qemu_migration.c 140.5 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
        mig->persistent = virDomainDefParseNode(caps, doc, nodes[0],
923 924 925 926 927 928 929 930 931
                                                -1, VIR_DOMAIN_XML_INACTIVE);
        if (!mig->persistent) {
            /* virDomainDefParseNode already reported
             * an error for us */
            goto error;
        }
        VIR_FREE(nodes);
    }

932 933 934 935 936
    if ((flags & QEMU_MIGRATION_COOKIE_NETWORK) &&
        virXPathBoolean("count(./network) > 0", ctxt) &&
        (!(mig->network = qemuMigrationCookieNetworkXMLParse(ctxt))))
        goto error;

937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956
    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);
    }

957
    virObjectUnref(caps);
958 959 960 961
    return 0;

error:
    VIR_FREE(tmp);
962
    VIR_FREE(nodes);
963
    virObjectUnref(caps);
964 965 966 967 968 969
    return -1;
}


static int
qemuMigrationCookieXMLParseStr(qemuMigrationCookiePtr mig,
970
                               virQEMUDriverPtr driver,
971
                               const char *xml,
E
Eric Blake 已提交
972
                               unsigned int flags)
973 974 975
{
    xmlDocPtr doc = NULL;
    xmlXPathContextPtr ctxt = NULL;
976
    int ret = -1;
977 978 979

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

980
    if (!(doc = virXMLParseStringCtxt(xml, _("(qemu_migration_cookie)"), &ctxt)))
981 982
        goto cleanup;

983
    ret = qemuMigrationCookieXMLParse(mig, driver, doc, ctxt, flags);
984 985 986 987 988 989 990 991 992 993 994

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

    return ret;
}


static int
qemuMigrationBakeCookie(qemuMigrationCookiePtr mig,
995
                        virQEMUDriverPtr driver,
996
                        virDomainObjPtr dom,
997 998
                        char **cookieout,
                        int *cookieoutlen,
E
Eric Blake 已提交
999
                        unsigned int flags)
1000
{
1001 1002
    if (!cookieout || !cookieoutlen)
        return 0;
1003 1004 1005

    *cookieoutlen = 0;

1006 1007 1008 1009
    if (flags & QEMU_MIGRATION_COOKIE_GRAPHICS &&
        qemuMigrationCookieAddGraphics(mig, driver, dom) < 0)
        return -1;

1010 1011 1012 1013
    if (flags & QEMU_MIGRATION_COOKIE_LOCKSTATE &&
        qemuMigrationCookieAddLockstate(mig, driver, dom) < 0)
        return -1;

1014 1015 1016 1017
    if (flags & QEMU_MIGRATION_COOKIE_PERSISTENT &&
        qemuMigrationCookieAddPersistent(mig, dom) < 0)
        return -1;

1018 1019 1020 1021 1022
    if (flags & QEMU_MIGRATION_COOKIE_NETWORK &&
        qemuMigrationCookieAddNetwork(mig, driver, dom) < 0) {
        return -1;
    }

1023 1024 1025 1026
    if ((flags & QEMU_MIGRATION_COOKIE_NBD) &&
        qemuMigrationCookieAddNBD(mig, driver, dom) < 0)
        return -1;

1027
    if (!(*cookieout = qemuMigrationCookieXMLFormatStr(driver, mig)))
1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038
        return -1;

    *cookieoutlen = strlen(*cookieout) + 1;

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

    return 0;
}


static qemuMigrationCookiePtr
1039
qemuMigrationEatCookie(virQEMUDriverPtr driver,
1040
                       virDomainObjPtr dom,
1041 1042
                       const char *cookiein,
                       int cookieinlen,
E
Eric Blake 已提交
1043
                       unsigned int flags)
1044 1045 1046 1047 1048 1049
{
    qemuMigrationCookiePtr mig = NULL;

    /* Parse & validate incoming cookie (if any) */
    if (cookiein && cookieinlen &&
        cookiein[cookieinlen-1] != '\0') {
1050 1051
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Migration cookie was not NULL terminated"));
1052 1053 1054 1055 1056 1057 1058 1059 1060 1061
        goto error;
    }

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

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

    if (cookiein && cookieinlen &&
        qemuMigrationCookieXMLParseStr(mig,
1062
                                       driver,
1063 1064 1065 1066
                                       cookiein,
                                       flags) < 0)
        goto error;

1067 1068 1069
    if (mig->flags & QEMU_MIGRATION_COOKIE_LOCKSTATE) {
        if (!mig->lockDriver) {
            if (virLockManagerPluginUsesState(driver->lockManager)) {
1070 1071 1072
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("Missing %s lock state for migration cookie"),
                               virLockManagerPluginGetName(driver->lockManager));
1073 1074 1075 1076
                goto error;
            }
        } else if (STRNEQ(mig->lockDriver,
                          virLockManagerPluginGetName(driver->lockManager))) {
1077 1078 1079 1080
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Source host lock driver %s different from target %s"),
                           mig->lockDriver,
                           virLockManagerPluginGetName(driver->lockManager));
1081 1082 1083 1084
            goto error;
        }
    }

1085 1086 1087 1088 1089 1090
    return mig;

error:
    qemuMigrationCookieFree(mig);
    return NULL;
}
1091

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
/**
 * 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;
}

1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 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
/**
 * 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;
}
1330

1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354

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;
}

1355 1356 1357 1358 1359 1360 1361 1362 1363
/* 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. */
1364
bool
1365
qemuMigrationIsAllowed(virQEMUDriverPtr driver, virDomainObjPtr vm,
1366
                       virDomainDefPtr def, bool remote)
1367
{
1368
    int nsnapshots;
1369 1370
    bool forbid;
    int i;
1371 1372 1373

    if (vm) {
        if (qemuProcessAutoDestroyActive(driver, vm)) {
1374 1375
            virReportError(VIR_ERR_OPERATION_INVALID,
                           "%s", _("domain is marked for auto destroy"));
1376 1377
            return false;
        }
1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390

        /* 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;
            }
1391
        }
1392

E
Eric Blake 已提交
1393 1394
        if (virDomainHasDiskMirror(vm)) {
            virReportError(VIR_ERR_OPERATION_INVALID, "%s",
1395
                           _("domain has an active block job"));
E
Eric Blake 已提交
1396 1397
            return false;
        }
1398 1399 1400

        def = vm->def;
    }
1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415

    /* 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",
1416
                       _("domain has assigned non-USB host devices"));
1417 1418 1419 1420 1421 1422
        return false;
    }

    return true;
}

1423 1424 1425 1426 1427 1428 1429 1430
static bool
qemuMigrationIsSafe(virDomainDefPtr def)
{
    int i;

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

1431 1432
        /* Our code elsewhere guarantees shared disks are either readonly (in
         * which case cache mode doesn't matter) or used with cache=none */
1433
        if (disk->src &&
1434 1435 1436
            !disk->shared &&
            !disk->readonly &&
            disk->cachemode != VIR_DOMAIN_DISK_CACHE_DISABLE) {
1437
            int cfs;
1438 1439 1440 1441 1442 1443

            if (disk->type == VIR_DOMAIN_DISK_TYPE_FILE) {
                if ((cfs = virStorageFileIsClusterFS(disk->src)) == 1)
                    continue;
                else if (cfs < 0)
                    return false;
1444 1445 1446
            } else if (disk->type == VIR_DOMAIN_DISK_TYPE_NETWORK &&
                       disk->protocol == VIR_DOMAIN_DISK_PROTOCOL_RBD) {
                continue;
1447
            }
1448

1449 1450 1451
            virReportError(VIR_ERR_MIGRATE_UNSAFE, "%s",
                           _("Migration may lead to data corruption if disks"
                             " use cache != none"));
1452 1453 1454 1455 1456 1457 1458
            return false;
        }
    }

    return true;
}

1459 1460 1461 1462
/** qemuMigrationSetOffline
 * Pause domain for non-live migration.
 */
int
1463
qemuMigrationSetOffline(virQEMUDriverPtr driver,
1464 1465 1466
                        virDomainObjPtr vm)
{
    int ret;
1467
    VIR_DEBUG("driver=%p vm=%p", driver, vm);
1468 1469
    ret = qemuProcessStopCPUs(driver, vm, VIR_DOMAIN_PAUSED_MIGRATION,
                              QEMU_ASYNC_JOB_MIGRATION_OUT);
1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483
    if (ret == 0) {
        virDomainEventPtr event;

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

    return ret;
}


1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524
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;
}


1525
static int
1526
qemuMigrationUpdateJobStatus(virQEMUDriverPtr driver,
1527
                             virDomainObjPtr vm,
1528 1529
                             const char *job,
                             enum qemuDomainAsyncJob asyncJob)
1530
{
1531
    qemuDomainObjPrivatePtr priv = vm->privateData;
1532
    int ret;
1533 1534
    bool wait_for_spice = false;
    bool spice_migrated = false;
1535 1536 1537
    qemuMonitorMigrationStatus status;

    memset(&status, 0, sizeof(status));
1538

1539 1540 1541 1542
    /* 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 &&
1543
        virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_SEAMLESS_MIGRATION))
1544 1545
        wait_for_spice = true;

1546 1547 1548
    ret = qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob);
    if (ret < 0) {
        /* Guest already exited; nothing further to update.  */
1549 1550
        return -1;
    }
1551
    ret = qemuMonitorGetMigrationStatus(priv->mon, &status);
1552 1553

    /* If qemu says migrated, check spice */
1554 1555 1556
    if (wait_for_spice &&
        ret == 0 &&
        status.status == QEMU_MONITOR_MIGRATION_STATUS_COMPLETED)
1557 1558 1559
        ret = qemuMonitorGetSpiceMigrationStatus(priv->mon,
                                                 &spice_migrated);

1560
    qemuDomainObjExitMonitor(driver, vm);
1561

1562 1563
    priv->job.status = status;

1564
    if (ret < 0 || virTimeMillisNow(&priv->job.info.timeElapsed) < 0) {
1565
        priv->job.info.type = VIR_DOMAIN_JOB_FAILED;
1566 1567
        return -1;
    }
1568
    priv->job.info.timeElapsed -= priv->job.start;
1569

1570
    ret = -1;
1571
    switch (priv->job.status.status) {
1572
    case QEMU_MONITOR_MIGRATION_STATUS_INACTIVE:
1573
        priv->job.info.type = VIR_DOMAIN_JOB_NONE;
1574 1575
        virReportError(VIR_ERR_OPERATION_FAILED,
                       _("%s: %s"), job, _("is not active"));
1576 1577 1578
        break;

    case QEMU_MONITOR_MIGRATION_STATUS_ACTIVE:
1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593
        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;
1594 1595 1596 1597 1598

        ret = 0;
        break;

    case QEMU_MONITOR_MIGRATION_STATUS_COMPLETED:
1599 1600
        if ((wait_for_spice && spice_migrated) || (!wait_for_spice))
            priv->job.info.type = VIR_DOMAIN_JOB_COMPLETED;
1601 1602 1603 1604
        ret = 0;
        break;

    case QEMU_MONITOR_MIGRATION_STATUS_ERROR:
1605
        priv->job.info.type = VIR_DOMAIN_JOB_FAILED;
1606 1607
        virReportError(VIR_ERR_OPERATION_FAILED,
                       _("%s: %s"), job, _("unexpectedly failed"));
1608 1609 1610
        break;

    case QEMU_MONITOR_MIGRATION_STATUS_CANCELLED:
1611
        priv->job.info.type = VIR_DOMAIN_JOB_CANCELLED;
1612 1613
        virReportError(VIR_ERR_OPERATION_ABORTED,
                       _("%s: %s"), job, _("canceled by client"));
1614 1615 1616 1617 1618 1619 1620
        break;
    }

    return ret;
}


1621
static int
1622
qemuMigrationWaitForCompletion(virQEMUDriverPtr driver, virDomainObjPtr vm,
1623 1624
                               enum qemuDomainAsyncJob asyncJob,
                               virConnectPtr dconn)
1625
{
1626
    qemuDomainObjPrivatePtr priv = vm->privateData;
1627 1628
    const char *job;

1629 1630
    switch (priv->job.asyncJob) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
1631 1632
        job = _("migration job");
        break;
1633
    case QEMU_ASYNC_JOB_SAVE:
1634 1635
        job = _("domain save job");
        break;
1636
    case QEMU_ASYNC_JOB_DUMP:
1637 1638 1639 1640 1641
        job = _("domain core dump job");
        break;
    default:
        job = _("job");
    }
1642

1643
    priv->job.info.type = VIR_DOMAIN_JOB_UNBOUNDED;
1644

1645
    while (priv->job.info.type == VIR_DOMAIN_JOB_UNBOUNDED) {
1646 1647 1648
        /* Poll every 50ms for progress & to allow cancellation */
        struct timespec ts = { .tv_sec = 0, .tv_nsec = 50 * 1000 * 1000ull };

1649
        if (qemuMigrationUpdateJobStatus(driver, vm, job, asyncJob) < 0)
1650 1651
            goto cleanup;

1652
        if (dconn && virConnectIsAlive(dconn) <= 0) {
1653 1654
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("Lost connection to destination host"));
1655 1656 1657
            goto cleanup;
        }

1658
        virObjectUnlock(vm);
1659 1660 1661

        nanosleep(&ts, NULL);

1662
        virObjectLock(vm);
1663 1664 1665
    }

cleanup:
1666
    if (priv->job.info.type == VIR_DOMAIN_JOB_COMPLETED)
1667 1668 1669
        return 0;
    else
        return -1;
1670 1671 1672
}


1673
static int
1674
qemuDomainMigrateGraphicsRelocate(virQEMUDriverPtr driver,
1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692
                                  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;

1693 1694
    ret = qemuDomainObjEnterMonitorAsync(driver, vm,
                                         QEMU_ASYNC_JOB_MIGRATION_OUT);
1695 1696 1697 1698 1699 1700 1701
    if (ret == 0) {
        ret = qemuMonitorGraphicsRelocate(priv->mon,
                                          cookie->graphics->type,
                                          cookie->remoteHostname,
                                          cookie->graphics->port,
                                          cookie->graphics->tlsPort,
                                          cookie->graphics->tlsSubject);
1702
        qemuDomainObjExitMonitor(driver, vm);
1703
    }
1704 1705 1706 1707 1708

    return ret;
}


1709
static int
1710
qemuDomainMigrateOPDRelocate(virQEMUDriverPtr driver ATTRIBUTE_UNUSED,
1711 1712 1713
                             virDomainObjPtr vm,
                             qemuMigrationCookiePtr cookie)
{
1714 1715
    virDomainNetDefPtr netptr;
    int ret = -1;
1716 1717 1718 1719 1720 1721 1722 1723 1724
    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:
1725
           break;
1726
        case VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH:
1727 1728 1729 1730 1731 1732 1733 1734
            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;
1735 1736 1737 1738 1739
        default:
            break;
        }
    }

1740 1741
    ret = 0;
cleanup:
1742 1743 1744 1745
    return ret;
}


1746 1747 1748 1749 1750 1751 1752
/* 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
1753
qemuMigrationCleanup(virQEMUDriverPtr driver,
1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806
                     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;
}

1807
/* The caller is supposed to lock the vm and start a migration job. */
1808
char *qemuMigrationBegin(virQEMUDriverPtr driver,
1809
                         virDomainObjPtr vm,
1810
                         const char *xmlin,
1811
                         const char *dname,
1812
                         char **cookieout,
1813 1814
                         int *cookieoutlen,
                         unsigned long flags)
1815 1816 1817
{
    char *rv = NULL;
    qemuMigrationCookiePtr mig = NULL;
1818
    virDomainDefPtr def = NULL;
1819
    qemuDomainObjPrivatePtr priv = vm->privateData;
1820
    virCapsPtr caps = NULL;
1821
    unsigned int cookieFlags = QEMU_MIGRATION_COOKIE_LOCKSTATE;
1822

1823
    VIR_DEBUG("driver=%p, vm=%p, xmlin=%s, dname=%s,"
1824
              " cookieout=%p, cookieoutlen=%p, flags=%lx",
1825
              driver, vm, NULLSTR(xmlin), NULLSTR(dname),
1826
              cookieout, cookieoutlen, flags);
1827

1828 1829 1830
    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
        goto cleanup;

1831 1832 1833 1834 1835 1836
    /* 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);
1837

1838
    if (!qemuMigrationIsAllowed(driver, vm, NULL, true))
1839 1840
        goto cleanup;

1841 1842 1843
    if (!(flags & VIR_MIGRATE_UNSAFE) && !qemuMigrationIsSafe(vm->def))
        goto cleanup;

1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854
    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;
        }
    }

1855
    if (!(mig = qemuMigrationEatCookie(driver, vm, NULL, 0, 0)))
1856 1857 1858 1859
        goto cleanup;

    if (qemuMigrationBakeCookie(mig, driver, vm,
                                cookieout, cookieoutlen,
1860
                                cookieFlags) < 0)
1861 1862
        goto cleanup;

L
liguang 已提交
1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884
    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;
        }
    }

1885
    if (xmlin) {
1886
        if (!(def = virDomainDefParseString(caps, xmlin,
M
Matthias Bolte 已提交
1887
                                            QEMU_EXPECTED_VIRT_TYPES,
1888 1889 1890
                                            VIR_DOMAIN_XML_INACTIVE)))
            goto cleanup;

1891
        if (STRNEQ(def->name, vm->def->name)) {
1892 1893
            virReportError(VIR_ERR_INVALID_ARG, "%s",
                           _("target domain name doesn't match source name"));
1894 1895 1896
            goto cleanup;
        }

1897
        if (!virDomainDefCheckABIStability(vm->def, def))
1898 1899
            goto cleanup;

1900
        rv = qemuDomainDefFormatLive(driver, def, false, true);
1901
    } else {
1902
        rv = qemuDomainDefFormatLive(driver, vm->def, false, true);
1903
    }
1904 1905 1906

cleanup:
    qemuMigrationCookieFree(mig);
1907
    virObjectUnref(caps);
1908
    virDomainDefFree(def);
1909 1910 1911 1912
    return rv;
}


1913 1914
/* Prepare is the first step, and it runs on the destination host.
 */
1915

1916
static void
1917
qemuMigrationPrepareCleanup(virQEMUDriverPtr driver,
1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932
                            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);
}

1933
static int
1934
qemuMigrationPrepareAny(virQEMUDriverPtr driver,
1935 1936 1937 1938 1939 1940 1941 1942
                        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
1943 1944
                        virStreamPtr st,
                        unsigned long flags)
1945 1946 1947 1948 1949
{
    virDomainDefPtr def = NULL;
    virDomainObjPtr vm = NULL;
    virDomainEventPtr event = NULL;
    int ret = -1;
1950
    int dataFD[2] = { -1, -1 };
1951
    qemuDomainObjPrivatePtr priv = NULL;
J
Jiri Denemark 已提交
1952
    unsigned long long now;
1953
    qemuMigrationCookiePtr mig = NULL;
1954
    bool tunnel = !!st;
J
Jiri Denemark 已提交
1955
    char *origname = NULL;
J
Jiri Denemark 已提交
1956
    char *xmlout = NULL;
L
liguang 已提交
1957
    unsigned int cookieFlags;
1958
    virCapsPtr caps = NULL;
1959

1960
    if (virTimeMillisNow(&now) < 0)
1961 1962
        return -1;

L
liguang 已提交
1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984
    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;
        }
    }

1985 1986 1987 1988
    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
        goto cleanup;

    if (!(def = virDomainDefParseString(caps, dom_xml,
M
Matthias Bolte 已提交
1989
                                        QEMU_EXPECTED_VIRT_TYPES,
1990 1991 1992
                                        VIR_DOMAIN_XML_INACTIVE)))
        goto cleanup;

1993
    if (!qemuMigrationIsAllowed(driver, NULL, def, true))
1994 1995 1996 1997
        goto cleanup;

    /* Target domain name, maybe renamed. */
    if (dname) {
J
Jiri Denemark 已提交
1998
        origname = def->name;
1999 2000 2001 2002 2003
        def->name = strdup(dname);
        if (def->name == NULL)
            goto cleanup;
    }

J
Jiri Denemark 已提交
2004 2005 2006 2007 2008
    /* Let migration hook filter domain XML */
    if (virHookPresent(VIR_HOOK_DRIVER_QEMU)) {
        char *xml;
        int hookret;

2009
        if (!(xml = qemuDomainDefFormatXML(driver, def,
2010 2011
                                           VIR_DOMAIN_XML_SECURE |
                                           VIR_DOMAIN_XML_MIGRATABLE)))
J
Jiri Denemark 已提交
2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
            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);
2029
                newdef = virDomainDefParseString(caps, xmlout,
J
Jiri Denemark 已提交
2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045
                                                 QEMU_EXPECTED_VIRT_TYPES,
                                                 VIR_DOMAIN_XML_INACTIVE);
                if (!newdef)
                    goto cleanup;

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

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

2046
    if (!(vm = virDomainObjListAdd(driver->domains,
2047
                                   caps,
2048 2049 2050 2051
                                   def,
                                   VIR_DOMAIN_OBJ_LIST_ADD_LIVE |
                                   VIR_DOMAIN_OBJ_LIST_ADD_CHECK_LIVE,
                                   NULL)))
2052
        goto cleanup;
2053

2054 2055
    def = NULL;
    priv = vm->privateData;
J
Jiri Denemark 已提交
2056 2057
    priv->origname = origname;
    origname = NULL;
2058

2059
    if (!(mig = qemuMigrationEatCookie(driver, vm, cookiein, cookieinlen,
2060 2061
                                       QEMU_MIGRATION_COOKIE_LOCKSTATE |
                                       QEMU_MIGRATION_COOKIE_NBD)))
2062 2063
        goto cleanup;

2064
    if (qemuMigrationJobStart(driver, vm, QEMU_ASYNC_JOB_MIGRATION_IN) < 0)
2065
        goto cleanup;
2066
    qemuMigrationJobSetPhase(driver, vm, QEMU_MIGRATION_PHASE_PREPARE);
2067 2068 2069 2070

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

L
liguang 已提交
2071 2072 2073
    if (flags & VIR_MIGRATE_OFFLINE)
        goto done;

2074 2075
    if (tunnel &&
        (pipe(dataFD) < 0 || virSetCloseExec(dataFD[1]) < 0)) {
2076 2077
        virReportSystemError(errno, "%s",
                             _("cannot create pipe for tunnelled migration"));
2078 2079 2080 2081
        goto endjob;
    }

    /* Start the QEMU daemon, with the same command-line arguments plus
2082
     * -incoming $migrateFrom
2083
     */
2084 2085 2086 2087
    if (qemuProcessStart(dconn, driver, vm, migrateFrom, dataFD[0], NULL, NULL,
                         VIR_NETDEV_VPORT_PROFILE_OP_MIGRATE_IN_START,
                         VIR_QEMU_PROCESS_START_PAUSED |
                         VIR_QEMU_PROCESS_START_AUTODESROY) < 0) {
2088
        virDomainAuditStart(vm, "migrated", false);
2089 2090 2091 2092 2093 2094
        /* Note that we don't set an error here because qemuProcessStart
         * should have already done that.
         */
        goto endjob;
    }

2095 2096 2097 2098
    if (tunnel) {
        if (virFDStreamOpen(st, dataFD[1]) < 0) {
            virReportSystemError(errno, "%s",
                                 _("cannot pass pipe for tunnelled migration"));
2099
            goto stop;
2100
        }
2101
        dataFD[1] = -1; /* 'st' owns the FD now & will close it */
2102 2103
    }

2104 2105 2106 2107 2108
    if (flags & VIR_MIGRATE_COMPRESSED &&
        qemuMigrationSetCompression(driver, vm,
                                    QEMU_ASYNC_JOB_MIGRATION_IN) < 0)
        goto stop;

2109 2110 2111 2112 2113 2114 2115 2116
    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");
    }
2117

L
liguang 已提交
2118 2119 2120 2121 2122 2123
done:
    if (flags & VIR_MIGRATE_OFFLINE)
        cookieFlags = 0;
    else
        cookieFlags = QEMU_MIGRATION_COOKIE_GRAPHICS;

2124 2125 2126 2127 2128 2129 2130
    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 {
2131 2132 2133 2134
            if (qemuMigrationStartNBDServer(driver, vm) < 0) {
                /* error already reported */
                goto endjob;
            }
2135 2136 2137 2138 2139 2140
            cookieFlags |= QEMU_MIGRATION_COOKIE_NBD;
        }
    }

    if (qemuMigrationBakeCookie(mig, driver, vm, cookieout,
                                cookieoutlen, cookieFlags) < 0) {
2141 2142 2143 2144 2145 2146 2147
        /* 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");
    }

2148 2149 2150
    if (qemuDomainCleanupAdd(vm, qemuMigrationPrepareCleanup) < 0)
        goto endjob;

L
liguang 已提交
2151 2152 2153 2154 2155 2156
    if (!(flags & VIR_MIGRATE_OFFLINE)) {
        virDomainAuditStart(vm, "migrated", true);
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STARTED,
                                         VIR_DOMAIN_EVENT_STARTED_MIGRATED);
    }
2157

2158 2159 2160 2161
    /* 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.
     */
2162
    if (!qemuMigrationJobContinue(vm)) {
2163
        vm = NULL;
2164 2165
        virReportError(VIR_ERR_OPERATION_FAILED,
                       "%s", _("domain disappeared"));
2166
        goto cleanup;
2167
    }
2168

2169
    ret = 0;
2170 2171

cleanup:
J
Jiri Denemark 已提交
2172
    VIR_FREE(origname);
J
Jiri Denemark 已提交
2173
    VIR_FREE(xmlout);
2174
    virDomainDefFree(def);
2175 2176
    VIR_FORCE_CLOSE(dataFD[0]);
    VIR_FORCE_CLOSE(dataFD[1]);
2177 2178
    if (vm) {
        if (ret >= 0 || vm->persistent)
2179
            virObjectUnlock(vm);
2180 2181
        else
            qemuDomainRemoveInactive(driver, vm);
2182 2183 2184 2185
        if (ret < 0 && priv->nbdPort) {
            virPortAllocatorRelease(driver->remotePorts, priv->nbdPort);
            priv->nbdPort = 0;
        }
2186
    }
2187 2188
    if (event)
        qemuDomainEventQueue(driver, event);
2189
    qemuMigrationCookieFree(mig);
2190
    virObjectUnref(caps);
2191
    return ret;
2192

2193 2194 2195 2196
stop:
    virDomainAuditStart(vm, "migrated", false);
    qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_FAILED, 0);

2197
endjob:
2198
    if (!qemuMigrationJobFinish(driver, vm)) {
2199 2200 2201
        vm = NULL;
    }
    goto cleanup;
2202 2203 2204
}


2205 2206 2207 2208 2209
/*
 * This version starts an empty VM listening on a localhost TCP port, and
 * sets up the corresponding virStream to handle the incoming data.
 */
int
2210
qemuMigrationPrepareTunnel(virQEMUDriverPtr driver,
2211 2212 2213 2214 2215 2216 2217
                           virConnectPtr dconn,
                           const char *cookiein,
                           int cookieinlen,
                           char **cookieout,
                           int *cookieoutlen,
                           virStreamPtr st,
                           const char *dname,
L
liguang 已提交
2218 2219
                           const char *dom_xml,
                           unsigned long flags)
2220 2221 2222 2223
{
    int ret;

    VIR_DEBUG("driver=%p, dconn=%p, cookiein=%s, cookieinlen=%d, "
2224 2225
              "cookieout=%p, cookieoutlen=%p, st=%p, dname=%s, dom_xml=%s "
              "flags=%lx",
2226
              driver, dconn, NULLSTR(cookiein), cookieinlen,
2227
              cookieout, cookieoutlen, st, NULLSTR(dname), dom_xml, flags);
2228 2229 2230 2231 2232 2233

    /* 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
2234
                                  "stdio", st, flags);
2235 2236 2237 2238
    return ret;
}


2239
int
2240
qemuMigrationPrepareDirect(virQEMUDriverPtr driver,
2241
                           virConnectPtr dconn,
2242 2243 2244 2245
                           const char *cookiein,
                           int cookieinlen,
                           char **cookieout,
                           int *cookieoutlen,
2246 2247 2248
                           const char *uri_in,
                           char **uri_out,
                           const char *dname,
L
liguang 已提交
2249 2250
                           const char *dom_xml,
                           unsigned long flags)
2251 2252 2253 2254 2255 2256 2257
{
    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
2258

2259 2260
    VIR_DEBUG("driver=%p, dconn=%p, cookiein=%s, cookieinlen=%d, "
              "cookieout=%p, cookieoutlen=%p, uri_in=%s, uri_out=%p, "
2261
              "dname=%s, dom_xml=%s flags=%lx",
2262 2263
              driver, dconn, NULLSTR(cookiein), cookieinlen,
              cookieout, cookieoutlen, NULLSTR(uri_in), uri_out,
2264
              NULLSTR(dname), dom_xml, flags);
2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284

    /* 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")) {
2285 2286 2287
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("hostname on destination resolved to localhost,"
                             " but migration requires an FQDN"));
2288 2289 2290 2291 2292
            goto cleanup;
        }

        /* XXX this really should have been a properly well-formed
         * URI, but we can't add in tcp:// now without breaking
2293
         * compatibility with old targets. We at least make the
2294 2295 2296
         * new targets accept both syntaxes though.
         */
        /* Caller frees */
2297
        if (virAsprintf(uri_out, "tcp:%s:%d", hostname, this_port) < 0) {
2298 2299 2300 2301 2302 2303 2304 2305
            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.
         */
2306
        if (!STRPREFIX(uri_in, "tcp:")) {
2307 2308 2309
            virReportError(VIR_ERR_INVALID_ARG, "%s",
                           _("only tcp URIs are supported for KVM/QEMU"
                             " migrations"));
2310 2311 2312 2313
            goto cleanup;
        }

        /* Get the port number. */
2314
        p = strrchr(uri_in, ':');
2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328
        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 */
2329 2330
            this_port = virParseNumber(&p);
            if (this_port == -1 || p-uri_in != strlen(uri_in)) {
2331 2332
                virReportError(VIR_ERR_INVALID_ARG,
                               "%s", _("URI ended with incorrect ':port'"));
2333 2334 2335 2336 2337 2338 2339 2340
                goto cleanup;
            }
        }
    }

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

2341
    /* QEMU will be started with -incoming tcp:0.0.0.0:port */
2342
    snprintf(migrateFrom, sizeof(migrateFrom), "tcp:0.0.0.0:%d", this_port);
2343

2344 2345
    ret = qemuMigrationPrepareAny(driver, dconn, cookiein, cookieinlen,
                                  cookieout, cookieoutlen, dname, dom_xml,
L
liguang 已提交
2346
                                  migrateFrom, NULL, flags);
2347 2348 2349 2350 2351 2352 2353 2354
cleanup:
    VIR_FREE(hostname);
    if (ret != 0)
        VIR_FREE(*uri_out);
    return ret;
}


2355 2356
enum qemuMigrationDestinationType {
    MIGRATION_DEST_HOST,
2357
    MIGRATION_DEST_CONNECT_HOST,
2358
    MIGRATION_DEST_UNIX,
2359
    MIGRATION_DEST_FD,
2360
};
2361

2362 2363 2364 2365
enum qemuMigrationForwardType {
    MIGRATION_FWD_DIRECT,
    MIGRATION_FWD_STREAM,
};
2366

2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377
typedef struct _qemuMigrationSpec qemuMigrationSpec;
typedef qemuMigrationSpec *qemuMigrationSpecPtr;
struct _qemuMigrationSpec {
    enum qemuMigrationDestinationType destType;
    union {
        struct {
            const char *name;
            int port;
        } host;

        struct {
2378
            char *file;
2379 2380
            int sock;
        } unix_socket;
2381 2382 2383 2384 2385

        struct {
            int qemu;
            int local;
        } fd;
2386 2387 2388 2389 2390 2391 2392
    } dest;

    enum qemuMigrationForwardType fwdType;
    union {
        virStreamPtr stream;
    } fwd;
};
2393 2394 2395

#define TUNNEL_SEND_BUF_SIZE 65536

2396 2397 2398 2399 2400 2401 2402
typedef struct _qemuMigrationIOThread qemuMigrationIOThread;
typedef qemuMigrationIOThread *qemuMigrationIOThreadPtr;
struct _qemuMigrationIOThread {
    virThread thread;
    virStreamPtr st;
    int sock;
    virError err;
2403 2404
    int wakeupRecvFD;
    int wakeupSendFD;
2405 2406 2407
};

static void qemuMigrationIOFunc(void *arg)
2408
{
2409
    qemuMigrationIOThreadPtr data = arg;
2410 2411 2412 2413 2414 2415 2416
    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);
2417 2418 2419

    if (VIR_ALLOC_N(buffer, TUNNEL_SEND_BUF_SIZE) < 0) {
        virReportOOMError();
2420
        goto abrt;
2421 2422
    }

2423 2424 2425
    fds[0].fd = data->sock;
    fds[1].fd = data->wakeupRecvFD;

2426
    for (;;) {
2427 2428 2429 2430 2431 2432 2433 2434 2435 2436
        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;
2437
            virReportSystemError(errno, "%s",
2438 2439
                                 _("poll failed in migration tunnel"));
            goto abrt;
2440
        }
2441 2442 2443 2444 2445 2446 2447

        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");
2448
            break;
2449
        }
2450

2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466
        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;
            }
2467 2468
        }

2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485
        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;
            }
        }
    }
2486

2487 2488
    if (virStreamFinish(data->st) < 0)
        goto error;
2489

2490 2491
    VIR_FREE(buffer);

2492 2493
    return;

2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505
abrt:
    err = virSaveLastError();
    if (err && err->code == VIR_ERR_OK) {
        virFreeError(err);
        err = NULL;
    }
    virStreamAbort(data->st);
    if (err) {
        virSetError(err);
        virFreeError(err);
    }

2506 2507 2508
error:
    virCopyLastError(&data->err);
    virResetLastError();
2509
    VIR_FREE(buffer);
2510 2511 2512 2513 2514 2515 2516
}


static qemuMigrationIOThreadPtr
qemuMigrationStartTunnel(virStreamPtr st,
                         int sock)
{
2517 2518
    qemuMigrationIOThreadPtr io = NULL;
    int wakeupFD[2] = { -1, -1 };
2519

2520 2521 2522 2523
    if (pipe2(wakeupFD, O_CLOEXEC) < 0) {
        virReportSystemError(errno, "%s",
                             _("Unable to make pipe"));
        goto error;
2524 2525
    }

2526 2527 2528
    if (VIR_ALLOC(io) < 0)
        goto no_memory;

2529 2530
    io->st = st;
    io->sock = sock;
2531 2532
    io->wakeupRecvFD = wakeupFD[0];
    io->wakeupSendFD = wakeupFD[1];
2533 2534 2535 2536 2537 2538

    if (virThreadCreate(&io->thread, true,
                        qemuMigrationIOFunc,
                        io) < 0) {
        virReportSystemError(errno, "%s",
                             _("Unable to create migration thread"));
2539
        goto error;
2540 2541 2542
    }

    return io;
2543 2544 2545 2546 2547 2548 2549 2550

no_memory:
    virReportOOMError();
error:
    VIR_FORCE_CLOSE(wakeupFD[0]);
    VIR_FORCE_CLOSE(wakeupFD[1]);
    VIR_FREE(io);
    return NULL;
2551 2552 2553
}

static int
2554
qemuMigrationStopTunnel(qemuMigrationIOThreadPtr io, bool error)
2555 2556
{
    int rv = -1;
2557 2558 2559 2560 2561 2562 2563 2564 2565
    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;
    }

2566 2567 2568 2569
    virThreadJoin(&io->thread);

    /* Forward error from the IO thread, to this thread */
    if (io->err.code != VIR_ERR_OK) {
2570 2571 2572 2573
        if (error)
            rv = 0;
        else
            virSetError(&io->err);
2574 2575 2576 2577 2578 2579 2580
        virResetError(&io->err);
        goto cleanup;
    }

    rv = 0;

cleanup:
2581 2582
    VIR_FORCE_CLOSE(io->wakeupSendFD);
    VIR_FORCE_CLOSE(io->wakeupRecvFD);
2583 2584
    VIR_FREE(io);
    return rv;
2585 2586
}

2587
static int
2588
qemuMigrationConnect(virQEMUDriverPtr driver,
2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609
                     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);
2610
        virObjectUnref(sock);
2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624
    }
    if (virSecurityManagerClearSocketLabel(driver->securityManager, vm->def) < 0 ||
        spec->dest.fd.qemu == -1)
        goto cleanup;

    ret = 0;

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

2625
static int
2626
qemuMigrationRun(virQEMUDriverPtr driver,
2627 2628 2629 2630 2631 2632 2633
                 virDomainObjPtr vm,
                 const char *cookiein,
                 int cookieinlen,
                 char **cookieout,
                 int *cookieoutlen,
                 unsigned long flags,
                 unsigned long resource,
2634 2635
                 qemuMigrationSpecPtr spec,
                 virConnectPtr dconn)
2636
{
2637
    int ret = -1;
2638 2639
    unsigned int migrate_flags = QEMU_MONITOR_MIGRATE_BACKGROUND;
    qemuDomainObjPrivatePtr priv = vm->privateData;
2640
    qemuMigrationCookiePtr mig = NULL;
2641
    qemuMigrationIOThreadPtr iothread = NULL;
2642
    int fd = -1;
2643
    unsigned long migrate_speed = resource ? resource : priv->migMaxBandwidth;
2644
    virErrorPtr orig_err = NULL;
2645
    unsigned int cookieFlags = 0;
2646 2647 2648 2649 2650 2651 2652

    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);
2653

2654 2655 2656 2657 2658 2659 2660 2661 2662 2663
    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;
    }

2664 2665
    if (virLockManagerPluginUsesState(driver->lockManager) &&
        !cookieout) {
2666 2667 2668 2669
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Migration with lock driver %s requires"
                         " cookie support"),
                       virLockManagerPluginGetName(driver->lockManager));
2670 2671 2672
        return -1;
    }

2673 2674 2675
    mig = qemuMigrationEatCookie(driver, vm, cookiein, cookieinlen,
                                 cookieFlags | QEMU_MIGRATION_COOKIE_GRAPHICS);
    if (!mig)
2676 2677 2678 2679 2680
        goto cleanup;

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

2681 2682 2683 2684 2685 2686 2687
    /* 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;
    }

2688
    /* Before EnterMonitor, since qemuMigrationSetOffline already does that */
2689 2690 2691 2692 2693 2694
    if (!(flags & VIR_MIGRATE_LIVE) &&
        virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING) {
        if (qemuMigrationSetOffline(driver, vm) < 0)
            goto cleanup;
    }

2695 2696 2697 2698 2699
    if (flags & VIR_MIGRATE_COMPRESSED &&
        qemuMigrationSetCompression(driver, vm,
                                    QEMU_ASYNC_JOB_MIGRATION_OUT) < 0)
        goto cleanup;

2700 2701
    if (qemuDomainObjEnterMonitorAsync(driver, vm,
                                       QEMU_ASYNC_JOB_MIGRATION_OUT) < 0)
2702 2703
        goto cleanup;

2704 2705 2706 2707
    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 */
2708
        qemuDomainObjExitMonitor(driver, vm);
2709 2710 2711 2712 2713 2714
        virReportError(VIR_ERR_OPERATION_ABORTED, _("%s: %s"),
                       qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                       _("canceled by client"));
        goto cleanup;
    }

2715
    if (qemuMonitorSetMigrationSpeed(priv->mon, migrate_speed) < 0) {
2716
        qemuDomainObjExitMonitor(driver, vm);
2717 2718
        goto cleanup;
    }
2719

2720 2721
    /* connect to the destination qemu if needed */
    if (spec->destType == MIGRATION_DEST_CONNECT_HOST &&
2722
        qemuMigrationConnect(driver, vm, spec) < 0) {
2723
        qemuDomainObjExitMonitor(driver, vm);
2724
        goto cleanup;
2725
    }
2726

2727 2728 2729 2730 2731 2732 2733
    switch (spec->destType) {
    case MIGRATION_DEST_HOST:
        ret = qemuMonitorMigrateToHost(priv->mon, migrate_flags,
                                       spec->dest.host.name,
                                       spec->dest.host.port);
        break;

2734 2735 2736 2737
    case MIGRATION_DEST_CONNECT_HOST:
        /* handled above and transformed into MIGRATION_DEST_FD */
        break;

2738
    case MIGRATION_DEST_UNIX:
2739
        if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_UNIX)) {
2740 2741 2742 2743 2744 2745 2746 2747 2748
            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;
2749 2750

    case MIGRATION_DEST_FD:
2751
        if (spec->fwdType != MIGRATION_FWD_DIRECT) {
2752
            fd = spec->dest.fd.local;
2753 2754
            spec->dest.fd.local = -1;
        }
2755 2756 2757 2758
        ret = qemuMonitorMigrateToFd(priv->mon, migrate_flags,
                                     spec->dest.fd.qemu);
        VIR_FORCE_CLOSE(spec->dest.fd.qemu);
        break;
2759
    }
2760
    qemuDomainObjExitMonitor(driver, vm);
2761
    if (ret < 0)
2762 2763
        goto cleanup;
    ret = -1;
2764 2765

    if (!virDomainObjIsActive(vm)) {
2766 2767
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("guest unexpectedly quit"));
2768 2769 2770 2771 2772 2773
        goto cleanup;
    }

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

2774 2775 2776 2777 2778 2779 2780 2781
    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;
2782

2783 2784 2785 2786 2787 2788 2789
        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;
        }
2790 2791
    }

2792 2793
    if (spec->fwdType != MIGRATION_FWD_DIRECT &&
        !(iothread = qemuMigrationStartTunnel(spec->fwd.stream, fd)))
2794 2795
        goto cancel;

2796
    if (qemuMigrationWaitForCompletion(driver, vm,
2797 2798
                                       QEMU_ASYNC_JOB_MIGRATION_OUT,
                                       dconn) < 0)
2799
        goto cleanup;
2800

2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813
    /* 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;
    }

2814
    ret = 0;
2815

2816
cleanup:
2817 2818 2819
    if (ret < 0 && !orig_err)
        orig_err = virSaveLastError();

2820
    if (spec->fwdType != MIGRATION_FWD_DIRECT) {
2821
        if (iothread && qemuMigrationStopTunnel(iothread, ret < 0) < 0)
2822
            ret = -1;
2823
        VIR_FORCE_CLOSE(fd);
2824
    }
2825

2826 2827
    cookieFlags |= (QEMU_MIGRATION_COOKIE_PERSISTENT |
                    QEMU_MIGRATION_COOKIE_NETWORK);
2828
    if (ret == 0 &&
2829 2830
        qemuMigrationBakeCookie(mig, driver, vm, cookieout,
                                cookieoutlen, cookieFlags) < 0) {
2831
        VIR_WARN("Unable to encode migration cookie");
2832
    }
2833

2834 2835
    qemuMigrationCookieFree(mig);

2836 2837 2838 2839 2840
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }

2841 2842
    return ret;

2843
cancel:
2844 2845
    orig_err = virSaveLastError();

2846
    if (virDomainObjIsActive(vm)) {
2847 2848
        if (qemuDomainObjEnterMonitorAsync(driver, vm,
                                           QEMU_ASYNC_JOB_MIGRATION_OUT) == 0) {
2849
            qemuMonitorMigrateCancel(priv->mon);
2850
            qemuDomainObjExitMonitor(driver, vm);
2851
        }
2852
    }
2853 2854 2855 2856 2857 2858
    goto cleanup;
}

/* Perform migration using QEMU's native TCP migrate support,
 * not encrypted obviously
 */
2859
static int doNativeMigrate(virQEMUDriverPtr driver,
2860 2861 2862 2863 2864 2865 2866
                           virDomainObjPtr vm,
                           const char *uri,
                           const char *cookiein,
                           int cookieinlen,
                           char **cookieout,
                           int *cookieoutlen,
                           unsigned long flags,
2867 2868
                           unsigned long resource,
                           virConnectPtr dconn)
2869
{
2870
    qemuDomainObjPrivatePtr priv = vm->privateData;
M
Martin Kletzander 已提交
2871
    virURIPtr uribits = NULL;
2872
    int ret = -1;
2873 2874 2875 2876 2877 2878 2879 2880
    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://")) {
2881
        char *tmp;
2882
        /* HACK: source host generates bogus URIs, so fix them up */
2883
        if (virAsprintf(&tmp, "tcp://%s", uri + strlen("tcp:")) < 0) {
2884 2885 2886
            virReportOOMError();
            return -1;
        }
M
Martin Kletzander 已提交
2887
        uribits = virURIParse(tmp);
2888
        VIR_FREE(tmp);
2889
    } else {
M
Martin Kletzander 已提交
2890
        uribits = virURIParse(uri);
2891
    }
2892
    if (!uribits)
2893 2894
        return -1;

2895
    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_FD))
2896 2897
        spec.destType = MIGRATION_DEST_CONNECT_HOST;
    else
2898
        spec.destType = MIGRATION_DEST_HOST;
2899 2900 2901
    spec.dest.host.name = uribits->server;
    spec.dest.host.port = uribits->port;
    spec.fwdType = MIGRATION_FWD_DIRECT;
2902

2903
    ret = qemuMigrationRun(driver, vm, cookiein, cookieinlen, cookieout,
2904
                           cookieoutlen, flags, resource, &spec, dconn);
2905 2906 2907 2908

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

2909
    virURIFree(uribits);
2910 2911 2912 2913 2914

    return ret;
}


2915
static int doTunnelMigrate(virQEMUDriverPtr driver,
2916 2917 2918 2919 2920 2921 2922
                           virDomainObjPtr vm,
                           virStreamPtr st,
                           const char *cookiein,
                           int cookieinlen,
                           char **cookieout,
                           int *cookieoutlen,
                           unsigned long flags,
2923 2924
                           unsigned long resource,
                           virConnectPtr dconn)
2925 2926
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
2927
    virNetSocketPtr sock = NULL;
2928 2929
    int ret = -1;
    qemuMigrationSpec spec;
2930
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
2931 2932 2933 2934 2935 2936

    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);

2937 2938 2939
    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)) {
2940 2941
        virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                       _("Source qemu is too old to support tunnelled migration"));
2942
        virObjectUnref(cfg);
2943
        return -1;
2944 2945 2946 2947 2948
    }

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

2949
    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_FD)) {
2950 2951 2952 2953 2954 2955
        int fds[2];

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

2956
        if (pipe2(fds, O_CLOEXEC) == 0) {
2957 2958 2959 2960
            spec.dest.fd.qemu = fds[1];
            spec.dest.fd.local = fds[0];
        }
        if (spec.dest.fd.qemu == -1 ||
2961
            virSecurityManagerSetImageFDLabel(driver->securityManager, vm->def,
2962
                                              spec.dest.fd.qemu) < 0) {
2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973
            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",
2974
                        cfg->libDir, vm->def->name) < 0) {
2975 2976 2977 2978 2979
            virReportOOMError();
            goto cleanup;
        }

        if (virNetSocketNewListenUNIX(spec.dest.unix_socket.file, 0700,
2980
                                      cfg->user, cfg->group,
2981 2982 2983 2984 2985 2986 2987
                                      &sock) < 0 ||
            virNetSocketListen(sock, 1) < 0)
            goto cleanup;

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

2988
    ret = qemuMigrationRun(driver, vm, cookiein, cookieinlen, cookieout,
2989
                           cookieoutlen, flags, resource, &spec, dconn);
2990

2991
cleanup:
2992 2993 2994 2995
    if (spec.destType == MIGRATION_DEST_FD) {
        VIR_FORCE_CLOSE(spec.dest.fd.qemu);
        VIR_FORCE_CLOSE(spec.dest.fd.local);
    } else {
2996
        virObjectUnref(sock);
2997 2998
        VIR_FREE(spec.dest.unix_socket.file);
    }
2999

3000
    virObjectUnref(cfg);
3001 3002 3003 3004
    return ret;
}


3005 3006 3007 3008
/* 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 */
3009
static int doPeer2PeerMigrate2(virQEMUDriverPtr driver,
3010
                               virConnectPtr sconn ATTRIBUTE_UNUSED,
3011 3012
                               virConnectPtr dconn,
                               virDomainObjPtr vm,
3013
                               const char *dconnuri,
3014 3015 3016
                               unsigned long flags,
                               const char *dname,
                               unsigned long resource)
3017 3018 3019
{
    virDomainPtr ddomain = NULL;
    char *uri_out = NULL;
3020
    char *cookie = NULL;
3021 3022 3023 3024 3025
    char *dom_xml = NULL;
    int cookielen = 0, ret;
    virErrorPtr orig_err = NULL;
    int cancelled;
    virStreamPtr st = NULL;
3026
    VIR_DEBUG("driver=%p, sconn=%p, dconn=%p, vm=%p, dconnuri=%s, "
3027
              "flags=%lx, dname=%s, resource=%lu",
3028 3029
              driver, sconn, dconn, vm, NULLSTR(dconnuri),
              flags, NULLSTR(dname), resource);
3030

3031 3032 3033 3034 3035
    /* 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,
3036 3037
                                        QEMU_DOMAIN_FORMAT_LIVE_FLAGS |
                                        VIR_DOMAIN_XML_MIGRATABLE)))
3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052
        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;

3053
        qemuDomainObjEnterRemote(vm);
3054 3055
        ret = dconn->driver->domainMigratePrepareTunnel
            (dconn, st, flags, dname, resource, dom_xml);
3056
        qemuDomainObjExitRemote(vm);
3057
    } else {
3058
        qemuDomainObjEnterRemote(vm);
3059 3060 3061
        ret = dconn->driver->domainMigratePrepare2
            (dconn, &cookie, &cookielen, NULL, &uri_out,
             flags, dname, resource, dom_xml);
3062
        qemuDomainObjExitRemote(vm);
3063 3064 3065
    }
    VIR_FREE(dom_xml);
    if (ret == -1)
3066 3067 3068
        goto cleanup;

    /* the domain may have shutdown or crashed while we had the locks dropped
3069
     * in qemuDomainObjEnterRemote, so check again
3070 3071
     */
    if (!virDomainObjIsActive(vm)) {
3072 3073
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("guest unexpectedly quit"));
3074 3075 3076
        goto cleanup;
    }

3077 3078
    if (!(flags & VIR_MIGRATE_TUNNELLED) &&
        (uri_out == NULL)) {
3079 3080
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("domainMigratePrepare2 did not set uri"));
3081 3082
        cancelled = 1;
        goto finish;
3083 3084
    }

3085 3086 3087 3088
    /* Perform the migration.  The driver isn't supposed to return
     * until the migration is complete.
     */
    VIR_DEBUG("Perform %p", sconn);
3089
    qemuMigrationJobSetPhase(driver, vm, QEMU_MIGRATION_PHASE_PERFORM2);
3090
    if (flags & VIR_MIGRATE_TUNNELLED)
3091 3092
        ret = doTunnelMigrate(driver, vm, st,
                              NULL, 0, NULL, NULL,
3093
                              flags, resource, dconn);
3094 3095 3096 3097
    else
        ret = doNativeMigrate(driver, vm, uri_out,
                              cookie, cookielen,
                              NULL, NULL, /* No out cookie with v2 migration */
3098
                              flags, resource, dconn);
3099 3100 3101 3102

    /* Perform failed. Make sure Finish doesn't overwrite the error */
    if (ret < 0)
        orig_err = virSaveLastError();
3103

3104 3105 3106 3107
    /* If Perform returns < 0, then we need to cancel the VM
     * startup on the destination
     */
    cancelled = ret < 0 ? 1 : 0;
3108 3109

finish:
3110 3111 3112 3113
    /* 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.
     */
3114
    dname = dname ? dname : vm->def->name;
3115
    VIR_DEBUG("Finish2 %p ret=%d", dconn, ret);
3116
    qemuDomainObjEnterRemote(vm);
3117
    ddomain = dconn->driver->domainMigrateFinish2
3118
        (dconn, dname, cookie, cookielen,
3119
         uri_out ? uri_out : dconnuri, flags, cancelled);
3120
    qemuDomainObjExitRemote(vm);
3121

3122 3123
cleanup:
    if (ddomain) {
3124
        virObjectUnref(ddomain);
3125 3126 3127 3128
        ret = 0;
    } else {
        ret = -1;
    }
3129

3130
    virObjectUnref(st);
3131 3132 3133 3134 3135 3136

    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
    VIR_FREE(uri_out);
3137
    VIR_FREE(cookie);
3138 3139

    return ret;
3140 3141 3142
}


3143 3144 3145 3146
/* 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 */
3147
static int doPeer2PeerMigrate3(virQEMUDriverPtr driver,
3148 3149 3150
                               virConnectPtr sconn,
                               virConnectPtr dconn,
                               virDomainObjPtr vm,
3151
                               const char *xmlin,
3152
                               const char *dconnuri,
3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168
                               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;
3169
    VIR_DEBUG("driver=%p, sconn=%p, dconn=%p, vm=%p, xmlin=%s, "
3170
              "dconnuri=%s, uri=%s, flags=%lx, dname=%s, resource=%lu",
3171 3172 3173
              driver, sconn, dconn, vm, NULLSTR(xmlin),
              NULLSTR(dconnuri), NULLSTR(uri), flags,
              NULLSTR(dname), resource);
3174

3175 3176 3177 3178 3179
    /* 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.  */

3180
    dom_xml = qemuMigrationBegin(driver, vm, xmlin, dname,
3181
                                 &cookieout, &cookieoutlen, flags);
3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196
    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;

3197
        qemuDomainObjEnterRemote(vm);
3198 3199 3200 3201
        ret = dconn->driver->domainMigratePrepareTunnel3
            (dconn, st, cookiein, cookieinlen,
             &cookieout, &cookieoutlen,
             flags, dname, resource, dom_xml);
3202
        qemuDomainObjExitRemote(vm);
3203
    } else {
3204
        qemuDomainObjEnterRemote(vm);
3205 3206
        ret = dconn->driver->domainMigratePrepare3
            (dconn, cookiein, cookieinlen, &cookieout, &cookieoutlen,
3207
             uri, &uri_out, flags, dname, resource, dom_xml);
3208
        qemuDomainObjExitRemote(vm);
3209 3210 3211 3212 3213
    }
    VIR_FREE(dom_xml);
    if (ret == -1)
        goto cleanup;

L
liguang 已提交
3214 3215 3216 3217 3218 3219 3220 3221
    if (flags & VIR_MIGRATE_OFFLINE) {
        VIR_DEBUG("Offline migration, skipping Perform phase");
        VIR_FREE(cookieout);
        cookieoutlen = 0;
        cancelled = 0;
        goto finish;
    }

3222 3223
    if (!(flags & VIR_MIGRATE_TUNNELLED) &&
        (uri_out == NULL)) {
3224 3225
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("domainMigratePrepare3 did not set uri"));
3226 3227 3228 3229 3230 3231 3232 3233 3234
        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.
     */
3235
    VIR_DEBUG("Perform3 %p uri=%s uri_out=%s", sconn, uri, uri_out);
3236
    qemuMigrationJobSetPhase(driver, vm, QEMU_MIGRATION_PHASE_PERFORM3);
3237 3238 3239 3240 3241 3242 3243 3244 3245
    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,
3246
                              flags, resource, dconn);
3247 3248 3249 3250
    else
        ret = doNativeMigrate(driver, vm, uri_out,
                              cookiein, cookieinlen,
                              &cookieout, &cookieoutlen,
3251
                              flags, resource, dconn);
3252 3253

    /* Perform failed. Make sure Finish doesn't overwrite the error */
3254
    if (ret < 0) {
3255
        orig_err = virSaveLastError();
3256 3257 3258 3259
    } else {
        qemuMigrationJobSetPhase(driver, vm,
                                 QEMU_MIGRATION_PHASE_PERFORM3_DONE);
    }
3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279

    /* 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;
3280
    qemuDomainObjEnterRemote(vm);
3281
    ddomain = dconn->driver->domainMigrateFinish3
3282
        (dconn, dname, cookiein, cookieinlen, &cookieout, &cookieoutlen,
3283
         dconnuri, uri_out ? uri_out : uri, flags, cancelled);
3284
    qemuDomainObjExitRemote(vm);
3285

3286 3287 3288 3289 3290 3291 3292
    /* 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.
3293
     */
3294
    cancelled = ddomain == NULL ? 1 : 0;
3295

3296 3297 3298 3299 3300 3301
    /* 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();

3302 3303 3304 3305
    /*
     * If cancelled, then src VM will be restarted, else
     * it will be killed
     */
3306
    VIR_DEBUG("Confirm3 %p cancelled=%d vm=%p", sconn, cancelled, vm);
3307 3308 3309 3310 3311 3312 3313
    VIR_FREE(cookiein);
    cookiein = cookieout;
    cookieinlen = cookieoutlen;
    cookieout = NULL;
    cookieoutlen = 0;
    ret = qemuMigrationConfirm(driver, sconn, vm,
                               cookiein, cookieinlen,
3314
                               flags, cancelled);
3315 3316 3317 3318
    /* 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.
     */
3319 3320 3321
    if (ret < 0)
        VIR_WARN("Guest %s probably left in 'paused' state on source",
                 vm->def->name);
3322 3323 3324

 cleanup:
    if (ddomain) {
3325
        virObjectUnref(ddomain);
3326 3327 3328 3329 3330
        ret = 0;
    } else {
        ret = -1;
    }

3331
    virObjectUnref(st);
3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344

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

    return ret;
}


3345
static int doPeer2PeerMigrate(virQEMUDriverPtr driver,
3346
                              virConnectPtr sconn,
3347
                              virDomainObjPtr vm,
3348
                              const char *xmlin,
3349
                              const char *dconnuri,
3350 3351 3352
                              const char *uri,
                              unsigned long flags,
                              const char *dname,
3353 3354
                              unsigned long resource,
                              bool *v3proto)
3355 3356 3357 3358
{
    int ret = -1;
    virConnectPtr dconn = NULL;
    bool p2p;
3359
    virErrorPtr orig_err = NULL;
3360
    bool offline = false;
3361
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
3362

3363
    VIR_DEBUG("driver=%p, sconn=%p, vm=%p, xmlin=%s, dconnuri=%s, "
3364
              "uri=%s, flags=%lx, dname=%s, resource=%lu",
3365 3366
              driver, sconn, vm, NULLSTR(xmlin), NULLSTR(dconnuri),
              NULLSTR(uri), flags, NULLSTR(dname), resource);
3367 3368 3369 3370 3371

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

3372
    qemuDomainObjEnterRemote(vm);
3373
    dconn = virConnectOpen(dconnuri);
3374
    qemuDomainObjExitRemote(vm);
3375
    if (dconn == NULL) {
3376 3377
        virReportError(VIR_ERR_OPERATION_FAILED,
                       _("Failed to connect to remote libvirt URI %s"), dconnuri);
3378
        virObjectUnref(cfg);
3379 3380 3381
        return -1;
    }

3382 3383
    if (virConnectSetKeepAlive(dconn, cfg->keepAliveInterval,
                               cfg->keepAliveCount) < 0)
3384 3385
        goto cleanup;

3386
    qemuDomainObjEnterRemote(vm);
3387 3388
    p2p = VIR_DRV_SUPPORTS_FEATURE(dconn->driver, dconn,
                                   VIR_DRV_FEATURE_MIGRATION_P2P);
3389
        /* v3proto reflects whether the caller used Perform3, but with
3390
         * p2p migrate, regardless of whether Perform2 or Perform3
3391 3392 3393 3394
         * 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
3395 3396 3397
    if (flags & VIR_MIGRATE_OFFLINE)
        offline = VIR_DRV_SUPPORTS_FEATURE(dconn->driver, dconn,
                                           VIR_DRV_FEATURE_MIGRATION_OFFLINE);
3398
    qemuDomainObjExitRemote(vm);
3399

3400
    if (!p2p) {
3401 3402
        virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                       _("Destination libvirt does not support peer-to-peer migration protocol"));
3403 3404 3405
        goto cleanup;
    }

L
liguang 已提交
3406 3407 3408 3409 3410 3411 3412
    if (flags & VIR_MIGRATE_OFFLINE && !offline) {
        virReportError(VIR_ERR_ARGUMENT_UNSUPPORTED, "%s",
                       _("offline migration is not supported by "
                         "the destination host"));
        goto cleanup;
    }

3413
    /* domain may have been stopped while we were talking to remote daemon */
L
liguang 已提交
3414
    if (!virDomainObjIsActive(vm) && !(flags & VIR_MIGRATE_OFFLINE)) {
3415 3416
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("guest unexpectedly quit"));
3417 3418 3419
        goto cleanup;
    }

3420 3421 3422 3423 3424 3425 3426
    /* 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;

3427
    if (*v3proto)
3428
        ret = doPeer2PeerMigrate3(driver, sconn, dconn, vm, xmlin,
3429
                                  dconnuri, uri, flags, dname, resource);
3430 3431
    else
        ret = doPeer2PeerMigrate2(driver, sconn, dconn, vm,
3432
                                  dconnuri, flags, dname, resource);
3433 3434

cleanup:
3435
    orig_err = virSaveLastError();
3436
    qemuDomainObjEnterRemote(vm);
3437
    virConnectClose(dconn);
3438
    qemuDomainObjExitRemote(vm);
3439 3440 3441 3442
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
3443
    virObjectUnref(cfg);
3444 3445 3446 3447
    return ret;
}


3448 3449 3450 3451 3452 3453
/*
 * 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
3454
qemuMigrationPerformJob(virQEMUDriverPtr driver,
3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467
                        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)
3468 3469 3470 3471
{
    virDomainEventPtr event = NULL;
    int ret = -1;
    int resume = 0;
3472
    virErrorPtr orig_err = NULL;
3473
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
3474

3475
    if (qemuMigrationJobStart(driver, vm, QEMU_ASYNC_JOB_MIGRATION_OUT) < 0)
3476 3477
        goto cleanup;

L
liguang 已提交
3478
    if (!virDomainObjIsActive(vm) && !(flags & VIR_MIGRATE_OFFLINE)) {
3479 3480
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("domain is not running"));
3481 3482 3483
        goto endjob;
    }

3484
    if (!qemuMigrationIsAllowed(driver, vm, NULL, true))
3485
        goto endjob;
3486

3487
    if (!(flags & VIR_MIGRATE_UNSAFE) && !qemuMigrationIsSafe(vm->def))
3488
        goto endjob;
3489

J
Jiri Denemark 已提交
3490
    resume = virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING;
3491 3492

    if ((flags & (VIR_MIGRATE_TUNNELLED | VIR_MIGRATE_PEER2PEER))) {
3493 3494 3495
        ret = doPeer2PeerMigrate(driver, conn, vm, xmlin,
                                 dconnuri, uri, flags, dname,
                                 resource, &v3proto);
3496
    } else {
3497 3498 3499
        qemuMigrationJobSetPhase(driver, vm, QEMU_MIGRATION_PHASE_PERFORM2);
        ret = doNativeMigrate(driver, vm, uri, cookiein, cookieinlen,
                              cookieout, cookieoutlen,
3500
                              flags, resource, NULL);
3501
    }
3502 3503
    if (ret < 0)
        goto endjob;
3504

3505 3506 3507 3508
    /*
     * In v3 protocol, the source VM is not killed off until the
     * confirm step.
     */
3509
    if (!v3proto) {
3510 3511
        qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_MIGRATED,
                        VIR_QEMU_PROCESS_STOP_MIGRATED);
3512
        virDomainAuditStop(vm, "migrated");
3513 3514 3515
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         VIR_DOMAIN_EVENT_STOPPED_MIGRATED);
3516
    }
3517
    resume = 0;
3518 3519

endjob:
3520 3521 3522
    if (ret < 0)
        orig_err = virSaveLastError();

J
Jiri Denemark 已提交
3523
    if (resume && virDomainObjGetState(vm, NULL) == VIR_DOMAIN_PAUSED) {
3524
        /* we got here through some sort of failure; start the domain again */
J
Jiri Denemark 已提交
3525
        if (qemuProcessStartCPUs(driver, vm, conn,
3526 3527
                                 VIR_DOMAIN_RUNNING_MIGRATION_CANCELED,
                                 QEMU_ASYNC_JOB_MIGRATION_OUT) < 0) {
3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539
            /* 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);
    }
3540

3541
    if (!qemuMigrationJobFinish(driver, vm)) {
3542 3543 3544 3545 3546
        vm = NULL;
    } else if (!virDomainObjIsActive(vm) &&
               (!vm->persistent ||
                (ret == 0 && (flags & VIR_MIGRATE_UNDEFINE_SOURCE)))) {
        if (flags & VIR_MIGRATE_UNDEFINE_SOURCE)
3547
            virDomainDeleteConfig(cfg->configDir, cfg->autostartDir, vm);
3548
        qemuDomainRemoveInactive(driver, vm);
3549 3550 3551
        vm = NULL;
    }

3552 3553 3554 3555 3556
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }

3557 3558
cleanup:
    if (vm)
3559
        virObjectUnlock(vm);
3560 3561
    if (event)
        qemuDomainEventQueue(driver, event);
3562
    virObjectUnref(cfg);
3563 3564 3565 3566 3567 3568 3569
    return ret;
}

/*
 * This implements perform phase of v3 migration protocol.
 */
static int
3570
qemuMigrationPerformPhase(virQEMUDriverPtr driver,
3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583
                          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;
3584
    bool hasrefs;
3585 3586 3587 3588 3589 3590 3591 3592 3593 3594

    /* 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);
3595 3596
    virQEMUCloseCallbacksUnset(driver->closeCallbacks, vm,
                               qemuMigrationCleanup);
3597 3598 3599 3600

    resume = virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING;
    ret = doNativeMigrate(driver, vm, uri, cookiein, cookieinlen,
                          cookieout, cookieoutlen,
3601
                          flags, resource, NULL);
3602 3603 3604 3605 3606

    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,
3607 3608
                                 VIR_DOMAIN_RUNNING_MIGRATION_CANCELED,
                                 QEMU_ASYNC_JOB_MIGRATION_OUT) < 0) {
3609 3610 3611 3612 3613 3614
            /* 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);
3615
        }
3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626

        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);

3627 3628
    if (virQEMUCloseCallbacksSet(driver->closeCallbacks, vm, conn,
                                 qemuMigrationCleanup) < 0)
3629 3630
        goto endjob;

3631 3632
endjob:
    if (ret < 0)
3633
        hasrefs = qemuMigrationJobFinish(driver, vm);
3634
    else
3635 3636
        hasrefs = qemuMigrationJobContinue(vm);
    if (!hasrefs) {
3637 3638
        vm = NULL;
    } else if (!virDomainObjIsActive(vm) && !vm->persistent) {
3639
        qemuDomainRemoveInactive(driver, vm);
3640
        vm = NULL;
3641
    }
3642 3643 3644

cleanup:
    if (vm)
3645
        virObjectUnlock(vm);
3646 3647 3648 3649 3650
    if (event)
        qemuDomainEventQueue(driver, event);
    return ret;
}

3651
int
3652
qemuMigrationPerform(virQEMUDriverPtr driver,
3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676
                     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) {
3677 3678
            virReportError(VIR_ERR_OPERATION_INVALID,
                           "%s", _("received unexpected cookie with P2P migration"));
3679 3680 3681 3682 3683 3684 3685 3686 3687
            return -1;
        }

        return qemuMigrationPerformJob(driver, conn, vm, xmlin, dconnuri, uri,
                                       cookiein, cookieinlen, cookieout,
                                       cookieoutlen, flags, dname, resource,
                                       v3proto);
    } else {
        if (dconnuri) {
3688 3689
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("Unexpected dconnuri parameter with non-peer2peer migration"));
3690 3691 3692 3693 3694 3695 3696
            return -1;
        }

        if (v3proto) {
            return qemuMigrationPerformPhase(driver, conn, vm, uri,
                                             cookiein, cookieinlen,
                                             cookieout, cookieoutlen,
3697
                                             flags, resource);
3698 3699 3700 3701 3702 3703 3704 3705
        } else {
            return qemuMigrationPerformJob(driver, conn, vm, xmlin, dconnuri,
                                           uri, cookiein, cookieinlen,
                                           cookieout, cookieoutlen, flags,
                                           dname, resource, v3proto);
        }
    }
}
3706

3707
static int
3708 3709 3710 3711 3712 3713 3714
qemuMigrationVPAssociatePortProfiles(virDomainDefPtr def) {
    int i;
    int last_good_net = -1;
    virDomainNetDefPtr net;

    for (i = 0; i < def->nnets; i++) {
        net = def->nets[i];
3715
        if (virDomainNetGetActualType(net) == VIR_DOMAIN_NET_TYPE_DIRECT) {
3716
            if (virNetDevVPortProfileAssociate(net->ifname,
3717
                                               virDomainNetGetActualVirtPortProfile(net),
3718
                                               &net->mac,
3719
                                               virDomainNetGetActualDirectDev(net),
3720
                                               -1,
3721
                                               def->uuid,
3722 3723
                                               VIR_NETDEV_VPORT_PROFILE_OP_MIGRATE_IN_FINISH,
                                               false) < 0) {
3724 3725 3726
                virReportError(VIR_ERR_OPERATION_FAILED,
                               _("Port profile Associate failed for %s"),
                               net->ifname);
3727
                goto err_exit;
3728 3729
            }
            VIR_DEBUG("Port profile Associate succeeded for %s", net->ifname);
3730

3731
            if (virNetDevMacVLanVPortProfileRegisterCallback(net->ifname, &net->mac,
3732 3733 3734 3735
                                                             virDomainNetGetActualDirectDev(net), def->uuid,
                                                             virDomainNetGetActualVirtPortProfile(net),
                                                             VIR_NETDEV_VPORT_PROFILE_OP_CREATE))
                goto err_exit;
3736 3737 3738 3739
        }
        last_good_net = i;
    }

3740
    return 0;
3741 3742 3743 3744

err_exit:
    for (i = 0; i < last_good_net; i++) {
        net = def->nets[i];
3745
        if (virDomainNetGetActualType(net) == VIR_DOMAIN_NET_TYPE_DIRECT) {
3746
            ignore_value(virNetDevVPortProfileDisassociate(net->ifname,
3747
                                                           virDomainNetGetActualVirtPortProfile(net),
3748
                                                           &net->mac,
3749
                                                           virDomainNetGetActualDirectDev(net),
3750
                                                           -1,
3751
                                                           VIR_NETDEV_VPORT_PROFILE_OP_MIGRATE_IN_FINISH));
3752 3753
        }
    }
3754
    return -1;
3755 3756 3757 3758
}


virDomainPtr
3759
qemuMigrationFinish(virQEMUDriverPtr driver,
3760 3761
                    virConnectPtr dconn,
                    virDomainObjPtr vm,
3762 3763 3764 3765
                    const char *cookiein,
                    int cookieinlen,
                    char **cookieout,
                    int *cookieoutlen,
3766
                    unsigned long flags,
3767 3768
                    int retcode,
                    bool v3proto)
3769 3770 3771 3772
{
    virDomainPtr dom = NULL;
    virDomainEventPtr event = NULL;
    int newVM = 1;
3773
    qemuMigrationCookiePtr mig = NULL;
3774
    virErrorPtr orig_err = NULL;
3775
    int cookie_flags = 0;
J
Jiri Denemark 已提交
3776
    qemuDomainObjPrivatePtr priv = vm->privateData;
3777
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
3778
    virCapsPtr caps = NULL;
3779

3780
    VIR_DEBUG("driver=%p, dconn=%p, vm=%p, cookiein=%s, cookieinlen=%d, "
3781
              "cookieout=%p, cookieoutlen=%p, flags=%lx, retcode=%d",
3782 3783
              driver, dconn, vm, NULLSTR(cookiein), cookieinlen,
              cookieout, cookieoutlen, flags, retcode);
3784

3785 3786 3787
    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
        goto cleanup;

3788
    if (!qemuMigrationJobIsActive(vm, QEMU_ASYNC_JOB_MIGRATION_IN))
3789 3790
        goto cleanup;

3791 3792 3793
    qemuMigrationJobStartPhase(driver, vm,
                               v3proto ? QEMU_MIGRATION_PHASE_FINISH3
                                       : QEMU_MIGRATION_PHASE_FINISH2);
3794

3795 3796
    qemuDomainCleanupRemove(vm, qemuMigrationPrepareCleanup);

3797
    cookie_flags = QEMU_MIGRATION_COOKIE_NETWORK;
3798 3799 3800 3801 3802
    if (flags & VIR_MIGRATE_PERSIST_DEST)
        cookie_flags |= QEMU_MIGRATION_COOKIE_PERSISTENT;

    if (!(mig = qemuMigrationEatCookie(driver, vm, cookiein,
                                       cookieinlen, cookie_flags)))
3803
        goto endjob;
3804 3805 3806 3807 3808

    /* 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
3809
        if (!virDomainObjIsActive(vm) && !(flags & VIR_MIGRATE_OFFLINE)) {
3810 3811
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("guest unexpectedly quit"));
3812
            goto endjob;
3813 3814
        }

L
liguang 已提交
3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827
        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");
3828
        }
3829

3830 3831
        qemuMigrationStopNBDServer(driver, vm, mig);

3832
        if (flags & VIR_MIGRATE_PERSIST_DEST) {
3833
            virDomainDefPtr vmdef;
3834 3835 3836
            if (vm->persistent)
                newVM = 0;
            vm->persistent = 1;
3837 3838 3839
            if (mig->persistent)
                vm->newDef = vmdef = mig->persistent;
            else
3840
                vmdef = virDomainObjGetPersistentDef(caps, vm);
3841
            if (!vmdef || virDomainSaveConfig(cfg->configDir, vmdef) < 0) {
3842 3843 3844 3845 3846 3847 3848 3849 3850
                /* 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.
                 */
3851 3852

                /*
3853 3854
                 * However, in v3 protocol, the source VM is still available
                 * to restart during confirm() step, so we kill it off now.
3855 3856
                 */
                if (v3proto) {
L
liguang 已提交
3857 3858 3859 3860 3861
                    if (!(flags & VIR_MIGRATE_OFFLINE)) {
                        qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_FAILED,
                                        VIR_QEMU_PROCESS_STOP_MIGRATED);
                        virDomainAuditStop(vm, "failed");
                    }
3862 3863
                    if (newVM)
                        vm->persistent = 0;
3864
                }
A
Alex Jia 已提交
3865
                if (!vmdef)
3866 3867
                    virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                                   _("can't get vmdef"));
3868 3869 3870 3871 3872 3873 3874 3875 3876 3877 3878 3879 3880
                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
3881
        if (!(flags & VIR_MIGRATE_PAUSED) && !(flags & VIR_MIGRATE_OFFLINE)) {
3882 3883 3884 3885
            /* 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
3886
            if (qemuProcessStartCPUs(driver, vm, dconn,
3887 3888
                                     VIR_DOMAIN_RUNNING_MIGRATED,
                                     QEMU_ASYNC_JOB_MIGRATION_IN) < 0) {
3889
                if (virGetLastError() == NULL)
3890 3891
                    virReportError(VIR_ERR_INTERNAL_ERROR,
                                   "%s", _("resume operation failed"));
3892 3893 3894 3895 3896
                /* Need to save the current error, in case shutting
                 * down the process overwrites it
                 */
                orig_err = virSaveLastError();

3897 3898 3899 3900 3901 3902 3903 3904 3905
                /*
                 * 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) {
3906 3907
                    qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_FAILED,
                                    VIR_QEMU_PROCESS_STOP_MIGRATED);
3908
                    virDomainAuditStop(vm, "failed");
3909 3910 3911 3912
                    event = virDomainEventNewFromObj(vm,
                                                     VIR_DOMAIN_EVENT_STOPPED,
                                                     VIR_DOMAIN_EVENT_STOPPED_FAILED);
                }
3913 3914 3915 3916
                goto endjob;
            }
        }

3917
        dom = virGetDomain(dconn, vm->def->name, vm->def->uuid);
3918

L
liguang 已提交
3919
        if (!(flags & VIR_MIGRATE_OFFLINE)) {
3920
            event = virDomainEventNewFromObj(vm,
L
liguang 已提交
3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931
                                             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);
            }
3932
        }
L
liguang 已提交
3933 3934

        if (virDomainObjIsActive(vm) &&
3935
            virDomainSaveStatus(caps, cfg->stateDir, vm) < 0) {
3936 3937 3938
            VIR_WARN("Failed to save status on vm %s", vm->def->name);
            goto endjob;
        }
3939 3940 3941

        /* Guest is successfully running, so cancel previous auto destroy */
        qemuProcessAutoDestroyRemove(driver, vm);
L
liguang 已提交
3942
    } else if (!(flags & VIR_MIGRATE_OFFLINE)) {
3943 3944
        qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_FAILED,
                        VIR_QEMU_PROCESS_STOP_MIGRATED);
3945
        virDomainAuditStop(vm, "failed");
3946 3947 3948 3949 3950
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         VIR_DOMAIN_EVENT_STOPPED_FAILED);
    }

3951 3952 3953
    if (qemuMigrationBakeCookie(mig, driver, vm, cookieout, cookieoutlen, 0) < 0)
        VIR_WARN("Unable to encode migration cookie");

3954
endjob:
E
Eric Blake 已提交
3955 3956 3957
    if (qemuMigrationJobFinish(driver, vm) == 0) {
        vm = NULL;
    } else if (!vm->persistent && !virDomainObjIsActive(vm)) {
3958
        qemuDomainRemoveInactive(driver, vm);
E
Eric Blake 已提交
3959
        vm = NULL;
3960
    }
3961 3962

cleanup:
J
Jiri Denemark 已提交
3963 3964
    if (vm) {
        VIR_FREE(priv->origname);
3965
        virObjectUnlock(vm);
J
Jiri Denemark 已提交
3966
    }
3967 3968
    if (event)
        qemuDomainEventQueue(driver, event);
3969
    qemuMigrationCookieFree(mig);
3970 3971 3972 3973
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
3974
    virObjectUnref(caps);
3975
    virObjectUnref(cfg);
3976 3977
    return dom;
}
3978

3979

3980
int qemuMigrationConfirm(virQEMUDriverPtr driver,
3981 3982 3983 3984
                         virConnectPtr conn,
                         virDomainObjPtr vm,
                         const char *cookiein,
                         int cookieinlen,
E
Eric Blake 已提交
3985
                         unsigned int flags,
3986
                         int retcode)
3987 3988 3989 3990
{
    qemuMigrationCookiePtr mig;
    virDomainEventPtr event = NULL;
    int rv = -1;
3991
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
3992
    virCapsPtr caps = NULL;
3993

3994
    VIR_DEBUG("driver=%p, conn=%p, vm=%p, cookiein=%s, cookieinlen=%d, "
3995
              "flags=%x, retcode=%d",
3996 3997
              driver, conn, vm, NULLSTR(cookiein), cookieinlen,
              flags, retcode);
3998

3999
    virCheckFlags(QEMU_MIGRATION_FLAGS, -1);
E
Eric Blake 已提交
4000

4001 4002 4003
    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
        goto cleanup;

4004 4005 4006 4007 4008
    qemuMigrationJobSetPhase(driver, vm,
                             retcode == 0
                             ? QEMU_MIGRATION_PHASE_CONFIRM3
                             : QEMU_MIGRATION_PHASE_CONFIRM3_CANCELLED);

4009
    if (!(mig = qemuMigrationEatCookie(driver, vm, cookiein, cookieinlen, 0)))
4010
        goto cleanup;
4011

L
liguang 已提交
4012 4013 4014
    if (flags & VIR_MIGRATE_OFFLINE)
        goto done;

4015 4016 4017 4018
    /* Did the migration go as planned?  If yes, kill off the
     * domain object, but if no, resume CPUs
     */
    if (retcode == 0) {
4019 4020
        qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_MIGRATED,
                        VIR_QEMU_PROCESS_STOP_MIGRATED);
4021
        virDomainAuditStop(vm, "migrated");
4022 4023 4024 4025 4026 4027 4028 4029 4030 4031 4032

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

        /* 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,
4033 4034
                                 VIR_DOMAIN_RUNNING_MIGRATED,
                                 QEMU_ASYNC_JOB_MIGRATION_OUT) < 0) {
4035
            if (virGetLastError() == NULL)
4036 4037
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               "%s", _("resume operation failed"));
4038
            goto cleanup;
4039 4040 4041 4042 4043
        }

        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_MIGRATED);
4044
        if (virDomainSaveStatus(caps, cfg->stateDir, vm) < 0) {
4045
            VIR_WARN("Failed to save status on vm %s", vm->def->name);
4046
            goto cleanup;
4047 4048 4049
        }
    }

L
liguang 已提交
4050
done:
4051 4052 4053 4054 4055 4056
    qemuMigrationCookieFree(mig);
    rv = 0;

cleanup:
    if (event)
        qemuDomainEventQueue(driver, event);
4057
    virObjectUnref(caps);
4058
    virObjectUnref(cfg);
4059 4060 4061 4062
    return rv;
}


4063
/* Helper function called while vm is active.  */
4064
int
4065
qemuMigrationToFile(virQEMUDriverPtr driver, virDomainObjPtr vm,
4066 4067
                    int fd, off_t offset, const char *path,
                    const char *compressor,
E
Eric Blake 已提交
4068
                    bool bypassSecurityDriver,
4069
                    enum qemuDomainAsyncJob asyncJob)
4070 4071 4072 4073 4074 4075
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virCgroupPtr cgroup = NULL;
    int ret = -1;
    int rc;
    bool restoreLabel = false;
4076 4077
    virCommandPtr cmd = NULL;
    int pipeFD[2] = { -1, -1 };
4078
    unsigned long saveMigBandwidth = priv->migMaxBandwidth;
4079
    char *errbuf = NULL;
4080 4081 4082 4083 4084

    /* 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,
4085 4086
                                     QEMU_DOMAIN_MIG_BANDWIDTH_MAX);
        priv->migMaxBandwidth = QEMU_DOMAIN_MIG_BANDWIDTH_MAX;
4087
        qemuDomainObjExitMonitor(driver, vm);
4088
    }
4089

4090
    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_FD) &&
4091
        (!compressor || pipe(pipeFD) == 0)) {
4092
        /* All right! We can use fd migration, which means that qemu
4093 4094 4095
         * 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.  */
4096
        if (virSecurityManagerSetImageFDLabel(driver->securityManager, vm->def,
4097
                                              compressor ? pipeFD[1] : fd) < 0)
4098
            goto cleanup;
4099 4100 4101
        bypassSecurityDriver = true;
    } else {
        /* Phooey - we have to fall back on exec migration, where qemu
E
Eric Blake 已提交
4102 4103
         * has to popen() the file by name, and block devices have to be
         * given cgroup ACL permission.  We might also stumble on
4104 4105
         * a race present in some qemu versions where it does a wait()
         * that botches pclose.  */
E
Eric Blake 已提交
4106
        if (qemuCgroupControllerActive(driver,
4107 4108 4109
                                       VIR_CGROUP_CONTROLLER_DEVICES)) {
            if (virCgroupForDomain(driver->cgroup, vm->def->name,
                                   &cgroup, 0) != 0) {
4110 4111 4112
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("Unable to find cgroup for %s"),
                               vm->def->name);
4113 4114 4115 4116
                goto cleanup;
            }
            rc = virCgroupAllowDevicePath(cgroup, path,
                                          VIR_CGROUP_DEVICE_RW);
4117
            virDomainAuditCgroupPath(vm, cgroup, "allow", path, "rw", rc);
E
Eric Blake 已提交
4118 4119 4120 4121
            if (rc == 1) {
                /* path was not a device, no further need for cgroup */
                virCgroupFree(&cgroup);
            } else if (rc < 0) {
4122 4123 4124 4125 4126
                virReportSystemError(-rc,
                                     _("Unable to allow device %s for %s"),
                                     path, vm->def->name);
                goto cleanup;
            }
4127
        }
4128 4129
        if ((!bypassSecurityDriver) &&
            virSecurityManagerSetSavedStateLabel(driver->securityManager,
4130
                                                 vm->def, path) < 0)
4131
            goto cleanup;
4132
        restoreLabel = true;
4133 4134
    }

4135
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
4136 4137
        goto cleanup;

4138 4139 4140
    if (!compressor) {
        const char *args[] = { "cat", NULL };

4141
        if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_FD) &&
4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157
            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
        };
4158 4159 4160 4161
        if (pipeFD[0] != -1) {
            cmd = virCommandNewArgs(args);
            virCommandSetInputFD(cmd, pipeFD[0]);
            virCommandSetOutputFD(cmd, &fd);
4162 4163
            virCommandSetErrorBuffer(cmd, &errbuf);
            virCommandDoAsyncIO(cmd);
4164 4165 4166
            if (virSetCloseExec(pipeFD[1]) < 0) {
                virReportSystemError(errno, "%s",
                                     _("Unable to set cloexec flag"));
4167
                qemuDomainObjExitMonitor(driver, vm);
4168 4169 4170
                goto cleanup;
            }
            if (virCommandRunAsync(cmd, NULL) < 0) {
4171
                qemuDomainObjExitMonitor(driver, vm);
4172 4173 4174 4175 4176 4177 4178
                goto cleanup;
            }
            rc = qemuMonitorMigrateToFd(priv->mon,
                                        QEMU_MONITOR_MIGRATE_BACKGROUND,
                                        pipeFD[1]);
            if (VIR_CLOSE(pipeFD[0]) < 0 ||
                VIR_CLOSE(pipeFD[1]) < 0)
4179
                VIR_WARN("failed to close intermediate pipe");
4180 4181 4182 4183 4184
        } else {
            rc = qemuMonitorMigrateToFile(priv->mon,
                                          QEMU_MONITOR_MIGRATE_BACKGROUND,
                                          args, path, offset);
        }
4185
    }
4186
    qemuDomainObjExitMonitor(driver, vm);
4187 4188 4189 4190

    if (rc < 0)
        goto cleanup;

4191
    rc = qemuMigrationWaitForCompletion(driver, vm, asyncJob, NULL);
4192 4193 4194 4195

    if (rc < 0)
        goto cleanup;

4196 4197 4198
    if (cmd && virCommandWait(cmd, NULL) < 0)
        goto cleanup;

4199 4200 4201
    ret = 0;

cleanup:
4202 4203 4204 4205
    /* Restore max migration bandwidth */
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) == 0) {
        qemuMonitorSetMigrationSpeed(priv->mon, saveMigBandwidth);
        priv->migMaxBandwidth = saveMigBandwidth;
4206
        qemuDomainObjExitMonitor(driver, vm);
4207 4208
    }

4209 4210
    VIR_FORCE_CLOSE(pipeFD[0]);
    VIR_FORCE_CLOSE(pipeFD[1]);
4211 4212 4213 4214 4215
    if (cmd) {
        VIR_DEBUG("Compression binary stderr: %s", NULLSTR(errbuf));
        VIR_FREE(errbuf);
        virCommandFree(cmd);
    }
4216 4217
    if (restoreLabel && (!bypassSecurityDriver) &&
        virSecurityManagerRestoreSavedStateLabel(driver->securityManager,
4218
                                                 vm->def, path) < 0)
4219 4220 4221 4222 4223
        VIR_WARN("failed to restore save state label on %s", path);

    if (cgroup != NULL) {
        rc = virCgroupDenyDevicePath(cgroup, path,
                                     VIR_CGROUP_DEVICE_RWM);
4224
        virDomainAuditCgroupPath(vm, cgroup, "deny", path, "rwm", rc);
4225 4226 4227 4228 4229 4230 4231
        if (rc < 0)
            VIR_WARN("Unable to deny device %s for %s %d",
                     path, vm->def->name, rc);
        virCgroupFree(&cgroup);
    }
    return ret;
}
4232 4233

int
4234
qemuMigrationJobStart(virQEMUDriverPtr driver,
4235 4236 4237 4238 4239
                      virDomainObjPtr vm,
                      enum qemuDomainAsyncJob job)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

4240
    if (qemuDomainObjBeginAsyncJob(driver, vm, job) < 0)
4241 4242
        return -1;

4243
    if (job == QEMU_ASYNC_JOB_MIGRATION_IN) {
4244
        qemuDomainObjSetAsyncJobMask(vm, QEMU_JOB_NONE);
4245 4246
    } else {
        qemuDomainObjSetAsyncJobMask(vm, DEFAULT_JOB_MASK |
4247
                                     JOB_MASK(QEMU_JOB_SUSPEND) |
4248 4249
                                     JOB_MASK(QEMU_JOB_MIGRATION_OP));
    }
4250 4251 4252 4253 4254 4255 4256

    priv->job.info.type = VIR_DOMAIN_JOB_UNBOUNDED;

    return 0;
}

void
4257
qemuMigrationJobSetPhase(virQEMUDriverPtr driver,
4258 4259 4260 4261 4262 4263 4264 4265 4266 4267 4268 4269 4270 4271 4272 4273
                         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
4274
qemuMigrationJobStartPhase(virQEMUDriverPtr driver,
4275 4276 4277
                           virDomainObjPtr vm,
                           enum qemuMigrationJobPhase phase)
{
4278
    virObjectRef(vm);
4279 4280 4281
    qemuMigrationJobSetPhase(driver, vm, phase);
}

4282
bool
4283 4284
qemuMigrationJobContinue(virDomainObjPtr vm)
{
4285
    qemuDomainObjReleaseAsyncJob(vm);
4286
    return virObjectUnref(vm);
4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302
}

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");

4303
        virReportError(VIR_ERR_OPERATION_INVALID, msg, vm->def->name);
4304 4305 4306 4307 4308
        return false;
    }
    return true;
}

4309
bool
4310
qemuMigrationJobFinish(virQEMUDriverPtr driver, virDomainObjPtr vm)
4311 4312 4313
{
    return qemuDomainObjEndAsyncJob(driver, vm);
}