qemu_migration.c 57.5 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
/*
 * qemu_migration.c: QEMU migration handling
 *
 * 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 <sys/time.h>
25 26
#include <gnutls/gnutls.h>
#include <gnutls/x509.h>
27 28 29 30 31 32 33

#include "qemu_migration.h"
#include "qemu_monitor.h"
#include "qemu_domain.h"
#include "qemu_process.h"
#include "qemu_capabilities.h"
#include "qemu_audit.h"
34
#include "qemu_cgroup.h"
35 36 37 38 39 40 41 42

#include "logging.h"
#include "virterror_internal.h"
#include "memory.h"
#include "util.h"
#include "files.h"
#include "datatypes.h"
#include "fdstream.h"
43 44
#include "uuid.h"

45 46 47 48 49

#define VIR_FROM_THIS VIR_FROM_QEMU

#define timeval_to_ms(tv)       (((tv).tv_sec * 1000ull) + ((tv).tv_usec / 1000))

50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307
typedef struct _qemuMigrationCookie qemuMigrationCookie;
typedef qemuMigrationCookie *qemuMigrationCookiePtr;
struct _qemuMigrationCookie {
    int flags;

    /* Host properties */
    unsigned char hostuuid[VIR_UUID_BUFLEN];
    char *hostname;

    /* Guest properties */
    unsigned char uuid[VIR_UUID_BUFLEN];
    char *name;
};


static void qemuMigrationCookieFree(qemuMigrationCookiePtr mig)
{
    if (!mig)
        return;

    VIR_FREE(mig->hostname);
    VIR_FREE(mig->name);
    VIR_FREE(mig);
}


static qemuMigrationCookiePtr
qemuMigrationCookieNew(virDomainObjPtr dom)
{
    qemuMigrationCookiePtr mig = NULL;

    if (VIR_ALLOC(mig) < 0)
        goto no_memory;

    if (!(mig->name = strdup(dom->def->name)))
        goto no_memory;
    memcpy(mig->uuid, dom->def->uuid, VIR_UUID_BUFLEN);

    if (!(mig->hostname = virGetHostname(NULL)))
        goto no_memory;
    if (virGetHostUUID(mig->hostuuid) < 0) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("Unable to obtain host UUID"));
        goto error;
    }

    return mig;

no_memory:
    virReportOOMError();
error:
    qemuMigrationCookieFree(mig);
    return NULL;
}


static void qemuMigrationCookieXMLFormat(virBufferPtr buf,
                                         qemuMigrationCookiePtr mig)
{
    char uuidstr[VIR_UUID_STRING_BUFLEN];
    char hostuuidstr[VIR_UUID_STRING_BUFLEN];

    virUUIDFormat(mig->uuid, uuidstr);
    virUUIDFormat(mig->hostuuid, hostuuidstr);

    virBufferAsprintf(buf, "<qemu-migration>\n");
    virBufferEscapeString(buf, "  <name>%s</name>\n", mig->name);
    virBufferAsprintf(buf, "  <uuid>%s</uuid>\n", uuidstr);
    virBufferEscapeString(buf, "  <hostname>%s</hostname>\n", mig->hostname);
    virBufferAsprintf(buf, "  <hostuuid>%s</hostuuid>\n", hostuuidstr);
    virBufferAddLit(buf, "</qemu-migration>\n");
}


static char *qemuMigrationCookieXMLFormatStr(qemuMigrationCookiePtr mig)
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;

    qemuMigrationCookieXMLFormat(&buf, mig);

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

    return virBufferContentAndReset(&buf);
}


static int
qemuMigrationCookieXMLParse(qemuMigrationCookiePtr mig,
                            xmlXPathContextPtr ctxt,
                            int flags ATTRIBUTE_UNUSED)
{
    char uuidstr[VIR_UUID_STRING_BUFLEN];
    char *tmp;

    /* We don't store the uuid, name, hostname, or hostuuid
     * values. We just compare them to local data to do some
     * sanity checking on migration operation
     */

    /* Extract domain name */
    if (!(tmp = virXPathString("string(./name[1])", ctxt))) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("missing name element in migration data"));
        goto error;
    }
    if (STRNEQ(tmp, mig->name)) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Incoming cookie data had unexpected name %s vs %s"),
                        tmp, mig->name);
        goto error;
    }
    VIR_FREE(tmp);

    /* Extract domain uuid */
    tmp = virXPathString("string(./uuid[1])", ctxt);
    if (!tmp) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("missing uuid element in migration data"));
        goto error;
    }
    virUUIDFormat(mig->uuid, uuidstr);
    if (STRNEQ(tmp, uuidstr)) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Incoming cookie data had unexpected UUID %s vs %s"),
                        tmp, uuidstr);
    }
    VIR_FREE(tmp);

    /* Check & forbid "localhost" migration */
    if (!(tmp = virXPathString("string(./hostname[1])", ctxt))) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("missing hostname element in migration data"));
        goto error;
    }
    if (STREQ(tmp, mig->hostname)) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Attempt to migrate guest to the same host %s"),
                        tmp);
        goto error;
    }
    VIR_FREE(tmp);

    if (!(tmp = virXPathString("string(./hostuuid[1])", ctxt))) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("missing hostuuid element in migration data"));
        goto error;
    }
    virUUIDFormat(mig->hostuuid, uuidstr);
    if (STREQ(tmp, uuidstr)) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Attempt to migrate guest to the same host %s"),
                        tmp);
        goto error;
    }
    VIR_FREE(tmp);

    return 0;

error:
    VIR_FREE(tmp);
    return -1;
}


static int
qemuMigrationCookieXMLParseStr(qemuMigrationCookiePtr mig,
                               const char *xml,
                               int flags)
{
    xmlDocPtr doc = NULL;
    xmlXPathContextPtr ctxt = NULL;
    int ret;

    VIR_DEBUG("xml=%s", NULLSTR(xml));

    if (!(doc = virXMLParseString(xml, "qemumigration.xml")))
        goto cleanup;

    if ((ctxt = xmlXPathNewContext(doc)) == NULL) {
        virReportOOMError();
        goto cleanup;
    }

    ctxt->node = xmlDocGetRootElement(doc);

    ret = qemuMigrationCookieXMLParse(mig, ctxt, flags);

cleanup:
    xmlXPathFreeContext(ctxt);
    xmlFreeDoc(doc);

    return ret;
}


static int
qemuMigrationBakeCookie(qemuMigrationCookiePtr mig,
                        struct qemud_driver *driver ATTRIBUTE_UNUSED,
                        virDomainObjPtr dom ATTRIBUTE_UNUSED,
                        char **cookieout,
                        int *cookieoutlen,
                        int flags ATTRIBUTE_UNUSED)
{
    if (!cookieout || !cookieoutlen) {
        qemuReportError(VIR_ERR_INVALID_ARG, "%s",
                        _("missing migration cookie data"));
        return -1;
    }

    *cookieoutlen = 0;

    if (!(*cookieout = qemuMigrationCookieXMLFormatStr(mig)))
        return -1;

    *cookieoutlen = strlen(*cookieout) + 1;

    VIR_DEBUG("cookielen=%d cookie=%s", *cookieoutlen, *cookieout);

    return 0;
}


static qemuMigrationCookiePtr
qemuMigrationEatCookie(virDomainObjPtr dom,
                       const char *cookiein,
                       int cookieinlen,
                       int flags)
{
    qemuMigrationCookiePtr mig = NULL;

    /* Parse & validate incoming cookie (if any) */
    if (cookiein && cookieinlen &&
        cookiein[cookieinlen-1] != '\0') {
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("Migration cookie was not NULL terminated"));
        goto error;
    }

    VIR_DEBUG("cookielen=%d cookie='%s'", cookieinlen, NULLSTR(cookiein));

    if (!(mig = qemuMigrationCookieNew(dom)))
        return NULL;

    if (cookiein && cookieinlen &&
        qemuMigrationCookieXMLParseStr(mig,
                                       cookiein,
                                       flags) < 0)
        goto error;

    return mig;

error:
    qemuMigrationCookieFree(mig);
    return NULL;
}
308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329

bool
qemuMigrationIsAllowed(virDomainDefPtr def)
{
    if (def->nhostdevs > 0) {
        qemuReportError(VIR_ERR_OPERATION_INVALID,
            "%s", _("Domain with assigned host devices cannot be migrated"));
        return false;
    }

    return true;
}

/** qemuMigrationSetOffline
 * Pause domain for non-live migration.
 */
int
qemuMigrationSetOffline(struct qemud_driver *driver,
                        virDomainObjPtr vm)
{
    int ret;

J
Jiri Denemark 已提交
330
    ret = qemuProcessStopCPUs(driver, vm, VIR_DOMAIN_PAUSED_MIGRATION);
331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386
    if (ret == 0) {
        virDomainEventPtr event;

        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_SUSPENDED,
                                         VIR_DOMAIN_EVENT_SUSPENDED_MIGRATED);
        if (event)
            qemuDomainEventQueue(driver, event);
    }

    return ret;
}


int
qemuMigrationWaitForCompletion(struct qemud_driver *driver, virDomainObjPtr vm)
{
    int ret = -1;
    int status;
    unsigned long long memProcessed;
    unsigned long long memRemaining;
    unsigned long long memTotal;
    qemuDomainObjPrivatePtr priv = vm->privateData;

    priv->jobInfo.type = VIR_DOMAIN_JOB_UNBOUNDED;

    while (priv->jobInfo.type == VIR_DOMAIN_JOB_UNBOUNDED) {
        /* Poll every 50ms for progress & to allow cancellation */
        struct timespec ts = { .tv_sec = 0, .tv_nsec = 50 * 1000 * 1000ull };
        struct timeval now;
        int rc;
        const char *job;

        switch (priv->jobActive) {
            case QEMU_JOB_MIGRATION_OUT:
                job = _("migration job");
                break;
            case QEMU_JOB_SAVE:
                job = _("domain save job");
                break;
            case QEMU_JOB_DUMP:
                job = _("domain core dump job");
                break;
            default:
                job = _("job");
        }


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

        if (priv->jobSignals & QEMU_JOB_SIGNAL_CANCEL) {
            priv->jobSignals ^= QEMU_JOB_SIGNAL_CANCEL;
387
            VIR_DEBUG("Cancelling job at client request");
388 389 390 391
            qemuDomainObjEnterMonitorWithDriver(driver, vm);
            rc = qemuMonitorMigrateCancel(priv->mon);
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            if (rc < 0) {
392
                VIR_WARN("Unable to cancel job");
393 394 395
            }
        } else if (priv->jobSignals & QEMU_JOB_SIGNAL_SUSPEND) {
            priv->jobSignals ^= QEMU_JOB_SIGNAL_SUSPEND;
396
            VIR_DEBUG("Pausing domain for non-live migration");
397
            if (qemuMigrationSetOffline(driver, vm) < 0)
398
                VIR_WARN("Unable to pause domain");
399 400 401 402 403 404 405 406 407 408
        } else if (priv->jobSignals & QEMU_JOB_SIGNAL_MIGRATE_DOWNTIME) {
            unsigned long long ms = priv->jobSignalsData.migrateDowntime;

            priv->jobSignals ^= QEMU_JOB_SIGNAL_MIGRATE_DOWNTIME;
            priv->jobSignalsData.migrateDowntime = 0;
            VIR_DEBUG("Setting migration downtime to %llums", ms);
            qemuDomainObjEnterMonitorWithDriver(driver, vm);
            rc = qemuMonitorSetMigrationDowntime(priv->mon, ms);
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            if (rc < 0)
409
                VIR_WARN("Unable to set migration downtime");
410 411 412 413 414 415 416 417 418 419
        } else if (priv->jobSignals & QEMU_JOB_SIGNAL_MIGRATE_SPEED) {
            unsigned long bandwidth = priv->jobSignalsData.migrateBandwidth;

            priv->jobSignals ^= QEMU_JOB_SIGNAL_MIGRATE_SPEED;
            priv->jobSignalsData.migrateBandwidth = 0;
            VIR_DEBUG("Setting migration bandwidth to %luMbs", bandwidth);
            qemuDomainObjEnterMonitorWithDriver(driver, vm);
            rc = qemuMonitorSetMigrationSpeed(priv->mon, bandwidth);
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            if (rc < 0)
420
                VIR_WARN("Unable to set migration speed");
421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509
        }

        /* Repeat check because the job signals might have caused
         * guest to die
         */
        if (!virDomainObjIsActive(vm)) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR, _("%s: %s"),
                            job, _("guest unexpectedly quit"));
            goto cleanup;
        }

        qemuDomainObjEnterMonitorWithDriver(driver, vm);
        rc = qemuMonitorGetMigrationStatus(priv->mon,
                                           &status,
                                           &memProcessed,
                                           &memRemaining,
                                           &memTotal);
        qemuDomainObjExitMonitorWithDriver(driver, vm);

        if (rc < 0) {
            priv->jobInfo.type = VIR_DOMAIN_JOB_FAILED;
            goto cleanup;
        }

        if (gettimeofday(&now, NULL) < 0) {
            priv->jobInfo.type = VIR_DOMAIN_JOB_FAILED;
            virReportSystemError(errno, "%s",
                                 _("cannot get time of day"));
            goto cleanup;
        }
        priv->jobInfo.timeElapsed = timeval_to_ms(now) - priv->jobStart;

        switch (status) {
        case QEMU_MONITOR_MIGRATION_STATUS_INACTIVE:
            priv->jobInfo.type = VIR_DOMAIN_JOB_NONE;
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("%s: %s"), job, _("is not active"));
            break;

        case QEMU_MONITOR_MIGRATION_STATUS_ACTIVE:
            priv->jobInfo.dataTotal = memTotal;
            priv->jobInfo.dataRemaining = memRemaining;
            priv->jobInfo.dataProcessed = memProcessed;

            priv->jobInfo.memTotal = memTotal;
            priv->jobInfo.memRemaining = memRemaining;
            priv->jobInfo.memProcessed = memProcessed;
            break;

        case QEMU_MONITOR_MIGRATION_STATUS_COMPLETED:
            priv->jobInfo.type = VIR_DOMAIN_JOB_COMPLETED;
            ret = 0;
            break;

        case QEMU_MONITOR_MIGRATION_STATUS_ERROR:
            priv->jobInfo.type = VIR_DOMAIN_JOB_FAILED;
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("%s: %s"), job, _("unexpectedly failed"));
            break;

        case QEMU_MONITOR_MIGRATION_STATUS_CANCELLED:
            priv->jobInfo.type = VIR_DOMAIN_JOB_CANCELLED;
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("%s: %s"), job, _("canceled by client"));
            break;
        }

        virDomainObjUnlock(vm);
        qemuDriverUnlock(driver);

        nanosleep(&ts, NULL);

        qemuDriverLock(driver);
        virDomainObjLock(vm);
    }

cleanup:
    return ret;
}


/* Prepare is the first step, and it runs on the destination host.
 *
 * This version starts an empty VM listening on a localhost TCP port, and
 * sets up the corresponding virStream to handle the incoming data.
 */
int
qemuMigrationPrepareTunnel(struct qemud_driver *driver,
                           virConnectPtr dconn,
510 511 512 513
                           const char *cookiein,
                           int cookieinlen,
                           char **cookieout,
                           int *cookieoutlen,
514 515 516 517 518 519 520 521 522
                           virStreamPtr st,
                           const char *dname,
                           const char *dom_xml)
{
    virDomainDefPtr def = NULL;
    virDomainObjPtr vm = NULL;
    virDomainEventPtr event = NULL;
    int ret = -1;
    int internalret;
523
    int dataFD[2] = { -1, -1 };
524 525
    qemuDomainObjPrivatePtr priv = NULL;
    struct timeval now;
526
    qemuMigrationCookiePtr mig = NULL;
527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561

    if (gettimeofday(&now, NULL) < 0) {
        virReportSystemError(errno, "%s",
                             _("cannot get time of day"));
        return -1;
    }

    /* Parse the domain XML. */
    if (!(def = virDomainDefParseString(driver->caps, dom_xml,
                                        VIR_DOMAIN_XML_INACTIVE)))
        goto cleanup;

    if (!qemuMigrationIsAllowed(def))
        goto cleanup;

    /* Target domain name, maybe renamed. */
    if (dname) {
        VIR_FREE(def->name);
        def->name = strdup(dname);
        if (def->name == NULL)
            goto cleanup;
    }

    if (virDomainObjIsDuplicate(&driver->domains, def, 1) < 0)
        goto cleanup;

    if (!(vm = virDomainAssignDef(driver->caps,
                                  &driver->domains,
                                  def, true))) {
        /* virDomainAssignDef already set the error */
        goto cleanup;
    }
    def = NULL;
    priv = vm->privateData;

562 563 564
    if (!(mig = qemuMigrationEatCookie(vm, cookiein, cookieinlen, 0)))
        goto cleanup;

565 566 567 568 569 570 571
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;
    priv->jobActive = QEMU_JOB_MIGRATION_OUT;

    /* Domain starts inactive, even if the domain XML had an id field. */
    vm->def->id = -1;

572
    if (pipe(dataFD) < 0 ||
573
        virSetCloseExec(dataFD[1]) < 0) {
574 575
        virReportSystemError(errno, "%s",
                             _("cannot create pipe for tunnelled migration"));
576 577 578 579
        goto endjob;
    }

    /* Start the QEMU daemon, with the same command-line arguments plus
580
     * -incoming stdio (which qemu_command might convert to exec:cat or fd:n)
581
     */
582
    internalret = qemuProcessStart(dconn, driver, vm, "stdio", true, dataFD[0],
583
                                   NULL, VIR_VM_OP_MIGRATE_IN_START);
584
    if (internalret < 0) {
585
        qemuAuditDomainStart(vm, "migrated", false);
586 587 588 589 590 591 592 593 594 595
        /* Note that we don't set an error here because qemuProcessStart
         * should have already done that.
         */
        if (!vm->persistent) {
            virDomainRemoveInactive(&driver->domains, vm);
            vm = NULL;
        }
        goto endjob;
    }

596
    if (virFDStreamOpen(st, dataFD[1]) < 0) {
597
        qemuAuditDomainStart(vm, "migrated", false);
J
Jiri Denemark 已提交
598
        qemuProcessStop(driver, vm, 0, VIR_DOMAIN_SHUTOFF_FAILED);
599 600 601 602 603
        if (!vm->persistent) {
            if (qemuDomainObjEndJob(vm) > 0)
                virDomainRemoveInactive(&driver->domains, vm);
            vm = NULL;
        }
604 605
        virReportSystemError(errno, "%s",
                             _("cannot pass pipe for tunnelled migration"));
606 607
        goto endjob;
    }
608
    dataFD[1] = -1; /* 'st' owns the FD now & will close it */
609

610
    qemuAuditDomainStart(vm, "migrated", true);
611 612 613 614

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_MIGRATED);
615 616 617 618 619 620 621 622 623

    if (qemuMigrationBakeCookie(mig, driver, vm, cookieout, cookieoutlen, 0) < 0) {
        /* We could tear down the whole guest here, but
         * cookie data is (so far) non-critical, so that
         * seems a little harsh. We'll just warn for now.
         */
        VIR_WARN("Unable to encode migration cookie");
    }

624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644
    ret = 0;

endjob:
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;

    /* We set a fake job active which is held across
     * API calls until the finish() call. This prevents
     * any other APIs being invoked while incoming
     * migration is taking place
     */
    if (vm &&
        virDomainObjIsActive(vm)) {
        priv->jobActive = QEMU_JOB_MIGRATION_IN;
        priv->jobInfo.type = VIR_DOMAIN_JOB_UNBOUNDED;
        priv->jobStart = timeval_to_ms(now);
    }

cleanup:
    virDomainDefFree(def);
645 646
    VIR_FORCE_CLOSE(dataFD[0]);
    VIR_FORCE_CLOSE(dataFD[1]);
647 648 649 650
    if (vm)
        virDomainObjUnlock(vm);
    if (event)
        qemuDomainEventQueue(driver, event);
651
    qemuMigrationCookieFree(mig);
652 653 654 655 656 657 658
    return ret;
}


int
qemuMigrationPrepareDirect(struct qemud_driver *driver,
                           virConnectPtr dconn,
659 660 661 662
                           const char *cookiein,
                           int cookieinlen,
                           char **cookieout,
                           int *cookieoutlen,
663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679
                           const char *uri_in,
                           char **uri_out,
                           const char *dname,
                           const char *dom_xml)
{
    static int port = 0;
    virDomainDefPtr def = NULL;
    virDomainObjPtr vm = NULL;
    int this_port;
    char *hostname = NULL;
    char migrateFrom [64];
    const char *p;
    virDomainEventPtr event = NULL;
    int ret = -1;
    int internalret;
    qemuDomainObjPrivatePtr priv = NULL;
    struct timeval now;
680
    qemuMigrationCookiePtr mig = NULL;
681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789

    if (gettimeofday(&now, NULL) < 0) {
        virReportSystemError(errno, "%s",
                             _("cannot get time of day"));
        return -1;
    }

    /* The URI passed in may be NULL or a string "tcp://somehostname:port".
     *
     * If the URI passed in is NULL then we allocate a port number
     * from our pool of port numbers and return a URI of
     * "tcp://ourhostname:port".
     *
     * If the URI passed in is not NULL then we try to parse out the
     * port number and use that (note that the hostname is assumed
     * to be a correct hostname which refers to the target machine).
     */
    if (uri_in == NULL) {
        this_port = QEMUD_MIGRATION_FIRST_PORT + port++;
        if (port == QEMUD_MIGRATION_NUM_PORTS) port = 0;

        /* Get hostname */
        if ((hostname = virGetHostname(NULL)) == NULL)
            goto cleanup;

        if (STRPREFIX(hostname, "localhost")) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("hostname on destination resolved to localhost, but migration requires an FQDN"));
            goto cleanup;
        }

        /* XXX this really should have been a properly well-formed
         * URI, but we can't add in tcp:// now without breaking
         * compatability with old targets. We at least make the
         * new targets accept both syntaxes though.
         */
        /* Caller frees */
        internalret = virAsprintf(uri_out, "tcp:%s:%d", hostname, this_port);
        if (internalret < 0) {
            virReportOOMError();
            goto cleanup;
        }
    } else {
        /* Check the URI starts with "tcp:".  We will escape the
         * URI when passing it to the qemu monitor, so bad
         * characters in hostname part don't matter.
         */
        if (!STRPREFIX (uri_in, "tcp:")) {
            qemuReportError (VIR_ERR_INVALID_ARG,
                             "%s", _("only tcp URIs are supported for KVM/QEMU migrations"));
            goto cleanup;
        }

        /* Get the port number. */
        p = strrchr (uri_in, ':');
        if (p == strchr(uri_in, ':')) {
            /* Generate a port */
            this_port = QEMUD_MIGRATION_FIRST_PORT + port++;
            if (port == QEMUD_MIGRATION_NUM_PORTS)
                port = 0;

            /* Caller frees */
            if (virAsprintf(uri_out, "%s:%d", uri_in, this_port) < 0) {
                virReportOOMError();
                goto cleanup;
            }

        } else {
            p++; /* definitely has a ':' in it, see above */
            this_port = virParseNumber (&p);
            if (this_port == -1 || p-uri_in != strlen (uri_in)) {
                qemuReportError(VIR_ERR_INVALID_ARG,
                                "%s", _("URI ended with incorrect ':port'"));
                goto cleanup;
            }
        }
    }

    if (*uri_out)
        VIR_DEBUG("Generated uri_out=%s", *uri_out);

    /* Parse the domain XML. */
    if (!(def = virDomainDefParseString(driver->caps, dom_xml,
                                        VIR_DOMAIN_XML_INACTIVE)))
        goto cleanup;

    if (!qemuMigrationIsAllowed(def))
        goto cleanup;

    /* Target domain name, maybe renamed. */
    if (dname) {
        VIR_FREE(def->name);
        def->name = strdup(dname);
        if (def->name == NULL)
            goto cleanup;
    }

    if (virDomainObjIsDuplicate(&driver->domains, def, 1) < 0)
        goto cleanup;

    if (!(vm = virDomainAssignDef(driver->caps,
                                  &driver->domains,
                                  def, true))) {
        /* virDomainAssignDef already set the error */
        goto cleanup;
    }
    def = NULL;
    priv = vm->privateData;

790 791 792
    if (!(mig = qemuMigrationEatCookie(vm, cookiein, cookieinlen, 0)))
        goto cleanup;

793 794 795 796 797 798 799 800 801 802 803 804 805
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;
    priv->jobActive = QEMU_JOB_MIGRATION_OUT;

    /* Domain starts inactive, even if the domain XML had an id field. */
    vm->def->id = -1;

    /* Start the QEMU daemon, with the same command-line arguments plus
     * -incoming tcp:0.0.0.0:port
     */
    snprintf (migrateFrom, sizeof (migrateFrom), "tcp:0.0.0.0:%d", this_port);
    if (qemuProcessStart(dconn, driver, vm, migrateFrom, true,
                         -1, NULL, VIR_VM_OP_MIGRATE_IN_START) < 0) {
806
        qemuAuditDomainStart(vm, "migrated", false);
807 808 809 810 811 812 813 814 815 816 817
        /* Note that we don't set an error here because qemuProcessStart
         * should have already done that.
         */
        if (!vm->persistent) {
            if (qemuDomainObjEndJob(vm) > 0)
                virDomainRemoveInactive(&driver->domains, vm);
            vm = NULL;
        }
        goto endjob;
    }

818 819 820 821 822 823 824 825
    if (qemuMigrationBakeCookie(mig, driver, vm, cookieout, cookieoutlen, 0) < 0) {
        /* We could tear down the whole guest here, but
         * cookie data is (so far) non-critical, so that
         * seems a little harsh. We'll just warn for now.
         */
        VIR_WARN("Unable to encode migration cookie");
    }

826
    qemuAuditDomainStart(vm, "migrated", true);
827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_MIGRATED);
    ret = 0;

endjob:
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;

    /* We set a fake job active which is held across
     * API calls until the finish() call. This prevents
     * any other APIs being invoked while incoming
     * migration is taking place
     */
    if (vm &&
        virDomainObjIsActive(vm)) {
        priv->jobActive = QEMU_JOB_MIGRATION_IN;
        priv->jobInfo.type = VIR_DOMAIN_JOB_UNBOUNDED;
        priv->jobStart = timeval_to_ms(now);
    }

cleanup:
    VIR_FREE(hostname);
    virDomainDefFree(def);
    if (ret != 0)
        VIR_FREE(*uri_out);
    if (vm)
        virDomainObjUnlock(vm);
    if (event)
        qemuDomainEventQueue(driver, event);
858
    qemuMigrationCookieFree(mig);
859 860 861 862 863 864 865 866 867 868
    return ret;
}


/* Perform migration using QEMU's native TCP migrate support,
 * not encrypted obviously
 */
static int doNativeMigrate(struct qemud_driver *driver,
                           virDomainObjPtr vm,
                           const char *uri,
869 870 871 872
                           const char *cookiein,
                           int cookieinlen,
                           char **cookieout,
                           int *cookieoutlen,
873 874 875 876 877 878 879 880
                           unsigned int flags,
                           const char *dname ATTRIBUTE_UNUSED,
                           unsigned long resource)
{
    int ret = -1;
    xmlURIPtr uribits = NULL;
    qemuDomainObjPrivatePtr priv = vm->privateData;
    unsigned int background_flags = QEMU_MONITOR_MIGRATE_BACKGROUND;
881 882 883 884
    qemuMigrationCookiePtr mig = NULL;

    if (!(mig = qemuMigrationEatCookie(vm, cookiein, cookieinlen, 0)))
        goto cleanup;
885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927

    /* Issue the migrate command. */
    if (STRPREFIX(uri, "tcp:") && !STRPREFIX(uri, "tcp://")) {
        /* HACK: source host generates bogus URIs, so fix them up */
        char *tmpuri;
        if (virAsprintf(&tmpuri, "tcp://%s", uri + strlen("tcp:")) < 0) {
            virReportOOMError();
            goto cleanup;
        }
        uribits = xmlParseURI(tmpuri);
        VIR_FREE(tmpuri);
    } else {
        uribits = xmlParseURI(uri);
    }
    if (!uribits) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("cannot parse URI %s"), uri);
        goto cleanup;
    }

    qemuDomainObjEnterMonitorWithDriver(driver, vm);
    if (resource > 0 &&
        qemuMonitorSetMigrationSpeed(priv->mon, resource) < 0) {
        qemuDomainObjExitMonitorWithDriver(driver, vm);
        goto cleanup;
    }

    if (flags & VIR_MIGRATE_NON_SHARED_DISK)
        background_flags |= QEMU_MONITOR_MIGRATE_NON_SHARED_DISK;

    if (flags & VIR_MIGRATE_NON_SHARED_INC)
        background_flags |= QEMU_MONITOR_MIGRATE_NON_SHARED_INC;

    if (qemuMonitorMigrateToHost(priv->mon, background_flags, uribits->server,
                                 uribits->port) < 0) {
        qemuDomainObjExitMonitorWithDriver(driver, vm);
        goto cleanup;
    }
    qemuDomainObjExitMonitorWithDriver(driver, vm);

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

928 929 930
    if (qemuMigrationBakeCookie(mig, driver, vm, cookieout, cookieoutlen, 0) < 0)
        VIR_WARN("Unable to encode migration cookie");

931 932 933
    ret = 0;

cleanup:
934
    qemuMigrationCookieFree(mig);
935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 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 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064
    xmlFreeURI(uribits);
    return ret;
}


#define TUNNEL_SEND_BUF_SIZE 65536

static int doTunnelSendAll(virStreamPtr st,
                           int sock)
{
    char *buffer;
    int nbytes = TUNNEL_SEND_BUF_SIZE;

    if (VIR_ALLOC_N(buffer, TUNNEL_SEND_BUF_SIZE) < 0) {
        virReportOOMError();
        virStreamAbort(st);
        return -1;
    }

    /* XXX should honour the 'resource' parameter here */
    for (;;) {
        nbytes = saferead(sock, buffer, nbytes);
        if (nbytes < 0) {
            virReportSystemError(errno, "%s",
                                 _("tunnelled migration failed to read from qemu"));
            virStreamAbort(st);
            VIR_FREE(buffer);
            return -1;
        }
        else if (nbytes == 0)
            /* EOF; get out of here */
            break;

        if (virStreamSend(st, buffer, nbytes) < 0) {
            qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
                            _("Failed to write migration data to remote libvirtd"));
            VIR_FREE(buffer);
            return -1;
        }
    }

    VIR_FREE(buffer);

    if (virStreamFinish(st) < 0)
        /* virStreamFinish set the error for us */
        return -1;

    return 0;
}

static int doTunnelMigrate(struct qemud_driver *driver,
                           virConnectPtr dconn,
                           virDomainObjPtr vm,
                           const char *dom_xml,
                           const char *uri,
                           unsigned long flags,
                           const char *dname,
                           unsigned long resource)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int client_sock = -1;
    int qemu_sock = -1;
    struct sockaddr_un sa_qemu, sa_client;
    socklen_t addrlen;
    virDomainPtr ddomain = NULL;
    int retval = -1;
    virStreamPtr st = NULL;
    char *unixfile = NULL;
    int internalret;
    int status;
    unsigned long long transferred, remaining, total;
    unsigned int background_flags = QEMU_MONITOR_MIGRATE_BACKGROUND;

    /*
     * The order of operations is important here to avoid touching
     * the source VM until we are very sure we can successfully
     * start the migration operation.
     *
     *   1. setup local support infrastructure (eg sockets)
     *   2. setup destination fully
     *   3. start migration on source
     */


    /* Stage 1. setup local support infrastructure */

    if (virAsprintf(&unixfile, "%s/qemu.tunnelmigrate.src.%s",
                    driver->libDir, vm->def->name) < 0) {
        virReportOOMError();
        goto cleanup;
    }

    qemu_sock = socket(AF_UNIX, SOCK_STREAM, 0);
    if (qemu_sock < 0) {
        virReportSystemError(errno, "%s",
                             _("cannot open tunnelled migration socket"));
        goto cleanup;
    }
    memset(&sa_qemu, 0, sizeof(sa_qemu));
    sa_qemu.sun_family = AF_UNIX;
    if (virStrcpy(sa_qemu.sun_path, unixfile,
                  sizeof(sa_qemu.sun_path)) == NULL) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Unix socket '%s' too big for destination"),
                        unixfile);
        goto cleanup;
    }
    unlink(unixfile);
    if (bind(qemu_sock, (struct sockaddr *)&sa_qemu, sizeof(sa_qemu)) < 0) {
        virReportSystemError(errno,
                             _("Cannot bind to unix socket '%s' for tunnelled migration"),
                             unixfile);
        goto cleanup;
    }
    if (listen(qemu_sock, 1) < 0) {
        virReportSystemError(errno,
                             _("Cannot listen on unix socket '%s' for tunnelled migration"),
                             unixfile);
        goto cleanup;
    }

    if (chown(unixfile, driver->user, driver->group) < 0) {
        virReportSystemError(errno,
                             _("Cannot change unix socket '%s' owner"),
                             unixfile);
        goto cleanup;
    }

    /* check that this qemu version supports the unix migration */

1065 1066
    if (!qemuCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_UNIX) &&
        !qemuCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_EXEC)) {
1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("Source qemu is too old to support tunnelled migration"));
        goto cleanup;
    }


    /* Stage 2. setup destination fully
     *
     * Once stage 2 has completed successfully, we *must* call finish
     * to cleanup the target whether we succeed or fail
     */
    st = virStreamNew(dconn, 0);
    if (st == NULL)
        /* virStreamNew only fails on OOM, and it reports the error itself */
        goto cleanup;

    qemuDomainObjEnterRemoteWithDriver(driver, vm);
    internalret = dconn->driver->domainMigratePrepareTunnel(dconn, st,
                                                            flags, dname,
                                                            resource, dom_xml);
    qemuDomainObjExitRemoteWithDriver(driver, vm);

    if (internalret < 0)
        /* domainMigratePrepareTunnel sets the error for us */
        goto cleanup;

    /* the domain may have shutdown or crashed while we had the locks dropped
     * in qemuDomainObjEnterRemoteWithDriver, so check again
     */
    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("guest unexpectedly quit"));
        goto cleanup;
    }

    /*   3. start migration on source */
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
    if (flags & VIR_MIGRATE_NON_SHARED_DISK)
        background_flags |= QEMU_MONITOR_MIGRATE_NON_SHARED_DISK;
    if (flags & VIR_MIGRATE_NON_SHARED_INC)
        background_flags |= QEMU_MONITOR_MIGRATE_NON_SHARED_INC;
1108
    if (qemuCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_UNIX)) {
1109 1110 1111
        internalret = qemuMonitorMigrateToUnix(priv->mon, background_flags,
                                               unixfile);
    }
1112
    else if (qemuCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_EXEC)) {
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 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212
        const char *args[] = { "nc", "-U", unixfile, NULL };
        internalret = qemuMonitorMigrateToCommand(priv->mon, QEMU_MONITOR_MIGRATE_BACKGROUND, args);
    } else {
        internalret = -1;
    }
    qemuDomainObjExitMonitorWithDriver(driver, vm);
    if (internalret < 0) {
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("tunnelled migration monitor command failed"));
        goto finish;
    }

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

    /* From this point onwards we *must* call cancel to abort the
     * migration on source if anything goes wrong */

    /* it is also possible that the migrate didn't fail initially, but
     * rather failed later on.  Check the output of "info migrate"
     */
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
    if (qemuMonitorGetMigrationStatus(priv->mon,
                                      &status,
                                      &transferred,
                                      &remaining,
                                      &total) < 0) {
        qemuDomainObjExitMonitorWithDriver(driver, vm);
        goto cancel;
    }
    qemuDomainObjExitMonitorWithDriver(driver, vm);

    if (status == QEMU_MONITOR_MIGRATION_STATUS_ERROR) {
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s",_("migrate failed"));
        goto cancel;
    }

    addrlen = sizeof(sa_client);
    while ((client_sock = accept(qemu_sock, (struct sockaddr *)&sa_client, &addrlen)) < 0) {
        if (errno == EAGAIN || errno == EINTR)
            continue;
        virReportSystemError(errno, "%s",
                             _("tunnelled migration failed to accept from qemu"));
        goto cancel;
    }

    retval = doTunnelSendAll(st, client_sock);

cancel:
    if (retval != 0 && virDomainObjIsActive(vm)) {
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
        qemuMonitorMigrateCancel(priv->mon);
        qemuDomainObjExitMonitorWithDriver(driver, vm);
    }

finish:
    dname = dname ? dname : vm->def->name;
    qemuDomainObjEnterRemoteWithDriver(driver, vm);
    ddomain = dconn->driver->domainMigrateFinish2
        (dconn, dname, NULL, 0, uri, flags, retval);
    qemuDomainObjExitRemoteWithDriver(driver, vm);

cleanup:
    VIR_FORCE_CLOSE(client_sock);
    VIR_FORCE_CLOSE(qemu_sock);

    if (ddomain)
        virUnrefDomain(ddomain);

    if (unixfile) {
        unlink(unixfile);
        VIR_FREE(unixfile);
    }

    if (st)
        /* don't call virStreamFree(), because that resets any pending errors */
        virUnrefStream(st);
    return retval;
}


/* This is essentially a simplified re-impl of
 * virDomainMigrateVersion2 from libvirt.c, but running in source
 * libvirtd context, instead of client app context */
static int doNonTunnelMigrate(struct qemud_driver *driver,
                              virConnectPtr dconn,
                              virDomainObjPtr vm,
                              const char *dom_xml,
                              const char *uri ATTRIBUTE_UNUSED,
                              unsigned long flags,
                              const char *dname,
                              unsigned long resource)
{
    virDomainPtr ddomain = NULL;
    int retval = -1;
    char *uri_out = NULL;
1213 1214
    char *cookie = NULL;
    int cookielen = 0;
1215 1216 1217 1218 1219 1220
    int rc;

    qemuDomainObjEnterRemoteWithDriver(driver, vm);
    /* NB we don't pass 'uri' into this, since that's the libvirtd
     * URI in this context - so we let dest pick it */
    rc = dconn->driver->domainMigratePrepare2(dconn,
1221 1222
                                              &cookie,
                                              &cookielen,
1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246
                                              NULL, /* uri */
                                              &uri_out,
                                              flags, dname,
                                              resource, dom_xml);
    qemuDomainObjExitRemoteWithDriver(driver, vm);
    if (rc < 0)
        /* domainMigratePrepare2 sets the error for us */
        goto cleanup;

    /* the domain may have shutdown or crashed while we had the locks dropped
     * in qemuDomainObjEnterRemoteWithDriver, so check again
     */
    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("guest unexpectedly quit"));
        goto cleanup;
    }

    if (uri_out == NULL) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("domainMigratePrepare2 did not set uri"));
        goto cleanup;
    }

1247 1248 1249 1250
    if (doNativeMigrate(driver, vm, uri_out,
                        cookie, cookielen,
                        NULL, NULL, /* No out cookie with v2 migration */
                        flags, dname, resource) < 0)
1251 1252 1253 1254 1255 1256 1257 1258
        goto finish;

    retval = 0;

finish:
    dname = dname ? dname : vm->def->name;
    qemuDomainObjEnterRemoteWithDriver(driver, vm);
    ddomain = dconn->driver->domainMigrateFinish2
1259
        (dconn, dname, cookie, cookielen, uri_out, flags, retval);
1260 1261 1262 1263 1264 1265
    qemuDomainObjExitRemoteWithDriver(driver, vm);

    if (ddomain)
        virUnrefDomain(ddomain);

cleanup:
1266
    VIR_FREE(cookie);
1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341
    return retval;
}


static int doPeer2PeerMigrate(struct qemud_driver *driver,
                              virDomainObjPtr vm,
                              const char *uri,
                              unsigned long flags,
                              const char *dname,
                              unsigned long resource)
{
    int ret = -1;
    virConnectPtr dconn = NULL;
    char *dom_xml;
    bool p2p;

    /* the order of operations is important here; we make sure the
     * destination side is completely setup before we touch the source
     */

    qemuDomainObjEnterRemoteWithDriver(driver, vm);
    dconn = virConnectOpen(uri);
    qemuDomainObjExitRemoteWithDriver(driver, vm);
    if (dconn == NULL) {
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("Failed to connect to remote libvirt URI %s"), uri);
        return -1;
    }

    qemuDomainObjEnterRemoteWithDriver(driver, vm);
    p2p = VIR_DRV_SUPPORTS_FEATURE(dconn->driver, dconn,
                                   VIR_DRV_FEATURE_MIGRATION_P2P);
    qemuDomainObjExitRemoteWithDriver(driver, vm);
    if (!p2p) {
        qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
                        _("Destination libvirt does not support peer-to-peer migration protocol"));
        goto cleanup;
    }

    /* domain may have been stopped while we were talking to remote daemon */
    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("guest unexpectedly quit"));
        goto cleanup;
    }

    dom_xml = qemuDomainFormatXML(driver, vm,
                                  VIR_DOMAIN_XML_SECURE |
                                  VIR_DOMAIN_XML_UPDATE_CPU);
    if (!dom_xml) {
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to get domain xml"));
        goto cleanup;
    }

    if (flags & VIR_MIGRATE_TUNNELLED)
        ret = doTunnelMigrate(driver, dconn, vm, dom_xml, uri, flags, dname, resource);
    else
        ret = doNonTunnelMigrate(driver, dconn, vm, dom_xml, uri, flags, dname, resource);

cleanup:
    VIR_FREE(dom_xml);
    /* don't call virConnectClose(), because that resets any pending errors */
    qemuDomainObjEnterRemoteWithDriver(driver, vm);
    virUnrefConnect(dconn);
    qemuDomainObjExitRemoteWithDriver(driver, vm);

    return ret;
}


int qemuMigrationPerform(struct qemud_driver *driver,
                         virConnectPtr conn,
                         virDomainObjPtr vm,
                         const char *uri,
1342 1343 1344 1345
                         const char *cookiein,
                         int cookieinlen,
                         char **cookieout,
                         int *cookieoutlen,
1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367
                         unsigned long flags,
                         const char *dname,
                         unsigned long resource)
{
    virDomainEventPtr event = NULL;
    int ret = -1;
    int resume = 0;
    qemuDomainObjPrivatePtr priv = vm->privateData;

    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;
    priv->jobActive = QEMU_JOB_MIGRATION_OUT;

    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
        goto endjob;
    }

    memset(&priv->jobInfo, 0, sizeof(priv->jobInfo));
    priv->jobInfo.type = VIR_DOMAIN_JOB_UNBOUNDED;

J
Jiri Denemark 已提交
1368 1369 1370
    resume = virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING;
    if (!(flags & VIR_MIGRATE_LIVE) &&
        virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING) {
1371 1372 1373 1374 1375
        if (qemuMigrationSetOffline(driver, vm) < 0)
            goto endjob;
    }

    if ((flags & (VIR_MIGRATE_TUNNELLED | VIR_MIGRATE_PEER2PEER))) {
1376 1377 1378 1379 1380 1381
        if (cookieinlen) {
            qemuReportError(VIR_ERR_OPERATION_INVALID,
                            "%s", _("received unexpected cookie with P2P migration"));
            goto endjob;
        }

1382 1383 1384 1385
        if (doPeer2PeerMigrate(driver, vm, uri, flags, dname, resource) < 0)
            /* doPeer2PeerMigrate already set the error, so just get out */
            goto endjob;
    } else {
1386 1387 1388
        if (doNativeMigrate(driver, vm, uri, cookiein, cookieinlen,
                            cookieout, cookieoutlen,
                            flags, dname, resource) < 0)
1389 1390 1391 1392
            goto endjob;
    }

    /* Clean up the source domain. */
J
Jiri Denemark 已提交
1393
    qemuProcessStop(driver, vm, 1, VIR_DOMAIN_SHUTOFF_MIGRATED);
1394
    qemuAuditDomainStop(vm, "migrated");
1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405
    resume = 0;

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_MIGRATED);
    if (!vm->persistent || (flags & VIR_MIGRATE_UNDEFINE_SOURCE)) {
        virDomainDeleteConfig(driver->configDir, driver->autostartDir, vm);
        if (qemuDomainObjEndJob(vm) > 0)
            virDomainRemoveInactive(&driver->domains, vm);
        vm = NULL;
    }
1406

1407 1408 1409
    ret = 0;

endjob:
J
Jiri Denemark 已提交
1410
    if (resume && virDomainObjGetState(vm, NULL) == VIR_DOMAIN_PAUSED) {
1411
        /* we got here through some sort of failure; start the domain again */
J
Jiri Denemark 已提交
1412 1413
        if (qemuProcessStartCPUs(driver, vm, conn,
                                 VIR_DOMAIN_RUNNING_MIGRATION_CANCELED) < 0) {
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 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
            /* Hm, we already know we are in error here.  We don't want to
             * overwrite the previous error, though, so we just throw something
             * to the logs and hope for the best
             */
            VIR_ERROR(_("Failed to resume guest %s after failure"),
                      vm->def->name);
        }

        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_MIGRATED);
    }
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
    if (event)
        qemuDomainEventQueue(driver, event);
    return ret;
}


#if WITH_MACVTAP
static void
qemuMigrationVPAssociatePortProfiles(virDomainDefPtr def) {
    int i;
    int last_good_net = -1;
    virDomainNetDefPtr net;

    for (i = 0; i < def->nnets; i++) {
        net = def->nets[i];
        if (net->type == VIR_DOMAIN_NET_TYPE_DIRECT) {
            if (vpAssociatePortProfileId(net->ifname,
                                         net->mac,
                                         net->data.direct.linkdev,
                                         &net->data.direct.virtPortProfile,
                                         def->uuid,
                                         VIR_VM_OP_MIGRATE_IN_FINISH) != 0)
                goto err_exit;
        }
        last_good_net = i;
    }

    return;

err_exit:
    for (i = 0; i < last_good_net; i++) {
        net = def->nets[i];
        if (net->type == VIR_DOMAIN_NET_TYPE_DIRECT) {
            vpDisassociatePortProfileId(net->ifname,
                                        net->mac,
                                        net->data.direct.linkdev,
                                        &net->data.direct.virtPortProfile,
                                        VIR_VM_OP_MIGRATE_IN_FINISH);
        }
    }
}
#else /* !WITH_MACVTAP */
static void
qemuMigrationVPAssociatePortProfiles(virDomainDefPtr def ATTRIBUTE_UNUSED) { }
#endif /* WITH_MACVTAP */


virDomainPtr
qemuMigrationFinish(struct qemud_driver *driver,
                    virConnectPtr dconn,
                    virDomainObjPtr vm,
1484 1485 1486 1487
                    const char *cookiein,
                    int cookieinlen,
                    char **cookieout,
                    int *cookieoutlen,
1488 1489 1490 1491 1492 1493 1494
                    unsigned long flags,
                    int retcode)
{
    virDomainPtr dom = NULL;
    virDomainEventPtr event = NULL;
    int newVM = 1;
    qemuDomainObjPrivatePtr priv = NULL;
1495
    qemuMigrationCookiePtr mig = NULL;
1496 1497 1498 1499 1500 1501 1502 1503 1504 1505

    priv = vm->privateData;
    if (priv->jobActive != QEMU_JOB_MIGRATION_IN) {
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("domain '%s' is not processing incoming migration"), vm->def->name);
        goto cleanup;
    }
    priv->jobActive = QEMU_JOB_NONE;
    memset(&priv->jobInfo, 0, sizeof(priv->jobInfo));

1506 1507 1508
    if (!(mig = qemuMigrationEatCookie(vm, cookiein, cookieinlen, 0)))
        goto cleanup;

1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

    /* Did the migration go as planned?  If yes, return the domain
     * object, but if no, clean up the empty qemu process.
     */
    if (retcode == 0) {
        if (!virDomainObjIsActive(vm)) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("guest unexpectedly quit"));
            goto cleanup;
        }

        qemuMigrationVPAssociatePortProfiles(vm->def);

        if (flags & VIR_MIGRATE_PERSIST_DEST) {
            if (vm->persistent)
                newVM = 0;
            vm->persistent = 1;

            if (virDomainSaveConfig(driver->configDir, vm->def) < 0) {
                /* Hmpf.  Migration was successful, but making it persistent
                 * was not.  If we report successful, then when this domain
                 * shuts down, management tools are in for a surprise.  On the
                 * other hand, if we report failure, then the management tools
                 * might try to restart the domain on the source side, even
                 * though the domain is actually running on the destination.
                 * Return a NULL dom pointer, and hope that this is a rare
                 * situation and management tools are smart.
                 */
                vm = NULL;
                goto endjob;
            }

            event = virDomainEventNewFromObj(vm,
                                             VIR_DOMAIN_EVENT_DEFINED,
                                             newVM ?
                                             VIR_DOMAIN_EVENT_DEFINED_ADDED :
                                             VIR_DOMAIN_EVENT_DEFINED_UPDATED);
            if (event)
                qemuDomainEventQueue(driver, event);
            event = NULL;

        }
        dom = virGetDomain (dconn, vm->def->name, vm->def->uuid);

        if (!(flags & VIR_MIGRATE_PAUSED)) {
            /* run 'cont' on the destination, which allows migration on qemu
             * >= 0.10.6 to work properly.  This isn't strictly necessary on
             * older qemu's, but it also doesn't hurt anything there
             */
J
Jiri Denemark 已提交
1560 1561
            if (qemuProcessStartCPUs(driver, vm, dconn,
                                     VIR_DOMAIN_RUNNING_MIGRATED) < 0) {
1562 1563 1564 1565 1566 1567 1568 1569 1570 1571
                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_MIGRATED);
J
Jiri Denemark 已提交
1572 1573
        if (virDomainObjGetState(vm, NULL) == VIR_DOMAIN_PAUSED) {
            virDomainObjSetState(vm, VIR_DOMAIN_PAUSED, VIR_DOMAIN_PAUSED_USER);
1574 1575 1576 1577 1578 1579 1580 1581 1582 1583
            qemuDomainEventQueue(driver, event);
            event = virDomainEventNewFromObj(vm,
                                             VIR_DOMAIN_EVENT_SUSPENDED,
                                             VIR_DOMAIN_EVENT_SUSPENDED_PAUSED);
        }
        if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0) {
            VIR_WARN("Failed to save status on vm %s", vm->def->name);
            goto endjob;
        }
    } else {
J
Jiri Denemark 已提交
1584
        qemuProcessStop(driver, vm, 1, VIR_DOMAIN_SHUTOFF_FAILED);
1585
        qemuAuditDomainStop(vm, "failed");
1586 1587 1588 1589 1590 1591 1592 1593 1594 1595
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         VIR_DOMAIN_EVENT_STOPPED_FAILED);
        if (!vm->persistent) {
            if (qemuDomainObjEndJob(vm) > 0)
                virDomainRemoveInactive(&driver->domains, vm);
            vm = NULL;
        }
    }

1596 1597 1598
    if (qemuMigrationBakeCookie(mig, driver, vm, cookieout, cookieoutlen, 0) < 0)
        VIR_WARN("Unable to encode migration cookie");

1599 1600 1601 1602 1603 1604 1605 1606 1607 1608
endjob:
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
    if (event)
        qemuDomainEventQueue(driver, event);
1609
    qemuMigrationCookieFree(mig);
1610 1611
    return dom;
}
1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624

/* Helper function called while driver lock is held and vm is active.  */
int
qemuMigrationToFile(struct qemud_driver *driver, virDomainObjPtr vm,
                    int fd, off_t offset, const char *path,
                    const char *compressor,
                    bool is_reg, bool bypassSecurityDriver)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virCgroupPtr cgroup = NULL;
    int ret = -1;
    int rc;
    bool restoreLabel = false;
1625 1626
    virCommandPtr cmd = NULL;
    int pipeFD[2] = { -1, -1 };
1627

1628
    if (qemuCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_FD) &&
1629
        (!compressor || pipe(pipeFD) == 0)) {
1630
        /* All right! We can use fd migration, which means that qemu
1631 1632 1633 1634 1635 1636
         * doesn't have to open() the file, so while we still have to
         * grant SELinux access, we can do it on fd and avoid cleanup
         * later, as well as skip futzing with cgroup.  */
        if (virSecurityManagerSetFDLabel(driver->securityManager, vm,
                                         compressor ? pipeFD[1] : fd) < 0)
            goto cleanup;
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
        bypassSecurityDriver = true;
    } else {
        /* Phooey - we have to fall back on exec migration, where qemu
         * has to popen() the file by name.  We might also stumble on
         * a race present in some qemu versions where it does a wait()
         * that botches pclose.  */
        if (!is_reg &&
            qemuCgroupControllerActive(driver,
                                       VIR_CGROUP_CONTROLLER_DEVICES)) {
            if (virCgroupForDomain(driver->cgroup, vm->def->name,
                                   &cgroup, 0) != 0) {
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("Unable to find cgroup for %s"),
                                vm->def->name);
                goto cleanup;
            }
            rc = virCgroupAllowDevicePath(cgroup, path,
                                          VIR_CGROUP_DEVICE_RW);
            qemuAuditCgroupPath(vm, cgroup, "allow", path, "rw", rc);
            if (rc < 0) {
                virReportSystemError(-rc,
                                     _("Unable to allow device %s for %s"),
                                     path, vm->def->name);
                goto cleanup;
            }
1662
        }
1663 1664 1665
        if ((!bypassSecurityDriver) &&
            virSecurityManagerSetSavedStateLabel(driver->securityManager,
                                                 vm, path) < 0)
1666
            goto cleanup;
1667
        restoreLabel = true;
1668 1669
    }

1670
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
1671 1672 1673
    if (!compressor) {
        const char *args[] = { "cat", NULL };

1674
        if (qemuCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_FD) &&
1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690
            priv->monConfig->type == VIR_DOMAIN_CHR_TYPE_UNIX) {
            rc = qemuMonitorMigrateToFd(priv->mon,
                                        QEMU_MONITOR_MIGRATE_BACKGROUND,
                                        fd);
        } else {
            rc = qemuMonitorMigrateToFile(priv->mon,
                                          QEMU_MONITOR_MIGRATE_BACKGROUND,
                                          args, path, offset);
        }
    } else {
        const char *prog = compressor;
        const char *args[] = {
            prog,
            "-c",
            NULL
        };
1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709
        if (pipeFD[0] != -1) {
            cmd = virCommandNewArgs(args);
            virCommandSetInputFD(cmd, pipeFD[0]);
            virCommandSetOutputFD(cmd, &fd);
            if (virSetCloseExec(pipeFD[1]) < 0) {
                virReportSystemError(errno, "%s",
                                     _("Unable to set cloexec flag"));
                qemuDomainObjExitMonitorWithDriver(driver, vm);
                goto cleanup;
            }
            if (virCommandRunAsync(cmd, NULL) < 0) {
                qemuDomainObjExitMonitorWithDriver(driver, vm);
                goto cleanup;
            }
            rc = qemuMonitorMigrateToFd(priv->mon,
                                        QEMU_MONITOR_MIGRATE_BACKGROUND,
                                        pipeFD[1]);
            if (VIR_CLOSE(pipeFD[0]) < 0 ||
                VIR_CLOSE(pipeFD[1]) < 0)
1710
                VIR_WARN("failed to close intermediate pipe");
1711 1712 1713 1714 1715
        } else {
            rc = qemuMonitorMigrateToFile(priv->mon,
                                          QEMU_MONITOR_MIGRATE_BACKGROUND,
                                          args, path, offset);
        }
1716
    }
1717
    qemuDomainObjExitMonitorWithDriver(driver, vm);
1718 1719 1720 1721 1722 1723 1724 1725 1726

    if (rc < 0)
        goto cleanup;

    rc = qemuMigrationWaitForCompletion(driver, vm);

    if (rc < 0)
        goto cleanup;

1727 1728 1729
    if (cmd && virCommandWait(cmd, NULL) < 0)
        goto cleanup;

1730 1731 1732
    ret = 0;

cleanup:
1733 1734 1735
    VIR_FORCE_CLOSE(pipeFD[0]);
    VIR_FORCE_CLOSE(pipeFD[1]);
    virCommandFree(cmd);
1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751
    if (restoreLabel && (!bypassSecurityDriver) &&
        virSecurityManagerRestoreSavedStateLabel(driver->securityManager,
                                                 vm, path) < 0)
        VIR_WARN("failed to restore save state label on %s", path);

    if (cgroup != NULL) {
        rc = virCgroupDenyDevicePath(cgroup, path,
                                     VIR_CGROUP_DEVICE_RWM);
        qemuAuditCgroupPath(vm, cgroup, "deny", path, "rwm", rc);
        if (rc < 0)
            VIR_WARN("Unable to deny device %s for %s %d",
                     path, vm->def->name, rc);
        virCgroupFree(&cgroup);
    }
    return ret;
}