qemu_migration.c 143.3 KB
Newer Older
1

2 3 4
/*
 * qemu_migration.c: QEMU migration handling
 *
5
 * Copyright (C) 2006-2013 Red Hat, Inc.
6 7 8 9 10 11 12 13 14 15 16 17
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
18
 * License along with this library.  If not, see
O
Osier Yang 已提交
19
 * <http://www.gnu.org/licenses/>.
20 21 22 23 24
 *
 */

#include <config.h>

J
Ján Tomko 已提交
25 26
#include <netdb.h>
#include <sys/socket.h>
27
#include <sys/time.h>
28
#ifdef WITH_GNUTLS
29 30 31
# include <gnutls/gnutls.h>
# include <gnutls/x509.h>
#endif
32
#include <fcntl.h>
33
#include <poll.h>
34 35 36 37 38 39

#include "qemu_migration.h"
#include "qemu_monitor.h"
#include "qemu_domain.h"
#include "qemu_process.h"
#include "qemu_capabilities.h"
40
#include "qemu_command.h"
41
#include "qemu_cgroup.h"
42

43
#include "domain_audit.h"
44
#include "virlog.h"
45
#include "virerror.h"
46
#include "viralloc.h"
E
Eric Blake 已提交
47
#include "virfile.h"
48 49
#include "datatypes.h"
#include "fdstream.h"
50
#include "viruuid.h"
51
#include "virtime.h"
52
#include "locking/domain_lock.h"
53
#include "rpc/virnetsocket.h"
54
#include "virstoragefile.h"
M
Martin Kletzander 已提交
55
#include "viruri.h"
56
#include "virhook.h"
57
#include "virstring.h"
58 59 60

#define VIR_FROM_THIS VIR_FROM_QEMU

61 62 63 64 65 66 67 68 69 70 71 72 73
VIR_ENUM_IMPL(qemuMigrationJobPhase, QEMU_MIGRATION_PHASE_LAST,
              "none",
              "perform2",
              "begin3",
              "perform3",
              "perform3_done",
              "confirm3_cancelled",
              "confirm3",
              "prepare",
              "finish2",
              "finish3",
);

74
enum qemuMigrationCookieFlags {
75
    QEMU_MIGRATION_COOKIE_FLAG_GRAPHICS,
76
    QEMU_MIGRATION_COOKIE_FLAG_LOCKSTATE,
77
    QEMU_MIGRATION_COOKIE_FLAG_PERSISTENT,
78
    QEMU_MIGRATION_COOKIE_FLAG_NETWORK,
79
    QEMU_MIGRATION_COOKIE_FLAG_NBD,
80 81 82 83 84 85 86

    QEMU_MIGRATION_COOKIE_FLAG_LAST
};

VIR_ENUM_DECL(qemuMigrationCookieFlag);
VIR_ENUM_IMPL(qemuMigrationCookieFlag,
              QEMU_MIGRATION_COOKIE_FLAG_LAST,
87 88 89 90 91
              "graphics",
              "lockstate",
              "persistent",
              "network",
              "nbd");
92 93 94

enum qemuMigrationCookieFeatures {
    QEMU_MIGRATION_COOKIE_GRAPHICS  = (1 << QEMU_MIGRATION_COOKIE_FLAG_GRAPHICS),
95
    QEMU_MIGRATION_COOKIE_LOCKSTATE = (1 << QEMU_MIGRATION_COOKIE_FLAG_LOCKSTATE),
96
    QEMU_MIGRATION_COOKIE_PERSISTENT = (1 << QEMU_MIGRATION_COOKIE_FLAG_PERSISTENT),
97
    QEMU_MIGRATION_COOKIE_NETWORK = (1 << QEMU_MIGRATION_COOKIE_FLAG_NETWORK),
98
    QEMU_MIGRATION_COOKIE_NBD = (1 << QEMU_MIGRATION_COOKIE_FLAG_NBD),
99 100 101 102 103 104 105 106 107 108 109 110
};

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

111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131
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;
};

132 133 134 135 136 137
typedef struct _qemuMigrationCookieNBD qemuMigrationCookieNBD;
typedef qemuMigrationCookieNBD *qemuMigrationCookieNBDPtr;
struct _qemuMigrationCookieNBD {
    int port; /* on which port does NBD server listen for incoming data */
};

138 139 140
typedef struct _qemuMigrationCookie qemuMigrationCookie;
typedef qemuMigrationCookie *qemuMigrationCookiePtr;
struct _qemuMigrationCookie {
E
Eric Blake 已提交
141 142
    unsigned int flags;
    unsigned int flagsMandatory;
143 144

    /* Host properties */
145 146 147 148
    unsigned char localHostuuid[VIR_UUID_BUFLEN];
    unsigned char remoteHostuuid[VIR_UUID_BUFLEN];
    char *localHostname;
    char *remoteHostname;
149 150 151 152

    /* Guest properties */
    unsigned char uuid[VIR_UUID_BUFLEN];
    char *name;
153

154 155 156 157
    /* If (flags & QEMU_MIGRATION_COOKIE_LOCKSTATE) */
    char *lockState;
    char *lockDriver;

158 159
    /* If (flags & QEMU_MIGRATION_COOKIE_GRAPHICS) */
    qemuMigrationCookieGraphicsPtr graphics;
160 161 162

    /* If (flags & QEMU_MIGRATION_COOKIE_PERSISTENT) */
    virDomainDefPtr persistent;
163 164 165

    /* If (flags & QEMU_MIGRATION_COOKIE_NETWORK) */
    qemuMigrationCookieNetworkPtr network;
166 167 168

    /* If (flags & QEMU_MIGRATION_COOKIE_NBD) */
    qemuMigrationCookieNBDPtr nbd;
169 170
};

171 172 173 174 175 176 177 178 179
static void qemuMigrationCookieGraphicsFree(qemuMigrationCookieGraphicsPtr grap)
{
    if (!grap)
        return;
    VIR_FREE(grap->listen);
    VIR_FREE(grap->tlsSubject);
    VIR_FREE(grap);
}

180

181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197
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);
}


198 199 200 201 202
static void qemuMigrationCookieFree(qemuMigrationCookiePtr mig)
{
    if (!mig)
        return;

203 204 205
    if (mig->flags & QEMU_MIGRATION_COOKIE_GRAPHICS)
        qemuMigrationCookieGraphicsFree(mig->graphics);

206 207 208
    if (mig->flags & QEMU_MIGRATION_COOKIE_NETWORK)
        qemuMigrationCookieNetworkFree(mig->network);

209 210
    VIR_FREE(mig->localHostname);
    VIR_FREE(mig->remoteHostname);
211
    VIR_FREE(mig->name);
212 213
    VIR_FREE(mig->lockState);
    VIR_FREE(mig->lockDriver);
214
    VIR_FREE(mig->nbd);
215 216 217 218
    VIR_FREE(mig);
}


219
#ifdef WITH_GNUTLS
220 221 222 223 224 225 226 227 228 229 230 231 232 233 234
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) {
235 236
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unable to read server cert %s"), certfile);
237 238 239 240 241
        goto error;
    }

    ret = gnutls_x509_crt_init(&cert);
    if (ret < 0) {
242 243 244
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("cannot initialize cert object: %s"),
                       gnutls_strerror(ret));
245 246 247 248 249 250 251 252
        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) {
253 254 255
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("cannot load cert data from %s: %s"),
                       certfile, gnutls_strerror(ret));
256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277
        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;
}
278
#endif
279 280

static qemuMigrationCookieGraphicsPtr
281
qemuMigrationCookieGraphicsAlloc(virQEMUDriverPtr driver,
282 283 284 285
                                 virDomainGraphicsDefPtr def)
{
    qemuMigrationCookieGraphicsPtr mig = NULL;
    const char *listenAddr;
286
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
287 288 289 290 291 292 293

    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;
294
        listenAddr = virDomainGraphicsListenGetAddress(def, 0);
295
        if (!listenAddr)
296
            listenAddr = cfg->vncListen;
297

298
#ifdef WITH_GNUTLS
299 300
        if (cfg->vncTLS &&
            !(mig->tlsSubject = qemuDomainExtractTLSSubject(cfg->vncTLSx509certdir)))
301
            goto error;
302
#endif
303 304
    } else {
        mig->port = def->data.spice.port;
305
        if (cfg->spiceTLS)
306 307 308
            mig->tlsPort = def->data.spice.tlsPort;
        else
            mig->tlsPort = -1;
309
        listenAddr = virDomainGraphicsListenGetAddress(def, 0);
310
        if (!listenAddr)
311
            listenAddr = cfg->spiceListen;
312

313
#ifdef WITH_GNUTLS
314 315
        if (cfg->spiceTLS &&
            !(mig->tlsSubject = qemuDomainExtractTLSSubject(cfg->spiceTLSx509certdir)))
316
            goto error;
317
#endif
318
    }
319 320
    if (VIR_STRDUP(mig->listen, listenAddr) < 0)
        goto error;
321

322
    virObjectUnref(cfg);
323 324 325 326 327 328
    return mig;

no_memory:
    virReportOOMError();
error:
    qemuMigrationCookieGraphicsFree(mig);
329
    virObjectUnref(cfg);
330 331 332 333
    return NULL;
}


334
static qemuMigrationCookieNetworkPtr
335
qemuMigrationCookieNetworkAlloc(virQEMUDriverPtr driver ATTRIBUTE_UNUSED,
336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362
                                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:
363
               break;
364
            case VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH:
365 366 367 368 369 370 371 372
                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;
373 374 375 376 377 378 379 380 381
            default:
                break;
            }
        }
    }
    return mig;

no_memory:
    virReportOOMError();
382
error:
383 384 385 386
    qemuMigrationCookieNetworkFree(mig);
    return NULL;
}

387 388 389
static qemuMigrationCookiePtr
qemuMigrationCookieNew(virDomainObjPtr dom)
{
J
Jiri Denemark 已提交
390
    qemuDomainObjPrivatePtr priv = dom->privateData;
391
    qemuMigrationCookiePtr mig = NULL;
J
Jiri Denemark 已提交
392
    const char *name;
393 394 395 396

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

J
Jiri Denemark 已提交
397 398 399 400
    if (priv->origname)
        name = priv->origname;
    else
        name = dom->def->name;
401 402
    if (VIR_STRDUP(mig->name, name) < 0)
        goto error;
403 404
    memcpy(mig->uuid, dom->def->uuid, VIR_UUID_BUFLEN);

405
    if (!(mig->localHostname = virGetHostname()))
406
        goto error;
407
    if (virGetHostUUID(mig->localHostuuid) < 0) {
408 409
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Unable to obtain host UUID"));
410 411 412 413 414 415 416 417 418 419 420 421 422
        goto error;
    }

    return mig;

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


423 424
static int
qemuMigrationCookieAddGraphics(qemuMigrationCookiePtr mig,
425
                               virQEMUDriverPtr driver,
426 427 428
                               virDomainObjPtr dom)
{
    if (mig->flags & QEMU_MIGRATION_COOKIE_GRAPHICS) {
429 430
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Migration graphics data already present"));
431 432 433 434 435
        return -1;
    }

    if (dom->def->ngraphics == 1 &&
        (dom->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC ||
436 437 438 439 440 441
         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;
    }
442 443 444 445 446

    return 0;
}


447 448
static int
qemuMigrationCookieAddLockstate(qemuMigrationCookiePtr mig,
449
                                virQEMUDriverPtr driver,
450 451 452 453 454
                                virDomainObjPtr dom)
{
    qemuDomainObjPrivatePtr priv = dom->privateData;

    if (mig->flags & QEMU_MIGRATION_COOKIE_LOCKSTATE) {
455 456
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Migration lockstate data already present"));
457 458 459 460
        return -1;
    }

    if (virDomainObjGetState(dom, NULL) == VIR_DOMAIN_PAUSED) {
461
        if (VIR_STRDUP(mig->lockState, priv->lockState) < 0)
462 463 464 465 466 467
            return -1;
    } else {
        if (virDomainLockProcessInquire(driver->lockManager, dom, &mig->lockState) < 0)
            return -1;
    }

468
    if (VIR_STRDUP(mig->lockDriver, virLockManagerPluginGetName(driver->lockManager)) < 0) {
469 470 471 472 473 474 475 476 477 478 479
        VIR_FREE(mig->lockState);
        return -1;
    }

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

    return 0;
}


480 481 482 483 484
static int
qemuMigrationCookieAddPersistent(qemuMigrationCookiePtr mig,
                                 virDomainObjPtr dom)
{
    if (mig->flags & QEMU_MIGRATION_COOKIE_PERSISTENT) {
485 486
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Migration persistent data already present"));
487 488 489 490 491 492 493 494 495 496 497 498 499
        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;
}


500 501
static int
qemuMigrationCookieAddNetwork(qemuMigrationCookiePtr mig,
502
                              virQEMUDriverPtr driver,
503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520
                              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;
}

521

522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542
static int
qemuMigrationCookieAddNBD(qemuMigrationCookiePtr mig,
                          virQEMUDriverPtr driver ATTRIBUTE_UNUSED,
                          virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

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

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

    return 0;
}


543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560
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");
    }
}


561 562 563 564 565 566 567 568 569 570 571
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) {
572
                virBufferAddLit(buf, "  <network>\n");
573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591
                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");
}


592
static int
593
qemuMigrationCookieXMLFormat(virQEMUDriverPtr driver,
594
                             virBufferPtr buf,
595
                             qemuMigrationCookiePtr mig)
596 597 598
{
    char uuidstr[VIR_UUID_STRING_BUFLEN];
    char hostuuidstr[VIR_UUID_STRING_BUFLEN];
599
    int i;
600 601

    virUUIDFormat(mig->uuid, uuidstr);
602
    virUUIDFormat(mig->localHostuuid, hostuuidstr);
603

604
    virBufferAddLit(buf, "<qemu-migration>\n");
605 606
    virBufferEscapeString(buf, "  <name>%s</name>\n", mig->name);
    virBufferAsprintf(buf, "  <uuid>%s</uuid>\n", uuidstr);
607
    virBufferEscapeString(buf, "  <hostname>%s</hostname>\n", mig->localHostname);
608
    virBufferAsprintf(buf, "  <hostuuid>%s</hostuuid>\n", hostuuidstr);
609

610
    for (i = 0; i < QEMU_MIGRATION_COOKIE_FLAG_LAST; i++) {
611 612 613 614 615
        if (mig->flagsMandatory & (1 << i))
            virBufferAsprintf(buf, "  <feature name='%s'/>\n",
                              qemuMigrationCookieFlagTypeToString(i));
    }

616 617
    if ((mig->flags & QEMU_MIGRATION_COOKIE_GRAPHICS) &&
        mig->graphics)
618 619
        qemuMigrationCookieGraphicsXMLFormat(buf, mig->graphics);

620 621 622 623 624 625 626 627 628
    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");
    }

629 630
    if ((mig->flags & QEMU_MIGRATION_COOKIE_PERSISTENT) &&
        mig->persistent) {
631
        virBufferAdjustIndent(buf, 2);
632 633 634
        if (qemuDomainDefFormatBuf(driver,
                                   mig->persistent,
                                   VIR_DOMAIN_XML_INACTIVE |
635 636
                                   VIR_DOMAIN_XML_SECURE |
                                   VIR_DOMAIN_XML_MIGRATABLE,
637
                                   buf) < 0)
638 639
            return -1;
        virBufferAdjustIndent(buf, -2);
640 641
    }

642 643 644
    if ((mig->flags & QEMU_MIGRATION_COOKIE_NETWORK) && mig->network)
        qemuMigrationCookieNetworkXMLFormat(buf, mig->network);

645 646 647 648 649 650 651
    if ((mig->flags & QEMU_MIGRATION_COOKIE_NBD) && mig->nbd) {
        virBufferAddLit(buf, "  <nbd");
        if (mig->nbd->port)
            virBufferAsprintf(buf, " port='%d'", mig->nbd->port);
        virBufferAddLit(buf, "/>\n");
    }

652
    virBufferAddLit(buf, "</qemu-migration>\n");
653
    return 0;
654 655 656
}


657
static char *qemuMigrationCookieXMLFormatStr(virQEMUDriverPtr driver,
658
                                             qemuMigrationCookiePtr mig)
659 660 661
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;

662
    if (qemuMigrationCookieXMLFormat(driver, &buf, mig) < 0) {
663 664 665
        virBufferFreeAndReset(&buf);
        return NULL;
    }
666 667 668

    if (virBufferError(&buf)) {
        virReportOOMError();
669
        virBufferFreeAndReset(&buf);
670 671 672 673 674 675 676
        return NULL;
    }

    return virBufferContentAndReset(&buf);
}


677 678 679 680 681 682 683 684 685 686
static qemuMigrationCookieGraphicsPtr
qemuMigrationCookieGraphicsXMLParse(xmlXPathContextPtr ctxt)
{
    qemuMigrationCookieGraphicsPtr grap;
    char *tmp;

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

    if (!(tmp = virXPathString("string(./graphics/@type)", ctxt))) {
687 688
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("missing type attribute in migration data"));
689 690 691
        goto error;
    }
    if ((grap->type = virDomainGraphicsTypeFromString(tmp)) < 0) {
692 693
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unknown graphics type %s"), tmp);
694 695 696
        VIR_FREE(tmp);
        goto error;
    }
E
Eric Blake 已提交
697
    VIR_FREE(tmp);
698
    if (virXPathInt("string(./graphics/@port)", ctxt, &grap->port) < 0) {
699 700
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("missing port attribute in migration data"));
701 702 703 704
        goto error;
    }
    if (grap->type == VIR_DOMAIN_GRAPHICS_TYPE_SPICE) {
        if (virXPathInt("string(./graphics/@tlsPort)", ctxt, &grap->tlsPort) < 0) {
705 706
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("missing tlsPort attribute in migration data"));
707 708 709 710
            goto error;
        }
    }
    if (!(grap->listen = virXPathString("string(./graphics/@listen)", ctxt))) {
711 712
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("missing listen attribute in migration data"));
713 714 715
        goto error;
    }
    /* Optional */
716
    grap->tlsSubject = virXPathString("string(./graphics/cert[@info='subject']/@value)", ctxt);
717 718 719 720 721 722 723 724 725 726 727

    return grap;

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


728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747
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;
748
    if (VIR_ALLOC_N(optr->net, optr->nnets) < 0)
749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779
        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;
}


780 781
static int
qemuMigrationCookieXMLParse(qemuMigrationCookiePtr mig,
782
                            virQEMUDriverPtr driver,
783
                            xmlDocPtr doc,
784
                            xmlXPathContextPtr ctxt,
E
Eric Blake 已提交
785
                            unsigned int flags)
786 787
{
    char uuidstr[VIR_UUID_STRING_BUFLEN];
788
    char *tmp = NULL;
789 790
    xmlNodePtr *nodes = NULL;
    int i, n;
791 792 793 794
    virCapsPtr caps = NULL;

    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
        goto error;
795 796 797 798 799 800 801 802

    /* 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))) {
803 804
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("missing name element in migration data"));
805 806 807
        goto error;
    }
    if (STRNEQ(tmp, mig->name)) {
808 809 810
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Incoming cookie data had unexpected name %s vs %s"),
                       tmp, mig->name);
811 812 813 814 815 816 817
        goto error;
    }
    VIR_FREE(tmp);

    /* Extract domain uuid */
    tmp = virXPathString("string(./uuid[1])", ctxt);
    if (!tmp) {
818 819
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("missing uuid element in migration data"));
820 821 822 823
        goto error;
    }
    virUUIDFormat(mig->uuid, uuidstr);
    if (STRNEQ(tmp, uuidstr)) {
824 825 826
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Incoming cookie data had unexpected UUID %s vs %s"),
                       tmp, uuidstr);
827 828 829 830
    }
    VIR_FREE(tmp);

    /* Check & forbid "localhost" migration */
831
    if (!(mig->remoteHostname = virXPathString("string(./hostname[1])", ctxt))) {
832 833
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("missing hostname element in migration data"));
834 835
        goto error;
    }
836
    if (STREQ(mig->remoteHostname, mig->localHostname)) {
837 838 839
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Attempt to migrate guest to the same host %s"),
                       mig->remoteHostname);
840 841 842 843
        goto error;
    }

    if (!(tmp = virXPathString("string(./hostuuid[1])", ctxt))) {
844 845 846
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       "%s", _("missing hostuuid element in migration data"));
        goto error;
847
    }
848 849 850 851 852 853 854 855 856 857 858 859
    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);
860

861 862
    /* Check to ensure all mandatory features from XML are also
     * present in 'flags' */
863
    if ((n = virXPathNodeSet("./feature", ctxt, &nodes)) < 0)
864 865
        goto error;

866
    for (i = 0; i < n; i++) {
867 868 869
        int val;
        char *str = virXMLPropString(nodes[i], "name");
        if (!str) {
870 871
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("missing feature name"));
872 873 874 875
            goto error;
        }

        if ((val = qemuMigrationCookieFlagTypeFromString(str)) < 0) {
876 877 878
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Unknown migration cookie feature %s"),
                           str);
879 880 881 882 883
            VIR_FREE(str);
            goto error;
        }

        if ((flags & (1 << val)) == 0) {
884 885 886
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Unsupported migration cookie feature %s"),
                           str);
887 888 889 890 891 892
            VIR_FREE(str);
        }
        VIR_FREE(str);
    }
    VIR_FREE(nodes);

893 894 895 896 897
    if ((flags & QEMU_MIGRATION_COOKIE_GRAPHICS) &&
        virXPathBoolean("count(./graphics) > 0", ctxt) &&
        (!(mig->graphics = qemuMigrationCookieGraphicsXMLParse(ctxt))))
        goto error;

898 899 900 901
    if ((flags & QEMU_MIGRATION_COOKIE_LOCKSTATE) &&
        virXPathBoolean("count(./lockstate) > 0", ctxt)) {
        mig->lockDriver = virXPathString("string(./lockstate[1]/@driver)", ctxt);
        if (!mig->lockDriver) {
902 903
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("Missing lock driver name in migration cookie"));
904 905 906 907 908 909 910
            goto error;
        }
        mig->lockState = virXPathString("string(./lockstate[1]/leases[1])", ctxt);
        if (mig->lockState && STREQ(mig->lockState, ""))
            VIR_FREE(mig->lockState);
    }

911 912 913
    if ((flags & QEMU_MIGRATION_COOKIE_PERSISTENT) &&
        virXPathBoolean("count(./domain) > 0", ctxt)) {
        if ((n = virXPathNodeSet("./domain", ctxt, &nodes)) > 1) {
914 915 916 917
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Too many domain elements in "
                             "migration cookie: %d"),
                           n);
918 919
            goto error;
        }
920 921
        mig->persistent = virDomainDefParseNode(doc, nodes[0],
                                                caps, driver->xmlopt,
922 923 924 925 926 927 928 929 930
                                                -1, VIR_DOMAIN_XML_INACTIVE);
        if (!mig->persistent) {
            /* virDomainDefParseNode already reported
             * an error for us */
            goto error;
        }
        VIR_FREE(nodes);
    }

931 932 933 934 935
    if ((flags & QEMU_MIGRATION_COOKIE_NETWORK) &&
        virXPathBoolean("count(./network) > 0", ctxt) &&
        (!(mig->network = qemuMigrationCookieNetworkXMLParse(ctxt))))
        goto error;

936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955
    if (flags & QEMU_MIGRATION_COOKIE_NBD &&
        virXPathBoolean("boolean(./nbd)", ctxt)) {
        char *port;

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

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

956
    virObjectUnref(caps);
957 958 959 960
    return 0;

error:
    VIR_FREE(tmp);
961
    VIR_FREE(nodes);
962
    virObjectUnref(caps);
963 964 965 966 967 968
    return -1;
}


static int
qemuMigrationCookieXMLParseStr(qemuMigrationCookiePtr mig,
969
                               virQEMUDriverPtr driver,
970
                               const char *xml,
E
Eric Blake 已提交
971
                               unsigned int flags)
972 973 974
{
    xmlDocPtr doc = NULL;
    xmlXPathContextPtr ctxt = NULL;
975
    int ret = -1;
976 977 978

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

979
    if (!(doc = virXMLParseStringCtxt(xml, _("(qemu_migration_cookie)"), &ctxt)))
980 981
        goto cleanup;

982
    ret = qemuMigrationCookieXMLParse(mig, driver, doc, ctxt, flags);
983 984 985 986 987 988 989 990 991 992 993

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

    return ret;
}


static int
qemuMigrationBakeCookie(qemuMigrationCookiePtr mig,
994
                        virQEMUDriverPtr driver,
995
                        virDomainObjPtr dom,
996 997
                        char **cookieout,
                        int *cookieoutlen,
E
Eric Blake 已提交
998
                        unsigned int flags)
999
{
1000 1001
    if (!cookieout || !cookieoutlen)
        return 0;
1002 1003 1004

    *cookieoutlen = 0;

1005 1006 1007 1008
    if (flags & QEMU_MIGRATION_COOKIE_GRAPHICS &&
        qemuMigrationCookieAddGraphics(mig, driver, dom) < 0)
        return -1;

1009 1010 1011 1012
    if (flags & QEMU_MIGRATION_COOKIE_LOCKSTATE &&
        qemuMigrationCookieAddLockstate(mig, driver, dom) < 0)
        return -1;

1013 1014 1015 1016
    if (flags & QEMU_MIGRATION_COOKIE_PERSISTENT &&
        qemuMigrationCookieAddPersistent(mig, dom) < 0)
        return -1;

1017 1018 1019 1020 1021
    if (flags & QEMU_MIGRATION_COOKIE_NETWORK &&
        qemuMigrationCookieAddNetwork(mig, driver, dom) < 0) {
        return -1;
    }

1022 1023 1024 1025
    if ((flags & QEMU_MIGRATION_COOKIE_NBD) &&
        qemuMigrationCookieAddNBD(mig, driver, dom) < 0)
        return -1;

1026
    if (!(*cookieout = qemuMigrationCookieXMLFormatStr(driver, mig)))
1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037
        return -1;

    *cookieoutlen = strlen(*cookieout) + 1;

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

    return 0;
}


static qemuMigrationCookiePtr
1038
qemuMigrationEatCookie(virQEMUDriverPtr driver,
1039
                       virDomainObjPtr dom,
1040 1041
                       const char *cookiein,
                       int cookieinlen,
E
Eric Blake 已提交
1042
                       unsigned int flags)
1043 1044 1045 1046 1047 1048
{
    qemuMigrationCookiePtr mig = NULL;

    /* Parse & validate incoming cookie (if any) */
    if (cookiein && cookieinlen &&
        cookiein[cookieinlen-1] != '\0') {
1049 1050
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Migration cookie was not NULL terminated"));
1051 1052 1053 1054 1055 1056 1057 1058 1059 1060
        goto error;
    }

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

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

    if (cookiein && cookieinlen &&
        qemuMigrationCookieXMLParseStr(mig,
1061
                                       driver,
1062 1063 1064 1065
                                       cookiein,
                                       flags) < 0)
        goto error;

1066 1067 1068
    if (mig->flags & QEMU_MIGRATION_COOKIE_LOCKSTATE) {
        if (!mig->lockDriver) {
            if (virLockManagerPluginUsesState(driver->lockManager)) {
1069 1070 1071
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("Missing %s lock state for migration cookie"),
                               virLockManagerPluginGetName(driver->lockManager));
1072 1073 1074 1075
                goto error;
            }
        } else if (STRNEQ(mig->lockDriver,
                          virLockManagerPluginGetName(driver->lockManager))) {
1076 1077 1078 1079
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Source host lock driver %s different from target %s"),
                           mig->lockDriver,
                           virLockManagerPluginGetName(driver->lockManager));
1080 1081 1082 1083
            goto error;
        }
    }

1084 1085 1086 1087 1088 1089
    return mig;

error:
    qemuMigrationCookieFree(mig);
    return NULL;
}
1090

1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104
/**
 * qemuMigrationStartNBDServer:
 * @driver: qemu driver
 * @vm: domain
 *
 * Starts NBD server. This is a newer method to copy
 * storage during migration than using 'blk' and 'inc'
 * arguments in 'migrate' monitor command.
 * Error is reported here.
 *
 * Returns 0 on success, -1 otherwise.
 */
static int
qemuMigrationStartNBDServer(virQEMUDriverPtr driver,
J
Ján Tomko 已提交
1105 1106
                            virDomainObjPtr vm,
                            const char *listenAddr)
1107 1108 1109 1110 1111 1112
{
    int ret = -1;
    qemuDomainObjPrivatePtr priv = vm->privateData;
    unsigned short port = 0;
    char *diskAlias = NULL;
    size_t i;
1113 1114 1115 1116 1117 1118
    const char *host;

    if (STREQ(listenAddr, "[::]"))
        host = "::";
    else
        host = listenAddr;
1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139

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

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

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

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

        if (!port &&
            ((virPortAllocatorAcquire(driver->remotePorts, &port) < 0) ||
1140
             (qemuMonitorNBDServerStart(priv->mon, host, port) < 0))) {
1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161
            qemuDomainObjExitMonitor(driver, vm);
            goto cleanup;
        }

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

    priv->nbdPort = port;
    ret = 0;

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

1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195
/**
 * qemuMigrationDriveMirror:
 * @driver: qemu driver
 * @vm: domain
 * @mig: migration cookie
 * @host: where are we migrating to
 * @speed: how much should the copying be limited
 * @migrate_flags: migrate monitor command flags
 *
 * Run drive-mirror to feed NBD server running on dst and wait
 * till the process switches into another phase where writes go
 * simultaneously to both source and destination. And this switch
 * is what we are waiting for before proceeding with the next
 * disk. On success, update @migrate_flags so we don't tell
 * 'migrate' command to do the very same operation.
 *
 * Returns 0 on success (@migrate_flags updated),
 *        -1 otherwise.
 */
static int
qemuMigrationDriveMirror(virQEMUDriverPtr driver,
                         virDomainObjPtr vm,
                         qemuMigrationCookiePtr mig,
                         const char *host,
                         unsigned long speed,
                         unsigned int *migrate_flags)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int ret = -1;
    int mon_ret;
    int port;
    size_t i, lastGood = 0;
    char *diskAlias = NULL;
    char *nbd_dest = NULL;
1196
    char *hoststr = NULL;
1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215
    unsigned int mirror_flags = VIR_DOMAIN_BLOCK_REBASE_REUSE_EXT;
    virErrorPtr err = NULL;

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

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

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

1216 1217 1218 1219 1220 1221 1222 1223 1224 1225
    /* escape literal IPv6 address */
    if (strchr(host, ':')) {
        if (virAsprintf(&hoststr, "[%s]", host) < 0) {
            virReportOOMError();
            goto error;
        }
    } else if (VIR_STRDUP(hoststr, host) < 0) {
        goto error;
    }

1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241
    if (*migrate_flags & QEMU_MONITOR_MIGRATE_NON_SHARED_INC)
        mirror_flags |= VIR_DOMAIN_BLOCK_REBASE_SHALLOW;

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

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

        VIR_FREE(diskAlias);
        VIR_FREE(nbd_dest);
        if ((virAsprintf(&diskAlias, "%s%s",
                         QEMU_DRIVE_HOST_PREFIX, disk->info.alias) < 0) ||
            (virAsprintf(&nbd_dest, "nbd:%s:%d:exportname=%s",
1242
                         hoststr, port, diskAlias) < 0)) {
1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310
            virReportOOMError();
            goto error;
        }

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

        if (mon_ret < 0)
            goto error;

        lastGood = i;

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

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

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

            if (mon_ret < 0)
                goto error;

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

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

            virObjectUnlock(vm);

            nanosleep(&ts, NULL);

            virObjectLock(vm);
        }
    }

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

cleanup:
    VIR_FREE(diskAlias);
    VIR_FREE(nbd_dest);
1311
    VIR_FREE(hoststr);
1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346
    return ret;

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

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

        VIR_FREE(diskAlias);
        if (virAsprintf(&diskAlias, "%s%s",
                        QEMU_DRIVE_HOST_PREFIX, disk->info.alias) < 0) {
            virReportOOMError();
            continue;
        }
        if (qemuDomainObjEnterMonitorAsync(driver, vm,
                                           QEMU_ASYNC_JOB_MIGRATION_OUT) == 0) {
            if (qemuMonitorBlockJob(priv->mon, diskAlias, NULL, 0,
                                    NULL, BLOCK_JOB_ABORT, true) < 0) {
                VIR_WARN("Unable to cancel block-job on '%s'", diskAlias);
            }
            qemuDomainObjExitMonitor(driver, vm);
        } else {
            VIR_WARN("Unable to enter monitor. No block job cancelled");
        }
    }
    if (err)
        virSetError(err);
    virFreeError(err);
    goto cleanup;
}
1347

1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371

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

    if (!mig->nbd)
        return;

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

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

    qemuDomainObjExitMonitor(driver, vm);

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

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 1405 1406 1407 1408 1409 1410 1411
static void
qemuMigrationCancelDriveMirror(qemuMigrationCookiePtr mig,
                               virQEMUDriverPtr driver,
                               virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    size_t i;
    char *diskAlias = NULL;

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

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

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

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

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

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

cleanup:
    VIR_FREE(diskAlias);
    return;
}

1412 1413 1414 1415 1416 1417 1418 1419 1420
/* 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. */
1421
bool
1422
qemuMigrationIsAllowed(virQEMUDriverPtr driver, virDomainObjPtr vm,
1423
                       virDomainDefPtr def, bool remote)
1424
{
1425
    int nsnapshots;
1426 1427
    bool forbid;
    int i;
1428 1429 1430

    if (vm) {
        if (qemuProcessAutoDestroyActive(driver, vm)) {
1431 1432
            virReportError(VIR_ERR_OPERATION_INVALID,
                           "%s", _("domain is marked for auto destroy"));
1433 1434
            return false;
        }
1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447

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

            if (nsnapshots > 0) {
                virReportError(VIR_ERR_OPERATION_INVALID,
                               _("cannot migrate domain with %d snapshots"),
                               nsnapshots);
                return false;
            }
1448
        }
1449

E
Eric Blake 已提交
1450 1451
        if (virDomainHasDiskMirror(vm)) {
            virReportError(VIR_ERR_OPERATION_INVALID, "%s",
1452
                           _("domain has an active block job"));
E
Eric Blake 已提交
1453 1454
            return false;
        }
1455 1456 1457

        def = vm->def;
    }
1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472

    /* 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",
1473
                       _("domain has assigned non-USB host devices"));
1474 1475 1476 1477 1478 1479
        return false;
    }

    return true;
}

1480 1481 1482 1483 1484
static bool
qemuMigrationIsSafe(virDomainDefPtr def)
{
    int i;

1485
    for (i = 0; i < def->ndisks; i++) {
1486 1487
        virDomainDiskDefPtr disk = def->disks[i];

1488 1489
        /* Our code elsewhere guarantees shared disks are either readonly (in
         * which case cache mode doesn't matter) or used with cache=none */
1490
        if (disk->src &&
1491 1492 1493
            !disk->shared &&
            !disk->readonly &&
            disk->cachemode != VIR_DOMAIN_DISK_CACHE_DISABLE) {
1494
            int rc;
1495 1496

            if (disk->type == VIR_DOMAIN_DISK_TYPE_FILE) {
1497 1498 1499
                if ((rc = virStorageFileIsSharedFS(disk->src)) < 0)
                    return false;
                else if (rc == 0)
1500
                    continue;
1501
                if ((rc = virStorageFileIsClusterFS(disk->src)) < 0)
1502
                    return false;
1503 1504
                else if (rc == 1)
                    continue;
1505 1506 1507
            } else if (disk->type == VIR_DOMAIN_DISK_TYPE_NETWORK &&
                       disk->protocol == VIR_DOMAIN_DISK_PROTOCOL_RBD) {
                continue;
1508
            }
1509

1510 1511 1512
            virReportError(VIR_ERR_MIGRATE_UNSAFE, "%s",
                           _("Migration may lead to data corruption if disks"
                             " use cache != none"));
1513 1514 1515 1516 1517 1518 1519
            return false;
        }
    }

    return true;
}

1520 1521 1522 1523
/** qemuMigrationSetOffline
 * Pause domain for non-live migration.
 */
int
1524
qemuMigrationSetOffline(virQEMUDriverPtr driver,
1525 1526 1527
                        virDomainObjPtr vm)
{
    int ret;
1528
    VIR_DEBUG("driver=%p vm=%p", driver, vm);
1529 1530
    ret = qemuProcessStopCPUs(driver, vm, VIR_DOMAIN_PAUSED_MIGRATION,
                              QEMU_ASYNC_JOB_MIGRATION_OUT);
1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544
    if (ret == 0) {
        virDomainEventPtr event;

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

    return ret;
}


1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585
static int
qemuMigrationSetCompression(virQEMUDriverPtr driver,
                            virDomainObjPtr vm,
                            enum qemuDomainAsyncJob job)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int ret;

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

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

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

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

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


1586
static int
1587
qemuMigrationUpdateJobStatus(virQEMUDriverPtr driver,
1588
                             virDomainObjPtr vm,
1589 1590
                             const char *job,
                             enum qemuDomainAsyncJob asyncJob)
1591
{
1592
    qemuDomainObjPrivatePtr priv = vm->privateData;
1593
    int ret;
1594 1595
    bool wait_for_spice = false;
    bool spice_migrated = false;
1596 1597 1598
    qemuMonitorMigrationStatus status;

    memset(&status, 0, sizeof(status));
1599

1600 1601 1602 1603
    /* 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 &&
1604
        virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_SEAMLESS_MIGRATION))
1605 1606
        wait_for_spice = true;

1607 1608 1609
    ret = qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob);
    if (ret < 0) {
        /* Guest already exited; nothing further to update.  */
1610 1611
        return -1;
    }
1612
    ret = qemuMonitorGetMigrationStatus(priv->mon, &status);
1613 1614

    /* If qemu says migrated, check spice */
1615 1616 1617
    if (wait_for_spice &&
        ret == 0 &&
        status.status == QEMU_MONITOR_MIGRATION_STATUS_COMPLETED)
1618 1619 1620
        ret = qemuMonitorGetSpiceMigrationStatus(priv->mon,
                                                 &spice_migrated);

1621
    qemuDomainObjExitMonitor(driver, vm);
1622

1623 1624
    priv->job.status = status;

1625
    if (ret < 0 || virTimeMillisNow(&priv->job.info.timeElapsed) < 0) {
1626
        priv->job.info.type = VIR_DOMAIN_JOB_FAILED;
1627 1628
        return -1;
    }
1629
    priv->job.info.timeElapsed -= priv->job.start;
1630

1631
    ret = -1;
1632
    switch (priv->job.status.status) {
1633
    case QEMU_MONITOR_MIGRATION_STATUS_INACTIVE:
1634
        priv->job.info.type = VIR_DOMAIN_JOB_NONE;
1635 1636
        virReportError(VIR_ERR_OPERATION_FAILED,
                       _("%s: %s"), job, _("is not active"));
1637 1638 1639
        break;

    case QEMU_MONITOR_MIGRATION_STATUS_ACTIVE:
1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654
        priv->job.info.fileTotal = priv->job.status.disk_total;
        priv->job.info.fileRemaining = priv->job.status.disk_remaining;
        priv->job.info.fileProcessed = priv->job.status.disk_transferred;

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

        priv->job.info.dataTotal =
            priv->job.status.ram_total + priv->job.status.disk_total;
        priv->job.info.dataRemaining =
            priv->job.status.ram_remaining + priv->job.status.disk_remaining;
        priv->job.info.dataProcessed =
            priv->job.status.ram_transferred +
            priv->job.status.disk_transferred;
1655 1656 1657 1658 1659

        ret = 0;
        break;

    case QEMU_MONITOR_MIGRATION_STATUS_COMPLETED:
1660 1661
        if ((wait_for_spice && spice_migrated) || (!wait_for_spice))
            priv->job.info.type = VIR_DOMAIN_JOB_COMPLETED;
1662 1663 1664 1665
        ret = 0;
        break;

    case QEMU_MONITOR_MIGRATION_STATUS_ERROR:
1666
        priv->job.info.type = VIR_DOMAIN_JOB_FAILED;
1667 1668
        virReportError(VIR_ERR_OPERATION_FAILED,
                       _("%s: %s"), job, _("unexpectedly failed"));
1669 1670 1671
        break;

    case QEMU_MONITOR_MIGRATION_STATUS_CANCELLED:
1672
        priv->job.info.type = VIR_DOMAIN_JOB_CANCELLED;
1673 1674
        virReportError(VIR_ERR_OPERATION_ABORTED,
                       _("%s: %s"), job, _("canceled by client"));
1675 1676 1677 1678 1679 1680 1681
        break;
    }

    return ret;
}


1682
static int
1683
qemuMigrationWaitForCompletion(virQEMUDriverPtr driver, virDomainObjPtr vm,
1684 1685
                               enum qemuDomainAsyncJob asyncJob,
                               virConnectPtr dconn)
1686
{
1687
    qemuDomainObjPrivatePtr priv = vm->privateData;
1688 1689
    const char *job;

1690 1691
    switch (priv->job.asyncJob) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
1692 1693
        job = _("migration job");
        break;
1694
    case QEMU_ASYNC_JOB_SAVE:
1695 1696
        job = _("domain save job");
        break;
1697
    case QEMU_ASYNC_JOB_DUMP:
1698 1699 1700 1701 1702
        job = _("domain core dump job");
        break;
    default:
        job = _("job");
    }
1703

1704
    priv->job.info.type = VIR_DOMAIN_JOB_UNBOUNDED;
1705

1706
    while (priv->job.info.type == VIR_DOMAIN_JOB_UNBOUNDED) {
1707 1708 1709
        /* Poll every 50ms for progress & to allow cancellation */
        struct timespec ts = { .tv_sec = 0, .tv_nsec = 50 * 1000 * 1000ull };

1710
        if (qemuMigrationUpdateJobStatus(driver, vm, job, asyncJob) < 0)
1711 1712
            goto cleanup;

1713
        if (dconn && virConnectIsAlive(dconn) <= 0) {
1714 1715
            virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("Lost connection to destination host"));
1716 1717 1718
            goto cleanup;
        }

1719
        virObjectUnlock(vm);
1720 1721 1722

        nanosleep(&ts, NULL);

1723
        virObjectLock(vm);
1724 1725 1726
    }

cleanup:
1727
    if (priv->job.info.type == VIR_DOMAIN_JOB_COMPLETED)
1728 1729 1730
        return 0;
    else
        return -1;
1731 1732 1733
}


1734
static int
1735
qemuDomainMigrateGraphicsRelocate(virQEMUDriverPtr driver,
1736 1737 1738 1739 1740
                                  virDomainObjPtr vm,
                                  qemuMigrationCookiePtr cookie)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int ret;
1741
    char *listenAddress;
1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754

    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;

1755
    listenAddress = cookie->graphics->listen;
1756 1757 1758 1759 1760
    if (!listenAddress ||
        STREQ(listenAddress, "0.0.0.0") ||
        STREQ(listenAddress, "::"))
        listenAddress = cookie->remoteHostname;

1761 1762
    ret = qemuDomainObjEnterMonitorAsync(driver, vm,
                                         QEMU_ASYNC_JOB_MIGRATION_OUT);
1763 1764 1765
    if (ret == 0) {
        ret = qemuMonitorGraphicsRelocate(priv->mon,
                                          cookie->graphics->type,
1766
                                          listenAddress,
1767 1768 1769
                                          cookie->graphics->port,
                                          cookie->graphics->tlsPort,
                                          cookie->graphics->tlsSubject);
1770
        qemuDomainObjExitMonitor(driver, vm);
1771
    }
1772 1773 1774 1775 1776

    return ret;
}


1777
static int
1778
qemuDomainMigrateOPDRelocate(virQEMUDriverPtr driver ATTRIBUTE_UNUSED,
1779 1780 1781
                             virDomainObjPtr vm,
                             qemuMigrationCookiePtr cookie)
{
1782 1783
    virDomainNetDefPtr netptr;
    int ret = -1;
1784 1785 1786 1787 1788 1789 1790 1791 1792
    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:
1793
           break;
1794
        case VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH:
1795 1796 1797 1798 1799 1800 1801 1802
            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;
1803 1804 1805 1806 1807
        default:
            break;
        }
    }

1808 1809
    ret = 0;
cleanup:
1810 1811 1812 1813
    return ret;
}


1814 1815 1816 1817 1818 1819 1820
/* 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
1821
qemuMigrationCleanup(virQEMUDriverPtr driver,
1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874
                     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;
}

1875
/* The caller is supposed to lock the vm and start a migration job. */
1876
char *qemuMigrationBegin(virQEMUDriverPtr driver,
1877
                         virDomainObjPtr vm,
1878
                         const char *xmlin,
1879
                         const char *dname,
1880
                         char **cookieout,
1881 1882
                         int *cookieoutlen,
                         unsigned long flags)
1883 1884 1885
{
    char *rv = NULL;
    qemuMigrationCookiePtr mig = NULL;
1886
    virDomainDefPtr def = NULL;
1887
    qemuDomainObjPrivatePtr priv = vm->privateData;
1888
    virCapsPtr caps = NULL;
1889
    unsigned int cookieFlags = QEMU_MIGRATION_COOKIE_LOCKSTATE;
1890

1891
    VIR_DEBUG("driver=%p, vm=%p, xmlin=%s, dname=%s,"
1892
              " cookieout=%p, cookieoutlen=%p, flags=%lx",
1893
              driver, vm, NULLSTR(xmlin), NULLSTR(dname),
1894
              cookieout, cookieoutlen, flags);
1895

1896 1897 1898
    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
        goto cleanup;

1899 1900 1901 1902 1903 1904
    /* 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);
1905

1906
    if (!qemuMigrationIsAllowed(driver, vm, NULL, true))
1907 1908
        goto cleanup;

1909 1910 1911
    if (!(flags & VIR_MIGRATE_UNSAFE) && !qemuMigrationIsSafe(vm->def))
        goto cleanup;

1912 1913 1914
    if (flags & (VIR_MIGRATE_NON_SHARED_DISK | VIR_MIGRATE_NON_SHARED_INC) &&
        virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_NBD_SERVER)) {
        /* TODO support NBD for TUNNELLED migration */
1915 1916 1917 1918
        if (flags & VIR_MIGRATE_TUNNELLED) {
            virReportError(VIR_ERR_ARGUMENT_UNSUPPORTED, "%s",
                _("NBD in tunnelled migration is currently not supported"));
            goto cleanup;
1919
        }
1920 1921
        cookieFlags |= QEMU_MIGRATION_COOKIE_NBD;
        priv->nbdPort = 0;
1922 1923
    }

1924
    if (!(mig = qemuMigrationEatCookie(driver, vm, NULL, 0, 0)))
1925 1926 1927 1928
        goto cleanup;

    if (qemuMigrationBakeCookie(mig, driver, vm,
                                cookieout, cookieoutlen,
1929
                                cookieFlags) < 0)
1930 1931
        goto cleanup;

L
liguang 已提交
1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953
    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;
        }
    }

1954
    if (xmlin) {
1955
        if (!(def = virDomainDefParseString(xmlin, caps, driver->xmlopt,
M
Matthias Bolte 已提交
1956
                                            QEMU_EXPECTED_VIRT_TYPES,
1957 1958 1959
                                            VIR_DOMAIN_XML_INACTIVE)))
            goto cleanup;

1960
        if (!virDomainDefCheckABIStability(vm->def, def))
1961 1962
            goto cleanup;

1963
        rv = qemuDomainDefFormatLive(driver, def, false, true);
1964
    } else {
1965
        rv = qemuDomainDefFormatLive(driver, vm->def, false, true);
1966
    }
1967 1968 1969

cleanup:
    qemuMigrationCookieFree(mig);
1970
    virObjectUnref(caps);
1971
    virDomainDefFree(def);
1972 1973 1974 1975
    return rv;
}


1976 1977
/* Prepare is the first step, and it runs on the destination host.
 */
1978

1979
static void
1980
qemuMigrationPrepareCleanup(virQEMUDriverPtr driver,
1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995
                            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);
}

1996
static int
1997
qemuMigrationPrepareAny(virQEMUDriverPtr driver,
1998 1999 2000 2001 2002
                        virConnectPtr dconn,
                        const char *cookiein,
                        int cookieinlen,
                        char **cookieout,
                        int *cookieoutlen,
2003
                        virDomainDefPtr def,
L
liguang 已提交
2004
                        virStreamPtr st,
J
Ján Tomko 已提交
2005
                        unsigned int port,
L
liguang 已提交
2006
                        unsigned long flags)
2007 2008 2009 2010
{
    virDomainObjPtr vm = NULL;
    virDomainEventPtr event = NULL;
    int ret = -1;
2011
    int dataFD[2] = { -1, -1 };
2012
    qemuDomainObjPrivatePtr priv = NULL;
J
Jiri Denemark 已提交
2013
    unsigned long long now;
2014
    qemuMigrationCookiePtr mig = NULL;
2015
    bool tunnel = !!st;
J
Jiri Denemark 已提交
2016
    char *origname = NULL;
J
Jiri Denemark 已提交
2017
    char *xmlout = NULL;
L
liguang 已提交
2018
    unsigned int cookieFlags;
2019
    virCapsPtr caps = NULL;
J
Ján Tomko 已提交
2020 2021
    const char *listenAddr = NULL;
    char *migrateFrom = NULL;
2022

2023
    if (virTimeMillisNow(&now) < 0)
2024 2025
        return -1;

L
liguang 已提交
2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047
    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;
        }
    }

2048 2049 2050
    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
        goto cleanup;

2051
    if (!qemuMigrationIsAllowed(driver, NULL, def, true))
2052 2053
        goto cleanup;

J
Jiri Denemark 已提交
2054 2055 2056 2057 2058
    /* Let migration hook filter domain XML */
    if (virHookPresent(VIR_HOOK_DRIVER_QEMU)) {
        char *xml;
        int hookret;

2059
        if (!(xml = qemuDomainDefFormatXML(driver, def,
2060 2061
                                           VIR_DOMAIN_XML_SECURE |
                                           VIR_DOMAIN_XML_MIGRATABLE)))
J
Jiri Denemark 已提交
2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078
            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);
2079
                newdef = virDomainDefParseString(xmlout, caps, driver->xmlopt,
J
Jiri Denemark 已提交
2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095
                                                 QEMU_EXPECTED_VIRT_TYPES,
                                                 VIR_DOMAIN_XML_INACTIVE);
                if (!newdef)
                    goto cleanup;

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

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

J
Ján Tomko 已提交
2096 2097 2098 2099
    if (tunnel) {
        /* QEMU will be started with -incoming stdio
         * (which qemu_command might convert to exec:cat or fd:n)
         */
2100
        if (VIR_STRDUP(migrateFrom, "stdio") < 0)
J
Ján Tomko 已提交
2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132
            goto cleanup;
    } else {
        virQEMUCapsPtr qemuCaps = NULL;
        struct addrinfo *info = NULL;
        struct addrinfo hints = { .ai_flags = AI_ADDRCONFIG,
                                  .ai_socktype = SOCK_STREAM };

        if (!(qemuCaps = virQEMUCapsCacheLookupCopy(driver->qemuCapsCache,
                                                    def->emulator)))
            goto cleanup;

        /* Listen on :: instead of 0.0.0.0 if QEMU understands it
         * and there is at least one IPv6 address configured
         */
        if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_IPV6_MIGRATION) &&
            getaddrinfo("::", NULL, &hints, &info) == 0) {
            freeaddrinfo(info);
            listenAddr = "[::]";
        } else {
            listenAddr = "0.0.0.0";
        }
        virObjectUnref(qemuCaps);

        /* QEMU will be started with -incoming [::]:port
         * or -incoming 0.0.0.0:port
         */
        if (virAsprintf(&migrateFrom, "tcp:%s:%d", listenAddr, port) < 0) {
            virReportOOMError();
            goto cleanup;
        }
    }

2133
    if (!(vm = virDomainObjListAdd(driver->domains, def,
2134
                                   driver->xmlopt,
2135 2136 2137
                                   VIR_DOMAIN_OBJ_LIST_ADD_LIVE |
                                   VIR_DOMAIN_OBJ_LIST_ADD_CHECK_LIVE,
                                   NULL)))
2138
        goto cleanup;
2139

2140 2141
    def = NULL;
    priv = vm->privateData;
J
Jiri Denemark 已提交
2142 2143
    priv->origname = origname;
    origname = NULL;
2144

2145
    if (!(mig = qemuMigrationEatCookie(driver, vm, cookiein, cookieinlen,
2146 2147
                                       QEMU_MIGRATION_COOKIE_LOCKSTATE |
                                       QEMU_MIGRATION_COOKIE_NBD)))
2148 2149
        goto cleanup;

2150
    if (qemuMigrationJobStart(driver, vm, QEMU_ASYNC_JOB_MIGRATION_IN) < 0)
2151
        goto cleanup;
2152
    qemuMigrationJobSetPhase(driver, vm, QEMU_MIGRATION_PHASE_PREPARE);
2153 2154 2155 2156

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

L
liguang 已提交
2157 2158 2159
    if (flags & VIR_MIGRATE_OFFLINE)
        goto done;

2160 2161
    if (tunnel &&
        (pipe(dataFD) < 0 || virSetCloseExec(dataFD[1]) < 0)) {
2162 2163
        virReportSystemError(errno, "%s",
                             _("cannot create pipe for tunnelled migration"));
2164 2165 2166
        goto endjob;
    }

2167 2168 2169 2170
    if (qemuDomainObjBeginNestedJob(driver, vm,
                                    QEMU_ASYNC_JOB_MIGRATION_IN) < 0)
        goto endjob;

2171
    /* Start the QEMU daemon, with the same command-line arguments plus
2172
     * -incoming $migrateFrom
2173
     */
2174 2175 2176
    if (qemuProcessStart(dconn, driver, vm, migrateFrom, dataFD[0], NULL, NULL,
                         VIR_NETDEV_VPORT_PROFILE_OP_MIGRATE_IN_START,
                         VIR_QEMU_PROCESS_START_PAUSED |
2177
                         VIR_QEMU_PROCESS_START_AUTODESTROY) < 0) {
2178
        virDomainAuditStart(vm, "migrated", false);
2179 2180
        if (qemuDomainObjEndJob(driver, vm) < 0)
            vm = NULL;
2181 2182 2183
        goto endjob;
    }

2184 2185 2186 2187
    if (tunnel) {
        if (virFDStreamOpen(st, dataFD[1]) < 0) {
            virReportSystemError(errno, "%s",
                                 _("cannot pass pipe for tunnelled migration"));
2188
            goto stop;
2189
        }
2190
        dataFD[1] = -1; /* 'st' owns the FD now & will close it */
2191 2192
    }

2193 2194 2195 2196 2197
    if (flags & VIR_MIGRATE_COMPRESSED &&
        qemuMigrationSetCompression(driver, vm,
                                    QEMU_ASYNC_JOB_MIGRATION_IN) < 0)
        goto stop;

2198 2199 2200 2201 2202 2203 2204 2205
    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");
    }
2206

L
liguang 已提交
2207 2208 2209 2210 2211 2212
done:
    if (flags & VIR_MIGRATE_OFFLINE)
        cookieFlags = 0;
    else
        cookieFlags = QEMU_MIGRATION_COOKIE_GRAPHICS;

2213 2214 2215
    if (mig->nbd &&
        flags & (VIR_MIGRATE_NON_SHARED_DISK | VIR_MIGRATE_NON_SHARED_INC) &&
        virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_NBD_SERVER)) {
2216 2217 2218
        if (qemuMigrationStartNBDServer(driver, vm, listenAddr) < 0) {
            /* error already reported */
            goto endjob;
2219
        }
2220
        cookieFlags |= QEMU_MIGRATION_COOKIE_NBD;
2221 2222 2223 2224
    }

    if (qemuMigrationBakeCookie(mig, driver, vm, cookieout,
                                cookieoutlen, cookieFlags) < 0) {
2225 2226 2227 2228 2229 2230 2231
        /* 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");
    }

2232 2233 2234
    if (qemuDomainCleanupAdd(vm, qemuMigrationPrepareCleanup) < 0)
        goto endjob;

L
liguang 已提交
2235 2236 2237 2238 2239 2240
    if (!(flags & VIR_MIGRATE_OFFLINE)) {
        virDomainAuditStart(vm, "migrated", true);
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STARTED,
                                         VIR_DOMAIN_EVENT_STARTED_MIGRATED);
    }
2241

2242 2243 2244 2245
    /* 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.
     */
2246
    if (!qemuMigrationJobContinue(vm)) {
2247
        vm = NULL;
2248 2249
        virReportError(VIR_ERR_OPERATION_FAILED,
                       "%s", _("domain disappeared"));
2250
        goto cleanup;
2251
    }
2252

2253
    ret = 0;
2254 2255

cleanup:
J
Ján Tomko 已提交
2256
    VIR_FREE(migrateFrom);
J
Jiri Denemark 已提交
2257
    VIR_FREE(origname);
J
Jiri Denemark 已提交
2258
    VIR_FREE(xmlout);
2259
    virDomainDefFree(def);
2260 2261
    VIR_FORCE_CLOSE(dataFD[0]);
    VIR_FORCE_CLOSE(dataFD[1]);
2262 2263
    if (vm) {
        if (ret >= 0 || vm->persistent)
2264
            virObjectUnlock(vm);
2265 2266
        else
            qemuDomainRemoveInactive(driver, vm);
2267 2268 2269 2270
        if (ret < 0 && priv->nbdPort) {
            virPortAllocatorRelease(driver->remotePorts, priv->nbdPort);
            priv->nbdPort = 0;
        }
2271
    }
2272 2273
    if (event)
        qemuDomainEventQueue(driver, event);
2274
    qemuMigrationCookieFree(mig);
2275
    virObjectUnref(caps);
2276
    return ret;
2277

2278 2279 2280 2281
stop:
    virDomainAuditStart(vm, "migrated", false);
    qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_FAILED, 0);

2282
endjob:
2283
    if (vm && !qemuMigrationJobFinish(driver, vm)) {
2284 2285 2286
        vm = NULL;
    }
    goto cleanup;
2287 2288 2289
}


2290 2291 2292 2293 2294
/*
 * This version starts an empty VM listening on a localhost TCP port, and
 * sets up the corresponding virStream to handle the incoming data.
 */
int
2295
qemuMigrationPrepareTunnel(virQEMUDriverPtr driver,
2296 2297 2298 2299 2300 2301
                           virConnectPtr dconn,
                           const char *cookiein,
                           int cookieinlen,
                           char **cookieout,
                           int *cookieoutlen,
                           virStreamPtr st,
2302
                           virDomainDefPtr def,
L
liguang 已提交
2303
                           unsigned long flags)
2304 2305 2306 2307
{
    int ret;

    VIR_DEBUG("driver=%p, dconn=%p, cookiein=%s, cookieinlen=%d, "
2308
              "cookieout=%p, cookieoutlen=%p, st=%p, def=%p, "
2309
              "flags=%lx",
2310
              driver, dconn, NULLSTR(cookiein), cookieinlen,
2311
              cookieout, cookieoutlen, st, def, flags);
2312 2313

    ret = qemuMigrationPrepareAny(driver, dconn, cookiein, cookieinlen,
2314
                                  cookieout, cookieoutlen, def,
J
Ján Tomko 已提交
2315
                                  st, 0, flags);
2316 2317 2318 2319
    return ret;
}


2320
int
2321
qemuMigrationPrepareDirect(virQEMUDriverPtr driver,
2322
                           virConnectPtr dconn,
2323 2324 2325 2326
                           const char *cookiein,
                           int cookieinlen,
                           char **cookieout,
                           int *cookieoutlen,
2327 2328
                           const char *uri_in,
                           char **uri_out,
2329
                           virDomainDefPtr def,
L
liguang 已提交
2330
                           unsigned long flags)
2331 2332 2333 2334 2335
{
    static int port = 0;
    int this_port;
    char *hostname = NULL;
    const char *p;
J
Ján Tomko 已提交
2336
    char *uri_str = NULL;
2337
    int ret = -1;
J
Ján Tomko 已提交
2338
    virURIPtr uri;
J
Jiri Denemark 已提交
2339

2340 2341
    VIR_DEBUG("driver=%p, dconn=%p, cookiein=%s, cookieinlen=%d, "
              "cookieout=%p, cookieoutlen=%p, uri_in=%s, uri_out=%p, "
2342
              "def=%p, flags=%lx",
2343 2344
              driver, dconn, NULLSTR(cookiein), cookieinlen,
              cookieout, cookieoutlen, NULLSTR(uri_in), uri_out,
2345
              def, flags);
2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361

    /* 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 */
2362
        if ((hostname = virGetHostname()) == NULL)
2363 2364 2365
            goto cleanup;

        if (STRPREFIX(hostname, "localhost")) {
2366 2367 2368
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("hostname on destination resolved to localhost,"
                             " but migration requires an FQDN"));
2369 2370 2371 2372 2373
            goto cleanup;
        }

        /* XXX this really should have been a properly well-formed
         * URI, but we can't add in tcp:// now without breaking
2374
         * compatibility with old targets. We at least make the
2375 2376 2377
         * new targets accept both syntaxes though.
         */
        /* Caller frees */
2378
        if (virAsprintf(uri_out, "tcp:%s:%d", hostname, this_port) < 0) {
2379 2380 2381 2382 2383 2384 2385 2386
            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.
         */
J
Ján Tomko 已提交
2387
        if (!(p = STRSKIP(uri_in, "tcp:"))) {
2388 2389 2390
            virReportError(VIR_ERR_INVALID_ARG, "%s",
                           _("only tcp URIs are supported for KVM/QEMU"
                             " migrations"));
2391 2392 2393
            goto cleanup;
        }

J
Ján Tomko 已提交
2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419
        /* Convert uri_in to well-formed URI with // after tcp: */
        if (!(STRPREFIX(uri_in, "tcp://"))) {
            if (virAsprintf(&uri_str, "tcp://%s", p) < 0) {
                virReportOOMError();
                goto cleanup;
            }
        }

        uri = virURIParse(uri_str ? uri_str : uri_in);
        VIR_FREE(uri_str);

        if (uri == NULL) {
            virReportError(VIR_ERR_INVALID_ARG, _("unable to parse URI: %s"),
                           uri_in);
            goto cleanup;
        }

        if (uri->server == NULL) {
            virReportError(VIR_ERR_INVALID_ARG, _("missing host in migration"
                                                  " URI: %s"), uri_in);
            goto cleanup;
        } else {
            hostname = uri->server;
        }

        if (uri->port == 0) {
2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431
            /* 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 {
J
Ján Tomko 已提交
2432
            this_port = uri->port;
2433 2434 2435 2436 2437 2438
        }
    }

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

2439
    ret = qemuMigrationPrepareAny(driver, dconn, cookiein, cookieinlen,
2440
                                  cookieout, cookieoutlen, def,
J
Ján Tomko 已提交
2441
                                  NULL, this_port, flags);
2442 2443 2444 2445 2446 2447 2448 2449
cleanup:
    VIR_FREE(hostname);
    if (ret != 0)
        VIR_FREE(*uri_out);
    return ret;
}


2450 2451
enum qemuMigrationDestinationType {
    MIGRATION_DEST_HOST,
2452
    MIGRATION_DEST_CONNECT_HOST,
2453
    MIGRATION_DEST_UNIX,
2454
    MIGRATION_DEST_FD,
2455
};
2456

2457 2458 2459 2460
enum qemuMigrationForwardType {
    MIGRATION_FWD_DIRECT,
    MIGRATION_FWD_STREAM,
};
2461

2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472
typedef struct _qemuMigrationSpec qemuMigrationSpec;
typedef qemuMigrationSpec *qemuMigrationSpecPtr;
struct _qemuMigrationSpec {
    enum qemuMigrationDestinationType destType;
    union {
        struct {
            const char *name;
            int port;
        } host;

        struct {
2473
            char *file;
2474 2475
            int sock;
        } unix_socket;
2476 2477 2478 2479 2480

        struct {
            int qemu;
            int local;
        } fd;
2481 2482 2483 2484 2485 2486 2487
    } dest;

    enum qemuMigrationForwardType fwdType;
    union {
        virStreamPtr stream;
    } fwd;
};
2488 2489 2490

#define TUNNEL_SEND_BUF_SIZE 65536

2491 2492 2493 2494 2495 2496 2497
typedef struct _qemuMigrationIOThread qemuMigrationIOThread;
typedef qemuMigrationIOThread *qemuMigrationIOThreadPtr;
struct _qemuMigrationIOThread {
    virThread thread;
    virStreamPtr st;
    int sock;
    virError err;
2498 2499
    int wakeupRecvFD;
    int wakeupSendFD;
2500 2501 2502
};

static void qemuMigrationIOFunc(void *arg)
2503
{
2504
    qemuMigrationIOThreadPtr data = arg;
2505 2506 2507 2508 2509 2510 2511
    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);
2512 2513 2514

    if (VIR_ALLOC_N(buffer, TUNNEL_SEND_BUF_SIZE) < 0) {
        virReportOOMError();
2515
        goto abrt;
2516 2517
    }

2518 2519 2520
    fds[0].fd = data->sock;
    fds[1].fd = data->wakeupRecvFD;

2521
    for (;;) {
2522 2523 2524 2525 2526 2527 2528 2529 2530 2531
        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;
2532
            virReportSystemError(errno, "%s",
2533 2534
                                 _("poll failed in migration tunnel"));
            goto abrt;
2535
        }
2536 2537 2538 2539 2540 2541 2542

        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");
2543
            break;
2544
        }
2545

2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561
        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;
            }
2562 2563
        }

2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580
        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;
            }
        }
    }
2581

2582 2583
    if (virStreamFinish(data->st) < 0)
        goto error;
2584

2585 2586
    VIR_FREE(buffer);

2587 2588
    return;

2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600
abrt:
    err = virSaveLastError();
    if (err && err->code == VIR_ERR_OK) {
        virFreeError(err);
        err = NULL;
    }
    virStreamAbort(data->st);
    if (err) {
        virSetError(err);
        virFreeError(err);
    }

2601 2602 2603
error:
    virCopyLastError(&data->err);
    virResetLastError();
2604
    VIR_FREE(buffer);
2605 2606 2607 2608 2609 2610 2611
}


static qemuMigrationIOThreadPtr
qemuMigrationStartTunnel(virStreamPtr st,
                         int sock)
{
2612 2613
    qemuMigrationIOThreadPtr io = NULL;
    int wakeupFD[2] = { -1, -1 };
2614

2615 2616 2617 2618
    if (pipe2(wakeupFD, O_CLOEXEC) < 0) {
        virReportSystemError(errno, "%s",
                             _("Unable to make pipe"));
        goto error;
2619 2620
    }

2621 2622 2623
    if (VIR_ALLOC(io) < 0)
        goto no_memory;

2624 2625
    io->st = st;
    io->sock = sock;
2626 2627
    io->wakeupRecvFD = wakeupFD[0];
    io->wakeupSendFD = wakeupFD[1];
2628 2629 2630 2631 2632 2633

    if (virThreadCreate(&io->thread, true,
                        qemuMigrationIOFunc,
                        io) < 0) {
        virReportSystemError(errno, "%s",
                             _("Unable to create migration thread"));
2634
        goto error;
2635 2636 2637
    }

    return io;
2638 2639 2640 2641 2642 2643 2644 2645

no_memory:
    virReportOOMError();
error:
    VIR_FORCE_CLOSE(wakeupFD[0]);
    VIR_FORCE_CLOSE(wakeupFD[1]);
    VIR_FREE(io);
    return NULL;
2646 2647 2648
}

static int
2649
qemuMigrationStopTunnel(qemuMigrationIOThreadPtr io, bool error)
2650 2651
{
    int rv = -1;
2652 2653 2654 2655 2656 2657 2658 2659 2660
    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;
    }

2661 2662 2663 2664
    virThreadJoin(&io->thread);

    /* Forward error from the IO thread, to this thread */
    if (io->err.code != VIR_ERR_OK) {
2665 2666 2667 2668
        if (error)
            rv = 0;
        else
            virSetError(&io->err);
2669 2670 2671 2672 2673 2674 2675
        virResetError(&io->err);
        goto cleanup;
    }

    rv = 0;

cleanup:
2676 2677
    VIR_FORCE_CLOSE(io->wakeupSendFD);
    VIR_FORCE_CLOSE(io->wakeupRecvFD);
2678 2679
    VIR_FREE(io);
    return rv;
2680 2681
}

2682
static int
2683
qemuMigrationConnect(virQEMUDriverPtr driver,
2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704
                     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);
2705
        virObjectUnref(sock);
2706 2707 2708 2709 2710
    }
    if (virSecurityManagerClearSocketLabel(driver->securityManager, vm->def) < 0 ||
        spec->dest.fd.qemu == -1)
        goto cleanup;

2711 2712 2713 2714 2715 2716 2717
    /* Migration expects a blocking FD */
    if (virSetBlocking(spec->dest.fd.qemu, true) < 0) {
        virReportSystemError(errno, _("Unable to set FD %d blocking"),
                             spec->dest.fd.qemu);
        goto cleanup;
    }

2718 2719 2720 2721 2722 2723 2724 2725 2726
    ret = 0;

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

2727
static int
2728
qemuMigrationRun(virQEMUDriverPtr driver,
2729 2730 2731 2732 2733 2734 2735
                 virDomainObjPtr vm,
                 const char *cookiein,
                 int cookieinlen,
                 char **cookieout,
                 int *cookieoutlen,
                 unsigned long flags,
                 unsigned long resource,
2736 2737
                 qemuMigrationSpecPtr spec,
                 virConnectPtr dconn)
2738
{
2739
    int ret = -1;
2740 2741
    unsigned int migrate_flags = QEMU_MONITOR_MIGRATE_BACKGROUND;
    qemuDomainObjPrivatePtr priv = vm->privateData;
2742
    qemuMigrationCookiePtr mig = NULL;
2743
    qemuMigrationIOThreadPtr iothread = NULL;
2744
    int fd = -1;
2745
    unsigned long migrate_speed = resource ? resource : priv->migMaxBandwidth;
2746
    virErrorPtr orig_err = NULL;
2747
    unsigned int cookieFlags = 0;
2748 2749 2750 2751 2752 2753 2754

    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);
2755

2756 2757 2758 2759 2760 2761 2762 2763 2764 2765
    if (flags & VIR_MIGRATE_NON_SHARED_DISK) {
        migrate_flags |= QEMU_MONITOR_MIGRATE_NON_SHARED_DISK;
        cookieFlags |= QEMU_MIGRATION_COOKIE_NBD;
    }

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

2766 2767
    if (virLockManagerPluginUsesState(driver->lockManager) &&
        !cookieout) {
2768 2769 2770 2771
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Migration with lock driver %s requires"
                         " cookie support"),
                       virLockManagerPluginGetName(driver->lockManager));
2772 2773 2774
        return -1;
    }

2775 2776 2777
    mig = qemuMigrationEatCookie(driver, vm, cookiein, cookieinlen,
                                 cookieFlags | QEMU_MIGRATION_COOKIE_GRAPHICS);
    if (!mig)
2778 2779 2780 2781 2782
        goto cleanup;

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

2783 2784 2785 2786 2787 2788 2789
    /* this will update migrate_flags on success */
    if (qemuMigrationDriveMirror(driver, vm, mig, spec->dest.host.name,
                                 migrate_speed, &migrate_flags) < 0) {
        /* error reported by helper func */
        goto cleanup;
    }

2790
    /* Before EnterMonitor, since qemuMigrationSetOffline already does that */
2791 2792 2793 2794 2795 2796
    if (!(flags & VIR_MIGRATE_LIVE) &&
        virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING) {
        if (qemuMigrationSetOffline(driver, vm) < 0)
            goto cleanup;
    }

2797 2798 2799 2800 2801
    if (flags & VIR_MIGRATE_COMPRESSED &&
        qemuMigrationSetCompression(driver, vm,
                                    QEMU_ASYNC_JOB_MIGRATION_OUT) < 0)
        goto cleanup;

2802 2803
    if (qemuDomainObjEnterMonitorAsync(driver, vm,
                                       QEMU_ASYNC_JOB_MIGRATION_OUT) < 0)
2804 2805
        goto cleanup;

2806 2807 2808 2809
    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 */
2810
        qemuDomainObjExitMonitor(driver, vm);
2811 2812 2813 2814 2815 2816
        virReportError(VIR_ERR_OPERATION_ABORTED, _("%s: %s"),
                       qemuDomainAsyncJobTypeToString(priv->job.asyncJob),
                       _("canceled by client"));
        goto cleanup;
    }

2817
    if (qemuMonitorSetMigrationSpeed(priv->mon, migrate_speed) < 0) {
2818
        qemuDomainObjExitMonitor(driver, vm);
2819 2820
        goto cleanup;
    }
2821

2822 2823
    /* connect to the destination qemu if needed */
    if (spec->destType == MIGRATION_DEST_CONNECT_HOST &&
2824
        qemuMigrationConnect(driver, vm, spec) < 0) {
2825
        qemuDomainObjExitMonitor(driver, vm);
2826
        goto cleanup;
2827
    }
2828

2829 2830 2831 2832 2833 2834 2835
    switch (spec->destType) {
    case MIGRATION_DEST_HOST:
        ret = qemuMonitorMigrateToHost(priv->mon, migrate_flags,
                                       spec->dest.host.name,
                                       spec->dest.host.port);
        break;

2836 2837 2838 2839
    case MIGRATION_DEST_CONNECT_HOST:
        /* handled above and transformed into MIGRATION_DEST_FD */
        break;

2840
    case MIGRATION_DEST_UNIX:
2841
        if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_UNIX)) {
2842 2843 2844 2845 2846 2847 2848 2849 2850
            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;
2851 2852

    case MIGRATION_DEST_FD:
2853
        if (spec->fwdType != MIGRATION_FWD_DIRECT) {
2854
            fd = spec->dest.fd.local;
2855 2856
            spec->dest.fd.local = -1;
        }
2857 2858 2859 2860
        ret = qemuMonitorMigrateToFd(priv->mon, migrate_flags,
                                     spec->dest.fd.qemu);
        VIR_FORCE_CLOSE(spec->dest.fd.qemu);
        break;
2861
    }
2862
    qemuDomainObjExitMonitor(driver, vm);
2863
    if (ret < 0)
2864 2865
        goto cleanup;
    ret = -1;
2866 2867

    if (!virDomainObjIsActive(vm)) {
2868 2869
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("guest unexpectedly quit"));
2870 2871 2872 2873 2874 2875
        goto cleanup;
    }

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

2876 2877 2878 2879 2880 2881 2882 2883
    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;
2884

2885 2886 2887 2888 2889 2890 2891
        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;
        }
2892 2893
    }

2894 2895
    if (spec->fwdType != MIGRATION_FWD_DIRECT &&
        !(iothread = qemuMigrationStartTunnel(spec->fwd.stream, fd)))
2896 2897
        goto cancel;

2898
    if (qemuMigrationWaitForCompletion(driver, vm,
2899 2900
                                       QEMU_ASYNC_JOB_MIGRATION_OUT,
                                       dconn) < 0)
2901
        goto cleanup;
2902

2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915
    /* 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;
    }

2916
    ret = 0;
2917

2918
cleanup:
2919 2920 2921
    if (ret < 0 && !orig_err)
        orig_err = virSaveLastError();

2922 2923 2924
    /* cancel any outstanding NBD jobs */
    qemuMigrationCancelDriveMirror(mig, driver, vm);

2925
    if (spec->fwdType != MIGRATION_FWD_DIRECT) {
2926
        if (iothread && qemuMigrationStopTunnel(iothread, ret < 0) < 0)
2927
            ret = -1;
2928
        VIR_FORCE_CLOSE(fd);
2929
    }
2930

2931 2932 2933
    cookieFlags |= QEMU_MIGRATION_COOKIE_NETWORK;
    if (flags & VIR_MIGRATE_PERSIST_DEST)
        cookieFlags |= QEMU_MIGRATION_COOKIE_PERSISTENT;
2934
    if (ret == 0 &&
2935 2936
        qemuMigrationBakeCookie(mig, driver, vm, cookieout,
                                cookieoutlen, cookieFlags) < 0) {
2937
        VIR_WARN("Unable to encode migration cookie");
2938
    }
2939

2940 2941
    qemuMigrationCookieFree(mig);

2942 2943 2944 2945 2946
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }

2947 2948
    return ret;

2949
cancel:
2950 2951
    orig_err = virSaveLastError();

2952
    if (virDomainObjIsActive(vm)) {
2953 2954
        if (qemuDomainObjEnterMonitorAsync(driver, vm,
                                           QEMU_ASYNC_JOB_MIGRATION_OUT) == 0) {
2955
            qemuMonitorMigrateCancel(priv->mon);
2956
            qemuDomainObjExitMonitor(driver, vm);
2957
        }
2958
    }
2959 2960 2961 2962 2963 2964
    goto cleanup;
}

/* Perform migration using QEMU's native TCP migrate support,
 * not encrypted obviously
 */
2965
static int doNativeMigrate(virQEMUDriverPtr driver,
2966 2967 2968 2969 2970 2971 2972
                           virDomainObjPtr vm,
                           const char *uri,
                           const char *cookiein,
                           int cookieinlen,
                           char **cookieout,
                           int *cookieoutlen,
                           unsigned long flags,
2973 2974
                           unsigned long resource,
                           virConnectPtr dconn)
2975
{
2976
    qemuDomainObjPrivatePtr priv = vm->privateData;
M
Martin Kletzander 已提交
2977
    virURIPtr uribits = NULL;
2978
    int ret = -1;
2979 2980 2981 2982 2983 2984 2985 2986
    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://")) {
2987
        char *tmp;
2988
        /* HACK: source host generates bogus URIs, so fix them up */
2989
        if (virAsprintf(&tmp, "tcp://%s", uri + strlen("tcp:")) < 0) {
2990 2991 2992
            virReportOOMError();
            return -1;
        }
M
Martin Kletzander 已提交
2993
        uribits = virURIParse(tmp);
2994
        VIR_FREE(tmp);
2995
    } else {
M
Martin Kletzander 已提交
2996
        uribits = virURIParse(uri);
2997
    }
2998
    if (!uribits)
2999 3000
        return -1;

3001
    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_FD))
3002 3003
        spec.destType = MIGRATION_DEST_CONNECT_HOST;
    else
3004
        spec.destType = MIGRATION_DEST_HOST;
3005 3006 3007
    spec.dest.host.name = uribits->server;
    spec.dest.host.port = uribits->port;
    spec.fwdType = MIGRATION_FWD_DIRECT;
3008

3009
    ret = qemuMigrationRun(driver, vm, cookiein, cookieinlen, cookieout,
3010
                           cookieoutlen, flags, resource, &spec, dconn);
3011 3012 3013 3014

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

3015
    virURIFree(uribits);
3016 3017 3018 3019 3020

    return ret;
}


3021
static int doTunnelMigrate(virQEMUDriverPtr driver,
3022 3023 3024 3025 3026 3027 3028
                           virDomainObjPtr vm,
                           virStreamPtr st,
                           const char *cookiein,
                           int cookieinlen,
                           char **cookieout,
                           int *cookieoutlen,
                           unsigned long flags,
3029 3030
                           unsigned long resource,
                           virConnectPtr dconn)
3031 3032
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
3033
    virNetSocketPtr sock = NULL;
3034 3035
    int ret = -1;
    qemuMigrationSpec spec;
3036
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
3037 3038 3039 3040 3041 3042

    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);

3043 3044 3045
    if (!virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_FD) &&
        !virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_UNIX) &&
        !virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_EXEC)) {
3046 3047
        virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                       _("Source qemu is too old to support tunnelled migration"));
3048
        virObjectUnref(cfg);
3049
        return -1;
3050 3051 3052 3053 3054
    }

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

3055
    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_FD)) {
3056 3057 3058 3059 3060 3061
        int fds[2];

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

3062
        if (pipe2(fds, O_CLOEXEC) == 0) {
3063 3064 3065 3066
            spec.dest.fd.qemu = fds[1];
            spec.dest.fd.local = fds[0];
        }
        if (spec.dest.fd.qemu == -1 ||
3067
            virSecurityManagerSetImageFDLabel(driver->securityManager, vm->def,
3068
                                              spec.dest.fd.qemu) < 0) {
3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079
            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",
3080
                        cfg->libDir, vm->def->name) < 0) {
3081 3082 3083 3084 3085
            virReportOOMError();
            goto cleanup;
        }

        if (virNetSocketNewListenUNIX(spec.dest.unix_socket.file, 0700,
3086
                                      cfg->user, cfg->group,
3087 3088 3089 3090 3091 3092 3093
                                      &sock) < 0 ||
            virNetSocketListen(sock, 1) < 0)
            goto cleanup;

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

3094
    ret = qemuMigrationRun(driver, vm, cookiein, cookieinlen, cookieout,
3095
                           cookieoutlen, flags, resource, &spec, dconn);
3096

3097
cleanup:
3098 3099 3100 3101
    if (spec.destType == MIGRATION_DEST_FD) {
        VIR_FORCE_CLOSE(spec.dest.fd.qemu);
        VIR_FORCE_CLOSE(spec.dest.fd.local);
    } else {
3102
        virObjectUnref(sock);
3103 3104
        VIR_FREE(spec.dest.unix_socket.file);
    }
3105

3106
    virObjectUnref(cfg);
3107 3108 3109 3110
    return ret;
}


3111 3112 3113 3114
/* 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 */
3115
static int doPeer2PeerMigrate2(virQEMUDriverPtr driver,
3116
                               virConnectPtr sconn ATTRIBUTE_UNUSED,
3117 3118
                               virConnectPtr dconn,
                               virDomainObjPtr vm,
3119
                               const char *dconnuri,
3120 3121 3122
                               unsigned long flags,
                               const char *dname,
                               unsigned long resource)
3123 3124 3125
{
    virDomainPtr ddomain = NULL;
    char *uri_out = NULL;
3126
    char *cookie = NULL;
3127 3128 3129 3130 3131
    char *dom_xml = NULL;
    int cookielen = 0, ret;
    virErrorPtr orig_err = NULL;
    int cancelled;
    virStreamPtr st = NULL;
3132
    VIR_DEBUG("driver=%p, sconn=%p, dconn=%p, vm=%p, dconnuri=%s, "
3133
              "flags=%lx, dname=%s, resource=%lu",
3134 3135
              driver, sconn, dconn, vm, NULLSTR(dconnuri),
              flags, NULLSTR(dname), resource);
3136

3137 3138 3139 3140 3141
    /* 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,
3142 3143
                                        QEMU_DOMAIN_FORMAT_LIVE_FLAGS |
                                        VIR_DOMAIN_XML_MIGRATABLE)))
3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158
        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;

3159
        qemuDomainObjEnterRemote(vm);
3160 3161
        ret = dconn->driver->domainMigratePrepareTunnel
            (dconn, st, flags, dname, resource, dom_xml);
3162
        qemuDomainObjExitRemote(vm);
3163
    } else {
3164
        qemuDomainObjEnterRemote(vm);
3165 3166 3167
        ret = dconn->driver->domainMigratePrepare2
            (dconn, &cookie, &cookielen, NULL, &uri_out,
             flags, dname, resource, dom_xml);
3168
        qemuDomainObjExitRemote(vm);
3169 3170 3171
    }
    VIR_FREE(dom_xml);
    if (ret == -1)
3172 3173 3174
        goto cleanup;

    /* the domain may have shutdown or crashed while we had the locks dropped
3175
     * in qemuDomainObjEnterRemote, so check again
3176 3177
     */
    if (!virDomainObjIsActive(vm)) {
3178 3179
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("guest unexpectedly quit"));
3180 3181 3182
        goto cleanup;
    }

3183 3184
    if (!(flags & VIR_MIGRATE_TUNNELLED) &&
        (uri_out == NULL)) {
3185 3186
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("domainMigratePrepare2 did not set uri"));
3187 3188
        cancelled = 1;
        goto finish;
3189 3190
    }

3191 3192 3193 3194
    /* Perform the migration.  The driver isn't supposed to return
     * until the migration is complete.
     */
    VIR_DEBUG("Perform %p", sconn);
3195
    qemuMigrationJobSetPhase(driver, vm, QEMU_MIGRATION_PHASE_PERFORM2);
3196
    if (flags & VIR_MIGRATE_TUNNELLED)
3197 3198
        ret = doTunnelMigrate(driver, vm, st,
                              NULL, 0, NULL, NULL,
3199
                              flags, resource, dconn);
3200 3201 3202 3203
    else
        ret = doNativeMigrate(driver, vm, uri_out,
                              cookie, cookielen,
                              NULL, NULL, /* No out cookie with v2 migration */
3204
                              flags, resource, dconn);
3205 3206 3207 3208

    /* Perform failed. Make sure Finish doesn't overwrite the error */
    if (ret < 0)
        orig_err = virSaveLastError();
3209

3210 3211 3212 3213
    /* If Perform returns < 0, then we need to cancel the VM
     * startup on the destination
     */
    cancelled = ret < 0 ? 1 : 0;
3214 3215

finish:
3216 3217 3218 3219
    /* 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.
     */
3220
    dname = dname ? dname : vm->def->name;
3221
    VIR_DEBUG("Finish2 %p ret=%d", dconn, ret);
3222
    qemuDomainObjEnterRemote(vm);
3223
    ddomain = dconn->driver->domainMigrateFinish2
3224
        (dconn, dname, cookie, cookielen,
3225
         uri_out ? uri_out : dconnuri, flags, cancelled);
3226
    qemuDomainObjExitRemote(vm);
3227

3228 3229
cleanup:
    if (ddomain) {
3230
        virObjectUnref(ddomain);
3231 3232 3233 3234
        ret = 0;
    } else {
        ret = -1;
    }
3235

3236
    virObjectUnref(st);
3237 3238 3239 3240 3241 3242

    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
    VIR_FREE(uri_out);
3243
    VIR_FREE(cookie);
3244 3245

    return ret;
3246 3247 3248
}


3249 3250 3251 3252
/* 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 */
3253
static int doPeer2PeerMigrate3(virQEMUDriverPtr driver,
3254 3255 3256
                               virConnectPtr sconn,
                               virConnectPtr dconn,
                               virDomainObjPtr vm,
3257
                               const char *xmlin,
3258
                               const char *dconnuri,
3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272 3273 3274
                               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;
3275
    VIR_DEBUG("driver=%p, sconn=%p, dconn=%p, vm=%p, xmlin=%s, "
3276
              "dconnuri=%s, uri=%s, flags=%lx, dname=%s, resource=%lu",
3277 3278 3279
              driver, sconn, dconn, vm, NULLSTR(xmlin),
              NULLSTR(dconnuri), NULLSTR(uri), flags,
              NULLSTR(dname), resource);
3280

3281 3282 3283 3284 3285
    /* 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.  */

3286
    dom_xml = qemuMigrationBegin(driver, vm, xmlin, dname,
3287
                                 &cookieout, &cookieoutlen, flags);
3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302
    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;

3303
        qemuDomainObjEnterRemote(vm);
3304 3305 3306 3307
        ret = dconn->driver->domainMigratePrepareTunnel3
            (dconn, st, cookiein, cookieinlen,
             &cookieout, &cookieoutlen,
             flags, dname, resource, dom_xml);
3308
        qemuDomainObjExitRemote(vm);
3309
    } else {
3310
        qemuDomainObjEnterRemote(vm);
3311 3312
        ret = dconn->driver->domainMigratePrepare3
            (dconn, cookiein, cookieinlen, &cookieout, &cookieoutlen,
3313
             uri, &uri_out, flags, dname, resource, dom_xml);
3314
        qemuDomainObjExitRemote(vm);
3315 3316 3317 3318 3319
    }
    VIR_FREE(dom_xml);
    if (ret == -1)
        goto cleanup;

L
liguang 已提交
3320 3321 3322 3323 3324 3325 3326 3327
    if (flags & VIR_MIGRATE_OFFLINE) {
        VIR_DEBUG("Offline migration, skipping Perform phase");
        VIR_FREE(cookieout);
        cookieoutlen = 0;
        cancelled = 0;
        goto finish;
    }

3328 3329
    if (!(flags & VIR_MIGRATE_TUNNELLED) &&
        (uri_out == NULL)) {
3330 3331
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("domainMigratePrepare3 did not set uri"));
3332 3333 3334 3335 3336 3337 3338 3339 3340
        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.
     */
3341
    VIR_DEBUG("Perform3 %p uri=%s uri_out=%s", sconn, uri, uri_out);
3342
    qemuMigrationJobSetPhase(driver, vm, QEMU_MIGRATION_PHASE_PERFORM3);
3343 3344 3345 3346 3347 3348 3349 3350 3351
    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,
3352
                              flags, resource, dconn);
3353 3354 3355 3356
    else
        ret = doNativeMigrate(driver, vm, uri_out,
                              cookiein, cookieinlen,
                              &cookieout, &cookieoutlen,
3357
                              flags, resource, dconn);
3358 3359

    /* Perform failed. Make sure Finish doesn't overwrite the error */
3360
    if (ret < 0) {
3361
        orig_err = virSaveLastError();
3362 3363 3364 3365
    } else {
        qemuMigrationJobSetPhase(driver, vm,
                                 QEMU_MIGRATION_PHASE_PERFORM3_DONE);
    }
3366 3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385

    /* 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;
3386
    qemuDomainObjEnterRemote(vm);
3387
    ddomain = dconn->driver->domainMigrateFinish3
3388
        (dconn, dname, cookiein, cookieinlen, &cookieout, &cookieoutlen,
3389
         dconnuri, uri_out ? uri_out : uri, flags, cancelled);
3390
    qemuDomainObjExitRemote(vm);
3391

3392 3393 3394 3395 3396 3397 3398
    /* 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.
3399
     */
3400
    cancelled = ddomain == NULL ? 1 : 0;
3401

3402 3403 3404 3405 3406 3407
    /* 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();

3408 3409 3410 3411
    /*
     * If cancelled, then src VM will be restarted, else
     * it will be killed
     */
3412
    VIR_DEBUG("Confirm3 %p cancelled=%d vm=%p", sconn, cancelled, vm);
3413 3414 3415 3416 3417 3418 3419
    VIR_FREE(cookiein);
    cookiein = cookieout;
    cookieinlen = cookieoutlen;
    cookieout = NULL;
    cookieoutlen = 0;
    ret = qemuMigrationConfirm(driver, sconn, vm,
                               cookiein, cookieinlen,
3420
                               flags, cancelled);
3421 3422 3423 3424
    /* 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.
     */
3425 3426 3427
    if (ret < 0)
        VIR_WARN("Guest %s probably left in 'paused' state on source",
                 vm->def->name);
3428 3429 3430

 cleanup:
    if (ddomain) {
3431
        virObjectUnref(ddomain);
3432 3433 3434 3435 3436
        ret = 0;
    } else {
        ret = -1;
    }

3437
    virObjectUnref(st);
3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450

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

    return ret;
}


3451
static int doPeer2PeerMigrate(virQEMUDriverPtr driver,
3452
                              virConnectPtr sconn,
3453
                              virDomainObjPtr vm,
3454
                              const char *xmlin,
3455
                              const char *dconnuri,
3456 3457 3458
                              const char *uri,
                              unsigned long flags,
                              const char *dname,
3459 3460
                              unsigned long resource,
                              bool *v3proto)
3461 3462 3463 3464
{
    int ret = -1;
    virConnectPtr dconn = NULL;
    bool p2p;
3465
    virErrorPtr orig_err = NULL;
3466
    bool offline = false;
3467
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
3468

3469
    VIR_DEBUG("driver=%p, sconn=%p, vm=%p, xmlin=%s, dconnuri=%s, "
3470
              "uri=%s, flags=%lx, dname=%s, resource=%lu",
3471 3472
              driver, sconn, vm, NULLSTR(xmlin), NULLSTR(dconnuri),
              NULLSTR(uri), flags, NULLSTR(dname), resource);
3473 3474 3475 3476 3477

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

3478
    qemuDomainObjEnterRemote(vm);
3479
    dconn = virConnectOpen(dconnuri);
3480
    qemuDomainObjExitRemote(vm);
3481
    if (dconn == NULL) {
3482
        virReportError(VIR_ERR_OPERATION_FAILED,
3483 3484
                       _("Failed to connect to remote libvirt URI %s: %s"),
                       dconnuri, virGetLastErrorMessage());
3485
        virObjectUnref(cfg);
3486 3487 3488
        return -1;
    }

3489 3490
    if (virConnectSetKeepAlive(dconn, cfg->keepAliveInterval,
                               cfg->keepAliveCount) < 0)
3491 3492
        goto cleanup;

3493
    qemuDomainObjEnterRemote(vm);
3494 3495
    p2p = VIR_DRV_SUPPORTS_FEATURE(dconn->driver, dconn,
                                   VIR_DRV_FEATURE_MIGRATION_P2P);
3496
        /* v3proto reflects whether the caller used Perform3, but with
3497
         * p2p migrate, regardless of whether Perform2 or Perform3
3498 3499 3500 3501
         * 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
3502 3503 3504
    if (flags & VIR_MIGRATE_OFFLINE)
        offline = VIR_DRV_SUPPORTS_FEATURE(dconn->driver, dconn,
                                           VIR_DRV_FEATURE_MIGRATION_OFFLINE);
3505
    qemuDomainObjExitRemote(vm);
3506

3507
    if (!p2p) {
3508 3509
        virReportError(VIR_ERR_OPERATION_FAILED, "%s",
                       _("Destination libvirt does not support peer-to-peer migration protocol"));
3510 3511 3512
        goto cleanup;
    }

L
liguang 已提交
3513 3514 3515 3516 3517 3518 3519
    if (flags & VIR_MIGRATE_OFFLINE && !offline) {
        virReportError(VIR_ERR_ARGUMENT_UNSUPPORTED, "%s",
                       _("offline migration is not supported by "
                         "the destination host"));
        goto cleanup;
    }

3520
    /* domain may have been stopped while we were talking to remote daemon */
L
liguang 已提交
3521
    if (!virDomainObjIsActive(vm) && !(flags & VIR_MIGRATE_OFFLINE)) {
3522 3523
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("guest unexpectedly quit"));
3524 3525 3526
        goto cleanup;
    }

3527 3528 3529 3530 3531 3532 3533
    /* 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;

3534
    if (*v3proto)
3535
        ret = doPeer2PeerMigrate3(driver, sconn, dconn, vm, xmlin,
3536
                                  dconnuri, uri, flags, dname, resource);
3537 3538
    else
        ret = doPeer2PeerMigrate2(driver, sconn, dconn, vm,
3539
                                  dconnuri, flags, dname, resource);
3540 3541

cleanup:
3542
    orig_err = virSaveLastError();
3543
    qemuDomainObjEnterRemote(vm);
3544
    virConnectClose(dconn);
3545
    qemuDomainObjExitRemote(vm);
3546 3547 3548 3549
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
3550
    virObjectUnref(cfg);
3551 3552 3553 3554
    return ret;
}


3555 3556 3557 3558 3559 3560
/*
 * 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
3561
qemuMigrationPerformJob(virQEMUDriverPtr driver,
3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574
                        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)
3575 3576 3577 3578
{
    virDomainEventPtr event = NULL;
    int ret = -1;
    int resume = 0;
3579
    virErrorPtr orig_err = NULL;
3580
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
3581

3582
    if (qemuMigrationJobStart(driver, vm, QEMU_ASYNC_JOB_MIGRATION_OUT) < 0)
3583 3584
        goto cleanup;

L
liguang 已提交
3585
    if (!virDomainObjIsActive(vm) && !(flags & VIR_MIGRATE_OFFLINE)) {
3586 3587
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("domain is not running"));
3588 3589 3590
        goto endjob;
    }

3591
    if (!qemuMigrationIsAllowed(driver, vm, NULL, true))
3592
        goto endjob;
3593

3594
    if (!(flags & VIR_MIGRATE_UNSAFE) && !qemuMigrationIsSafe(vm->def))
3595
        goto endjob;
3596

J
Jiri Denemark 已提交
3597
    resume = virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING;
3598 3599

    if ((flags & (VIR_MIGRATE_TUNNELLED | VIR_MIGRATE_PEER2PEER))) {
3600 3601 3602
        ret = doPeer2PeerMigrate(driver, conn, vm, xmlin,
                                 dconnuri, uri, flags, dname,
                                 resource, &v3proto);
3603
    } else {
3604 3605 3606
        qemuMigrationJobSetPhase(driver, vm, QEMU_MIGRATION_PHASE_PERFORM2);
        ret = doNativeMigrate(driver, vm, uri, cookiein, cookieinlen,
                              cookieout, cookieoutlen,
3607
                              flags, resource, NULL);
3608
    }
3609 3610
    if (ret < 0)
        goto endjob;
3611

3612 3613 3614 3615
    /*
     * In v3 protocol, the source VM is not killed off until the
     * confirm step.
     */
3616
    if (!v3proto) {
3617 3618
        qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_MIGRATED,
                        VIR_QEMU_PROCESS_STOP_MIGRATED);
3619
        virDomainAuditStop(vm, "migrated");
3620 3621 3622
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         VIR_DOMAIN_EVENT_STOPPED_MIGRATED);
3623
    }
3624
    resume = 0;
3625 3626

endjob:
3627 3628 3629
    if (ret < 0)
        orig_err = virSaveLastError();

J
Jiri Denemark 已提交
3630
    if (resume && virDomainObjGetState(vm, NULL) == VIR_DOMAIN_PAUSED) {
3631
        /* we got here through some sort of failure; start the domain again */
J
Jiri Denemark 已提交
3632
        if (qemuProcessStartCPUs(driver, vm, conn,
3633 3634
                                 VIR_DOMAIN_RUNNING_MIGRATION_CANCELED,
                                 QEMU_ASYNC_JOB_MIGRATION_OUT) < 0) {
3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646
            /* 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);
    }
3647

3648
    if (!qemuMigrationJobFinish(driver, vm)) {
3649 3650 3651 3652 3653
        vm = NULL;
    } else if (!virDomainObjIsActive(vm) &&
               (!vm->persistent ||
                (ret == 0 && (flags & VIR_MIGRATE_UNDEFINE_SOURCE)))) {
        if (flags & VIR_MIGRATE_UNDEFINE_SOURCE)
3654
            virDomainDeleteConfig(cfg->configDir, cfg->autostartDir, vm);
3655
        qemuDomainRemoveInactive(driver, vm);
3656 3657 3658
        vm = NULL;
    }

3659 3660 3661 3662 3663
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }

3664 3665
cleanup:
    if (vm)
3666
        virObjectUnlock(vm);
3667 3668
    if (event)
        qemuDomainEventQueue(driver, event);
3669
    virObjectUnref(cfg);
3670 3671 3672 3673 3674 3675 3676
    return ret;
}

/*
 * This implements perform phase of v3 migration protocol.
 */
static int
3677
qemuMigrationPerformPhase(virQEMUDriverPtr driver,
3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690
                          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;
3691
    bool hasrefs;
3692 3693 3694 3695 3696 3697 3698 3699 3700 3701

    /* 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);
3702 3703
    virQEMUCloseCallbacksUnset(driver->closeCallbacks, vm,
                               qemuMigrationCleanup);
3704 3705 3706 3707

    resume = virDomainObjGetState(vm, NULL) == VIR_DOMAIN_RUNNING;
    ret = doNativeMigrate(driver, vm, uri, cookiein, cookieinlen,
                          cookieout, cookieoutlen,
3708
                          flags, resource, NULL);
3709 3710 3711 3712 3713

    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,
3714 3715
                                 VIR_DOMAIN_RUNNING_MIGRATION_CANCELED,
                                 QEMU_ASYNC_JOB_MIGRATION_OUT) < 0) {
3716 3717 3718 3719 3720 3721
            /* 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);
3722
        }
3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733

        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);

3734 3735
    if (virQEMUCloseCallbacksSet(driver->closeCallbacks, vm, conn,
                                 qemuMigrationCleanup) < 0)
3736 3737
        goto endjob;

3738 3739
endjob:
    if (ret < 0)
3740
        hasrefs = qemuMigrationJobFinish(driver, vm);
3741
    else
3742 3743
        hasrefs = qemuMigrationJobContinue(vm);
    if (!hasrefs) {
3744 3745
        vm = NULL;
    } else if (!virDomainObjIsActive(vm) && !vm->persistent) {
3746
        qemuDomainRemoveInactive(driver, vm);
3747
        vm = NULL;
3748
    }
3749 3750 3751

cleanup:
    if (vm)
3752
        virObjectUnlock(vm);
3753 3754 3755 3756 3757
    if (event)
        qemuDomainEventQueue(driver, event);
    return ret;
}

3758
int
3759
qemuMigrationPerform(virQEMUDriverPtr driver,
3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783
                     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) {
3784 3785
            virReportError(VIR_ERR_OPERATION_INVALID,
                           "%s", _("received unexpected cookie with P2P migration"));
3786 3787 3788 3789 3790 3791 3792 3793 3794
            return -1;
        }

        return qemuMigrationPerformJob(driver, conn, vm, xmlin, dconnuri, uri,
                                       cookiein, cookieinlen, cookieout,
                                       cookieoutlen, flags, dname, resource,
                                       v3proto);
    } else {
        if (dconnuri) {
3795 3796
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("Unexpected dconnuri parameter with non-peer2peer migration"));
3797 3798 3799 3800 3801 3802 3803
            return -1;
        }

        if (v3proto) {
            return qemuMigrationPerformPhase(driver, conn, vm, uri,
                                             cookiein, cookieinlen,
                                             cookieout, cookieoutlen,
3804
                                             flags, resource);
3805 3806 3807 3808 3809 3810 3811 3812
        } else {
            return qemuMigrationPerformJob(driver, conn, vm, xmlin, dconnuri,
                                           uri, cookiein, cookieinlen,
                                           cookieout, cookieoutlen, flags,
                                           dname, resource, v3proto);
        }
    }
}
3813

3814
static int
3815 3816 3817 3818 3819 3820 3821
qemuMigrationVPAssociatePortProfiles(virDomainDefPtr def) {
    int i;
    int last_good_net = -1;
    virDomainNetDefPtr net;

    for (i = 0; i < def->nnets; i++) {
        net = def->nets[i];
3822
        if (virDomainNetGetActualType(net) == VIR_DOMAIN_NET_TYPE_DIRECT) {
3823
            if (virNetDevVPortProfileAssociate(net->ifname,
3824
                                               virDomainNetGetActualVirtPortProfile(net),
3825
                                               &net->mac,
3826
                                               virDomainNetGetActualDirectDev(net),
3827
                                               -1,
3828
                                               def->uuid,
3829 3830
                                               VIR_NETDEV_VPORT_PROFILE_OP_MIGRATE_IN_FINISH,
                                               false) < 0) {
3831 3832 3833
                virReportError(VIR_ERR_OPERATION_FAILED,
                               _("Port profile Associate failed for %s"),
                               net->ifname);
3834
                goto err_exit;
3835 3836
            }
            VIR_DEBUG("Port profile Associate succeeded for %s", net->ifname);
3837

3838
            if (virNetDevMacVLanVPortProfileRegisterCallback(net->ifname, &net->mac,
3839 3840 3841 3842
                                                             virDomainNetGetActualDirectDev(net), def->uuid,
                                                             virDomainNetGetActualVirtPortProfile(net),
                                                             VIR_NETDEV_VPORT_PROFILE_OP_CREATE))
                goto err_exit;
3843 3844 3845 3846
        }
        last_good_net = i;
    }

3847
    return 0;
3848 3849 3850 3851

err_exit:
    for (i = 0; i < last_good_net; i++) {
        net = def->nets[i];
3852
        if (virDomainNetGetActualType(net) == VIR_DOMAIN_NET_TYPE_DIRECT) {
3853
            ignore_value(virNetDevVPortProfileDisassociate(net->ifname,
3854
                                                           virDomainNetGetActualVirtPortProfile(net),
3855
                                                           &net->mac,
3856
                                                           virDomainNetGetActualDirectDev(net),
3857
                                                           -1,
3858
                                                           VIR_NETDEV_VPORT_PROFILE_OP_MIGRATE_IN_FINISH));
3859 3860
        }
    }
3861
    return -1;
3862 3863 3864 3865
}


virDomainPtr
3866
qemuMigrationFinish(virQEMUDriverPtr driver,
3867 3868
                    virConnectPtr dconn,
                    virDomainObjPtr vm,
3869 3870 3871 3872
                    const char *cookiein,
                    int cookieinlen,
                    char **cookieout,
                    int *cookieoutlen,
3873
                    unsigned long flags,
3874 3875
                    int retcode,
                    bool v3proto)
3876 3877 3878 3879
{
    virDomainPtr dom = NULL;
    virDomainEventPtr event = NULL;
    int newVM = 1;
3880
    qemuMigrationCookiePtr mig = NULL;
3881
    virErrorPtr orig_err = NULL;
3882
    int cookie_flags = 0;
J
Jiri Denemark 已提交
3883
    qemuDomainObjPrivatePtr priv = vm->privateData;
3884
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);
3885
    virCapsPtr caps = NULL;
3886

3887
    VIR_DEBUG("driver=%p, dconn=%p, vm=%p, cookiein=%s, cookieinlen=%d, "
3888
              "cookieout=%p, cookieoutlen=%p, flags=%lx, retcode=%d",
3889 3890
              driver, dconn, vm, NULLSTR(cookiein), cookieinlen,
              cookieout, cookieoutlen, flags, retcode);
3891

3892 3893 3894
    if (!(caps = virQEMUDriverGetCapabilities(driver, false)))
        goto cleanup;

3895
    if (!qemuMigrationJobIsActive(vm, QEMU_ASYNC_JOB_MIGRATION_IN))
3896 3897
        goto cleanup;

3898 3899 3900
    qemuMigrationJobStartPhase(driver, vm,
                               v3proto ? QEMU_MIGRATION_PHASE_FINISH3
                                       : QEMU_MIGRATION_PHASE_FINISH2);
3901

3902 3903
    qemuDomainCleanupRemove(vm, qemuMigrationPrepareCleanup);

3904
    cookie_flags = QEMU_MIGRATION_COOKIE_NETWORK;
3905 3906 3907 3908 3909
    if (flags & VIR_MIGRATE_PERSIST_DEST)
        cookie_flags |= QEMU_MIGRATION_COOKIE_PERSISTENT;

    if (!(mig = qemuMigrationEatCookie(driver, vm, cookiein,
                                       cookieinlen, cookie_flags)))
3910
        goto endjob;
3911 3912 3913 3914 3915

    /* 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
3916
        if (!virDomainObjIsActive(vm) && !(flags & VIR_MIGRATE_OFFLINE)) {
3917 3918
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("guest unexpectedly quit"));
3919
            goto endjob;
3920 3921
        }

L
liguang 已提交
3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934
        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");
3935
        }
3936

3937 3938
        qemuMigrationStopNBDServer(driver, vm, mig);

3939
        if (flags & VIR_MIGRATE_PERSIST_DEST) {
3940
            virDomainDefPtr vmdef;
3941 3942 3943
            if (vm->persistent)
                newVM = 0;
            vm->persistent = 1;
3944 3945 3946
            if (mig->persistent)
                vm->newDef = vmdef = mig->persistent;
            else
3947
                vmdef = virDomainObjGetPersistentDef(caps, driver->xmlopt, vm);
3948
            if (!vmdef || virDomainSaveConfig(cfg->configDir, vmdef) < 0) {
3949 3950 3951 3952 3953 3954 3955 3956 3957
                /* 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.
                 */
3958 3959

                /*
3960 3961
                 * However, in v3 protocol, the source VM is still available
                 * to restart during confirm() step, so we kill it off now.
3962 3963
                 */
                if (v3proto) {
L
liguang 已提交
3964 3965 3966 3967 3968
                    if (!(flags & VIR_MIGRATE_OFFLINE)) {
                        qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_FAILED,
                                        VIR_QEMU_PROCESS_STOP_MIGRATED);
                        virDomainAuditStop(vm, "failed");
                    }
3969 3970
                    if (newVM)
                        vm->persistent = 0;
3971
                }
A
Alex Jia 已提交
3972
                if (!vmdef)
3973 3974
                    virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                                   _("can't get vmdef"));
3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985 3986 3987
                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
3988
        if (!(flags & VIR_MIGRATE_PAUSED) && !(flags & VIR_MIGRATE_OFFLINE)) {
3989 3990 3991 3992
            /* 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
3993
            if (qemuProcessStartCPUs(driver, vm, dconn,
3994 3995
                                     VIR_DOMAIN_RUNNING_MIGRATED,
                                     QEMU_ASYNC_JOB_MIGRATION_IN) < 0) {
3996
                if (virGetLastError() == NULL)
3997 3998
                    virReportError(VIR_ERR_INTERNAL_ERROR,
                                   "%s", _("resume operation failed"));
3999 4000 4001 4002 4003
                /* Need to save the current error, in case shutting
                 * down the process overwrites it
                 */
                orig_err = virSaveLastError();

4004 4005 4006 4007 4008 4009 4010 4011 4012
                /*
                 * 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) {
4013 4014
                    qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_FAILED,
                                    VIR_QEMU_PROCESS_STOP_MIGRATED);
4015
                    virDomainAuditStop(vm, "failed");
4016 4017 4018 4019
                    event = virDomainEventNewFromObj(vm,
                                                     VIR_DOMAIN_EVENT_STOPPED,
                                                     VIR_DOMAIN_EVENT_STOPPED_FAILED);
                }
4020 4021 4022 4023
                goto endjob;
            }
        }

4024
        dom = virGetDomain(dconn, vm->def->name, vm->def->uuid);
4025

L
liguang 已提交
4026
        if (!(flags & VIR_MIGRATE_OFFLINE)) {
4027
            event = virDomainEventNewFromObj(vm,
L
liguang 已提交
4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038
                                             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);
            }
4039
        }
L
liguang 已提交
4040 4041

        if (virDomainObjIsActive(vm) &&
4042
            virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm) < 0) {
4043 4044 4045
            VIR_WARN("Failed to save status on vm %s", vm->def->name);
            goto endjob;
        }
4046 4047 4048

        /* Guest is successfully running, so cancel previous auto destroy */
        qemuProcessAutoDestroyRemove(driver, vm);
L
liguang 已提交
4049
    } else if (!(flags & VIR_MIGRATE_OFFLINE)) {
4050 4051
        qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_FAILED,
                        VIR_QEMU_PROCESS_STOP_MIGRATED);
4052
        virDomainAuditStop(vm, "failed");
4053 4054 4055 4056 4057
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         VIR_DOMAIN_EVENT_STOPPED_FAILED);
    }

4058 4059 4060
    if (qemuMigrationBakeCookie(mig, driver, vm, cookieout, cookieoutlen, 0) < 0)
        VIR_WARN("Unable to encode migration cookie");

4061
endjob:
E
Eric Blake 已提交
4062 4063 4064
    if (qemuMigrationJobFinish(driver, vm) == 0) {
        vm = NULL;
    } else if (!vm->persistent && !virDomainObjIsActive(vm)) {
4065
        qemuDomainRemoveInactive(driver, vm);
E
Eric Blake 已提交
4066
        vm = NULL;
4067
    }
4068 4069

cleanup:
J
Jiri Denemark 已提交
4070 4071
    if (vm) {
        VIR_FREE(priv->origname);
4072
        virObjectUnlock(vm);
J
Jiri Denemark 已提交
4073
    }
4074 4075
    if (event)
        qemuDomainEventQueue(driver, event);
4076
    qemuMigrationCookieFree(mig);
4077 4078 4079 4080
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
4081
    virObjectUnref(caps);
4082
    virObjectUnref(cfg);
4083 4084
    return dom;
}
4085

4086

4087
int qemuMigrationConfirm(virQEMUDriverPtr driver,
4088 4089 4090 4091
                         virConnectPtr conn,
                         virDomainObjPtr vm,
                         const char *cookiein,
                         int cookieinlen,
E
Eric Blake 已提交
4092
                         unsigned int flags,
4093
                         int retcode)
4094 4095 4096 4097
{
    qemuMigrationCookiePtr mig;
    virDomainEventPtr event = NULL;
    int rv = -1;
4098 4099
    virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver);

4100
    VIR_DEBUG("driver=%p, conn=%p, vm=%p, cookiein=%s, cookieinlen=%d, "
4101
              "flags=%x, retcode=%d",
4102 4103
              driver, conn, vm, NULLSTR(cookiein), cookieinlen,
              flags, retcode);
4104

4105
    virCheckFlags(QEMU_MIGRATION_FLAGS, -1);
E
Eric Blake 已提交
4106

4107 4108 4109 4110 4111
    qemuMigrationJobSetPhase(driver, vm,
                             retcode == 0
                             ? QEMU_MIGRATION_PHASE_CONFIRM3
                             : QEMU_MIGRATION_PHASE_CONFIRM3_CANCELLED);

4112
    if (!(mig = qemuMigrationEatCookie(driver, vm, cookiein, cookieinlen, 0)))
4113
        goto cleanup;
4114

L
liguang 已提交
4115 4116 4117
    if (flags & VIR_MIGRATE_OFFLINE)
        goto done;

4118 4119 4120 4121
    /* Did the migration go as planned?  If yes, kill off the
     * domain object, but if no, resume CPUs
     */
    if (retcode == 0) {
4122 4123
        qemuProcessStop(driver, vm, VIR_DOMAIN_SHUTOFF_MIGRATED,
                        VIR_QEMU_PROCESS_STOP_MIGRATED);
4124
        virDomainAuditStop(vm, "migrated");
4125 4126 4127 4128 4129 4130

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

4131 4132 4133
        /* cancel any outstanding NBD jobs */
        qemuMigrationCancelDriveMirror(mig, driver, vm);

4134 4135 4136 4137 4138
        /* 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,
4139 4140
                                 VIR_DOMAIN_RUNNING_MIGRATED,
                                 QEMU_ASYNC_JOB_MIGRATION_OUT) < 0) {
4141
            if (virGetLastError() == NULL)
4142 4143
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               "%s", _("resume operation failed"));
4144
            goto cleanup;
4145 4146 4147 4148 4149
        }

        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_MIGRATED);
4150
        if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm) < 0) {
4151
            VIR_WARN("Failed to save status on vm %s", vm->def->name);
4152
            goto cleanup;
4153 4154 4155
        }
    }

L
liguang 已提交
4156
done:
4157 4158 4159 4160 4161 4162
    qemuMigrationCookieFree(mig);
    rv = 0;

cleanup:
    if (event)
        qemuDomainEventQueue(driver, event);
4163
    virObjectUnref(cfg);
4164 4165 4166 4167
    return rv;
}


4168
/* Helper function called while vm is active.  */
4169
int
4170
qemuMigrationToFile(virQEMUDriverPtr driver, virDomainObjPtr vm,
4171 4172
                    int fd, off_t offset, const char *path,
                    const char *compressor,
E
Eric Blake 已提交
4173
                    bool bypassSecurityDriver,
4174
                    enum qemuDomainAsyncJob asyncJob)
4175 4176 4177 4178 4179
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int ret = -1;
    int rc;
    bool restoreLabel = false;
4180 4181
    virCommandPtr cmd = NULL;
    int pipeFD[2] = { -1, -1 };
4182
    unsigned long saveMigBandwidth = priv->migMaxBandwidth;
4183
    char *errbuf = NULL;
4184 4185 4186 4187 4188

    /* 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,
4189 4190
                                     QEMU_DOMAIN_MIG_BANDWIDTH_MAX);
        priv->migMaxBandwidth = QEMU_DOMAIN_MIG_BANDWIDTH_MAX;
4191
        qemuDomainObjExitMonitor(driver, vm);
4192
    }
4193

4194
    if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_FD) &&
4195
        (!compressor || pipe(pipeFD) == 0)) {
4196
        /* All right! We can use fd migration, which means that qemu
4197 4198 4199
         * 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.  */
4200
        if (virSecurityManagerSetImageFDLabel(driver->securityManager, vm->def,
4201
                                              compressor ? pipeFD[1] : fd) < 0)
4202
            goto cleanup;
4203 4204 4205
        bypassSecurityDriver = true;
    } else {
        /* Phooey - we have to fall back on exec migration, where qemu
E
Eric Blake 已提交
4206 4207
         * has to popen() the file by name, and block devices have to be
         * given cgroup ACL permission.  We might also stumble on
4208 4209
         * a race present in some qemu versions where it does a wait()
         * that botches pclose.  */
4210 4211 4212
        if (virCgroupHasController(priv->cgroup,
                                   VIR_CGROUP_CONTROLLER_DEVICES)) {
            rc = virCgroupAllowDevicePath(priv->cgroup, path,
4213
                                          VIR_CGROUP_DEVICE_RW);
4214
            virDomainAuditCgroupPath(vm, priv->cgroup, "allow", path, "rw", rc);
E
Eric Blake 已提交
4215 4216 4217
            if (rc == 1) {
                /* path was not a device, no further need for cgroup */
            } else if (rc < 0) {
4218 4219 4220 4221 4222
                virReportSystemError(-rc,
                                     _("Unable to allow device %s for %s"),
                                     path, vm->def->name);
                goto cleanup;
            }
4223
        }
4224 4225
        if ((!bypassSecurityDriver) &&
            virSecurityManagerSetSavedStateLabel(driver->securityManager,
4226
                                                 vm->def, path) < 0)
4227
            goto cleanup;
4228
        restoreLabel = true;
4229 4230
    }

4231
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) < 0)
4232 4233
        goto cleanup;

4234 4235 4236
    if (!compressor) {
        const char *args[] = { "cat", NULL };

4237
        if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_MIGRATE_QEMU_FD) &&
4238 4239 4240 4241 4242 4243 4244 4245 4246 4247 4248 4249 4250 4251 4252 4253
            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
        };
4254 4255 4256 4257
        if (pipeFD[0] != -1) {
            cmd = virCommandNewArgs(args);
            virCommandSetInputFD(cmd, pipeFD[0]);
            virCommandSetOutputFD(cmd, &fd);
4258 4259
            virCommandSetErrorBuffer(cmd, &errbuf);
            virCommandDoAsyncIO(cmd);
4260 4261 4262
            if (virSetCloseExec(pipeFD[1]) < 0) {
                virReportSystemError(errno, "%s",
                                     _("Unable to set cloexec flag"));
4263
                qemuDomainObjExitMonitor(driver, vm);
4264 4265 4266
                goto cleanup;
            }
            if (virCommandRunAsync(cmd, NULL) < 0) {
4267
                qemuDomainObjExitMonitor(driver, vm);
4268 4269 4270 4271 4272 4273 4274
                goto cleanup;
            }
            rc = qemuMonitorMigrateToFd(priv->mon,
                                        QEMU_MONITOR_MIGRATE_BACKGROUND,
                                        pipeFD[1]);
            if (VIR_CLOSE(pipeFD[0]) < 0 ||
                VIR_CLOSE(pipeFD[1]) < 0)
4275
                VIR_WARN("failed to close intermediate pipe");
4276 4277 4278 4279 4280
        } else {
            rc = qemuMonitorMigrateToFile(priv->mon,
                                          QEMU_MONITOR_MIGRATE_BACKGROUND,
                                          args, path, offset);
        }
4281
    }
4282
    qemuDomainObjExitMonitor(driver, vm);
4283 4284 4285 4286

    if (rc < 0)
        goto cleanup;

4287
    rc = qemuMigrationWaitForCompletion(driver, vm, asyncJob, NULL);
4288 4289 4290 4291

    if (rc < 0)
        goto cleanup;

4292 4293 4294
    if (cmd && virCommandWait(cmd, NULL) < 0)
        goto cleanup;

4295 4296 4297
    ret = 0;

cleanup:
4298 4299 4300 4301
    /* Restore max migration bandwidth */
    if (qemuDomainObjEnterMonitorAsync(driver, vm, asyncJob) == 0) {
        qemuMonitorSetMigrationSpeed(priv->mon, saveMigBandwidth);
        priv->migMaxBandwidth = saveMigBandwidth;
4302
        qemuDomainObjExitMonitor(driver, vm);
4303 4304
    }

4305 4306
    VIR_FORCE_CLOSE(pipeFD[0]);
    VIR_FORCE_CLOSE(pipeFD[1]);
4307 4308 4309 4310 4311
    if (cmd) {
        VIR_DEBUG("Compression binary stderr: %s", NULLSTR(errbuf));
        VIR_FREE(errbuf);
        virCommandFree(cmd);
    }
4312 4313
    if (restoreLabel && (!bypassSecurityDriver) &&
        virSecurityManagerRestoreSavedStateLabel(driver->securityManager,
4314
                                                 vm->def, path) < 0)
4315 4316
        VIR_WARN("failed to restore save state label on %s", path);

4317 4318 4319
    if (virCgroupHasController(priv->cgroup,
                               VIR_CGROUP_CONTROLLER_DEVICES)) {
        rc = virCgroupDenyDevicePath(priv->cgroup, path,
4320
                                     VIR_CGROUP_DEVICE_RWM);
4321
        virDomainAuditCgroupPath(vm, priv->cgroup, "deny", path, "rwm", rc);
4322 4323 4324 4325 4326 4327
        if (rc < 0)
            VIR_WARN("Unable to deny device %s for %s %d",
                     path, vm->def->name, rc);
    }
    return ret;
}
4328 4329

int
4330
qemuMigrationJobStart(virQEMUDriverPtr driver,
4331 4332 4333 4334 4335
                      virDomainObjPtr vm,
                      enum qemuDomainAsyncJob job)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

4336
    if (qemuDomainObjBeginAsyncJob(driver, vm, job) < 0)
4337 4338
        return -1;

4339
    if (job == QEMU_ASYNC_JOB_MIGRATION_IN) {
4340
        qemuDomainObjSetAsyncJobMask(vm, QEMU_JOB_NONE);
4341 4342
    } else {
        qemuDomainObjSetAsyncJobMask(vm, DEFAULT_JOB_MASK |
4343
                                     JOB_MASK(QEMU_JOB_SUSPEND) |
4344 4345
                                     JOB_MASK(QEMU_JOB_MIGRATION_OP));
    }
4346 4347 4348 4349 4350 4351 4352

    priv->job.info.type = VIR_DOMAIN_JOB_UNBOUNDED;

    return 0;
}

void
4353
qemuMigrationJobSetPhase(virQEMUDriverPtr driver,
4354 4355 4356 4357 4358 4359 4360 4361 4362 4363 4364 4365 4366 4367 4368 4369
                         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
4370
qemuMigrationJobStartPhase(virQEMUDriverPtr driver,
4371 4372 4373
                           virDomainObjPtr vm,
                           enum qemuMigrationJobPhase phase)
{
4374
    virObjectRef(vm);
4375 4376 4377
    qemuMigrationJobSetPhase(driver, vm, phase);
}

4378
bool
4379 4380
qemuMigrationJobContinue(virDomainObjPtr vm)
{
4381
    qemuDomainObjReleaseAsyncJob(vm);
4382
    return virObjectUnref(vm);
4383 4384 4385 4386 4387 4388 4389 4390 4391 4392 4393 4394 4395 4396 4397 4398
}

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");

4399
        virReportError(VIR_ERR_OPERATION_INVALID, msg, vm->def->name);
4400 4401 4402 4403 4404
        return false;
    }
    return true;
}

4405
bool
4406
qemuMigrationJobFinish(virQEMUDriverPtr driver, virDomainObjPtr vm)
4407 4408 4409
{
    return qemuDomainObjEndAsyncJob(driver, vm);
}