You need to sign in or sign up before continuing.
qemu_migration.c 125.3 KB
Newer Older
1

2 3 4
/*
 * qemu_migration.c: QEMU migration handling
 *
M
Martin Kletzander 已提交
5
 * Copyright (C) 2006-2012 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 27
#include <gnutls/gnutls.h>
#include <gnutls/x509.h>
28
#include <fcntl.h>
29
#include <poll.h>
30 31 32 33 34 35

#include "qemu_migration.h"
#include "qemu_monitor.h"
#include "qemu_domain.h"
#include "qemu_process.h"
#include "qemu_capabilities.h"
36
#include "qemu_cgroup.h"
37

38
#include "domain_audit.h"
39 40 41 42
#include "logging.h"
#include "virterror_internal.h"
#include "memory.h"
#include "util.h"
E
Eric Blake 已提交
43
#include "virfile.h"
44 45
#include "datatypes.h"
#include "fdstream.h"
46
#include "uuid.h"
47
#include "virtime.h"
48
#include "locking/domain_lock.h"
49
#include "rpc/virnetsocket.h"
50
#include "storage_file.h"
M
Martin Kletzander 已提交
51
#include "viruri.h"
J
Jiri Denemark 已提交
52
#include "hooks.h"
53

54 55 56

#define VIR_FROM_THIS VIR_FROM_QEMU

57 58 59 60 61 62 63 64 65 66 67 68 69
VIR_ENUM_IMPL(qemuMigrationJobPhase, QEMU_MIGRATION_PHASE_LAST,
              "none",
              "perform2",
              "begin3",
              "perform3",
              "perform3_done",
              "confirm3_cancelled",
              "confirm3",
              "prepare",
              "finish2",
              "finish3",
);

70
enum qemuMigrationCookieFlags {
71
    QEMU_MIGRATION_COOKIE_FLAG_GRAPHICS,
72
    QEMU_MIGRATION_COOKIE_FLAG_LOCKSTATE,
73
    QEMU_MIGRATION_COOKIE_FLAG_PERSISTENT,
74
    QEMU_MIGRATION_COOKIE_FLAG_NETWORK,
75 76 77 78 79 80 81

    QEMU_MIGRATION_COOKIE_FLAG_LAST
};

VIR_ENUM_DECL(qemuMigrationCookieFlag);
VIR_ENUM_IMPL(qemuMigrationCookieFlag,
              QEMU_MIGRATION_COOKIE_FLAG_LAST,
82
              "graphics", "lockstate", "persistent", "network");
83 84 85

enum qemuMigrationCookieFeatures {
    QEMU_MIGRATION_COOKIE_GRAPHICS  = (1 << QEMU_MIGRATION_COOKIE_FLAG_GRAPHICS),
86
    QEMU_MIGRATION_COOKIE_LOCKSTATE = (1 << QEMU_MIGRATION_COOKIE_FLAG_LOCKSTATE),
87
    QEMU_MIGRATION_COOKIE_PERSISTENT = (1 << QEMU_MIGRATION_COOKIE_FLAG_PERSISTENT),
88
    QEMU_MIGRATION_COOKIE_NETWORK = (1 << QEMU_MIGRATION_COOKIE_FLAG_NETWORK),
89 90 91 92 93 94 95 96 97 98 99 100
};

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

101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121
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;
};

122 123 124
typedef struct _qemuMigrationCookie qemuMigrationCookie;
typedef qemuMigrationCookie *qemuMigrationCookiePtr;
struct _qemuMigrationCookie {
E
Eric Blake 已提交
125 126
    unsigned int flags;
    unsigned int flagsMandatory;
127 128

    /* Host properties */
129 130 131 132
    unsigned char localHostuuid[VIR_UUID_BUFLEN];
    unsigned char remoteHostuuid[VIR_UUID_BUFLEN];
    char *localHostname;
    char *remoteHostname;
133 134 135 136

    /* Guest properties */
    unsigned char uuid[VIR_UUID_BUFLEN];
    char *name;
137

138 139 140 141
    /* If (flags & QEMU_MIGRATION_COOKIE_LOCKSTATE) */
    char *lockState;
    char *lockDriver;

142 143
    /* If (flags & QEMU_MIGRATION_COOKIE_GRAPHICS) */
    qemuMigrationCookieGraphicsPtr graphics;
144 145 146

    /* If (flags & QEMU_MIGRATION_COOKIE_PERSISTENT) */
    virDomainDefPtr persistent;
147 148 149

    /* If (flags & QEMU_MIGRATION_COOKIE_NETWORK) */
    qemuMigrationCookieNetworkPtr network;
150 151
};

152 153 154 155 156 157 158 159 160
static void qemuMigrationCookieGraphicsFree(qemuMigrationCookieGraphicsPtr grap)
{
    if (!grap)
        return;
    VIR_FREE(grap->listen);
    VIR_FREE(grap->tlsSubject);
    VIR_FREE(grap);
}

161

162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178
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);
}


179 180 181 182 183
static void qemuMigrationCookieFree(qemuMigrationCookiePtr mig)
{
    if (!mig)
        return;

184 185 186
    if (mig->flags & QEMU_MIGRATION_COOKIE_GRAPHICS)
        qemuMigrationCookieGraphicsFree(mig->graphics);

187 188 189
    if (mig->flags & QEMU_MIGRATION_COOKIE_NETWORK)
        qemuMigrationCookieNetworkFree(mig->network);

190 191
    VIR_FREE(mig->localHostname);
    VIR_FREE(mig->remoteHostname);
192
    VIR_FREE(mig->name);
193 194
    VIR_FREE(mig->lockState);
    VIR_FREE(mig->lockDriver);
195 196 197 198
    VIR_FREE(mig);
}


199 200 201 202 203 204 205 206 207 208 209 210 211 212 213
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) {
214 215
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unable to read server cert %s"), certfile);
216 217 218 219 220
        goto error;
    }

    ret = gnutls_x509_crt_init(&cert);
    if (ret < 0) {
221 222 223
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("cannot initialize cert object: %s"),
                       gnutls_strerror(ret));
224 225 226 227 228 229 230 231
        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) {
232 233 234
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("cannot load cert data from %s: %s"),
                       certfile, gnutls_strerror(ret));
235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259
        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;
}


static qemuMigrationCookieGraphicsPtr
260
qemuMigrationCookieGraphicsAlloc(virQEMUDriverPtr driver,
261 262 263 264 265 266 267 268 269 270 271
                                 virDomainGraphicsDefPtr def)
{
    qemuMigrationCookieGraphicsPtr mig = NULL;
    const char *listenAddr;

    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;
272
        listenAddr = virDomainGraphicsListenGetAddress(def, 0);
273 274 275 276 277 278 279 280 281 282 283 284
        if (!listenAddr)
            listenAddr = driver->vncListen;

        if (driver->vncTLS &&
            !(mig->tlsSubject = qemuDomainExtractTLSSubject(driver->vncTLSx509certdir)))
            goto error;
    } else {
        mig->port = def->data.spice.port;
        if (driver->spiceTLS)
            mig->tlsPort = def->data.spice.tlsPort;
        else
            mig->tlsPort = -1;
285
        listenAddr = virDomainGraphicsListenGetAddress(def, 0);
286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305
        if (!listenAddr)
            listenAddr = driver->spiceListen;

        if (driver->spiceTLS &&
            !(mig->tlsSubject = qemuDomainExtractTLSSubject(driver->spiceTLSx509certdir)))
            goto error;
    }
    if (!(mig->listen = strdup(listenAddr)))
        goto no_memory;

    return mig;

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


306
static qemuMigrationCookieNetworkPtr
307
qemuMigrationCookieNetworkAlloc(virQEMUDriverPtr driver ATTRIBUTE_UNUSED,
308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334
                                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:
335
               break;
336
            case VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH:
337 338 339 340 341 342 343 344
                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;
345 346 347 348 349 350 351 352 353
            default:
                break;
            }
        }
    }
    return mig;

no_memory:
    virReportOOMError();
354
error:
355 356 357 358
    qemuMigrationCookieNetworkFree(mig);
    return NULL;
}

359 360 361
static qemuMigrationCookiePtr
qemuMigrationCookieNew(virDomainObjPtr dom)
{
J
Jiri Denemark 已提交
362
    qemuDomainObjPrivatePtr priv = dom->privateData;
363
    qemuMigrationCookiePtr mig = NULL;
J
Jiri Denemark 已提交
364
    const char *name;
365 366 367 368

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

J
Jiri Denemark 已提交
369 370 371 372 373
    if (priv->origname)
        name = priv->origname;
    else
        name = dom->def->name;
    if (!(mig->name = strdup(name)))
374 375 376
        goto no_memory;
    memcpy(mig->uuid, dom->def->uuid, VIR_UUID_BUFLEN);

377
    if (!(mig->localHostname = virGetHostname(NULL)))
378
        goto error;
379
    if (virGetHostUUID(mig->localHostuuid) < 0) {
380 381
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Unable to obtain host UUID"));
382 383 384 385 386 387 388 389 390 391 392 393 394
        goto error;
    }

    return mig;

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


395 396
static int
qemuMigrationCookieAddGraphics(qemuMigrationCookiePtr mig,
397
                               virQEMUDriverPtr driver,
398 399 400
                               virDomainObjPtr dom)
{
    if (mig->flags & QEMU_MIGRATION_COOKIE_GRAPHICS) {
401 402
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Migration graphics data already present"));
403 404 405 406 407
        return -1;
    }

    if (dom->def->ngraphics == 1 &&
        (dom->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC ||
408 409 410 411 412 413
         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;
    }
414 415 416 417 418

    return 0;
}


419 420
static int
qemuMigrationCookieAddLockstate(qemuMigrationCookiePtr mig,
421
                                virQEMUDriverPtr driver,
422 423 424 425 426
                                virDomainObjPtr dom)
{
    qemuDomainObjPrivatePtr priv = dom->privateData;

    if (mig->flags & QEMU_MIGRATION_COOKIE_LOCKSTATE) {
427 428
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Migration lockstate data already present"));
429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452
        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;
}


453 454 455 456 457
static int
qemuMigrationCookieAddPersistent(qemuMigrationCookiePtr mig,
                                 virDomainObjPtr dom)
{
    if (mig->flags & QEMU_MIGRATION_COOKIE_PERSISTENT) {
458 459
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Migration persistent data already present"));
460 461 462 463 464 465 466 467 468 469 470 471 472
        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;
}


473 474
static int
qemuMigrationCookieAddNetwork(qemuMigrationCookiePtr mig,
475
                              virQEMUDriverPtr driver,
476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493
                              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;
}

494

495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512
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");
    }
}


513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543
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");
}


544
static int
545
qemuMigrationCookieXMLFormat(virQEMUDriverPtr driver,
546
                             virBufferPtr buf,
547
                             qemuMigrationCookiePtr mig)
548 549 550
{
    char uuidstr[VIR_UUID_STRING_BUFLEN];
    char hostuuidstr[VIR_UUID_STRING_BUFLEN];
551
    int i;
552 553

    virUUIDFormat(mig->uuid, uuidstr);
554
    virUUIDFormat(mig->localHostuuid, hostuuidstr);
555 556 557 558

    virBufferAsprintf(buf, "<qemu-migration>\n");
    virBufferEscapeString(buf, "  <name>%s</name>\n", mig->name);
    virBufferAsprintf(buf, "  <uuid>%s</uuid>\n", uuidstr);
559
    virBufferEscapeString(buf, "  <hostname>%s</hostname>\n", mig->localHostname);
560
    virBufferAsprintf(buf, "  <hostuuid>%s</hostuuid>\n", hostuuidstr);
561

562 563 564 565 566 567
    for (i = 0 ; i < QEMU_MIGRATION_COOKIE_FLAG_LAST ; i++) {
        if (mig->flagsMandatory & (1 << i))
            virBufferAsprintf(buf, "  <feature name='%s'/>\n",
                              qemuMigrationCookieFlagTypeToString(i));
    }

568 569
    if ((mig->flags & QEMU_MIGRATION_COOKIE_GRAPHICS) &&
        mig->graphics)
570 571
        qemuMigrationCookieGraphicsXMLFormat(buf, mig->graphics);

572 573 574 575 576 577 578 579 580
    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");
    }

581 582
    if ((mig->flags & QEMU_MIGRATION_COOKIE_PERSISTENT) &&
        mig->persistent) {
583
        virBufferAdjustIndent(buf, 2);
584 585 586
        if (qemuDomainDefFormatBuf(driver,
                                   mig->persistent,
                                   VIR_DOMAIN_XML_INACTIVE |
587 588
                                   VIR_DOMAIN_XML_SECURE |
                                   VIR_DOMAIN_XML_MIGRATABLE,
589
                                   buf) < 0)
590 591
            return -1;
        virBufferAdjustIndent(buf, -2);
592 593
    }

594 595 596
    if ((mig->flags & QEMU_MIGRATION_COOKIE_NETWORK) && mig->network)
        qemuMigrationCookieNetworkXMLFormat(buf, mig->network);

597
    virBufferAddLit(buf, "</qemu-migration>\n");
598
    return 0;
599 600 601
}


602
static char *qemuMigrationCookieXMLFormatStr(virQEMUDriverPtr driver,
603
                                             qemuMigrationCookiePtr mig)
604 605 606
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;

607
    if (qemuMigrationCookieXMLFormat(driver, &buf, mig) < 0) {
608 609 610
        virBufferFreeAndReset(&buf);
        return NULL;
    }
611 612 613

    if (virBufferError(&buf)) {
        virReportOOMError();
614
        virBufferFreeAndReset(&buf);
615 616 617 618 619 620 621
        return NULL;
    }

    return virBufferContentAndReset(&buf);
}


622 623 624 625 626 627 628 629 630 631
static qemuMigrationCookieGraphicsPtr
qemuMigrationCookieGraphicsXMLParse(xmlXPathContextPtr ctxt)
{
    qemuMigrationCookieGraphicsPtr grap;
    char *tmp;

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

    if (!(tmp = virXPathString("string(./graphics/@type)", ctxt))) {
632 633
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("missing type attribute in migration data"));
634 635 636
        goto error;
    }
    if ((grap->type = virDomainGraphicsTypeFromString(tmp)) < 0) {
637 638
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unknown graphics type %s"), tmp);
639 640 641
        VIR_FREE(tmp);
        goto error;
    }
E
Eric Blake 已提交
642
    VIR_FREE(tmp);
643
    if (virXPathInt("string(./graphics/@port)", ctxt, &grap->port) < 0) {
644 645
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("missing port attribute in migration data"));
646 647 648 649
        goto error;
    }
    if (grap->type == VIR_DOMAIN_GRAPHICS_TYPE_SPICE) {
        if (virXPathInt("string(./graphics/@tlsPort)", ctxt, &grap->tlsPort) < 0) {
650 651
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("missing tlsPort attribute in migration data"));
652 653 654 655
            goto error;
        }
    }
    if (!(grap->listen = virXPathString("string(./graphics/@listen)", ctxt))) {
656 657
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("missing listen attribute in migration data"));
658 659 660
        goto error;
    }
    /* Optional */
661
    grap->tlsSubject = virXPathString("string(./graphics/cert[@info='subject']/@value)", ctxt);
662 663 664 665 666 667 668 669 670 671 672

    return grap;

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


673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692
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;
693
    if (VIR_ALLOC_N(optr->net, optr->nnets) < 0)
694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724
        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;
}


725 726
static int
qemuMigrationCookieXMLParse(qemuMigrationCookiePtr mig,
727
                            virQEMUDriverPtr driver,
728
                            xmlDocPtr doc,
729
                            xmlXPathContextPtr ctxt,
E
Eric Blake 已提交
730
                            unsigned int flags)
731 732 733
{
    char uuidstr[VIR_UUID_STRING_BUFLEN];
    char *tmp;
734 735
    xmlNodePtr *nodes = NULL;
    int i, n;
736 737 738 739 740 741 742 743

    /* 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))) {
744 745
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("missing name element in migration data"));
746 747 748
        goto error;
    }
    if (STRNEQ(tmp, mig->name)) {
749 750 751
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Incoming cookie data had unexpected name %s vs %s"),
                       tmp, mig->name);
752 753 754 755 756 757 758
        goto error;
    }
    VIR_FREE(tmp);

    /* Extract domain uuid */
    tmp = virXPathString("string(./uuid[1])", ctxt);
    if (!tmp) {
759 760
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("missing uuid element in migration data"));
761 762 763 764
        goto error;
    }
    virUUIDFormat(mig->uuid, uuidstr);
    if (STRNEQ(tmp, uuidstr)) {
765 766 767
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Incoming cookie data had unexpected UUID %s vs %s"),
                       tmp, uuidstr);
768 769 770 771
    }
    VIR_FREE(tmp);

    /* Check & forbid "localhost" migration */
772
    if (!(mig->remoteHostname = virXPathString("string(./hostname[1])", ctxt))) {
773 774
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("missing hostname element in migration data"));
775 776
        goto error;
    }
777
    if (STREQ(mig->remoteHostname, mig->localHostname)) {
778 779 780
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Attempt to migrate guest to the same host %s"),
                       mig->remoteHostname);
781 782 783 784
        goto error;
    }

    if (!(tmp = virXPathString("string(./hostuuid[1])", ctxt))) {
785 786 787
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("missing hostuuid element in migration data"));
        goto error;
788
    }
789 790 791 792 793 794 795 796 797 798 799 800
    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);
801

802 803 804 805 806 807 808 809 810
    /* Check to ensure all mandatory features from XML are also
     * present in 'flags' */
    if ((n = virXPathNodeSet("./features", ctxt, &nodes)) < 0)
        goto error;

    for (i = 0 ; i < n ; i++) {
        int val;
        char *str = virXMLPropString(nodes[i], "name");
        if (!str) {
811 812
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("missing feature name"));
813 814 815 816
            goto error;
        }

        if ((val = qemuMigrationCookieFlagTypeFromString(str)) < 0) {
817 818 819
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Unknown migration cookie feature %s"),
                           str);
820 821 822 823 824
            VIR_FREE(str);
            goto error;
        }

        if ((flags & (1 << val)) == 0) {
825 826 827
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Unsupported migration cookie feature %s"),
                           str);
828 829 830 831 832 833
            VIR_FREE(str);
        }
        VIR_FREE(str);
    }
    VIR_FREE(nodes);

834 835 836 837 838
    if ((flags & QEMU_MIGRATION_COOKIE_GRAPHICS) &&
        virXPathBoolean("count(./graphics) > 0", ctxt) &&
        (!(mig->graphics = qemuMigrationCookieGraphicsXMLParse(ctxt))))
        goto error;

839 840 841 842
    if ((flags & QEMU_MIGRATION_COOKIE_LOCKSTATE) &&
        virXPathBoolean("count(./lockstate) > 0", ctxt)) {
        mig->lockDriver = virXPathString("string(./lockstate[1]/@driver)", ctxt);
        if (!mig->lockDriver) {
843 844
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("Missing lock driver name in migration cookie"));
845 846 847 848 849 850 851
            goto error;
        }
        mig->lockState = virXPathString("string(./lockstate[1]/leases[1])", ctxt);
        if (mig->lockState && STREQ(mig->lockState, ""))
            VIR_FREE(mig->lockState);
    }

852 853 854
    if ((flags & QEMU_MIGRATION_COOKIE_PERSISTENT) &&
        virXPathBoolean("count(./domain) > 0", ctxt)) {
        if ((n = virXPathNodeSet("./domain", ctxt, &nodes)) > 1) {
855 856 857 858
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Too many domain elements in "
                             "migration cookie: %d"),
                           n);
859 860 861 862 863 864 865 866 867 868 869 870
            goto error;
        }
        mig->persistent = virDomainDefParseNode(driver->caps, doc, nodes[0],
                                                -1, VIR_DOMAIN_XML_INACTIVE);
        if (!mig->persistent) {
            /* virDomainDefParseNode already reported
             * an error for us */
            goto error;
        }
        VIR_FREE(nodes);
    }

871 872 873 874 875
    if ((flags & QEMU_MIGRATION_COOKIE_NETWORK) &&
        virXPathBoolean("count(./network) > 0", ctxt) &&
        (!(mig->network = qemuMigrationCookieNetworkXMLParse(ctxt))))
        goto error;

876 877 878 879
    return 0;

error:
    VIR_FREE(tmp);
880
    VIR_FREE(nodes);
881 882 883 884 885 886
    return -1;
}


static int
qemuMigrationCookieXMLParseStr(qemuMigrationCookiePtr mig,
887
                               virQEMUDriverPtr driver,
888
                               const char *xml,
E
Eric Blake 已提交
889
                               unsigned int flags)
890 891 892
{
    xmlDocPtr doc = NULL;
    xmlXPathContextPtr ctxt = NULL;
893
    int ret = -1;
894 895 896

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

897
    if (!(doc = virXMLParseStringCtxt(xml, _("(qemu_migration_cookie)"), &ctxt)))
898 899
        goto cleanup;

900
    ret = qemuMigrationCookieXMLParse(mig, driver, doc, ctxt, flags);
901 902 903 904 905 906 907 908 909 910 911

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

    return ret;
}


static int
qemuMigrationBakeCookie(qemuMigrationCookiePtr mig,
912
                        virQEMUDriverPtr driver,
913
                        virDomainObjPtr dom,
914 915
                        char **cookieout,
                        int *cookieoutlen,
E
Eric Blake 已提交
916
                        unsigned int flags)
917
{
918 919
    if (!cookieout || !cookieoutlen)
        return 0;
920 921 922

    *cookieoutlen = 0;

923 924 925 926
    if (flags & QEMU_MIGRATION_COOKIE_GRAPHICS &&
        qemuMigrationCookieAddGraphics(mig, driver, dom) < 0)
        return -1;

927 928 929 930
    if (flags & QEMU_MIGRATION_COOKIE_LOCKSTATE &&
        qemuMigrationCookieAddLockstate(mig, driver, dom) < 0)
        return -1;

931 932 933 934
    if (flags & QEMU_MIGRATION_COOKIE_PERSISTENT &&
        qemuMigrationCookieAddPersistent(mig, dom) < 0)
        return -1;

935 936 937 938 939
    if (flags & QEMU_MIGRATION_COOKIE_NETWORK &&
        qemuMigrationCookieAddNetwork(mig, driver, dom) < 0) {
        return -1;
    }

940
    if (!(*cookieout = qemuMigrationCookieXMLFormatStr(driver, mig)))
941 942 943 944 945 946 947 948 949 950 951
        return -1;

    *cookieoutlen = strlen(*cookieout) + 1;

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

    return 0;
}


static qemuMigrationCookiePtr
952
qemuMigrationEatCookie(virQEMUDriverPtr driver,
953
                       virDomainObjPtr dom,
954 955
                       const char *cookiein,
                       int cookieinlen,
E
Eric Blake 已提交
956
                       unsigned int flags)
957 958 959 960 961 962
{
    qemuMigrationCookiePtr mig = NULL;

    /* Parse & validate incoming cookie (if any) */
    if (cookiein && cookieinlen &&
        cookiein[cookieinlen-1] != '\0') {
963 964
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Migration cookie was not NULL terminated"));
965 966 967 968 969 970 971 972 973 974
        goto error;
    }

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

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

    if (cookiein && cookieinlen &&
        qemuMigrationCookieXMLParseStr(mig,
975
                                       driver,
976 977 978 979
                                       cookiein,
                                       flags) < 0)
        goto error;

980 981 982
    if (mig->flags & QEMU_MIGRATION_COOKIE_LOCKSTATE) {
        if (!mig->lockDriver) {
            if (virLockManagerPluginUsesState(driver->lockManager)) {
983 984 985
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("Missing %s lock state for migration cookie"),
                               virLockManagerPluginGetName(driver->lockManager));
986 987 988 989
                goto error;
            }
        } else if (STRNEQ(mig->lockDriver,
                          virLockManagerPluginGetName(driver->lockManager))) {
990 991 992 993
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Source host lock driver %s different from target %s"),
                           mig->lockDriver,
                           virLockManagerPluginGetName(driver->lockManager));
994 995 996 997
            goto error;
        }
    }

998 999 1000 1001 1002 1003
    return mig;

error:
    qemuMigrationCookieFree(mig);
    return NULL;
}
1004

1005 1006 1007 1008 1009 1010 1011 1012 1013 1014
/* 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. */
static bool
1015
qemuMigrationIsAllowed(virQEMUDriverPtr driver, virDomainObjPtr vm,
1016
                       virDomainDefPtr def)
1017
{
1018
    int nsnapshots;
1019 1020
    bool forbid;
    int i;
1021 1022 1023

    if (vm) {
        if (qemuProcessAutoDestroyActive(driver, vm)) {
1024 1025
            virReportError(VIR_ERR_OPERATION_INVALID,
                           "%s", _("domain is marked for auto destroy"));
1026 1027
            return false;
        }
1028
        if ((nsnapshots = virDomainSnapshotObjListNum(vm->snapshots, NULL,
1029
                                                      0))) {
1030 1031 1032
            virReportError(VIR_ERR_OPERATION_INVALID,
                           _("cannot migrate domain with %d snapshots"),
                           nsnapshots);
1033 1034
            return false;
        }
E
Eric Blake 已提交
1035 1036 1037 1038 1039
        if (virDomainHasDiskMirror(vm)) {
            virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                           _("cannot migrate domain with active block job"));
            return false;
        }
1040 1041 1042

        def = vm->def;
    }
1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059

    /* 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",
                       _("Domain with assigned non-USB host devices "
                         "cannot be migrated"));
1060 1061 1062 1063 1064 1065
        return false;
    }

    return true;
}

1066 1067 1068 1069 1070 1071 1072 1073
static bool
qemuMigrationIsSafe(virDomainDefPtr def)
{
    int i;

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

1074 1075
        /* Our code elsewhere guarantees shared disks are either readonly (in
         * which case cache mode doesn't matter) or used with cache=none */
1076
        if (disk->src &&
1077 1078 1079
            !disk->shared &&
            !disk->readonly &&
            disk->cachemode != VIR_DOMAIN_DISK_CACHE_DISABLE) {
1080
            int cfs;
1081 1082 1083 1084 1085 1086

            if (disk->type == VIR_DOMAIN_DISK_TYPE_FILE) {
                if ((cfs = virStorageFileIsClusterFS(disk->src)) == 1)
                    continue;
                else if (cfs < 0)
                    return false;
1087 1088 1089
            } else if (disk->type == VIR_DOMAIN_DISK_TYPE_NETWORK &&
                       disk->protocol == VIR_DOMAIN_DISK_PROTOCOL_RBD) {
                continue;
1090
            }
1091

1092 1093 1094
            virReportError(VIR_ERR_MIGRATE_UNSAFE, "%s",
                           _("Migration may lead to data corruption if disks"
                             " use cache != none"));
1095 1096 1097 1098 1099 1100 1101
            return false;
        }
    }

    return true;
}

1102 1103 1104 1105
/** qemuMigrationSetOffline
 * Pause domain for non-live migration.
 */
int
1106
qemuMigrationSetOffline(virQEMUDriverPtr driver,
1107 1108 1109
                        virDomainObjPtr vm)
{
    int ret;
1110
    VIR_DEBUG("driver=%p vm=%p", driver, vm);
1111 1112
    ret = qemuProcessStopCPUs(driver, vm, VIR_DOMAIN_PAUSED_MIGRATION,
                              QEMU_ASYNC_JOB_MIGRATION_OUT);
1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126
    if (ret == 0) {
        virDomainEventPtr event;

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

    return ret;
}


1127
static int
1128
qemuMigrationUpdateJobStatus(virQEMUDriverPtr driver,
1129
                             virDomainObjPtr vm,
1130 1131
                             const char *job,
                             enum qemuDomainAsyncJob asyncJob)
1132
{
1133
    qemuDomainObjPrivatePtr priv = vm->privateData;
1134
    int ret;
1135
    int status;
1136 1137
    bool wait_for_spice = false;
    bool spice_migrated = false;
1138 1139 1140
    unsigned long long memProcessed;
    unsigned long long memRemaining;
    unsigned long long memTotal;
1141

1142 1143 1144 1145 1146 1147 1148
    /* 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 &&
        qemuCapsGet(priv->caps, QEMU_CAPS_SEAMLESS_MIGRATION))
        wait_for_spice = true;

1149 1150 1151
    ret = qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob);
    if (ret < 0) {
        /* Guest already exited; nothing further to update.  */
1152 1153
        return -1;
    }
1154 1155 1156 1157 1158
    ret = qemuMonitorGetMigrationStatus(priv->mon,
                                        &status,
                                        &memProcessed,
                                        &memRemaining,
                                        &memTotal);
1159 1160 1161 1162 1163 1164 1165

    /* If qemu says migrated, check spice */
    if (wait_for_spice && (ret == 0) &&
        (status == QEMU_MONITOR_MIGRATION_STATUS_COMPLETED))
        ret = qemuMonitorGetSpiceMigrationStatus(priv->mon,
                                                 &spice_migrated);

1166
    qemuDomainObjExitMonitorWithDriver(driver, vm);
1167

1168
    if (ret < 0 || virTimeMillisNow(&priv->job.info.timeElapsed) < 0) {
1169
        priv->job.info.type = VIR_DOMAIN_JOB_FAILED;
1170 1171
        return -1;
    }
1172
    priv->job.info.timeElapsed -= priv->job.start;
1173

1174
    ret = -1;
1175 1176
    switch (status) {
    case QEMU_MONITOR_MIGRATION_STATUS_INACTIVE:
1177
        priv->job.info.type = VIR_DOMAIN_JOB_NONE;
1178 1179
        virReportError(VIR_ERR_OPERATION_FAILED,
                       _("%s: %s"), job, _("is not active"));
1180 1181 1182
        break;

    case QEMU_MONITOR_MIGRATION_STATUS_ACTIVE:
1183 1184 1185
        priv->job.info.dataTotal = memTotal;
        priv->job.info.dataRemaining = memRemaining;
        priv->job.info.dataProcessed = memProcessed;
1186

1187 1188 1189
        priv->job.info.memTotal = memTotal;
        priv->job.info.memRemaining = memRemaining;
        priv->job.info.memProcessed = memProcessed;
1190 1191 1192 1193 1194

        ret = 0;
        break;

    case QEMU_MONITOR_MIGRATION_STATUS_COMPLETED:
1195 1196
        if ((wait_for_spice && spice_migrated) || (!wait_for_spice))
            priv->job.info.type = VIR_DOMAIN_JOB_COMPLETED;
1197 1198 1199 1200
        ret = 0;
        break;

    case QEMU_MONITOR_MIGRATION_STATUS_ERROR:
1201
        priv->job.info.type = VIR_DOMAIN_JOB_FAILED;
1202 1203
        virReportError(VIR_ERR_OPERATION_FAILED,
                       _("%s: %s"), job, _("unexpectedly failed"));
1204 1205 1206
        break;

    case QEMU_MONITOR_MIGRATION_STATUS_CANCELLED:
1207
        priv->job.info.type = VIR_DOMAIN_JOB_CANCELLED;
1208 1209
        virReportError(VIR_ERR_OPERATION_ABORTED,
                       _("%s: %s"), job, _("canceled by client"));
1210 1211 1212 1213 1214 1215 1216
        break;
    }

    return ret;
}


1217
static int
1218
qemuMigrationWaitForCompletion(virQEMUDriverPtr driver, virDomainObjPtr vm,
1219 1220
                               enum qemuDomainAsyncJob asyncJob,
                               virConnectPtr dconn)
1221
{
1222
    qemuDomainObjPrivatePtr priv = vm->privateData;
1223 1224
    const char *job;

1225 1226
    switch (priv->job.asyncJob) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
1227 1228
        job = _("migration job");
        break;
1229
    case QEMU_ASYNC_JOB_SAVE:
1230 1231
        job = _("domain save job");
        break;
1232
    case QEMU_ASYNC_JOB_DUMP:
1233 1234 1235 1236 1237
        job = _("domain core dump job");
        break;
    default:
        job = _("job");
    }
1238

1239
    priv->job.info.type = VIR_DOMAIN_JOB_UNBOUNDED;
1240

1241
    while (priv->job.info.type == VIR_DOMAIN_JOB_UNBOUNDED) {
1242 1243 1244
        /* Poll every 50ms for progress & to allow cancellation */
        struct timespec ts = { .tv_sec = 0, .tv_nsec = 50 * 1000 * 1000ull };

1245
        if (qemuMigrationUpdateJobStatus(driver, vm, job, asyncJob) < 0)
1246 1247
            goto cleanup;

1248
        if (dconn && virConnectIsAlive(dconn) <= 0) {
1249 1250
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("Lost connection to destination host"));
1251 1252 1253
            goto cleanup;
        }

1254 1255 1256 1257 1258 1259 1260 1261 1262 1263
        virDomainObjUnlock(vm);
        qemuDriverUnlock(driver);

        nanosleep(&ts, NULL);

        qemuDriverLock(driver);
        virDomainObjLock(vm);
    }

cleanup:
1264
    if (priv->job.info.type == VIR_DOMAIN_JOB_COMPLETED)
1265 1266 1267
        return 0;
    else
        return -1;
1268 1269 1270
}


1271
static int
1272
qemuDomainMigrateGraphicsRelocate(virQEMUDriverPtr driver,
1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290
                                  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;

1291 1292
    ret = qemuDomainObjEnterMonitorAsync(driver, vm,
                                         QEMU_ASYNC_JOB_MIGRATION_OUT);
1293 1294 1295 1296 1297 1298 1299 1300 1301
    if (ret == 0) {
        ret = qemuMonitorGraphicsRelocate(priv->mon,
                                          cookie->graphics->type,
                                          cookie->remoteHostname,
                                          cookie->graphics->port,
                                          cookie->graphics->tlsPort,
                                          cookie->graphics->tlsSubject);
        qemuDomainObjExitMonitorWithDriver(driver, vm);
    }
1302 1303 1304 1305 1306

    return ret;
}


1307
static int
1308
qemuDomainMigrateOPDRelocate(virQEMUDriverPtr driver ATTRIBUTE_UNUSED,
1309 1310 1311
                             virDomainObjPtr vm,
                             qemuMigrationCookiePtr cookie)
{
1312 1313
    virDomainNetDefPtr netptr;
    int ret = -1;
1314 1315 1316 1317 1318 1319 1320 1321 1322
    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:
1323
           break;
1324
        case VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH:
1325 1326 1327 1328 1329 1330 1331 1332
            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;
1333 1334 1335 1336 1337
        default:
            break;
        }
    }

1338 1339
    ret = 0;
cleanup:
1340 1341 1342 1343
    return ret;
}


1344 1345 1346 1347 1348 1349 1350
/* 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
1351
qemuMigrationCleanup(virQEMUDriverPtr driver,
1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404
                     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;
}

1405
/* The caller is supposed to lock the vm and start a migration job. */
1406
char *qemuMigrationBegin(virQEMUDriverPtr driver,
1407
                         virDomainObjPtr vm,
1408
                         const char *xmlin,
1409
                         const char *dname,
1410
                         char **cookieout,
1411 1412
                         int *cookieoutlen,
                         unsigned long flags)
1413 1414 1415
{
    char *rv = NULL;
    qemuMigrationCookiePtr mig = NULL;
1416
    virDomainDefPtr def = NULL;
1417 1418
    qemuDomainObjPrivatePtr priv = vm->privateData;

1419
    VIR_DEBUG("driver=%p, vm=%p, xmlin=%s, dname=%s,"
1420
              " cookieout=%p, cookieoutlen=%p, flags=%lx",
1421
              driver, vm, NULLSTR(xmlin), NULLSTR(dname),
1422
              cookieout, cookieoutlen, flags);
1423

1424 1425 1426 1427 1428 1429
    /* 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);
1430

1431
    if (!qemuMigrationIsAllowed(driver, vm, NULL))
1432 1433
        goto cleanup;

1434 1435 1436
    if (!(flags & VIR_MIGRATE_UNSAFE) && !qemuMigrationIsSafe(vm->def))
        goto cleanup;

1437
    if (!(mig = qemuMigrationEatCookie(driver, vm, NULL, 0, 0)))
1438 1439 1440 1441
        goto cleanup;

    if (qemuMigrationBakeCookie(mig, driver, vm,
                                cookieout, cookieoutlen,
1442
                                QEMU_MIGRATION_COOKIE_LOCKSTATE) < 0)
1443 1444
        goto cleanup;

L
liguang 已提交
1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466
    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;
        }
    }

1467 1468
    if (xmlin) {
        if (!(def = virDomainDefParseString(driver->caps, xmlin,
M
Matthias Bolte 已提交
1469
                                            QEMU_EXPECTED_VIRT_TYPES,
1470 1471 1472
                                            VIR_DOMAIN_XML_INACTIVE)))
            goto cleanup;

1473
        if (STRNEQ(def->name, vm->def->name)) {
1474 1475
            virReportError(VIR_ERR_INVALID_ARG, "%s",
                           _("target domain name doesn't match source name"));
1476 1477 1478
            goto cleanup;
        }

1479
        if (!virDomainDefCheckABIStability(vm->def, def))
1480 1481
            goto cleanup;

1482
        rv = qemuDomainDefFormatLive(driver, def, false, true);
1483
    } else {
1484
        rv = qemuDomainDefFormatLive(driver, vm->def, false, true);
1485
    }
1486 1487 1488

cleanup:
    qemuMigrationCookieFree(mig);
1489
    virDomainDefFree(def);
1490 1491 1492 1493
    return rv;
}


1494 1495
/* Prepare is the first step, and it runs on the destination host.
 */
1496

1497
static void
1498
qemuMigrationPrepareCleanup(virQEMUDriverPtr driver,
1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513
                            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);
}

1514
static int
1515
qemuMigrationPrepareAny(virQEMUDriverPtr driver,
1516 1517 1518 1519 1520 1521 1522 1523
                        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
1524 1525
                        virStreamPtr st,
                        unsigned long flags)
1526 1527 1528 1529 1530
{
    virDomainDefPtr def = NULL;
    virDomainObjPtr vm = NULL;
    virDomainEventPtr event = NULL;
    int ret = -1;
1531
    int dataFD[2] = { -1, -1 };
1532
    qemuDomainObjPrivatePtr priv = NULL;
J
Jiri Denemark 已提交
1533
    unsigned long long now;
1534
    qemuMigrationCookiePtr mig = NULL;
1535
    bool tunnel = !!st;
J
Jiri Denemark 已提交
1536
    char *origname = NULL;
J
Jiri Denemark 已提交
1537
    char *xmlout = NULL;
L
liguang 已提交
1538
    unsigned int cookieFlags;
1539

1540
    if (virTimeMillisNow(&now) < 0)
1541 1542
        return -1;

L
liguang 已提交
1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564
    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;
        }
    }

1565
    if (!(def = virDomainDefParseString(driver->caps, dom_xml,
M
Matthias Bolte 已提交
1566
                                        QEMU_EXPECTED_VIRT_TYPES,
1567 1568 1569
                                        VIR_DOMAIN_XML_INACTIVE)))
        goto cleanup;

1570
    if (!qemuMigrationIsAllowed(driver, NULL, def))
1571 1572 1573 1574
        goto cleanup;

    /* Target domain name, maybe renamed. */
    if (dname) {
J
Jiri Denemark 已提交
1575
        origname = def->name;
1576 1577 1578 1579 1580
        def->name = strdup(dname);
        if (def->name == NULL)
            goto cleanup;
    }

J
Jiri Denemark 已提交
1581 1582 1583 1584 1585
    /* Let migration hook filter domain XML */
    if (virHookPresent(VIR_HOOK_DRIVER_QEMU)) {
        char *xml;
        int hookret;

1586
        if (!(xml = qemuDomainDefFormatXML(driver, def,
1587 1588
                                           VIR_DOMAIN_XML_SECURE |
                                           VIR_DOMAIN_XML_MIGRATABLE)))
J
Jiri Denemark 已提交
1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622
            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);
                newdef = virDomainDefParseString(driver->caps, xmlout,
                                                 QEMU_EXPECTED_VIRT_TYPES,
                                                 VIR_DOMAIN_XML_INACTIVE);
                if (!newdef)
                    goto cleanup;

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

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

1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633
    if (virDomainObjIsDuplicate(&driver->domains, def, 1) < 0)
        goto cleanup;

    if (!(vm = virDomainAssignDef(driver->caps,
                                  &driver->domains,
                                  def, true))) {
        /* virDomainAssignDef already set the error */
        goto cleanup;
    }
    def = NULL;
    priv = vm->privateData;
J
Jiri Denemark 已提交
1634 1635
    priv->origname = origname;
    origname = NULL;
1636

1637 1638
    if (!(mig = qemuMigrationEatCookie(driver, vm, cookiein, cookieinlen,
                                       QEMU_MIGRATION_COOKIE_LOCKSTATE)))
1639 1640
        goto cleanup;

1641
    if (qemuMigrationJobStart(driver, vm, QEMU_ASYNC_JOB_MIGRATION_IN) < 0)
1642
        goto cleanup;
1643
    qemuMigrationJobSetPhase(driver, vm, QEMU_MIGRATION_PHASE_PREPARE);
1644 1645 1646 1647

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

L
liguang 已提交
1648 1649 1650
    if (flags & VIR_MIGRATE_OFFLINE)
        goto done;

1651 1652
    if (tunnel &&
        (pipe(dataFD) < 0 || virSetCloseExec(dataFD[1]) < 0)) {
1653 1654
        virReportSystemError(errno, "%s",
                             _("cannot create pipe for tunnelled migration"));
1655 1656 1657 1658
        goto endjob;
    }

    /* Start the QEMU daemon, with the same command-line arguments plus
1659
     * -incoming $migrateFrom
1660
     */
1661 1662 1663 1664
    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) {
1665
        virDomainAuditStart(vm, "migrated", false);
1666 1667 1668 1669 1670 1671
        /* Note that we don't set an error here because qemuProcessStart
         * should have already done that.
         */
        goto endjob;
    }

1672 1673 1674 1675 1676
    if (tunnel) {
        if (virFDStreamOpen(st, dataFD[1]) < 0) {
            virReportSystemError(errno, "%s",
                                 _("cannot pass pipe for tunnelled migration"));
            virDomainAuditStart(vm, "migrated", false);
1677
            qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_FAILED, 0);
1678
            goto endjob;
1679
        }
1680
        dataFD[1] = -1; /* 'st' owns the FD now & will close it */
1681 1682
    }

1683 1684 1685 1686 1687 1688 1689 1690
    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");
    }
1691

L
liguang 已提交
1692 1693 1694 1695 1696 1697
done:
    if (flags & VIR_MIGRATE_OFFLINE)
        cookieFlags = 0;
    else
        cookieFlags = QEMU_MIGRATION_COOKIE_GRAPHICS;

1698
    if (qemuMigrationBakeCookie(mig, driver, vm, cookieout, cookieoutlen,
L
liguang 已提交
1699
                                cookieFlags) < 0) {
1700 1701 1702 1703 1704 1705 1706
        /* 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");
    }

1707 1708 1709
    if (qemuDomainCleanupAdd(vm, qemuMigrationPrepareCleanup) < 0)
        goto endjob;

L
liguang 已提交
1710 1711 1712 1713 1714 1715
    if (!(flags & VIR_MIGRATE_OFFLINE)) {
        virDomainAuditStart(vm, "migrated", true);
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STARTED,
                                         VIR_DOMAIN_EVENT_STARTED_MIGRATED);
    }
1716

1717 1718 1719 1720
    /* 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.
     */
1721
    if (!qemuMigrationJobContinue(vm)) {
1722
        vm = NULL;
1723 1724
        virReportError(VIR_ERR_OPERATION_FAILED,
                       "%s", _("domain disappeared"));
1725
        goto cleanup;
1726
    }
1727

1728
    ret = 0;
1729 1730

cleanup:
J
Jiri Denemark 已提交
1731
    VIR_FREE(origname);
J
Jiri Denemark 已提交
1732
    VIR_FREE(xmlout);
1733
    virDomainDefFree(def);
1734 1735
    VIR_FORCE_CLOSE(dataFD[0]);
    VIR_FORCE_CLOSE(dataFD[1]);
1736 1737 1738 1739 1740 1741
    if (vm) {
        if (ret >= 0 || vm->persistent)
            virDomainObjUnlock(vm);
        else
            qemuDomainRemoveInactive(driver, vm);
    }
1742 1743
    if (event)
        qemuDomainEventQueue(driver, event);
1744
    qemuMigrationCookieFree(mig);
1745
    return ret;
1746 1747

endjob:
1748
    if (!qemuMigrationJobFinish(driver, vm)) {
1749 1750 1751
        vm = NULL;
    }
    goto cleanup;
1752 1753 1754
}


1755 1756 1757 1758 1759
/*
 * This version starts an empty VM listening on a localhost TCP port, and
 * sets up the corresponding virStream to handle the incoming data.
 */
int
1760
qemuMigrationPrepareTunnel(virQEMUDriverPtr driver,
1761 1762 1763 1764 1765 1766 1767
                           virConnectPtr dconn,
                           const char *cookiein,
                           int cookieinlen,
                           char **cookieout,
                           int *cookieoutlen,
                           virStreamPtr st,
                           const char *dname,
L
liguang 已提交
1768 1769
                           const char *dom_xml,
                           unsigned long flags)
1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782
{
    int ret;

    VIR_DEBUG("driver=%p, dconn=%p, cookiein=%s, cookieinlen=%d, "
              "cookieout=%p, cookieoutlen=%p, st=%p, dname=%s, dom_xml=%s",
              driver, dconn, NULLSTR(cookiein), cookieinlen,
              cookieout, cookieoutlen, st, NULLSTR(dname), dom_xml);

    /* 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
1783
                                  "stdio", st, flags);
1784 1785 1786 1787
    return ret;
}


1788
int
1789
qemuMigrationPrepareDirect(virQEMUDriverPtr driver,
1790
                           virConnectPtr dconn,
1791 1792 1793 1794
                           const char *cookiein,
                           int cookieinlen,
                           char **cookieout,
                           int *cookieoutlen,
1795 1796 1797
                           const char *uri_in,
                           char **uri_out,
                           const char *dname,
L
liguang 已提交
1798 1799
                           const char *dom_xml,
                           unsigned long flags)
1800 1801 1802 1803 1804 1805 1806
{
    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
1807

1808 1809 1810 1811 1812 1813
    VIR_DEBUG("driver=%p, dconn=%p, cookiein=%s, cookieinlen=%d, "
              "cookieout=%p, cookieoutlen=%p, uri_in=%s, uri_out=%p, "
              "dname=%s, dom_xml=%s",
              driver, dconn, NULLSTR(cookiein), cookieinlen,
              cookieout, cookieoutlen, NULLSTR(uri_in), uri_out,
              NULLSTR(dname), dom_xml);
1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833

    /* 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")) {
1834 1835 1836
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("hostname on destination resolved to localhost,"
                             " but migration requires an FQDN"));
1837 1838 1839 1840 1841
            goto cleanup;
        }

        /* XXX this really should have been a properly well-formed
         * URI, but we can't add in tcp:// now without breaking
1842
         * compatibility with old targets. We at least make the
1843 1844 1845
         * new targets accept both syntaxes though.
         */
        /* Caller frees */
1846
        if (virAsprintf(uri_out, "tcp:%s:%d", hostname, this_port) < 0) {
1847 1848 1849 1850 1851 1852 1853 1854
            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.
         */
1855
        if (!STRPREFIX(uri_in, "tcp:")) {
1856 1857 1858
            virReportError(VIR_ERR_INVALID_ARG, "%s",
                           _("only tcp URIs are supported for KVM/QEMU"
                             " migrations"));
1859 1860 1861 1862
            goto cleanup;
        }

        /* Get the port number. */
1863
        p = strrchr(uri_in, ':');
1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877
        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 */
1878 1879
            this_port = virParseNumber(&p);
            if (this_port == -1 || p-uri_in != strlen(uri_in)) {
1880 1881
                virReportError(VIR_ERR_INVALID_ARG,
                               "%s", _("URI ended with incorrect ':port'"));
1882 1883 1884 1885 1886 1887 1888 1889
                goto cleanup;
            }
        }
    }

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

1890
    /* QEMU will be started with -incoming tcp:0.0.0.0:port */
1891
    snprintf(migrateFrom, sizeof(migrateFrom), "tcp:0.0.0.0:%d", this_port);
1892

1893 1894
    ret = qemuMigrationPrepareAny(driver, dconn, cookiein, cookieinlen,
                                  cookieout, cookieoutlen, dname, dom_xml,
L
liguang 已提交
1895
                                  migrateFrom, NULL, flags);
1896 1897 1898 1899 1900 1901 1902 1903
cleanup:
    VIR_FREE(hostname);
    if (ret != 0)
        VIR_FREE(*uri_out);
    return ret;
}


1904 1905
enum qemuMigrationDestinationType {
    MIGRATION_DEST_HOST,
1906
    MIGRATION_DEST_CONNECT_HOST,
1907
    MIGRATION_DEST_UNIX,
1908
    MIGRATION_DEST_FD,
1909
};
1910

1911 1912 1913 1914
enum qemuMigrationForwardType {
    MIGRATION_FWD_DIRECT,
    MIGRATION_FWD_STREAM,
};
1915

1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926
typedef struct _qemuMigrationSpec qemuMigrationSpec;
typedef qemuMigrationSpec *qemuMigrationSpecPtr;
struct _qemuMigrationSpec {
    enum qemuMigrationDestinationType destType;
    union {
        struct {
            const char *name;
            int port;
        } host;

        struct {
1927
            char *file;
1928 1929
            int sock;
        } unix_socket;
1930 1931 1932 1933 1934

        struct {
            int qemu;
            int local;
        } fd;
1935 1936 1937 1938 1939 1940 1941
    } dest;

    enum qemuMigrationForwardType fwdType;
    union {
        virStreamPtr stream;
    } fwd;
};
1942 1943 1944

#define TUNNEL_SEND_BUF_SIZE 65536

1945 1946 1947 1948 1949 1950 1951
typedef struct _qemuMigrationIOThread qemuMigrationIOThread;
typedef qemuMigrationIOThread *qemuMigrationIOThreadPtr;
struct _qemuMigrationIOThread {
    virThread thread;
    virStreamPtr st;
    int sock;
    virError err;
1952 1953
    int wakeupRecvFD;
    int wakeupSendFD;
1954 1955 1956
};

static void qemuMigrationIOFunc(void *arg)
1957
{
1958
    qemuMigrationIOThreadPtr data = arg;
1959 1960 1961 1962 1963 1964 1965
    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);
1966 1967 1968

    if (VIR_ALLOC_N(buffer, TUNNEL_SEND_BUF_SIZE) < 0) {
        virReportOOMError();
1969
        goto abrt;
1970 1971
    }

1972 1973 1974
    fds[0].fd = data->sock;
    fds[1].fd = data->wakeupRecvFD;

1975
    for (;;) {
1976 1977 1978 1979 1980 1981 1982 1983 1984 1985
        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;
1986
            virReportSystemError(errno, "%s",
1987 1988
                                 _("poll failed in migration tunnel"));
            goto abrt;
1989
        }
1990 1991 1992 1993 1994 1995 1996

        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");
1997
            break;
1998
        }
1999

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
        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;
            }
2016 2017
        }

2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034
        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;
            }
        }
    }
2035

2036 2037
    if (virStreamFinish(data->st) < 0)
        goto error;
2038

2039 2040
    VIR_FREE(buffer);

2041 2042
    return;

2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054
abrt:
    err = virSaveLastError();
    if (err && err->code == VIR_ERR_OK) {
        virFreeError(err);
        err = NULL;
    }
    virStreamAbort(data->st);
    if (err) {
        virSetError(err);
        virFreeError(err);
    }

2055 2056 2057
error:
    virCopyLastError(&data->err);
    virResetLastError();
2058
    VIR_FREE(buffer);
2059 2060 2061 2062 2063 2064 2065
}


static qemuMigrationIOThreadPtr
qemuMigrationStartTunnel(virStreamPtr st,
                         int sock)
{
2066 2067
    qemuMigrationIOThreadPtr io = NULL;
    int wakeupFD[2] = { -1, -1 };
2068

2069 2070 2071 2072
    if (pipe2(wakeupFD, O_CLOEXEC) < 0) {
        virReportSystemError(errno, "%s",
                             _("Unable to make pipe"));
        goto error;
2073 2074
    }

2075 2076 2077
    if (VIR_ALLOC(io) < 0)
        goto no_memory;

2078 2079
    io->st = st;
    io->sock = sock;
2080 2081
    io->wakeupRecvFD = wakeupFD[0];
    io->wakeupSendFD = wakeupFD[1];
2082 2083 2084 2085 2086 2087

    if (virThreadCreate(&io->thread, true,
                        qemuMigrationIOFunc,
                        io) < 0) {
        virReportSystemError(errno, "%s",
                             _("Unable to create migration thread"));
2088
        goto error;
2089 2090 2091
    }

    return io;
2092 2093 2094 2095 2096 2097 2098 2099

no_memory:
    virReportOOMError();
error:
    VIR_FORCE_CLOSE(wakeupFD[0]);
    VIR_FORCE_CLOSE(wakeupFD[1]);
    VIR_FREE(io);
    return NULL;
2100 2101 2102
}

static int
2103
qemuMigrationStopTunnel(qemuMigrationIOThreadPtr io, bool error)
2104 2105
{
    int rv = -1;
2106 2107 2108 2109 2110 2111 2112 2113 2114
    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;
    }

2115 2116 2117 2118
    virThreadJoin(&io->thread);

    /* Forward error from the IO thread, to this thread */
    if (io->err.code != VIR_ERR_OK) {
2119 2120 2121 2122
        if (error)
            rv = 0;
        else
            virSetError(&io->err);
2123 2124 2125 2126 2127 2128 2129
        virResetError(&io->err);
        goto cleanup;
    }

    rv = 0;

cleanup:
2130 2131
    VIR_FORCE_CLOSE(io->wakeupSendFD);
    VIR_FORCE_CLOSE(io->wakeupRecvFD);
2132 2133
    VIR_FREE(io);
    return rv;
2134 2135
}

2136
static int
2137
qemuMigrationConnect(virQEMUDriverPtr driver,
2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158
                     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);
2159
        virObjectUnref(sock);
2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173
    }
    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;
}

2174
static int
2175
qemuMigrationRun(virQEMUDriverPtr driver,
2176 2177 2178 2179 2180 2181 2182
                 virDomainObjPtr vm,
                 const char *cookiein,
                 int cookieinlen,
                 char **cookieout,
                 int *cookieoutlen,
                 unsigned long flags,
                 unsigned long resource,
2183 2184
                 qemuMigrationSpecPtr spec,
                 virConnectPtr dconn)
2185
{
2186
    int ret = -1;
2187 2188
    unsigned int migrate_flags = QEMU_MONITOR_MIGRATE_BACKGROUND;
    qemuDomainObjPrivatePtr priv = vm->privateData;
2189
    qemuMigrationCookiePtr mig = NULL;
2190
    qemuMigrationIOThreadPtr iothread = NULL;
2191
    int fd = -1;
2192
    unsigned long migrate_speed = resource ? resource : priv->migMaxBandwidth;
2193
    virErrorPtr orig_err = NULL;
2194 2195 2196 2197 2198 2199 2200

    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);
2201

2202 2203
    if (virLockManagerPluginUsesState(driver->lockManager) &&
        !cookieout) {
2204 2205 2206 2207
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Migration with lock driver %s requires"
                         " cookie support"),
                       virLockManagerPluginGetName(driver->lockManager));
2208 2209 2210 2211
        return -1;
    }

    if (!(mig = qemuMigrationEatCookie(driver, vm, cookiein, cookieinlen,
2212 2213 2214 2215 2216 2217
                                       QEMU_MIGRATION_COOKIE_GRAPHICS)))
        goto cleanup;

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

2218
    /* Before EnterMonitor, since qemuMigrationSetOffline already does that */
2219 2220 2221 2222 2223 2224
    if (!(flags & VIR_MIGRATE_LIVE) &&
        virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING) {
        if (qemuMigrationSetOffline(driver, vm) < 0)
            goto cleanup;
    }

2225 2226
    if (qemuDomainObjEnterMonitorAsync(driver, vm,
                                       QEMU_ASYNC_JOB_MIGRATION_OUT) < 0)
2227 2228
        goto cleanup;

2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239
    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 */
        qemuDomainObjExitMonitorWithDriver(driver, vm);
        virReportError(VIR_ERR_OPERATION_ABORTED, _("%s: %s"),
                       qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                       _("canceled by client"));
        goto cleanup;
    }

2240
    if (qemuMonitorSetMigrationSpeed(priv->mon, migrate_speed) < 0) {
2241 2242 2243
        qemuDomainObjExitMonitorWithDriver(driver, vm);
        goto cleanup;
    }
2244

2245
    if (flags & VIR_MIGRATE_NON_SHARED_DISK)
2246 2247
        migrate_flags |= QEMU_MONITOR_MIGRATE_NON_SHARED_DISK;

2248
    if (flags & VIR_MIGRATE_NON_SHARED_INC)
2249 2250
        migrate_flags |= QEMU_MONITOR_MIGRATE_NON_SHARED_INC;

2251 2252
    /* connect to the destination qemu if needed */
    if (spec->destType == MIGRATION_DEST_CONNECT_HOST &&
2253 2254
        qemuMigrationConnect(driver, vm, spec) < 0) {
        qemuDomainObjExitMonitorWithDriver(driver, vm);
2255
        goto cleanup;
2256
    }
2257

2258 2259 2260 2261 2262 2263 2264
    switch (spec->destType) {
    case MIGRATION_DEST_HOST:
        ret = qemuMonitorMigrateToHost(priv->mon, migrate_flags,
                                       spec->dest.host.name,
                                       spec->dest.host.port);
        break;

2265 2266 2267 2268
    case MIGRATION_DEST_CONNECT_HOST:
        /* handled above and transformed into MIGRATION_DEST_FD */
        break;

2269
    case MIGRATION_DEST_UNIX:
2270
        if (qemuCapsGet(priv->caps, QEMU_CAPS_MIGRATE_QEMU_UNIX)) {
2271 2272 2273 2274 2275 2276 2277 2278 2279
            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;
2280 2281

    case MIGRATION_DEST_FD:
2282
        if (spec->fwdType != MIGRATION_FWD_DIRECT) {
2283
            fd = spec->dest.fd.local;
2284 2285
            spec->dest.fd.local = -1;
        }
2286 2287 2288 2289
        ret = qemuMonitorMigrateToFd(priv->mon, migrate_flags,
                                     spec->dest.fd.qemu);
        VIR_FORCE_CLOSE(spec->dest.fd.qemu);
        break;
2290 2291
    }
    qemuDomainObjExitMonitorWithDriver(driver, vm);
2292
    if (ret < 0)
2293 2294
        goto cleanup;
    ret = -1;
2295 2296

    if (!virDomainObjIsActive(vm)) {
2297 2298
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("guest unexpectedly quit"));
2299 2300 2301 2302 2303 2304
        goto cleanup;
    }

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

2305 2306 2307 2308 2309 2310 2311 2312
    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;
2313

2314 2315 2316 2317 2318 2319 2320
        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;
        }
2321 2322
    }

2323 2324
    if (spec->fwdType != MIGRATION_FWD_DIRECT &&
        !(iothread = qemuMigrationStartTunnel(spec->fwd.stream, fd)))
2325 2326
        goto cancel;

2327
    if (qemuMigrationWaitForCompletion(driver, vm,
2328 2329
                                       QEMU_ASYNC_JOB_MIGRATION_OUT,
                                       dconn) < 0)
2330
        goto cleanup;
2331

2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344
    /* 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;
    }

2345
    ret = 0;
2346

2347
cleanup:
2348 2349 2350
    if (ret < 0 && !orig_err)
        orig_err = virSaveLastError();

2351
    if (spec->fwdType != MIGRATION_FWD_DIRECT) {
2352
        if (iothread && qemuMigrationStopTunnel(iothread, ret < 0) < 0)
2353
            ret = -1;
2354
        VIR_FORCE_CLOSE(fd);
2355
    }
2356

2357
    if (ret == 0 &&
2358
        qemuMigrationBakeCookie(mig, driver, vm, cookieout, cookieoutlen,
2359 2360
                                QEMU_MIGRATION_COOKIE_PERSISTENT |
                                QEMU_MIGRATION_COOKIE_NETWORK) < 0) {
2361
        VIR_WARN("Unable to encode migration cookie");
2362
    }
2363

2364 2365
    qemuMigrationCookieFree(mig);

2366 2367 2368 2369 2370
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }

2371 2372
    return ret;

2373
cancel:
2374 2375
    orig_err = virSaveLastError();

2376
    if (virDomainObjIsActive(vm)) {
2377 2378
        if (qemuDomainObjEnterMonitorAsync(driver, vm,
                                           QEMU_ASYNC_JOB_MIGRATION_OUT) == 0) {
2379 2380 2381
            qemuMonitorMigrateCancel(priv->mon);
            qemuDomainObjExitMonitorWithDriver(driver, vm);
        }
2382
    }
2383 2384 2385 2386 2387 2388
    goto cleanup;
}

/* Perform migration using QEMU's native TCP migrate support,
 * not encrypted obviously
 */
2389
static int doNativeMigrate(virQEMUDriverPtr driver,
2390 2391 2392 2393 2394 2395 2396
                           virDomainObjPtr vm,
                           const char *uri,
                           const char *cookiein,
                           int cookieinlen,
                           char **cookieout,
                           int *cookieoutlen,
                           unsigned long flags,
2397 2398
                           unsigned long resource,
                           virConnectPtr dconn)
2399
{
2400
    qemuDomainObjPrivatePtr priv = vm->privateData;
M
Martin Kletzander 已提交
2401
    virURIPtr uribits = NULL;
2402
    int ret = -1;
2403 2404 2405 2406 2407 2408 2409 2410
    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://")) {
2411
        char *tmp;
2412
        /* HACK: source host generates bogus URIs, so fix them up */
2413
        if (virAsprintf(&tmp, "tcp://%s", uri + strlen("tcp:")) < 0) {
2414 2415 2416
            virReportOOMError();
            return -1;
        }
M
Martin Kletzander 已提交
2417
        uribits = virURIParse(tmp);
2418
        VIR_FREE(tmp);
2419
    } else {
M
Martin Kletzander 已提交
2420
        uribits = virURIParse(uri);
2421
    }
2422
    if (!uribits)
2423 2424
        return -1;

2425
    if (qemuCapsGet(priv->caps, QEMU_CAPS_MIGRATE_QEMU_FD))
2426 2427
        spec.destType = MIGRATION_DEST_CONNECT_HOST;
    else
2428
        spec.destType = MIGRATION_DEST_HOST;
2429 2430 2431
    spec.dest.host.name = uribits->server;
    spec.dest.host.port = uribits->port;
    spec.fwdType = MIGRATION_FWD_DIRECT;
2432

2433
    ret = qemuMigrationRun(driver, vm, cookiein, cookieinlen, cookieout,
2434
                           cookieoutlen, flags, resource, &spec, dconn);
2435 2436 2437 2438

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

2439
    virURIFree(uribits);
2440 2441 2442 2443 2444

    return ret;
}


2445
static int doTunnelMigrate(virQEMUDriverPtr driver,
2446 2447 2448 2449 2450 2451 2452
                           virDomainObjPtr vm,
                           virStreamPtr st,
                           const char *cookiein,
                           int cookieinlen,
                           char **cookieout,
                           int *cookieoutlen,
                           unsigned long flags,
2453 2454
                           unsigned long resource,
                           virConnectPtr dconn)
2455 2456
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
2457
    virNetSocketPtr sock = NULL;
2458 2459 2460 2461 2462 2463 2464 2465
    int ret = -1;
    qemuMigrationSpec spec;

    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);

2466 2467 2468
    if (!qemuCapsGet(priv->caps, QEMU_CAPS_MIGRATE_QEMU_FD) &&
        !qemuCapsGet(priv->caps, QEMU_CAPS_MIGRATE_QEMU_UNIX) &&
        !qemuCapsGet(priv->caps, QEMU_CAPS_MIGRATE_QEMU_EXEC)) {
2469 2470
        virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                       _("Source qemu is too old to support tunnelled migration"));
2471
        return -1;
2472 2473 2474 2475 2476
    }

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

2477
    if (qemuCapsGet(priv->caps, QEMU_CAPS_MIGRATE_QEMU_FD)) {
2478 2479 2480 2481 2482 2483
        int fds[2];

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

2484
        if (pipe2(fds, O_CLOEXEC) == 0) {
2485 2486 2487 2488
            spec.dest.fd.qemu = fds[1];
            spec.dest.fd.local = fds[0];
        }
        if (spec.dest.fd.qemu == -1 ||
2489
            virSecurityManagerSetImageFDLabel(driver->securityManager, vm->def,
2490
                                              spec.dest.fd.qemu) < 0) {
2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515
            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",
                        driver->libDir, vm->def->name) < 0) {
            virReportOOMError();
            goto cleanup;
        }

        if (virNetSocketNewListenUNIX(spec.dest.unix_socket.file, 0700,
                                      driver->user, driver->group,
                                      &sock) < 0 ||
            virNetSocketListen(sock, 1) < 0)
            goto cleanup;

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

2516
    ret = qemuMigrationRun(driver, vm, cookiein, cookieinlen, cookieout,
2517
                           cookieoutlen, flags, resource, &spec, dconn);
2518

2519
cleanup:
2520 2521 2522 2523
    if (spec.destType == MIGRATION_DEST_FD) {
        VIR_FORCE_CLOSE(spec.dest.fd.qemu);
        VIR_FORCE_CLOSE(spec.dest.fd.local);
    } else {
2524
        virObjectUnref(sock);
2525 2526
        VIR_FREE(spec.dest.unix_socket.file);
    }
2527 2528 2529 2530 2531

    return ret;
}


2532 2533 2534 2535
/* 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 */
2536
static int doPeer2PeerMigrate2(virQEMUDriverPtr driver,
2537
                               virConnectPtr sconn ATTRIBUTE_UNUSED,
2538 2539
                               virConnectPtr dconn,
                               virDomainObjPtr vm,
2540
                               const char *dconnuri,
2541 2542 2543
                               unsigned long flags,
                               const char *dname,
                               unsigned long resource)
2544 2545 2546
{
    virDomainPtr ddomain = NULL;
    char *uri_out = NULL;
2547
    char *cookie = NULL;
2548 2549 2550 2551 2552
    char *dom_xml = NULL;
    int cookielen = 0, ret;
    virErrorPtr orig_err = NULL;
    int cancelled;
    virStreamPtr st = NULL;
2553
    VIR_DEBUG("driver=%p, sconn=%p, dconn=%p, vm=%p, dconnuri=%s, "
2554
              "flags=%lx, dname=%s, resource=%lu",
2555 2556
              driver, sconn, dconn, vm, NULLSTR(dconnuri),
              flags, NULLSTR(dname), resource);
2557

2558 2559 2560 2561 2562
    /* 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,
2563 2564
                                        QEMU_DOMAIN_FORMAT_LIVE_FLAGS |
                                        VIR_DOMAIN_XML_MIGRATABLE)))
2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592
        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;

        qemuDomainObjEnterRemoteWithDriver(driver, vm);
        ret = dconn->driver->domainMigratePrepareTunnel
            (dconn, st, flags, dname, resource, dom_xml);
        qemuDomainObjExitRemoteWithDriver(driver, vm);
    } else {
        qemuDomainObjEnterRemoteWithDriver(driver, vm);
        ret = dconn->driver->domainMigratePrepare2
            (dconn, &cookie, &cookielen, NULL, &uri_out,
             flags, dname, resource, dom_xml);
        qemuDomainObjExitRemoteWithDriver(driver, vm);
    }
    VIR_FREE(dom_xml);
    if (ret == -1)
2593 2594 2595 2596 2597 2598
        goto cleanup;

    /* the domain may have shutdown or crashed while we had the locks dropped
     * in qemuDomainObjEnterRemoteWithDriver, so check again
     */
    if (!virDomainObjIsActive(vm)) {
2599 2600
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("guest unexpectedly quit"));
2601 2602 2603
        goto cleanup;
    }

2604 2605
    if (!(flags & VIR_MIGRATE_TUNNELLED) &&
        (uri_out == NULL)) {
2606 2607
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("domainMigratePrepare2 did not set uri"));
2608 2609
        cancelled = 1;
        goto finish;
2610 2611
    }

2612 2613 2614 2615
    /* Perform the migration.  The driver isn't supposed to return
     * until the migration is complete.
     */
    VIR_DEBUG("Perform %p", sconn);
2616
    qemuMigrationJobSetPhase(driver, vm, QEMU_MIGRATION_PHASE_PERFORM2);
2617
    if (flags & VIR_MIGRATE_TUNNELLED)
2618 2619
        ret = doTunnelMigrate(driver, vm, st,
                              NULL, 0, NULL, NULL,
2620
                              flags, resource, dconn);
2621 2622 2623 2624
    else
        ret = doNativeMigrate(driver, vm, uri_out,
                              cookie, cookielen,
                              NULL, NULL, /* No out cookie with v2 migration */
2625
                              flags, resource, dconn);
2626 2627 2628 2629

    /* Perform failed. Make sure Finish doesn't overwrite the error */
    if (ret < 0)
        orig_err = virSaveLastError();
2630

2631 2632 2633 2634
    /* If Perform returns < 0, then we need to cancel the VM
     * startup on the destination
     */
    cancelled = ret < 0 ? 1 : 0;
2635 2636

finish:
2637 2638 2639 2640
    /* 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.
     */
2641
    dname = dname ? dname : vm->def->name;
2642
    VIR_DEBUG("Finish2 %p ret=%d", dconn, ret);
2643 2644
    qemuDomainObjEnterRemoteWithDriver(driver, vm);
    ddomain = dconn->driver->domainMigrateFinish2
2645
        (dconn, dname, cookie, cookielen,
2646
         uri_out ? uri_out : dconnuri, flags, cancelled);
2647 2648
    qemuDomainObjExitRemoteWithDriver(driver, vm);

2649 2650
cleanup:
    if (ddomain) {
2651
        virObjectUnref(ddomain);
2652 2653 2654 2655
        ret = 0;
    } else {
        ret = -1;
    }
2656

2657
    virObjectUnref(st);
2658 2659 2660 2661 2662 2663

    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
    VIR_FREE(uri_out);
2664
    VIR_FREE(cookie);
2665 2666

    return ret;
2667 2668 2669
}


2670 2671 2672 2673
/* 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 */
2674
static int doPeer2PeerMigrate3(virQEMUDriverPtr driver,
2675 2676 2677
                               virConnectPtr sconn,
                               virConnectPtr dconn,
                               virDomainObjPtr vm,
2678
                               const char *xmlin,
2679
                               const char *dconnuri,
2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695
                               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;
2696
    VIR_DEBUG("driver=%p, sconn=%p, dconn=%p, vm=%p, xmlin=%s, "
2697
              "dconnuri=%s, uri=%s, flags=%lx, dname=%s, resource=%lu",
2698 2699 2700
              driver, sconn, dconn, vm, NULLSTR(xmlin),
              NULLSTR(dconnuri), NULLSTR(uri), flags,
              NULLSTR(dname), resource);
2701

2702 2703 2704 2705 2706
    /* 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.  */

2707
    dom_xml = qemuMigrationBegin(driver, vm, xmlin, dname,
2708
                                 &cookieout, &cookieoutlen, flags);
2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733
    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;

        qemuDomainObjEnterRemoteWithDriver(driver, vm);
        ret = dconn->driver->domainMigratePrepareTunnel3
            (dconn, st, cookiein, cookieinlen,
             &cookieout, &cookieoutlen,
             flags, dname, resource, dom_xml);
        qemuDomainObjExitRemoteWithDriver(driver, vm);
    } else {
        qemuDomainObjEnterRemoteWithDriver(driver, vm);
        ret = dconn->driver->domainMigratePrepare3
            (dconn, cookiein, cookieinlen, &cookieout, &cookieoutlen,
2734
             uri, &uri_out, flags, dname, resource, dom_xml);
2735 2736 2737 2738 2739 2740
        qemuDomainObjExitRemoteWithDriver(driver, vm);
    }
    VIR_FREE(dom_xml);
    if (ret == -1)
        goto cleanup;

L
liguang 已提交
2741 2742 2743 2744 2745 2746 2747 2748
    if (flags & VIR_MIGRATE_OFFLINE) {
        VIR_DEBUG("Offline migration, skipping Perform phase");
        VIR_FREE(cookieout);
        cookieoutlen = 0;
        cancelled = 0;
        goto finish;
    }

2749 2750
    if (!(flags & VIR_MIGRATE_TUNNELLED) &&
        (uri_out == NULL)) {
2751 2752
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("domainMigratePrepare3 did not set uri"));
2753 2754 2755 2756 2757 2758 2759 2760 2761
        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.
     */
2762
    VIR_DEBUG("Perform3 %p uri=%s uri_out=%s", sconn, uri, uri_out);
2763
    qemuMigrationJobSetPhase(driver, vm, QEMU_MIGRATION_PHASE_PERFORM3);
2764 2765 2766 2767 2768 2769 2770 2771 2772
    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,
2773
                              flags, resource, dconn);
2774 2775 2776 2777
    else
        ret = doNativeMigrate(driver, vm, uri_out,
                              cookiein, cookieinlen,
                              &cookieout, &cookieoutlen,
2778
                              flags, resource, dconn);
2779 2780

    /* Perform failed. Make sure Finish doesn't overwrite the error */
2781
    if (ret < 0) {
2782
        orig_err = virSaveLastError();
2783 2784 2785 2786
    } else {
        qemuMigrationJobSetPhase(driver, vm,
                                 QEMU_MIGRATION_PHASE_PERFORM3_DONE);
    }
2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807

    /* 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;
    qemuDomainObjEnterRemoteWithDriver(driver, vm);
2808
    ddomain = dconn->driver->domainMigrateFinish3
2809
        (dconn, dname, cookiein, cookieinlen, &cookieout, &cookieoutlen,
2810
         dconnuri, uri_out ? uri_out : uri, flags, cancelled);
2811 2812
    qemuDomainObjExitRemoteWithDriver(driver, vm);

2813 2814 2815 2816 2817 2818 2819
    /* 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.
2820
     */
2821
    cancelled = ddomain == NULL ? 1 : 0;
2822

2823 2824 2825 2826 2827 2828
    /* 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();

2829 2830 2831 2832
    /*
     * If cancelled, then src VM will be restarted, else
     * it will be killed
     */
2833
    VIR_DEBUG("Confirm3 %p cancelled=%d vm=%p", sconn, cancelled, vm);
2834 2835 2836 2837 2838 2839 2840
    VIR_FREE(cookiein);
    cookiein = cookieout;
    cookieinlen = cookieoutlen;
    cookieout = NULL;
    cookieoutlen = 0;
    ret = qemuMigrationConfirm(driver, sconn, vm,
                               cookiein, cookieinlen,
2841
                               flags, cancelled);
2842 2843 2844 2845
    /* 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.
     */
2846 2847 2848
    if (ret < 0)
        VIR_WARN("Guest %s probably left in 'paused' state on source",
                 vm->def->name);
2849 2850 2851

 cleanup:
    if (ddomain) {
2852
        virObjectUnref(ddomain);
2853 2854 2855 2856 2857
        ret = 0;
    } else {
        ret = -1;
    }

2858
    virObjectUnref(st);
2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871

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

    return ret;
}


2872
static int doPeer2PeerMigrate(virQEMUDriverPtr driver,
2873
                              virConnectPtr sconn,
2874
                              virDomainObjPtr vm,
2875
                              const char *xmlin,
2876
                              const char *dconnuri,
2877 2878 2879
                              const char *uri,
                              unsigned long flags,
                              const char *dname,
2880 2881
                              unsigned long resource,
                              bool *v3proto)
2882 2883 2884 2885
{
    int ret = -1;
    virConnectPtr dconn = NULL;
    bool p2p;
2886
    virErrorPtr orig_err = NULL;
L
liguang 已提交
2887
    bool offline;
2888

2889
    VIR_DEBUG("driver=%p, sconn=%p, vm=%p, xmlin=%s, dconnuri=%s, "
2890
              "uri=%s, flags=%lx, dname=%s, resource=%lu",
2891 2892
              driver, sconn, vm, NULLSTR(xmlin), NULLSTR(dconnuri),
              NULLSTR(uri), flags, NULLSTR(dname), resource);
2893 2894 2895 2896 2897 2898

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

    qemuDomainObjEnterRemoteWithDriver(driver, vm);
2899
    dconn = virConnectOpen(dconnuri);
2900 2901
    qemuDomainObjExitRemoteWithDriver(driver, vm);
    if (dconn == NULL) {
2902 2903
        virReportError(VIR_ERR_OPERATION_FAILED,
                       _("Failed to connect to remote libvirt URI %s"), dconnuri);
2904 2905 2906
        return -1;
    }

2907 2908 2909 2910
    if (virConnectSetKeepAlive(dconn, driver->keepAliveInterval,
                               driver->keepAliveCount) < 0)
        goto cleanup;

2911 2912 2913
    qemuDomainObjEnterRemoteWithDriver(driver, vm);
    p2p = VIR_DRV_SUPPORTS_FEATURE(dconn->driver, dconn,
                                   VIR_DRV_FEATURE_MIGRATION_P2P);
2914
        /* v3proto reflects whether the caller used Perform3, but with
2915
         * p2p migrate, regardless of whether Perform2 or Perform3
2916 2917 2918 2919
         * 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
2920 2921 2922
    if (flags & VIR_MIGRATE_OFFLINE)
        offline = VIR_DRV_SUPPORTS_FEATURE(dconn->driver, dconn,
                                           VIR_DRV_FEATURE_MIGRATION_OFFLINE);
2923
    qemuDomainObjExitRemoteWithDriver(driver, vm);
2924

2925
    if (!p2p) {
2926 2927
        virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                       _("Destination libvirt does not support peer-to-peer migration protocol"));
2928 2929 2930
        goto cleanup;
    }

L
liguang 已提交
2931 2932 2933 2934 2935 2936 2937
    if (flags & VIR_MIGRATE_OFFLINE && !offline) {
        virReportError(VIR_ERR_ARGUMENT_UNSUPPORTED, "%s",
                       _("offline migration is not supported by "
                         "the destination host"));
        goto cleanup;
    }

2938
    /* domain may have been stopped while we were talking to remote daemon */
L
liguang 已提交
2939
    if (!virDomainObjIsActive(vm) && !(flags & VIR_MIGRATE_OFFLINE)) {
2940 2941
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("guest unexpectedly quit"));
2942 2943 2944
        goto cleanup;
    }

2945 2946 2947 2948 2949 2950 2951
    /* 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;

2952
    if (*v3proto)
2953
        ret = doPeer2PeerMigrate3(driver, sconn, dconn, vm, xmlin,
2954
                                  dconnuri, uri, flags, dname, resource);
2955 2956
    else
        ret = doPeer2PeerMigrate2(driver, sconn, dconn, vm,
2957
                                  dconnuri, flags, dname, resource);
2958 2959

cleanup:
2960
    orig_err = virSaveLastError();
2961
    qemuDomainObjEnterRemoteWithDriver(driver, vm);
2962
    virConnectClose(dconn);
2963
    qemuDomainObjExitRemoteWithDriver(driver, vm);
2964 2965 2966 2967
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
2968 2969 2970 2971 2972

    return ret;
}


2973 2974 2975 2976 2977 2978
/*
 * 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
2979
qemuMigrationPerformJob(virQEMUDriverPtr driver,
2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992
                        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)
2993 2994 2995 2996
{
    virDomainEventPtr event = NULL;
    int ret = -1;
    int resume = 0;
2997
    virErrorPtr orig_err = NULL;
2998

2999
    if (qemuMigrationJobStart(driver, vm, QEMU_ASYNC_JOB_MIGRATION_OUT) < 0)
3000 3001
        goto cleanup;

L
liguang 已提交
3002
    if (!virDomainObjIsActive(vm) && !(flags & VIR_MIGRATE_OFFLINE)) {
3003 3004
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("domain is not running"));
3005 3006 3007
        goto endjob;
    }

3008
    if (!qemuMigrationIsAllowed(driver, vm, NULL))
3009
        goto endjob;
3010

3011
    if (!(flags & VIR_MIGRATE_UNSAFE) && !qemuMigrationIsSafe(vm->def))
3012
        goto endjob;
3013

J
Jiri Denemark 已提交
3014
    resume = virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING;
3015 3016

    if ((flags & (VIR_MIGRATE_TUNNELLED | VIR_MIGRATE_PEER2PEER))) {
3017 3018 3019
        ret = doPeer2PeerMigrate(driver, conn, vm, xmlin,
                                 dconnuri, uri, flags, dname,
                                 resource, &v3proto);
3020
    } else {
3021 3022 3023
        qemuMigrationJobSetPhase(driver, vm, QEMU_MIGRATION_PHASE_PERFORM2);
        ret = doNativeMigrate(driver, vm, uri, cookiein, cookieinlen,
                              cookieout, cookieoutlen,
3024
                              flags, resource, NULL);
3025
    }
3026 3027
    if (ret < 0)
        goto endjob;
3028

3029 3030 3031 3032
    /*
     * In v3 protocol, the source VM is not killed off until the
     * confirm step.
     */
3033
    if (!v3proto) {
3034 3035
        qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_MIGRATED,
                        VIR_QEMU_PROCESS_STOP_MIGRATED);
3036
        virDomainAuditStop(vm, "migrated");
3037 3038 3039
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         VIR_DOMAIN_EVENT_STOPPED_MIGRATED);
3040
    }
3041
    resume = 0;
3042 3043

endjob:
3044 3045 3046
    if (ret < 0)
        orig_err = virSaveLastError();

J
Jiri Denemark 已提交
3047
    if (resume && virDomainObjGetState(vm, NULL) == VIR_DOMAIN_PAUSED) {
3048
        /* we got here through some sort of failure; start the domain again */
J
Jiri Denemark 已提交
3049
        if (qemuProcessStartCPUs(driver, vm, conn,
3050 3051
                                 VIR_DOMAIN_RUNNING_MIGRATION_CANCELED,
                                 QEMU_ASYNC_JOB_MIGRATION_OUT) < 0) {
3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063
            /* 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);
    }
3064

3065
    if (!qemuMigrationJobFinish(driver, vm)) {
3066 3067 3068 3069 3070 3071
        vm = NULL;
    } else if (!virDomainObjIsActive(vm) &&
               (!vm->persistent ||
                (ret == 0 && (flags & VIR_MIGRATE_UNDEFINE_SOURCE)))) {
        if (flags & VIR_MIGRATE_UNDEFINE_SOURCE)
            virDomainDeleteConfig(driver->configDir, driver->autostartDir, vm);
3072
        qemuDomainRemoveInactive(driver, vm);
3073 3074 3075
        vm = NULL;
    }

3076 3077 3078 3079 3080
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }

3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092
cleanup:
    if (vm)
        virDomainObjUnlock(vm);
    if (event)
        qemuDomainEventQueue(driver, event);
    return ret;
}

/*
 * This implements perform phase of v3 migration protocol.
 */
static int
3093
qemuMigrationPerformPhase(virQEMUDriverPtr driver,
3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106
                          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;
3107
    bool hasrefs;
3108 3109 3110 3111 3112 3113 3114 3115 3116 3117

    /* 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);
3118
    qemuDriverCloseCallbackUnset(driver, vm, qemuMigrationCleanup);
3119 3120 3121 3122

    resume = virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING;
    ret = doNativeMigrate(driver, vm, uri, cookiein, cookieinlen,
                          cookieout, cookieoutlen,
3123
                          flags, resource, NULL);
3124 3125 3126 3127 3128

    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,
3129 3130
                                 VIR_DOMAIN_RUNNING_MIGRATION_CANCELED,
                                 QEMU_ASYNC_JOB_MIGRATION_OUT) < 0) {
3131 3132 3133 3134 3135 3136
            /* 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);
3137
        }
3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148

        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);

3149 3150 3151 3152
    if (qemuDriverCloseCallbackSet(driver, vm, conn,
                                   qemuMigrationCleanup) < 0)
        goto endjob;

3153 3154
endjob:
    if (ret < 0)
3155
        hasrefs = qemuMigrationJobFinish(driver, vm);
3156
    else
3157 3158
        hasrefs = qemuMigrationJobContinue(vm);
    if (!hasrefs) {
3159 3160
        vm = NULL;
    } else if (!virDomainObjIsActive(vm) && !vm->persistent) {
3161
        qemuDomainRemoveInactive(driver, vm);
3162
        vm = NULL;
3163
    }
3164 3165 3166 3167 3168 3169 3170 3171 3172

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

3173
int
3174
qemuMigrationPerform(virQEMUDriverPtr driver,
3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198
                     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) {
3199 3200
            virReportError(VIR_ERR_OPERATION_INVALID,
                           "%s", _("received unexpected cookie with P2P migration"));
3201 3202 3203 3204 3205 3206 3207 3208 3209
            return -1;
        }

        return qemuMigrationPerformJob(driver, conn, vm, xmlin, dconnuri, uri,
                                       cookiein, cookieinlen, cookieout,
                                       cookieoutlen, flags, dname, resource,
                                       v3proto);
    } else {
        if (dconnuri) {
3210 3211
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("Unexpected dconnuri parameter with non-peer2peer migration"));
3212 3213 3214 3215 3216 3217 3218
            return -1;
        }

        if (v3proto) {
            return qemuMigrationPerformPhase(driver, conn, vm, uri,
                                             cookiein, cookieinlen,
                                             cookieout, cookieoutlen,
3219
                                             flags, resource);
3220 3221 3222 3223 3224 3225 3226 3227
        } else {
            return qemuMigrationPerformJob(driver, conn, vm, xmlin, dconnuri,
                                           uri, cookiein, cookieinlen,
                                           cookieout, cookieoutlen, flags,
                                           dname, resource, v3proto);
        }
    }
}
3228

3229
static int
3230 3231 3232 3233 3234 3235 3236
qemuMigrationVPAssociatePortProfiles(virDomainDefPtr def) {
    int i;
    int last_good_net = -1;
    virDomainNetDefPtr net;

    for (i = 0; i < def->nnets; i++) {
        net = def->nets[i];
3237
        if (virDomainNetGetActualType(net) == VIR_DOMAIN_NET_TYPE_DIRECT) {
3238
            if (virNetDevVPortProfileAssociate(net->ifname,
3239
                                               virDomainNetGetActualVirtPortProfile(net),
3240
                                               &net->mac,
3241
                                               virDomainNetGetActualDirectDev(net),
3242
                                               -1,
3243
                                               def->uuid,
3244 3245
                                               VIR_NETDEV_VPORT_PROFILE_OP_MIGRATE_IN_FINISH,
                                               false) < 0) {
3246 3247 3248
                virReportError(VIR_ERR_OPERATION_FAILED,
                               _("Port profile Associate failed for %s"),
                               net->ifname);
3249
                goto err_exit;
3250 3251
            }
            VIR_DEBUG("Port profile Associate succeeded for %s", net->ifname);
3252

3253
            if (virNetDevMacVLanVPortProfileRegisterCallback(net->ifname, &net->mac,
3254 3255 3256 3257
                                                             virDomainNetGetActualDirectDev(net), def->uuid,
                                                             virDomainNetGetActualVirtPortProfile(net),
                                                             VIR_NETDEV_VPORT_PROFILE_OP_CREATE))
                goto err_exit;
3258 3259 3260 3261
        }
        last_good_net = i;
    }

3262
    return 0;
3263 3264 3265 3266

err_exit:
    for (i = 0; i < last_good_net; i++) {
        net = def->nets[i];
3267
        if (virDomainNetGetActualType(net) == VIR_DOMAIN_NET_TYPE_DIRECT) {
3268
            ignore_value(virNetDevVPortProfileDisassociate(net->ifname,
3269
                                                           virDomainNetGetActualVirtPortProfile(net),
3270
                                                           &net->mac,
3271
                                                           virDomainNetGetActualDirectDev(net),
3272
                                                           -1,
3273
                                                           VIR_NETDEV_VPORT_PROFILE_OP_MIGRATE_IN_FINISH));
3274 3275
        }
    }
3276
    return -1;
3277 3278 3279 3280
}


virDomainPtr
3281
qemuMigrationFinish(virQEMUDriverPtr driver,
3282 3283
                    virConnectPtr dconn,
                    virDomainObjPtr vm,
3284 3285 3286 3287
                    const char *cookiein,
                    int cookieinlen,
                    char **cookieout,
                    int *cookieoutlen,
3288
                    unsigned long flags,
3289 3290
                    int retcode,
                    bool v3proto)
3291 3292 3293 3294
{
    virDomainPtr dom = NULL;
    virDomainEventPtr event = NULL;
    int newVM = 1;
3295
    qemuMigrationCookiePtr mig = NULL;
3296
    virErrorPtr orig_err = NULL;
3297
    int cookie_flags = 0;
J
Jiri Denemark 已提交
3298
    qemuDomainObjPrivatePtr priv = vm->privateData;
3299

3300
    VIR_DEBUG("driver=%p, dconn=%p, vm=%p, cookiein=%s, cookieinlen=%d, "
3301
              "cookieout=%p, cookieoutlen=%p, flags=%lx, retcode=%d",
3302 3303
              driver, dconn, vm, NULLSTR(cookiein), cookieinlen,
              cookieout, cookieoutlen, flags, retcode);
3304

3305
    if (!qemuMigrationJobIsActive(vm, QEMU_ASYNC_JOB_MIGRATION_IN))
3306 3307
        goto cleanup;

3308 3309 3310
    qemuMigrationJobStartPhase(driver, vm,
                               v3proto ? QEMU_MIGRATION_PHASE_FINISH3
                                       : QEMU_MIGRATION_PHASE_FINISH2);
3311

3312 3313
    qemuDomainCleanupRemove(vm, qemuMigrationPrepareCleanup);

3314
    cookie_flags = QEMU_MIGRATION_COOKIE_NETWORK;
3315 3316 3317 3318 3319
    if (flags & VIR_MIGRATE_PERSIST_DEST)
        cookie_flags |= QEMU_MIGRATION_COOKIE_PERSISTENT;

    if (!(mig = qemuMigrationEatCookie(driver, vm, cookiein,
                                       cookieinlen, cookie_flags)))
3320
        goto endjob;
3321 3322 3323 3324 3325

    /* 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
3326
        if (!virDomainObjIsActive(vm) && !(flags & VIR_MIGRATE_OFFLINE)) {
3327 3328
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("guest unexpectedly quit"));
3329
            goto endjob;
3330 3331
        }

L
liguang 已提交
3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344
        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");
3345
        }
3346 3347

        if (flags & VIR_MIGRATE_PERSIST_DEST) {
3348
            virDomainDefPtr vmdef;
3349 3350 3351
            if (vm->persistent)
                newVM = 0;
            vm->persistent = 1;
3352 3353 3354 3355
            if (mig->persistent)
                vm->newDef = vmdef = mig->persistent;
            else
                vmdef = virDomainObjGetPersistentDef(driver->caps, vm);
A
Alex Jia 已提交
3356
            if (!vmdef || virDomainSaveConfig(driver->configDir, vmdef) < 0) {
3357 3358 3359 3360 3361 3362 3363 3364 3365
                /* 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.
                 */
3366 3367

                /*
3368 3369
                 * However, in v3 protocol, the source VM is still available
                 * to restart during confirm() step, so we kill it off now.
3370 3371
                 */
                if (v3proto) {
L
liguang 已提交
3372 3373 3374 3375 3376
                    if (!(flags & VIR_MIGRATE_OFFLINE)) {
                        qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_FAILED,
                                        VIR_QEMU_PROCESS_STOP_MIGRATED);
                        virDomainAuditStop(vm, "failed");
                    }
3377 3378
                    if (newVM)
                        vm->persistent = 0;
3379
                }
A
Alex Jia 已提交
3380
                if (!vmdef)
3381 3382
                    virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                                   _("can't get vmdef"));
3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395
                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
3396
        if (!(flags & VIR_MIGRATE_PAUSED) && !(flags & VIR_MIGRATE_OFFLINE)) {
3397 3398 3399 3400
            /* 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
3401
            if (qemuProcessStartCPUs(driver, vm, dconn,
3402 3403
                                     VIR_DOMAIN_RUNNING_MIGRATED,
                                     QEMU_ASYNC_JOB_MIGRATION_IN) < 0) {
3404
                if (virGetLastError() == NULL)
3405 3406
                    virReportError(VIR_ERR_INTERNAL_ERROR,
                                   "%s", _("resume operation failed"));
3407 3408 3409 3410 3411
                /* Need to save the current error, in case shutting
                 * down the process overwrites it
                 */
                orig_err = virSaveLastError();

3412 3413 3414 3415 3416 3417 3418 3419 3420
                /*
                 * 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) {
3421 3422
                    qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_FAILED,
                                    VIR_QEMU_PROCESS_STOP_MIGRATED);
3423
                    virDomainAuditStop(vm, "failed");
3424 3425 3426 3427
                    event = virDomainEventNewFromObj(vm,
                                                     VIR_DOMAIN_EVENT_STOPPED,
                                                     VIR_DOMAIN_EVENT_STOPPED_FAILED);
                }
3428 3429 3430 3431
                goto endjob;
            }
        }

3432
        dom = virGetDomain(dconn, vm->def->name, vm->def->uuid);
3433

L
liguang 已提交
3434
        if (!(flags & VIR_MIGRATE_OFFLINE)) {
3435
            event = virDomainEventNewFromObj(vm,
L
liguang 已提交
3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446
                                             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);
            }
3447
        }
L
liguang 已提交
3448 3449 3450

        if (virDomainObjIsActive(vm) &&
            virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0) {
3451 3452 3453
            VIR_WARN("Failed to save status on vm %s", vm->def->name);
            goto endjob;
        }
3454 3455 3456

        /* Guest is successfully running, so cancel previous auto destroy */
        qemuProcessAutoDestroyRemove(driver, vm);
L
liguang 已提交
3457
    } else if (!(flags & VIR_MIGRATE_OFFLINE)) {
3458 3459
        qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_FAILED,
                        VIR_QEMU_PROCESS_STOP_MIGRATED);
3460
        virDomainAuditStop(vm, "failed");
3461 3462 3463 3464 3465
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         VIR_DOMAIN_EVENT_STOPPED_FAILED);
    }

3466 3467 3468
    if (qemuMigrationBakeCookie(mig, driver, vm, cookieout, cookieoutlen, 0) < 0)
        VIR_WARN("Unable to encode migration cookie");

3469
endjob:
E
Eric Blake 已提交
3470 3471 3472
    if (qemuMigrationJobFinish(driver, vm) == 0) {
        vm = NULL;
    } else if (!vm->persistent && !virDomainObjIsActive(vm)) {
3473
        qemuDomainRemoveInactive(driver, vm);
E
Eric Blake 已提交
3474
        vm = NULL;
3475
    }
3476 3477

cleanup:
J
Jiri Denemark 已提交
3478 3479
    if (vm) {
        VIR_FREE(priv->origname);
3480
        virDomainObjUnlock(vm);
J
Jiri Denemark 已提交
3481
    }
3482 3483
    if (event)
        qemuDomainEventQueue(driver, event);
3484
    qemuMigrationCookieFree(mig);
3485 3486 3487 3488
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
3489 3490
    return dom;
}
3491

3492

3493
int qemuMigrationConfirm(virQEMUDriverPtr driver,
3494 3495 3496 3497
                         virConnectPtr conn,
                         virDomainObjPtr vm,
                         const char *cookiein,
                         int cookieinlen,
E
Eric Blake 已提交
3498
                         unsigned int flags,
3499
                         int retcode)
3500 3501 3502 3503
{
    qemuMigrationCookiePtr mig;
    virDomainEventPtr event = NULL;
    int rv = -1;
3504
    VIR_DEBUG("driver=%p, conn=%p, vm=%p, cookiein=%s, cookieinlen=%d, "
3505
              "flags=%x, retcode=%d",
3506 3507
              driver, conn, vm, NULLSTR(cookiein), cookieinlen,
              flags, retcode);
3508

3509
    virCheckFlags(QEMU_MIGRATION_FLAGS, -1);
E
Eric Blake 已提交
3510

3511 3512 3513 3514 3515
    qemuMigrationJobSetPhase(driver, vm,
                             retcode == 0
                             ? QEMU_MIGRATION_PHASE_CONFIRM3
                             : QEMU_MIGRATION_PHASE_CONFIRM3_CANCELLED);

3516
    if (!(mig = qemuMigrationEatCookie(driver, vm, cookiein, cookieinlen, 0)))
3517 3518
        return -1;

L
liguang 已提交
3519 3520 3521
    if (flags & VIR_MIGRATE_OFFLINE)
        goto done;

3522 3523 3524 3525
    /* Did the migration go as planned?  If yes, kill off the
     * domain object, but if no, resume CPUs
     */
    if (retcode == 0) {
3526 3527
        qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_MIGRATED,
                        VIR_QEMU_PROCESS_STOP_MIGRATED);
3528
        virDomainAuditStop(vm, "migrated");
3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539

        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,
3540 3541
                                 VIR_DOMAIN_RUNNING_MIGRATED,
                                 QEMU_ASYNC_JOB_MIGRATION_OUT) < 0) {
3542
            if (virGetLastError() == NULL)
3543 3544
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               "%s", _("resume operation failed"));
3545
            goto cleanup;
3546 3547 3548 3549 3550 3551 3552
        }

        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_MIGRATED);
        if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0) {
            VIR_WARN("Failed to save status on vm %s", vm->def->name);
3553
            goto cleanup;
3554 3555 3556
        }
    }

L
liguang 已提交
3557
done:
3558 3559 3560 3561 3562 3563 3564 3565 3566 3567
    qemuMigrationCookieFree(mig);
    rv = 0;

cleanup:
    if (event)
        qemuDomainEventQueue(driver, event);
    return rv;
}


3568 3569
/* Helper function called while driver lock is held and vm is active.  */
int
3570
qemuMigrationToFile(virQEMUDriverPtr driver, virDomainObjPtr vm,
3571 3572
                    int fd, off_t offset, const char *path,
                    const char *compressor,
E
Eric Blake 已提交
3573
                    bool bypassSecurityDriver,
3574
                    enum qemuDomainAsyncJob asyncJob)
3575 3576 3577 3578 3579 3580
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virCgroupPtr cgroup = NULL;
    int ret = -1;
    int rc;
    bool restoreLabel = false;
3581 3582
    virCommandPtr cmd = NULL;
    int pipeFD[2] = { -1, -1 };
3583 3584 3585 3586 3587 3588
    unsigned long saveMigBandwidth = priv->migMaxBandwidth;

    /* 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,
3589 3590
                                     QEMU_DOMAIN_MIG_BANDWIDTH_MAX);
        priv->migMaxBandwidth = QEMU_DOMAIN_MIG_BANDWIDTH_MAX;
3591 3592
        qemuDomainObjExitMonitorWithDriver(driver, vm);
    }
3593

3594
    if (qemuCapsGet(priv->caps, QEMU_CAPS_MIGRATE_QEMU_FD) &&
3595
        (!compressor || pipe(pipeFD) == 0)) {
3596
        /* All right! We can use fd migration, which means that qemu
3597 3598 3599
         * 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.  */
3600
        if (virSecurityManagerSetImageFDLabel(driver->securityManager, vm->def,
3601
                                              compressor ? pipeFD[1] : fd) < 0)
3602
            goto cleanup;
3603 3604 3605
        bypassSecurityDriver = true;
    } else {
        /* Phooey - we have to fall back on exec migration, where qemu
E
Eric Blake 已提交
3606 3607
         * has to popen() the file by name, and block devices have to be
         * given cgroup ACL permission.  We might also stumble on
3608 3609
         * a race present in some qemu versions where it does a wait()
         * that botches pclose.  */
E
Eric Blake 已提交
3610
        if (qemuCgroupControllerActive(driver,
3611 3612 3613
                                       VIR_CGROUP_CONTROLLER_DEVICES)) {
            if (virCgroupForDomain(driver->cgroup, vm->def->name,
                                   &cgroup, 0) != 0) {
3614 3615 3616
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("Unable to find cgroup for %s"),
                               vm->def->name);
3617 3618 3619 3620
                goto cleanup;
            }
            rc = virCgroupAllowDevicePath(cgroup, path,
                                          VIR_CGROUP_DEVICE_RW);
3621
            virDomainAuditCgroupPath(vm, cgroup, "allow", path, "rw", rc);
E
Eric Blake 已提交
3622 3623 3624 3625
            if (rc == 1) {
                /* path was not a device, no further need for cgroup */
                virCgroupFree(&cgroup);
            } else if (rc < 0) {
3626 3627 3628 3629 3630
                virReportSystemError(-rc,
                                     _("Unable to allow device %s for %s"),
                                     path, vm->def->name);
                goto cleanup;
            }
3631
        }
3632 3633
        if ((!bypassSecurityDriver) &&
            virSecurityManagerSetSavedStateLabel(driver->securityManager,
3634
                                                 vm->def, path) < 0)
3635
            goto cleanup;
3636
        restoreLabel = true;
3637 3638
    }

3639
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
3640 3641
        goto cleanup;

3642 3643 3644
    if (!compressor) {
        const char *args[] = { "cat", NULL };

3645
        if (qemuCapsGet(priv->caps, QEMU_CAPS_MIGRATE_QEMU_FD) &&
3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661
            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
        };
3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680
        if (pipeFD[0] != -1) {
            cmd = virCommandNewArgs(args);
            virCommandSetInputFD(cmd, pipeFD[0]);
            virCommandSetOutputFD(cmd, &fd);
            if (virSetCloseExec(pipeFD[1]) < 0) {
                virReportSystemError(errno, "%s",
                                     _("Unable to set cloexec flag"));
                qemuDomainObjExitMonitorWithDriver(driver, vm);
                goto cleanup;
            }
            if (virCommandRunAsync(cmd, NULL) < 0) {
                qemuDomainObjExitMonitorWithDriver(driver, vm);
                goto cleanup;
            }
            rc = qemuMonitorMigrateToFd(priv->mon,
                                        QEMU_MONITOR_MIGRATE_BACKGROUND,
                                        pipeFD[1]);
            if (VIR_CLOSE(pipeFD[0]) < 0 ||
                VIR_CLOSE(pipeFD[1]) < 0)
3681
                VIR_WARN("failed to close intermediate pipe");
3682 3683 3684 3685 3686
        } else {
            rc = qemuMonitorMigrateToFile(priv->mon,
                                          QEMU_MONITOR_MIGRATE_BACKGROUND,
                                          args, path, offset);
        }
3687
    }
3688
    qemuDomainObjExitMonitorWithDriver(driver, vm);
3689 3690 3691 3692

    if (rc < 0)
        goto cleanup;

3693
    rc = qemuMigrationWaitForCompletion(driver, vm, asyncJob, NULL);
3694 3695 3696 3697

    if (rc < 0)
        goto cleanup;

3698 3699 3700
    if (cmd && virCommandWait(cmd, NULL) < 0)
        goto cleanup;

3701 3702 3703
    ret = 0;

cleanup:
3704 3705 3706 3707 3708 3709 3710
    /* Restore max migration bandwidth */
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) == 0) {
        qemuMonitorSetMigrationSpeed(priv->mon, saveMigBandwidth);
        priv->migMaxBandwidth = saveMigBandwidth;
        qemuDomainObjExitMonitorWithDriver(driver, vm);
    }

3711 3712 3713
    VIR_FORCE_CLOSE(pipeFD[0]);
    VIR_FORCE_CLOSE(pipeFD[1]);
    virCommandFree(cmd);
3714 3715
    if (restoreLabel && (!bypassSecurityDriver) &&
        virSecurityManagerRestoreSavedStateLabel(driver->securityManager,
3716
                                                 vm->def, path) < 0)
3717 3718 3719 3720 3721
        VIR_WARN("failed to restore save state label on %s", path);

    if (cgroup != NULL) {
        rc = virCgroupDenyDevicePath(cgroup, path,
                                     VIR_CGROUP_DEVICE_RWM);
3722
        virDomainAuditCgroupPath(vm, cgroup, "deny", path, "rwm", rc);
3723 3724 3725 3726 3727 3728 3729
        if (rc < 0)
            VIR_WARN("Unable to deny device %s for %s %d",
                     path, vm->def->name, rc);
        virCgroupFree(&cgroup);
    }
    return ret;
}
3730 3731

int
3732
qemuMigrationJobStart(virQEMUDriverPtr driver,
3733 3734 3735 3736 3737 3738 3739 3740
                      virDomainObjPtr vm,
                      enum qemuDomainAsyncJob job)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

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

3741
    if (job == QEMU_ASYNC_JOB_MIGRATION_IN) {
3742
        qemuDomainObjSetAsyncJobMask(vm, QEMU_JOB_NONE);
3743 3744
    } else {
        qemuDomainObjSetAsyncJobMask(vm, DEFAULT_JOB_MASK |
3745
                                     JOB_MASK(QEMU_JOB_SUSPEND) |
3746 3747
                                     JOB_MASK(QEMU_JOB_MIGRATION_OP));
    }
3748 3749 3750 3751 3752 3753 3754

    priv->job.info.type = VIR_DOMAIN_JOB_UNBOUNDED;

    return 0;
}

void
3755
qemuMigrationJobSetPhase(virQEMUDriverPtr driver,
3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771
                         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
3772
qemuMigrationJobStartPhase(virQEMUDriverPtr driver,
3773 3774 3775
                           virDomainObjPtr vm,
                           enum qemuMigrationJobPhase phase)
{
3776
    virObjectRef(vm);
3777 3778 3779
    qemuMigrationJobSetPhase(driver, vm, phase);
}

3780
bool
3781 3782
qemuMigrationJobContinue(virDomainObjPtr vm)
{
3783
    qemuDomainObjReleaseAsyncJob(vm);
3784
    return virObjectUnref(vm);
3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796 3797 3798 3799 3800
}

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");

3801
        virReportError(VIR_ERR_OPERATION_INVALID, msg, vm->def->name);
3802 3803 3804 3805 3806
        return false;
    }
    return true;
}

3807
bool
3808
qemuMigrationJobFinish(virQEMUDriverPtr driver, virDomainObjPtr vm)
3809 3810 3811
{
    return qemuDomainObjEndAsyncJob(driver, vm);
}