qemu_domain.c 47.9 KB
Newer Older
1 2 3
/*
 * qemu_domain.h: QEMU domain private state
 *
4
 * Copyright (C) 2006-2011 Red Hat, Inc.
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
 * Copyright (C) 2006 Daniel P. Berrange
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with this library; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307  USA
 *
 * Author: Daniel P. Berrange <berrange@redhat.com>
 */

#include <config.h>

#include "qemu_domain.h"
#include "qemu_command.h"
28
#include "qemu_capabilities.h"
29
#include "qemu_migration.h"
30 31 32 33
#include "memory.h"
#include "logging.h"
#include "virterror_internal.h"
#include "c-ctype.h"
34
#include "cpu/cpu.h"
35
#include "ignore-value.h"
36
#include "uuid.h"
E
Eric Blake 已提交
37
#include "virfile.h"
38
#include "domain_event.h"
39
#include "virtime.h"
40

41
#include <sys/time.h>
42
#include <fcntl.h>
43

44 45 46 47 48 49
#include <libxml/xpathInternals.h>

#define VIR_FROM_THIS VIR_FROM_QEMU

#define QEMU_NAMESPACE_HREF "http://libvirt.org/schemas/domain/qemu/1.0"

50 51 52 53 54 55 56
VIR_ENUM_DECL(qemuDomainJob)
VIR_ENUM_IMPL(qemuDomainJob, QEMU_JOB_LAST,
              "none",
              "query",
              "destroy",
              "suspend",
              "modify",
57
              "abort",
58
              "migration operation",
59 60 61 62 63 64 65 66 67 68 69 70 71
              "none",   /* async job is never stored in job.active */
              "async nested",
);

VIR_ENUM_DECL(qemuDomainAsyncJob)
VIR_ENUM_IMPL(qemuDomainAsyncJob, QEMU_ASYNC_JOB_LAST,
              "none",
              "migration out",
              "migration in",
              "save",
              "dump",
);

72

J
Jiri Denemark 已提交
73 74 75 76 77 78 79
const char *
qemuDomainAsyncJobPhaseToString(enum qemuDomainAsyncJob job,
                                int phase ATTRIBUTE_UNUSED)
{
    switch (job) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
    case QEMU_ASYNC_JOB_MIGRATION_IN:
80 81
        return qemuMigrationJobPhaseTypeToString(phase);

J
Jiri Denemark 已提交
82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101
    case QEMU_ASYNC_JOB_SAVE:
    case QEMU_ASYNC_JOB_DUMP:
    case QEMU_ASYNC_JOB_NONE:
    case QEMU_ASYNC_JOB_LAST:
        ; /* fall through */
    }

    return "none";
}

int
qemuDomainAsyncJobPhaseFromString(enum qemuDomainAsyncJob job,
                                  const char *phase)
{
    if (!phase)
        return 0;

    switch (job) {
    case QEMU_ASYNC_JOB_MIGRATION_OUT:
    case QEMU_ASYNC_JOB_MIGRATION_IN:
102 103
        return qemuMigrationJobPhaseTypeFromString(phase);

J
Jiri Denemark 已提交
104 105 106 107 108 109 110 111 112 113 114 115 116
    case QEMU_ASYNC_JOB_SAVE:
    case QEMU_ASYNC_JOB_DUMP:
    case QEMU_ASYNC_JOB_NONE:
    case QEMU_ASYNC_JOB_LAST:
        ; /* fall through */
    }

    if (STREQ(phase, "none"))
        return 0;
    else
        return -1;
}

117 118 119 120 121

/* driver must be locked before calling */
void qemuDomainEventQueue(struct qemud_driver *driver,
                          virDomainEventPtr event)
{
122
    virDomainEventStateQueue(driver->domainEventState, event);
123 124 125
}


126 127 128 129 130 131 132 133
static int
qemuDomainObjInitJob(qemuDomainObjPrivatePtr priv)
{
    memset(&priv->job, 0, sizeof(priv->job));

    if (virCondInit(&priv->job.cond) < 0)
        return -1;

134 135 136 137 138
    if (virCondInit(&priv->job.asyncCond) < 0) {
        ignore_value(virCondDestroy(&priv->job.cond));
        return -1;
    }

139 140 141 142 143 144 145 146 147
    return 0;
}

static void
qemuDomainObjResetJob(qemuDomainObjPrivatePtr priv)
{
    struct qemuDomainJobObj *job = &priv->job;

    job->active = QEMU_JOB_NONE;
148 149 150 151 152 153 154 155
}

static void
qemuDomainObjResetAsyncJob(qemuDomainObjPrivatePtr priv)
{
    struct qemuDomainJobObj *job = &priv->job;

    job->asyncJob = QEMU_ASYNC_JOB_NONE;
J
Jiri Denemark 已提交
156
    job->phase = 0;
157
    job->mask = DEFAULT_JOB_MASK;
158 159 160 161
    job->start = 0;
    memset(&job->info, 0, sizeof(job->info));
}

162 163 164 165 166 167 168 169 170
void
qemuDomainObjRestoreJob(virDomainObjPtr obj,
                        struct qemuDomainJobObj *job)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    memset(job, 0, sizeof(*job));
    job->active = priv->job.active;
    job->asyncJob = priv->job.asyncJob;
J
Jiri Denemark 已提交
171
    job->phase = priv->job.phase;
172 173 174 175 176

    qemuDomainObjResetJob(priv);
    qemuDomainObjResetAsyncJob(priv);
}

177 178 179 180
static void
qemuDomainObjFreeJob(qemuDomainObjPrivatePtr priv)
{
    ignore_value(virCondDestroy(&priv->job.cond));
181
    ignore_value(virCondDestroy(&priv->job.asyncCond));
182 183 184
}


185 186 187 188 189 190 191
static void *qemuDomainObjPrivateAlloc(void)
{
    qemuDomainObjPrivatePtr priv;

    if (VIR_ALLOC(priv) < 0)
        return NULL;

192
    if (qemuDomainObjInitJob(priv) < 0)
193
        goto error;
194

195 196
    priv->migMaxBandwidth = QEMU_DOMAIN_DEFAULT_MIG_BANDWIDTH_MAX;

197
    return priv;
198 199 200 201

error:
    VIR_FREE(priv);
    return NULL;
202 203 204 205 206 207
}

static void qemuDomainObjPrivateFree(void *data)
{
    qemuDomainObjPrivatePtr priv = data;

208 209
    qemuCapsFree(priv->qemuCaps);

210
    qemuDomainPCIAddressSetFree(priv->pciaddrs);
211
    virDomainChrSourceDefFree(priv->monConfig);
212
    qemuDomainObjFreeJob(priv);
213
    VIR_FREE(priv->vcpupids);
214
    VIR_FREE(priv->lockState);
J
Jiri Denemark 已提交
215
    VIR_FREE(priv->origname);
216 217 218

    /* This should never be non-NULL if we get here, but just in case... */
    if (priv->mon) {
219
        VIR_ERROR(_("Unexpected QEMU monitor still active during domain deletion"));
220 221 222 223 224 225 226 227 228 229 230 231 232
        qemuMonitorClose(priv->mon);
    }
    VIR_FREE(priv);
}


static int qemuDomainObjPrivateXMLFormat(virBufferPtr buf, void *data)
{
    qemuDomainObjPrivatePtr priv = data;
    const char *monitorpath;

    /* priv->monitor_chr is set only for qemu */
    if (priv->monConfig) {
233
        switch (priv->monConfig->type) {
234
        case VIR_DOMAIN_CHR_TYPE_UNIX:
235
            monitorpath = priv->monConfig->data.nix.path;
236 237 238
            break;
        default:
        case VIR_DOMAIN_CHR_TYPE_PTY:
239
            monitorpath = priv->monConfig->data.file.path;
240 241 242 243 244 245
            break;
        }

        virBufferEscapeString(buf, "  <monitor path='%s'", monitorpath);
        if (priv->monJSON)
            virBufferAddLit(buf, " json='1'");
246
        virBufferAsprintf(buf, " type='%s'/>\n",
247
                          virDomainChrTypeToString(priv->monConfig->type));
248 249 250 251 252 253 254
    }


    if (priv->nvcpupids) {
        int i;
        virBufferAddLit(buf, "  <vcpus>\n");
        for (i = 0 ; i < priv->nvcpupids ; i++) {
255
            virBufferAsprintf(buf, "    <vcpu pid='%d'/>\n", priv->vcpupids[i]);
256 257 258 259
        }
        virBufferAddLit(buf, "  </vcpus>\n");
    }

260 261 262 263 264
    if (priv->qemuCaps) {
        int i;
        virBufferAddLit(buf, "  <qemuCaps>\n");
        for (i = 0 ; i < QEMU_CAPS_LAST ; i++) {
            if (qemuCapsGet(priv->qemuCaps, i)) {
265
                virBufferAsprintf(buf, "    <flag name='%s'/>\n",
266 267 268 269 270 271
                                  qemuCapsTypeToString(i));
            }
        }
        virBufferAddLit(buf, "  </qemuCaps>\n");
    }

272 273 274
    if (priv->lockState)
        virBufferAsprintf(buf, "  <lockstate>%s</lockstate>\n", priv->lockState);

275
    if (priv->job.active || priv->job.asyncJob) {
J
Jiri Denemark 已提交
276
        virBufferAsprintf(buf, "  <job type='%s' async='%s'",
277 278
                          qemuDomainJobTypeToString(priv->job.active),
                          qemuDomainAsyncJobTypeToString(priv->job.asyncJob));
J
Jiri Denemark 已提交
279 280 281 282 283 284
        if (priv->job.phase) {
            virBufferAsprintf(buf, " phase='%s'",
                              qemuDomainAsyncJobPhaseToString(
                                    priv->job.asyncJob, priv->job.phase));
        }
        virBufferAddLit(buf, "/>\n");
285 286
    }

287 288 289
    if (priv->fakeReboot)
        virBufferAsprintf(buf, "  <fakereboot/>\n");

290 291 292 293 294 295 296 297 298 299
    return 0;
}

static int qemuDomainObjPrivateXMLParse(xmlXPathContextPtr ctxt, void *data)
{
    qemuDomainObjPrivatePtr priv = data;
    char *monitorpath;
    char *tmp;
    int n, i;
    xmlNodePtr *nodes = NULL;
300
    virBitmapPtr qemuCaps = NULL;
301 302 303 304 305 306 307 308 309 310 311 312 313 314 315

    if (VIR_ALLOC(priv->monConfig) < 0) {
        virReportOOMError();
        goto error;
    }

    if (!(monitorpath =
          virXPathString("string(./monitor[1]/@path)", ctxt))) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("no monitor path"));
        goto error;
    }

    tmp = virXPathString("string(./monitor[1]/@type)", ctxt);
    if (tmp)
316
        priv->monConfig->type = virDomainChrTypeFromString(tmp);
317
    else
318
        priv->monConfig->type = VIR_DOMAIN_CHR_TYPE_PTY;
319 320 321 322 323 324 325 326
    VIR_FREE(tmp);

    if (virXPathBoolean("count(./monitor[@json = '1']) > 0", ctxt)) {
        priv->monJSON = 1;
    } else {
        priv->monJSON = 0;
    }

327
    switch (priv->monConfig->type) {
328
    case VIR_DOMAIN_CHR_TYPE_PTY:
329
        priv->monConfig->data.file.path = monitorpath;
330 331
        break;
    case VIR_DOMAIN_CHR_TYPE_UNIX:
332
        priv->monConfig->data.nix.path = monitorpath;
333 334 335 336 337
        break;
    default:
        VIR_FREE(monitorpath);
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("unsupported monitor type '%s'"),
338
                        virDomainChrTypeToString(priv->monConfig->type));
339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365
        goto error;
    }

    n = virXPathNodeSet("./vcpus/vcpu", ctxt, &nodes);
    if (n < 0)
        goto error;
    if (n) {
        priv->nvcpupids = n;
        if (VIR_REALLOC_N(priv->vcpupids, priv->nvcpupids) < 0) {
            virReportOOMError();
            goto error;
        }

        for (i = 0 ; i < n ; i++) {
            char *pidstr = virXMLPropString(nodes[i], "pid");
            if (!pidstr)
                goto error;

            if (virStrToLong_i(pidstr, NULL, 10, &(priv->vcpupids[i])) < 0) {
                VIR_FREE(pidstr);
                goto error;
            }
            VIR_FREE(pidstr);
        }
        VIR_FREE(nodes);
    }

366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381
    if ((n = virXPathNodeSet("./qemuCaps/flag", ctxt, &nodes)) < 0) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("failed to parse qemu capabilities flags"));
        goto error;
    }
    if (n > 0) {
        if (!(qemuCaps = qemuCapsNew()))
            goto error;

        for (i = 0 ; i < n ; i++) {
            char *str = virXMLPropString(nodes[i], "name");
            if (str) {
                int flag = qemuCapsTypeFromString(str);
                if (flag < 0) {
                    qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                    _("Unknown qemu capabilities flag %s"), str);
382
                    VIR_FREE(str);
383 384
                    goto error;
                }
385
                VIR_FREE(str);
386 387 388 389 390 391 392 393
                qemuCapsSet(qemuCaps, flag);
            }
        }

        priv->qemuCaps = qemuCaps;
    }
    VIR_FREE(nodes);

394
    priv->lockState = virXPathString("string(./lockstate)", ctxt);
395

396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419
    if ((tmp = virXPathString("string(./job[1]/@type)", ctxt))) {
        int type;

        if ((type = qemuDomainJobTypeFromString(tmp)) < 0) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("Unknown job type %s"), tmp);
            VIR_FREE(tmp);
            goto error;
        }
        VIR_FREE(tmp);
        priv->job.active = type;
    }

    if ((tmp = virXPathString("string(./job[1]/@async)", ctxt))) {
        int async;

        if ((async = qemuDomainAsyncJobTypeFromString(tmp)) < 0) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("Unknown async job type %s"), tmp);
            VIR_FREE(tmp);
            goto error;
        }
        VIR_FREE(tmp);
        priv->job.asyncJob = async;
J
Jiri Denemark 已提交
420 421 422 423 424 425 426 427 428 429 430

        if ((tmp = virXPathString("string(./job[1]/@phase)", ctxt))) {
            priv->job.phase = qemuDomainAsyncJobPhaseFromString(async, tmp);
            if (priv->job.phase < 0) {
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("Unknown job phase %s"), tmp);
                VIR_FREE(tmp);
                goto error;
            }
            VIR_FREE(tmp);
        }
431 432
    }

433 434
    priv->fakeReboot = virXPathBoolean("boolean(./fakereboot)", ctxt) == 1;

435 436 437
    return 0;

error:
438
    virDomainChrSourceDefFree(priv->monConfig);
439 440
    priv->monConfig = NULL;
    VIR_FREE(nodes);
441
    qemuCapsFree(qemuCaps);
442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467
    return -1;
}


static void
qemuDomainDefNamespaceFree(void *nsdata)
{
    qemuDomainCmdlineDefPtr cmd = nsdata;
    unsigned int i;

    if (!cmd)
        return;

    for (i = 0; i < cmd->num_args; i++)
        VIR_FREE(cmd->args[i]);
    for (i = 0; i < cmd->num_env; i++) {
        VIR_FREE(cmd->env_name[i]);
        VIR_FREE(cmd->env_value[i]);
    }
    VIR_FREE(cmd->args);
    VIR_FREE(cmd->env_name);
    VIR_FREE(cmd->env_value);
    VIR_FREE(cmd);
}

static int
P
Philipp Hahn 已提交
468 469
qemuDomainDefNamespaceParse(xmlDocPtr xml ATTRIBUTE_UNUSED,
                            xmlNodePtr root ATTRIBUTE_UNUSED,
470 471 472 473
                            xmlXPathContextPtr ctxt,
                            void **data)
{
    qemuDomainCmdlineDefPtr cmd = NULL;
P
Philipp Hahn 已提交
474
    bool uses_qemu_ns = false;
475 476 477
    xmlNodePtr *nodes = NULL;
    int n, i;

P
Philipp Hahn 已提交
478
    if (xmlXPathRegisterNs(ctxt, BAD_CAST "qemu", BAD_CAST QEMU_NAMESPACE_HREF) < 0) {
479
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
P
Philipp Hahn 已提交
480 481
                        _("Failed to register xml namespace '%s'"),
                        QEMU_NAMESPACE_HREF);
482 483 484 485 486 487 488 489 490 491 492 493
        return -1;
    }

    if (VIR_ALLOC(cmd) < 0) {
        virReportOOMError();
        return -1;
    }

    /* first handle the extra command-line arguments */
    n = virXPathNodeSet("./qemu:commandline/qemu:arg", ctxt, &nodes);
    if (n < 0)
        goto error;
P
Philipp Hahn 已提交
494
    uses_qemu_ns |= n > 0;
495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514

    if (n && VIR_ALLOC_N(cmd->args, n) < 0)
        goto no_memory;

    for (i = 0; i < n; i++) {
        cmd->args[cmd->num_args] = virXMLPropString(nodes[i], "value");
        if (cmd->args[cmd->num_args] == NULL) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            "%s", _("No qemu command-line argument specified"));
            goto error;
        }
        cmd->num_args++;
    }

    VIR_FREE(nodes);

    /* now handle the extra environment variables */
    n = virXPathNodeSet("./qemu:commandline/qemu:env", ctxt, &nodes);
    if (n < 0)
        goto error;
P
Philipp Hahn 已提交
515
    uses_qemu_ns |= n > 0;
516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556

    if (n && VIR_ALLOC_N(cmd->env_name, n) < 0)
        goto no_memory;

    if (n && VIR_ALLOC_N(cmd->env_value, n) < 0)
        goto no_memory;

    for (i = 0; i < n; i++) {
        char *tmp;

        tmp = virXMLPropString(nodes[i], "name");
        if (tmp == NULL) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            "%s", _("No qemu environment name specified"));
            goto error;
        }
        if (tmp[0] == '\0') {
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            "%s", _("Empty qemu environment name specified"));
            goto error;
        }
        if (!c_isalpha(tmp[0]) && tmp[0] != '_') {
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            "%s", _("Invalid environment name, it must begin with a letter or underscore"));
            goto error;
        }
        if (strspn(tmp, "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789_") != strlen(tmp)) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            "%s", _("Invalid environment name, it must contain only alphanumerics and underscore"));
            goto error;
        }

        cmd->env_name[cmd->num_env] = tmp;

        cmd->env_value[cmd->num_env] = virXMLPropString(nodes[i], "value");
        /* a NULL value for command is allowed, since it might be empty */
        cmd->num_env++;
    }

    VIR_FREE(nodes);

P
Philipp Hahn 已提交
557 558 559 560
    if (uses_qemu_ns)
        *data = cmd;
    else
        VIR_FREE(cmd);
561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587

    return 0;

no_memory:
    virReportOOMError();

error:
    VIR_FREE(nodes);
    qemuDomainDefNamespaceFree(cmd);
    return -1;
}

static int
qemuDomainDefNamespaceFormatXML(virBufferPtr buf,
                                void *nsdata)
{
    qemuDomainCmdlineDefPtr cmd = nsdata;
    unsigned int i;

    if (!cmd->num_args && !cmd->num_env)
        return 0;

    virBufferAddLit(buf, "  <qemu:commandline>\n");
    for (i = 0; i < cmd->num_args; i++)
        virBufferEscapeString(buf, "    <qemu:arg value='%s'/>\n",
                              cmd->args[i]);
    for (i = 0; i < cmd->num_env; i++) {
588
        virBufferAsprintf(buf, "    <qemu:env name='%s'", cmd->env_name[i]);
589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622
        if (cmd->env_value[i])
            virBufferEscapeString(buf, " value='%s'", cmd->env_value[i]);
        virBufferAddLit(buf, "/>\n");
    }
    virBufferAddLit(buf, "  </qemu:commandline>\n");

    return 0;
}

static const char *
qemuDomainDefNamespaceHref(void)
{
    return "xmlns:qemu='" QEMU_NAMESPACE_HREF "'";
}


void qemuDomainSetPrivateDataHooks(virCapsPtr caps)
{
    /* Domain XML parser hooks */
    caps->privateDataAllocFunc = qemuDomainObjPrivateAlloc;
    caps->privateDataFreeFunc = qemuDomainObjPrivateFree;
    caps->privateDataXMLFormat = qemuDomainObjPrivateXMLFormat;
    caps->privateDataXMLParse = qemuDomainObjPrivateXMLParse;

}

void qemuDomainSetNamespaceHooks(virCapsPtr caps)
{
    /* Domain Namespace XML parser hooks */
    caps->ns.parse = qemuDomainDefNamespaceParse;
    caps->ns.free = qemuDomainDefNamespaceFree;
    caps->ns.format = qemuDomainDefNamespaceFormatXML;
    caps->ns.href = qemuDomainDefNamespaceHref;
}
623

624
static void
625
qemuDomainObjSaveJob(struct qemud_driver *driver, virDomainObjPtr obj)
626
{
627 628 629 630 631
    if (!virDomainObjIsActive(obj)) {
        /* don't write the state file yet, it will be written once the domain
         * gets activated */
        return;
    }
632

633 634
    if (virDomainSaveStatus(driver->caps, driver->stateDir, obj) < 0)
        VIR_WARN("Failed to save status on vm %s", obj->def->name);
635 636
}

J
Jiri Denemark 已提交
637 638 639 640 641 642 643 644 645 646 647 648 649 650
void
qemuDomainObjSetJobPhase(struct qemud_driver *driver,
                         virDomainObjPtr obj,
                         int phase)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    if (!priv->job.asyncJob)
        return;

    priv->job.phase = phase;
    qemuDomainObjSaveJob(driver, obj);
}

651
void
652 653
qemuDomainObjSetAsyncJobMask(virDomainObjPtr obj,
                             unsigned long long allowedJobs)
654 655 656
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

657 658 659 660 661 662 663
    if (!priv->job.asyncJob)
        return;

    priv->job.mask = allowedJobs | JOB_MASK(QEMU_JOB_DESTROY);
}

void
664
qemuDomainObjDiscardAsyncJob(struct qemud_driver *driver, virDomainObjPtr obj)
665 666 667 668 669 670
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    if (priv->job.active == QEMU_JOB_ASYNC_NESTED)
        qemuDomainObjResetJob(priv);
    qemuDomainObjResetAsyncJob(priv);
671
    qemuDomainObjSaveJob(driver, obj);
672 673 674
}

static bool
675
qemuDomainNestedJobAllowed(qemuDomainObjPrivatePtr priv, enum qemuDomainJob job)
676 677
{
    return !priv->job.asyncJob || (priv->job.mask & JOB_MASK(job)) != 0;
678 679
}

680 681 682 683 684 685
bool
qemuDomainJobAllowed(qemuDomainObjPrivatePtr priv, enum qemuDomainJob job)
{
    return !priv->job.active && qemuDomainNestedJobAllowed(priv, job);
}

686 687 688
/* Give up waiting for mutex after 30 seconds */
#define QEMU_JOB_WAIT_TIME (1000ull * 30)

689 690 691 692
/*
 * obj must be locked before calling; driver_locked says if qemu_driver is
 * locked or not.
 */
693
static int ATTRIBUTE_NONNULL(1)
694 695
qemuDomainObjBeginJobInternal(struct qemud_driver *driver,
                              bool driver_locked,
696 697 698
                              virDomainObjPtr obj,
                              enum qemuDomainJob job,
                              enum qemuDomainAsyncJob asyncJob)
699 700
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
J
Jiri Denemark 已提交
701
    unsigned long long now;
702
    unsigned long long then;
703
    bool nested = job == QEMU_JOB_ASYNC_NESTED;
704

705 706
    priv->jobs_queued++;

707
    if (virTimeMillisNow(&now) < 0)
708
        return -1;
J
Jiri Denemark 已提交
709
    then = now + QEMU_JOB_WAIT_TIME;
710 711

    virDomainObjRef(obj);
712 713
    if (driver_locked)
        qemuDriverUnlock(driver);
714

715
retry:
716 717 718 719 720
    if (driver->max_queued &&
        priv->jobs_queued > driver->max_queued) {
        goto error;
    }

721
    while (!nested && !qemuDomainNestedJobAllowed(priv, job)) {
722 723 724 725
        if (virCondWaitUntil(&priv->job.asyncCond, &obj->lock, then) < 0)
            goto error;
    }

726
    while (priv->job.active) {
727 728
        if (virCondWaitUntil(&priv->job.cond, &obj->lock, then) < 0)
            goto error;
729
    }
730 731 732

    /* No job is active but a new async job could have been started while obj
     * was unlocked, so we need to recheck it. */
733
    if (!nested && !qemuDomainNestedJobAllowed(priv, job))
734 735
        goto retry;

736
    qemuDomainObjResetJob(priv);
737 738

    if (job != QEMU_JOB_ASYNC) {
739 740 741
        VIR_DEBUG("Starting job: %s (async=%s)",
                   qemuDomainJobTypeToString(job),
                   qemuDomainAsyncJobTypeToString(priv->job.asyncJob));
742 743
        priv->job.active = job;
    } else {
744 745
        VIR_DEBUG("Starting async job: %s",
                  qemuDomainAsyncJobTypeToString(asyncJob));
746 747 748 749
        qemuDomainObjResetAsyncJob(priv);
        priv->job.asyncJob = asyncJob;
        priv->job.start = now;
    }
750

751 752 753 754 755 756
    if (driver_locked) {
        virDomainObjUnlock(obj);
        qemuDriverLock(driver);
        virDomainObjLock(obj);
    }

757 758
    qemuDomainObjSaveJob(driver, obj);

759
    return 0;
760 761 762 763 764

error:
    if (errno == ETIMEDOUT)
        qemuReportError(VIR_ERR_OPERATION_TIMEOUT,
                        "%s", _("cannot acquire state change lock"));
765 766 767 768 769
    else if (driver->max_queued &&
             priv->jobs_queued > driver->max_queued)
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("cannot acquire state change lock "
                                "due to max_queued limit"));
770 771 772
    else
        virReportSystemError(errno,
                             "%s", _("cannot acquire job mutex"));
773
    priv->jobs_queued--;
774 775 776 777 778 779 780 781
    if (driver_locked) {
        virDomainObjUnlock(obj);
        qemuDriverLock(driver);
        virDomainObjLock(obj);
    }
    /* Safe to ignore value since ref count was incremented above */
    ignore_value(virDomainObjUnref(obj));
    return -1;
782 783 784
}

/*
785 786 787 788 789 790 791 792
 * obj must be locked before calling, qemud_driver must NOT be locked
 *
 * This must be called by anything that will change the VM state
 * in any way, or anything that will use the QEMU monitor.
 *
 * Upon successful return, the object will have its ref count increased,
 * successful calls must be followed by EndJob eventually
 */
793 794 795
int qemuDomainObjBeginJob(struct qemud_driver *driver,
                          virDomainObjPtr obj,
                          enum qemuDomainJob job)
796
{
797
    return qemuDomainObjBeginJobInternal(driver, false, obj, job,
798 799 800
                                         QEMU_ASYNC_JOB_NONE);
}

801 802
int qemuDomainObjBeginAsyncJob(struct qemud_driver *driver,
                               virDomainObjPtr obj,
803
                               enum qemuDomainAsyncJob asyncJob)
804
{
805
    return qemuDomainObjBeginJobInternal(driver, false, obj, QEMU_JOB_ASYNC,
806
                                         asyncJob);
807 808 809 810 811
}

/*
 * obj must be locked before calling. If qemud_driver is passed, it MUST be
 * locked; otherwise it MUST NOT be locked.
812 813 814
 *
 * This must be called by anything that will change the VM state
 * in any way, or anything that will use the QEMU monitor.
815 816 817
 *
 * Upon successful return, the object will have its ref count increased,
 * successful calls must be followed by EndJob eventually
818 819
 */
int qemuDomainObjBeginJobWithDriver(struct qemud_driver *driver,
820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840
                                    virDomainObjPtr obj,
                                    enum qemuDomainJob job)
{
    if (job <= QEMU_JOB_NONE || job >= QEMU_JOB_ASYNC) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("Attempt to start invalid job"));
        return -1;
    }

    return qemuDomainObjBeginJobInternal(driver, true, obj, job,
                                         QEMU_ASYNC_JOB_NONE);
}

int qemuDomainObjBeginAsyncJobWithDriver(struct qemud_driver *driver,
                                         virDomainObjPtr obj,
                                         enum qemuDomainAsyncJob asyncJob)
{
    return qemuDomainObjBeginJobInternal(driver, true, obj, QEMU_JOB_ASYNC,
                                         asyncJob);
}

841 842 843 844 845 846 847 848 849
/*
 * obj must be locked before calling, qemud_driver does not matter
 *
 * To be called after completing the work associated with the
 * earlier qemuDomainBeginJob() call
 *
 * Returns remaining refcount on 'obj', maybe 0 to indicated it
 * was deleted
 */
850
int qemuDomainObjEndJob(struct qemud_driver *driver, virDomainObjPtr obj)
851 852 853
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

854 855
    priv->jobs_queued--;

856 857 858 859
    VIR_DEBUG("Stopping job: %s (async=%s)",
              qemuDomainJobTypeToString(priv->job.active),
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob));

860
    qemuDomainObjResetJob(priv);
861
    qemuDomainObjSaveJob(driver, obj);
862
    virCondSignal(&priv->job.cond);
863 864 865 866

    return virDomainObjUnref(obj);
}

867
int
868
qemuDomainObjEndAsyncJob(struct qemud_driver *driver, virDomainObjPtr obj)
869 870
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
871

872 873
    priv->jobs_queued--;

874 875 876
    VIR_DEBUG("Stopping async job: %s",
              qemuDomainAsyncJobTypeToString(priv->job.asyncJob));

877
    qemuDomainObjResetAsyncJob(priv);
878
    qemuDomainObjSaveJob(driver, obj);
879 880 881 882 883
    virCondBroadcast(&priv->job.asyncCond);

    return virDomainObjUnref(obj);
}

884
static int
885
qemuDomainObjEnterMonitorInternal(struct qemud_driver *driver,
886
                                  bool driver_locked,
887 888
                                  virDomainObjPtr obj,
                                  enum qemuDomainAsyncJob asyncJob)
889 890 891
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

892 893 894 895 896 897
    if (asyncJob != QEMU_ASYNC_JOB_NONE) {
        if (asyncJob != priv->job.asyncJob) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("unepxected async job %d"), asyncJob);
            return -1;
        }
898 899 900
        if (qemuDomainObjBeginJobInternal(driver, driver_locked, obj,
                                          QEMU_JOB_ASYNC_NESTED,
                                          QEMU_ASYNC_JOB_NONE) < 0)
901 902 903 904
            return -1;
        if (!virDomainObjIsActive(obj)) {
            qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
                            _("domain is no longer running"));
905 906
            /* Still referenced by the containing async job.  */
            ignore_value(qemuDomainObjEndJob(driver, obj));
907 908 909 910
            return -1;
        }
    }

911 912
    qemuMonitorLock(priv->mon);
    qemuMonitorRef(priv->mon);
913
    ignore_value(virTimeMillisNow(&priv->monStart));
914
    virDomainObjUnlock(obj);
915
    if (driver_locked)
916
        qemuDriverUnlock(driver);
917 918

    return 0;
919 920
}

921
static void ATTRIBUTE_NONNULL(1)
922
qemuDomainObjExitMonitorInternal(struct qemud_driver *driver,
923
                                 bool driver_locked,
924
                                 virDomainObjPtr obj)
925 926 927 928 929 930 931 932 933
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
    int refs;

    refs = qemuMonitorUnref(priv->mon);

    if (refs > 0)
        qemuMonitorUnlock(priv->mon);

934
    if (driver_locked)
935
        qemuDriverLock(driver);
936 937
    virDomainObjLock(obj);

938
    priv->monStart = 0;
939 940 941
    if (refs == 0) {
        priv->mon = NULL;
    }
942

943 944 945 946 947 948 949 950 951
    if (priv->job.active == QEMU_JOB_ASYNC_NESTED) {
        qemuDomainObjResetJob(priv);
        qemuDomainObjSaveJob(driver, obj);
        virCondSignal(&priv->job.cond);

        /* safe to ignore since the surrounding async job increased
         * the reference counter as well */
        ignore_value(virDomainObjUnref(obj));
    }
952 953
}

954 955 956 957
/*
 * obj must be locked before calling, qemud_driver must be unlocked
 *
 * To be called immediately before any QEMU monitor API call
958
 * Must have already either called qemuDomainObjBeginJob() and checked
959
 * that the VM is still active; may not be used for nested async jobs.
960 961 962
 *
 * To be followed with qemuDomainObjExitMonitor() once complete
 */
963 964
void qemuDomainObjEnterMonitor(struct qemud_driver *driver,
                               virDomainObjPtr obj)
965
{
966 967
    ignore_value(qemuDomainObjEnterMonitorInternal(driver, false, obj,
                                                   QEMU_ASYNC_JOB_NONE));
968 969 970 971 972 973
}

/* obj must NOT be locked before calling, qemud_driver must be unlocked
 *
 * Should be paired with an earlier qemuDomainObjEnterMonitor() call
 */
974 975
void qemuDomainObjExitMonitor(struct qemud_driver *driver,
                              virDomainObjPtr obj)
976
{
977
    qemuDomainObjExitMonitorInternal(driver, false, obj);
978
}
979 980 981 982 983

/*
 * obj must be locked before calling, qemud_driver must be locked
 *
 * To be called immediately before any QEMU monitor API call
984
 * Must have already either called qemuDomainObjBeginJobWithDriver() and
985
 * checked that the VM is still active; may not be used for nested async jobs.
986 987 988
 *
 * To be followed with qemuDomainObjExitMonitorWithDriver() once complete
 */
989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012
void qemuDomainObjEnterMonitorWithDriver(struct qemud_driver *driver,
                                         virDomainObjPtr obj)
{
    ignore_value(qemuDomainObjEnterMonitorInternal(driver, true, obj,
                                                   QEMU_ASYNC_JOB_NONE));
}

/*
 * obj and qemud_driver must be locked before calling
 *
 * To be called immediately before any QEMU monitor API call.
 * Must have already either called qemuDomainObjBeginJobWithDriver()
 * and checked that the VM is still active, with asyncJob of
 * QEMU_ASYNC_JOB_NONE; or already called qemuDomainObjBeginAsyncJob,
 * with the same asyncJob.
 *
 * Returns 0 if job was started, in which case this must be followed with
 * qemuDomainObjExitMonitorWithDriver(); or -1 if the job could not be
 * started (probably because the vm exited in the meantime).
 */
int
qemuDomainObjEnterMonitorAsync(struct qemud_driver *driver,
                               virDomainObjPtr obj,
                               enum qemuDomainAsyncJob asyncJob)
1013
{
1014
    return qemuDomainObjEnterMonitorInternal(driver, true, obj, asyncJob);
1015 1016 1017 1018 1019 1020 1021 1022 1023 1024
}

/* obj must NOT be locked before calling, qemud_driver must be unlocked,
 * and will be locked after returning
 *
 * Should be paired with an earlier qemuDomainObjEnterMonitorWithDriver() call
 */
void qemuDomainObjExitMonitorWithDriver(struct qemud_driver *driver,
                                        virDomainObjPtr obj)
{
1025
    qemuDomainObjExitMonitorInternal(driver, true, obj);
1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040
}

void qemuDomainObjEnterRemoteWithDriver(struct qemud_driver *driver,
                                        virDomainObjPtr obj)
{
    virDomainObjRef(obj);
    virDomainObjUnlock(obj);
    qemuDriverUnlock(driver);
}

void qemuDomainObjExitRemoteWithDriver(struct qemud_driver *driver,
                                       virDomainObjPtr obj)
{
    qemuDriverLock(driver);
    virDomainObjLock(obj);
1041 1042 1043
    /* Safe to ignore value, since we incremented ref in
     * qemuDomainObjEnterRemoteWithDriver */
    ignore_value(virDomainObjUnref(obj));
1044
}
1045 1046


1047 1048
char *qemuDomainDefFormatXML(struct qemud_driver *driver,
                             virDomainDefPtr def,
E
Eric Blake 已提交
1049
                             unsigned int flags)
1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077
{
    char *ret = NULL;
    virCPUDefPtr cpu = NULL;
    virCPUDefPtr def_cpu;

    def_cpu = def->cpu;

    /* Update guest CPU requirements according to host CPU */
    if ((flags & VIR_DOMAIN_XML_UPDATE_CPU) && def_cpu && def_cpu->model) {
        if (!driver->caps || !driver->caps->host.cpu) {
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            "%s", _("cannot get host CPU capabilities"));
            goto cleanup;
        }

        if (!(cpu = virCPUDefCopy(def_cpu))
            || cpuUpdate(cpu, driver->caps->host.cpu))
            goto cleanup;
        def->cpu = cpu;
    }

    ret = virDomainDefFormat(def, flags);

cleanup:
    def->cpu = def_cpu;
    virCPUDefFree(cpu);
    return ret;
}
1078

1079 1080
char *qemuDomainFormatXML(struct qemud_driver *driver,
                          virDomainObjPtr vm,
E
Eric Blake 已提交
1081
                          unsigned int flags)
1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093
{
    virDomainDefPtr def;

    if ((flags & VIR_DOMAIN_XML_INACTIVE) && vm->newDef)
        def = vm->newDef;
    else
        def = vm->def;

    return qemuDomainDefFormatXML(driver, def, flags);
}


1094
void qemuDomainObjTaint(struct qemud_driver *driver,
1095
                        virDomainObjPtr obj,
1096 1097
                        enum virDomainTaintFlags taint,
                        int logFD)
1098
{
1099 1100
    virErrorPtr orig_err = NULL;

1101 1102 1103 1104 1105 1106 1107 1108 1109
    if (virDomainObjTaint(obj, taint)) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(obj->def->uuid, uuidstr);

        VIR_WARN("Domain id=%d name='%s' uuid=%s is tainted: %s",
                 obj->def->id,
                 obj->def->name,
                 uuidstr,
                 virDomainTaintTypeToString(taint));
1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123

        /* We don't care about errors logging taint info, so
         * preserve original error, and clear any error that
         * is raised */
        orig_err = virSaveLastError();
        if (qemuDomainAppendLog(driver, obj, logFD,
                                "Domain id=%d is tainted: %s\n",
                                obj->def->id,
                                virDomainTaintTypeToString(taint)) < 0)
            virResetLastError();
        if (orig_err) {
            virSetError(orig_err);
            virFreeError(orig_err);
        }
1124 1125 1126 1127 1128
    }
}


void qemuDomainObjCheckTaint(struct qemud_driver *driver,
1129 1130
                             virDomainObjPtr obj,
                             int logFD)
1131 1132 1133
{
    int i;

1134 1135 1136 1137
    if (driver->privileged &&
        (!driver->clearEmulatorCapabilities ||
         driver->user == 0 ||
         driver->group == 0))
1138
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_HIGH_PRIVILEGES, logFD);
1139 1140 1141 1142

    if (obj->def->namespaceData) {
        qemuDomainCmdlineDefPtr qemucmd = obj->def->namespaceData;
        if (qemucmd->num_args || qemucmd->num_env)
1143
            qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_CUSTOM_ARGV, logFD);
1144 1145 1146
    }

    for (i = 0 ; i < obj->def->ndisks ; i++)
1147
        qemuDomainObjCheckDiskTaint(driver, obj, obj->def->disks[i], logFD);
1148 1149

    for (i = 0 ; i < obj->def->nnets ; i++)
1150
        qemuDomainObjCheckNetTaint(driver, obj, obj->def->nets[i], logFD);
1151 1152 1153 1154 1155
}


void qemuDomainObjCheckDiskTaint(struct qemud_driver *driver,
                                 virDomainObjPtr obj,
1156 1157
                                 virDomainDiskDefPtr disk,
                                 int logFD)
1158 1159 1160
{
    if (!disk->driverType &&
        driver->allowDiskFormatProbing)
1161
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_DISK_PROBING, logFD);
1162 1163 1164 1165 1166
}


void qemuDomainObjCheckNetTaint(struct qemud_driver *driver,
                                virDomainObjPtr obj,
1167 1168
                                virDomainNetDefPtr net,
                                int logFD)
1169 1170 1171 1172 1173
{
    if ((net->type == VIR_DOMAIN_NET_TYPE_ETHERNET &&
         net->data.ethernet.script != NULL) ||
        (net->type == VIR_DOMAIN_NET_TYPE_BRIDGE &&
         net->data.bridge.script != NULL))
1174
        qemuDomainObjTaint(driver, obj, VIR_DOMAIN_TAINT_SHELL_SCRIPTS, logFD);
1175
}
1176 1177 1178 1179 1180


static int
qemuDomainOpenLogHelper(struct qemud_driver *driver,
                        virDomainObjPtr vm,
E
Eric Blake 已提交
1181
                        int oflags,
1182 1183 1184 1185 1186 1187 1188 1189 1190 1191
                        mode_t mode)
{
    char *logfile;
    int fd = -1;

    if (virAsprintf(&logfile, "%s/%s.log", driver->logDir, vm->def->name) < 0) {
        virReportOOMError();
        return -1;
    }

E
Eric Blake 已提交
1192
    if ((fd = open(logfile, oflags, mode)) < 0) {
1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210
        virReportSystemError(errno, _("failed to create logfile %s"),
                             logfile);
        goto cleanup;
    }
    if (virSetCloseExec(fd) < 0) {
        virReportSystemError(errno, _("failed to set close-on-exec flag on %s"),
                             logfile);
        VIR_FORCE_CLOSE(fd);
        goto cleanup;
    }

cleanup:
    VIR_FREE(logfile);
    return fd;
}


int
E
Eric Blake 已提交
1211 1212
qemuDomainCreateLog(struct qemud_driver *driver, virDomainObjPtr vm,
                    bool append)
1213
{
E
Eric Blake 已提交
1214
    int oflags;
1215

E
Eric Blake 已提交
1216
    oflags = O_CREAT | O_WRONLY;
1217 1218
    /* Only logrotate files in /var/log, so only append if running privileged */
    if (driver->privileged || append)
E
Eric Blake 已提交
1219
        oflags |= O_APPEND;
1220
    else
E
Eric Blake 已提交
1221
        oflags |= O_TRUNC;
1222

E
Eric Blake 已提交
1223
    return qemuDomainOpenLogHelper(driver, vm, oflags, S_IRUSR | S_IWUSR);
1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260
}


int
qemuDomainOpenLog(struct qemud_driver *driver, virDomainObjPtr vm, off_t pos)
{
    int fd;
    off_t off;
    int whence;

    if ((fd = qemuDomainOpenLogHelper(driver, vm, O_RDONLY, 0)) < 0)
        return -1;

    if (pos < 0) {
        off = 0;
        whence = SEEK_END;
    } else {
        off = pos;
        whence = SEEK_SET;
    }

    if (lseek(fd, off, whence) < 0) {
        if (whence == SEEK_END)
            virReportSystemError(errno,
                                 _("unable to seek to end of log for %s"),
                                 vm->def->name);
        else
            virReportSystemError(errno,
                                 _("unable to seek to %lld from start for %s"),
                                 (long long)off, vm->def->name);
        VIR_FORCE_CLOSE(fd);
    }

    return fd;
}


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
int qemuDomainAppendLog(struct qemud_driver *driver,
                        virDomainObjPtr obj,
                        int logFD,
                        const char *fmt, ...)
{
    int fd = logFD;
    va_list argptr;
    char *message = NULL;
    int ret = -1;

    va_start(argptr, fmt);

    if ((fd == -1) &&
        (fd = qemuDomainCreateLog(driver, obj, true)) < 0)
        goto cleanup;

    if (virVasprintf(&message, fmt, argptr) < 0) {
        virReportOOMError();
        goto cleanup;
    }
    if (safewrite(fd, message, strlen(message)) < 0) {
        virReportSystemError(errno, _("Unable to write to domain logfile %s"),
                             obj->def->name);
        goto cleanup;
    }

    ret = 0;

cleanup:
    va_end(argptr);

    if (fd != logFD)
        VIR_FORCE_CLOSE(fd);

O
Osier Yang 已提交
1295
    VIR_FREE(message);
1296 1297
    return ret;
}
1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354

/* Locate an appropriate 'qemu-img' binary.  */
const char *
qemuFindQemuImgBinary(struct qemud_driver *driver)
{
    if (!driver->qemuImgBinary) {
        driver->qemuImgBinary = virFindFileInPath("kvm-img");
        if (!driver->qemuImgBinary)
            driver->qemuImgBinary = virFindFileInPath("qemu-img");
        if (!driver->qemuImgBinary)
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            "%s", _("unable to find kvm-img or qemu-img"));
    }

    return driver->qemuImgBinary;
}

int
qemuDomainSnapshotWriteMetadata(virDomainObjPtr vm,
                                virDomainSnapshotObjPtr snapshot,
                                char *snapshotDir)
{
    char *newxml = NULL;
    int ret = -1;
    char *snapDir = NULL;
    char *snapFile = NULL;
    char uuidstr[VIR_UUID_STRING_BUFLEN];
    char *tmp;

    virUUIDFormat(vm->def->uuid, uuidstr);
    newxml = virDomainSnapshotDefFormat(uuidstr, snapshot->def,
                                        VIR_DOMAIN_XML_SECURE, 1);
    if (newxml == NULL) {
        virReportOOMError();
        return -1;
    }

    if (virAsprintf(&snapDir, "%s/%s", snapshotDir, vm->def->name) < 0) {
        virReportOOMError();
        goto cleanup;
    }
    if (virFileMakePath(snapDir) < 0) {
        virReportSystemError(errno, _("cannot create snapshot directory '%s'"),
                             snapDir);
        goto cleanup;
    }

    if (virAsprintf(&snapFile, "%s/%s.xml", snapDir, snapshot->def->name) < 0) {
        virReportOOMError();
        goto cleanup;
    }

    if (virAsprintf(&tmp, "snapshot-edit %s", vm->def->name) < 0) {
        virReportOOMError();
        goto cleanup;
    }

1355 1356
    ret = virXMLSaveFile(snapFile, snapshot->def->name, tmp, newxml);
    VIR_FREE(tmp);
1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376

cleanup:
    VIR_FREE(snapFile);
    VIR_FREE(snapDir);
    VIR_FREE(newxml);
    return ret;
}

/* The domain is expected to be locked and inactive. Return -1 on normal
 * failure, 1 if we skipped a disk due to try_all.  */
int
qemuDomainSnapshotForEachQcow2(struct qemud_driver *driver,
                               virDomainObjPtr vm,
                               virDomainSnapshotObjPtr snap,
                               const char *op,
                               bool try_all)
{
    const char *qemuimgarg[] = { NULL, "snapshot", NULL, NULL, NULL, NULL };
    int i;
    bool skipped = false;
1377 1378 1379 1380 1381 1382 1383 1384
    virDomainDefPtr def;

    /* Prefer action on the disks in use at the time the snapshot was
     * created; but fall back to current definition if dealing with a
     * snapshot created prior to libvirt 0.9.5.  */
    def = snap->def->dom;
    if (!def)
        def = vm->def;
1385 1386 1387 1388 1389 1390 1391 1392 1393 1394

    qemuimgarg[0] = qemuFindQemuImgBinary(driver);
    if (qemuimgarg[0] == NULL) {
        /* qemuFindQemuImgBinary set the error */
        return -1;
    }

    qemuimgarg[2] = op;
    qemuimgarg[3] = snap->def->name;

1395
    for (i = 0; i < def->ndisks; i++) {
1396 1397 1398 1399
        /* FIXME: we also need to handle LVM here */
        /* FIXME: if we fail halfway through this loop, we are in an
         * inconsistent state.  I'm not quite sure what to do about that
         */
1400 1401 1402
        if (def->disks[i]->device == VIR_DOMAIN_DISK_DEVICE_DISK) {
            if (!def->disks[i]->driverType ||
                STRNEQ(def->disks[i]->driverType, "qcow2")) {
1403 1404 1405 1406 1407
                if (try_all) {
                    /* Continue on even in the face of error, since other
                     * disks in this VM may have the same snapshot name.
                     */
                    VIR_WARN("skipping snapshot action on %s",
1408
                             def->disks[i]->dst);
1409 1410 1411 1412 1413 1414
                    skipped = true;
                    continue;
                }
                qemuReportError(VIR_ERR_OPERATION_INVALID,
                                _("Disk device '%s' does not support"
                                  " snapshotting"),
1415
                                def->disks[i]->dst);
1416 1417 1418
                return -1;
            }

1419
            qemuimgarg[4] = def->disks[i]->src;
1420 1421 1422 1423

            if (virRun(qemuimgarg, NULL) < 0) {
                if (try_all) {
                    VIR_WARN("skipping snapshot action on %s",
1424
                             def->disks[i]->dst);
1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542
                    skipped = true;
                    continue;
                }
                return -1;
            }
        }
    }

    return skipped ? 1 : 0;
}

/* Discard one snapshot (or its metadata), without reparenting any children.  */
int
qemuDomainSnapshotDiscard(struct qemud_driver *driver,
                          virDomainObjPtr vm,
                          virDomainSnapshotObjPtr snap,
                          bool update_current,
                          bool metadata_only)
{
    char *snapFile = NULL;
    int ret = -1;
    qemuDomainObjPrivatePtr priv;
    virDomainSnapshotObjPtr parentsnap = NULL;

    if (!metadata_only) {
        if (!virDomainObjIsActive(vm)) {
            /* Ignore any skipped disks */
            if (qemuDomainSnapshotForEachQcow2(driver, vm, snap, "-d",
                                               true) < 0)
                goto cleanup;
        } else {
            priv = vm->privateData;
            qemuDomainObjEnterMonitorWithDriver(driver, vm);
            /* we continue on even in the face of error */
            qemuMonitorDeleteSnapshot(priv->mon, snap->def->name);
            qemuDomainObjExitMonitorWithDriver(driver, vm);
        }
    }

    if (virAsprintf(&snapFile, "%s/%s/%s.xml", driver->snapshotDir,
                    vm->def->name, snap->def->name) < 0) {
        virReportOOMError();
        goto cleanup;
    }

    if (snap == vm->current_snapshot) {
        if (update_current && snap->def->parent) {
            parentsnap = virDomainSnapshotFindByName(&vm->snapshots,
                                                     snap->def->parent);
            if (!parentsnap) {
                VIR_WARN("missing parent snapshot matching name '%s'",
                         snap->def->parent);
            } else {
                parentsnap->def->current = true;
                if (qemuDomainSnapshotWriteMetadata(vm, parentsnap,
                                                    driver->snapshotDir) < 0) {
                    VIR_WARN("failed to set parent snapshot '%s' as current",
                             snap->def->parent);
                    parentsnap->def->current = false;
                    parentsnap = NULL;
                }
            }
        }
        vm->current_snapshot = parentsnap;
    }

    if (unlink(snapFile) < 0)
        VIR_WARN("Failed to unlink %s", snapFile);
    virDomainSnapshotObjListRemove(&vm->snapshots, snap);

    ret = 0;

cleanup:
    VIR_FREE(snapFile);

    return ret;
}

/* Hash iterator callback to discard multiple snapshots.  */
void qemuDomainSnapshotDiscardAll(void *payload,
                                  const void *name ATTRIBUTE_UNUSED,
                                  void *data)
{
    virDomainSnapshotObjPtr snap = payload;
    struct qemu_snap_remove *curr = data;
    int err;

    if (snap->def->current)
        curr->current = true;
    err = qemuDomainSnapshotDiscard(curr->driver, curr->vm, snap, false,
                                    curr->metadata_only);
    if (err && !curr->err)
        curr->err = err;
}

int
qemuDomainSnapshotDiscardAllMetadata(struct qemud_driver *driver,
                                     virDomainObjPtr vm)
{
    struct qemu_snap_remove rem;

    rem.driver = driver;
    rem.vm = vm;
    rem.metadata_only = true;
    rem.err = 0;
    virHashForEach(vm->snapshots.objs, qemuDomainSnapshotDiscardAll, &rem);

    return rem.err;
}

/*
 * The caller must hold a lock on both driver and vm, and there must
 * be no remaining references to vm.
 */
void
qemuDomainRemoveInactive(struct qemud_driver *driver,
                         virDomainObjPtr vm)
{
1543 1544
    char *snapDir;

1545 1546 1547 1548 1549
    /* Remove any snapshot metadata prior to removing the domain */
    if (qemuDomainSnapshotDiscardAllMetadata(driver, vm) < 0) {
        VIR_WARN("unable to remove all snapshots for domain %s",
                 vm->def->name);
    }
1550 1551 1552 1553 1554 1555 1556 1557 1558
    else if (virAsprintf(&snapDir, "%s/%s", driver->snapshotDir,
                         vm->def->name) < 0) {
        VIR_WARN("unable to remove snapshot directory %s/%s",
                 driver->snapshotDir, vm->def->name);
    } else {
        if (rmdir(snapDir) < 0 && errno != ENOENT)
            VIR_WARN("unable to remove snapshot directory %s", snapDir);
        VIR_FREE(snapDir);
    }
1559 1560
    virDomainRemoveInactive(&driver->domains, vm);
}
1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576

void
qemuDomainSetFakeReboot(struct qemud_driver *driver,
                        virDomainObjPtr vm,
                        bool value)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;

    if (priv->fakeReboot == value)
        return;

    priv->fakeReboot = value;

    if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
        VIR_WARN("Failed to save status on vm %s", vm->def->name);
}
M
Michal Privoznik 已提交
1577 1578 1579 1580 1581 1582 1583 1584 1585

int
qemuDomainCheckDiskPresence(struct qemud_driver *driver,
                            virDomainObjPtr vm,
                            bool start_with_state)
{
    int ret = -1;
    int i;
    virDomainDiskDefPtr disk;
M
Michal Privoznik 已提交
1586
    char uuid[VIR_UUID_STRING_BUFLEN];
1587
    virDomainEventPtr event = NULL;
M
Michal Privoznik 已提交
1588 1589 1590 1591 1592 1593 1594 1595 1596

    virUUIDFormat(vm->def->uuid, uuid);

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

        if (!disk->startupPolicy || !disk->src)
            continue;

M
Michal Privoznik 已提交
1597 1598 1599 1600
        if (virFileAccessibleAs(disk->src, F_OK,
                                driver->user,
                                driver->group) >= 0) {
            /* disk accessible */
M
Michal Privoznik 已提交
1601 1602 1603 1604 1605 1606 1607 1608
            continue;
        }

        switch ((enum virDomainStartupPolicy) disk->startupPolicy) {
            case VIR_DOMAIN_STARTUP_POLICY_OPTIONAL:
                break;

            case VIR_DOMAIN_STARTUP_POLICY_MANDATORY:
M
Michal Privoznik 已提交
1609
                virReportSystemError(errno,
M
Michal Privoznik 已提交
1610 1611 1612 1613 1614 1615 1616
                                     _("cannot access file '%s'"),
                                     disk->src);
                goto cleanup;
                break;

            case VIR_DOMAIN_STARTUP_POLICY_REQUISITE:
                if (!start_with_state) {
M
Michal Privoznik 已提交
1617
                    virReportSystemError(errno,
M
Michal Privoznik 已提交
1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629
                                         _("cannot access file '%s'"),
                                         disk->src);
                    goto cleanup;
                }
                break;

            case VIR_DOMAIN_STARTUP_POLICY_DEFAULT:
            case VIR_DOMAIN_STARTUP_POLICY_LAST:
                /* this should never happen */
                break;
        }

M
Michal Privoznik 已提交
1630 1631
        VIR_DEBUG("Dropping disk '%s' on domain '%s' (UUID '%s') "
                  "due to inaccessible source '%s'",
M
Michal Privoznik 已提交
1632 1633
                  disk->dst, vm->def->name, uuid, disk->src);

1634
        event = virDomainEventDiskChangeNewFromObj(vm, disk->src, NULL, disk->info.alias,
1635
                                                   VIR_DOMAIN_EVENT_DISK_CHANGE_MISSING_ON_START);
1636 1637 1638
        if (event)
            qemuDomainEventQueue(driver, event);

M
Michal Privoznik 已提交
1639 1640 1641 1642 1643 1644 1645 1646
        VIR_FREE(disk->src);
    }

    ret = 0;

cleanup:
    return ret;
}