qemu_migration.c 142.3 KB
Newer Older
1

2 3 4
/*
 * qemu_migration.c: QEMU migration handling
 *
5
 * Copyright (C) 2006-2013 Red Hat, Inc.
6 7 8 9 10 11 12 13 14 15 16 17
 *
 * 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
18
 * License along with this library.  If not, see
O
Osier Yang 已提交
19
 * <http://www.gnu.org/licenses/>.
20 21 22 23 24 25
 *
 */

#include <config.h>

#include <sys/time.h>
26
#ifdef WITH_GNUTLS
27 28 29
# include <gnutls/gnutls.h>
# include <gnutls/x509.h>
#endif
30
#include <fcntl.h>
31
#include <poll.h>
32 33 34 35 36 37

#include "qemu_migration.h"
#include "qemu_monitor.h"
#include "qemu_domain.h"
#include "qemu_process.h"
#include "qemu_capabilities.h"
38
#include "qemu_command.h"
39
#include "qemu_cgroup.h"
40

41
#include "domain_audit.h"
42
#include "virlog.h"
43
#include "virerror.h"
44
#include "viralloc.h"
45
#include "virutil.h"
E
Eric Blake 已提交
46
#include "virfile.h"
47 48
#include "datatypes.h"
#include "fdstream.h"
49
#include "viruuid.h"
50
#include "virtime.h"
51
#include "locking/domain_lock.h"
52
#include "rpc/virnetsocket.h"
53
#include "virstoragefile.h"
M
Martin Kletzander 已提交
54
#include "viruri.h"
55
#include "virhook.h"
56

57 58 59

#define VIR_FROM_THIS VIR_FROM_QEMU

60 61 62 63 64 65 66 67 68 69 70 71 72
VIR_ENUM_IMPL(qemuMigrationJobPhase, QEMU_MIGRATION_PHASE_LAST,
              "none",
              "perform2",
              "begin3",
              "perform3",
              "perform3_done",
              "confirm3_cancelled",
              "confirm3",
              "prepare",
              "finish2",
              "finish3",
);

73
enum qemuMigrationCookieFlags {
74
    QEMU_MIGRATION_COOKIE_FLAG_GRAPHICS,
75
    QEMU_MIGRATION_COOKIE_FLAG_LOCKSTATE,
76
    QEMU_MIGRATION_COOKIE_FLAG_PERSISTENT,
77
    QEMU_MIGRATION_COOKIE_FLAG_NETWORK,
78
    QEMU_MIGRATION_COOKIE_FLAG_NBD,
79 80 81 82 83 84 85

    QEMU_MIGRATION_COOKIE_FLAG_LAST
};

VIR_ENUM_DECL(qemuMigrationCookieFlag);
VIR_ENUM_IMPL(qemuMigrationCookieFlag,
              QEMU_MIGRATION_COOKIE_FLAG_LAST,
86 87 88 89 90
              "graphics",
              "lockstate",
              "persistent",
              "network",
              "nbd");
91 92 93

enum qemuMigrationCookieFeatures {
    QEMU_MIGRATION_COOKIE_GRAPHICS  = (1 << QEMU_MIGRATION_COOKIE_FLAG_GRAPHICS),
94
    QEMU_MIGRATION_COOKIE_LOCKSTATE = (1 << QEMU_MIGRATION_COOKIE_FLAG_LOCKSTATE),
95
    QEMU_MIGRATION_COOKIE_PERSISTENT = (1 << QEMU_MIGRATION_COOKIE_FLAG_PERSISTENT),
96
    QEMU_MIGRATION_COOKIE_NETWORK = (1 << QEMU_MIGRATION_COOKIE_FLAG_NETWORK),
97
    QEMU_MIGRATION_COOKIE_NBD = (1 << QEMU_MIGRATION_COOKIE_FLAG_NBD),
98 99 100 101 102 103 104 105 106 107 108 109
};

typedef struct _qemuMigrationCookieGraphics qemuMigrationCookieGraphics;
typedef qemuMigrationCookieGraphics *qemuMigrationCookieGraphicsPtr;
struct _qemuMigrationCookieGraphics {
    int type;
    int port;
    int tlsPort;
    char *listen;
    char *tlsSubject;
};

110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130
typedef struct _qemuMigrationCookieNetData qemuMigrationCookieNetData;
typedef qemuMigrationCookieNetData *qemuMigrationCookieNetDataPtr;
struct _qemuMigrationCookieNetData {
    int vporttype; /* enum virNetDevVPortProfile */

    /*
     * Array of pointers to saved data. Each VIF will have it's own
     * data to transfer.
     */
    char *portdata;
};

typedef struct _qemuMigrationCookieNetwork qemuMigrationCookieNetwork;
typedef qemuMigrationCookieNetwork *qemuMigrationCookieNetworkPtr;
struct _qemuMigrationCookieNetwork {
    /* How many virtual NICs are we saving data for? */
    int nnets;

    qemuMigrationCookieNetDataPtr net;
};

131 132 133 134 135 136
typedef struct _qemuMigrationCookieNBD qemuMigrationCookieNBD;
typedef qemuMigrationCookieNBD *qemuMigrationCookieNBDPtr;
struct _qemuMigrationCookieNBD {
    int port; /* on which port does NBD server listen for incoming data */
};

137 138 139
typedef struct _qemuMigrationCookie qemuMigrationCookie;
typedef qemuMigrationCookie *qemuMigrationCookiePtr;
struct _qemuMigrationCookie {
E
Eric Blake 已提交
140 141
    unsigned int flags;
    unsigned int flagsMandatory;
142 143

    /* Host properties */
144 145 146 147
    unsigned char localHostuuid[VIR_UUID_BUFLEN];
    unsigned char remoteHostuuid[VIR_UUID_BUFLEN];
    char *localHostname;
    char *remoteHostname;
148 149 150 151

    /* Guest properties */
    unsigned char uuid[VIR_UUID_BUFLEN];
    char *name;
152

153 154 155 156
    /* If (flags & QEMU_MIGRATION_COOKIE_LOCKSTATE) */
    char *lockState;
    char *lockDriver;

157 158
    /* If (flags & QEMU_MIGRATION_COOKIE_GRAPHICS) */
    qemuMigrationCookieGraphicsPtr graphics;
159 160 161

    /* If (flags & QEMU_MIGRATION_COOKIE_PERSISTENT) */
    virDomainDefPtr persistent;
162 163 164

    /* If (flags & QEMU_MIGRATION_COOKIE_NETWORK) */
    qemuMigrationCookieNetworkPtr network;
165 166 167

    /* If (flags & QEMU_MIGRATION_COOKIE_NBD) */
    qemuMigrationCookieNBDPtr nbd;
168 169
};

170 171 172 173 174 175 176 177 178
static void qemuMigrationCookieGraphicsFree(qemuMigrationCookieGraphicsPtr grap)
{
    if (!grap)
        return;
    VIR_FREE(grap->listen);
    VIR_FREE(grap->tlsSubject);
    VIR_FREE(grap);
}

179

180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196
static void
qemuMigrationCookieNetworkFree(qemuMigrationCookieNetworkPtr network)
{
    int i;

    if (!network)
        return;

    if (network->net) {
        for (i = 0; i < network->nnets; i++)
            VIR_FREE(network->net[i].portdata);
    }
    VIR_FREE(network->net);
    VIR_FREE(network);
}


197 198 199 200 201
static void qemuMigrationCookieFree(qemuMigrationCookiePtr mig)
{
    if (!mig)
        return;

202 203 204
    if (mig->flags & QEMU_MIGRATION_COOKIE_GRAPHICS)
        qemuMigrationCookieGraphicsFree(mig->graphics);

205 206 207
    if (mig->flags & QEMU_MIGRATION_COOKIE_NETWORK)
        qemuMigrationCookieNetworkFree(mig->network);

208 209
    VIR_FREE(mig->localHostname);
    VIR_FREE(mig->remoteHostname);
210
    VIR_FREE(mig->name);
211 212
    VIR_FREE(mig->lockState);
    VIR_FREE(mig->lockDriver);
213
    VIR_FREE(mig->nbd);
214 215 216 217
    VIR_FREE(mig);
}


218
#ifdef WITH_GNUTLS
219 220 221 222 223 224 225 226 227 228 229 230 231 232 233
static char *
qemuDomainExtractTLSSubject(const char *certdir)
{
    char *certfile = NULL;
    char *subject = NULL;
    char *pemdata = NULL;
    gnutls_datum_t pemdatum;
    gnutls_x509_crt_t cert;
    int ret;
    size_t subjectlen;

    if (virAsprintf(&certfile, "%s/server-cert.pem", certdir) < 0)
        goto no_memory;

    if (virFileReadAll(certfile, 8192, &pemdata) < 0) {
234 235
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unable to read server cert %s"), certfile);
236 237 238 239 240
        goto error;
    }

    ret = gnutls_x509_crt_init(&cert);
    if (ret < 0) {
241 242 243
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("cannot initialize cert object: %s"),
                       gnutls_strerror(ret));
244 245 246 247 248 249 250 251
        goto error;
    }

    pemdatum.data = (unsigned char *)pemdata;
    pemdatum.size = strlen(pemdata);

    ret = gnutls_x509_crt_import(cert, &pemdatum, GNUTLS_X509_FMT_PEM);
    if (ret < 0) {
252 253 254
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("cannot load cert data from %s: %s"),
                       certfile, gnutls_strerror(ret));
255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276
        goto error;
    }

    subjectlen = 1024;
    if (VIR_ALLOC_N(subject, subjectlen+1) < 0)
        goto no_memory;

    gnutls_x509_crt_get_dn(cert, subject, &subjectlen);
    subject[subjectlen] = '\0';

    VIR_FREE(certfile);
    VIR_FREE(pemdata);

    return subject;

no_memory:
    virReportOOMError();
error:
    VIR_FREE(certfile);
    VIR_FREE(pemdata);
    return NULL;
}
277
#endif
278 279

static qemuMigrationCookieGraphicsPtr
280
qemuMigrationCookieGraphicsAlloc(virQEMUDriverPtr driver,
281 282 283 284
                                 virDomainGraphicsDefPtr def)
{
    qemuMigrationCookieGraphicsPtr mig = NULL;
    const char *listenAddr;
285
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
286 287 288 289 290 291 292

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

    mig->type = def->type;
    if (mig->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC) {
        mig->port = def->data.vnc.port;
293
        listenAddr = virDomainGraphicsListenGetAddress(def, 0);
294
        if (!listenAddr)
295
            listenAddr = cfg->vncListen;
296

297
#ifdef WITH_GNUTLS
298 299
        if (cfg->vncTLS &&
            !(mig->tlsSubject = qemuDomainExtractTLSSubject(cfg->vncTLSx509certdir)))
300
            goto error;
301
#endif
302 303
    } else {
        mig->port = def->data.spice.port;
304
        if (cfg->spiceTLS)
305 306 307
            mig->tlsPort = def->data.spice.tlsPort;
        else
            mig->tlsPort = -1;
308
        listenAddr = virDomainGraphicsListenGetAddress(def, 0);
309
        if (!listenAddr)
310
            listenAddr = cfg->spiceListen;
311

312
#ifdef WITH_GNUTLS
313 314
        if (cfg->spiceTLS &&
            !(mig->tlsSubject = qemuDomainExtractTLSSubject(cfg->spiceTLSx509certdir)))
315
            goto error;
316
#endif
317 318 319 320
    }
    if (!(mig->listen = strdup(listenAddr)))
        goto no_memory;

321
    virObjectUnref(cfg);
322 323 324 325
    return mig;

no_memory:
    virReportOOMError();
326
#ifdef WITH_GNUTLS
327
error:
328
#endif
329
    qemuMigrationCookieGraphicsFree(mig);
330
    virObjectUnref(cfg);
331 332 333 334
    return NULL;
}


335
static qemuMigrationCookieNetworkPtr
336
qemuMigrationCookieNetworkAlloc(virQEMUDriverPtr driver ATTRIBUTE_UNUSED,
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
                                virDomainDefPtr def)
{
    qemuMigrationCookieNetworkPtr mig;
    int i;

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

    mig->nnets = def->nnets;

    if (VIR_ALLOC_N(mig->net, def->nnets) <0)
        goto no_memory;

    for (i = 0; i < def->nnets; i++) {
        virDomainNetDefPtr netptr;
        virNetDevVPortProfilePtr vport;

        netptr = def->nets[i];
        vport = virDomainNetGetActualVirtPortProfile(netptr);

        if (vport) {
            mig->net[i].vporttype = vport->virtPortType;

            switch (vport->virtPortType) {
            case VIR_NETDEV_VPORT_PROFILE_NONE:
            case VIR_NETDEV_VPORT_PROFILE_8021QBG:
            case VIR_NETDEV_VPORT_PROFILE_8021QBH:
364
               break;
365
            case VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH:
366 367 368 369 370 371 372 373
                if (virNetDevOpenvswitchGetMigrateData(&mig->net[i].portdata,
                                                       netptr->ifname) != 0) {
                        virReportSystemError(VIR_ERR_INTERNAL_ERROR,
                                             _("Unable to run command to get OVS port data for "
                                             "interface %s"), netptr->ifname);
                        goto error;
                }
                break;
374 375 376 377 378 379 380 381 382
            default:
                break;
            }
        }
    }
    return mig;

no_memory:
    virReportOOMError();
383
error:
384 385 386 387
    qemuMigrationCookieNetworkFree(mig);
    return NULL;
}

388 389 390
static qemuMigrationCookiePtr
qemuMigrationCookieNew(virDomainObjPtr dom)
{
J
Jiri Denemark 已提交
391
    qemuDomainObjPrivatePtr priv = dom->privateData;
392
    qemuMigrationCookiePtr mig = NULL;
J
Jiri Denemark 已提交
393
    const char *name;
394 395 396 397

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

J
Jiri Denemark 已提交
398 399 400 401 402
    if (priv->origname)
        name = priv->origname;
    else
        name = dom->def->name;
    if (!(mig->name = strdup(name)))
403 404 405
        goto no_memory;
    memcpy(mig->uuid, dom->def->uuid, VIR_UUID_BUFLEN);

406
    if (!(mig->localHostname = virGetHostname(NULL)))
407
        goto error;
408
    if (virGetHostUUID(mig->localHostuuid) < 0) {
409 410
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Unable to obtain host UUID"));
411 412 413 414 415 416 417 418 419 420 421 422 423
        goto error;
    }

    return mig;

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


424 425
static int
qemuMigrationCookieAddGraphics(qemuMigrationCookiePtr mig,
426
                               virQEMUDriverPtr driver,
427 428 429
                               virDomainObjPtr dom)
{
    if (mig->flags & QEMU_MIGRATION_COOKIE_GRAPHICS) {
430 431
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Migration graphics data already present"));
432 433 434 435 436
        return -1;
    }

    if (dom->def->ngraphics == 1 &&
        (dom->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC ||
437 438 439 440 441 442
         dom->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_SPICE)) {
        if (!(mig->graphics =
              qemuMigrationCookieGraphicsAlloc(driver, dom->def->graphics[0])))
            return -1;
        mig->flags |= QEMU_MIGRATION_COOKIE_GRAPHICS;
    }
443 444 445 446 447

    return 0;
}


448 449
static int
qemuMigrationCookieAddLockstate(qemuMigrationCookiePtr mig,
450
                                virQEMUDriverPtr driver,
451 452 453 454 455
                                virDomainObjPtr dom)
{
    qemuDomainObjPrivatePtr priv = dom->privateData;

    if (mig->flags & QEMU_MIGRATION_COOKIE_LOCKSTATE) {
456 457
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Migration lockstate data already present"));
458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481
        return -1;
    }

    if (virDomainObjGetState(dom, NULL) == VIR_DOMAIN_PAUSED) {
        if (priv->lockState &&
            !(mig->lockState = strdup(priv->lockState)))
            return -1;
    } else {
        if (virDomainLockProcessInquire(driver->lockManager, dom, &mig->lockState) < 0)
            return -1;
    }

    if (!(mig->lockDriver = strdup(virLockManagerPluginGetName(driver->lockManager)))) {
        VIR_FREE(mig->lockState);
        return -1;
    }

    mig->flags |= QEMU_MIGRATION_COOKIE_LOCKSTATE;
    mig->flagsMandatory |= QEMU_MIGRATION_COOKIE_LOCKSTATE;

    return 0;
}


482 483 484 485 486
static int
qemuMigrationCookieAddPersistent(qemuMigrationCookiePtr mig,
                                 virDomainObjPtr dom)
{
    if (mig->flags & QEMU_MIGRATION_COOKIE_PERSISTENT) {
487 488
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Migration persistent data already present"));
489 490 491 492 493 494 495 496 497 498 499 500 501
        return -1;
    }

    if (!dom->newDef)
        return 0;

    mig->persistent = dom->newDef;
    mig->flags |= QEMU_MIGRATION_COOKIE_PERSISTENT;
    mig->flagsMandatory |= QEMU_MIGRATION_COOKIE_PERSISTENT;
    return 0;
}


502 503
static int
qemuMigrationCookieAddNetwork(qemuMigrationCookiePtr mig,
504
                              virQEMUDriverPtr driver,
505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522
                              virDomainObjPtr dom)
{
    if (mig->flags & QEMU_MIGRATION_COOKIE_NETWORK) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Network migration data already present"));
        return -1;
    }

    if (dom->def->nnets > 0) {
        mig->network = qemuMigrationCookieNetworkAlloc(driver, dom->def);
        if (!mig->network)
            return -1;
        mig->flags |= QEMU_MIGRATION_COOKIE_NETWORK;
    }

    return 0;
}

523

524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544
static int
qemuMigrationCookieAddNBD(qemuMigrationCookiePtr mig,
                          virQEMUDriverPtr driver ATTRIBUTE_UNUSED,
                          virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    /* It is not a bug if there already is a NBD data */
    if (!mig->nbd &&
        VIR_ALLOC(mig->nbd) < 0) {
        virReportOOMError();
        return -1;
    }

    mig->nbd->port = priv->nbdPort;
    mig->flags |= QEMU_MIGRATION_COOKIE_NBD;

    return 0;
}


545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562
static void qemuMigrationCookieGraphicsXMLFormat(virBufferPtr buf,
                                                 qemuMigrationCookieGraphicsPtr grap)
{
    virBufferAsprintf(buf, "  <graphics type='%s' port='%d' listen='%s'",
                      virDomainGraphicsTypeToString(grap->type),
                      grap->port, grap->listen);
    if (grap->type == VIR_DOMAIN_GRAPHICS_TYPE_SPICE)
        virBufferAsprintf(buf, " tlsPort='%d'", grap->tlsPort);
    if (grap->tlsSubject) {
        virBufferAddLit(buf, ">\n");
        virBufferEscapeString(buf, "    <cert info='subject' value='%s'/>\n", grap->tlsSubject);
        virBufferAddLit(buf, "  </graphics>\n");
    } else {
        virBufferAddLit(buf, "/>\n");
    }
}


563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593
static void
qemuMigrationCookieNetworkXMLFormat(virBufferPtr buf,
                                    qemuMigrationCookieNetworkPtr optr)
{
    int i;
    bool empty = true;

    for (i = 0; i < optr->nnets; i++) {
        /* If optr->net[i].vporttype is not set, there is nothing to transfer */
        if (optr->net[i].vporttype != VIR_NETDEV_VPORT_PROFILE_NONE) {
            if (empty) {
                virBufferAsprintf(buf, "  <network>\n");
                empty = false;
            }
            virBufferAsprintf(buf, "    <interface index='%d' vporttype='%s'",
                              i, virNetDevVPortTypeToString(optr->net[i].vporttype));
            if (optr->net[i].portdata) {
                virBufferAddLit(buf, ">\n");
                virBufferEscapeString(buf, "      <portdata>%s</portdata>\n",
                                      optr->net[i].portdata);
                virBufferAddLit(buf, "    </interface>\n");
            } else {
                virBufferAddLit(buf, "/>\n");
            }
        }
    }
    if (!empty)
        virBufferAddLit(buf, "  </network>\n");
}


594
static int
595
qemuMigrationCookieXMLFormat(virQEMUDriverPtr driver,
596
                             virBufferPtr buf,
597
                             qemuMigrationCookiePtr mig)
598 599 600
{
    char uuidstr[VIR_UUID_STRING_BUFLEN];
    char hostuuidstr[VIR_UUID_STRING_BUFLEN];
601
    int i;
602 603

    virUUIDFormat(mig->uuid, uuidstr);
604
    virUUIDFormat(mig->localHostuuid, hostuuidstr);
605 606 607 608

    virBufferAsprintf(buf, "<qemu-migration>\n");
    virBufferEscapeString(buf, "  <name>%s</name>\n", mig->name);
    virBufferAsprintf(buf, "  <uuid>%s</uuid>\n", uuidstr);
609
    virBufferEscapeString(buf, "  <hostname>%s</hostname>\n", mig->localHostname);
610
    virBufferAsprintf(buf, "  <hostuuid>%s</hostuuid>\n", hostuuidstr);
611

612 613 614 615 616 617
    for (i = 0 ; i < QEMU_MIGRATION_COOKIE_FLAG_LAST ; i++) {
        if (mig->flagsMandatory & (1 << i))
            virBufferAsprintf(buf, "  <feature name='%s'/>\n",
                              qemuMigrationCookieFlagTypeToString(i));
    }

618 619
    if ((mig->flags & QEMU_MIGRATION_COOKIE_GRAPHICS) &&
        mig->graphics)
620 621
        qemuMigrationCookieGraphicsXMLFormat(buf, mig->graphics);

622 623 624 625 626 627 628 629 630
    if ((mig->flags & QEMU_MIGRATION_COOKIE_LOCKSTATE) &&
        mig->lockState) {
        virBufferAsprintf(buf, "  <lockstate driver='%s'>\n",
                          mig->lockDriver);
        virBufferAsprintf(buf, "    <leases>%s</leases>\n",
                          mig->lockState);
        virBufferAddLit(buf, "  </lockstate>\n");
    }

631 632
    if ((mig->flags & QEMU_MIGRATION_COOKIE_PERSISTENT) &&
        mig->persistent) {
633
        virBufferAdjustIndent(buf, 2);
634 635 636
        if (qemuDomainDefFormatBuf(driver,
                                   mig->persistent,
                                   VIR_DOMAIN_XML_INACTIVE |
637 638
                                   VIR_DOMAIN_XML_SECURE |
                                   VIR_DOMAIN_XML_MIGRATABLE,
639
                                   buf) < 0)
640 641
            return -1;
        virBufferAdjustIndent(buf, -2);
642 643
    }

644 645 646
    if ((mig->flags & QEMU_MIGRATION_COOKIE_NETWORK) && mig->network)
        qemuMigrationCookieNetworkXMLFormat(buf, mig->network);

647 648 649 650 651 652 653
    if ((mig->flags & QEMU_MIGRATION_COOKIE_NBD) && mig->nbd) {
        virBufferAddLit(buf, "  <nbd");
        if (mig->nbd->port)
            virBufferAsprintf(buf, " port='%d'", mig->nbd->port);
        virBufferAddLit(buf, "/>\n");
    }

654
    virBufferAddLit(buf, "</qemu-migration>\n");
655
    return 0;
656 657 658
}


659
static char *qemuMigrationCookieXMLFormatStr(virQEMUDriverPtr driver,
660
                                             qemuMigrationCookiePtr mig)
661 662 663
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;

664
    if (qemuMigrationCookieXMLFormat(driver, &buf, mig) < 0) {
665 666 667
        virBufferFreeAndReset(&buf);
        return NULL;
    }
668 669 670

    if (virBufferError(&buf)) {
        virReportOOMError();
671
        virBufferFreeAndReset(&buf);
672 673 674 675 676 677 678
        return NULL;
    }

    return virBufferContentAndReset(&buf);
}


679 680 681 682 683 684 685 686 687 688
static qemuMigrationCookieGraphicsPtr
qemuMigrationCookieGraphicsXMLParse(xmlXPathContextPtr ctxt)
{
    qemuMigrationCookieGraphicsPtr grap;
    char *tmp;

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

    if (!(tmp = virXPathString("string(./graphics/@type)", ctxt))) {
689 690
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("missing type attribute in migration data"));
691 692 693
        goto error;
    }
    if ((grap->type = virDomainGraphicsTypeFromString(tmp)) < 0) {
694 695
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unknown graphics type %s"), tmp);
696 697 698
        VIR_FREE(tmp);
        goto error;
    }
E
Eric Blake 已提交
699
    VIR_FREE(tmp);
700
    if (virXPathInt("string(./graphics/@port)", ctxt, &grap->port) < 0) {
701 702
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("missing port attribute in migration data"));
703 704 705 706
        goto error;
    }
    if (grap->type == VIR_DOMAIN_GRAPHICS_TYPE_SPICE) {
        if (virXPathInt("string(./graphics/@tlsPort)", ctxt, &grap->tlsPort) < 0) {
707 708
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("missing tlsPort attribute in migration data"));
709 710 711 712
            goto error;
        }
    }
    if (!(grap->listen = virXPathString("string(./graphics/@listen)", ctxt))) {
713 714
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("missing listen attribute in migration data"));
715 716 717
        goto error;
    }
    /* Optional */
718
    grap->tlsSubject = virXPathString("string(./graphics/cert[@info='subject']/@value)", ctxt);
719 720 721 722 723 724 725 726 727 728 729

    return grap;

no_memory:
    virReportOOMError();
error:
    qemuMigrationCookieGraphicsFree(grap);
    return NULL;
}


730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749
static qemuMigrationCookieNetworkPtr
qemuMigrationCookieNetworkXMLParse(xmlXPathContextPtr ctxt)
{
    qemuMigrationCookieNetworkPtr optr;
    int i;
    int n;
    xmlNodePtr *interfaces = NULL;
    char *vporttype;
    xmlNodePtr save_ctxt = ctxt->node;

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

    if ((n = virXPathNodeSet("./network/interface", ctxt, &interfaces)) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("missing interface information"));
        goto error;
    }

    optr->nnets = n;
750
    if (VIR_ALLOC_N(optr->net, optr->nnets) < 0)
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
        goto no_memory;

    for (i = 0; i < n; i++) {
        /* portdata is optional, and may not exist */
        ctxt->node = interfaces[i];
        optr->net[i].portdata = virXPathString("string(./portdata[1])", ctxt);

        if (!(vporttype = virXMLPropString(interfaces[i], "vporttype"))) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("missing vporttype attribute in migration data"));
            goto error;
        }
        optr->net[i].vporttype = virNetDevVPortTypeFromString(vporttype);
    }

    VIR_FREE(interfaces);

cleanup:
    ctxt->node = save_ctxt;
    return optr;

no_memory:
    virReportOOMError();
error:
    VIR_FREE(interfaces);
    qemuMigrationCookieNetworkFree(optr);
    optr = NULL;
    goto cleanup;
}


782 783
static int
qemuMigrationCookieXMLParse(qemuMigrationCookiePtr mig,
784
                            virQEMUDriverPtr driver,
785
                            xmlDocPtr doc,
786
                            xmlXPathContextPtr ctxt,
E
Eric Blake 已提交
787
                            unsigned int flags)
788 789
{
    char uuidstr[VIR_UUID_STRING_BUFLEN];
790
    char *tmp = NULL;
791 792
    xmlNodePtr *nodes = NULL;
    int i, n;
793 794 795 796
    virCapsPtr caps = NULL;

    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
        goto error;
797 798 799 800 801 802 803 804

    /* 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))) {
805 806
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("missing name element in migration data"));
807 808 809
        goto error;
    }
    if (STRNEQ(tmp, mig->name)) {
810 811 812
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Incoming cookie data had unexpected name %s vs %s"),
                       tmp, mig->name);
813 814 815 816 817 818 819
        goto error;
    }
    VIR_FREE(tmp);

    /* Extract domain uuid */
    tmp = virXPathString("string(./uuid[1])", ctxt);
    if (!tmp) {
820 821
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("missing uuid element in migration data"));
822 823 824 825
        goto error;
    }
    virUUIDFormat(mig->uuid, uuidstr);
    if (STRNEQ(tmp, uuidstr)) {
826 827 828
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Incoming cookie data had unexpected UUID %s vs %s"),
                       tmp, uuidstr);
829 830 831 832
    }
    VIR_FREE(tmp);

    /* Check & forbid "localhost" migration */
833
    if (!(mig->remoteHostname = virXPathString("string(./hostname[1])", ctxt))) {
834 835
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("missing hostname element in migration data"));
836 837
        goto error;
    }
838
    if (STREQ(mig->remoteHostname, mig->localHostname)) {
839 840 841
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Attempt to migrate guest to the same host %s"),
                       mig->remoteHostname);
842 843 844 845
        goto error;
    }

    if (!(tmp = virXPathString("string(./hostuuid[1])", ctxt))) {
846 847 848
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("missing hostuuid element in migration data"));
        goto error;
849
    }
850 851 852 853 854 855 856 857 858 859 860 861
    if (virUUIDParse(tmp, mig->remoteHostuuid) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("malformed hostuuid element in migration data"));
        goto error;
    }
    if (memcmp(mig->remoteHostuuid, mig->localHostuuid, VIR_UUID_BUFLEN) == 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Attempt to migrate guest to the same host %s"),
                       tmp);
        goto error;
    }
    VIR_FREE(tmp);
862

863 864
    /* Check to ensure all mandatory features from XML are also
     * present in 'flags' */
865
    if ((n = virXPathNodeSet("./feature", ctxt, &nodes)) < 0)
866 867 868 869 870 871
        goto error;

    for (i = 0 ; i < n ; i++) {
        int val;
        char *str = virXMLPropString(nodes[i], "name");
        if (!str) {
872 873
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("missing feature name"));
874 875 876 877
            goto error;
        }

        if ((val = qemuMigrationCookieFlagTypeFromString(str)) < 0) {
878 879 880
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Unknown migration cookie feature %s"),
                           str);
881 882 883 884 885
            VIR_FREE(str);
            goto error;
        }

        if ((flags & (1 << val)) == 0) {
886 887 888
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Unsupported migration cookie feature %s"),
                           str);
889 890 891 892 893 894
            VIR_FREE(str);
        }
        VIR_FREE(str);
    }
    VIR_FREE(nodes);

895 896 897 898 899
    if ((flags & QEMU_MIGRATION_COOKIE_GRAPHICS) &&
        virXPathBoolean("count(./graphics) > 0", ctxt) &&
        (!(mig->graphics = qemuMigrationCookieGraphicsXMLParse(ctxt))))
        goto error;

900 901 902 903
    if ((flags & QEMU_MIGRATION_COOKIE_LOCKSTATE) &&
        virXPathBoolean("count(./lockstate) > 0", ctxt)) {
        mig->lockDriver = virXPathString("string(./lockstate[1]/@driver)", ctxt);
        if (!mig->lockDriver) {
904 905
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("Missing lock driver name in migration cookie"));
906 907 908 909 910 911 912
            goto error;
        }
        mig->lockState = virXPathString("string(./lockstate[1]/leases[1])", ctxt);
        if (mig->lockState && STREQ(mig->lockState, ""))
            VIR_FREE(mig->lockState);
    }

913 914 915
    if ((flags & QEMU_MIGRATION_COOKIE_PERSISTENT) &&
        virXPathBoolean("count(./domain) > 0", ctxt)) {
        if ((n = virXPathNodeSet("./domain", ctxt, &nodes)) > 1) {
916 917 918 919
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Too many domain elements in "
                             "migration cookie: %d"),
                           n);
920 921
            goto error;
        }
922 923
        mig->persistent = virDomainDefParseNode(caps, driver->xmlconf,
                                                doc, nodes[0],
924 925 926 927 928 929 930 931 932
                                                -1, VIR_DOMAIN_XML_INACTIVE);
        if (!mig->persistent) {
            /* virDomainDefParseNode already reported
             * an error for us */
            goto error;
        }
        VIR_FREE(nodes);
    }

933 934 935 936 937
    if ((flags & QEMU_MIGRATION_COOKIE_NETWORK) &&
        virXPathBoolean("count(./network) > 0", ctxt) &&
        (!(mig->network = qemuMigrationCookieNetworkXMLParse(ctxt))))
        goto error;

938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957
    if (flags & QEMU_MIGRATION_COOKIE_NBD &&
        virXPathBoolean("boolean(./nbd)", ctxt)) {
        char *port;

        if (VIR_ALLOC(mig->nbd) < 0) {
            virReportOOMError();
            goto error;
        }

        port = virXPathString("string(./nbd/@port)", ctxt);
        if (port && virStrToLong_i(port, NULL, 10, &mig->nbd->port) < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Malformed nbd port '%s'"),
                           port);
            VIR_FREE(port);
            goto error;
        }
        VIR_FREE(port);
    }

958
    virObjectUnref(caps);
959 960 961 962
    return 0;

error:
    VIR_FREE(tmp);
963
    VIR_FREE(nodes);
964
    virObjectUnref(caps);
965 966 967 968 969 970
    return -1;
}


static int
qemuMigrationCookieXMLParseStr(qemuMigrationCookiePtr mig,
971
                               virQEMUDriverPtr driver,
972
                               const char *xml,
E
Eric Blake 已提交
973
                               unsigned int flags)
974 975 976
{
    xmlDocPtr doc = NULL;
    xmlXPathContextPtr ctxt = NULL;
977
    int ret = -1;
978 979 980

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

981
    if (!(doc = virXMLParseStringCtxt(xml, _("(qemu_migration_cookie)"), &ctxt)))
982 983
        goto cleanup;

984
    ret = qemuMigrationCookieXMLParse(mig, driver, doc, ctxt, flags);
985 986 987 988 989 990 991 992 993 994 995

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

    return ret;
}


static int
qemuMigrationBakeCookie(qemuMigrationCookiePtr mig,
996
                        virQEMUDriverPtr driver,
997
                        virDomainObjPtr dom,
998 999
                        char **cookieout,
                        int *cookieoutlen,
E
Eric Blake 已提交
1000
                        unsigned int flags)
1001
{
1002 1003
    if (!cookieout || !cookieoutlen)
        return 0;
1004 1005 1006

    *cookieoutlen = 0;

1007 1008 1009 1010
    if (flags & QEMU_MIGRATION_COOKIE_GRAPHICS &&
        qemuMigrationCookieAddGraphics(mig, driver, dom) < 0)
        return -1;

1011 1012 1013 1014
    if (flags & QEMU_MIGRATION_COOKIE_LOCKSTATE &&
        qemuMigrationCookieAddLockstate(mig, driver, dom) < 0)
        return -1;

1015 1016 1017 1018
    if (flags & QEMU_MIGRATION_COOKIE_PERSISTENT &&
        qemuMigrationCookieAddPersistent(mig, dom) < 0)
        return -1;

1019 1020 1021 1022 1023
    if (flags & QEMU_MIGRATION_COOKIE_NETWORK &&
        qemuMigrationCookieAddNetwork(mig, driver, dom) < 0) {
        return -1;
    }

1024 1025 1026 1027
    if ((flags & QEMU_MIGRATION_COOKIE_NBD) &&
        qemuMigrationCookieAddNBD(mig, driver, dom) < 0)
        return -1;

1028
    if (!(*cookieout = qemuMigrationCookieXMLFormatStr(driver, mig)))
1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039
        return -1;

    *cookieoutlen = strlen(*cookieout) + 1;

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

    return 0;
}


static qemuMigrationCookiePtr
1040
qemuMigrationEatCookie(virQEMUDriverPtr driver,
1041
                       virDomainObjPtr dom,
1042 1043
                       const char *cookiein,
                       int cookieinlen,
E
Eric Blake 已提交
1044
                       unsigned int flags)
1045 1046 1047 1048 1049 1050
{
    qemuMigrationCookiePtr mig = NULL;

    /* Parse & validate incoming cookie (if any) */
    if (cookiein && cookieinlen &&
        cookiein[cookieinlen-1] != '\0') {
1051 1052
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Migration cookie was not NULL terminated"));
1053 1054 1055 1056 1057 1058 1059 1060 1061 1062
        goto error;
    }

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

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

    if (cookiein && cookieinlen &&
        qemuMigrationCookieXMLParseStr(mig,
1063
                                       driver,
1064 1065 1066 1067
                                       cookiein,
                                       flags) < 0)
        goto error;

1068 1069 1070
    if (mig->flags & QEMU_MIGRATION_COOKIE_LOCKSTATE) {
        if (!mig->lockDriver) {
            if (virLockManagerPluginUsesState(driver->lockManager)) {
1071 1072 1073
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("Missing %s lock state for migration cookie"),
                               virLockManagerPluginGetName(driver->lockManager));
1074 1075 1076 1077
                goto error;
            }
        } else if (STRNEQ(mig->lockDriver,
                          virLockManagerPluginGetName(driver->lockManager))) {
1078 1079 1080 1081
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Source host lock driver %s different from target %s"),
                           mig->lockDriver,
                           virLockManagerPluginGetName(driver->lockManager));
1082 1083 1084 1085
            goto error;
        }
    }

1086 1087 1088 1089 1090 1091
    return mig;

error:
    qemuMigrationCookieFree(mig);
    return NULL;
}
1092

1093 1094 1095 1096 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 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157
/**
 * qemuMigrationStartNBDServer:
 * @driver: qemu driver
 * @vm: domain
 *
 * Starts NBD server. This is a newer method to copy
 * storage during migration than using 'blk' and 'inc'
 * arguments in 'migrate' monitor command.
 * Error is reported here.
 *
 * Returns 0 on success, -1 otherwise.
 */
static int
qemuMigrationStartNBDServer(virQEMUDriverPtr driver,
                            virDomainObjPtr vm)
{
    int ret = -1;
    qemuDomainObjPrivatePtr priv = vm->privateData;
    unsigned short port = 0;
    const char *listenAddr = "0.0.0.0";
    char *diskAlias = NULL;
    size_t i;

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

        /* skip shared, RO and source-less disks */
        if (disk->shared || disk->readonly || !disk->src)
            continue;

        VIR_FREE(diskAlias);
        if (virAsprintf(&diskAlias, "%s%s",
                        QEMU_DRIVE_HOST_PREFIX, disk->info.alias) < 0) {
            virReportOOMError();
            goto cleanup;
        }

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

        if (!port &&
            ((virPortAllocatorAcquire(driver->remotePorts, &port) < 0) ||
             (qemuMonitorNBDServerStart(priv->mon, listenAddr, port) < 0))) {
            qemuDomainObjExitMonitor(driver, vm);
            goto cleanup;
        }

        if (qemuMonitorNBDServerAdd(priv->mon, diskAlias, true) < 0) {
            qemuDomainObjExitMonitor(driver, vm);
            goto cleanup;
        }
        qemuDomainObjExitMonitor(driver, vm);
    }

    priv->nbdPort = port;
    ret = 0;

cleanup:
    VIR_FREE(diskAlias);
    if ((ret < 0) && port)
        virPortAllocatorRelease(driver->remotePorts, port);
    return ret;
}

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 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 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
/**
 * qemuMigrationDriveMirror:
 * @driver: qemu driver
 * @vm: domain
 * @mig: migration cookie
 * @host: where are we migrating to
 * @speed: how much should the copying be limited
 * @migrate_flags: migrate monitor command flags
 *
 * Run drive-mirror to feed NBD server running on dst and wait
 * till the process switches into another phase where writes go
 * simultaneously to both source and destination. And this switch
 * is what we are waiting for before proceeding with the next
 * disk. On success, update @migrate_flags so we don't tell
 * 'migrate' command to do the very same operation.
 *
 * Returns 0 on success (@migrate_flags updated),
 *        -1 otherwise.
 */
static int
qemuMigrationDriveMirror(virQEMUDriverPtr driver,
                         virDomainObjPtr vm,
                         qemuMigrationCookiePtr mig,
                         const char *host,
                         unsigned long speed,
                         unsigned int *migrate_flags)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int ret = -1;
    int mon_ret;
    int port;
    size_t i, lastGood = 0;
    char *diskAlias = NULL;
    char *nbd_dest = NULL;
    unsigned int mirror_flags = VIR_DOMAIN_BLOCK_REBASE_REUSE_EXT;
    virErrorPtr err = NULL;

    if (!(*migrate_flags & (QEMU_MONITOR_MIGRATE_NON_SHARED_DISK |
                            QEMU_MONITOR_MIGRATE_NON_SHARED_INC)))
        return 0;

    if (!mig->nbd) {
        /* Destination doesn't support NBD server.
         * Fall back to previous implementation. */
        VIR_DEBUG("Destination doesn't support NBD server "
                  "Falling back to previous implementation.");
        return 0;
    }

    /* steal NBD port and thus prevent its propagation back to destination */
    port = mig->nbd->port;
    mig->nbd->port = 0;

    if (*migrate_flags & QEMU_MONITOR_MIGRATE_NON_SHARED_INC)
        mirror_flags |= VIR_DOMAIN_BLOCK_REBASE_SHALLOW;

    for (i = 0; i < vm->def->ndisks; i++) {
        virDomainDiskDefPtr disk = vm->def->disks[i];
        virDomainBlockJobInfo info;

        /* skip shared, RO and source-less disks */
        if (disk->shared || disk->readonly || !disk->src)
            continue;

        VIR_FREE(diskAlias);
        VIR_FREE(nbd_dest);
        if ((virAsprintf(&diskAlias, "%s%s",
                         QEMU_DRIVE_HOST_PREFIX, disk->info.alias) < 0) ||
            (virAsprintf(&nbd_dest, "nbd:%s:%d:exportname=%s",
                         host, port, diskAlias) < 0)) {
            virReportOOMError();
            goto error;
        }

        if (qemuDomainObjEnterMonitorAsync(driver, vm,
                                           QEMU_ASYNC_JOB_MIGRATION_OUT) < 0)
            goto error;
        mon_ret = qemuMonitorDriveMirror(priv->mon, diskAlias, nbd_dest,
                                         NULL, speed, mirror_flags);
        qemuDomainObjExitMonitor(driver, vm);

        if (mon_ret < 0)
            goto error;

        lastGood = i;

        /* wait for completion */
        while (true) {
            /* Poll every 500ms for progress & to allow cancellation */
            struct timespec ts = { .tv_sec = 0, .tv_nsec = 500 * 1000 * 1000ull };

            memset(&info, 0, sizeof(info));

            if (qemuDomainObjEnterMonitorAsync(driver, vm,
                                               QEMU_ASYNC_JOB_MIGRATION_OUT) < 0)
                goto error;
            if (priv->job.asyncAbort) {
                /* explicitly do this *after* we entered the monitor,
                 * as this is a critical section so we are guaranteed
                 * priv->job.asyncAbort will not change */
                qemuDomainObjExitMonitor(driver, vm);
                virReportError(VIR_ERR_OPERATION_ABORTED, _("%s: %s"),
                               qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                               _("canceled by client"));
                goto error;
            }
            mon_ret = qemuMonitorBlockJob(priv->mon, diskAlias, NULL, 0,
                                          &info, BLOCK_JOB_INFO, true);
            qemuDomainObjExitMonitor(driver, vm);

            if (mon_ret < 0)
                goto error;

            if (info.cur == info.end) {
                VIR_DEBUG("Drive mirroring of '%s' completed", diskAlias);
                break;
            }

            /* XXX Frankly speaking, we should listen to the events,
             * instead of doing this. But this works for now and we
             * are doing something similar in migration itself anyway */

            virObjectUnlock(vm);

            nanosleep(&ts, NULL);

            virObjectLock(vm);
        }
    }

    /* Okay, copied. Modify migrate_flags */
    *migrate_flags &= ~(QEMU_MONITOR_MIGRATE_NON_SHARED_DISK |
                        QEMU_MONITOR_MIGRATE_NON_SHARED_INC);
    ret = 0;

cleanup:
    VIR_FREE(diskAlias);
    VIR_FREE(nbd_dest);
    return ret;

error:
    /* don't overwrite any errors */
    err = virSaveLastError();
    /* cancel any outstanding jobs */
    while (lastGood) {
        virDomainDiskDefPtr disk = vm->def->disks[--lastGood];

        /* skip shared, RO disks */
        if (disk->shared || disk->readonly || !disk->src)
            continue;

        VIR_FREE(diskAlias);
        if (virAsprintf(&diskAlias, "%s%s",
                        QEMU_DRIVE_HOST_PREFIX, disk->info.alias) < 0) {
            virReportOOMError();
            continue;
        }
        if (qemuDomainObjEnterMonitorAsync(driver, vm,
                                           QEMU_ASYNC_JOB_MIGRATION_OUT) == 0) {
            if (qemuMonitorBlockJob(priv->mon, diskAlias, NULL, 0,
                                    NULL, BLOCK_JOB_ABORT, true) < 0) {
                VIR_WARN("Unable to cancel block-job on '%s'", diskAlias);
            }
            qemuDomainObjExitMonitor(driver, vm);
        } else {
            VIR_WARN("Unable to enter monitor. No block job cancelled");
        }
    }
    if (err)
        virSetError(err);
    virFreeError(err);
    goto cleanup;
}
1331

1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355

static void
qemuMigrationStopNBDServer(virQEMUDriverPtr driver,
                           virDomainObjPtr vm,
                           qemuMigrationCookiePtr mig)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    if (!mig->nbd)
        return;

    if (qemuDomainObjEnterMonitorAsync(driver, vm,
                                       QEMU_ASYNC_JOB_MIGRATION_IN) < 0)
        return;

    if (qemuMonitorNBDServerStop(priv->mon) < 0)
        VIR_WARN("Unable to stop NBD server");

    qemuDomainObjExitMonitor(driver, vm);

    virPortAllocatorRelease(driver->remotePorts, priv->nbdPort);
    priv->nbdPort = 0;
}

1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 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
static void
qemuMigrationCancelDriveMirror(qemuMigrationCookiePtr mig,
                               virQEMUDriverPtr driver,
                               virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    size_t i;
    char *diskAlias = NULL;

    VIR_DEBUG("mig=%p nbdPort=%d", mig->nbd, priv->nbdPort);

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

        /* skip shared, RO and source-less disks */
        if (disk->shared || disk->readonly || !disk->src)
            continue;

        VIR_FREE(diskAlias);
        if (virAsprintf(&diskAlias, "%s%s",
                        QEMU_DRIVE_HOST_PREFIX, disk->info.alias) < 0) {
            virReportOOMError();
            goto cleanup;
        }

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

        if (qemuMonitorBlockJob(priv->mon, diskAlias, NULL, 0,
                                NULL, BLOCK_JOB_ABORT, true) < 0)
            VIR_WARN("Unable to stop block job on %s", diskAlias);
        qemuDomainObjExitMonitor(driver, vm);
    }

cleanup:
    VIR_FREE(diskAlias);
    return;
}

1396 1397 1398 1399 1400 1401 1402 1403 1404
/* Validate whether the domain is safe to migrate.  If vm is NULL,
 * then this is being run in the v2 Prepare stage on the destination
 * (where we only have the target xml); if vm is provided, then this
 * is being run in either v2 Perform or v3 Begin (where we also have
 * access to all of the domain's metadata, such as whether it is
 * marked autodestroy or has snapshots).  While it would be nice to
 * assume that checking on source is sufficient to prevent ever
 * talking to the destination in the first place, we are stuck with
 * the fact that older servers did not do checks on the source. */
1405
bool
1406
qemuMigrationIsAllowed(virQEMUDriverPtr driver, virDomainObjPtr vm,
1407
                       virDomainDefPtr def, bool remote)
1408
{
1409
    int nsnapshots;
1410 1411
    bool forbid;
    int i;
1412 1413 1414

    if (vm) {
        if (qemuProcessAutoDestroyActive(driver, vm)) {
1415 1416
            virReportError(VIR_ERR_OPERATION_INVALID,
                           "%s", _("domain is marked for auto destroy"));
1417 1418
            return false;
        }
1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431

        /* perform these checks only when migrating to remote hosts */
        if (remote) {
            nsnapshots = virDomainSnapshotObjListNum(vm->snapshots, NULL, 0);
            if (nsnapshots < 0)
                return false;

            if (nsnapshots > 0) {
                virReportError(VIR_ERR_OPERATION_INVALID,
                               _("cannot migrate domain with %d snapshots"),
                               nsnapshots);
                return false;
            }
1432
        }
1433

E
Eric Blake 已提交
1434 1435
        if (virDomainHasDiskMirror(vm)) {
            virReportError(VIR_ERR_OPERATION_INVALID, "%s",
1436
                           _("domain has an active block job"));
E
Eric Blake 已提交
1437 1438
            return false;
        }
1439 1440 1441

        def = vm->def;
    }
1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456

    /* Migration with USB host devices is allowed, all other devices are
     * forbidden.
     */
    forbid = false;
    for (i = 0; i < def->nhostdevs; i++) {
        virDomainHostdevDefPtr hostdev = def->hostdevs[i];
        if (hostdev->mode != VIR_DOMAIN_HOSTDEV_MODE_SUBSYS ||
            hostdev->source.subsys.type != VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB) {
            forbid = true;
            break;
        }
    }
    if (forbid) {
        virReportError(VIR_ERR_OPERATION_INVALID, "%s",
1457
                       _("domain has assigned non-USB host devices"));
1458 1459 1460 1461 1462 1463
        return false;
    }

    return true;
}

1464 1465 1466 1467 1468 1469 1470 1471
static bool
qemuMigrationIsSafe(virDomainDefPtr def)
{
    int i;

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

1472 1473
        /* Our code elsewhere guarantees shared disks are either readonly (in
         * which case cache mode doesn't matter) or used with cache=none */
1474
        if (disk->src &&
1475 1476 1477
            !disk->shared &&
            !disk->readonly &&
            disk->cachemode != VIR_DOMAIN_DISK_CACHE_DISABLE) {
1478
            int cfs;
1479 1480 1481 1482 1483 1484

            if (disk->type == VIR_DOMAIN_DISK_TYPE_FILE) {
                if ((cfs = virStorageFileIsClusterFS(disk->src)) == 1)
                    continue;
                else if (cfs < 0)
                    return false;
1485 1486 1487
            } else if (disk->type == VIR_DOMAIN_DISK_TYPE_NETWORK &&
                       disk->protocol == VIR_DOMAIN_DISK_PROTOCOL_RBD) {
                continue;
1488
            }
1489

1490 1491 1492
            virReportError(VIR_ERR_MIGRATE_UNSAFE, "%s",
                           _("Migration may lead to data corruption if disks"
                             " use cache != none"));
1493 1494 1495 1496 1497 1498 1499
            return false;
        }
    }

    return true;
}

1500 1501 1502 1503
/** qemuMigrationSetOffline
 * Pause domain for non-live migration.
 */
int
1504
qemuMigrationSetOffline(virQEMUDriverPtr driver,
1505 1506 1507
                        virDomainObjPtr vm)
{
    int ret;
1508
    VIR_DEBUG("driver=%p vm=%p", driver, vm);
1509 1510
    ret = qemuProcessStopCPUs(driver, vm, VIR_DOMAIN_PAUSED_MIGRATION,
                              QEMU_ASYNC_JOB_MIGRATION_OUT);
1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524
    if (ret == 0) {
        virDomainEventPtr event;

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

    return ret;
}


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 1560 1561 1562 1563 1564 1565
static int
qemuMigrationSetCompression(virQEMUDriverPtr driver,
                            virDomainObjPtr vm,
                            enum qemuDomainAsyncJob job)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int ret;

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

    ret = qemuMonitorGetMigrationCapability(
                priv->mon,
                QEMU_MONITOR_MIGRATION_CAPS_XBZRLE);

    if (ret < 0) {
        goto cleanup;
    } else if (ret == 0) {
        if (job == QEMU_ASYNC_JOB_MIGRATION_IN) {
            virReportError(VIR_ERR_ARGUMENT_UNSUPPORTED, "%s",
                           _("Compressed migration is not supported by "
                             "target QEMU binary"));
        } else {
            virReportError(VIR_ERR_ARGUMENT_UNSUPPORTED, "%s",
                           _("Compressed migration is not supported by "
                             "source QEMU binary"));
        }
        ret = -1;
        goto cleanup;
    }

    ret = qemuMonitorSetMigrationCapability(
                priv->mon,
                QEMU_MONITOR_MIGRATION_CAPS_XBZRLE);

cleanup:
    qemuDomainObjExitMonitor(driver, vm);
    return ret;
}


1566
static int
1567
qemuMigrationUpdateJobStatus(virQEMUDriverPtr driver,
1568
                             virDomainObjPtr vm,
1569 1570
                             const char *job,
                             enum qemuDomainAsyncJob asyncJob)
1571
{
1572
    qemuDomainObjPrivatePtr priv = vm->privateData;
1573
    int ret;
1574 1575
    bool wait_for_spice = false;
    bool spice_migrated = false;
1576 1577 1578
    qemuMonitorMigrationStatus status;

    memset(&status, 0, sizeof(status));
1579

1580 1581 1582 1583
    /* If guest uses SPICE and supports seamles_migration we have to hold up
     * migration finish until SPICE server transfers its data */
    if (vm->def->ngraphics == 1 &&
        vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_SPICE &&
1584
        virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_SEAMLESS_MIGRATION))
1585 1586
        wait_for_spice = true;

1587 1588 1589
    ret = qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob);
    if (ret < 0) {
        /* Guest already exited; nothing further to update.  */
1590 1591
        return -1;
    }
1592
    ret = qemuMonitorGetMigrationStatus(priv->mon, &status);
1593 1594

    /* If qemu says migrated, check spice */
1595 1596 1597
    if (wait_for_spice &&
        ret == 0 &&
        status.status == QEMU_MONITOR_MIGRATION_STATUS_COMPLETED)
1598 1599 1600
        ret = qemuMonitorGetSpiceMigrationStatus(priv->mon,
                                                 &spice_migrated);

1601
    qemuDomainObjExitMonitor(driver, vm);
1602

1603 1604
    priv->job.status = status;

1605
    if (ret < 0 || virTimeMillisNow(&priv->job.info.timeElapsed) < 0) {
1606
        priv->job.info.type = VIR_DOMAIN_JOB_FAILED;
1607 1608
        return -1;
    }
1609
    priv->job.info.timeElapsed -= priv->job.start;
1610

1611
    ret = -1;
1612
    switch (priv->job.status.status) {
1613
    case QEMU_MONITOR_MIGRATION_STATUS_INACTIVE:
1614
        priv->job.info.type = VIR_DOMAIN_JOB_NONE;
1615 1616
        virReportError(VIR_ERR_OPERATION_FAILED,
                       _("%s: %s"), job, _("is not active"));
1617 1618 1619
        break;

    case QEMU_MONITOR_MIGRATION_STATUS_ACTIVE:
1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634
        priv->job.info.fileTotal = priv->job.status.disk_total;
        priv->job.info.fileRemaining = priv->job.status.disk_remaining;
        priv->job.info.fileProcessed = priv->job.status.disk_transferred;

        priv->job.info.memTotal = priv->job.status.ram_total;
        priv->job.info.memRemaining = priv->job.status.ram_remaining;
        priv->job.info.memProcessed = priv->job.status.ram_transferred;

        priv->job.info.dataTotal =
            priv->job.status.ram_total + priv->job.status.disk_total;
        priv->job.info.dataRemaining =
            priv->job.status.ram_remaining + priv->job.status.disk_remaining;
        priv->job.info.dataProcessed =
            priv->job.status.ram_transferred +
            priv->job.status.disk_transferred;
1635 1636 1637 1638 1639

        ret = 0;
        break;

    case QEMU_MONITOR_MIGRATION_STATUS_COMPLETED:
1640 1641
        if ((wait_for_spice && spice_migrated) || (!wait_for_spice))
            priv->job.info.type = VIR_DOMAIN_JOB_COMPLETED;
1642 1643 1644 1645
        ret = 0;
        break;

    case QEMU_MONITOR_MIGRATION_STATUS_ERROR:
1646
        priv->job.info.type = VIR_DOMAIN_JOB_FAILED;
1647 1648
        virReportError(VIR_ERR_OPERATION_FAILED,
                       _("%s: %s"), job, _("unexpectedly failed"));
1649 1650 1651
        break;

    case QEMU_MONITOR_MIGRATION_STATUS_CANCELLED:
1652
        priv->job.info.type = VIR_DOMAIN_JOB_CANCELLED;
1653 1654
        virReportError(VIR_ERR_OPERATION_ABORTED,
                       _("%s: %s"), job, _("canceled by client"));
1655 1656 1657 1658 1659 1660 1661
        break;
    }

    return ret;
}


1662
static int
1663
qemuMigrationWaitForCompletion(virQEMUDriverPtr driver, virDomainObjPtr vm,
1664 1665
                               enum qemuDomainAsyncJob asyncJob,
                               virConnectPtr dconn)
1666
{
1667
    qemuDomainObjPrivatePtr priv = vm->privateData;
1668 1669
    const char *job;

1670 1671
    switch (priv->job.asyncJob) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
1672 1673
        job = _("migration job");
        break;
1674
    case QEMU_ASYNC_JOB_SAVE:
1675 1676
        job = _("domain save job");
        break;
1677
    case QEMU_ASYNC_JOB_DUMP:
1678 1679 1680 1681 1682
        job = _("domain core dump job");
        break;
    default:
        job = _("job");
    }
1683

1684
    priv->job.info.type = VIR_DOMAIN_JOB_UNBOUNDED;
1685

1686
    while (priv->job.info.type == VIR_DOMAIN_JOB_UNBOUNDED) {
1687 1688 1689
        /* Poll every 50ms for progress & to allow cancellation */
        struct timespec ts = { .tv_sec = 0, .tv_nsec = 50 * 1000 * 1000ull };

1690
        if (qemuMigrationUpdateJobStatus(driver, vm, job, asyncJob) < 0)
1691 1692
            goto cleanup;

1693
        if (dconn && virConnectIsAlive(dconn) <= 0) {
1694 1695
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("Lost connection to destination host"));
1696 1697 1698
            goto cleanup;
        }

1699
        virObjectUnlock(vm);
1700 1701 1702

        nanosleep(&ts, NULL);

1703
        virObjectLock(vm);
1704 1705 1706
    }

cleanup:
1707
    if (priv->job.info.type == VIR_DOMAIN_JOB_COMPLETED)
1708 1709 1710
        return 0;
    else
        return -1;
1711 1712 1713
}


1714
static int
1715
qemuDomainMigrateGraphicsRelocate(virQEMUDriverPtr driver,
1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733
                                  virDomainObjPtr vm,
                                  qemuMigrationCookiePtr cookie)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int ret;

    if (!cookie)
        return 0;

    if (!cookie->graphics)
        return 0;

    /* QEMU doesn't support VNC relocation yet, so
     * skip it to avoid generating an error
     */
    if (cookie->graphics->type != VIR_DOMAIN_GRAPHICS_TYPE_SPICE)
        return 0;

1734 1735
    ret = qemuDomainObjEnterMonitorAsync(driver, vm,
                                         QEMU_ASYNC_JOB_MIGRATION_OUT);
1736 1737 1738 1739 1740 1741 1742
    if (ret == 0) {
        ret = qemuMonitorGraphicsRelocate(priv->mon,
                                          cookie->graphics->type,
                                          cookie->remoteHostname,
                                          cookie->graphics->port,
                                          cookie->graphics->tlsPort,
                                          cookie->graphics->tlsSubject);
1743
        qemuDomainObjExitMonitor(driver, vm);
1744
    }
1745 1746 1747 1748 1749

    return ret;
}


1750
static int
1751
qemuDomainMigrateOPDRelocate(virQEMUDriverPtr driver ATTRIBUTE_UNUSED,
1752 1753 1754
                             virDomainObjPtr vm,
                             qemuMigrationCookiePtr cookie)
{
1755 1756
    virDomainNetDefPtr netptr;
    int ret = -1;
1757 1758 1759 1760 1761 1762 1763 1764 1765
    int i;

    for (i = 0; i < cookie->network->nnets; i++) {
        netptr = vm->def->nets[i];

        switch (cookie->network->net[i].vporttype) {
        case VIR_NETDEV_VPORT_PROFILE_NONE:
        case VIR_NETDEV_VPORT_PROFILE_8021QBG:
        case VIR_NETDEV_VPORT_PROFILE_8021QBH:
1766
           break;
1767
        case VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH:
1768 1769 1770 1771 1772 1773 1774 1775
            if (virNetDevOpenvswitchSetMigrateData(cookie->network->net[i].portdata,
                                                   netptr->ifname) != 0) {
                virReportSystemError(VIR_ERR_INTERNAL_ERROR,
                                     _("Unable to run command to set OVS port data for "
                                     "interface %s"), netptr->ifname);
                goto cleanup;
            }
            break;
1776 1777 1778 1779 1780
        default:
            break;
        }
    }

1781 1782
    ret = 0;
cleanup:
1783 1784 1785 1786
    return ret;
}


1787 1788 1789 1790 1791 1792 1793
/* This is called for outgoing non-p2p migrations when a connection to the
 * client which initiated the migration was closed but we were waiting for it
 * to follow up with the next phase, that is, in between
 * qemuDomainMigrateBegin3 and qemuDomainMigratePerform3 or
 * qemuDomainMigratePerform3 and qemuDomainMigrateConfirm3.
 */
virDomainObjPtr
1794
qemuMigrationCleanup(virQEMUDriverPtr driver,
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
                     virDomainObjPtr vm,
                     virConnectPtr conn)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    VIR_DEBUG("vm=%s, conn=%p, asyncJob=%s, phase=%s",
              vm->def->name, conn,
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
              qemuDomainAsyncJobPhaseToString(priv->job.asyncJob,
                                              priv->job.phase));

    if (!qemuMigrationJobIsActive(vm, QEMU_ASYNC_JOB_MIGRATION_OUT))
        goto cleanup;

    VIR_DEBUG("The connection which started outgoing migration of domain %s"
              " was closed; canceling the migration",
              vm->def->name);

    switch ((enum qemuMigrationJobPhase) priv->job.phase) {
    case QEMU_MIGRATION_PHASE_BEGIN3:
        /* just forget we were about to migrate */
        qemuDomainObjDiscardAsyncJob(driver, vm);
        break;

    case QEMU_MIGRATION_PHASE_PERFORM3_DONE:
        VIR_WARN("Migration of domain %s finished but we don't know if the"
                 " domain was successfully started on destination or not",
                 vm->def->name);
        /* clear the job and let higher levels decide what to do */
        qemuDomainObjDiscardAsyncJob(driver, vm);
        break;

    case QEMU_MIGRATION_PHASE_PERFORM3:
        /* cannot be seen without an active migration API; unreachable */
    case QEMU_MIGRATION_PHASE_CONFIRM3:
    case QEMU_MIGRATION_PHASE_CONFIRM3_CANCELLED:
        /* all done; unreachable */
    case QEMU_MIGRATION_PHASE_PREPARE:
    case QEMU_MIGRATION_PHASE_FINISH2:
    case QEMU_MIGRATION_PHASE_FINISH3:
        /* incoming migration; unreachable */
    case QEMU_MIGRATION_PHASE_PERFORM2:
        /* single phase outgoing migration; unreachable */
    case QEMU_MIGRATION_PHASE_NONE:
    case QEMU_MIGRATION_PHASE_LAST:
        /* unreachable */
        ;
    }

cleanup:
    return vm;
}

1848
/* The caller is supposed to lock the vm and start a migration job. */
1849
char *qemuMigrationBegin(virQEMUDriverPtr driver,
1850
                         virDomainObjPtr vm,
1851
                         const char *xmlin,
1852
                         const char *dname,
1853
                         char **cookieout,
1854 1855
                         int *cookieoutlen,
                         unsigned long flags)
1856 1857 1858
{
    char *rv = NULL;
    qemuMigrationCookiePtr mig = NULL;
1859
    virDomainDefPtr def = NULL;
1860
    qemuDomainObjPrivatePtr priv = vm->privateData;
1861
    virCapsPtr caps = NULL;
1862
    unsigned int cookieFlags = QEMU_MIGRATION_COOKIE_LOCKSTATE;
1863

1864
    VIR_DEBUG("driver=%p, vm=%p, xmlin=%s, dname=%s,"
1865
              " cookieout=%p, cookieoutlen=%p, flags=%lx",
1866
              driver, vm, NULLSTR(xmlin), NULLSTR(dname),
1867
              cookieout, cookieoutlen, flags);
1868

1869 1870 1871
    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
        goto cleanup;

1872 1873 1874 1875 1876 1877
    /* Only set the phase if we are inside QEMU_ASYNC_JOB_MIGRATION_OUT.
     * Otherwise we will start the async job later in the perform phase losing
     * change protection.
     */
    if (priv->job.asyncJob == QEMU_ASYNC_JOB_MIGRATION_OUT)
        qemuMigrationJobSetPhase(driver, vm, QEMU_MIGRATION_PHASE_BEGIN3);
1878

1879
    if (!qemuMigrationIsAllowed(driver, vm, NULL, true))
1880 1881
        goto cleanup;

1882 1883 1884
    if (!(flags & VIR_MIGRATE_UNSAFE) && !qemuMigrationIsSafe(vm->def))
        goto cleanup;

1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895
    if (flags & (VIR_MIGRATE_NON_SHARED_DISK | VIR_MIGRATE_NON_SHARED_INC) &&
        virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_NBD_SERVER)) {
        /* TODO support NBD for TUNNELLED migration */
        if (flags & VIR_MIGRATE_TUNNELLED)
            VIR_DEBUG("NBD in tunnelled migration is currently not supported");
        else {
            cookieFlags |= QEMU_MIGRATION_COOKIE_NBD;
            priv->nbdPort = 0;
        }
    }

1896
    if (!(mig = qemuMigrationEatCookie(driver, vm, NULL, 0, 0)))
1897 1898 1899 1900
        goto cleanup;

    if (qemuMigrationBakeCookie(mig, driver, vm,
                                cookieout, cookieoutlen,
1901
                                cookieFlags) < 0)
1902 1903
        goto cleanup;

L
liguang 已提交
1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925
    if (flags & VIR_MIGRATE_OFFLINE) {
        if (flags & (VIR_MIGRATE_NON_SHARED_DISK |
                     VIR_MIGRATE_NON_SHARED_INC)) {
            virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                           _("offline migration cannot handle "
                             "non-shared storage"));
            goto cleanup;
        }
        if (!(flags & VIR_MIGRATE_PERSIST_DEST)) {
            virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                           _("offline migration must be specified with "
                             "the persistent flag set"));
            goto cleanup;
        }
        if (flags & VIR_MIGRATE_TUNNELLED) {
            virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                           _("tunnelled offline migration does not "
                             "make sense"));
            goto cleanup;
        }
    }

1926
    if (xmlin) {
1927
        if (!(def = virDomainDefParseString(caps, driver->xmlconf, xmlin,
M
Matthias Bolte 已提交
1928
                                            QEMU_EXPECTED_VIRT_TYPES,
1929 1930 1931
                                            VIR_DOMAIN_XML_INACTIVE)))
            goto cleanup;

1932
        if (STRNEQ(def->name, vm->def->name)) {
1933 1934
            virReportError(VIR_ERR_INVALID_ARG, "%s",
                           _("target domain name doesn't match source name"));
1935 1936 1937
            goto cleanup;
        }

1938
        if (!virDomainDefCheckABIStability(vm->def, def))
1939 1940
            goto cleanup;

1941
        rv = qemuDomainDefFormatLive(driver, def, false, true);
1942
    } else {
1943
        rv = qemuDomainDefFormatLive(driver, vm->def, false, true);
1944
    }
1945 1946 1947

cleanup:
    qemuMigrationCookieFree(mig);
1948
    virObjectUnref(caps);
1949
    virDomainDefFree(def);
1950 1951 1952 1953
    return rv;
}


1954 1955
/* Prepare is the first step, and it runs on the destination host.
 */
1956

1957
static void
1958
qemuMigrationPrepareCleanup(virQEMUDriverPtr driver,
1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973
                            virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    VIR_DEBUG("driver=%p, vm=%s, job=%s, asyncJob=%s",
              driver,
              vm->def->name,
              qemuDomainJobTypeToString(priv->job.active),
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob));

    if (!qemuMigrationJobIsActive(vm, QEMU_ASYNC_JOB_MIGRATION_IN))
        return;
    qemuDomainObjDiscardAsyncJob(driver, vm);
}

1974
static int
1975
qemuMigrationPrepareAny(virQEMUDriverPtr driver,
1976 1977 1978 1979 1980 1981 1982 1983
                        virConnectPtr dconn,
                        const char *cookiein,
                        int cookieinlen,
                        char **cookieout,
                        int *cookieoutlen,
                        const char *dname,
                        const char *dom_xml,
                        const char *migrateFrom,
L
liguang 已提交
1984 1985
                        virStreamPtr st,
                        unsigned long flags)
1986 1987 1988 1989 1990
{
    virDomainDefPtr def = NULL;
    virDomainObjPtr vm = NULL;
    virDomainEventPtr event = NULL;
    int ret = -1;
1991
    int dataFD[2] = { -1, -1 };
1992
    qemuDomainObjPrivatePtr priv = NULL;
J
Jiri Denemark 已提交
1993
    unsigned long long now;
1994
    qemuMigrationCookiePtr mig = NULL;
1995
    bool tunnel = !!st;
J
Jiri Denemark 已提交
1996
    char *origname = NULL;
J
Jiri Denemark 已提交
1997
    char *xmlout = NULL;
L
liguang 已提交
1998
    unsigned int cookieFlags;
1999
    virCapsPtr caps = NULL;
2000

2001
    if (virTimeMillisNow(&now) < 0)
2002 2003
        return -1;

L
liguang 已提交
2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025
    if (flags & VIR_MIGRATE_OFFLINE) {
        if (flags & (VIR_MIGRATE_NON_SHARED_DISK |
                     VIR_MIGRATE_NON_SHARED_INC)) {
            virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                           _("offline migration cannot handle "
                             "non-shared storage"));
            goto cleanup;
        }
        if (!(flags & VIR_MIGRATE_PERSIST_DEST)) {
            virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                           _("offline migration must be specified with "
                             "the persistent flag set"));
            goto cleanup;
        }
        if (tunnel) {
            virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                           _("tunnelled offline migration does not "
                             "make sense"));
            goto cleanup;
        }
    }

2026 2027 2028
    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
        goto cleanup;

2029
    if (!(def = virDomainDefParseString(caps, driver->xmlconf, dom_xml,
M
Matthias Bolte 已提交
2030
                                        QEMU_EXPECTED_VIRT_TYPES,
2031 2032 2033
                                        VIR_DOMAIN_XML_INACTIVE)))
        goto cleanup;

2034
    if (!qemuMigrationIsAllowed(driver, NULL, def, true))
2035 2036 2037 2038
        goto cleanup;

    /* Target domain name, maybe renamed. */
    if (dname) {
J
Jiri Denemark 已提交
2039
        origname = def->name;
2040 2041 2042 2043 2044
        def->name = strdup(dname);
        if (def->name == NULL)
            goto cleanup;
    }

J
Jiri Denemark 已提交
2045 2046 2047 2048 2049
    /* Let migration hook filter domain XML */
    if (virHookPresent(VIR_HOOK_DRIVER_QEMU)) {
        char *xml;
        int hookret;

2050
        if (!(xml = qemuDomainDefFormatXML(driver, def,
2051 2052
                                           VIR_DOMAIN_XML_SECURE |
                                           VIR_DOMAIN_XML_MIGRATABLE)))
J
Jiri Denemark 已提交
2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069
            goto cleanup;

        hookret = virHookCall(VIR_HOOK_DRIVER_QEMU, def->name,
                              VIR_HOOK_QEMU_OP_MIGRATE, VIR_HOOK_SUBOP_BEGIN,
                              NULL, xml, &xmlout);
        VIR_FREE(xml);

        if (hookret < 0) {
            goto cleanup;
        } else if (hookret == 0) {
            if (!*xmlout) {
                VIR_DEBUG("Migrate hook filter returned nothing; using the"
                          " original XML");
            } else {
                virDomainDefPtr newdef;

                VIR_DEBUG("Using hook-filtered domain XML: %s", xmlout);
2070
                newdef = virDomainDefParseString(caps, driver->xmlconf, xmlout,
J
Jiri Denemark 已提交
2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086
                                                 QEMU_EXPECTED_VIRT_TYPES,
                                                 VIR_DOMAIN_XML_INACTIVE);
                if (!newdef)
                    goto cleanup;

                if (!virDomainDefCheckABIStability(def, newdef)) {
                    virDomainDefFree(newdef);
                    goto cleanup;
                }

                virDomainDefFree(def);
                def = newdef;
            }
        }
    }

2087
    if (!(vm = virDomainObjListAdd(driver->domains,
2088
                                   driver->xmlconf,
2089 2090 2091 2092
                                   def,
                                   VIR_DOMAIN_OBJ_LIST_ADD_LIVE |
                                   VIR_DOMAIN_OBJ_LIST_ADD_CHECK_LIVE,
                                   NULL)))
2093
        goto cleanup;
2094

2095 2096
    def = NULL;
    priv = vm->privateData;
J
Jiri Denemark 已提交
2097 2098
    priv->origname = origname;
    origname = NULL;
2099

2100
    if (!(mig = qemuMigrationEatCookie(driver, vm, cookiein, cookieinlen,
2101 2102
                                       QEMU_MIGRATION_COOKIE_LOCKSTATE |
                                       QEMU_MIGRATION_COOKIE_NBD)))
2103 2104
        goto cleanup;

2105
    if (qemuMigrationJobStart(driver, vm, QEMU_ASYNC_JOB_MIGRATION_IN) < 0)
2106
        goto cleanup;
2107
    qemuMigrationJobSetPhase(driver, vm, QEMU_MIGRATION_PHASE_PREPARE);
2108 2109 2110 2111

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

L
liguang 已提交
2112 2113 2114
    if (flags & VIR_MIGRATE_OFFLINE)
        goto done;

2115 2116
    if (tunnel &&
        (pipe(dataFD) < 0 || virSetCloseExec(dataFD[1]) < 0)) {
2117 2118
        virReportSystemError(errno, "%s",
                             _("cannot create pipe for tunnelled migration"));
2119 2120 2121
        goto endjob;
    }

2122 2123 2124 2125
    if (qemuDomainObjBeginNestedJob(driver, vm,
                                    QEMU_ASYNC_JOB_MIGRATION_IN) < 0)
        goto endjob;

2126
    /* Start the QEMU daemon, with the same command-line arguments plus
2127
     * -incoming $migrateFrom
2128
     */
2129 2130 2131
    if (qemuProcessStart(dconn, driver, vm, migrateFrom, dataFD[0], NULL, NULL,
                         VIR_NETDEV_VPORT_PROFILE_OP_MIGRATE_IN_START,
                         VIR_QEMU_PROCESS_START_PAUSED |
2132
                         VIR_QEMU_PROCESS_START_AUTODESTROY) < 0) {
2133
        virDomainAuditStart(vm, "migrated", false);
2134 2135
        if (qemuDomainObjEndJob(driver, vm) < 0)
            vm = NULL;
2136 2137 2138
        goto endjob;
    }

2139 2140 2141 2142
    if (tunnel) {
        if (virFDStreamOpen(st, dataFD[1]) < 0) {
            virReportSystemError(errno, "%s",
                                 _("cannot pass pipe for tunnelled migration"));
2143
            goto stop;
2144
        }
2145
        dataFD[1] = -1; /* 'st' owns the FD now & will close it */
2146 2147
    }

2148 2149 2150 2151 2152
    if (flags & VIR_MIGRATE_COMPRESSED &&
        qemuMigrationSetCompression(driver, vm,
                                    QEMU_ASYNC_JOB_MIGRATION_IN) < 0)
        goto stop;

2153 2154 2155 2156 2157 2158 2159 2160
    if (mig->lockState) {
        VIR_DEBUG("Received lockstate %s", mig->lockState);
        VIR_FREE(priv->lockState);
        priv->lockState = mig->lockState;
        mig->lockState = NULL;
    } else {
        VIR_DEBUG("Received no lockstate");
    }
2161

L
liguang 已提交
2162 2163 2164 2165 2166 2167
done:
    if (flags & VIR_MIGRATE_OFFLINE)
        cookieFlags = 0;
    else
        cookieFlags = QEMU_MIGRATION_COOKIE_GRAPHICS;

2168 2169 2170 2171 2172 2173 2174
    if (mig->nbd &&
        flags & (VIR_MIGRATE_NON_SHARED_DISK | VIR_MIGRATE_NON_SHARED_INC) &&
        virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_NBD_SERVER)) {
        /* TODO support NBD for TUNNELLED migration */
        if (flags & VIR_MIGRATE_TUNNELLED)
            VIR_DEBUG("NBD in tunnelled migration is currently not supported");
        else {
2175 2176 2177 2178
            if (qemuMigrationStartNBDServer(driver, vm) < 0) {
                /* error already reported */
                goto endjob;
            }
2179 2180 2181 2182 2183 2184
            cookieFlags |= QEMU_MIGRATION_COOKIE_NBD;
        }
    }

    if (qemuMigrationBakeCookie(mig, driver, vm, cookieout,
                                cookieoutlen, cookieFlags) < 0) {
2185 2186 2187 2188 2189 2190 2191
        /* 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");
    }

2192 2193 2194
    if (qemuDomainCleanupAdd(vm, qemuMigrationPrepareCleanup) < 0)
        goto endjob;

L
liguang 已提交
2195 2196 2197 2198 2199 2200
    if (!(flags & VIR_MIGRATE_OFFLINE)) {
        virDomainAuditStart(vm, "migrated", true);
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STARTED,
                                         VIR_DOMAIN_EVENT_STARTED_MIGRATED);
    }
2201

2202 2203 2204 2205
    /* We keep the job active across API calls until the finish() call.
     * This prevents any other APIs being invoked while incoming
     * migration is taking place.
     */
2206
    if (!qemuMigrationJobContinue(vm)) {
2207
        vm = NULL;
2208 2209
        virReportError(VIR_ERR_OPERATION_FAILED,
                       "%s", _("domain disappeared"));
2210
        goto cleanup;
2211
    }
2212

2213
    ret = 0;
2214 2215

cleanup:
J
Jiri Denemark 已提交
2216
    VIR_FREE(origname);
J
Jiri Denemark 已提交
2217
    VIR_FREE(xmlout);
2218
    virDomainDefFree(def);
2219 2220
    VIR_FORCE_CLOSE(dataFD[0]);
    VIR_FORCE_CLOSE(dataFD[1]);
2221 2222
    if (vm) {
        if (ret >= 0 || vm->persistent)
2223
            virObjectUnlock(vm);
2224 2225
        else
            qemuDomainRemoveInactive(driver, vm);
2226 2227 2228 2229
        if (ret < 0 && priv->nbdPort) {
            virPortAllocatorRelease(driver->remotePorts, priv->nbdPort);
            priv->nbdPort = 0;
        }
2230
    }
2231 2232
    if (event)
        qemuDomainEventQueue(driver, event);
2233
    qemuMigrationCookieFree(mig);
2234
    virObjectUnref(caps);
2235
    return ret;
2236

2237 2238 2239 2240
stop:
    virDomainAuditStart(vm, "migrated", false);
    qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_FAILED, 0);

2241
endjob:
2242
    if (vm && !qemuMigrationJobFinish(driver, vm)) {
2243 2244 2245
        vm = NULL;
    }
    goto cleanup;
2246 2247 2248
}


2249 2250 2251 2252 2253
/*
 * This version starts an empty VM listening on a localhost TCP port, and
 * sets up the corresponding virStream to handle the incoming data.
 */
int
2254
qemuMigrationPrepareTunnel(virQEMUDriverPtr driver,
2255 2256 2257 2258 2259 2260 2261
                           virConnectPtr dconn,
                           const char *cookiein,
                           int cookieinlen,
                           char **cookieout,
                           int *cookieoutlen,
                           virStreamPtr st,
                           const char *dname,
L
liguang 已提交
2262 2263
                           const char *dom_xml,
                           unsigned long flags)
2264 2265 2266 2267
{
    int ret;

    VIR_DEBUG("driver=%p, dconn=%p, cookiein=%s, cookieinlen=%d, "
2268 2269
              "cookieout=%p, cookieoutlen=%p, st=%p, dname=%s, dom_xml=%s "
              "flags=%lx",
2270
              driver, dconn, NULLSTR(cookiein), cookieinlen,
2271
              cookieout, cookieoutlen, st, NULLSTR(dname), dom_xml, flags);
2272 2273 2274 2275 2276 2277

    /* QEMU will be started with -incoming stdio (which qemu_command might
     * convert to exec:cat or fd:n)
     */
    ret = qemuMigrationPrepareAny(driver, dconn, cookiein, cookieinlen,
                                  cookieout, cookieoutlen, dname, dom_xml,
L
liguang 已提交
2278
                                  "stdio", st, flags);
2279 2280 2281 2282
    return ret;
}


2283
int
2284
qemuMigrationPrepareDirect(virQEMUDriverPtr driver,
2285
                           virConnectPtr dconn,
2286 2287 2288 2289
                           const char *cookiein,
                           int cookieinlen,
                           char **cookieout,
                           int *cookieoutlen,
2290 2291 2292
                           const char *uri_in,
                           char **uri_out,
                           const char *dname,
L
liguang 已提交
2293 2294
                           const char *dom_xml,
                           unsigned long flags)
2295 2296 2297 2298 2299 2300 2301
{
    static int port = 0;
    int this_port;
    char *hostname = NULL;
    char migrateFrom [64];
    const char *p;
    int ret = -1;
J
Jiri Denemark 已提交
2302

2303 2304
    VIR_DEBUG("driver=%p, dconn=%p, cookiein=%s, cookieinlen=%d, "
              "cookieout=%p, cookieoutlen=%p, uri_in=%s, uri_out=%p, "
2305
              "dname=%s, dom_xml=%s flags=%lx",
2306 2307
              driver, dconn, NULLSTR(cookiein), cookieinlen,
              cookieout, cookieoutlen, NULLSTR(uri_in), uri_out,
2308
              NULLSTR(dname), dom_xml, flags);
2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328

    /* 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")) {
2329 2330 2331
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("hostname on destination resolved to localhost,"
                             " but migration requires an FQDN"));
2332 2333 2334 2335 2336
            goto cleanup;
        }

        /* XXX this really should have been a properly well-formed
         * URI, but we can't add in tcp:// now without breaking
2337
         * compatibility with old targets. We at least make the
2338 2339 2340
         * new targets accept both syntaxes though.
         */
        /* Caller frees */
2341
        if (virAsprintf(uri_out, "tcp:%s:%d", hostname, this_port) < 0) {
2342 2343 2344 2345 2346 2347 2348 2349
            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.
         */
2350
        if (!STRPREFIX(uri_in, "tcp:")) {
2351 2352 2353
            virReportError(VIR_ERR_INVALID_ARG, "%s",
                           _("only tcp URIs are supported for KVM/QEMU"
                             " migrations"));
2354 2355 2356 2357
            goto cleanup;
        }

        /* Get the port number. */
2358
        p = strrchr(uri_in, ':');
2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372
        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 */
2373 2374
            this_port = virParseNumber(&p);
            if (this_port == -1 || p-uri_in != strlen(uri_in)) {
2375 2376
                virReportError(VIR_ERR_INVALID_ARG,
                               "%s", _("URI ended with incorrect ':port'"));
2377 2378 2379 2380 2381 2382 2383 2384
                goto cleanup;
            }
        }
    }

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

2385
    /* QEMU will be started with -incoming tcp:0.0.0.0:port */
2386
    snprintf(migrateFrom, sizeof(migrateFrom), "tcp:0.0.0.0:%d", this_port);
2387

2388 2389
    ret = qemuMigrationPrepareAny(driver, dconn, cookiein, cookieinlen,
                                  cookieout, cookieoutlen, dname, dom_xml,
L
liguang 已提交
2390
                                  migrateFrom, NULL, flags);
2391 2392 2393 2394 2395 2396 2397 2398
cleanup:
    VIR_FREE(hostname);
    if (ret != 0)
        VIR_FREE(*uri_out);
    return ret;
}


2399 2400
enum qemuMigrationDestinationType {
    MIGRATION_DEST_HOST,
2401
    MIGRATION_DEST_CONNECT_HOST,
2402
    MIGRATION_DEST_UNIX,
2403
    MIGRATION_DEST_FD,
2404
};
2405

2406 2407 2408 2409
enum qemuMigrationForwardType {
    MIGRATION_FWD_DIRECT,
    MIGRATION_FWD_STREAM,
};
2410

2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421
typedef struct _qemuMigrationSpec qemuMigrationSpec;
typedef qemuMigrationSpec *qemuMigrationSpecPtr;
struct _qemuMigrationSpec {
    enum qemuMigrationDestinationType destType;
    union {
        struct {
            const char *name;
            int port;
        } host;

        struct {
2422
            char *file;
2423 2424
            int sock;
        } unix_socket;
2425 2426 2427 2428 2429

        struct {
            int qemu;
            int local;
        } fd;
2430 2431 2432 2433 2434 2435 2436
    } dest;

    enum qemuMigrationForwardType fwdType;
    union {
        virStreamPtr stream;
    } fwd;
};
2437 2438 2439

#define TUNNEL_SEND_BUF_SIZE 65536

2440 2441 2442 2443 2444 2445 2446
typedef struct _qemuMigrationIOThread qemuMigrationIOThread;
typedef qemuMigrationIOThread *qemuMigrationIOThreadPtr;
struct _qemuMigrationIOThread {
    virThread thread;
    virStreamPtr st;
    int sock;
    virError err;
2447 2448
    int wakeupRecvFD;
    int wakeupSendFD;
2449 2450 2451
};

static void qemuMigrationIOFunc(void *arg)
2452
{
2453
    qemuMigrationIOThreadPtr data = arg;
2454 2455 2456 2457 2458 2459 2460
    char *buffer = NULL;
    struct pollfd fds[2];
    int timeout = -1;
    virErrorPtr err = NULL;

    VIR_DEBUG("Running migration tunnel; stream=%p, sock=%d",
              data->st, data->sock);
2461 2462 2463

    if (VIR_ALLOC_N(buffer, TUNNEL_SEND_BUF_SIZE) < 0) {
        virReportOOMError();
2464
        goto abrt;
2465 2466
    }

2467 2468 2469
    fds[0].fd = data->sock;
    fds[1].fd = data->wakeupRecvFD;

2470
    for (;;) {
2471 2472 2473 2474 2475 2476 2477 2478 2479 2480
        int ret;

        fds[0].events = fds[1].events = POLLIN;
        fds[0].revents = fds[1].revents = 0;

        ret = poll(fds, ARRAY_CARDINALITY(fds), timeout);

        if (ret < 0) {
            if (errno == EAGAIN || errno == EINTR)
                continue;
2481
            virReportSystemError(errno, "%s",
2482 2483
                                 _("poll failed in migration tunnel"));
            goto abrt;
2484
        }
2485 2486 2487 2488 2489 2490 2491

        if (ret == 0) {
            /* We were asked to gracefully stop but reading would block. This
             * can only happen if qemu told us migration finished but didn't
             * close the migration fd. We handle this in the same way as EOF.
             */
            VIR_DEBUG("QEMU forgot to close migration fd");
2492
            break;
2493
        }
2494

2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510
        if (fds[1].revents & (POLLIN | POLLERR | POLLHUP)) {
            char stop = 0;

            if (saferead(data->wakeupRecvFD, &stop, 1) != 1) {
                virReportSystemError(errno, "%s",
                                     _("failed to read from wakeup fd"));
                goto abrt;
            }

            VIR_DEBUG("Migration tunnel was asked to %s",
                      stop ? "abort" : "finish");
            if (stop) {
                goto abrt;
            } else {
                timeout = 0;
            }
2511 2512
        }

2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529
        if (fds[0].revents & (POLLIN | POLLERR | POLLHUP)) {
            int nbytes;

            nbytes = saferead(data->sock, buffer, TUNNEL_SEND_BUF_SIZE);
            if (nbytes > 0) {
                if (virStreamSend(data->st, buffer, nbytes) < 0)
                    goto error;
            } else if (nbytes < 0) {
                virReportSystemError(errno, "%s",
                        _("tunnelled migration failed to read from qemu"));
                goto abrt;
            } else {
                /* EOF; get out of here */
                break;
            }
        }
    }
2530

2531 2532
    if (virStreamFinish(data->st) < 0)
        goto error;
2533

2534 2535
    VIR_FREE(buffer);

2536 2537
    return;

2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549
abrt:
    err = virSaveLastError();
    if (err && err->code == VIR_ERR_OK) {
        virFreeError(err);
        err = NULL;
    }
    virStreamAbort(data->st);
    if (err) {
        virSetError(err);
        virFreeError(err);
    }

2550 2551 2552
error:
    virCopyLastError(&data->err);
    virResetLastError();
2553
    VIR_FREE(buffer);
2554 2555 2556 2557 2558 2559 2560
}


static qemuMigrationIOThreadPtr
qemuMigrationStartTunnel(virStreamPtr st,
                         int sock)
{
2561 2562
    qemuMigrationIOThreadPtr io = NULL;
    int wakeupFD[2] = { -1, -1 };
2563

2564 2565 2566 2567
    if (pipe2(wakeupFD, O_CLOEXEC) < 0) {
        virReportSystemError(errno, "%s",
                             _("Unable to make pipe"));
        goto error;
2568 2569
    }

2570 2571 2572
    if (VIR_ALLOC(io) < 0)
        goto no_memory;

2573 2574
    io->st = st;
    io->sock = sock;
2575 2576
    io->wakeupRecvFD = wakeupFD[0];
    io->wakeupSendFD = wakeupFD[1];
2577 2578 2579 2580 2581 2582

    if (virThreadCreate(&io->thread, true,
                        qemuMigrationIOFunc,
                        io) < 0) {
        virReportSystemError(errno, "%s",
                             _("Unable to create migration thread"));
2583
        goto error;
2584 2585 2586
    }

    return io;
2587 2588 2589 2590 2591 2592 2593 2594

no_memory:
    virReportOOMError();
error:
    VIR_FORCE_CLOSE(wakeupFD[0]);
    VIR_FORCE_CLOSE(wakeupFD[1]);
    VIR_FREE(io);
    return NULL;
2595 2596 2597
}

static int
2598
qemuMigrationStopTunnel(qemuMigrationIOThreadPtr io, bool error)
2599 2600
{
    int rv = -1;
2601 2602 2603 2604 2605 2606 2607 2608 2609
    char stop = error ? 1 : 0;

    /* make sure the thread finishes its job and is joinable */
    if (safewrite(io->wakeupSendFD, &stop, 1) != 1) {
        virReportSystemError(errno, "%s",
                             _("failed to wakeup migration tunnel"));
        goto cleanup;
    }

2610 2611 2612 2613
    virThreadJoin(&io->thread);

    /* Forward error from the IO thread, to this thread */
    if (io->err.code != VIR_ERR_OK) {
2614 2615 2616 2617
        if (error)
            rv = 0;
        else
            virSetError(&io->err);
2618 2619 2620 2621 2622 2623 2624
        virResetError(&io->err);
        goto cleanup;
    }

    rv = 0;

cleanup:
2625 2626
    VIR_FORCE_CLOSE(io->wakeupSendFD);
    VIR_FORCE_CLOSE(io->wakeupRecvFD);
2627 2628
    VIR_FREE(io);
    return rv;
2629 2630
}

2631
static int
2632
qemuMigrationConnect(virQEMUDriverPtr driver,
2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653
                     virDomainObjPtr vm,
                     qemuMigrationSpecPtr spec)
{
    virNetSocketPtr sock;
    const char *host;
    char *port = NULL;
    int ret = -1;

    host = spec->dest.host.name;
    if (virAsprintf(&port, "%d", spec->dest.host.port) < 0) {
        virReportOOMError();
        return -1;
    }

    spec->destType = MIGRATION_DEST_FD;
    spec->dest.fd.qemu = -1;

    if (virSecurityManagerSetSocketLabel(driver->securityManager, vm->def) < 0)
        goto cleanup;
    if (virNetSocketNewConnectTCP(host, port, &sock) == 0) {
        spec->dest.fd.qemu = virNetSocketDupFD(sock, true);
2654
        virObjectUnref(sock);
2655 2656 2657 2658 2659
    }
    if (virSecurityManagerClearSocketLabel(driver->securityManager, vm->def) < 0 ||
        spec->dest.fd.qemu == -1)
        goto cleanup;

2660 2661 2662 2663 2664 2665 2666
    /* Migration expects a blocking FD */
    if (virSetBlocking(spec->dest.fd.qemu, true) < 0) {
        virReportSystemError(errno, _("Unable to set FD %d blocking"),
                             spec->dest.fd.qemu);
        goto cleanup;
    }

2667 2668 2669 2670 2671 2672 2673 2674 2675
    ret = 0;

cleanup:
    VIR_FREE(port);
    if (ret < 0)
        VIR_FORCE_CLOSE(spec->dest.fd.qemu);
    return ret;
}

2676
static int
2677
qemuMigrationRun(virQEMUDriverPtr driver,
2678 2679 2680 2681 2682 2683 2684
                 virDomainObjPtr vm,
                 const char *cookiein,
                 int cookieinlen,
                 char **cookieout,
                 int *cookieoutlen,
                 unsigned long flags,
                 unsigned long resource,
2685 2686
                 qemuMigrationSpecPtr spec,
                 virConnectPtr dconn)
2687
{
2688
    int ret = -1;
2689 2690
    unsigned int migrate_flags = QEMU_MONITOR_MIGRATE_BACKGROUND;
    qemuDomainObjPrivatePtr priv = vm->privateData;
2691
    qemuMigrationCookiePtr mig = NULL;
2692
    qemuMigrationIOThreadPtr iothread = NULL;
2693
    int fd = -1;
2694
    unsigned long migrate_speed = resource ? resource : priv->migMaxBandwidth;
2695
    virErrorPtr orig_err = NULL;
2696
    unsigned int cookieFlags = 0;
2697 2698 2699 2700 2701 2702 2703

    VIR_DEBUG("driver=%p, vm=%p, cookiein=%s, cookieinlen=%d, "
              "cookieout=%p, cookieoutlen=%p, flags=%lx, resource=%lu, "
              "spec=%p (dest=%d, fwd=%d)",
              driver, vm, NULLSTR(cookiein), cookieinlen,
              cookieout, cookieoutlen, flags, resource,
              spec, spec->destType, spec->fwdType);
2704

2705 2706 2707 2708 2709 2710 2711 2712 2713 2714
    if (flags & VIR_MIGRATE_NON_SHARED_DISK) {
        migrate_flags |= QEMU_MONITOR_MIGRATE_NON_SHARED_DISK;
        cookieFlags |= QEMU_MIGRATION_COOKIE_NBD;
    }

    if (flags & VIR_MIGRATE_NON_SHARED_INC) {
        migrate_flags |= QEMU_MONITOR_MIGRATE_NON_SHARED_INC;
        cookieFlags |= QEMU_MIGRATION_COOKIE_NBD;
    }

2715 2716
    if (virLockManagerPluginUsesState(driver->lockManager) &&
        !cookieout) {
2717 2718 2719 2720
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Migration with lock driver %s requires"
                         " cookie support"),
                       virLockManagerPluginGetName(driver->lockManager));
2721 2722 2723
        return -1;
    }

2724 2725 2726
    mig = qemuMigrationEatCookie(driver, vm, cookiein, cookieinlen,
                                 cookieFlags | QEMU_MIGRATION_COOKIE_GRAPHICS);
    if (!mig)
2727 2728 2729 2730 2731
        goto cleanup;

    if (qemuDomainMigrateGraphicsRelocate(driver, vm, mig) < 0)
        VIR_WARN("unable to provide data for graphics client relocation");

2732 2733 2734 2735 2736 2737 2738
    /* this will update migrate_flags on success */
    if (qemuMigrationDriveMirror(driver, vm, mig, spec->dest.host.name,
                                 migrate_speed, &migrate_flags) < 0) {
        /* error reported by helper func */
        goto cleanup;
    }

2739
    /* Before EnterMonitor, since qemuMigrationSetOffline already does that */
2740 2741 2742 2743 2744 2745
    if (!(flags & VIR_MIGRATE_LIVE) &&
        virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING) {
        if (qemuMigrationSetOffline(driver, vm) < 0)
            goto cleanup;
    }

2746 2747 2748 2749 2750
    if (flags & VIR_MIGRATE_COMPRESSED &&
        qemuMigrationSetCompression(driver, vm,
                                    QEMU_ASYNC_JOB_MIGRATION_OUT) < 0)
        goto cleanup;

2751 2752
    if (qemuDomainObjEnterMonitorAsync(driver, vm,
                                       QEMU_ASYNC_JOB_MIGRATION_OUT) < 0)
2753 2754
        goto cleanup;

2755 2756 2757 2758
    if (priv->job.asyncAbort) {
        /* explicitly do this *after* we entered the monitor,
         * as this is a critical section so we are guaranteed
         * priv->job.asyncAbort will not change */
2759
        qemuDomainObjExitMonitor(driver, vm);
2760 2761 2762 2763 2764 2765
        virReportError(VIR_ERR_OPERATION_ABORTED, _("%s: %s"),
                       qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                       _("canceled by client"));
        goto cleanup;
    }

2766
    if (qemuMonitorSetMigrationSpeed(priv->mon, migrate_speed) < 0) {
2767
        qemuDomainObjExitMonitor(driver, vm);
2768 2769
        goto cleanup;
    }
2770

2771 2772
    /* connect to the destination qemu if needed */
    if (spec->destType == MIGRATION_DEST_CONNECT_HOST &&
2773
        qemuMigrationConnect(driver, vm, spec) < 0) {
2774
        qemuDomainObjExitMonitor(driver, vm);
2775
        goto cleanup;
2776
    }
2777

2778 2779 2780 2781 2782 2783 2784
    switch (spec->destType) {
    case MIGRATION_DEST_HOST:
        ret = qemuMonitorMigrateToHost(priv->mon, migrate_flags,
                                       spec->dest.host.name,
                                       spec->dest.host.port);
        break;

2785 2786 2787 2788
    case MIGRATION_DEST_CONNECT_HOST:
        /* handled above and transformed into MIGRATION_DEST_FD */
        break;

2789
    case MIGRATION_DEST_UNIX:
2790
        if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_UNIX)) {
2791 2792 2793 2794 2795 2796 2797 2798 2799
            ret = qemuMonitorMigrateToUnix(priv->mon, migrate_flags,
                                           spec->dest.unix_socket.file);
        } else {
            const char *args[] = {
                "nc", "-U", spec->dest.unix_socket.file, NULL
            };
            ret = qemuMonitorMigrateToCommand(priv->mon, migrate_flags, args);
        }
        break;
2800 2801

    case MIGRATION_DEST_FD:
2802
        if (spec->fwdType != MIGRATION_FWD_DIRECT) {
2803
            fd = spec->dest.fd.local;
2804 2805
            spec->dest.fd.local = -1;
        }
2806 2807 2808 2809
        ret = qemuMonitorMigrateToFd(priv->mon, migrate_flags,
                                     spec->dest.fd.qemu);
        VIR_FORCE_CLOSE(spec->dest.fd.qemu);
        break;
2810
    }
2811
    qemuDomainObjExitMonitor(driver, vm);
2812
    if (ret < 0)
2813 2814
        goto cleanup;
    ret = -1;
2815 2816

    if (!virDomainObjIsActive(vm)) {
2817 2818
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("guest unexpectedly quit"));
2819 2820 2821 2822 2823 2824
        goto cleanup;
    }

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

2825 2826 2827 2828 2829 2830 2831 2832
    if (spec->destType == MIGRATION_DEST_UNIX) {
        /* It is also possible that the migrate didn't fail initially, but
         * rather failed later on.  Check its status before waiting for a
         * connection from qemu which may never be initiated.
         */
        if (qemuMigrationUpdateJobStatus(driver, vm, _("migration job"),
                                         QEMU_ASYNC_JOB_MIGRATION_OUT) < 0)
            goto cancel;
2833

2834 2835 2836 2837 2838 2839 2840
        while ((fd = accept(spec->dest.unix_socket.sock, NULL, NULL)) < 0) {
            if (errno == EAGAIN || errno == EINTR)
                continue;
            virReportSystemError(errno, "%s",
                                 _("failed to accept connection from qemu"));
            goto cancel;
        }
2841 2842
    }

2843 2844
    if (spec->fwdType != MIGRATION_FWD_DIRECT &&
        !(iothread = qemuMigrationStartTunnel(spec->fwd.stream, fd)))
2845 2846
        goto cancel;

2847
    if (qemuMigrationWaitForCompletion(driver, vm,
2848 2849
                                       QEMU_ASYNC_JOB_MIGRATION_OUT,
                                       dconn) < 0)
2850
        goto cleanup;
2851

2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864
    /* When migration completed, QEMU will have paused the
     * CPUs for us, but unless we're using the JSON monitor
     * we won't have been notified of this, so might still
     * think we're running. For v2 protocol this doesn't
     * matter because we'll kill the VM soon, but for v3
     * this is important because we stay paused until the
     * confirm3 step, but need to release the lock state
     */
    if (virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING) {
        if (qemuMigrationSetOffline(driver, vm) < 0)
            goto cleanup;
    }

2865
    ret = 0;
2866

2867
cleanup:
2868 2869 2870
    if (ret < 0 && !orig_err)
        orig_err = virSaveLastError();

2871 2872 2873
    /* cancel any outstanding NBD jobs */
    qemuMigrationCancelDriveMirror(mig, driver, vm);

2874
    if (spec->fwdType != MIGRATION_FWD_DIRECT) {
2875
        if (iothread && qemuMigrationStopTunnel(iothread, ret < 0) < 0)
2876
            ret = -1;
2877
        VIR_FORCE_CLOSE(fd);
2878
    }
2879

2880 2881
    cookieFlags |= (QEMU_MIGRATION_COOKIE_PERSISTENT |
                    QEMU_MIGRATION_COOKIE_NETWORK);
2882
    if (ret == 0 &&
2883 2884
        qemuMigrationBakeCookie(mig, driver, vm, cookieout,
                                cookieoutlen, cookieFlags) < 0) {
2885
        VIR_WARN("Unable to encode migration cookie");
2886
    }
2887

2888 2889
    qemuMigrationCookieFree(mig);

2890 2891 2892 2893 2894
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }

2895 2896
    return ret;

2897
cancel:
2898 2899
    orig_err = virSaveLastError();

2900
    if (virDomainObjIsActive(vm)) {
2901 2902
        if (qemuDomainObjEnterMonitorAsync(driver, vm,
                                           QEMU_ASYNC_JOB_MIGRATION_OUT) == 0) {
2903
            qemuMonitorMigrateCancel(priv->mon);
2904
            qemuDomainObjExitMonitor(driver, vm);
2905
        }
2906
    }
2907 2908 2909 2910 2911 2912
    goto cleanup;
}

/* Perform migration using QEMU's native TCP migrate support,
 * not encrypted obviously
 */
2913
static int doNativeMigrate(virQEMUDriverPtr driver,
2914 2915 2916 2917 2918 2919 2920
                           virDomainObjPtr vm,
                           const char *uri,
                           const char *cookiein,
                           int cookieinlen,
                           char **cookieout,
                           int *cookieoutlen,
                           unsigned long flags,
2921 2922
                           unsigned long resource,
                           virConnectPtr dconn)
2923
{
2924
    qemuDomainObjPrivatePtr priv = vm->privateData;
M
Martin Kletzander 已提交
2925
    virURIPtr uribits = NULL;
2926
    int ret = -1;
2927 2928 2929 2930 2931 2932 2933 2934
    qemuMigrationSpec spec;

    VIR_DEBUG("driver=%p, vm=%p, uri=%s, cookiein=%s, cookieinlen=%d, "
              "cookieout=%p, cookieoutlen=%p, flags=%lx, resource=%lu",
              driver, vm, uri, NULLSTR(cookiein), cookieinlen,
              cookieout, cookieoutlen, flags, resource);

    if (STRPREFIX(uri, "tcp:") && !STRPREFIX(uri, "tcp://")) {
2935
        char *tmp;
2936
        /* HACK: source host generates bogus URIs, so fix them up */
2937
        if (virAsprintf(&tmp, "tcp://%s", uri + strlen("tcp:")) < 0) {
2938 2939 2940
            virReportOOMError();
            return -1;
        }
M
Martin Kletzander 已提交
2941
        uribits = virURIParse(tmp);
2942
        VIR_FREE(tmp);
2943
    } else {
M
Martin Kletzander 已提交
2944
        uribits = virURIParse(uri);
2945
    }
2946
    if (!uribits)
2947 2948
        return -1;

2949
    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_FD))
2950 2951
        spec.destType = MIGRATION_DEST_CONNECT_HOST;
    else
2952
        spec.destType = MIGRATION_DEST_HOST;
2953 2954 2955
    spec.dest.host.name = uribits->server;
    spec.dest.host.port = uribits->port;
    spec.fwdType = MIGRATION_FWD_DIRECT;
2956

2957
    ret = qemuMigrationRun(driver, vm, cookiein, cookieinlen, cookieout,
2958
                           cookieoutlen, flags, resource, &spec, dconn);
2959 2960 2961 2962

    if (spec.destType == MIGRATION_DEST_FD)
        VIR_FORCE_CLOSE(spec.dest.fd.qemu);

2963
    virURIFree(uribits);
2964 2965 2966 2967 2968

    return ret;
}


2969
static int doTunnelMigrate(virQEMUDriverPtr driver,
2970 2971 2972 2973 2974 2975 2976
                           virDomainObjPtr vm,
                           virStreamPtr st,
                           const char *cookiein,
                           int cookieinlen,
                           char **cookieout,
                           int *cookieoutlen,
                           unsigned long flags,
2977 2978
                           unsigned long resource,
                           virConnectPtr dconn)
2979 2980
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
2981
    virNetSocketPtr sock = NULL;
2982 2983
    int ret = -1;
    qemuMigrationSpec spec;
2984
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
2985 2986 2987 2988 2989 2990

    VIR_DEBUG("driver=%p, vm=%p, st=%p, cookiein=%s, cookieinlen=%d, "
              "cookieout=%p, cookieoutlen=%p, flags=%lx, resource=%lu",
              driver, vm, st, NULLSTR(cookiein), cookieinlen,
              cookieout, cookieoutlen, flags, resource);

2991 2992 2993
    if (!virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_FD) &&
        !virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_UNIX) &&
        !virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_EXEC)) {
2994 2995
        virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                       _("Source qemu is too old to support tunnelled migration"));
2996
        virObjectUnref(cfg);
2997
        return -1;
2998 2999 3000 3001 3002
    }

    spec.fwdType = MIGRATION_FWD_STREAM;
    spec.fwd.stream = st;

3003
    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_FD)) {
3004 3005 3006 3007 3008 3009
        int fds[2];

        spec.destType = MIGRATION_DEST_FD;
        spec.dest.fd.qemu = -1;
        spec.dest.fd.local = -1;

3010
        if (pipe2(fds, O_CLOEXEC) == 0) {
3011 3012 3013 3014
            spec.dest.fd.qemu = fds[1];
            spec.dest.fd.local = fds[0];
        }
        if (spec.dest.fd.qemu == -1 ||
3015
            virSecurityManagerSetImageFDLabel(driver->securityManager, vm->def,
3016
                                              spec.dest.fd.qemu) < 0) {
3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027
            virReportSystemError(errno, "%s",
                        _("cannot create pipe for tunnelled migration"));
            goto cleanup;
        }
    } else {
        spec.destType = MIGRATION_DEST_UNIX;
        spec.dest.unix_socket.sock = -1;
        spec.dest.unix_socket.file = NULL;

        if (virAsprintf(&spec.dest.unix_socket.file,
                        "%s/qemu.tunnelmigrate.src.%s",
3028
                        cfg->libDir, vm->def->name) < 0) {
3029 3030 3031 3032 3033
            virReportOOMError();
            goto cleanup;
        }

        if (virNetSocketNewListenUNIX(spec.dest.unix_socket.file, 0700,
3034
                                      cfg->user, cfg->group,
3035 3036 3037 3038 3039 3040 3041
                                      &sock) < 0 ||
            virNetSocketListen(sock, 1) < 0)
            goto cleanup;

        spec.dest.unix_socket.sock = virNetSocketGetFD(sock);
    }

3042
    ret = qemuMigrationRun(driver, vm, cookiein, cookieinlen, cookieout,
3043
                           cookieoutlen, flags, resource, &spec, dconn);
3044

3045
cleanup:
3046 3047 3048 3049
    if (spec.destType == MIGRATION_DEST_FD) {
        VIR_FORCE_CLOSE(spec.dest.fd.qemu);
        VIR_FORCE_CLOSE(spec.dest.fd.local);
    } else {
3050
        virObjectUnref(sock);
3051 3052
        VIR_FREE(spec.dest.unix_socket.file);
    }
3053

3054
    virObjectUnref(cfg);
3055 3056 3057 3058
    return ret;
}


3059 3060 3061 3062
/* This is essentially a re-impl of virDomainMigrateVersion2
 * from libvirt.c, but running in source libvirtd context,
 * instead of client app context & also adding in tunnel
 * handling */
3063
static int doPeer2PeerMigrate2(virQEMUDriverPtr driver,
3064
                               virConnectPtr sconn ATTRIBUTE_UNUSED,
3065 3066
                               virConnectPtr dconn,
                               virDomainObjPtr vm,
3067
                               const char *dconnuri,
3068 3069 3070
                               unsigned long flags,
                               const char *dname,
                               unsigned long resource)
3071 3072 3073
{
    virDomainPtr ddomain = NULL;
    char *uri_out = NULL;
3074
    char *cookie = NULL;
3075 3076 3077 3078 3079
    char *dom_xml = NULL;
    int cookielen = 0, ret;
    virErrorPtr orig_err = NULL;
    int cancelled;
    virStreamPtr st = NULL;
3080
    VIR_DEBUG("driver=%p, sconn=%p, dconn=%p, vm=%p, dconnuri=%s, "
3081
              "flags=%lx, dname=%s, resource=%lu",
3082 3083
              driver, sconn, dconn, vm, NULLSTR(dconnuri),
              flags, NULLSTR(dname), resource);
3084

3085 3086 3087 3088 3089
    /* In version 2 of the protocol, the prepare step is slightly
     * different.  We fetch the domain XML of the source domain
     * and pass it to Prepare2.
     */
    if (!(dom_xml = qemuDomainFormatXML(driver, vm,
3090 3091
                                        QEMU_DOMAIN_FORMAT_LIVE_FLAGS |
                                        VIR_DOMAIN_XML_MIGRATABLE)))
3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106
        return -1;

    if (virDomainObjGetState(vm, NULL) == VIR_DOMAIN_PAUSED)
        flags |= VIR_MIGRATE_PAUSED;

    VIR_DEBUG("Prepare2 %p", dconn);
    if (flags & VIR_MIGRATE_TUNNELLED) {
        /*
         * Tunnelled Migrate Version 2 does not support cookies
         * due to missing parameters in the prepareTunnel() API.
         */

        if (!(st = virStreamNew(dconn, 0)))
            goto cleanup;

3107
        qemuDomainObjEnterRemote(vm);
3108 3109
        ret = dconn->driver->domainMigratePrepareTunnel
            (dconn, st, flags, dname, resource, dom_xml);
3110
        qemuDomainObjExitRemote(vm);
3111
    } else {
3112
        qemuDomainObjEnterRemote(vm);
3113 3114 3115
        ret = dconn->driver->domainMigratePrepare2
            (dconn, &cookie, &cookielen, NULL, &uri_out,
             flags, dname, resource, dom_xml);
3116
        qemuDomainObjExitRemote(vm);
3117 3118 3119
    }
    VIR_FREE(dom_xml);
    if (ret == -1)
3120 3121 3122
        goto cleanup;

    /* the domain may have shutdown or crashed while we had the locks dropped
3123
     * in qemuDomainObjEnterRemote, so check again
3124 3125
     */
    if (!virDomainObjIsActive(vm)) {
3126 3127
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("guest unexpectedly quit"));
3128 3129 3130
        goto cleanup;
    }

3131 3132
    if (!(flags & VIR_MIGRATE_TUNNELLED) &&
        (uri_out == NULL)) {
3133 3134
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("domainMigratePrepare2 did not set uri"));
3135 3136
        cancelled = 1;
        goto finish;
3137 3138
    }

3139 3140 3141 3142
    /* Perform the migration.  The driver isn't supposed to return
     * until the migration is complete.
     */
    VIR_DEBUG("Perform %p", sconn);
3143
    qemuMigrationJobSetPhase(driver, vm, QEMU_MIGRATION_PHASE_PERFORM2);
3144
    if (flags & VIR_MIGRATE_TUNNELLED)
3145 3146
        ret = doTunnelMigrate(driver, vm, st,
                              NULL, 0, NULL, NULL,
3147
                              flags, resource, dconn);
3148 3149 3150 3151
    else
        ret = doNativeMigrate(driver, vm, uri_out,
                              cookie, cookielen,
                              NULL, NULL, /* No out cookie with v2 migration */
3152
                              flags, resource, dconn);
3153 3154 3155 3156

    /* Perform failed. Make sure Finish doesn't overwrite the error */
    if (ret < 0)
        orig_err = virSaveLastError();
3157

3158 3159 3160 3161
    /* If Perform returns < 0, then we need to cancel the VM
     * startup on the destination
     */
    cancelled = ret < 0 ? 1 : 0;
3162 3163

finish:
3164 3165 3166 3167
    /* In version 2 of the migration protocol, we pass the
     * status code from the sender to the destination host,
     * so it can do any cleanup if the migration failed.
     */
3168
    dname = dname ? dname : vm->def->name;
3169
    VIR_DEBUG("Finish2 %p ret=%d", dconn, ret);
3170
    qemuDomainObjEnterRemote(vm);
3171
    ddomain = dconn->driver->domainMigrateFinish2
3172
        (dconn, dname, cookie, cookielen,
3173
         uri_out ? uri_out : dconnuri, flags, cancelled);
3174
    qemuDomainObjExitRemote(vm);
3175

3176 3177
cleanup:
    if (ddomain) {
3178
        virObjectUnref(ddomain);
3179 3180 3181 3182
        ret = 0;
    } else {
        ret = -1;
    }
3183

3184
    virObjectUnref(st);
3185 3186 3187 3188 3189 3190

    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
    VIR_FREE(uri_out);
3191
    VIR_FREE(cookie);
3192 3193

    return ret;
3194 3195 3196
}


3197 3198 3199 3200
/* This is essentially a re-impl of virDomainMigrateVersion3
 * from libvirt.c, but running in source libvirtd context,
 * instead of client app context & also adding in tunnel
 * handling */
3201
static int doPeer2PeerMigrate3(virQEMUDriverPtr driver,
3202 3203 3204
                               virConnectPtr sconn,
                               virConnectPtr dconn,
                               virDomainObjPtr vm,
3205
                               const char *xmlin,
3206
                               const char *dconnuri,
3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222
                               const char *uri,
                               unsigned long flags,
                               const char *dname,
                               unsigned long resource)
{
    virDomainPtr ddomain = NULL;
    char *uri_out = NULL;
    char *cookiein = NULL;
    char *cookieout = NULL;
    char *dom_xml = NULL;
    int cookieinlen = 0;
    int cookieoutlen = 0;
    int ret = -1;
    virErrorPtr orig_err = NULL;
    int cancelled;
    virStreamPtr st = NULL;
3223
    VIR_DEBUG("driver=%p, sconn=%p, dconn=%p, vm=%p, xmlin=%s, "
3224
              "dconnuri=%s, uri=%s, flags=%lx, dname=%s, resource=%lu",
3225 3226 3227
              driver, sconn, dconn, vm, NULLSTR(xmlin),
              NULLSTR(dconnuri), NULLSTR(uri), flags,
              NULLSTR(dname), resource);
3228

3229 3230 3231 3232 3233
    /* Unlike the virDomainMigrateVersion3 counterpart, we don't need
     * to worry about auto-setting the VIR_MIGRATE_CHANGE_PROTECTION
     * bit here, because we are already running inside the context of
     * a single job.  */

3234
    dom_xml = qemuMigrationBegin(driver, vm, xmlin, dname,
3235
                                 &cookieout, &cookieoutlen, flags);
3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250
    if (!dom_xml)
        goto cleanup;

    if (virDomainObjGetState(vm, NULL) == VIR_DOMAIN_PAUSED)
        flags |= VIR_MIGRATE_PAUSED;

    VIR_DEBUG("Prepare3 %p", dconn);
    cookiein = cookieout;
    cookieinlen = cookieoutlen;
    cookieout = NULL;
    cookieoutlen = 0;
    if (flags & VIR_MIGRATE_TUNNELLED) {
        if (!(st = virStreamNew(dconn, 0)))
            goto cleanup;

3251
        qemuDomainObjEnterRemote(vm);
3252 3253 3254 3255
        ret = dconn->driver->domainMigratePrepareTunnel3
            (dconn, st, cookiein, cookieinlen,
             &cookieout, &cookieoutlen,
             flags, dname, resource, dom_xml);
3256
        qemuDomainObjExitRemote(vm);
3257
    } else {
3258
        qemuDomainObjEnterRemote(vm);
3259 3260
        ret = dconn->driver->domainMigratePrepare3
            (dconn, cookiein, cookieinlen, &cookieout, &cookieoutlen,
3261
             uri, &uri_out, flags, dname, resource, dom_xml);
3262
        qemuDomainObjExitRemote(vm);
3263 3264 3265 3266 3267
    }
    VIR_FREE(dom_xml);
    if (ret == -1)
        goto cleanup;

L
liguang 已提交
3268 3269 3270 3271 3272 3273 3274 3275
    if (flags & VIR_MIGRATE_OFFLINE) {
        VIR_DEBUG("Offline migration, skipping Perform phase");
        VIR_FREE(cookieout);
        cookieoutlen = 0;
        cancelled = 0;
        goto finish;
    }

3276 3277
    if (!(flags & VIR_MIGRATE_TUNNELLED) &&
        (uri_out == NULL)) {
3278 3279
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("domainMigratePrepare3 did not set uri"));
3280 3281 3282 3283 3284 3285 3286 3287 3288
        cancelled = 1;
        goto finish;
    }

    /* Perform the migration.  The driver isn't supposed to return
     * until the migration is complete. The src VM should remain
     * running, but in paused state until the destination can
     * confirm migration completion.
     */
3289
    VIR_DEBUG("Perform3 %p uri=%s uri_out=%s", sconn, uri, uri_out);
3290
    qemuMigrationJobSetPhase(driver, vm, QEMU_MIGRATION_PHASE_PERFORM3);
3291 3292 3293 3294 3295 3296 3297 3298 3299
    VIR_FREE(cookiein);
    cookiein = cookieout;
    cookieinlen = cookieoutlen;
    cookieout = NULL;
    cookieoutlen = 0;
    if (flags & VIR_MIGRATE_TUNNELLED)
        ret = doTunnelMigrate(driver, vm, st,
                              cookiein, cookieinlen,
                              &cookieout, &cookieoutlen,
3300
                              flags, resource, dconn);
3301 3302 3303 3304
    else
        ret = doNativeMigrate(driver, vm, uri_out,
                              cookiein, cookieinlen,
                              &cookieout, &cookieoutlen,
3305
                              flags, resource, dconn);
3306 3307

    /* Perform failed. Make sure Finish doesn't overwrite the error */
3308
    if (ret < 0) {
3309
        orig_err = virSaveLastError();
3310 3311 3312 3313
    } else {
        qemuMigrationJobSetPhase(driver, vm,
                                 QEMU_MIGRATION_PHASE_PERFORM3_DONE);
    }
3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333

    /* If Perform returns < 0, then we need to cancel the VM
     * startup on the destination
     */
    cancelled = ret < 0 ? 1 : 0;

finish:
    /*
     * The status code from the source is passed to the destination.
     * The dest can cleanup in the source indicated it failed to
     * send all migration data. Returns NULL for ddomain if
     * the dest was unable to complete migration.
     */
    VIR_DEBUG("Finish3 %p ret=%d", dconn, ret);
    VIR_FREE(cookiein);
    cookiein = cookieout;
    cookieinlen = cookieoutlen;
    cookieout = NULL;
    cookieoutlen = 0;
    dname = dname ? dname : vm->def->name;
3334
    qemuDomainObjEnterRemote(vm);
3335
    ddomain = dconn->driver->domainMigrateFinish3
3336
        (dconn, dname, cookiein, cookieinlen, &cookieout, &cookieoutlen,
3337
         dconnuri, uri_out ? uri_out : uri, flags, cancelled);
3338
    qemuDomainObjExitRemote(vm);
3339

3340 3341 3342 3343 3344 3345 3346
    /* If ddomain is NULL, then we were unable to start
     * the guest on the target, and must restart on the
     * source. There is a small chance that the ddomain
     * is NULL due to an RPC failure, in which case
     * ddomain could in fact be running on the dest.
     * The lock manager plugins should take care of
     * safety in this scenario.
3347
     */
3348
    cancelled = ddomain == NULL ? 1 : 0;
3349

3350 3351 3352 3353 3354 3355
    /* If finish3 set an error, and we don't have an earlier
     * one we need to preserve it in case confirm3 overwrites
     */
    if (!orig_err)
        orig_err = virSaveLastError();

3356 3357 3358 3359
    /*
     * If cancelled, then src VM will be restarted, else
     * it will be killed
     */
3360
    VIR_DEBUG("Confirm3 %p cancelled=%d vm=%p", sconn, cancelled, vm);
3361 3362 3363 3364 3365 3366 3367
    VIR_FREE(cookiein);
    cookiein = cookieout;
    cookieinlen = cookieoutlen;
    cookieout = NULL;
    cookieoutlen = 0;
    ret = qemuMigrationConfirm(driver, sconn, vm,
                               cookiein, cookieinlen,
3368
                               flags, cancelled);
3369 3370 3371 3372
    /* If Confirm3 returns -1, there's nothing more we can
     * do, but fortunately worst case is that there is a
     * domain left in 'paused' state on source.
     */
3373 3374 3375
    if (ret < 0)
        VIR_WARN("Guest %s probably left in 'paused' state on source",
                 vm->def->name);
3376 3377 3378

 cleanup:
    if (ddomain) {
3379
        virObjectUnref(ddomain);
3380 3381 3382 3383 3384
        ret = 0;
    } else {
        ret = -1;
    }

3385
    virObjectUnref(st);
3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398

    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
    VIR_FREE(uri_out);
    VIR_FREE(cookiein);
    VIR_FREE(cookieout);

    return ret;
}


3399
static int doPeer2PeerMigrate(virQEMUDriverPtr driver,
3400
                              virConnectPtr sconn,
3401
                              virDomainObjPtr vm,
3402
                              const char *xmlin,
3403
                              const char *dconnuri,
3404 3405 3406
                              const char *uri,
                              unsigned long flags,
                              const char *dname,
3407 3408
                              unsigned long resource,
                              bool *v3proto)
3409 3410 3411 3412
{
    int ret = -1;
    virConnectPtr dconn = NULL;
    bool p2p;
3413
    virErrorPtr orig_err = NULL;
3414
    bool offline = false;
3415
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
3416

3417
    VIR_DEBUG("driver=%p, sconn=%p, vm=%p, xmlin=%s, dconnuri=%s, "
3418
              "uri=%s, flags=%lx, dname=%s, resource=%lu",
3419 3420
              driver, sconn, vm, NULLSTR(xmlin), NULLSTR(dconnuri),
              NULLSTR(uri), flags, NULLSTR(dname), resource);
3421 3422 3423 3424 3425

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

3426
    qemuDomainObjEnterRemote(vm);
3427
    dconn = virConnectOpen(dconnuri);
3428
    qemuDomainObjExitRemote(vm);
3429
    if (dconn == NULL) {
3430 3431
        virReportError(VIR_ERR_OPERATION_FAILED,
                       _("Failed to connect to remote libvirt URI %s"), dconnuri);
3432
        virObjectUnref(cfg);
3433 3434 3435
        return -1;
    }

3436 3437
    if (virConnectSetKeepAlive(dconn, cfg->keepAliveInterval,
                               cfg->keepAliveCount) < 0)
3438 3439
        goto cleanup;

3440
    qemuDomainObjEnterRemote(vm);
3441 3442
    p2p = VIR_DRV_SUPPORTS_FEATURE(dconn->driver, dconn,
                                   VIR_DRV_FEATURE_MIGRATION_P2P);
3443
        /* v3proto reflects whether the caller used Perform3, but with
3444
         * p2p migrate, regardless of whether Perform2 or Perform3
3445 3446 3447 3448
         * were used, we decide protocol based on what target supports
         */
    *v3proto = VIR_DRV_SUPPORTS_FEATURE(dconn->driver, dconn,
                                        VIR_DRV_FEATURE_MIGRATION_V3);
L
liguang 已提交
3449 3450 3451
    if (flags & VIR_MIGRATE_OFFLINE)
        offline = VIR_DRV_SUPPORTS_FEATURE(dconn->driver, dconn,
                                           VIR_DRV_FEATURE_MIGRATION_OFFLINE);
3452
    qemuDomainObjExitRemote(vm);
3453

3454
    if (!p2p) {
3455 3456
        virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                       _("Destination libvirt does not support peer-to-peer migration protocol"));
3457 3458 3459
        goto cleanup;
    }

L
liguang 已提交
3460 3461 3462 3463 3464 3465 3466
    if (flags & VIR_MIGRATE_OFFLINE && !offline) {
        virReportError(VIR_ERR_ARGUMENT_UNSUPPORTED, "%s",
                       _("offline migration is not supported by "
                         "the destination host"));
        goto cleanup;
    }

3467
    /* domain may have been stopped while we were talking to remote daemon */
L
liguang 已提交
3468
    if (!virDomainObjIsActive(vm) && !(flags & VIR_MIGRATE_OFFLINE)) {
3469 3470
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("guest unexpectedly quit"));
3471 3472 3473
        goto cleanup;
    }

3474 3475 3476 3477 3478 3479 3480
    /* Change protection is only required on the source side (us), and
     * only for v3 migration when begin and perform are separate jobs.
     * But peer-2-peer is already a single job, and we still want to
     * talk to older destinations that would reject the flag.
     * Therefore it is safe to clear the bit here.  */
    flags &= ~VIR_MIGRATE_CHANGE_PROTECTION;

3481
    if (*v3proto)
3482
        ret = doPeer2PeerMigrate3(driver, sconn, dconn, vm, xmlin,
3483
                                  dconnuri, uri, flags, dname, resource);
3484 3485
    else
        ret = doPeer2PeerMigrate2(driver, sconn, dconn, vm,
3486
                                  dconnuri, flags, dname, resource);
3487 3488

cleanup:
3489
    orig_err = virSaveLastError();
3490
    qemuDomainObjEnterRemote(vm);
3491
    virConnectClose(dconn);
3492
    qemuDomainObjExitRemote(vm);
3493 3494 3495 3496
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
3497
    virObjectUnref(cfg);
3498 3499 3500 3501
    return ret;
}


3502 3503 3504 3505 3506 3507
/*
 * This implements perform part of the migration protocol when migration job
 * does not need to be active across several APIs, i.e., peer2peer migration or
 * perform phase of v2 non-peer2peer migration.
 */
static int
3508
qemuMigrationPerformJob(virQEMUDriverPtr driver,
3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521
                        virConnectPtr conn,
                        virDomainObjPtr vm,
                        const char *xmlin,
                        const char *dconnuri,
                        const char *uri,
                        const char *cookiein,
                        int cookieinlen,
                        char **cookieout,
                        int *cookieoutlen,
                        unsigned long flags,
                        const char *dname,
                        unsigned long resource,
                        bool v3proto)
3522 3523 3524 3525
{
    virDomainEventPtr event = NULL;
    int ret = -1;
    int resume = 0;
3526
    virErrorPtr orig_err = NULL;
3527
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
3528

3529
    if (qemuMigrationJobStart(driver, vm, QEMU_ASYNC_JOB_MIGRATION_OUT) < 0)
3530 3531
        goto cleanup;

L
liguang 已提交
3532
    if (!virDomainObjIsActive(vm) && !(flags & VIR_MIGRATE_OFFLINE)) {
3533 3534
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("domain is not running"));
3535 3536 3537
        goto endjob;
    }

3538
    if (!qemuMigrationIsAllowed(driver, vm, NULL, true))
3539
        goto endjob;
3540

3541
    if (!(flags & VIR_MIGRATE_UNSAFE) && !qemuMigrationIsSafe(vm->def))
3542
        goto endjob;
3543

J
Jiri Denemark 已提交
3544
    resume = virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING;
3545 3546

    if ((flags & (VIR_MIGRATE_TUNNELLED | VIR_MIGRATE_PEER2PEER))) {
3547 3548 3549
        ret = doPeer2PeerMigrate(driver, conn, vm, xmlin,
                                 dconnuri, uri, flags, dname,
                                 resource, &v3proto);
3550
    } else {
3551 3552 3553
        qemuMigrationJobSetPhase(driver, vm, QEMU_MIGRATION_PHASE_PERFORM2);
        ret = doNativeMigrate(driver, vm, uri, cookiein, cookieinlen,
                              cookieout, cookieoutlen,
3554
                              flags, resource, NULL);
3555
    }
3556 3557
    if (ret < 0)
        goto endjob;
3558

3559 3560 3561 3562
    /*
     * In v3 protocol, the source VM is not killed off until the
     * confirm step.
     */
3563
    if (!v3proto) {
3564 3565
        qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_MIGRATED,
                        VIR_QEMU_PROCESS_STOP_MIGRATED);
3566
        virDomainAuditStop(vm, "migrated");
3567 3568 3569
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         VIR_DOMAIN_EVENT_STOPPED_MIGRATED);
3570
    }
3571
    resume = 0;
3572 3573

endjob:
3574 3575 3576
    if (ret < 0)
        orig_err = virSaveLastError();

J
Jiri Denemark 已提交
3577
    if (resume && virDomainObjGetState(vm, NULL) == VIR_DOMAIN_PAUSED) {
3578
        /* we got here through some sort of failure; start the domain again */
J
Jiri Denemark 已提交
3579
        if (qemuProcessStartCPUs(driver, vm, conn,
3580 3581
                                 VIR_DOMAIN_RUNNING_MIGRATION_CANCELED,
                                 QEMU_ASYNC_JOB_MIGRATION_OUT) < 0) {
3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593
            /* 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);
    }
3594

3595
    if (!qemuMigrationJobFinish(driver, vm)) {
3596 3597 3598 3599 3600
        vm = NULL;
    } else if (!virDomainObjIsActive(vm) &&
               (!vm->persistent ||
                (ret == 0 && (flags & VIR_MIGRATE_UNDEFINE_SOURCE)))) {
        if (flags & VIR_MIGRATE_UNDEFINE_SOURCE)
3601
            virDomainDeleteConfig(cfg->configDir, cfg->autostartDir, vm);
3602
        qemuDomainRemoveInactive(driver, vm);
3603 3604 3605
        vm = NULL;
    }

3606 3607 3608 3609 3610
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }

3611 3612
cleanup:
    if (vm)
3613
        virObjectUnlock(vm);
3614 3615
    if (event)
        qemuDomainEventQueue(driver, event);
3616
    virObjectUnref(cfg);
3617 3618 3619 3620 3621 3622 3623
    return ret;
}

/*
 * This implements perform phase of v3 migration protocol.
 */
static int
3624
qemuMigrationPerformPhase(virQEMUDriverPtr driver,
3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637
                          virConnectPtr conn,
                          virDomainObjPtr vm,
                          const char *uri,
                          const char *cookiein,
                          int cookieinlen,
                          char **cookieout,
                          int *cookieoutlen,
                          unsigned long flags,
                          unsigned long resource)
{
    virDomainEventPtr event = NULL;
    int ret = -1;
    bool resume;
3638
    bool hasrefs;
3639 3640 3641 3642 3643 3644 3645 3646 3647 3648

    /* If we didn't start the job in the begin phase, start it now. */
    if (!(flags & VIR_MIGRATE_CHANGE_PROTECTION)) {
        if (qemuMigrationJobStart(driver, vm, QEMU_ASYNC_JOB_MIGRATION_OUT) < 0)
            goto cleanup;
    } else if (!qemuMigrationJobIsActive(vm, QEMU_ASYNC_JOB_MIGRATION_OUT)) {
        goto cleanup;
    }

    qemuMigrationJobStartPhase(driver, vm, QEMU_MIGRATION_PHASE_PERFORM3);
3649 3650
    virQEMUCloseCallbacksUnset(driver->closeCallbacks, vm,
                               qemuMigrationCleanup);
3651 3652 3653 3654

    resume = virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING;
    ret = doNativeMigrate(driver, vm, uri, cookiein, cookieinlen,
                          cookieout, cookieoutlen,
3655
                          flags, resource, NULL);
3656 3657 3658 3659 3660

    if (ret < 0 && resume &&
        virDomainObjGetState(vm, NULL) == VIR_DOMAIN_PAUSED) {
        /* we got here through some sort of failure; start the domain again */
        if (qemuProcessStartCPUs(driver, vm, conn,
3661 3662
                                 VIR_DOMAIN_RUNNING_MIGRATION_CANCELED,
                                 QEMU_ASYNC_JOB_MIGRATION_OUT) < 0) {
3663 3664 3665 3666 3667 3668
            /* 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);
3669
        }
3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680

        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_MIGRATED);
    }

    if (ret < 0)
        goto endjob;

    qemuMigrationJobSetPhase(driver, vm, QEMU_MIGRATION_PHASE_PERFORM3_DONE);

3681 3682
    if (virQEMUCloseCallbacksSet(driver->closeCallbacks, vm, conn,
                                 qemuMigrationCleanup) < 0)
3683 3684
        goto endjob;

3685 3686
endjob:
    if (ret < 0)
3687
        hasrefs = qemuMigrationJobFinish(driver, vm);
3688
    else
3689 3690
        hasrefs = qemuMigrationJobContinue(vm);
    if (!hasrefs) {
3691 3692
        vm = NULL;
    } else if (!virDomainObjIsActive(vm) && !vm->persistent) {
3693
        qemuDomainRemoveInactive(driver, vm);
3694
        vm = NULL;
3695
    }
3696 3697 3698

cleanup:
    if (vm)
3699
        virObjectUnlock(vm);
3700 3701 3702 3703 3704
    if (event)
        qemuDomainEventQueue(driver, event);
    return ret;
}

3705
int
3706
qemuMigrationPerform(virQEMUDriverPtr driver,
3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730
                     virConnectPtr conn,
                     virDomainObjPtr vm,
                     const char *xmlin,
                     const char *dconnuri,
                     const char *uri,
                     const char *cookiein,
                     int cookieinlen,
                     char **cookieout,
                     int *cookieoutlen,
                     unsigned long flags,
                     const char *dname,
                     unsigned long resource,
                     bool v3proto)
{
    VIR_DEBUG("driver=%p, conn=%p, vm=%p, xmlin=%s, dconnuri=%s, "
              "uri=%s, cookiein=%s, cookieinlen=%d, cookieout=%p, "
              "cookieoutlen=%p, flags=%lx, dname=%s, resource=%lu, v3proto=%d",
              driver, conn, vm, NULLSTR(xmlin), NULLSTR(dconnuri),
              NULLSTR(uri), NULLSTR(cookiein), cookieinlen,
              cookieout, cookieoutlen, flags, NULLSTR(dname),
              resource, v3proto);

    if ((flags & (VIR_MIGRATE_TUNNELLED | VIR_MIGRATE_PEER2PEER))) {
        if (cookieinlen) {
3731 3732
            virReportError(VIR_ERR_OPERATION_INVALID,
                           "%s", _("received unexpected cookie with P2P migration"));
3733 3734 3735 3736 3737 3738 3739 3740 3741
            return -1;
        }

        return qemuMigrationPerformJob(driver, conn, vm, xmlin, dconnuri, uri,
                                       cookiein, cookieinlen, cookieout,
                                       cookieoutlen, flags, dname, resource,
                                       v3proto);
    } else {
        if (dconnuri) {
3742 3743
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("Unexpected dconnuri parameter with non-peer2peer migration"));
3744 3745 3746 3747 3748 3749 3750
            return -1;
        }

        if (v3proto) {
            return qemuMigrationPerformPhase(driver, conn, vm, uri,
                                             cookiein, cookieinlen,
                                             cookieout, cookieoutlen,
3751
                                             flags, resource);
3752 3753 3754 3755 3756 3757 3758 3759
        } else {
            return qemuMigrationPerformJob(driver, conn, vm, xmlin, dconnuri,
                                           uri, cookiein, cookieinlen,
                                           cookieout, cookieoutlen, flags,
                                           dname, resource, v3proto);
        }
    }
}
3760

3761
static int
3762 3763 3764 3765 3766 3767 3768
qemuMigrationVPAssociatePortProfiles(virDomainDefPtr def) {
    int i;
    int last_good_net = -1;
    virDomainNetDefPtr net;

    for (i = 0; i < def->nnets; i++) {
        net = def->nets[i];
3769
        if (virDomainNetGetActualType(net) == VIR_DOMAIN_NET_TYPE_DIRECT) {
3770
            if (virNetDevVPortProfileAssociate(net->ifname,
3771
                                               virDomainNetGetActualVirtPortProfile(net),
3772
                                               &net->mac,
3773
                                               virDomainNetGetActualDirectDev(net),
3774
                                               -1,
3775
                                               def->uuid,
3776 3777
                                               VIR_NETDEV_VPORT_PROFILE_OP_MIGRATE_IN_FINISH,
                                               false) < 0) {
3778 3779 3780
                virReportError(VIR_ERR_OPERATION_FAILED,
                               _("Port profile Associate failed for %s"),
                               net->ifname);
3781
                goto err_exit;
3782 3783
            }
            VIR_DEBUG("Port profile Associate succeeded for %s", net->ifname);
3784

3785
            if (virNetDevMacVLanVPortProfileRegisterCallback(net->ifname, &net->mac,
3786 3787 3788 3789
                                                             virDomainNetGetActualDirectDev(net), def->uuid,
                                                             virDomainNetGetActualVirtPortProfile(net),
                                                             VIR_NETDEV_VPORT_PROFILE_OP_CREATE))
                goto err_exit;
3790 3791 3792 3793
        }
        last_good_net = i;
    }

3794
    return 0;
3795 3796 3797 3798

err_exit:
    for (i = 0; i < last_good_net; i++) {
        net = def->nets[i];
3799
        if (virDomainNetGetActualType(net) == VIR_DOMAIN_NET_TYPE_DIRECT) {
3800
            ignore_value(virNetDevVPortProfileDisassociate(net->ifname,
3801
                                                           virDomainNetGetActualVirtPortProfile(net),
3802
                                                           &net->mac,
3803
                                                           virDomainNetGetActualDirectDev(net),
3804
                                                           -1,
3805
                                                           VIR_NETDEV_VPORT_PROFILE_OP_MIGRATE_IN_FINISH));
3806 3807
        }
    }
3808
    return -1;
3809 3810 3811 3812
}


virDomainPtr
3813
qemuMigrationFinish(virQEMUDriverPtr driver,
3814 3815
                    virConnectPtr dconn,
                    virDomainObjPtr vm,
3816 3817 3818 3819
                    const char *cookiein,
                    int cookieinlen,
                    char **cookieout,
                    int *cookieoutlen,
3820
                    unsigned long flags,
3821 3822
                    int retcode,
                    bool v3proto)
3823 3824 3825 3826
{
    virDomainPtr dom = NULL;
    virDomainEventPtr event = NULL;
    int newVM = 1;
3827
    qemuMigrationCookiePtr mig = NULL;
3828
    virErrorPtr orig_err = NULL;
3829
    int cookie_flags = 0;
J
Jiri Denemark 已提交
3830
    qemuDomainObjPrivatePtr priv = vm->privateData;
3831
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
3832
    virCapsPtr caps = NULL;
3833

3834
    VIR_DEBUG("driver=%p, dconn=%p, vm=%p, cookiein=%s, cookieinlen=%d, "
3835
              "cookieout=%p, cookieoutlen=%p, flags=%lx, retcode=%d",
3836 3837
              driver, dconn, vm, NULLSTR(cookiein), cookieinlen,
              cookieout, cookieoutlen, flags, retcode);
3838

3839 3840 3841
    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
        goto cleanup;

3842
    if (!qemuMigrationJobIsActive(vm, QEMU_ASYNC_JOB_MIGRATION_IN))
3843 3844
        goto cleanup;

3845 3846 3847
    qemuMigrationJobStartPhase(driver, vm,
                               v3proto ? QEMU_MIGRATION_PHASE_FINISH3
                                       : QEMU_MIGRATION_PHASE_FINISH2);
3848

3849 3850
    qemuDomainCleanupRemove(vm, qemuMigrationPrepareCleanup);

3851
    cookie_flags = QEMU_MIGRATION_COOKIE_NETWORK;
3852 3853 3854 3855 3856
    if (flags & VIR_MIGRATE_PERSIST_DEST)
        cookie_flags |= QEMU_MIGRATION_COOKIE_PERSISTENT;

    if (!(mig = qemuMigrationEatCookie(driver, vm, cookiein,
                                       cookieinlen, cookie_flags)))
3857
        goto endjob;
3858 3859 3860 3861 3862

    /* Did the migration go as planned?  If yes, return the domain
     * object, but if no, clean up the empty qemu process.
     */
    if (retcode == 0) {
L
liguang 已提交
3863
        if (!virDomainObjIsActive(vm) && !(flags & VIR_MIGRATE_OFFLINE)) {
3864 3865
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("guest unexpectedly quit"));
3866
            goto endjob;
3867 3868
        }

L
liguang 已提交
3869 3870 3871 3872 3873 3874 3875 3876 3877 3878 3879 3880 3881
        if (!(flags & VIR_MIGRATE_OFFLINE)) {
            if (qemuMigrationVPAssociatePortProfiles(vm->def) < 0) {
                qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_FAILED,
                                VIR_QEMU_PROCESS_STOP_MIGRATED);
                virDomainAuditStop(vm, "failed");
                event = virDomainEventNewFromObj(vm,
                                                 VIR_DOMAIN_EVENT_STOPPED,
                                                 VIR_DOMAIN_EVENT_STOPPED_FAILED);
                goto endjob;
            }
            if (mig->network)
                if (qemuDomainMigrateOPDRelocate(driver, vm, mig) < 0)
                    VIR_WARN("unable to provide network data for relocation");
3882
        }
3883

3884 3885
        qemuMigrationStopNBDServer(driver, vm, mig);

3886
        if (flags & VIR_MIGRATE_PERSIST_DEST) {
3887
            virDomainDefPtr vmdef;
3888 3889 3890
            if (vm->persistent)
                newVM = 0;
            vm->persistent = 1;
3891 3892 3893
            if (mig->persistent)
                vm->newDef = vmdef = mig->persistent;
            else
3894
                vmdef = virDomainObjGetPersistentDef(caps, driver->xmlconf, vm);
3895
            if (!vmdef || virDomainSaveConfig(cfg->configDir, vmdef) < 0) {
3896 3897 3898 3899 3900 3901 3902 3903 3904
                /* 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.
                 */
3905 3906

                /*
3907 3908
                 * However, in v3 protocol, the source VM is still available
                 * to restart during confirm() step, so we kill it off now.
3909 3910
                 */
                if (v3proto) {
L
liguang 已提交
3911 3912 3913 3914 3915
                    if (!(flags & VIR_MIGRATE_OFFLINE)) {
                        qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_FAILED,
                                        VIR_QEMU_PROCESS_STOP_MIGRATED);
                        virDomainAuditStop(vm, "failed");
                    }
3916 3917
                    if (newVM)
                        vm->persistent = 0;
3918
                }
A
Alex Jia 已提交
3919
                if (!vmdef)
3920 3921
                    virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                                   _("can't get vmdef"));
3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934
                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;
        }

L
liguang 已提交
3935
        if (!(flags & VIR_MIGRATE_PAUSED) && !(flags & VIR_MIGRATE_OFFLINE)) {
3936 3937 3938 3939
            /* 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
3940
            if (qemuProcessStartCPUs(driver, vm, dconn,
3941 3942
                                     VIR_DOMAIN_RUNNING_MIGRATED,
                                     QEMU_ASYNC_JOB_MIGRATION_IN) < 0) {
3943
                if (virGetLastError() == NULL)
3944 3945
                    virReportError(VIR_ERR_INTERNAL_ERROR,
                                   "%s", _("resume operation failed"));
3946 3947 3948 3949 3950
                /* Need to save the current error, in case shutting
                 * down the process overwrites it
                 */
                orig_err = virSaveLastError();

3951 3952 3953 3954 3955 3956 3957 3958 3959
                /*
                 * In v3 protocol, the source VM is still available to
                 * restart during confirm() step, so we kill it off
                 * now.
                 * In v2 protocol, the source is dead, so we leave
                 * target in paused state, in case admin can fix
                 * things up
                 */
                if (v3proto) {
3960 3961
                    qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_FAILED,
                                    VIR_QEMU_PROCESS_STOP_MIGRATED);
3962
                    virDomainAuditStop(vm, "failed");
3963 3964 3965 3966
                    event = virDomainEventNewFromObj(vm,
                                                     VIR_DOMAIN_EVENT_STOPPED,
                                                     VIR_DOMAIN_EVENT_STOPPED_FAILED);
                }
3967 3968 3969 3970
                goto endjob;
            }
        }

3971
        dom = virGetDomain(dconn, vm->def->name, vm->def->uuid);
3972

L
liguang 已提交
3973
        if (!(flags & VIR_MIGRATE_OFFLINE)) {
3974
            event = virDomainEventNewFromObj(vm,
L
liguang 已提交
3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985
                                             VIR_DOMAIN_EVENT_RESUMED,
                                             VIR_DOMAIN_EVENT_RESUMED_MIGRATED);
            if (virDomainObjGetState(vm, NULL) == VIR_DOMAIN_PAUSED) {
                virDomainObjSetState(vm, VIR_DOMAIN_PAUSED,
                                     VIR_DOMAIN_PAUSED_USER);
                if (event)
                    qemuDomainEventQueue(driver, event);
                event = virDomainEventNewFromObj(vm,
                                                 VIR_DOMAIN_EVENT_SUSPENDED,
                                                 VIR_DOMAIN_EVENT_SUSPENDED_PAUSED);
            }
3986
        }
L
liguang 已提交
3987 3988

        if (virDomainObjIsActive(vm) &&
3989
            virDomainSaveStatus(driver->xmlconf, cfg->stateDir, vm) < 0) {
3990 3991 3992
            VIR_WARN("Failed to save status on vm %s", vm->def->name);
            goto endjob;
        }
3993 3994 3995

        /* Guest is successfully running, so cancel previous auto destroy */
        qemuProcessAutoDestroyRemove(driver, vm);
L
liguang 已提交
3996
    } else if (!(flags & VIR_MIGRATE_OFFLINE)) {
3997 3998
        qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_FAILED,
                        VIR_QEMU_PROCESS_STOP_MIGRATED);
3999
        virDomainAuditStop(vm, "failed");
4000 4001 4002 4003 4004
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         VIR_DOMAIN_EVENT_STOPPED_FAILED);
    }

4005 4006 4007
    if (qemuMigrationBakeCookie(mig, driver, vm, cookieout, cookieoutlen, 0) < 0)
        VIR_WARN("Unable to encode migration cookie");

4008
endjob:
E
Eric Blake 已提交
4009 4010 4011
    if (qemuMigrationJobFinish(driver, vm) == 0) {
        vm = NULL;
    } else if (!vm->persistent && !virDomainObjIsActive(vm)) {
4012
        qemuDomainRemoveInactive(driver, vm);
E
Eric Blake 已提交
4013
        vm = NULL;
4014
    }
4015 4016

cleanup:
J
Jiri Denemark 已提交
4017 4018
    if (vm) {
        VIR_FREE(priv->origname);
4019
        virObjectUnlock(vm);
J
Jiri Denemark 已提交
4020
    }
4021 4022
    if (event)
        qemuDomainEventQueue(driver, event);
4023
    qemuMigrationCookieFree(mig);
4024 4025 4026 4027
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
4028
    virObjectUnref(caps);
4029
    virObjectUnref(cfg);
4030 4031
    return dom;
}
4032

4033

4034
int qemuMigrationConfirm(virQEMUDriverPtr driver,
4035 4036 4037 4038
                         virConnectPtr conn,
                         virDomainObjPtr vm,
                         const char *cookiein,
                         int cookieinlen,
E
Eric Blake 已提交
4039
                         unsigned int flags,
4040
                         int retcode)
4041 4042 4043 4044
{
    qemuMigrationCookiePtr mig;
    virDomainEventPtr event = NULL;
    int rv = -1;
4045 4046
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);

4047
    VIR_DEBUG("driver=%p, conn=%p, vm=%p, cookiein=%s, cookieinlen=%d, "
4048
              "flags=%x, retcode=%d",
4049 4050
              driver, conn, vm, NULLSTR(cookiein), cookieinlen,
              flags, retcode);
4051

4052
    virCheckFlags(QEMU_MIGRATION_FLAGS, -1);
E
Eric Blake 已提交
4053

4054 4055 4056 4057 4058
    qemuMigrationJobSetPhase(driver, vm,
                             retcode == 0
                             ? QEMU_MIGRATION_PHASE_CONFIRM3
                             : QEMU_MIGRATION_PHASE_CONFIRM3_CANCELLED);

4059
    if (!(mig = qemuMigrationEatCookie(driver, vm, cookiein, cookieinlen, 0)))
4060
        goto cleanup;
4061

L
liguang 已提交
4062 4063 4064
    if (flags & VIR_MIGRATE_OFFLINE)
        goto done;

4065 4066 4067 4068
    /* Did the migration go as planned?  If yes, kill off the
     * domain object, but if no, resume CPUs
     */
    if (retcode == 0) {
4069 4070
        qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_MIGRATED,
                        VIR_QEMU_PROCESS_STOP_MIGRATED);
4071
        virDomainAuditStop(vm, "migrated");
4072 4073 4074 4075 4076 4077

        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         VIR_DOMAIN_EVENT_STOPPED_MIGRATED);
    } else {

4078 4079 4080
        /* cancel any outstanding NBD jobs */
        qemuMigrationCancelDriveMirror(mig, driver, vm);

4081 4082 4083 4084 4085
        /* 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
         */
        if (qemuProcessStartCPUs(driver, vm, conn,
4086 4087
                                 VIR_DOMAIN_RUNNING_MIGRATED,
                                 QEMU_ASYNC_JOB_MIGRATION_OUT) < 0) {
4088
            if (virGetLastError() == NULL)
4089 4090
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               "%s", _("resume operation failed"));
4091
            goto cleanup;
4092 4093 4094 4095 4096
        }

        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_MIGRATED);
4097
        if (virDomainSaveStatus(driver->xmlconf, cfg->stateDir, vm) < 0) {
4098
            VIR_WARN("Failed to save status on vm %s", vm->def->name);
4099
            goto cleanup;
4100 4101 4102
        }
    }

L
liguang 已提交
4103
done:
4104 4105 4106 4107 4108 4109
    qemuMigrationCookieFree(mig);
    rv = 0;

cleanup:
    if (event)
        qemuDomainEventQueue(driver, event);
4110
    virObjectUnref(cfg);
4111 4112 4113 4114
    return rv;
}


4115
/* Helper function called while vm is active.  */
4116
int
4117
qemuMigrationToFile(virQEMUDriverPtr driver, virDomainObjPtr vm,
4118 4119
                    int fd, off_t offset, const char *path,
                    const char *compressor,
E
Eric Blake 已提交
4120
                    bool bypassSecurityDriver,
4121
                    enum qemuDomainAsyncJob asyncJob)
4122 4123 4124 4125 4126 4127
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virCgroupPtr cgroup = NULL;
    int ret = -1;
    int rc;
    bool restoreLabel = false;
4128 4129
    virCommandPtr cmd = NULL;
    int pipeFD[2] = { -1, -1 };
4130
    unsigned long saveMigBandwidth = priv->migMaxBandwidth;
4131
    char *errbuf = NULL;
4132 4133 4134 4135 4136

    /* Increase migration bandwidth to unlimited since target is a file.
     * Failure to change migration speed is not fatal. */
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) == 0) {
        qemuMonitorSetMigrationSpeed(priv->mon,
4137 4138
                                     QEMU_DOMAIN_MIG_BANDWIDTH_MAX);
        priv->migMaxBandwidth = QEMU_DOMAIN_MIG_BANDWIDTH_MAX;
4139
        qemuDomainObjExitMonitor(driver, vm);
4140
    }
4141

4142
    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_FD) &&
4143
        (!compressor || pipe(pipeFD) == 0)) {
4144
        /* All right! We can use fd migration, which means that qemu
4145 4146 4147
         * 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.  */
4148
        if (virSecurityManagerSetImageFDLabel(driver->securityManager, vm->def,
4149
                                              compressor ? pipeFD[1] : fd) < 0)
4150
            goto cleanup;
4151 4152 4153
        bypassSecurityDriver = true;
    } else {
        /* Phooey - we have to fall back on exec migration, where qemu
E
Eric Blake 已提交
4154 4155
         * has to popen() the file by name, and block devices have to be
         * given cgroup ACL permission.  We might also stumble on
4156 4157
         * a race present in some qemu versions where it does a wait()
         * that botches pclose.  */
E
Eric Blake 已提交
4158
        if (qemuCgroupControllerActive(driver,
4159 4160 4161
                                       VIR_CGROUP_CONTROLLER_DEVICES)) {
            if (virCgroupForDomain(driver->cgroup, vm->def->name,
                                   &cgroup, 0) != 0) {
4162 4163 4164
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("Unable to find cgroup for %s"),
                               vm->def->name);
4165 4166 4167 4168
                goto cleanup;
            }
            rc = virCgroupAllowDevicePath(cgroup, path,
                                          VIR_CGROUP_DEVICE_RW);
4169
            virDomainAuditCgroupPath(vm, cgroup, "allow", path, "rw", rc);
E
Eric Blake 已提交
4170 4171 4172 4173
            if (rc == 1) {
                /* path was not a device, no further need for cgroup */
                virCgroupFree(&cgroup);
            } else if (rc < 0) {
4174 4175 4176 4177 4178
                virReportSystemError(-rc,
                                     _("Unable to allow device %s for %s"),
                                     path, vm->def->name);
                goto cleanup;
            }
4179
        }
4180 4181
        if ((!bypassSecurityDriver) &&
            virSecurityManagerSetSavedStateLabel(driver->securityManager,
4182
                                                 vm->def, path) < 0)
4183
            goto cleanup;
4184
        restoreLabel = true;
4185 4186
    }

4187
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
4188 4189
        goto cleanup;

4190 4191 4192
    if (!compressor) {
        const char *args[] = { "cat", NULL };

4193
        if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_FD) &&
4194 4195 4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209
            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
        };
4210 4211 4212 4213
        if (pipeFD[0] != -1) {
            cmd = virCommandNewArgs(args);
            virCommandSetInputFD(cmd, pipeFD[0]);
            virCommandSetOutputFD(cmd, &fd);
4214 4215
            virCommandSetErrorBuffer(cmd, &errbuf);
            virCommandDoAsyncIO(cmd);
4216 4217 4218
            if (virSetCloseExec(pipeFD[1]) < 0) {
                virReportSystemError(errno, "%s",
                                     _("Unable to set cloexec flag"));
4219
                qemuDomainObjExitMonitor(driver, vm);
4220 4221 4222
                goto cleanup;
            }
            if (virCommandRunAsync(cmd, NULL) < 0) {
4223
                qemuDomainObjExitMonitor(driver, vm);
4224 4225 4226 4227 4228 4229 4230
                goto cleanup;
            }
            rc = qemuMonitorMigrateToFd(priv->mon,
                                        QEMU_MONITOR_MIGRATE_BACKGROUND,
                                        pipeFD[1]);
            if (VIR_CLOSE(pipeFD[0]) < 0 ||
                VIR_CLOSE(pipeFD[1]) < 0)
4231
                VIR_WARN("failed to close intermediate pipe");
4232 4233 4234 4235 4236
        } else {
            rc = qemuMonitorMigrateToFile(priv->mon,
                                          QEMU_MONITOR_MIGRATE_BACKGROUND,
                                          args, path, offset);
        }
4237
    }
4238
    qemuDomainObjExitMonitor(driver, vm);
4239 4240 4241 4242

    if (rc < 0)
        goto cleanup;

4243
    rc = qemuMigrationWaitForCompletion(driver, vm, asyncJob, NULL);
4244 4245 4246 4247

    if (rc < 0)
        goto cleanup;

4248 4249 4250
    if (cmd && virCommandWait(cmd, NULL) < 0)
        goto cleanup;

4251 4252 4253
    ret = 0;

cleanup:
4254 4255 4256 4257
    /* Restore max migration bandwidth */
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) == 0) {
        qemuMonitorSetMigrationSpeed(priv->mon, saveMigBandwidth);
        priv->migMaxBandwidth = saveMigBandwidth;
4258
        qemuDomainObjExitMonitor(driver, vm);
4259 4260
    }

4261 4262
    VIR_FORCE_CLOSE(pipeFD[0]);
    VIR_FORCE_CLOSE(pipeFD[1]);
4263 4264 4265 4266 4267
    if (cmd) {
        VIR_DEBUG("Compression binary stderr: %s", NULLSTR(errbuf));
        VIR_FREE(errbuf);
        virCommandFree(cmd);
    }
4268 4269
    if (restoreLabel && (!bypassSecurityDriver) &&
        virSecurityManagerRestoreSavedStateLabel(driver->securityManager,
4270
                                                 vm->def, path) < 0)
4271 4272 4273 4274 4275
        VIR_WARN("failed to restore save state label on %s", path);

    if (cgroup != NULL) {
        rc = virCgroupDenyDevicePath(cgroup, path,
                                     VIR_CGROUP_DEVICE_RWM);
4276
        virDomainAuditCgroupPath(vm, cgroup, "deny", path, "rwm", rc);
4277 4278 4279 4280 4281 4282 4283
        if (rc < 0)
            VIR_WARN("Unable to deny device %s for %s %d",
                     path, vm->def->name, rc);
        virCgroupFree(&cgroup);
    }
    return ret;
}
4284 4285

int
4286
qemuMigrationJobStart(virQEMUDriverPtr driver,
4287 4288 4289 4290 4291
                      virDomainObjPtr vm,
                      enum qemuDomainAsyncJob job)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

4292
    if (qemuDomainObjBeginAsyncJob(driver, vm, job) < 0)
4293 4294
        return -1;

4295
    if (job == QEMU_ASYNC_JOB_MIGRATION_IN) {
4296
        qemuDomainObjSetAsyncJobMask(vm, QEMU_JOB_NONE);
4297 4298
    } else {
        qemuDomainObjSetAsyncJobMask(vm, DEFAULT_JOB_MASK |
4299
                                     JOB_MASK(QEMU_JOB_SUSPEND) |
4300 4301
                                     JOB_MASK(QEMU_JOB_MIGRATION_OP));
    }
4302 4303 4304 4305 4306 4307 4308

    priv->job.info.type = VIR_DOMAIN_JOB_UNBOUNDED;

    return 0;
}

void
4309
qemuMigrationJobSetPhase(virQEMUDriverPtr driver,
4310 4311 4312 4313 4314 4315 4316 4317 4318 4319 4320 4321 4322 4323 4324 4325
                         virDomainObjPtr vm,
                         enum qemuMigrationJobPhase phase)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    if (phase < priv->job.phase) {
        VIR_ERROR(_("migration protocol going backwards %s => %s"),
                  qemuMigrationJobPhaseTypeToString(priv->job.phase),
                  qemuMigrationJobPhaseTypeToString(phase));
        return;
    }

    qemuDomainObjSetJobPhase(driver, vm, phase);
}

void
4326
qemuMigrationJobStartPhase(virQEMUDriverPtr driver,
4327 4328 4329
                           virDomainObjPtr vm,
                           enum qemuMigrationJobPhase phase)
{
4330
    virObjectRef(vm);
4331 4332 4333
    qemuMigrationJobSetPhase(driver, vm, phase);
}

4334
bool
4335 4336
qemuMigrationJobContinue(virDomainObjPtr vm)
{
4337
    qemuDomainObjReleaseAsyncJob(vm);
4338
    return virObjectUnref(vm);
4339 4340 4341 4342 4343 4344 4345 4346 4347 4348 4349 4350 4351 4352 4353 4354
}

bool
qemuMigrationJobIsActive(virDomainObjPtr vm,
                         enum qemuDomainAsyncJob job)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    if (priv->job.asyncJob != job) {
        const char *msg;

        if (job == QEMU_ASYNC_JOB_MIGRATION_IN)
            msg = _("domain '%s' is not processing incoming migration");
        else
            msg = _("domain '%s' is not being migrated");

4355
        virReportError(VIR_ERR_OPERATION_INVALID, msg, vm->def->name);
4356 4357 4358 4359 4360
        return false;
    }
    return true;
}

4361
bool
4362
qemuMigrationJobFinish(virQEMUDriverPtr driver, virDomainObjPtr vm)
4363 4364 4365
{
    return qemuDomainObjEndAsyncJob(driver, vm);
}