qemu_process.c 113.9 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
/*
 * qemu_process.h: QEMU process management
 *
 * Copyright (C) 2006-2011 Red Hat, Inc.
 *
 * 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
 * License along with this library; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307  USA
 *
 */

#include <config.h>

#include <fcntl.h>
#include <unistd.h>
#include <signal.h>
#include <sys/stat.h>
28 29
#include <sys/time.h>
#include <sys/resource.h>
30 31 32 33 34 35 36 37 38 39

#include "qemu_process.h"
#include "qemu_domain.h"
#include "qemu_cgroup.h"
#include "qemu_capabilities.h"
#include "qemu_monitor.h"
#include "qemu_command.h"
#include "qemu_hostdev.h"
#include "qemu_hotplug.h"
#include "qemu_bridge_filter.h"
40
#include "qemu_migration.h"
41

42
#if HAVE_NUMACTL
E
Eric Blake 已提交
43
# define NUMA_VERSION1_COMPATIBILITY 1
44 45 46
# include <numa.h>
#endif

47 48 49 50 51
#include "datatypes.h"
#include "logging.h"
#include "virterror_internal.h"
#include "memory.h"
#include "hooks.h"
E
Eric Blake 已提交
52
#include "virfile.h"
53
#include "virpidfile.h"
54 55 56 57
#include "util.h"
#include "c-ctype.h"
#include "nodeinfo.h"
#include "processinfo.h"
58
#include "domain_audit.h"
59
#include "domain_nwfilter.h"
60
#include "locking/domain_lock.h"
61
#include "network/bridge_driver.h"
62
#include "uuid.h"
63 64 65 66

#define VIR_FROM_THIS VIR_FROM_QEMU

#define START_POSTFIX ": starting up\n"
67
#define ATTACH_POSTFIX ": attaching\n"
68 69 70 71 72 73 74 75 76 77 78 79 80 81 82
#define SHUTDOWN_POSTFIX ": shutting down\n"

/**
 * qemudRemoveDomainStatus
 *
 * remove all state files of a domain from statedir
 *
 * Returns 0 on success
 */
static int
qemuProcessRemoveDomainStatus(struct qemud_driver *driver,
                              virDomainObjPtr vm)
{
    char ebuf[1024];
    char *file = NULL;
83
    qemuDomainObjPrivatePtr priv = vm->privateData;
84 85 86 87 88 89 90 91 92 93 94

    if (virAsprintf(&file, "%s/%s.xml", driver->stateDir, vm->def->name) < 0) {
        virReportOOMError();
        return(-1);
    }

    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);

95 96 97
    if (priv->pidfile &&
        unlink(priv->pidfile) < 0 &&
        errno != ENOENT)
98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115
        VIR_WARN("Failed to remove PID file for %s: %s",
                 vm->def->name, virStrerror(errno, ebuf, sizeof(ebuf)));

    return 0;
}


/* XXX figure out how to remove this */
extern struct qemud_driver *qemu_driver;

/*
 * This is a callback registered with a qemuMonitorPtr  instance,
 * and to be invoked when the monitor console hits an end of file
 * condition, or error, thus indicating VM shutdown should be
 * performed
 */
static void
qemuProcessHandleMonitorEOF(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
116
                            virDomainObjPtr vm)
117 118 119 120
{
    struct qemud_driver *driver = qemu_driver;
    virDomainEventPtr event = NULL;
    qemuDomainObjPrivatePtr priv;
121 122 123
    int eventReason = VIR_DOMAIN_EVENT_STOPPED_SHUTDOWN;
    int stopReason = VIR_DOMAIN_SHUTOFF_SHUTDOWN;
    const char *auditReason = "shutdown";
124 125 126

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

127
    qemuDriverLock(driver);
128 129 130 131 132
    virDomainObjLock(vm);

    if (!virDomainObjIsActive(vm)) {
        VIR_DEBUG("Domain %p is not active, ignoring EOF", vm);
        virDomainObjUnlock(vm);
133
        qemuDriverUnlock(driver);
134 135 136 137
        return;
    }

    priv = vm->privateData;
138
    if (priv->monJSON && !priv->gotShutdown) {
139 140
        VIR_DEBUG("Monitor connection to '%s' closed without SHUTDOWN event; "
                  "assuming the domain crashed", vm->def->name);
141
        eventReason = VIR_DOMAIN_EVENT_STOPPED_FAILED;
142
        stopReason = VIR_DOMAIN_SHUTOFF_CRASHED;
143
        auditReason = "failed";
144 145 146 147
    }

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
148 149
                                     eventReason);
    qemuProcessStop(driver, vm, 0, stopReason);
150
    virDomainAuditStop(vm, auditReason);
151 152 153 154 155 156

    if (!vm->persistent)
        virDomainRemoveInactive(&driver->domains, vm);
    else
        virDomainObjUnlock(vm);

157
    if (event)
158
        qemuDomainEventQueue(driver, event);
159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180
    qemuDriverUnlock(driver);
}


/*
 * 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,
                              virDomainObjPtr vm)
{
    struct qemud_driver *driver = qemu_driver;
    virDomainEventPtr event = NULL;

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

    qemuDriverLock(driver);
    virDomainObjLock(vm);

181
    ((qemuDomainObjPrivatePtr) vm->privateData)->monError = true;
182 183 184 185 186
    event = virDomainEventControlErrorNewFromObj(vm);
    if (event)
        qemuDomainEventQueue(driver, event);

    virDomainObjUnlock(vm);
187
    qemuDriverUnlock(driver);
188 189 190 191 192 193 194
}


static virDomainDiskDefPtr
qemuProcessFindDomainDiskByPath(virDomainObjPtr vm,
                                const char *path)
{
195
    int i = virDomainDiskIndexByName(vm->def, path, true);
196

197 198
    if (i >= 0)
        return vm->def->disks[i];
199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258

    qemuReportError(VIR_ERR_INTERNAL_ERROR,
                    _("no disk found with path %s"),
                    path);
    return NULL;
}

static virDomainDiskDefPtr
qemuProcessFindDomainDiskByAlias(virDomainObjPtr vm,
                                 const char *alias)
{
    int i;

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

    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;
    }

    qemuReportError(VIR_ERR_INTERNAL_ERROR,
                    _("no disk found with alias %s"),
                    alias);
    return NULL;
}

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

    if (!disk->encryption) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("disk %s does not have any encryption information"),
                        disk->src);
        return -1;
    }
    enc = disk->encryption;

    if (!conn) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("cannot find secrets without a connection"));
        goto cleanup;
    }

    if (conn->secretDriver == NULL ||
        conn->secretDriver->lookupByUUID == NULL ||
        conn->secretDriver->getValue == NULL) {
259
        qemuReportError(VIR_ERR_OPERATION_INVALID, "%s",
260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276
                        _("secret storage not supported"));
        goto cleanup;
    }

    if (enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_QCOW ||
        enc->nsecrets != 1 ||
        enc->secrets[0]->type !=
        VIR_STORAGE_ENCRYPTION_SECRET_TYPE_PASSPHRASE) {
        qemuReportError(VIR_ERR_XML_ERROR,
                        _("invalid <encryption> for volume %s"), disk->src);
        goto cleanup;
    }

    secret = conn->secretDriver->lookupByUUID(conn,
                                              enc->secrets[0]->uuid);
    if (secret == NULL)
        goto cleanup;
277
    data = conn->secretDriver->getValue(secret, &size, 0,
278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358
                                        VIR_SECRET_GET_VALUE_INTERNAL_CALL);
    virUnrefSecret(secret);
    if (data == NULL)
        goto cleanup;

    if (memchr(data, '\0', size) != NULL) {
        memset(data, 0, size);
        VIR_FREE(data);
        qemuReportError(VIR_ERR_XML_ERROR,
                        _("format='qcow' passphrase for %s must not contain a "
                          "'\\0'"), disk->src);
        goto cleanup;
    }

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

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

    *secretRet = passphrase;
    *secretLen = size;

    ret = 0;

cleanup:
    return ret;
}

static int
qemuProcessFindVolumeQcowPassphrase(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                                    virConnectPtr conn,
                                    virDomainObjPtr vm,
                                    const char *path,
                                    char **secretRet,
                                    size_t *secretLen)
{
    virDomainDiskDefPtr disk;
    int ret = -1;

    virDomainObjLock(vm);
    disk = qemuProcessFindDomainDiskByPath(vm, path);

    if (!disk)
        goto cleanup;

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

cleanup:
    virDomainObjUnlock(vm);
    return ret;
}


static int
qemuProcessHandleReset(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                       virDomainObjPtr vm)
{
    struct qemud_driver *driver = qemu_driver;
    virDomainEventPtr event;

    virDomainObjLock(vm);
    event = virDomainEventRebootNewFromObj(vm);
    virDomainObjUnlock(vm);

    if (event) {
        qemuDriverLock(driver);
        qemuDomainEventQueue(driver, event);
        qemuDriverUnlock(driver);
    }

    return 0;
}


359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377
/*
 * 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)
{
    struct qemud_driver *driver = qemu_driver;
    virDomainObjPtr vm = opaque;
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virDomainEventPtr event = NULL;
    int ret = -1;
    VIR_DEBUG("vm=%p", vm);
    qemuDriverLock(driver);
    virDomainObjLock(vm);
378
    if (qemuDomainObjBeginJob(driver, vm, QEMU_JOB_MODIFY) < 0)
379 380 381 382 383 384 385 386
        goto cleanup;

    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("guest unexpectedly quit"));
        goto endjob;
    }

387
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
388 389 390 391 392 393 394 395 396 397 398 399 400
    if (qemuMonitorSystemReset(priv->mon) < 0) {
        qemuDomainObjExitMonitorWithDriver(driver, vm);
        goto endjob;
    }
    qemuDomainObjExitMonitorWithDriver(driver, vm);

    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("guest unexpectedly quit"));
        goto endjob;
    }

    if (qemuProcessStartCPUs(driver, vm, NULL,
401 402
                             VIR_DOMAIN_RUNNING_BOOTED,
                             QEMU_ASYNC_JOB_NONE) < 0) {
403 404 405 406 407 408 409 410 411 412 413 414
        if (virGetLastError() == NULL)
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            "%s", _("resume operation failed"));
        goto endjob;
    }
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_RESUMED,
                                     VIR_DOMAIN_EVENT_RESUMED_UNPAUSED);

    ret = 0;

endjob:
415
    if (qemuDomainObjEndJob(driver, vm) == 0)
416 417 418 419 420 421 422 423 424 425 426 427 428 429 430
        vm = NULL;

cleanup:
    if (vm) {
        if (ret == -1)
            qemuProcessKill(vm);
        if (virDomainObjUnref(vm) > 0)
            virDomainObjUnlock(vm);
    }
    if (event)
        qemuDomainEventQueue(driver, event);
    qemuDriverUnlock(driver);
}


431 432 433 434
static int
qemuProcessHandleShutdown(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                          virDomainObjPtr vm)
{
435 436 437
    qemuDomainObjPrivatePtr priv = vm->privateData;
    VIR_DEBUG("vm=%p", vm);

438
    virDomainObjLock(vm);
439 440 441 442 443 444 445 446
    priv->gotShutdown = true;
    if (priv->fakeReboot) {
        virDomainObjRef(vm);
        virThread th;
        if (virThreadCreate(&th,
                            false,
                            qemuProcessFakeReboot,
                            vm) < 0) {
447
            VIR_ERROR(_("Failed to create reboot thread, killing domain"));
448 449 450 451 452 453 454 455 456
            qemuProcessKill(vm);
            if (virDomainObjUnref(vm) == 0)
                vm = NULL;
        }
    } else {
        qemuProcessKill(vm);
    }
    if (vm)
        virDomainObjUnlock(vm);
457 458 459 460 461 462 463 464 465 466 467 468 469

    return 0;
}


static int
qemuProcessHandleStop(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                      virDomainObjPtr vm)
{
    struct qemud_driver *driver = qemu_driver;
    virDomainEventPtr event = NULL;

    virDomainObjLock(vm);
J
Jiri Denemark 已提交
470
    if (virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING) {
471
        qemuDomainObjPrivatePtr priv = vm->privateData;
472 473
        VIR_DEBUG("Transitioned guest %s to paused state due to unknown event",
                  vm->def->name);
474

J
Jiri Denemark 已提交
475
        virDomainObjSetState(vm, VIR_DOMAIN_PAUSED, VIR_DOMAIN_PAUSED_UNKNOWN);
476 477 478 479
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_SUSPENDED,
                                         VIR_DOMAIN_EVENT_SUSPENDED_PAUSED);

480 481 482 483 484
        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));

485 486 487 488
        if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0) {
            VIR_WARN("Unable to save status on vm %s after state change",
                     vm->def->name);
        }
489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517
    }
    virDomainObjUnlock(vm);

    if (event) {
        qemuDriverLock(driver);
        if (event)
            qemuDomainEventQueue(driver, event);
        qemuDriverUnlock(driver);
    }

    return 0;
}


static int
qemuProcessHandleRTCChange(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                           virDomainObjPtr vm,
                           long long offset)
{
    struct qemud_driver *driver = qemu_driver;
    virDomainEventPtr event;

    virDomainObjLock(vm);
    event = virDomainEventRTCChangeNewFromObj(vm, offset);

    if (vm->def->clock.offset == VIR_DOMAIN_CLOCK_OFFSET_VARIABLE)
        vm->def->clock.data.adjustment = offset;

    if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
518
        VIR_WARN("unable to save domain status with RTC change");
519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544

    virDomainObjUnlock(vm);

    if (event) {
        qemuDriverLock(driver);
        qemuDomainEventQueue(driver, event);
        qemuDriverUnlock(driver);
    }

    return 0;
}


static int
qemuProcessHandleWatchdog(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                          virDomainObjPtr vm,
                          int action)
{
    struct qemud_driver *driver = qemu_driver;
    virDomainEventPtr watchdogEvent = NULL;
    virDomainEventPtr lifecycleEvent = NULL;

    virDomainObjLock(vm);
    watchdogEvent = virDomainEventWatchdogNewFromObj(vm, action);

    if (action == VIR_DOMAIN_EVENT_WATCHDOG_PAUSE &&
J
Jiri Denemark 已提交
545
        virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING) {
546
        qemuDomainObjPrivatePtr priv = vm->privateData;
547 548
        VIR_DEBUG("Transitioned guest %s to paused state due to watchdog", vm->def->name);

J
Jiri Denemark 已提交
549
        virDomainObjSetState(vm, VIR_DOMAIN_PAUSED, VIR_DOMAIN_PAUSED_WATCHDOG);
550 551 552 553
        lifecycleEvent = virDomainEventNewFromObj(vm,
                                                  VIR_DOMAIN_EVENT_SUSPENDED,
                                                  VIR_DOMAIN_EVENT_SUSPENDED_WATCHDOG);

554 555 556 557 558
        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));

559 560 561 562
        if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0) {
            VIR_WARN("Unable to save status on vm %s after watchdog event",
                     vm->def->name);
        }
563 564 565 566 567 568 569
    }

    if (vm->def->watchdog->action == VIR_DOMAIN_WATCHDOG_ACTION_DUMP) {
        struct qemuDomainWatchdogEvent *wdEvent;
        if (VIR_ALLOC(wdEvent) == 0) {
            wdEvent->action = VIR_DOMAIN_WATCHDOG_ACTION_DUMP;
            wdEvent->vm = vm;
W
Wen Congyang 已提交
570 571 572 573
            /* Hold an extra reference because we can't allow 'vm' to be
             * deleted before handling watchdog event is finished.
             */
            virDomainObjRef(vm);
574
            if (virThreadPoolSendJob(driver->workerPool, 0, wdEvent) < 0) {
575
                if (virDomainObjUnref(vm) == 0)
576
                    vm = NULL;
W
Wen Congyang 已提交
577 578
                VIR_FREE(wdEvent);
            }
579
        } else {
580
            virReportOOMError();
581
        }
582 583
    }

584 585
    if (vm)
        virDomainObjUnlock(vm);
586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629

    if (watchdogEvent || lifecycleEvent) {
        qemuDriverLock(driver);
        if (watchdogEvent)
            qemuDomainEventQueue(driver, watchdogEvent);
        if (lifecycleEvent)
            qemuDomainEventQueue(driver, lifecycleEvent);
        qemuDriverUnlock(driver);
    }

    return 0;
}


static int
qemuProcessHandleIOError(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                         virDomainObjPtr vm,
                         const char *diskAlias,
                         int action,
                         const char *reason)
{
    struct qemud_driver *driver = qemu_driver;
    virDomainEventPtr ioErrorEvent = NULL;
    virDomainEventPtr ioErrorEvent2 = NULL;
    virDomainEventPtr lifecycleEvent = NULL;
    const char *srcPath;
    const char *devAlias;
    virDomainDiskDefPtr disk;

    virDomainObjLock(vm);
    disk = qemuProcessFindDomainDiskByAlias(vm, diskAlias);

    if (disk) {
        srcPath = disk->src;
        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
630
        virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING) {
631
        qemuDomainObjPrivatePtr priv = vm->privateData;
632 633
        VIR_DEBUG("Transitioned guest %s to paused state due to IO error", vm->def->name);

J
Jiri Denemark 已提交
634
        virDomainObjSetState(vm, VIR_DOMAIN_PAUSED, VIR_DOMAIN_PAUSED_IOERROR);
635 636 637 638
        lifecycleEvent = virDomainEventNewFromObj(vm,
                                                  VIR_DOMAIN_EVENT_SUSPENDED,
                                                  VIR_DOMAIN_EVENT_SUSPENDED_IOERROR);

639 640 641 642 643
        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));

644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662
        if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
            VIR_WARN("Unable to save status on vm %s after IO error", vm->def->name);
    }
    virDomainObjUnlock(vm);

    if (ioErrorEvent || ioErrorEvent2 || lifecycleEvent) {
        qemuDriverLock(driver);
        if (ioErrorEvent)
            qemuDomainEventQueue(driver, ioErrorEvent);
        if (ioErrorEvent2)
            qemuDomainEventQueue(driver, ioErrorEvent2);
        if (lifecycleEvent)
            qemuDomainEventQueue(driver, lifecycleEvent);
        qemuDriverUnlock(driver);
    }

    return 0;
}

663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692
static int
qemuProcessHandleBlockJob(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                          virDomainObjPtr vm,
                          const char *diskAlias,
                          int type,
                          int status)
{
    struct qemud_driver *driver = qemu_driver;
    virDomainEventPtr event = NULL;
    const char *path;
    virDomainDiskDefPtr disk;

    virDomainObjLock(vm);
    disk = qemuProcessFindDomainDiskByAlias(vm, diskAlias);

    if (disk) {
        path = disk->src;
        event = virDomainEventBlockJobNewFromObj(vm, path, type, status);
    }

    virDomainObjUnlock(vm);

    if (event) {
        qemuDriverLock(driver);
        qemuDomainEventQueue(driver, event);
        qemuDriverUnlock(driver);
    }

    return 0;
}
693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734

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,
                          const char *saslUsername)
{
    struct qemud_driver *driver = qemu_driver;
    virDomainEventPtr event;
    virDomainEventGraphicsAddressPtr localAddr = NULL;
    virDomainEventGraphicsAddressPtr remoteAddr = NULL;
    virDomainEventGraphicsSubjectPtr subject = NULL;
    int i;

    if (VIR_ALLOC(localAddr) < 0)
        goto no_memory;
    localAddr->family = localFamily;
    if (!(localAddr->service = strdup(localService)) ||
        !(localAddr->node = strdup(localNode)))
        goto no_memory;

    if (VIR_ALLOC(remoteAddr) < 0)
        goto no_memory;
    remoteAddr->family = remoteFamily;
    if (!(remoteAddr->service = strdup(remoteService)) ||
        !(remoteAddr->node = strdup(remoteNode)))
        goto no_memory;

    if (VIR_ALLOC(subject) < 0)
        goto no_memory;
    if (x509dname) {
        if (VIR_REALLOC_N(subject->identities, subject->nidentity+1) < 0)
            goto no_memory;
        subject->nidentity++;
735 736 737
        if (!(subject->identities[subject->nidentity-1].type = strdup("x509dname")) ||
            !(subject->identities[subject->nidentity-1].name = strdup(x509dname)))
            goto no_memory;
738 739 740 741 742
    }
    if (saslUsername) {
        if (VIR_REALLOC_N(subject->identities, subject->nidentity+1) < 0)
            goto no_memory;
        subject->nidentity++;
743 744 745
        if (!(subject->identities[subject->nidentity-1].type = strdup("saslUsername")) ||
            !(subject->identities[subject->nidentity-1].name = strdup(saslUsername)))
            goto no_memory;
746 747
    }

748
    virDomainObjLock(vm);
749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787
    event = virDomainEventGraphicsNewFromObj(vm, phase, localAddr, remoteAddr, authScheme, subject);
    virDomainObjUnlock(vm);

    if (event) {
        qemuDriverLock(driver);
        qemuDomainEventQueue(driver, event);
        qemuDriverUnlock(driver);
    }

    return 0;

no_memory:
    virReportOOMError();
    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) {
        for (i = 0 ; i < subject->nidentity ; i++) {
            VIR_FREE(subject->identities[i].type);
            VIR_FREE(subject->identities[i].name);
        }
        VIR_FREE(subject->identities);
        VIR_FREE(subject);
    }

    return -1;
}


static void qemuProcessHandleMonitorDestroy(qemuMonitorPtr mon,
                                            virDomainObjPtr vm)
{
788 789 790 791
    qemuDomainObjPrivatePtr priv;

    virDomainObjLock(vm);
    priv = vm->privateData;
792 793
    if (priv->mon == mon)
        priv->mon = NULL;
794 795
    if (virDomainObjUnref(vm) > 0)
        virDomainObjUnlock(vm);
796 797 798 799 800
}

static qemuMonitorCallbacks monitorCallbacks = {
    .destroy = qemuProcessHandleMonitorDestroy,
    .eofNotify = qemuProcessHandleMonitorEOF,
801
    .errorNotify = qemuProcessHandleMonitorError,
802 803 804 805 806 807 808 809
    .diskSecretLookup = qemuProcessFindVolumeQcowPassphrase,
    .domainShutdown = qemuProcessHandleShutdown,
    .domainStop = qemuProcessHandleStop,
    .domainReset = qemuProcessHandleReset,
    .domainRTCChange = qemuProcessHandleRTCChange,
    .domainWatchdog = qemuProcessHandleWatchdog,
    .domainIOError = qemuProcessHandleIOError,
    .domainGraphics = qemuProcessHandleGraphics,
810
    .domainBlockJob = qemuProcessHandleBlockJob,
811 812 813 814 815 816 817
};

static int
qemuConnectMonitor(struct qemud_driver *driver, virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int ret = -1;
818
    qemuMonitorPtr mon = NULL;
819

820 821
    if (virSecurityManagerSetDaemonSocketLabel(driver->securityManager,
                                               vm) < 0) {
822 823 824 825 826 827 828 829 830
        VIR_ERROR(_("Failed to set security context for monitor for %s"),
                  vm->def->name);
        goto error;
    }

    /* Hold an extra reference because we can't allow 'vm' to be
     * deleted while the monitor is active */
    virDomainObjRef(vm);

831 832 833 834 835 836 837 838 839 840 841 842
    ignore_value(virTimeMs(&priv->monStart));
    virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);

    mon = qemuMonitorOpen(vm,
                          priv->monConfig,
                          priv->monJSON,
                          &monitorCallbacks);

    qemuDriverLock(driver);
    virDomainObjLock(vm);
    priv->monStart = 0;
843

844
    /* Safe to ignore value since ref count was incremented above */
845
    if (mon == NULL)
846
        ignore_value(virDomainObjUnref(vm));
847

848 849 850 851 852 853
    if (!virDomainObjIsActive(vm)) {
        qemuMonitorClose(mon);
        goto error;
    }
    priv->mon = mon;

854 855 856 857 858 859 860 861 862 863 864 865
    if (virSecurityManagerClearSocketLabel(driver->securityManager, vm) < 0) {
        VIR_ERROR(_("Failed to clear security context for monitor for %s"),
                  vm->def->name);
        goto error;
    }

    if (priv->mon == NULL) {
        VIR_INFO("Failed to connect monitor for %s", vm->def->name);
        goto error;
    }


866
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892
    ret = qemuMonitorSetCapabilities(priv->mon);
    qemuDomainObjExitMonitorWithDriver(driver, vm);

error:

    return ret;
}

typedef int qemuProcessLogHandleOutput(virDomainObjPtr vm,
                                       const char *output,
                                       int fd);

/*
 * Returns -1 for error, 0 on success
 */
static int
qemuProcessReadLogOutput(virDomainObjPtr vm,
                         int fd,
                         char *buf,
                         size_t buflen,
                         qemuProcessLogHandleOutput func,
                         const char *what,
                         int timeout)
{
    int retries = (timeout*10);
    int got = 0;
893 894 895 896
    char *debug = NULL;
    int ret = -1;
    char *filter_next = buf;

897 898
    buf[0] = '\0';

899 900 901 902 903 904 905
    /* This relies on log message format generated by virLogFormatString() and
     * might need to be modified when message format changes. */
    if (virAsprintf(&debug, ": %d: debug : ", vm->pid) < 0) {
        virReportOOMError();
        return -1;
    }

906
    while (retries) {
907
        ssize_t func_ret, bytes;
908
        int isdead = 0;
909
        char *eol;
910 911 912 913 914 915 916 917

        func_ret = func(vm, buf, fd);

        if (kill(vm->pid, 0) == -1 && errno == ESRCH)
            isdead = 1;

        /* Any failures should be detected before we read the log, so we
         * always have something useful to report on failure. */
918 919
        bytes = saferead(fd, buf+got, buflen-got-1);
        if (bytes < 0) {
920 921 922
            virReportSystemError(errno,
                                 _("Failure while reading %s log output"),
                                 what);
923
            goto cleanup;
924 925
        }

926
        got += bytes;
927
        buf[got] = '\0';
928 929 930 931 932 933 934 935 936 937 938 939 940

        /* Filter out debug messages from intermediate libvirt process */
        while ((eol = strchr(filter_next, '\n'))) {
            *eol = '\0';
            if (strstr(filter_next, debug)) {
                memmove(filter_next, eol + 1, got - (eol - buf));
                got -= eol + 1 - filter_next;
            } else {
                filter_next = eol + 1;
                *eol = '\n';
            }
        }

941 942 943 944
        if (got == buflen-1) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("Out of space while reading %s log output: %s"),
                            what, buf);
945
            goto cleanup;
946 947 948 949 950 951
        }

        if (isdead) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("Process exited while reading %s log output: %s"),
                            what, buf);
952
            goto cleanup;
953 954
        }

955 956 957 958
        if (func_ret <= 0) {
            ret = func_ret;
            goto cleanup;
        }
959 960 961 962 963 964 965 966

        usleep(100*1000);
        retries--;
    }

    qemuReportError(VIR_ERR_INTERNAL_ERROR,
                    _("Timed out while reading %s log output: %s"),
                    what, buf);
967 968 969 970

cleanup:
    VIR_FREE(debug);
    return ret;
971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025
}


/*
 * Look at a chunk of data from the QEMU stdout logs and try to
 * find a TTY device, as indicated by a line like
 *
 * char device redirected to /dev/pts/3
 *
 * Returns -1 for error, 0 success, 1 continue reading
 */
static int
qemuProcessExtractTTYPath(const char *haystack,
                          size_t *offset,
                          char **path)
{
    static const char needle[] = "char device redirected to";
    char *tmp, *dev;

    VIR_FREE(*path);
    /* First look for our magic string */
    if (!(tmp = strstr(haystack + *offset, needle))) {
        return 1;
    }
    tmp += sizeof(needle);
    dev = tmp;

    /*
     * And look for first whitespace character and nul terminate
     * to mark end of the pty path
     */
    while (*tmp) {
        if (c_isspace(*tmp)) {
            *path = strndup(dev, tmp-dev);
            if (*path == NULL) {
                virReportOOMError();
                return -1;
            }

            /* ... now further update offset till we get EOL */
            *offset = tmp - haystack;
            return 0;
        }
        tmp++;
    }

    /*
     * We found a path, but didn't find any whitespace,
     * so it must be still incomplete - we should at
     * least see a \n - indicate that we want to carry
     * on trying again
     */
    return 1;
}

1026 1027 1028
static int
qemuProcessLookupPTYs(virDomainChrDefPtr *devices,
                      int count,
C
Cole Robinson 已提交
1029 1030
                      virHashTablePtr paths,
                      bool chardevfmt)
1031 1032
{
    int i;
C
Cole Robinson 已提交
1033
    const char *prefix = chardevfmt ? "char" : "";
1034 1035 1036 1037

    for (i = 0 ; i < count ; i++) {
        virDomainChrDefPtr chr = devices[i];
        if (chr->source.type == VIR_DOMAIN_CHR_TYPE_PTY) {
C
Cole Robinson 已提交
1038
            char id[32];
1039 1040
            const char *path;

C
Cole Robinson 已提交
1041 1042
            if (snprintf(id, sizeof(id), "%s%s",
                         prefix, chr->info.alias) >= sizeof(id))
1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074
                return -1;

            path = (const char *) virHashLookup(paths, id);
            if (path == NULL) {
                if (chr->source.data.file.path == NULL) {
                    /* neither the log output nor 'info chardev' had a
                     * pty path for this chardev, report an error
                     */
                    qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                    _("no assigned pty for device %s"), id);
                    return -1;
                } else {
                    /* 'info chardev' had no pty path for this chardev,
                     * but the log output had, so we're fine
                     */
                    continue;
                }
            }

            VIR_FREE(chr->source.data.file.path);
            chr->source.data.file.path = strdup(path);

            if (chr->source.data.file.path == NULL) {
                virReportOOMError();
                return -1;
            }
        }
    }

    return 0;
}

1075 1076
static int
qemuProcessFindCharDevicePTYsMonitor(virDomainObjPtr vm,
C
Cole Robinson 已提交
1077
                                     virBitmapPtr qemuCaps,
1078 1079
                                     virHashTablePtr paths)
{
C
Cole Robinson 已提交
1080 1081
    bool chardevfmt = qemuCapsGet(qemuCaps, QEMU_CAPS_CHARDEV);

1082
    if (qemuProcessLookupPTYs(vm->def->serials, vm->def->nserials,
C
Cole Robinson 已提交
1083
                              paths, chardevfmt) < 0)
1084 1085 1086
        return -1;

    if (qemuProcessLookupPTYs(vm->def->parallels, vm->def->nparallels,
C
Cole Robinson 已提交
1087
                              paths, chardevfmt) < 0)
1088
        return -1;
1089

1090
    if (qemuProcessLookupPTYs(vm->def->channels, vm->def->nchannels,
C
Cole Robinson 已提交
1091
                              paths, chardevfmt) < 0)
1092 1093 1094
        return -1;

    if (vm->def->console &&
C
Cole Robinson 已提交
1095
        qemuProcessLookupPTYs(&vm->def->console, 1, paths, chardevfmt) < 0)
1096
        return -1;
1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142

    return 0;
}

static int
qemuProcessFindCharDevicePTYs(virDomainObjPtr vm,
                              const char *output,
                              int fd ATTRIBUTE_UNUSED)
{
    size_t offset = 0;
    int ret, i;

    /* The order in which QEMU prints out the PTY paths is
       the order in which it procsses its serial and parallel
       device args. This code must match that ordering.... */

    /* first comes the serial devices */
    for (i = 0 ; i < vm->def->nserials ; i++) {
        virDomainChrDefPtr chr = vm->def->serials[i];
        if (chr->source.type == VIR_DOMAIN_CHR_TYPE_PTY) {
            if ((ret = qemuProcessExtractTTYPath(output, &offset,
                                                 &chr->source.data.file.path)) != 0)
                return ret;
        }
    }

    /* then the parallel devices */
    for (i = 0 ; i < vm->def->nparallels ; i++) {
        virDomainChrDefPtr chr = vm->def->parallels[i];
        if (chr->source.type == VIR_DOMAIN_CHR_TYPE_PTY) {
            if ((ret = qemuProcessExtractTTYPath(output, &offset,
                                                 &chr->source.data.file.path)) != 0)
                return ret;
        }
    }

    /* then the channel devices */
    for (i = 0 ; i < vm->def->nchannels ; i++) {
        virDomainChrDefPtr chr = vm->def->channels[i];
        if (chr->source.type == VIR_DOMAIN_CHR_TYPE_PTY) {
            if ((ret = qemuProcessExtractTTYPath(output, &offset,
                                                 &chr->source.data.file.path)) != 0)
                return ret;
        }
    }

1143 1144 1145 1146 1147 1148 1149 1150 1151 1152
    if (vm->def->console) {
        virDomainChrDefPtr chr = vm->def->console;
        if (chr->source.type == VIR_DOMAIN_CHR_TYPE_PTY &&
            chr->targetType == VIR_DOMAIN_CHR_CONSOLE_TARGET_TYPE_VIRTIO) {
            if ((ret = qemuProcessExtractTTYPath(output, &offset,
                                                 &chr->source.data.file.path)) != 0)
                return ret;
        }
    }

1153 1154 1155
    return 0;
}

1156
static void qemuProcessFreePtyPath(void *payload, const void *name ATTRIBUTE_UNUSED)
1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174
{
    VIR_FREE(payload);
}

static void
qemuProcessReadLogFD(int logfd, char *buf, int maxlen, int off)
{
    int ret;
    char *tmpbuf = buf + off;

    ret = saferead(logfd, tmpbuf, maxlen - off - 1);
    if (ret < 0) {
        ret = 0;
    }

    tmpbuf[ret] = '\0';
}

1175

1176 1177
static int
qemuProcessWaitForMonitor(struct qemud_driver* driver,
C
Cole Robinson 已提交
1178 1179 1180
                          virDomainObjPtr vm,
                          virBitmapPtr qemuCaps,
                          off_t pos)
1181
{
1182
    char *buf = NULL;
1183
    size_t buf_size = 4096; /* Plenty of space to get startup greeting */
1184
    int logfd = -1;
1185 1186
    int ret = -1;
    virHashTablePtr paths = NULL;
1187
    qemuDomainObjPrivatePtr priv;
1188

1189 1190 1191
    if (pos != -1) {
        if ((logfd = qemuDomainOpenLog(driver, vm, pos)) < 0)
            return -1;
1192

1193 1194 1195 1196
        if (VIR_ALLOC_N(buf, buf_size) < 0) {
            virReportOOMError();
            return -1;
        }
1197

1198 1199 1200 1201 1202
        if (qemuProcessReadLogOutput(vm, logfd, buf, buf_size,
                                     qemuProcessFindCharDevicePTYs,
                                     "console", 30) < 0)
            goto closelog;
    }
1203 1204 1205 1206 1207 1208 1209 1210 1211 1212

    VIR_DEBUG("Connect monitor to %p '%s'", vm, vm->def->name);
    if (qemuConnectMonitor(driver, vm) < 0) {
        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. */
1213
    paths = virHashCreate(0, qemuProcessFreePtyPath);
1214
    if (paths == NULL)
1215 1216
        goto cleanup;

1217
    priv = vm->privateData;
1218
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
1219 1220 1221 1222 1223
    ret = qemuMonitorGetPtyPaths(priv->mon, paths);
    qemuDomainObjExitMonitorWithDriver(driver, vm);

    VIR_DEBUG("qemuMonitorGetPtyPaths returned %i", ret);
    if (ret == 0)
C
Cole Robinson 已提交
1224
        ret = qemuProcessFindCharDevicePTYsMonitor(vm, qemuCaps, paths);
1225 1226

cleanup:
1227
    virHashFree(paths);
1228

E
Eric Blake 已提交
1229
    if (pos != -1 && kill(vm->pid, 0) == -1 && errno == ESRCH) {
1230 1231
        /* VM is dead, any other error raised in the interim is probably
         * not as important as the qemu cmdline output */
1232
        qemuProcessReadLogFD(logfd, buf, buf_size, strlen(buf));
1233 1234 1235 1236 1237 1238
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("process exited while connecting to monitor: %s"),
                        buf);
        ret = -1;
    }

1239
closelog:
1240
    if (VIR_CLOSE(logfd) < 0) {
1241
        char ebuf[1024];
1242 1243 1244 1245
        VIR_WARN("Unable to close logfile: %s",
                 virStrerror(errno, ebuf, sizeof ebuf));
    }

1246 1247
    VIR_FREE(buf);

1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270
    return ret;
}

static int
qemuProcessDetectVcpuPIDs(struct qemud_driver *driver,
                          virDomainObjPtr vm)
{
    pid_t *cpupids = NULL;
    int ncpupids;
    qemuDomainObjPrivatePtr priv = vm->privateData;

    if (vm->def->virtType != VIR_DOMAIN_VIRT_KVM) {
        priv->nvcpupids = 1;
        if (VIR_ALLOC_N(priv->vcpupids, priv->nvcpupids) < 0) {
            virReportOOMError();
            return -1;
        }
        priv->vcpupids[0] = vm->pid;
        return 0;
    }

    /* What follows is now all KVM specific */

1271
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295
    if ((ncpupids = qemuMonitorGetCPUInfo(priv->mon, &cpupids)) < 0) {
        qemuDomainObjExitMonitorWithDriver(driver, vm);
        return -1;
    }
    qemuDomainObjExitMonitorWithDriver(driver, vm);

    /* Treat failure to get VCPU<->PID mapping as non-fatal */
    if (ncpupids == 0)
        return 0;

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

    priv->nvcpupids = ncpupids;
    priv->vcpupids = cpupids;
    return 0;
}

1296 1297 1298 1299 1300 1301 1302 1303 1304

/*
 * Set NUMA memory policy for qemu process, to be run between
 * fork/exec of QEMU only.
 */
#if HAVE_NUMACTL
static int
qemuProcessInitNumaMemoryPolicy(virDomainObjPtr vm)
{
E
Eric Blake 已提交
1305
    nodemask_t mask;
1306 1307 1308 1309 1310
    int mode = -1;
    int node = -1;
    int ret = -1;
    int i = 0;
    int maxnode = 0;
E
Eric Blake 已提交
1311
    bool warned = false;
1312

1313 1314 1315
    if (!vm->def->numatune.memory.nodemask)
        return 0;

1316 1317 1318 1319 1320 1321 1322 1323
    VIR_DEBUG("Setting NUMA memory policy");

    if (numa_available() < 0) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("Host kernel is not aware of NUMA."));
        return -1;
    }

E
Eric Blake 已提交
1324
    maxnode = numa_max_node() + 1;
1325 1326

    /* Convert nodemask to NUMA bitmask. */
E
Eric Blake 已提交
1327
    nodemask_zero(&mask);
1328 1329
    for (i = 0; i < VIR_DOMAIN_CPUMASK_LEN; i++) {
        if (vm->def->numatune.memory.nodemask[i]) {
E
Eric Blake 已提交
1330 1331 1332 1333 1334 1335
            if (i > NUMA_NUM_NODES) {
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("Host cannot support NUMA node %d"), i);
                return -1;
            }
            if (i > maxnode && !warned) {
1336 1337
                VIR_WARN("nodeset is out of range, there is only %d NUMA "
                         "nodes on host", maxnode);
E
Eric Blake 已提交
1338
                warned = true;
1339
             }
E
Eric Blake 已提交
1340
            nodemask_set(&mask, i);
1341 1342 1343 1344 1345 1346 1347
        }
    }

    mode = vm->def->numatune.memory.mode;

    if (mode == VIR_DOMAIN_NUMATUNE_MEM_STRICT) {
        numa_set_bind_policy(1);
E
Eric Blake 已提交
1348
        numa_set_membind(&mask);
1349 1350 1351
        numa_set_bind_policy(0);
    } else if (mode == VIR_DOMAIN_NUMATUNE_MEM_PREFERRED) {
        int nnodes = 0;
E
Eric Blake 已提交
1352 1353
        for (i = 0; i < NUMA_NUM_NODES; i++) {
            if (nodemask_isset(&mask, i)) {
1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368
                node = i;
                nnodes++;
            }
        }

        if (nnodes != 1) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            "%s", _("NUMA memory tuning in 'preferred' mode "
                                    "only supports single node"));
            goto cleanup;
        }

        numa_set_bind_policy(0);
        numa_set_preferred(node);
    } else if (mode == VIR_DOMAIN_NUMATUNE_MEM_INTERLEAVE) {
E
Eric Blake 已提交
1369
        numa_set_interleave_mask(&mask);
1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398
    } else {
        /* XXX: Shouldn't go here, as we already do checking when
         * parsing domain XML.
         */
        qemuReportError(VIR_ERR_XML_ERROR,
                        "%s", _("Invalid mode for memory NUMA tuning."));
        goto cleanup;
    }

    ret = 0;

cleanup:
    return ret;
}
#else
static int
qemuProcessInitNumaMemoryPolicy(virDomainObjPtr vm)
{
    if (vm->def->numatune.memory.nodemask) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("libvirt is compiled without NUMA tuning support"));

        return -1;
    }

    return 0;
}
#endif

1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409
/*
 * To be run between fork/exec of QEMU only
 */
static int
qemuProcessInitCpuAffinity(virDomainObjPtr vm)
{
    int i, hostcpus, maxcpu = QEMUD_CPUMASK_LEN;
    virNodeInfo nodeinfo;
    unsigned char *cpumap;
    int cpumaplen;

1410
    VIR_DEBUG("Setting CPU affinity");
1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455

    if (nodeGetInfo(NULL, &nodeinfo) < 0)
        return -1;

    /* setaffinity fails if you set bits for CPUs which
     * aren't present, so we have to limit ourselves */
    hostcpus = VIR_NODEINFO_MAXCPUS(nodeinfo);
    if (maxcpu > hostcpus)
        maxcpu = hostcpus;

    cpumaplen = VIR_CPU_MAPLEN(maxcpu);
    if (VIR_ALLOC_N(cpumap, cpumaplen) < 0) {
        virReportOOMError();
        return -1;
    }

    if (vm->def->cpumask) {
        /* XXX why don't we keep 'cpumask' in the libvirt cpumap
         * format to start with ?!?! */
        for (i = 0 ; i < maxcpu && i < vm->def->cpumasklen ; i++)
            if (vm->def->cpumask[i])
                VIR_USE_CPU(cpumap, i);
    } 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 */
        for (i = 0 ; i < maxcpu ; i++)
            VIR_USE_CPU(cpumap, i);
    }

    /* We are pressuming we are running between fork/exec of QEMU
     * so use '0' to indicate our own process ID. No threads are
     * running at this point
     */
    if (virProcessInfoSetAffinity(0, /* Self */
                                  cpumap, cpumaplen, maxcpu) < 0) {
        VIR_FREE(cpumap);
        return -1;
    }
    VIR_FREE(cpumap);

    return 0;
}

1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489
/* set link states to down on interfaces at qemu start */
static int
qemuProcessSetLinkStates(virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virDomainDefPtr def = vm->def;
    int i;
    int ret = 0;

    for (i = 0; i < def->nnets; i++) {
        if (def->nets[i]->linkstate == VIR_DOMAIN_NET_INTERFACE_LINK_STATE_DOWN) {
            VIR_DEBUG("Setting link state: %s", def->nets[i]->info.alias);

            if (!qemuCapsGet(priv->qemuCaps, QEMU_CAPS_NETDEV)) {
                qemuReportError(VIR_ERR_NO_SUPPORT,
                                _("Setting of link state is not supported by this qemu"));
                return -1;
            }

            ret = qemuMonitorSetLink(priv->mon,
                                     def->nets[i]->info.alias,
                                     VIR_DOMAIN_NET_INTERFACE_LINK_STATE_DOWN);
            if (ret != 0) {
                qemuReportError(VIR_ERR_OPERATION_FAILED,
                               _("Couldn't set link state on interface: %s"), def->nets[i]->info.alias);
                break;
            }
        }
    }

    return ret;
}

/* Set CPU affinities for vcpus if vcpupin xml provided. */
1490 1491 1492 1493 1494 1495 1496 1497 1498
static int
qemuProcessSetVcpuAffinites(virConnectPtr conn,
                            virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virDomainDefPtr def = vm->def;
    virNodeInfo nodeinfo;
    pid_t vcpupid;
    unsigned char *cpumask;
1499
    int vcpu, cpumaplen, hostcpus, maxcpu, n;
E
Eric Blake 已提交
1500 1501
    unsigned char *cpumap = NULL;
    int ret = -1;
1502 1503 1504 1505 1506 1507 1508 1509 1510

    if (virNodeGetInfo(conn, &nodeinfo) != 0) {
        return  -1;
    }

    if (!def->cputune.nvcpupin)
        return 0;

    if (priv->vcpupids == NULL) {
1511
        qemuReportError(VIR_ERR_OPERATION_INVALID,
1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522
                        "%s", _("cpu affinity is not supported"));
        return -1;
    }

    hostcpus = VIR_NODEINFO_MAXCPUS(nodeinfo);
    cpumaplen = VIR_CPU_MAPLEN(hostcpus);
    maxcpu = cpumaplen * 8;

    if (maxcpu > hostcpus)
        maxcpu = hostcpus;

E
Eric Blake 已提交
1523 1524 1525 1526 1527
    if (VIR_ALLOC_N(cpumap, cpumaplen) < 0) {
        virReportOOMError();
        return -1;
    }

1528
    for (n = 0; n < def->cputune.nvcpupin; n++) {
1529
        int i;
1530
        vcpu = def->cputune.vcpupin[n]->vcpuid;
1531

E
Eric Blake 已提交
1532
        memset(cpumap, 0, cpumaplen);
1533
        cpumask = (unsigned char *)def->cputune.vcpupin[n]->cpumask;
1534 1535
        vcpupid = priv->vcpupids[vcpu];

1536
        for (i = 0 ; i < VIR_DOMAIN_CPUMASK_LEN ; i++) {
1537
            if (cpumask[i])
1538
                VIR_USE_CPU(cpumap, i);
1539 1540 1541 1542 1543 1544
        }

        if (virProcessInfoSetAffinity(vcpupid,
                                      cpumap,
                                      cpumaplen,
                                      maxcpu) < 0) {
E
Eric Blake 已提交
1545
            goto cleanup;
1546 1547 1548
        }
    }

E
Eric Blake 已提交
1549 1550 1551 1552
    ret = 0;
cleanup:
    VIR_FREE(cpumap);
    return ret;
1553
}
1554 1555 1556 1557

static int
qemuProcessInitPasswords(virConnectPtr conn,
                         struct qemud_driver *driver,
1558
                         virDomainObjPtr vm)
1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579
{
    int ret = 0;
    qemuDomainObjPrivatePtr priv = vm->privateData;

    if (vm->def->ngraphics == 1) {
        if (vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC) {
            ret = qemuDomainChangeGraphicsPasswords(driver, vm,
                                                    VIR_DOMAIN_GRAPHICS_TYPE_VNC,
                                                    &vm->def->graphics[0]->data.vnc.auth,
                                                    driver->vncPassword);
        } else if (vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_SPICE) {
            ret = qemuDomainChangeGraphicsPasswords(driver, vm,
                                                    VIR_DOMAIN_GRAPHICS_TYPE_SPICE,
                                                    &vm->def->graphics[0]->data.spice.auth,
                                                    driver->spicePassword);
        }
    }

    if (ret < 0)
        goto cleanup;

1580
    if (qemuCapsGet(priv->qemuCaps, QEMU_CAPS_DEVICE)) {
1581 1582 1583 1584 1585
        int i;

        for (i = 0 ; i < vm->def->ndisks ; i++) {
            char *secret;
            size_t secretLen;
E
Eric Blake 已提交
1586
            const char *alias;
1587 1588 1589 1590 1591 1592 1593 1594 1595 1596

            if (!vm->def->disks[i]->encryption ||
                !vm->def->disks[i]->src)
                continue;

            if (qemuProcessGetVolumeQcowPassphrase(conn,
                                                   vm->def->disks[i],
                                                   &secret, &secretLen) < 0)
                goto cleanup;

E
Eric Blake 已提交
1597
            alias = vm->def->disks[i]->info.alias;
1598
            qemuDomainObjEnterMonitorWithDriver(driver, vm);
E
Eric Blake 已提交
1599
            ret = qemuMonitorSetDrivePassphrase(priv->mon, alias, secret);
1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 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 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988
            VIR_FREE(secret);
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            if (ret < 0)
                goto cleanup;
        }
    }

cleanup:
    return ret;
}


#define QEMU_PCI_VENDOR_INTEL     0x8086
#define QEMU_PCI_VENDOR_LSI_LOGIC 0x1000
#define QEMU_PCI_VENDOR_REDHAT    0x1af4
#define QEMU_PCI_VENDOR_CIRRUS    0x1013
#define QEMU_PCI_VENDOR_REALTEK   0x10ec
#define QEMU_PCI_VENDOR_AMD       0x1022
#define QEMU_PCI_VENDOR_ENSONIQ   0x1274
#define QEMU_PCI_VENDOR_VMWARE    0x15ad
#define QEMU_PCI_VENDOR_QEMU      0x1234

#define QEMU_PCI_PRODUCT_DISK_VIRTIO 0x1001

#define QEMU_PCI_PRODUCT_BALLOON_VIRTIO 0x1002

#define QEMU_PCI_PRODUCT_NIC_NE2K     0x8029
#define QEMU_PCI_PRODUCT_NIC_PCNET    0x2000
#define QEMU_PCI_PRODUCT_NIC_RTL8139  0x8139
#define QEMU_PCI_PRODUCT_NIC_E1000    0x100E
#define QEMU_PCI_PRODUCT_NIC_VIRTIO   0x1000

#define QEMU_PCI_PRODUCT_VGA_CIRRUS 0x00b8
#define QEMU_PCI_PRODUCT_VGA_VMWARE 0x0405
#define QEMU_PCI_PRODUCT_VGA_STDVGA 0x1111

#define QEMU_PCI_PRODUCT_AUDIO_AC97    0x2415
#define QEMU_PCI_PRODUCT_AUDIO_ES1370  0x5000

#define QEMU_PCI_PRODUCT_CONTROLLER_PIIX 0x7010
#define QEMU_PCI_PRODUCT_CONTROLLER_LSI  0x0012

#define QEMU_PCI_PRODUCT_WATCHDOG_I63000ESB 0x25ab

static int
qemuProcessAssignNextPCIAddress(virDomainDeviceInfo *info,
                                int vendor,
                                int product,
                                qemuMonitorPCIAddress *addrs,
                                int naddrs)
{
    int found = 0;
    int i;

    VIR_DEBUG("Look for %x:%x out of %d", vendor, product, naddrs);

    for (i = 0 ; (i < naddrs) && !found; i++) {
        VIR_DEBUG("Maybe %x:%x", addrs[i].vendor, addrs[i].product);
        if (addrs[i].vendor == vendor &&
            addrs[i].product == product) {
            VIR_DEBUG("Match %d", i);
            found = 1;
            break;
        }
    }
    if (!found) {
        return -1;
    }

    /* Blank it out so this device isn't matched again */
    addrs[i].vendor = 0;
    addrs[i].product = 0;

    if (info->type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE)
        info->type = VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI;

    if (info->type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI) {
        info->addr.pci.domain = addrs[i].addr.domain;
        info->addr.pci.bus = addrs[i].addr.bus;
        info->addr.pci.slot = addrs[i].addr.slot;
        info->addr.pci.function = addrs[i].addr.function;
    }

    return 0;
}

static int
qemuProcessGetPCIDiskVendorProduct(virDomainDiskDefPtr def,
                                   unsigned *vendor,
                                   unsigned *product)
{
    switch (def->bus) {
    case VIR_DOMAIN_DISK_BUS_VIRTIO:
        *vendor = QEMU_PCI_VENDOR_REDHAT;
        *product = QEMU_PCI_PRODUCT_DISK_VIRTIO;
        break;

    default:
        return -1;
    }

    return 0;
}

static int
qemuProcessGetPCINetVendorProduct(virDomainNetDefPtr def,
                                  unsigned *vendor,
                                  unsigned *product)
{
    if (!def->model)
        return -1;

    if (STREQ(def->model, "ne2k_pci")) {
        *vendor = QEMU_PCI_VENDOR_REALTEK;
        *product = QEMU_PCI_PRODUCT_NIC_NE2K;
    } else if (STREQ(def->model, "pcnet")) {
        *vendor = QEMU_PCI_VENDOR_AMD;
        *product = QEMU_PCI_PRODUCT_NIC_PCNET;
    } else if (STREQ(def->model, "rtl8139")) {
        *vendor = QEMU_PCI_VENDOR_REALTEK;
        *product = QEMU_PCI_PRODUCT_NIC_RTL8139;
    } else if (STREQ(def->model, "e1000")) {
        *vendor = QEMU_PCI_VENDOR_INTEL;
        *product = QEMU_PCI_PRODUCT_NIC_E1000;
    } else if (STREQ(def->model, "virtio")) {
        *vendor = QEMU_PCI_VENDOR_REDHAT;
        *product = QEMU_PCI_PRODUCT_NIC_VIRTIO;
    } else {
        VIR_INFO("Unexpected NIC model %s, cannot get PCI address",
                 def->model);
        return -1;
    }
    return 0;
}

static int
qemuProcessGetPCIControllerVendorProduct(virDomainControllerDefPtr def,
                                         unsigned *vendor,
                                         unsigned *product)
{
    switch (def->type) {
    case VIR_DOMAIN_CONTROLLER_TYPE_SCSI:
        *vendor = QEMU_PCI_VENDOR_LSI_LOGIC;
        *product = QEMU_PCI_PRODUCT_CONTROLLER_LSI;
        break;

    case VIR_DOMAIN_CONTROLLER_TYPE_FDC:
        /* XXX we could put in the ISA bridge address, but
           that's not technically the FDC's address */
        return -1;

    case VIR_DOMAIN_CONTROLLER_TYPE_IDE:
        *vendor = QEMU_PCI_VENDOR_INTEL;
        *product = QEMU_PCI_PRODUCT_CONTROLLER_PIIX;
        break;

    default:
        VIR_INFO("Unexpected controller type %s, cannot get PCI address",
                 virDomainControllerTypeToString(def->type));
        return -1;
    }

    return 0;
}

static int
qemuProcessGetPCIVideoVendorProduct(virDomainVideoDefPtr def,
                                    unsigned *vendor,
                                    unsigned *product)
{
    switch (def->type) {
    case VIR_DOMAIN_VIDEO_TYPE_CIRRUS:
        *vendor = QEMU_PCI_VENDOR_CIRRUS;
        *product = QEMU_PCI_PRODUCT_VGA_CIRRUS;
        break;

    case VIR_DOMAIN_VIDEO_TYPE_VGA:
        *vendor = QEMU_PCI_VENDOR_QEMU;
        *product = QEMU_PCI_PRODUCT_VGA_STDVGA;
        break;

    case VIR_DOMAIN_VIDEO_TYPE_VMVGA:
        *vendor = QEMU_PCI_VENDOR_VMWARE;
        *product = QEMU_PCI_PRODUCT_VGA_VMWARE;
        break;

    default:
        return -1;
    }
    return 0;
}

static int
qemuProcessGetPCISoundVendorProduct(virDomainSoundDefPtr def,
                                    unsigned *vendor,
                                    unsigned *product)
{
    switch (def->model) {
    case VIR_DOMAIN_SOUND_MODEL_ES1370:
        *vendor = QEMU_PCI_VENDOR_ENSONIQ;
        *product = QEMU_PCI_PRODUCT_AUDIO_ES1370;
        break;

    case VIR_DOMAIN_SOUND_MODEL_AC97:
        *vendor = QEMU_PCI_VENDOR_INTEL;
        *product = QEMU_PCI_PRODUCT_AUDIO_AC97;
        break;

    default:
        return -1;
    }

    return 0;
}

static int
qemuProcessGetPCIWatchdogVendorProduct(virDomainWatchdogDefPtr def,
                                       unsigned *vendor,
                                       unsigned *product)
{
    switch (def->model) {
    case VIR_DOMAIN_WATCHDOG_MODEL_I6300ESB:
        *vendor = QEMU_PCI_VENDOR_INTEL;
        *product = QEMU_PCI_PRODUCT_WATCHDOG_I63000ESB;
        break;

    default:
        return -1;
    }

    return 0;
}


static int
qemuProcessGetPCIMemballoonVendorProduct(virDomainMemballoonDefPtr def,
                                         unsigned *vendor,
                                         unsigned *product)
{
    switch (def->model) {
    case VIR_DOMAIN_MEMBALLOON_MODEL_VIRTIO:
        *vendor = QEMU_PCI_VENDOR_REDHAT;
        *product = QEMU_PCI_PRODUCT_BALLOON_VIRTIO;
        break;

    default:
        return -1;
    }

    return 0;
}


/*
 * This entire method assumes that PCI devices in 'info pci'
 * match ordering of devices specified on the command line
 * wrt to devices of matching vendor+product
 *
 * XXXX this might not be a valid assumption if we assign
 * some static addrs on CLI. Have to check that...
 */
static int
qemuProcessDetectPCIAddresses(virDomainObjPtr vm,
                              qemuMonitorPCIAddress *addrs,
                              int naddrs)
{
    unsigned int vendor = 0, product = 0;
    int i;

    /* XXX should all these vendor/product IDs be kept in the
     * actual device data structure instead ?
     */

    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (qemuProcessGetPCIDiskVendorProduct(vm->def->disks[i], &vendor, &product) < 0)
            continue;

        if (qemuProcessAssignNextPCIAddress(&(vm->def->disks[i]->info),
                                            vendor, product,
                                            addrs, naddrs) < 0) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for VirtIO disk %s"),
                            vm->def->disks[i]->dst);
            return -1;
        }
    }

    for (i = 0 ; i < vm->def->nnets ; i++) {
        if (qemuProcessGetPCINetVendorProduct(vm->def->nets[i], &vendor, &product) < 0)
            continue;

        if (qemuProcessAssignNextPCIAddress(&(vm->def->nets[i]->info),
                                            vendor, product,
                                            addrs,  naddrs) < 0) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for %s NIC"),
                            vm->def->nets[i]->model);
            return -1;
        }
    }

    for (i = 0 ; i < vm->def->ncontrollers ; i++) {
        if (qemuProcessGetPCIControllerVendorProduct(vm->def->controllers[i], &vendor, &product) < 0)
            continue;

        if (qemuProcessAssignNextPCIAddress(&(vm->def->controllers[i]->info),
                                            vendor, product,
                                            addrs,  naddrs) < 0) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for controller %s"),
                            virDomainControllerTypeToString(vm->def->controllers[i]->type));
            return -1;
        }
    }

    for (i = 0 ; i < vm->def->nvideos ; i++) {
        if (qemuProcessGetPCIVideoVendorProduct(vm->def->videos[i], &vendor, &product) < 0)
            continue;

        if (qemuProcessAssignNextPCIAddress(&(vm->def->videos[i]->info),
                                            vendor, product,
                                            addrs,  naddrs) < 0) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for video adapter %s"),
                            virDomainVideoTypeToString(vm->def->videos[i]->type));
            return -1;
        }
    }

    for (i = 0 ; i < vm->def->nsounds ; i++) {
        if (qemuProcessGetPCISoundVendorProduct(vm->def->sounds[i], &vendor, &product) < 0)
            continue;

        if (qemuProcessAssignNextPCIAddress(&(vm->def->sounds[i]->info),
                                    vendor, product,
                                     addrs,  naddrs) < 0) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for sound adapter %s"),
                            virDomainSoundModelTypeToString(vm->def->sounds[i]->model));
            return -1;
        }
    }


    if (vm->def->watchdog &&
        qemuProcessGetPCIWatchdogVendorProduct(vm->def->watchdog, &vendor, &product) == 0) {
        if (qemuProcessAssignNextPCIAddress(&(vm->def->watchdog->info),
                                            vendor, product,
                                            addrs,  naddrs) < 0) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for watchdog %s"),
                            virDomainWatchdogModelTypeToString(vm->def->watchdog->model));
            return -1;
        }
    }

    if (vm->def->memballoon &&
        qemuProcessGetPCIMemballoonVendorProduct(vm->def->memballoon, &vendor, &product) == 0) {
        if (qemuProcessAssignNextPCIAddress(&(vm->def->memballoon->info),
                                            vendor, product,
                                            addrs, naddrs) < 0) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for balloon %s"),
                            virDomainMemballoonModelTypeToString(vm->def->memballoon->model));
            return -1;
        }
    }

    /* XXX console (virtio) */


    /* ... and now things we don't have in our xml */

    /* XXX USB controller ? */

    /* XXX what about other PCI devices (ie bridges) */

    return 0;
}

static int
qemuProcessInitPCIAddresses(struct qemud_driver *driver,
                            virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int naddrs;
    int ret;
    qemuMonitorPCIAddress *addrs = NULL;

1989
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091
    naddrs = qemuMonitorGetAllPCIAddresses(priv->mon,
                                           &addrs);
    qemuDomainObjExitMonitorWithDriver(driver, vm);

    ret = qemuProcessDetectPCIAddresses(vm, addrs, naddrs);

    VIR_FREE(addrs);

    return ret;
}


static int qemuProcessNextFreePort(struct qemud_driver *driver,
                                   int startPort)
{
    int i;

    for (i = startPort ; i < QEMU_VNC_PORT_MAX; i++) {
        int fd;
        int reuse = 1;
        struct sockaddr_in addr;
        bool used = false;

        if (virBitmapGetBit(driver->reservedVNCPorts,
                            i - QEMU_VNC_PORT_MIN, &used) < 0)
            VIR_DEBUG("virBitmapGetBit failed on bit %d", i - QEMU_VNC_PORT_MIN);

        if (used)
            continue;

        addr.sin_family = AF_INET;
        addr.sin_port = htons(i);
        addr.sin_addr.s_addr = htonl(INADDR_ANY);
        fd = socket(PF_INET, SOCK_STREAM, 0);
        if (fd < 0)
            return -1;

        if (setsockopt(fd, SOL_SOCKET, SO_REUSEADDR, (void*)&reuse, sizeof(reuse)) < 0) {
            VIR_FORCE_CLOSE(fd);
            break;
        }

        if (bind(fd, (struct sockaddr*)&addr, sizeof(addr)) == 0) {
            /* Not in use, lets grab it */
            VIR_FORCE_CLOSE(fd);
            /* Add port to bitmap of reserved ports */
            if (virBitmapSetBit(driver->reservedVNCPorts,
                                i - QEMU_VNC_PORT_MIN) < 0) {
                VIR_DEBUG("virBitmapSetBit failed on bit %d",
                          i - QEMU_VNC_PORT_MIN);
            }
            return i;
        }
        VIR_FORCE_CLOSE(fd);

        if (errno == EADDRINUSE) {
            /* In use, try next */
            continue;
        }
        /* Some other bad failure, get out.. */
        break;
    }
    return -1;
}


static void
qemuProcessReturnPort(struct qemud_driver *driver,
                      int port)
{
    if (port < QEMU_VNC_PORT_MIN)
        return;

    if (virBitmapClearBit(driver->reservedVNCPorts,
                          port - QEMU_VNC_PORT_MIN) < 0)
        VIR_DEBUG("Could not mark port %d as unused", port);
}


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

    if ((fd = open(dev->source.data.file.path,
                   O_CREAT | O_APPEND, S_IRUSR|S_IWUSR)) < 0) {
        virReportSystemError(errno,
                             _("Unable to pre-create chardev file '%s'"),
                             dev->source.data.file.path);
        return -1;
    }

    VIR_FORCE_CLOSE(fd);

    return 0;
}


2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110
static int
qemuProcessLimits(struct qemud_driver *driver)
{
    if (driver->maxProcesses > 0) {
        struct rlimit rlim;

        rlim.rlim_cur = rlim.rlim_max = driver->maxProcesses;
        if (setrlimit(RLIMIT_NPROC, &rlim) < 0) {
            virReportSystemError(errno,
                                 _("cannot limit number of processes to %d"),
                                 driver->maxProcesses);
            return -1;
        }
    }

    return 0;
}


2111 2112 2113 2114 2115 2116 2117 2118 2119
struct qemuProcessHookData {
    virConnectPtr conn;
    virDomainObjPtr vm;
    struct qemud_driver *driver;
};

static int qemuProcessHook(void *data)
{
    struct qemuProcessHookData *h = data;
2120
    int ret = -1;
2121
    int fd;
2122 2123 2124 2125 2126

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

    VIR_DEBUG("Obtaining domain lock");
2127 2128 2129 2130 2131 2132 2133 2134 2135
    /*
     * 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.
     */
    if (virSecurityManagerSetSocketLabel(h->driver->securityManager, h->vm) < 0)
        goto cleanup;
2136 2137 2138
    if (virDomainLockProcessStart(h->driver->lockManager,
                                  h->vm,
                                  /* QEMU is always pased initially */
2139 2140
                                  true,
                                  &fd) < 0)
2141
        goto cleanup;
2142 2143
    if (virSecurityManagerClearSocketLabel(h->driver->securityManager, h->vm) < 0)
        goto cleanup;
2144

2145
    if (qemuProcessLimits(h->driver) < 0)
2146
        goto cleanup;
2147

2148 2149 2150
    /* This must take place before exec(), so that all QEMU
     * memory allocation is on the correct NUMA node
     */
2151
    VIR_DEBUG("Moving procss to cgroup");
2152
    if (qemuAddToCgroup(h->driver, h->vm->def) < 0)
2153
        goto cleanup;
2154 2155 2156

    /* This must be done after cgroup placement to avoid resetting CPU
     * affinity */
2157
    VIR_DEBUG("Setup CPU affinity");
2158
    if (qemuProcessInitCpuAffinity(h->vm) < 0)
2159
        goto cleanup;
2160

2161 2162 2163
    if (qemuProcessInitNumaMemoryPolicy(h->vm) < 0)
        return -1;

2164
    VIR_DEBUG("Setting up security labelling");
2165
    if (virSecurityManagerSetProcessLabel(h->driver->securityManager, h->vm) < 0)
2166
        goto cleanup;
2167

2168 2169 2170 2171 2172
    ret = 0;

cleanup:
    VIR_DEBUG("Hook complete ret=%d", ret);
    return ret;
2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192
}

int
qemuProcessPrepareMonitorChr(struct qemud_driver *driver,
                             virDomainChrSourceDefPtr monConfig,
                             const char *vm)
{
    monConfig->type = VIR_DOMAIN_CHR_TYPE_UNIX;
    monConfig->data.nix.listen = true;

    if (virAsprintf(&monConfig->data.nix.path, "%s/%s.monitor",
                    driver->libDir, vm) < 0) {
        virReportOOMError();
        return -1;
    }

    return 0;
}


2193 2194 2195 2196 2197
/*
 * Precondition: Both driver and vm must be locked,
 * and a job must be active. This method will call
 * {Enter,Exit}MonitorWithDriver
 */
E
Eric Blake 已提交
2198 2199
int
qemuProcessStartCPUs(struct qemud_driver *driver, virDomainObjPtr vm,
2200 2201
                     virConnectPtr conn, virDomainRunningReason reason,
                     enum qemuDomainAsyncJob asyncJob)
2202 2203 2204 2205
{
    int ret;
    qemuDomainObjPrivatePtr priv = vm->privateData;

2206 2207 2208 2209 2210 2211 2212 2213 2214 2215
    VIR_DEBUG("Using lock state '%s'", NULLSTR(priv->lockState));
    if (virDomainLockProcessResume(driver->lockManager, vm, priv->lockState) < 0) {
        /* Don't free priv->lockState on error, because we need
         * to make sure we have state still present if the user
         * tries to resume again
         */
        return -1;
    }
    VIR_FREE(priv->lockState);

2216 2217 2218 2219 2220
    ret = qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob);
    if (ret == 0) {
        ret = qemuMonitorStartCPUs(priv->mon, conn);
        qemuDomainObjExitMonitorWithDriver(driver, vm);
    }
J
Jiri Denemark 已提交
2221

2222
    if (ret == 0) {
J
Jiri Denemark 已提交
2223
        virDomainObjSetState(vm, VIR_DOMAIN_RUNNING, reason);
2224 2225 2226 2227 2228
    } else {
        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));
    }
2229 2230 2231 2232 2233

    return ret;
}


J
Jiri Denemark 已提交
2234
int qemuProcessStopCPUs(struct qemud_driver *driver, virDomainObjPtr vm,
2235 2236
                        virDomainPausedReason reason,
                        enum qemuDomainAsyncJob asyncJob)
2237 2238
{
    int ret;
J
Jiri Denemark 已提交
2239 2240
    int oldState;
    int oldReason;
2241 2242
    qemuDomainObjPrivatePtr priv = vm->privateData;

2243
    VIR_FREE(priv->lockState);
J
Jiri Denemark 已提交
2244 2245 2246
    oldState = virDomainObjGetState(vm, &oldReason);
    virDomainObjSetState(vm, VIR_DOMAIN_PAUSED, reason);

2247
    ret = qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob);
2248 2249 2250 2251
    if (ret == 0) {
        ret = qemuMonitorStopCPUs(priv->mon);
        qemuDomainObjExitMonitorWithDriver(driver, vm);
    }
J
Jiri Denemark 已提交
2252

2253 2254 2255 2256 2257
    if (ret == 0) {
        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));
    } else {
J
Jiri Denemark 已提交
2258
        virDomainObjSetState(vm, oldState, oldReason);
2259
    }
J
Jiri Denemark 已提交
2260

2261 2262 2263 2264 2265
    return ret;
}



2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278
static int
qemuProcessNotifyNets(virDomainDefPtr def)
{
    int ii;

    for (ii = 0 ; ii < def->nnets ; ii++) {
        virDomainNetDefPtr net = def->nets[ii];
        if (networkNotifyActualDevice(net) < 0)
            return -1;
    }
    return 0;
}

2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301
static int
qemuProcessFiltersInstantiate(virConnectPtr conn,
                              virDomainDefPtr def)
{
    int err = 0;
    int i;

    if (!conn)
        return 1;

    for (i = 0 ; i < def->nnets ; i++) {
        virDomainNetDefPtr net = def->nets[i];
        if ((net->filter) && (net->ifname)) {
           if (virDomainConfNWFilterInstantiate(conn, net)) {
                err = 1;
                break;
            }
        }
    }

    return err;
}

2302 2303 2304 2305 2306 2307 2308 2309
static int
qemuProcessUpdateState(struct qemud_driver *driver, virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virDomainState state;
    bool running;
    int ret;

2310
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327
    ret = qemuMonitorGetStatus(priv->mon, &running);
    qemuDomainObjExitMonitorWithDriver(driver, vm);

    if (ret < 0 || !virDomainObjIsActive(vm))
        return -1;

    state = virDomainObjGetState(vm, NULL);

    if (state == VIR_DOMAIN_PAUSED && running) {
        VIR_DEBUG("Domain %s was unpaused while its monitor was disconnected;"
                  " changing state to running", vm->def->name);
        virDomainObjSetState(vm, VIR_DOMAIN_RUNNING,
                             VIR_DOMAIN_RUNNING_UNPAUSED);
    } else if (state == VIR_DOMAIN_RUNNING && !running) {
        VIR_DEBUG("Domain %s was paused while its monitor was disconnected;"
                  " changing state to paused", vm->def->name);
        virDomainObjSetState(vm, VIR_DOMAIN_PAUSED, VIR_DOMAIN_PAUSED_UNKNOWN);
2328 2329 2330 2331
    } else if (state == VIR_DOMAIN_SHUTOFF && running) {
        VIR_DEBUG("Domain %s finished booting; changing state to running",
                  vm->def->name);
        virDomainObjSetState(vm, VIR_DOMAIN_RUNNING, VIR_DOMAIN_RUNNING_BOOTED);
2332 2333 2334 2335 2336
    }

    return 0;
}

2337 2338 2339 2340 2341 2342 2343 2344 2345
static int
qemuProcessRecoverMigration(struct qemud_driver *driver,
                            virDomainObjPtr vm,
                            virConnectPtr conn,
                            enum qemuDomainAsyncJob job,
                            enum qemuMigrationJobPhase phase,
                            virDomainState state,
                            int reason)
{
2346 2347
    qemuDomainObjPrivatePtr priv = vm->privateData;

2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370
    if (job == QEMU_ASYNC_JOB_MIGRATION_IN) {
        switch (phase) {
        case QEMU_MIGRATION_PHASE_NONE:
        case QEMU_MIGRATION_PHASE_PERFORM2:
        case QEMU_MIGRATION_PHASE_BEGIN3:
        case QEMU_MIGRATION_PHASE_PERFORM3:
        case QEMU_MIGRATION_PHASE_PERFORM3_DONE:
        case QEMU_MIGRATION_PHASE_CONFIRM3_CANCELLED:
        case QEMU_MIGRATION_PHASE_CONFIRM3:
        case QEMU_MIGRATION_PHASE_LAST:
            break;

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

        case QEMU_MIGRATION_PHASE_FINISH2:
            /* source domain is already killed so let's just resume the domain
             * and hope we are all set */
            VIR_DEBUG("Incoming migration finished, resuming domain %s",
                      vm->def->name);
            if (qemuProcessStartCPUs(driver, vm, conn,
2371 2372
                                     VIR_DOMAIN_RUNNING_UNPAUSED,
                                     QEMU_ASYNC_JOB_NONE) < 0) {
2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402
                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 */
            VIR_DEBUG("Killing migrated domain %s", vm->def->name);
            return -1;
        }
    } else if (job == QEMU_ASYNC_JOB_MIGRATION_OUT) {
        switch (phase) {
        case QEMU_MIGRATION_PHASE_NONE:
        case QEMU_MIGRATION_PHASE_PREPARE:
        case QEMU_MIGRATION_PHASE_FINISH2:
        case QEMU_MIGRATION_PHASE_FINISH3:
        case QEMU_MIGRATION_PHASE_LAST:
            break;

        case QEMU_MIGRATION_PHASE_BEGIN3:
            /* nothing happen 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
             * domain */
            VIR_DEBUG("Canceling unfinished outgoing migration of domain %s",
                      vm->def->name);
2403
            qemuDomainObjEnterMonitor(driver, vm);
2404 2405
            ignore_value(qemuMonitorMigrateCancel(priv->mon));
            qemuDomainObjExitMonitor(driver, vm);
2406 2407 2408 2409 2410 2411
            /* resume the domain but only if it was paused as a result of
             * migration */
            if (state == VIR_DOMAIN_PAUSED &&
                (reason == VIR_DOMAIN_PAUSED_MIGRATION ||
                 reason == VIR_DOMAIN_PAUSED_UNKNOWN)) {
                if (qemuProcessStartCPUs(driver, vm, conn,
2412 2413
                                         VIR_DOMAIN_RUNNING_UNPAUSED,
                                         QEMU_ASYNC_JOB_NONE) < 0) {
2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432
                    VIR_WARN("Could not resume domain %s", vm->def->name);
                }
            }
            break;

        case QEMU_MIGRATION_PHASE_PERFORM3_DONE:
            /* migration finished but we didn't have a chance to get the result
             * of Finish3 step; third party needs to check what to do next
             */
            break;

        case QEMU_MIGRATION_PHASE_CONFIRM3_CANCELLED:
            /* Finish3 failed, we need to resume the domain */
            VIR_DEBUG("Resuming domain %s after failed migration",
                      vm->def->name);
            if (state == VIR_DOMAIN_PAUSED &&
                (reason == VIR_DOMAIN_PAUSED_MIGRATION ||
                 reason == VIR_DOMAIN_PAUSED_UNKNOWN)) {
                if (qemuProcessStartCPUs(driver, vm, conn,
2433 2434
                                         VIR_DOMAIN_RUNNING_UNPAUSED,
                                         QEMU_ASYNC_JOB_NONE) < 0) {
2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448
                    VIR_WARN("Could not resume domain %s", vm->def->name);
                }
            }
            break;

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

    return 0;
}

2449 2450 2451 2452 2453 2454
static int
qemuProcessRecoverJob(struct qemud_driver *driver,
                      virDomainObjPtr vm,
                      virConnectPtr conn,
                      const struct qemuDomainJobObj *job)
{
2455
    qemuDomainObjPrivatePtr priv = vm->privateData;
2456 2457 2458 2459 2460 2461 2462 2463
    virDomainState state;
    int reason;

    state = virDomainObjGetState(vm, &reason);

    switch (job->asyncJob) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
    case QEMU_ASYNC_JOB_MIGRATION_IN:
2464 2465 2466
        if (qemuProcessRecoverMigration(driver, vm, conn, job->asyncJob,
                                        job->phase, state, reason) < 0)
            return -1;
2467 2468 2469 2470
        break;

    case QEMU_ASYNC_JOB_SAVE:
    case QEMU_ASYNC_JOB_DUMP:
2471
        qemuDomainObjEnterMonitor(driver, vm);
2472 2473
        ignore_value(qemuMonitorMigrateCancel(priv->mon));
        qemuDomainObjExitMonitor(driver, vm);
2474
        /* resume the domain but only if it was paused as a result of
2475 2476 2477
         * running save/dump operation.  Although we are recovering an
         * async job, this function is run at startup and must resume
         * things using sync monitor connections.  */
2478 2479 2480 2481 2482 2483 2484
        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) ||
             reason == VIR_DOMAIN_PAUSED_UNKNOWN)) {
            if (qemuProcessStartCPUs(driver, vm, conn,
2485 2486
                                     VIR_DOMAIN_RUNNING_UNPAUSED,
                                     QEMU_ASYNC_JOB_NONE) < 0) {
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
                VIR_WARN("Could not resume domain %s after", vm->def->name);
            }
        }
        break;

    case QEMU_ASYNC_JOB_NONE:
    case QEMU_ASYNC_JOB_LAST:
        break;
    }

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

    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;

2520
    case QEMU_JOB_MIGRATION_OP:
2521
    case QEMU_JOB_ABORT:
2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532
    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;
}

2533 2534 2535
struct qemuProcessReconnectData {
    virConnectPtr conn;
    struct qemud_driver *driver;
2536 2537
    void *payload;
    struct qemuDomainJobObj oldjob;
2538 2539 2540 2541
};
/*
 * Open an existing VM's monitor, re-detect VCPU threads
 * and re-reserve the security labels in use
S
Stefan Berger 已提交
2542 2543 2544 2545
 *
 * We own the virConnectPtr we are passed here - whoever started
 * this thread function has increased the reference counter to it
 * so that we now have to close it.
2546 2547
 */
static void
2548
qemuProcessReconnect(void *opaque)
2549 2550 2551
{
    struct qemuProcessReconnectData *data = opaque;
    struct qemud_driver *driver = data->driver;
2552
    virDomainObjPtr obj = data->payload;
2553 2554
    qemuDomainObjPrivatePtr priv;
    virConnectPtr conn = data->conn;
2555
    struct qemuDomainJobObj oldjob;
2556

2557 2558 2559 2560 2561
    memcpy(&oldjob, &data->oldjob, sizeof(oldjob));

    VIR_FREE(data);

    qemuDriverLock(driver);
2562 2563
    virDomainObjLock(obj);

2564

2565 2566 2567 2568
    VIR_DEBUG("Reconnect monitor to %p '%s'", obj, obj->def->name);

    priv = obj->privateData;

2569 2570 2571
    /* Set fake job so that EnterMonitor* doesn't want to start a new one */
    priv->job.active = QEMU_JOB_MODIFY;

2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583
    /* Hold an extra reference because we can't allow 'vm' to be
     * deleted if qemuConnectMonitor() failed */
    virDomainObjRef(obj);

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

    if (qemuUpdateActivePciHostdevs(driver, obj->def) < 0) {
        goto error;
    }

2584 2585 2586
    if (qemuProcessUpdateState(driver, obj) < 0)
        goto error;

2587 2588 2589 2590 2591 2592
    if (virDomainObjGetState(obj, NULL) == VIR_DOMAIN_SHUTOFF) {
        VIR_DEBUG("Domain '%s' wasn't fully started yet, killing it",
                  obj->def->name);
        goto error;
    }

2593 2594 2595 2596 2597
    /* If upgrading from old libvirtd we won't have found any
     * caps in the domain status, so re-query them
     */
    if (!priv->qemuCaps &&
        qemuCapsExtractVersionInfo(obj->def->emulator, obj->def->os.arch,
2598
                                   NULL,
2599 2600 2601 2602
                                   &priv->qemuCaps) < 0)
        goto error;

    if (qemuCapsGet(priv->qemuCaps, QEMU_CAPS_DEVICE)) {
2603 2604 2605 2606 2607 2608 2609 2610 2611 2612
        priv->persistentAddrs = 1;

        if (!(priv->pciaddrs = qemuDomainPCIAddressSetCreate(obj->def)) ||
            qemuAssignDevicePCISlots(obj->def, priv->pciaddrs) < 0)
            goto error;
    }

    if (virSecurityManagerReserveLabel(driver->securityManager, obj) < 0)
        goto error;

2613 2614 2615
    if (qemuProcessNotifyNets(obj->def) < 0)
        goto error;

2616 2617 2618
    if (qemuProcessFiltersInstantiate(conn, obj->def))
        goto error;

2619 2620 2621
    if (qemuProcessRecoverJob(driver, obj, conn, &oldjob) < 0)
        goto error;

2622 2623
    priv->job.active = QEMU_JOB_NONE;

2624 2625 2626 2627
    /* update domain state XML with possibly updated state in virDomainObj */
    if (virDomainSaveStatus(driver->caps, driver->stateDir, obj) < 0)
        goto error;

2628 2629 2630 2631 2632
    if (obj->def->id >= driver->nextvmid)
        driver->nextvmid = obj->def->id + 1;

    if (virDomainObjUnref(obj) > 0)
        virDomainObjUnlock(obj);
2633

2634 2635 2636 2637 2638
    if (qemuDomainObjEndJob(driver, obj) == 0)
        obj = NULL;

    qemuDriverUnlock(driver);

S
Stefan Berger 已提交
2639 2640
    virConnectClose(conn);

2641 2642 2643
    return;

error:
2644 2645 2646
    if (qemuDomainObjEndJob(driver, obj) == 0)
        obj = NULL;

2647 2648 2649
    if (!virDomainObjIsActive(obj)) {
        if (virDomainObjUnref(obj) > 0)
            virDomainObjUnlock(obj);
2650
        qemuDriverUnlock(driver);
2651 2652 2653 2654 2655 2656 2657
        return;
    }

    if (virDomainObjUnref(obj) > 0) {
        /* 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 */
J
Jiri Denemark 已提交
2658
        qemuProcessStop(driver, obj, 0, VIR_DOMAIN_SHUTOFF_FAILED);
2659 2660 2661 2662 2663
        if (!obj->persistent)
            virDomainRemoveInactive(&driver->domains, obj);
        else
            virDomainObjUnlock(obj);
    }
2664
    qemuDriverUnlock(driver);
S
Stefan Berger 已提交
2665 2666

    virConnectClose(conn);
2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 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
}

static void
qemuProcessReconnectHelper(void *payload,
                           const void *name ATTRIBUTE_UNUSED,
                           void *opaque)
{
    virThread thread;
    struct qemuProcessReconnectData *src = opaque;
    struct qemuProcessReconnectData *data;
    virDomainObjPtr obj = payload;

    if (VIR_ALLOC(data) < 0) {
        virReportOOMError();
        return;
    }

    memcpy(data, src, sizeof(*data));
    data->payload = payload;

    /* This iterator is called with driver being locked.
     * We create a separate thread to run qemuProcessReconnect in it.
     * However, qemuProcessReconnect needs to:
     * 1. lock driver
     * 2. just before monitor reconnect do lightweight MonitorEnter
     *    (increase VM refcount, unlock VM & driver)
     * 3. reconnect to monitor
     * 4. do lightweight MonitorExit (lock driver & VM)
     * 5. continue reconnect process
     * 6. EndJob
     * 7. unlock driver
     *
     * It is necessary to NOT hold driver lock for the entire run
     * of reconnect, otherwise we will get blocked if there is
     * unresponsive qemu.
     * However, iterating over hash table MUST be done on locked
     * driver.
     *
     * NB, we can't do normal MonitorEnter & MonitorExit because
     * these two lock the monitor lock, which does not exists in
     * this early phase.
     */

    virDomainObjLock(obj);

    qemuDomainObjRestoreJob(obj, &data->oldjob);

    if (qemuDomainObjBeginJobWithDriver(src->driver, obj, QEMU_JOB_MODIFY) < 0)
        goto error;

S
Stefan Berger 已提交
2717 2718 2719 2720 2721 2722
    /* Since we close the connection later on, we have to make sure
     * that the threads we start see a valid connection throughout their
     * lifetime. We simply increase the reference counter here.
     */
    virConnectRef(data->conn);

2723
    if (virThreadCreate(&thread, true, qemuProcessReconnect, data) < 0) {
S
Stefan Berger 已提交
2724 2725 2726

        virConnectClose(data->conn);

2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("Could not create thread. QEMU initialization "
                          "might be incomplete"));
        if (qemuDomainObjEndJob(src->driver, obj) == 0) {
            obj = NULL;
        } else if (virDomainObjUnref(obj) > 0) {
           /* We can't spawn a thread and thus connect to monitor.
            * Kill qemu */
            qemuProcessStop(src->driver, obj, 0, VIR_DOMAIN_SHUTOFF_FAILED);
            if (!obj->persistent)
                virDomainRemoveInactive(&src->driver->domains, obj);
            else
                virDomainObjUnlock(obj);
        }
        goto error;
    }

    virDomainObjUnlock(obj);

    return;

error:
    VIR_FREE(data);
2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761
}

/**
 * qemuProcessReconnectAll
 *
 * Try to re-open the resources for live VMs that we care
 * about.
 */
void
qemuProcessReconnectAll(virConnectPtr conn, struct qemud_driver *driver)
{
    struct qemuProcessReconnectData data = {conn, driver};
2762
    virHashForEach(driver->domains.objs, qemuProcessReconnectHelper, &data);
2763 2764 2765 2766 2767 2768 2769
}

int qemuProcessStart(virConnectPtr conn,
                     struct qemud_driver *driver,
                     virDomainObjPtr vm,
                     const char *migrateFrom,
                     bool start_paused,
2770
                     bool autodestroy,
2771 2772
                     int stdin_fd,
                     const char *stdin_path,
2773
                     virDomainSnapshotObjPtr snapshot,
2774 2775 2776 2777 2778 2779 2780 2781 2782 2783
                     enum virVMOperationType vmop)
{
    int ret;
    off_t pos = -1;
    char ebuf[1024];
    int logfile = -1;
    char *timestamp;
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virCommandPtr cmd = NULL;
    struct qemuProcessHookData hookData;
E
Eric Blake 已提交
2784
    unsigned long cur_balloon;
2785 2786 2787 2788 2789

    hookData.conn = conn;
    hookData.vm = vm;
    hookData.driver = driver;

2790
    VIR_DEBUG("Beginning VM startup process");
2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801

    if (virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("VM is already active"));
        return -1;
    }

    /* 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
     */
2802
    VIR_DEBUG("Setting current domain def as transient");
2803 2804 2805 2806
    if (virDomainObjSetDefTransient(driver->caps, vm, true) < 0)
        goto cleanup;

    vm->def->id = driver->nextvmid++;
2807
    priv->fakeReboot = false;
2808
    virDomainObjSetState(vm, VIR_DOMAIN_SHUTOFF, VIR_DOMAIN_SHUTOFF_UNKNOWN);
2809

2810
    /* Run an early hook to set-up missing devices */
2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825
    if (virHookPresent(VIR_HOOK_DRIVER_QEMU)) {
        char *xml = virDomainDefFormat(vm->def, 0);
        int hookret;

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

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

2826
    /* Must be run before security labelling */
2827
    VIR_DEBUG("Preparing host devices");
2828 2829 2830
    if (qemuPrepareHostDevices(driver, vm->def) < 0)
        goto cleanup;

2831
    VIR_DEBUG("Preparing chr devices");
2832 2833 2834 2835 2836 2837 2838 2839
    if (virDomainChrDefForeach(vm->def,
                               true,
                               qemuProcessPrepareChardevDevice,
                               NULL) < 0)
        goto cleanup;

    /* If you are using a SecurityDriver with dynamic labelling,
       then generate a security label for isolation */
2840
    VIR_DEBUG("Generating domain security label (if required)");
2841
    if (virSecurityManagerGenLabel(driver->securityManager, vm) < 0) {
2842
        virDomainAuditSecurityLabel(vm, false);
2843 2844
        goto cleanup;
    }
2845
    virDomainAuditSecurityLabel(vm, true);
2846 2847 2848

    /* Ensure no historical cgroup for this VM is lying around bogus
     * settings */
2849
    VIR_DEBUG("Ensuring no historical cgroup is lying around");
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 2886 2887
    qemuRemoveCgroup(driver, vm, 1);

    if (vm->def->ngraphics == 1) {
        if (vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
            !vm->def->graphics[0]->data.vnc.socket &&
            vm->def->graphics[0]->data.vnc.autoport) {
            int port = qemuProcessNextFreePort(driver, QEMU_VNC_PORT_MIN);
            if (port < 0) {
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                "%s", _("Unable to find an unused VNC port"));
                goto cleanup;
            }
            vm->def->graphics[0]->data.vnc.port = port;
        } else if (vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_SPICE &&
                   vm->def->graphics[0]->data.spice.autoport) {
            int port = qemuProcessNextFreePort(driver, QEMU_VNC_PORT_MIN);
            int tlsPort = -1;
            if (port < 0) {
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                "%s", _("Unable to find an unused SPICE port"));
                goto cleanup;
            }

            if (driver->spiceTLS) {
                tlsPort = qemuProcessNextFreePort(driver, port + 1);
                if (tlsPort < 0) {
                    qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                    "%s", _("Unable to find an unused SPICE TLS port"));
                    qemuProcessReturnPort(driver, port);
                    goto cleanup;
                }
            }

            vm->def->graphics[0]->data.spice.port = port;
            vm->def->graphics[0]->data.spice.tlsPort = tlsPort;
        }
    }

2888
    if (virFileMakePath(driver->logDir) < 0) {
2889 2890 2891 2892 2893 2894
        virReportSystemError(errno,
                             _("cannot create log directory %s"),
                             driver->logDir);
        goto cleanup;
    }

2895
    VIR_DEBUG("Creating domain log file");
2896
    if ((logfile = qemuDomainCreateLog(driver, vm, false)) < 0)
2897 2898
        goto cleanup;

2899
    VIR_DEBUG("Determining emulator version");
2900 2901
    qemuCapsFree(priv->qemuCaps);
    priv->qemuCaps = NULL;
2902 2903
    if (qemuCapsExtractVersionInfo(vm->def->emulator, vm->def->os.arch,
                                   NULL,
2904
                                   &priv->qemuCaps) < 0)
2905 2906
        goto cleanup;

2907
    VIR_DEBUG("Setting up domain cgroup (if required)");
2908 2909 2910 2911 2912 2913 2914 2915
    if (qemuSetupCgroup(driver, vm) < 0)
        goto cleanup;

    if (VIR_ALLOC(priv->monConfig) < 0) {
        virReportOOMError();
        goto cleanup;
    }

2916
    VIR_DEBUG("Preparing monitor state");
2917 2918 2919 2920
    if (qemuProcessPrepareMonitorChr(driver, priv->monConfig, vm->def->name) < 0)
        goto cleanup;

#if HAVE_YAJL
2921
    if (qemuCapsGet(priv->qemuCaps, QEMU_CAPS_MONITOR_JSON))
2922 2923 2924 2925 2926
        priv->monJSON = 1;
    else
#endif
        priv->monJSON = 0;

2927 2928
    priv->monError = false;
    priv->monStart = 0;
2929 2930
    priv->gotShutdown = false;

2931
    VIR_FREE(priv->pidfile);
2932
    if (!(priv->pidfile = virPidFileBuildPath(driver->stateDir, vm->def->name))) {
2933 2934
        virReportSystemError(errno,
                             "%s", _("Failed to build pidfile path."));
2935 2936 2937
        goto cleanup;
    }

2938 2939
    if (unlink(priv->pidfile) < 0 &&
        errno != ENOENT) {
2940
        virReportSystemError(errno,
2941 2942
                             _("Cannot remove stale PID file %s"),
                             priv->pidfile);
2943 2944 2945 2946 2947 2948 2949 2950 2951 2952
        goto cleanup;
    }

    /*
     * 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
     */
2953
    if (qemuCapsGet(priv->qemuCaps, QEMU_CAPS_DEVICE)) {
2954
        VIR_DEBUG("Assigning domain PCI addresses");
2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972
        /* Populate cache with current addresses */
        if (priv->pciaddrs) {
            qemuDomainPCIAddressSetFree(priv->pciaddrs);
            priv->pciaddrs = NULL;
        }
        if (!(priv->pciaddrs = qemuDomainPCIAddressSetCreate(vm->def)))
            goto cleanup;


        /* Assign any remaining addresses */
        if (qemuAssignDevicePCISlots(vm->def, priv->pciaddrs) < 0)
            goto cleanup;

        priv->persistentAddrs = 1;
    } else {
        priv->persistentAddrs = 0;
    }

2973
    VIR_DEBUG("Building emulator command line");
2974
    if (!(cmd = qemuBuildCommandLine(conn, driver, vm->def, priv->monConfig,
2975
                                     priv->monJSON != 0, priv->qemuCaps,
2976
                                     migrateFrom, stdin_fd, snapshot, vmop)))
2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009
        goto cleanup;

    /* now that we know it is about to start call the hook if present */
    if (virHookPresent(VIR_HOOK_DRIVER_QEMU)) {
        char *xml = virDomainDefFormat(vm->def, 0);
        int hookret;

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

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

    if ((timestamp = virTimestamp()) == NULL) {
        virReportOOMError();
        goto cleanup;
    } else {
        if (safewrite(logfile, timestamp, strlen(timestamp)) < 0 ||
            safewrite(logfile, START_POSTFIX, strlen(START_POSTFIX)) < 0) {
            VIR_WARN("Unable to write timestamp to logfile: %s",
                     virStrerror(errno, ebuf, sizeof ebuf));
        }

        VIR_FREE(timestamp);
    }

    virCommandWriteArgLog(cmd, logfile);

3010
    qemuDomainObjCheckTaint(driver, vm, logfile);
3011

3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025
    if ((pos = lseek(logfile, 0, SEEK_END)) < 0)
        VIR_WARN("Unable to seek to end of logfile: %s",
                 virStrerror(errno, ebuf, sizeof ebuf));

    VIR_DEBUG("Clear emulator capabilities: %d",
              driver->clearEmulatorCapabilities);
    if (driver->clearEmulatorCapabilities)
        virCommandClearCaps(cmd);

    virCommandSetPreExecHook(cmd, qemuProcessHook, &hookData);

    virCommandSetOutputFD(cmd, &logfile);
    virCommandSetErrorFD(cmd, &logfile);
    virCommandNonblockingFDs(cmd);
3026
    virCommandSetPidFile(cmd, priv->pidfile);
3027
    virCommandDaemonize(cmd);
3028
    virCommandRequireHandshake(cmd);
3029 3030 3031

    ret = virCommandRun(cmd, NULL);

E
Eric Blake 已提交
3032
    /* wait for qemu process to show up */
3033
    if (ret == 0) {
3034
        if (virPidFileReadPath(priv->pidfile, &vm->pid) < 0) {
3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("Domain %s didn't show up"), vm->def->name);
            ret = -1;
        }
#if 0
    } else if (ret == -2) {
        /*
         * XXX this is bogus. It isn't safe to set vm->pid = child
         * because the child no longer exists.
         */

3046
        /* The virCommand process that launches the daemon failed. Pending on
3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057
         * when it failed (we can't determine for sure), there may be
         * extra info in the domain log (if the hook failed for example).
         *
         * Pretend like things succeeded, and let 'WaitForMonitor' report
         * the log contents for us.
         */
        vm->pid = child;
        ret = 0;
#endif
    }

3058 3059 3060 3061 3062
    VIR_DEBUG("Writing early domain status to disk");
    if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0) {
        goto cleanup;
    }

3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 3083 3084 3085 3086
    VIR_DEBUG("Waiting for handshake from child");
    if (virCommandHandshakeWait(cmd) < 0) {
        goto cleanup;
    }

    VIR_DEBUG("Setting domain security labels");
    if (virSecurityManagerSetAllLabel(driver->securityManager,
                                      vm, stdin_path) < 0)
        goto cleanup;

    if (stdin_fd != -1) {
        /* 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");

        if (fstat(stdin_fd, &stdin_sb) < 0) {
            virReportSystemError(errno,
                                 _("cannot stat fd %d"), stdin_fd);
            goto cleanup;
        }
        if (S_ISFIFO(stdin_sb.st_mode) &&
3087
            virSecurityManagerSetImageFDLabel(driver->securityManager, vm, stdin_fd) < 0)
3088 3089 3090 3091 3092 3093 3094 3095 3096
            goto cleanup;
    }

    VIR_DEBUG("Labelling done, completing handshake to child");
    if (virCommandHandshakeNotify(cmd) < 0) {
        goto cleanup;
    }
    VIR_DEBUG("Handshake complete, child running");

3097 3098 3099 3100 3101 3102 3103 3104 3105
    if (migrateFrom)
        start_paused = true;

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

    if (ret == -1) /* The VM failed to start */
        goto cleanup;

3106
    VIR_DEBUG("Waiting for monitor to show up");
C
Cole Robinson 已提交
3107
    if (qemuProcessWaitForMonitor(driver, vm, priv->qemuCaps, pos) < 0)
3108 3109
        goto cleanup;

3110
    VIR_DEBUG("Detecting VCPU PIDs");
3111 3112 3113
    if (qemuProcessDetectVcpuPIDs(driver, vm) < 0)
        goto cleanup;

3114 3115 3116 3117
    VIR_DEBUG("Setting cgroup for each VCPU(if required)");
    if (qemuSetupCgroupForVcpu(driver, vm) < 0)
        goto cleanup;

3118
    VIR_DEBUG("Setting VCPU affinities");
3119 3120 3121
    if (qemuProcessSetVcpuAffinites(conn, vm) < 0)
        goto cleanup;

3122
    VIR_DEBUG("Setting any required VM passwords");
3123
    if (qemuProcessInitPasswords(conn, driver, vm) < 0)
3124 3125 3126 3127
        goto cleanup;

    /* If we have -device, then addresses are assigned explicitly.
     * If not, then we have to detect dynamic ones here */
3128
    if (!qemuCapsGet(priv->qemuCaps, QEMU_CAPS_DEVICE)) {
3129
        VIR_DEBUG("Determining domain device PCI addresses");
3130 3131 3132 3133
        if (qemuProcessInitPCIAddresses(driver, vm) < 0)
            goto cleanup;
    }

3134 3135 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145
    /* set default link states */
    /* qemu doesn't support setting this on the command line, so
     * enter the monitor */
    VIR_DEBUG("Setting network link states");
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
    if (qemuProcessSetLinkStates(vm) < 0) {
        qemuDomainObjExitMonitorWithDriver(driver, vm);
        goto cleanup;
    }

    qemuDomainObjExitMonitorWithDriver(driver, vm);

3146 3147 3148 3149
    /* Technically, qemuProcessStart can be called from inside
     * QEMU_ASYNC_JOB_MIGRATION_IN, but we are okay treating this like
     * a sync job since no other job can call into the domain until
     * migration completes.  */
3150
    VIR_DEBUG("Setting initial memory amount");
E
Eric Blake 已提交
3151
    cur_balloon = vm->def->mem.cur_balloon;
3152
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
E
Eric Blake 已提交
3153
    if (qemuMonitorSetBalloon(priv->mon, cur_balloon) < 0) {
3154 3155 3156 3157 3158 3159
        qemuDomainObjExitMonitorWithDriver(driver, vm);
        goto cleanup;
    }
    qemuDomainObjExitMonitorWithDriver(driver, vm);

    if (!start_paused) {
3160
        VIR_DEBUG("Starting domain CPUs");
3161
        /* Allow the CPUS to start executing */
J
Jiri Denemark 已提交
3162
        if (qemuProcessStartCPUs(driver, vm, conn,
3163 3164
                                 VIR_DOMAIN_RUNNING_BOOTED,
                                 QEMU_ASYNC_JOB_NONE) < 0) {
3165 3166 3167 3168 3169
            if (virGetLastError() == NULL)
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                "%s", _("resume operation failed"));
            goto cleanup;
        }
J
Jiri Denemark 已提交
3170 3171 3172 3173 3174
    } else {
        virDomainObjSetState(vm, VIR_DOMAIN_PAUSED,
                             migrateFrom ?
                             VIR_DOMAIN_PAUSED_MIGRATION :
                             VIR_DOMAIN_PAUSED_USER);
3175 3176
    }

3177 3178 3179 3180
    if (autodestroy &&
        qemuProcessAutoDestroyAdd(driver, vm, conn) < 0)
        goto cleanup;

3181
    VIR_DEBUG("Writing domain status to disk");
3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195
    if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
        goto cleanup;

    virCommandFree(cmd);
    VIR_FORCE_CLOSE(logfile);

    return 0;

cleanup:
    /* We jump here if we failed to start the VM for any reason, or
     * if we failed to initialize the now running VM. kill it off and
     * pretend we never started it */
    virCommandFree(cmd);
    VIR_FORCE_CLOSE(logfile);
J
Jiri Denemark 已提交
3196
    qemuProcessStop(driver, vm, 0, VIR_DOMAIN_SHUTOFF_FAILED);
3197 3198 3199 3200 3201

    return -1;
}


3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225
void qemuProcessKill(virDomainObjPtr vm)
{
    int i;
    VIR_DEBUG("vm=%s pid=%d", vm->def->name, vm->pid);

    if (!virDomainObjIsActive(vm)) {
        VIR_DEBUG("VM '%s' not active", vm->def->name);
        return;
    }

    /* This loop sends SIGTERM, then waits a few iterations
     * (1.6 seconds) to see if it dies. If still alive then
     * it does SIGKILL, and waits a few more iterations (1.6
     * seconds more) to confirm that it has really gone.
     */
    for (i = 0 ; i < 15 ; i++) {
        int signum;
        if (i == 0)
            signum = SIGTERM;
        else if (i == 8)
            signum = SIGKILL;
        else
            signum = 0; /* Just check for existence */

3226 3227
        if (virKillProcess(vm->pid, signum) < 0) {
            if (errno != ESRCH) {
3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239
                char ebuf[1024];
                VIR_WARN("Failed to kill process %d %s",
                         vm->pid, virStrerror(errno, ebuf, sizeof ebuf));
            }
            break;
        }

        usleep(200 * 1000);
    }
}


3240 3241
void qemuProcessStop(struct qemud_driver *driver,
                     virDomainObjPtr vm,
J
Jiri Denemark 已提交
3242 3243
                     int migrated,
                     virDomainShutoffReason reason)
3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262
{
    int ret;
    int retries = 0;
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virErrorPtr orig_err;
    virDomainDefPtr def;
    int i;
    int logfile = -1;
    char *timestamp;
    char ebuf[1024];

    VIR_DEBUG("Shutting down VM '%s' pid=%d migrated=%d",
              vm->def->name, vm->pid, migrated);

    if (!virDomainObjIsActive(vm)) {
        VIR_DEBUG("VM '%s' not active", vm->def->name);
        return;
    }

3263
    if ((logfile = qemuDomainCreateLog(driver, vm, true)) < 0) {
3264 3265 3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318
        /* To not break the normal domain shutdown process, skip the
         * timestamp log writing if failed on opening log file. */
        VIR_WARN("Unable to open logfile: %s",
                  virStrerror(errno, ebuf, sizeof ebuf));
    } else {
        if ((timestamp = virTimestamp()) == NULL) {
            virReportOOMError();
        } else {
            if (safewrite(logfile, timestamp, strlen(timestamp)) < 0 ||
                safewrite(logfile, SHUTDOWN_POSTFIX,
                          strlen(SHUTDOWN_POSTFIX)) < 0) {
                VIR_WARN("Unable to write timestamp to logfile: %s",
                         virStrerror(errno, ebuf, sizeof ebuf));
            }

            VIR_FREE(timestamp);
        }

        if (VIR_CLOSE(logfile) < 0)
             VIR_WARN("Unable to close logfile: %s",
                      virStrerror(errno, ebuf, sizeof ebuf));
    }

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

    virDomainConfVMNWFilterTeardown(vm);

    if (driver->macFilter) {
        def = vm->def;
        for (i = 0 ; i < def->nnets ; i++) {
            virDomainNetDefPtr net = def->nets[i];
            if (net->ifname == NULL)
                continue;
            if ((errno = networkDisallowMacOnPort(driver, net->ifname,
                                                  net->mac))) {
                virReportSystemError(errno,
             _("failed to remove ebtables rule to allow MAC address on  '%s'"),
                                     net->ifname);
            }
        }
    }

    if (priv->mon)
        qemuMonitorClose(priv->mon);

    if (priv->monConfig) {
        if (priv->monConfig->type == VIR_DOMAIN_CHR_TYPE_UNIX)
            unlink(priv->monConfig->data.nix.path);
        virDomainChrSourceDefFree(priv->monConfig);
        priv->monConfig = NULL;
    }

    /* shut it off for sure */
3319
    qemuProcessKill(vm);
3320

3321 3322 3323
    /* Stop autodestroy in case guest is restarted */
    qemuProcessAutoDestroyRemove(driver, vm);

3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340
    /* now that we know it's stopped call the hook if present */
    if (virHookPresent(VIR_HOOK_DRIVER_QEMU)) {
        char *xml = virDomainDefFormat(vm->def, 0);

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

    /* Reset Security Labels */
    virSecurityManagerRestoreAllLabel(driver->securityManager,
                                      vm, migrated);
    virSecurityManagerReleaseLabel(driver->securityManager, vm);

    /* Clear out dynamically assigned labels */
    if (vm->def->seclabel.type == VIR_DOMAIN_SECLABEL_DYNAMIC) {
3341 3342
        if (!vm->def->seclabel.baselabel)
            VIR_FREE(vm->def->seclabel.model);
3343 3344
        VIR_FREE(vm->def->seclabel.label);
    }
3345
    VIR_FREE(vm->def->seclabel.imagelabel);
3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358

    virDomainDefClearDeviceAliases(vm->def);
    if (!priv->persistentAddrs) {
        virDomainDefClearPCIAddresses(vm->def);
        qemuDomainPCIAddressSetFree(priv->pciaddrs);
        priv->pciaddrs = NULL;
    }

    qemuDomainReAttachHostDevices(driver, vm->def);

    def = vm->def;
    for (i = 0; i < def->nnets; i++) {
        virDomainNetDefPtr net = def->nets[i];
3359
#if WITH_MACVTAP
3360 3361 3362 3363 3364 3365
        if (virDomainNetGetActualType(net) == VIR_DOMAIN_NET_TYPE_DIRECT) {
            delMacvtap(net->ifname, net->mac,
                       virDomainNetGetActualDirectDev(net),
                       virDomainNetGetActualDirectMode(net),
                       virDomainNetGetActualDirectVirtPortProfile(net),
                       driver->stateDir);
3366 3367 3368
            VIR_FREE(net->ifname);
        }
#endif
3369 3370 3371 3372 3373
        /* release the physical device (or any other resources used by
         * this interface in the network driver
         */
        networkReleaseActualDevice(net);
    }
3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401

retry:
    if ((ret = qemuRemoveCgroup(driver, vm, 0)) < 0) {
        if (ret == -EBUSY && (retries++ < 5)) {
            usleep(200*1000);
            goto retry;
        }
        VIR_WARN("Failed to remove cgroup for %s",
                 vm->def->name);
    }

    qemuProcessRemoveDomainStatus(driver, vm);

    /* Remove VNC port from port reservation bitmap, but only if it was
       reserved by the driver (autoport=yes)
    */
    if ((vm->def->ngraphics == 1) &&
        vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
        vm->def->graphics[0]->data.vnc.autoport) {
        qemuProcessReturnPort(driver, vm->def->graphics[0]->data.vnc.port);
    }
    if ((vm->def->ngraphics == 1) &&
        vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_SPICE &&
        vm->def->graphics[0]->data.spice.autoport) {
        qemuProcessReturnPort(driver, vm->def->graphics[0]->data.spice.port);
        qemuProcessReturnPort(driver, vm->def->graphics[0]->data.spice.tlsPort);
    }

3402
    vm->taint = 0;
3403 3404
    vm->pid = -1;
    vm->def->id = -1;
J
Jiri Denemark 已提交
3405
    virDomainObjSetState(vm, VIR_DOMAIN_SHUTOFF, reason);
3406 3407
    VIR_FREE(priv->vcpupids);
    priv->nvcpupids = 0;
3408 3409
    qemuCapsFree(priv->qemuCaps);
    priv->qemuCaps = NULL;
3410
    VIR_FREE(priv->pidfile);
3411

3412
    /* The "release" hook cleans up additional resources */
3413 3414 3415 3416 3417 3418 3419 3420 3421
    if (virHookPresent(VIR_HOOK_DRIVER_QEMU)) {
        char *xml = virDomainDefFormat(vm->def, 0);

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

3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433
    if (vm->newDef) {
        virDomainDefFree(vm->def);
        vm->def = vm->newDef;
        vm->def->id = -1;
        vm->newDef = NULL;
    }

    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
}
3434 3435


3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572
int qemuProcessAttach(virConnectPtr conn ATTRIBUTE_UNUSED,
                      struct qemud_driver *driver,
                      virDomainObjPtr vm,
                      int pid,
                      const char *pidfile,
                      virDomainChrSourceDefPtr monConfig,
                      bool monJSON)
{
    char ebuf[1024];
    int logfile = -1;
    char *timestamp;
    qemuDomainObjPrivatePtr priv = vm->privateData;
    bool running = true;
    virSecurityLabelPtr seclabel = NULL;

    VIR_DEBUG("Beginning VM attach process");

    if (virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("VM is already active"));
        return -1;
    }

    /* 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(driver->caps, vm, true) < 0)
        goto cleanup;

    vm->def->id = driver->nextvmid++;

    if (virFileMakePath(driver->logDir) < 0) {
        virReportSystemError(errno,
                             _("cannot create log directory %s"),
                             driver->logDir);
        goto cleanup;
    }

    VIR_FREE(priv->pidfile);
    if (pidfile &&
        !(priv->pidfile = strdup(pidfile)))
        goto no_memory;

    VIR_DEBUG("Detect security driver config");
    vm->def->seclabel.type = VIR_DOMAIN_SECLABEL_STATIC;
    if (VIR_ALLOC(seclabel) < 0)
        goto no_memory;
    if (virSecurityManagerGetProcessLabel(driver->securityManager,
                                          vm, seclabel) < 0)
        goto cleanup;
    if (!(vm->def->seclabel.model = strdup(driver->caps->host.secModel.model)))
        goto no_memory;
    if (!(vm->def->seclabel.label = strdup(seclabel->label)))
        goto no_memory;

    VIR_DEBUG("Creating domain log file");
    if ((logfile = qemuDomainCreateLog(driver, vm, false)) < 0)
        goto cleanup;

    VIR_DEBUG("Determining emulator version");
    qemuCapsFree(priv->qemuCaps);
    priv->qemuCaps = NULL;
    if (qemuCapsExtractVersionInfo(vm->def->emulator,
                                   vm->def->os.arch,
                                   NULL,
                                   &priv->qemuCaps) < 0)
        goto cleanup;

    VIR_DEBUG("Preparing monitor state");
    priv->monConfig = monConfig;
    monConfig = NULL;
    priv->monJSON = monJSON;

    priv->gotShutdown = false;

    /*
     * Normally PCI addresses are assigned in the virDomainCreate
     * or virDomainDefine methods. We might still need to assign
     * some here to cope with the question of upgrades. Regardless
     * we also need to populate the PCi address set cache for later
     * use in hotplug
     */
    if (qemuCapsGet(priv->qemuCaps, QEMU_CAPS_DEVICE)) {
        VIR_DEBUG("Assigning domain PCI addresses");
        /* Populate cache with current addresses */
        if (priv->pciaddrs) {
            qemuDomainPCIAddressSetFree(priv->pciaddrs);
            priv->pciaddrs = NULL;
        }
        if (!(priv->pciaddrs = qemuDomainPCIAddressSetCreate(vm->def)))
            goto cleanup;

        /* Assign any remaining addresses */
        if (qemuAssignDevicePCISlots(vm->def, priv->pciaddrs) < 0)
            goto cleanup;

        priv->persistentAddrs = 1;
    } else {
        priv->persistentAddrs = 0;
    }

    if ((timestamp = virTimestamp()) == NULL) {
        virReportOOMError();
        goto cleanup;
    } else {
        if (safewrite(logfile, timestamp, strlen(timestamp)) < 0 ||
            safewrite(logfile, ATTACH_POSTFIX, strlen(ATTACH_POSTFIX)) < 0) {
            VIR_WARN("Unable to write timestamp to logfile: %s",
                     virStrerror(errno, ebuf, sizeof ebuf));
        }

        VIR_FREE(timestamp);
    }

    qemuDomainObjTaint(driver, vm, VIR_DOMAIN_TAINT_EXTERNAL_LAUNCH, logfile);

    vm->pid = pid;

    VIR_DEBUG("Waiting for monitor to show up");
    if (qemuProcessWaitForMonitor(driver, vm, priv->qemuCaps, -1) < 0)
        goto cleanup;

    VIR_DEBUG("Detecting VCPU PIDs");
    if (qemuProcessDetectVcpuPIDs(driver, vm) < 0)
        goto cleanup;

    /* If we have -device, then addresses are assigned explicitly.
     * If not, then we have to detect dynamic ones here */
    if (!qemuCapsGet(priv->qemuCaps, QEMU_CAPS_DEVICE)) {
        VIR_DEBUG("Determining domain device PCI addresses");
        if (qemuProcessInitPCIAddresses(driver, vm) < 0)
            goto cleanup;
    }

    VIR_DEBUG("Getting initial memory amount");
3573
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 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 3615 3616 3617 3618
    if (qemuMonitorGetBalloonInfo(priv->mon, &vm->def->mem.cur_balloon) < 0) {
        qemuDomainObjExitMonitorWithDriver(driver, vm);
        goto cleanup;
    }
    if (qemuMonitorGetStatus(priv->mon, &running) < 0) {
        qemuDomainObjExitMonitorWithDriver(driver, vm);
        goto cleanup;
    }
    if (qemuMonitorGetVirtType(priv->mon, &vm->def->virtType) < 0) {
        qemuDomainObjExitMonitorWithDriver(driver, vm);
        goto cleanup;
    }
    qemuDomainObjExitMonitorWithDriver(driver, vm);

    if (!virDomainObjIsActive(vm))
        goto cleanup;

    if (running)
        virDomainObjSetState(vm, VIR_DOMAIN_RUNNING,
                             VIR_DOMAIN_RUNNING_UNPAUSED);
    else
        virDomainObjSetState(vm, VIR_DOMAIN_PAUSED, VIR_DOMAIN_PAUSED_UNKNOWN);

    VIR_DEBUG("Writing domain status to disk");
    if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
        goto cleanup;

    VIR_FORCE_CLOSE(logfile);
    VIR_FREE(seclabel);

    return 0;

no_memory:
    virReportOOMError();
cleanup:
    /* We jump here if we failed to start the VM for any reason, or
     * if we failed to initialize the now running VM. kill it off and
     * pretend we never started it */
    VIR_FORCE_CLOSE(logfile);
    VIR_FREE(seclabel);
    virDomainChrSourceDefFree(monConfig);
    return -1;
}


3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660
int qemuProcessAutoDestroyInit(struct qemud_driver *driver)
{
    if (!(driver->autodestroy = virHashCreate(5, NULL)))
        return -1;

    return 0;
}

struct qemuProcessAutoDestroyData {
    struct qemud_driver *driver;
    virConnectPtr conn;
};

static void qemuProcessAutoDestroyDom(void *payload,
                                      const void *name,
                                      void *opaque)
{
    struct qemuProcessAutoDestroyData *data = opaque;
    virConnectPtr conn = payload;
    const char *uuidstr = name;
    unsigned char uuid[VIR_UUID_BUFLEN];
    virDomainObjPtr dom;
    qemuDomainObjPrivatePtr priv;
    virDomainEventPtr event = NULL;

    VIR_DEBUG("conn=%p uuidstr=%s thisconn=%p", conn, uuidstr, data->conn);

    if (data->conn != conn)
        return;

    if (virUUIDParse(uuidstr, uuid) < 0) {
        VIR_WARN("Failed to parse %s", uuidstr);
        return;
    }

    if (!(dom = virDomainFindByUUID(&data->driver->domains,
                                    uuid))) {
        VIR_DEBUG("No domain object to kill");
        return;
    }

    priv = dom->privateData;
3661 3662
    if (priv->job.asyncJob) {
        VIR_DEBUG("vm=%s has long-term job active, cancelling",
3663
                  dom->def->name);
3664
        qemuDomainObjDiscardAsyncJob(data->driver, dom);
3665 3666
    }

3667 3668
    if (qemuDomainObjBeginJobWithDriver(data->driver, dom,
                                        QEMU_JOB_DESTROY) < 0)
3669 3670 3671 3672
        goto cleanup;

    VIR_DEBUG("Killing domain");
    qemuProcessStop(data->driver, dom, 1, VIR_DOMAIN_SHUTOFF_DESTROYED);
3673
    virDomainAuditStop(dom, "destroyed");
3674 3675 3676
    event = virDomainEventNewFromObj(dom,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_DESTROYED);
3677
    if (qemuDomainObjEndJob(data->driver, dom) == 0)
3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 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 3725 3726 3727 3728
        dom = NULL;
    if (dom && !dom->persistent)
        virDomainRemoveInactive(&data->driver->domains, dom);

cleanup:
    if (dom)
        virDomainObjUnlock(dom);
    if (event)
        qemuDomainEventQueue(data->driver, event);
    virHashRemoveEntry(data->driver->autodestroy, uuidstr);
}

/*
 * Precondition: driver is locked
 */
void qemuProcessAutoDestroyRun(struct qemud_driver *driver, virConnectPtr conn)
{
    struct qemuProcessAutoDestroyData data = {
        driver, conn
    };
    VIR_DEBUG("conn=%p", conn);
    virHashForEach(driver->autodestroy, qemuProcessAutoDestroyDom, &data);
}

void qemuProcessAutoDestroyShutdown(struct qemud_driver *driver)
{
    virHashFree(driver->autodestroy);
}

int qemuProcessAutoDestroyAdd(struct qemud_driver *driver,
                              virDomainObjPtr vm,
                              virConnectPtr conn)
{
    char uuidstr[VIR_UUID_STRING_BUFLEN];
    virUUIDFormat(vm->def->uuid, uuidstr);
    VIR_DEBUG("vm=%s uuid=%s conn=%p", vm->def->name, uuidstr, conn);
    if (virHashAddEntry(driver->autodestroy, uuidstr, conn) < 0)
        return -1;
    return 0;
}

int qemuProcessAutoDestroyRemove(struct qemud_driver *driver,
                                 virDomainObjPtr vm)
{
    char uuidstr[VIR_UUID_STRING_BUFLEN];
    virUUIDFormat(vm->def->uuid, uuidstr);
    VIR_DEBUG("vm=%s uuid=%s", vm->def->name, uuidstr);
    if (virHashRemoveEntry(driver->autodestroy, uuidstr) < 0)
        return -1;
    return 0;
}
3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739

bool qemuProcessAutoDestroyActive(struct qemud_driver *driver,
                                  virDomainObjPtr vm)
{
    char uuidstr[VIR_UUID_STRING_BUFLEN];
    virUUIDFormat(vm->def->uuid, uuidstr);
    VIR_DEBUG("vm=%s uuid=%s", vm->def->name, uuidstr);
    if (virHashLookup(driver->autodestroy, uuidstr) != NULL)
        return true;
    return false;
}