    }
6954 6955 6956 6957 6958 6959 6960 6961

    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
6962 6963 6964
    /* Remove the master key */
    qemuDomainMasterKeyRemove(priv);

6965
    /* Do this before we delete the tree and remove pidfile. */
6966
    qemuProcessKillManagedPRDaemon(vm);
6967

6968 6969
    virFileDeleteTree(priv->libDir);
    virFileDeleteTree(priv->channelTargetDir);
6970

6971 6972 6973 6974 6975 6976
    ignore_value(virDomainChrDefForeach(vm->def,
                                        false,
                                        qemuProcessCleanupChardevDevice,
                                        NULL));


6977
    /* shut it off for sure */
6978 6979 6980
    ignore_value(qemuProcessKill(vm,
                                 VIR_QEMU_PROCESS_KILL_FORCE|
                                 VIR_QEMU_PROCESS_KILL_NOCHECK));
6981

6982 6983
    qemuDomainCleanupRun(driver, vm);

6984 6985
    qemuExtDevicesStop(driver, vm->def);

6986
    /* Stop autodestroy in case guest is restarted */
6987
    qemuProcessAutoDestroyRemove(driver, vm);
6988

6989 6990
    /* now that we know it's stopped call the hook if present */
    if (virHookPresent(VIR_HOOK_DRIVER_QEMU)) {
6991
        char *xml = qemuDomainDefFormatXML(driver, vm->def, 0);
6992 6993

        /* we can't stop the operation even if the script raised an error */
6994 6995 6996
        ignore_value(virHookCall(VIR_HOOK_DRIVER_QEMU, vm->def->name,
                                 VIR_HOOK_QEMU_OP_STOPPED, VIR_HOOK_SUBOP_END,
                                 NULL, xml, NULL));
6997 6998 6999
        VIR_FREE(xml);
    }

7000 7001
    /* Reset Security Labels unless caller don't want us to */
    if (!(flags & VIR_QEMU_PROCESS_STOP_NO_RELABEL))
7002 7003 7004
        qemuSecurityRestoreAllLabel(driver, vm,
                                    !!(flags & VIR_QEMU_PROCESS_STOP_MIGRATED));

7005
    qemuSecurityReleaseLabel(driver->securityManager, vm->def);
7006

7007
    for (i = 0; i < vm->def->ndisks; i++) {
7008
        virDomainDeviceDef dev;
7009
        virDomainDiskDefPtr disk = vm->def->disks[i];
7010 7011 7012 7013

        dev.type = VIR_DOMAIN_DEVICE_DISK;
        dev.data.disk = disk;
        ignore_value(qemuRemoveSharedDevice(driver, &dev, vm->def->name));
7014 7015
    }

7016
    /* Clear out dynamically assigned labels */
7017
    for (i = 0; i < vm->def->nseclabels; i++) {
7018
        if (vm->def->seclabels[i]->type == VIR_DOMAIN_SECLABEL_DYNAMIC)
7019 7020
            VIR_FREE(vm->def->seclabels[i]->label);
        VIR_FREE(vm->def->seclabels[i]->imagelabel);
7021 7022
    }

7023
    qemuHostdevReAttachDomainDevices(driver, vm->def);
7024 7025 7026 7027

    def = vm->def;
    for (i = 0; i < def->nnets; i++) {
        virDomainNetDefPtr net = def->nets[i];
7028 7029 7030 7031
        vport = virDomainNetGetActualVirtPortProfile(net);

        switch (virDomainNetGetActualType(net)) {
        case VIR_DOMAIN_NET_TYPE_DIRECT:
7032
            ignore_value(virNetDevMacVLanDeleteWithVPortProfile(
7033
                             net->ifname, &net->mac,
7034 7035
                             virDomainNetGetActualDirectDev(net),
                             virDomainNetGetActualDirectMode(net),
7036
                             virDomainNetGetActualVirtPortProfile(net),
7037
                             cfg->stateDir));
7038
            break;
7039 7040 7041 7042 7043 7044
        case VIR_DOMAIN_NET_TYPE_ETHERNET:
            if (net->ifname) {
                ignore_value(virNetDevTapDelete(net->ifname, net->backend.tap));
                VIR_FREE(net->ifname);
            }
            break;
7045 7046 7047 7048
        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))
7049
                ignore_value(virNetDevTapDelete(net->ifname, net->backend.tap));
7050 7051
#endif
            break;
7052 7053 7054 7055 7056 7057 7058 7059 7060 7061 7062
        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;
7063
        }
7064 7065 7066
        /* release the physical device (or any other resources used by
         * this interface in the network driver
         */
7067 7068 7069 7070 7071 7072 7073 7074 7075
        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
7076

7077 7078
        /* kick the device out of the hostdev list too */
        virDomainNetRemoveHostdev(def, net);
7079
        virDomainNetReleaseActualDevice(vm->def, net);
7080
    }
7081

7082
 retry:
7083
    if ((ret = qemuRemoveCgroup(vm)) < 0) {
7084 7085 7086 7087 7088 7089 7090
        if (ret == -EBUSY && (retries++ < 5)) {
            usleep(200*1000);
            goto retry;
        }
        VIR_WARN("Failed to remove cgroup for %s",
                 vm->def->name);
    }
7091

7092 7093 7094
    /* 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 */
B
Bing Niu 已提交
7095 7096
    for (i = 0; i < vm->def->nresctrls; i++)
        virResctrlAllocRemove(vm->def->resctrls[i]->alloc);
7097

7098 7099
    qemuProcessRemoveDomainStatus(driver, vm);

7100 7101
    /* Remove VNC and Spice ports from port reservation bitmap, but only if
       they were reserved by the driver (autoport=yes)
7102
    */
7103
    for (i = 0; i < vm->def->ngraphics; ++i) {
7104
        virDomainGraphicsDefPtr graphics = vm->def->graphics[i];
7105 7106
        if (graphics->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC) {
            if (graphics->data.vnc.autoport) {
7107
                virPortAllocatorRelease(graphics->data.vnc.port);
7108
            } else if (graphics->data.vnc.portReserved) {
7109
                virPortAllocatorRelease(graphics->data.spice.port);
7110 7111
                graphics->data.vnc.portReserved = false;
            }
7112
            if (graphics->data.vnc.websocketGenerated) {
7113
                virPortAllocatorRelease(graphics->data.vnc.websocket);
7114 7115 7116
                graphics->data.vnc.websocketGenerated = false;
                graphics->data.vnc.websocket = -1;
            } else if (graphics->data.vnc.websocket) {
7117
                virPortAllocatorRelease(graphics->data.vnc.websocket);
7118
            }
7119
        }
7120 7121
        if (graphics->type == VIR_DOMAIN_GRAPHICS_TYPE_SPICE) {
            if (graphics->data.spice.autoport) {
7122 7123
                virPortAllocatorRelease(graphics->data.spice.port);
                virPortAllocatorRelease(graphics->data.spice.tlsPort);
7124 7125
            } else {
                if (graphics->data.spice.portReserved) {
7126
                    virPortAllocatorRelease(graphics->data.spice.port);
7127 7128 7129 7130
                    graphics->data.spice.portReserved = false;
                }

                if (graphics->data.spice.tlsPortReserved) {
7131
                    virPortAllocatorRelease(graphics->data.spice.tlsPort);
7132 7133 7134
                    graphics->data.spice.tlsPortReserved = false;
                }
            }
7135
        }
7136 7137
    }

7138
    vm->taint = 0;
7139
    vm->pid = -1;
J
Jiri Denemark 已提交
7140
    virDomainObjSetState(vm, VIR_DOMAIN_SHUTOFF, reason);
7141 7142
    for (i = 0; i < vm->def->niothreadids; i++)
        vm->def->iothreadids[i]->thread_id = 0;
7143

7144 7145
    /* clear all private data entries which are no longer needed */
    qemuDomainObjPrivateDataClear(priv);
7146

7147
    /* The "release" hook cleans up additional resources */
7148
    if (virHookPresent(VIR_HOOK_DRIVER_QEMU)) {
7149
        char *xml = qemuDomainDefFormatXML(driver, vm->def, 0);
7150 7151 7152

        /* we can't stop the operation even if the script raised an error */
        virHookCall(VIR_HOOK_DRIVER_QEMU, vm->def->name,
7153 7154
                    VIR_HOOK_QEMU_OP_RELEASE, VIR_HOOK_SUBOP_END,
                    NULL, xml, NULL);
7155 7156 7157
        VIR_FREE(xml);
    }

7158
    virDomainObjRemoveTransientDef(vm);
7159

7160 7161 7162 7163 7164
 endjob:
    if (asyncJob != QEMU_ASYNC_JOB_NONE)
        qemuDomainObjEndJob(driver, vm);

 cleanup:
7165 7166 7167 7168
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
7169
    virObjectUnref(cfg);
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
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;
7195
    virDomainVirtType virtType;
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 7284 7285

    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);
7286
    if (!(priv->qemuCaps = virQEMUCapsCacheLookupCopy(driver->qemuCapsCache,
7287 7288 7289 7290 7291 7292 7293 7294 7295 7296 7297
                                                      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;

7298 7299 7300 7301
    /* Attaching to running QEMU so we need to detect whether it was started
     * with -no-reboot. */
    qemuProcessPrepareAllowReboot(vm);

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
    /*
     * 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;
7347
    if (qemuMonitorGetVirtType(priv->mon, &virtType) < 0)
7348
        goto exit_monitor;
7349
    vm->def->virtType = virtType;
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 7419 7420 7421
    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;
}


7422
static void
7423 7424 7425 7426 7427 7428 7429 7430 7431 7432 7433 7434 7435 7436 7437 7438 7439 7440 7441 7442 7443 7444 7445
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)
7446
        return;
7447 7448 7449 7450 7451 7452 7453 7454 7455 7456 7457

    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);

7458 7459
    qemuDomainObjEndJob(driver, dom);

7460
    virObjectEventStateQueue(driver->domainEventState, event);
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 7541 7542 7543
}

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;
}


7544 7545 7546 7547 7548
static int
qemuProcessRefreshCPU(virQEMUDriverPtr driver,
                      virDomainObjPtr vm)
{
    virCapsPtr caps = virQEMUDriverGetCapabilities(driver, false);
7549
    qemuDomainObjPrivatePtr priv = vm->privateData;
7550
    virCPUDefPtr host = NULL;
7551
    virCPUDefPtr cpu = NULL;
7552 7553 7554 7555 7556 7557 7558 7559 7560 7561 7562 7563 7564 7565 7566 7567 7568 7569 7570 7571 7572
    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;

7573 7574 7575 7576 7577 7578 7579
        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)
7580 7581 7582 7583
            goto cleanup;

        if (qemuProcessUpdateCPU(driver, vm, QEMU_ASYNC_JOB_NONE) < 0)
            goto cleanup;
7584 7585 7586 7587 7588 7589 7590 7591
    } 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;
7592 7593 7594 7595 7596
    }

    ret = 0;

 cleanup:
7597
    virCPUDefFree(cpu);
7598 7599 7600 7601 7602 7603
    virCPUDefFree(host);
    virObjectUnref(caps);
    return ret;
}


7604 7605 7606 7607 7608 7609 7610 7611 7612 7613 7614 7615 7616 7617 7618 7619 7620 7621 7622 7623 7624 7625 7626 7627
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)
7628
{
7629 7630 7631 7632
    struct qemuProcessReconnectData *data = opaque;
    virQEMUDriverPtr driver = data->driver;
    virDomainObjPtr obj = data->obj;
    qemuDomainObjPrivatePtr priv;
7633
    qemuDomainJobObj oldjob;
7634 7635 7636
    int state;
    int reason;
    virQEMUDriverConfigPtr cfg;
7637
    size_t i;
7638 7639
    unsigned int stopFlags = 0;
    bool jobStarted = false;
7640
    virCapsPtr caps = NULL;
7641
    bool retry = true;
7642

7643
    VIR_FREE(data);
7644

7645 7646 7647 7648 7649 7650
    qemuDomainObjRestoreJob(obj, &oldjob);
    if (oldjob.asyncJob == QEMU_ASYNC_JOB_MIGRATION_IN)
        stopFlags |= VIR_QEMU_PROCESS_STOP_MIGRATED;

    cfg = virQEMUDriverGetConfig(driver);
    priv = obj->privateData;
7651

7652
    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
7653
        goto error;
7654

7655
    if (qemuDomainObjBeginJob(driver, obj, QEMU_JOB_MODIFY) < 0)
7656
        goto error;
7657
    jobStarted = true;
7658

7659 7660 7661 7662
    /* 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;
7663

7664 7665 7666
    /* Restore the masterKey */
    if (qemuDomainMasterKeyReadFile(priv) < 0)
        goto error;
7667

7668 7669 7670 7671
    /* 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);

7672 7673 7674 7675 7676 7677
    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);
7678 7679

    /* XXX check PID liveliness & EXE path */
7680
    if (qemuConnectMonitor(driver, obj, QEMU_ASYNC_JOB_NONE, retry, NULL) < 0)
7681
        goto error;
7682

7683
    if (qemuHostdevUpdateActiveDomainDevices(driver, obj->def) < 0)
7684
        goto error;
7685

7686 7687 7688 7689
    priv->machineName = qemuDomainGetMachineName(obj);
    if (!priv->machineName)
        goto error;

7690
    if (qemuConnectCgroup(obj) < 0)
7691
        goto error;
7692

7693
    if (qemuDomainPerfRestart(obj) < 0)
7694
        goto error;
7695

7696 7697 7698 7699
    /* XXX: Need to change as long as lock is introduced for
     * qemu_driver->sharedDevices.
     */
    for (i = 0; i < obj->def->ndisks; i++) {
7700
        virDomainDiskDefPtr disk = obj->def->disks[i];
7701
        virDomainDeviceDef dev;
7702

7703
        if (virDomainDiskTranslateSourcePool(disk) < 0)
7704
            goto error;
7705

7706 7707 7708 7709 7710
        /* 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 */
7711 7712
            virStorageSourceBackingStoreClear(disk->src);
            if (qemuDomainDetermineDiskChain(driver, obj, disk, false) < 0)
7713 7714 7715 7716
                goto error;
        } else {
            VIR_DEBUG("skipping backing chain detection for '%s'", disk->dst);
        }
7717

7718
        dev.type = VIR_DOMAIN_DEVICE_DISK;
7719
        dev.data.disk = disk;
7720
        if (qemuAddSharedDevice(driver, &dev, obj->def->name) < 0)
7721 7722 7723 7724
            goto error;
    }

    for (i = 0; i < obj->def->ngraphics; i++) {
7725
        if (qemuProcessGraphicsReservePorts(obj->def->graphics[i], true) < 0)
7726
            goto error;
7727
    }
7728

7729
    if (qemuProcessUpdateState(driver, obj) < 0)
7730 7731
        goto error;

7732 7733 7734 7735 7736 7737
    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);
7738
        goto error;
7739
    }
7740

7741 7742 7743 7744
    /* If upgrading from old libvirtd we won't have found any
     * caps in the domain status, so re-query them
     */
    if (!priv->qemuCaps &&
7745
        !(priv->qemuCaps = virQEMUCapsCacheLookupCopy(driver->qemuCapsCache,
7746 7747
                                                      obj->def->emulator,
                                                      obj->def->os.machine)))
7748
        goto error;
7749

7750 7751 7752 7753 7754 7755 7756 7757 7758 7759 7760 7761
    /* 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;
    }
7762

7763
    if (qemuProcessBuildDestroyMemoryPaths(driver, obj, NULL, true) < 0)
7764
        goto error;
7765

7766 7767
    if ((qemuDomainAssignAddresses(obj->def, priv->qemuCaps,
                                   driver, obj, false)) < 0) {
7768
        goto error;
7769
    }
7770

7771 7772 7773 7774 7775 7776
    /* if domain requests security driver we haven't loaded, report error, but
     * do not kill the domain
     */
    ignore_value(qemuSecurityCheckAllLabel(driver->securityManager,
                                           obj->def));

7777 7778
    if (qemuProcessRefreshCPU(driver, obj) < 0)
        goto error;
7779

7780
    if (qemuDomainRefreshVcpuInfo(driver, obj, QEMU_ASYNC_JOB_NONE, true) < 0)
7781
        goto error;
7782

7783
    qemuDomainVcpuPersistOrder(obj->def);
7784

7785 7786
    if (qemuSecurityReserveLabel(driver->securityManager, obj->def, obj->pid) < 0)
        goto error;
7787

7788 7789
    qemuProcessNotifyNets(obj->def);

7790
    if (qemuProcessFiltersInstantiate(obj->def, true))
7791
        goto error;
7792

7793
    if (qemuProcessRefreshDisks(driver, obj, QEMU_ASYNC_JOB_NONE) < 0)
7794
        goto error;
D
Daniel P. Berrange 已提交
7795

7796
    if (qemuBlockNodeNamesDetect(driver, obj, QEMU_ASYNC_JOB_NONE) < 0)
7797 7798
        goto error;

7799
    if (qemuRefreshVirtioChannelState(driver, obj, QEMU_ASYNC_JOB_NONE) < 0)
7800 7801
        goto error;

7802 7803 7804 7805
    /* 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)
7806
        goto error;
7807

7808
    if (qemuProcessRecoverJob(driver, obj, &oldjob, &stopFlags) < 0)
7809
        goto error;
7810

7811 7812
    if (qemuProcessUpdateDevices(driver, obj) < 0)
        goto error;
7813

7814 7815 7816
    if (qemuRefreshPRManagerState(driver, obj) < 0)
        goto error;

7817 7818 7819 7820 7821
    qemuProcessReconnectCheckMemAliasOrderMismatch(obj);

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

B
Bing Niu 已提交
7822 7823
    for (i = 0; i < obj->def->nresctrls; i++) {
        if (virResctrlAllocDeterminePath(obj->def->resctrls[i]->alloc,
7824 7825 7826 7827
                                         priv->machineName) < 0)
            goto error;
    }

7828 7829
    /* update domain state XML with possibly updated state in virDomainObj */
    if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, obj, driver->caps) < 0)
7830
        goto error;
7831

7832 7833
    /* Run an hook to allow admins to do some magic */
    if (virHookPresent(VIR_HOOK_DRIVER_QEMU)) {
7834
        char *xml = qemuDomainDefFormatXML(driver, obj->def, 0);
7835 7836
        int hookret;

7837 7838
        hookret = virHookCall(VIR_HOOK_DRIVER_QEMU, obj->def->name,
                              VIR_HOOK_QEMU_OP_RECONNECT, VIR_HOOK_SUBOP_BEGIN,
7839 7840 7841 7842 7843 7844 7845
                              NULL, xml, NULL);
        VIR_FREE(xml);

        /*
         * If the script raised an error abort the launch
         */
        if (hookret < 0)
7846
            goto error;
7847 7848
    }

7849 7850
    if (virAtomicIntInc(&driver->nactive) == 1 && driver->inhibitCallback)
        driver->inhibitCallback(true, driver->inhibitOpaque);
7851

7852
 cleanup:
7853 7854 7855
    if (jobStarted) {
        if (!virDomainObjIsActive(obj))
            qemuDomainRemoveInactive(driver, obj);
7856
        qemuDomainObjEndJob(driver, obj);
7857 7858 7859 7860
    } else {
        if (!virDomainObjIsActive(obj))
            qemuDomainRemoveInactiveJob(driver, obj);
    }
7861
    virDomainObjEndAPI(&obj);
7862
    virObjectUnref(cfg);
7863
    virObjectUnref(caps);
7864 7865
    virNWFilterUnlockFilterUpdates();
    return;
7866

7867 7868 7869 7870 7871
 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
7872 7873 7874 7875
         * 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;
7876 7877 7878 7879 7880
        /* 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);
7881
    }
7882
    goto cleanup;
7883
}
7884

7885 7886 7887
static int
qemuProcessReconnectHelper(virDomainObjPtr obj,
                           void *opaque)
7888
{
7889 7890 7891
    virThread thread;
    struct qemuProcessReconnectData *src = opaque;
    struct qemuProcessReconnectData *data;
7892

7893 7894 7895
    /* If the VM was inactive, we don't need to reconnect */
    if (!obj->pid)
        return 0;
7896

7897 7898
    if (VIR_ALLOC(data) < 0)
        return -1;
7899

7900 7901
    memcpy(data, src, sizeof(*data));
    data->obj = obj;
7902

7903 7904
    virNWFilterReadLockFilterUpdates();

7905 7906 7907 7908
    /* this lock and reference will be eventually transferred to the thread
     * that handles the reconnect */
    virObjectLock(obj);
    virObjectRef(obj);
7909

7910 7911 7912 7913 7914 7915 7916 7917 7918 7919 7920
    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);
7921
        qemuDomainRemoveInactiveJob(src->driver, obj);
7922

7923
        virDomainObjEndAPI(&obj);
7924
        virNWFilterUnlockFilterUpdates();
7925 7926
        VIR_FREE(data);
        return -1;
7927 7928
    }

7929 7930
    return 0;
}
7931

7932 7933 7934 7935 7936 7937 7938
/**
 * qemuProcessReconnectAll
 *
 * Try to re-open the resources for live VMs that we care
 * about.
 */
void
7939
qemuProcessReconnectAll(virQEMUDriverPtr driver)
7940
{
7941
    struct qemuProcessReconnectData data = {.driver = driver};
7942
    virDomainObjListForEach(driver->domains, qemuProcessReconnectHelper, &data);
7943
}