qemu_driver.c 392.2 KB
Newer Older
D
Daniel P. Berrange 已提交
1 2 3
/*
 * driver.c: core driver methods for managing qemu guests
 *
4
 * Copyright (C) 2006, 2007, 2008, 2009, 2010 Red Hat, Inc.
D
Daniel P. Berrange 已提交
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
 * 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>
 */

24
#include <config.h>
25

D
Daniel P. Berrange 已提交
26 27
#include <sys/types.h>
#include <sys/poll.h>
28
#include <sys/time.h>
D
Daniel P. Berrange 已提交
29 30 31
#include <dirent.h>
#include <limits.h>
#include <string.h>
32
#include <stdbool.h>
D
Daniel P. Berrange 已提交
33 34 35 36 37
#include <stdio.h>
#include <stdarg.h>
#include <stdlib.h>
#include <unistd.h>
#include <errno.h>
38
#include <sys/utsname.h>
39 40 41 42
#include <sys/stat.h>
#include <fcntl.h>
#include <signal.h>
#include <paths.h>
43 44
#include <pwd.h>
#include <stdio.h>
45
#include <sys/wait.h>
46
#include <sys/ioctl.h>
47
#include <sys/un.h>
D
Daniel P. Berrange 已提交
48

49 50
#include <libxml/xpathInternals.h>

51
#include "virterror_internal.h"
52
#include "logging.h"
53
#include "datatypes.h"
54 55
#include "qemu_driver.h"
#include "qemu_conf.h"
56
#include "qemu_monitor.h"
57
#include "qemu_bridge_filter.h"
J
Jim Meyering 已提交
58
#include "c-ctype.h"
59
#include "event.h"
60
#include "buf.h"
61
#include "util.h"
62
#include "nodeinfo.h"
63
#include "stats_linux.h"
64
#include "capabilities.h"
65
#include "memory.h"
66
#include "uuid.h"
67
#include "domain_conf.h"
68 69
#include "node_device_conf.h"
#include "pci.h"
70
#include "hostusb.h"
71
#include "processinfo.h"
72 73
#include "qemu_security_stacked.h"
#include "qemu_security_dac.h"
74
#include "cgroup.h"
C
Chris Lalancette 已提交
75
#include "libvirt_internal.h"
76
#include "xml.h"
77
#include "cpu/cpu.h"
78
#include "macvtap.h"
79
#include "domain_nwfilter.h"
80
#include "hooks.h"
81
#include "storage_file.h"
82

83

84 85
#define VIR_FROM_THIS VIR_FROM_QEMU

86 87 88
#define QEMU_VNC_PORT_MIN  5900
#define QEMU_VNC_PORT_MAX  65535

89 90
#define QEMU_NAMESPACE_HREF "http://libvirt.org/schemas/domain/qemu/1.0"

91 92 93 94 95 96
/* Only 1 job is allowed at any time
 * A job includes *all* monitor commands, even those just querying
 * information, not merely actions */
enum qemuDomainJob {
    QEMU_JOB_NONE = 0,  /* Always set to 0 for easy if (jobActive) conditions */
    QEMU_JOB_UNSPECIFIED,
97 98
    QEMU_JOB_MIGRATION_OUT,
    QEMU_JOB_MIGRATION_IN,
99 100 101 102 103
};

enum qemuDomainJobSignals {
    QEMU_JOB_SIGNAL_CANCEL  = 1 << 0, /* Request job cancellation */
    QEMU_JOB_SIGNAL_SUSPEND = 1 << 1, /* Request VM suspend to finish live migration offline */
104 105 106 107 108
    QEMU_JOB_SIGNAL_MIGRATE_DOWNTIME = 1 << 2, /* Request migration downtime change */
};

struct qemuDomainJobSignalsData {
    unsigned long long migrateDowntime; /* Data for QEMU_JOB_SIGNAL_MIGRATE_DOWNTIME */
109 110
};

111 112 113
typedef struct _qemuDomainObjPrivate qemuDomainObjPrivate;
typedef qemuDomainObjPrivate *qemuDomainObjPrivatePtr;
struct _qemuDomainObjPrivate {
114
    virCond jobCond; /* Use in conjunction with main virDomainObjPtr lock */
115 116
    enum qemuDomainJob jobActive;   /* Currently running job */
    unsigned int jobSignals;        /* Signals for running job */
117
    struct qemuDomainJobSignalsData jobSignalsData; /* Signal specific data */
118 119
    virDomainJobInfo jobInfo;
    unsigned long long jobStart;
120

121
    qemuMonitorPtr mon;
122
    virDomainChrDefPtr monConfig;
D
Daniel P. Berrange 已提交
123
    int monJSON;
124
    int monitor_warned;
125 126 127

    int nvcpupids;
    int *vcpupids;
128 129

    qemuDomainPCIAddressSetPtr pciaddrs;
130
    int persistentAddrs;
131 132
};

133 134
static int qemudShutdown(void);

135 136
static void qemuDriverLock(struct qemud_driver *driver)
{
137
    virMutexLock(&driver->lock);
138 139 140
}
static void qemuDriverUnlock(struct qemud_driver *driver)
{
141
    virMutexUnlock(&driver->lock);
142 143
}

144 145 146
static void qemuDomainEventFlush(int timer, void *opaque);
static void qemuDomainEventQueue(struct qemud_driver *driver,
                                 virDomainEventPtr event);
147

J
Jiri Denemark 已提交
148 149
static int qemudDomainObjStart(virConnectPtr conn,
                               struct qemud_driver *driver,
150 151
                               virDomainObjPtr vm,
                               bool start_paused);
J
Jiri Denemark 已提交
152

153 154
static int qemudStartVMDaemon(virConnectPtr conn,
                              struct qemud_driver *driver,
155
                              virDomainObjPtr vm,
156
                              const char *migrateFrom,
157
                              bool start_paused,
158 159
                              int stdin_fd,
                              const char *stdin_path);
160

161
static void qemudShutdownVMDaemon(struct qemud_driver *driver,
162 163
                                  virDomainObjPtr vm,
                                  int migrated);
164

165
static int qemudDomainGetMaxVcpus(virDomainPtr dom);
166

167
static int qemuDetectVcpuPIDs(struct qemud_driver *driver,
168
                              virDomainObjPtr vm);
169

170 171 172
static int qemuUpdateActivePciHostdevs(struct qemud_driver *driver,
                                       virDomainDefPtr def);

173 174 175
static int qemudVMFiltersInstantiate(virConnectPtr conn,
                                     virDomainDefPtr def);

J
Jim Meyering 已提交
176
static struct qemud_driver *qemu_driver = NULL;
177

178 179 180 181 182 183 184 185 186 187 188 189 190 191 192

static void *qemuDomainObjPrivateAlloc(void)
{
    qemuDomainObjPrivatePtr priv;

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

    return priv;
}

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

193
    qemuDomainPCIAddressSetFree(priv->pciaddrs);
194 195 196
    virDomainChrDefFree(priv->monConfig);
    VIR_FREE(priv->vcpupids);

197 198
    /* This should never be non-NULL if we get here, but just in case... */
    if (priv->mon) {
199
        VIR_ERROR0(_("Unexpected QEMU monitor still active during domain deletion"));
200 201 202 203 204 205
        qemuMonitorClose(priv->mon);
    }
    VIR_FREE(priv);
}


206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223
static int qemuDomainObjPrivateXMLFormat(virBufferPtr buf, void *data)
{
    qemuDomainObjPrivatePtr priv = data;
    const char *monitorpath;

    /* priv->monitor_chr is set only for qemu */
    if (priv->monConfig) {
        switch (priv->monConfig->type) {
        case VIR_DOMAIN_CHR_TYPE_UNIX:
            monitorpath = priv->monConfig->data.nix.path;
            break;
        default:
        case VIR_DOMAIN_CHR_TYPE_PTY:
            monitorpath = priv->monConfig->data.file.path;
            break;
        }

        virBufferEscapeString(buf, "  <monitor path='%s'", monitorpath);
D
Daniel P. Berrange 已提交
224 225
        if (priv->monJSON)
            virBufferAddLit(buf, " json='1'");
226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251
        virBufferVSprintf(buf, " type='%s'/>\n",
                          virDomainChrTypeToString(priv->monConfig->type));
    }


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

    return 0;
}

static int qemuDomainObjPrivateXMLParse(xmlXPathContextPtr ctxt, void *data)
{
    qemuDomainObjPrivatePtr priv = data;
    char *monitorpath;
    char *tmp;
    int n, i;
    xmlNodePtr *nodes = NULL;

    if (VIR_ALLOC(priv->monConfig) < 0) {
252
        virReportOOMError();
253 254 255
        goto error;
    }

256
    if (!(priv->monConfig->info.alias = strdup("monitor"))) {
257
        virReportOOMError();
258 259 260
        goto error;
    }

261
    if (!(monitorpath =
262
          virXPathString("string(./monitor[1]/@path)", ctxt))) {
263 264
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("no monitor path"));
265 266 267
        goto error;
    }

268
    tmp = virXPathString("string(./monitor[1]/@type)", ctxt);
269 270 271 272 273 274
    if (tmp)
        priv->monConfig->type = virDomainChrTypeFromString(tmp);
    else
        priv->monConfig->type = VIR_DOMAIN_CHR_TYPE_PTY;
    VIR_FREE(tmp);

275
    if (virXPathBoolean("count(./monitor[@json = '1']) > 0", ctxt)) {
D
Daniel P. Berrange 已提交
276
        priv->monJSON = 1;
277 278 279
    } else {
        priv->monJSON = 0;
    }
D
Daniel P. Berrange 已提交
280

281 282 283 284 285 286 287 288 289
    switch (priv->monConfig->type) {
    case VIR_DOMAIN_CHR_TYPE_PTY:
        priv->monConfig->data.file.path = monitorpath;
        break;
    case VIR_DOMAIN_CHR_TYPE_UNIX:
        priv->monConfig->data.nix.path = monitorpath;
        break;
    default:
        VIR_FREE(monitorpath);
290 291 292
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("unsupported monitor type '%s'"),
                        virDomainChrTypeToString(priv->monConfig->type));
293 294 295
        goto error;
    }

296
    n = virXPathNodeSet("./vcpus/vcpu", ctxt, &nodes);
297 298 299 300 301
    if (n < 0)
        goto error;
    if (n) {
        priv->nvcpupids = n;
        if (VIR_REALLOC_N(priv->vcpupids, priv->nvcpupids) < 0) {
302
            virReportOOMError();
303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322
            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);
    }

    return 0;

error:
323 324
    virDomainChrDefFree(priv->monConfig);
    priv->monConfig = NULL;
325 326 327 328 329 330
    VIR_FREE(nodes);
    return -1;
}



331 332 333 334 335 336 337 338 339
/*
 * 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
 */
340 341 342 343

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

344 345 346 347
static int qemuDomainObjBeginJob(virDomainObjPtr obj) ATTRIBUTE_RETURN_CHECK;
static int qemuDomainObjBeginJob(virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
348 349 350 351
    struct timeval now;
    unsigned long long then;

    if (gettimeofday(&now, NULL) < 0) {
352
        virReportSystemError(errno, "%s",
353 354 355 356 357
                             _("cannot get time of day"));
        return -1;
    }
    then = (now.tv_sec * 1000ull) + (now.tv_usec / 1000);
    then += QEMU_JOB_WAIT_TIME;
358 359 360 361

    virDomainObjRef(obj);

    while (priv->jobActive) {
362
        if (virCondWaitUntil(&priv->jobCond, &obj->lock, then) < 0) {
363
            virDomainObjUnref(obj);
364
            if (errno == ETIMEDOUT)
365 366
                qemuReportError(VIR_ERR_OPERATION_TIMEOUT,
                                "%s", _("cannot acquire state change lock"));
367
            else
368
                virReportSystemError(errno,
369
                                     "%s", _("cannot acquire job mutex"));
370 371 372
            return -1;
        }
    }
373 374
    priv->jobActive = QEMU_JOB_UNSPECIFIED;
    priv->jobSignals = 0;
375
    memset(&priv->jobSignalsData, 0, sizeof(priv->jobSignalsData));
376 377
    priv->jobStart = (now.tv_sec * 1000ull) + (now.tv_usec / 1000);
    memset(&priv->jobInfo, 0, sizeof(priv->jobInfo));
378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393

    return 0;
}

/*
 * obj must be locked before calling, qemud_driver must 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.
 */
static int qemuDomainObjBeginJobWithDriver(struct qemud_driver *driver,
                                           virDomainObjPtr obj) ATTRIBUTE_RETURN_CHECK;
static int qemuDomainObjBeginJobWithDriver(struct qemud_driver *driver,
                                           virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
394 395 396 397
    struct timeval now;
    unsigned long long then;

    if (gettimeofday(&now, NULL) < 0) {
398
        virReportSystemError(errno, "%s",
399 400 401 402 403
                             _("cannot get time of day"));
        return -1;
    }
    then = (now.tv_sec * 1000ull) + (now.tv_usec / 1000);
    then += QEMU_JOB_WAIT_TIME;
404 405 406 407 408

    virDomainObjRef(obj);
    qemuDriverUnlock(driver);

    while (priv->jobActive) {
409
        if (virCondWaitUntil(&priv->jobCond, &obj->lock, then) < 0) {
410
            virDomainObjUnref(obj);
411
            if (errno == ETIMEDOUT)
412 413
                qemuReportError(VIR_ERR_OPERATION_TIMEOUT,
                                "%s", _("cannot acquire state change lock"));
414
            else
415
                virReportSystemError(errno,
416
                                     "%s", _("cannot acquire job mutex"));
M
Matthias Bolte 已提交
417
            qemuDriverLock(driver);
418 419 420
            return -1;
        }
    }
421 422
    priv->jobActive = QEMU_JOB_UNSPECIFIED;
    priv->jobSignals = 0;
423
    memset(&priv->jobSignalsData, 0, sizeof(priv->jobSignalsData));
424 425
    priv->jobStart = (now.tv_sec * 1000ull) + (now.tv_usec / 1000);
    memset(&priv->jobInfo, 0, sizeof(priv->jobInfo));
426 427 428 429 430 431 432 433 434 435 436 437 438

    virDomainObjUnlock(obj);
    qemuDriverLock(driver);
    virDomainObjLock(obj);

    return 0;
}

/*
 * obj must be locked before calling, qemud_driver does not matter
 *
 * To be called after completing the work associated with the
 * earlier  qemuDomainBeginJob() call
439 440 441
 *
 * Returns remaining refcount on 'obj', maybe 0 to indicated it
 * was deleted
442
 */
443
static int ATTRIBUTE_RETURN_CHECK qemuDomainObjEndJob(virDomainObjPtr obj)
444 445 446
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

447 448
    priv->jobActive = QEMU_JOB_NONE;
    priv->jobSignals = 0;
449
    memset(&priv->jobSignalsData, 0, sizeof(priv->jobSignalsData));
450 451
    priv->jobStart = 0;
    memset(&priv->jobInfo, 0, sizeof(priv->jobInfo));
452 453
    virCondSignal(&priv->jobCond);

454
    return virDomainObjUnref(obj);
455 456 457 458 459 460 461 462 463 464 465
}


/*
 * obj must be locked before calling, qemud_driver must be unlocked
 *
 * To be called immediately before any QEMU monitor API call
 * Must have alrady called qemuDomainObjBeginJob().
 *
 * To be followed with qemuDomainObjExitMonitor() once complete
 */
466 467 468 469 470
static void qemuDomainObjEnterMonitor(virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    qemuMonitorLock(priv->mon);
471
    qemuMonitorRef(priv->mon);
472
    virDomainObjUnlock(obj);
473 474 475
}


476 477 478 479
/* obj must NOT be locked before calling, qemud_driver must be unlocked
 *
 * Should be paired with an earlier  qemuDomainObjEnterMonitor() call
 */
480 481 482
static void qemuDomainObjExitMonitor(virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
483 484 485 486 487 488
    int refs;

    refs = qemuMonitorUnref(priv->mon);

    if (refs > 0)
        qemuMonitorUnlock(priv->mon);
489

490
    virDomainObjLock(obj);
491 492 493 494 495

    if (refs == 0) {
        virDomainObjUnref(obj);
        priv->mon = NULL;
    }
496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511
}


/*
 * obj must be locked before calling, qemud_driver must be locked
 *
 * To be called immediately before any QEMU monitor API call
 * Must have alrady called qemuDomainObjBeginJob().
 *
 * To be followed with qemuDomainObjExitMonitorWithDriver() once complete
 */
static void qemuDomainObjEnterMonitorWithDriver(struct qemud_driver *driver, virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    qemuMonitorLock(priv->mon);
512
    qemuMonitorRef(priv->mon);
513 514 515 516 517 518 519 520 521 522 523 524 525
    virDomainObjUnlock(obj);
    qemuDriverUnlock(driver);
}


/* obj must NOT be locked before calling, qemud_driver must be unlocked,
 * and will be locked after returning
 *
 * Should be paired with an earlier  qemuDomainObjEnterMonitor() call
 */
static void qemuDomainObjExitMonitorWithDriver(struct qemud_driver *driver, virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
526 527 528 529 530 531
    int refs;

    refs = qemuMonitorUnref(priv->mon);

    if (refs > 0)
        qemuMonitorUnlock(priv->mon);
532 533 534

    qemuDriverLock(driver);
    virDomainObjLock(obj);
535 536 537 538 539

    if (refs == 0) {
        virDomainObjUnref(obj);
        priv->mon = NULL;
    }
540 541
}

542 543 544 545 546 547 548 549 550 551 552 553 554 555 556
static void qemuDomainObjEnterRemoteWithDriver(struct qemud_driver *driver,
                                               virDomainObjPtr obj)
{
    virDomainObjRef(obj);
    virDomainObjUnlock(obj);
    qemuDriverUnlock(driver);
}

static void qemuDomainObjExitRemoteWithDriver(struct qemud_driver *driver,
                                              virDomainObjPtr obj)
{
    qemuDriverLock(driver);
    virDomainObjLock(obj);
    virDomainObjUnref(obj);
}
557

558
void qemuDomainDefNamespaceFree(void *nsdata)
559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577
{
    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);
}

578 579 580 581
int qemuDomainDefNamespaceParse(xmlDocPtr xml,
                                xmlNodePtr root,
                                xmlXPathContextPtr ctxt,
                                void **data)
582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690
{
    qemuDomainCmdlineDefPtr cmd = NULL;
    xmlNsPtr ns;
    xmlNodePtr *nodes = NULL;
    int n, i;

    ns = xmlSearchNs(xml, root, BAD_CAST "qemu");
    if (!ns)
        /* this is fine; it just means there was no qemu namespace listed */
        return 0;

    if (STRNEQ((const char *)ns->href, QEMU_NAMESPACE_HREF)) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Found namespace '%s' doesn't match expected '%s'"),
                        ns->href, QEMU_NAMESPACE_HREF);
        return -1;
    }

    if (xmlXPathRegisterNs(ctxt, ns->prefix, ns->href) < 0) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Failed to register xml namespace '%s'"), ns->href);
        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)
        /* virXPathNodeSet already set the error */
        goto error;

    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)
        /* virXPathNodeSet already set the error */
        goto error;

    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);

    *data = cmd;

    return 0;

no_memory:
    virReportOOMError();

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

691 692
int qemuDomainDefNamespaceFormatXML(virBufferPtr buf,
                                    void *nsdata)
693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714
{
    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++) {
        virBufferVSprintf(buf, "    <qemu:env name='%s'", cmd->env_name[i]);
        if (cmd->env_value[i])
            virBufferEscapeString(buf, " value='%s'", cmd->env_value[i]);
        virBufferAddLit(buf, "/>\n");
    }
    virBufferAddLit(buf, "  </qemu:commandline>\n");

    return 0;
}

715
const char *qemuDomainDefNamespaceHref(void)
716 717 718 719
{
    return "xmlns:qemu='" QEMU_NAMESPACE_HREF "'";
}

720 721 722 723 724 725 726 727 728
static int qemuCgroupControllerActive(struct qemud_driver *driver,
                                      int controller)
{
    if (driver->cgroup == NULL)
        return 0;
    if (driver->cgroupControllers & (1 << controller))
        return 1;
    return 0;
}
729

730
static int
731
qemudLogFD(struct qemud_driver *driver, const char* name)
732 733 734
{
    char logfile[PATH_MAX];
    mode_t logmode;
G
Guido Günther 已提交
735
    int ret, fd = -1;
736

737 738
    if ((ret = snprintf(logfile, sizeof(logfile), "%s/%s.log",
                        driver->logDir, name))
G
Guido Günther 已提交
739
        < 0 || ret >= sizeof(logfile)) {
740
        virReportOOMError();
741 742 743 744
        return -1;
    }

    logmode = O_CREAT | O_WRONLY;
745 746
    /* Only logrotate files in /var/log, so only append if running privileged */
    if (driver->privileged)
747
        logmode |= O_APPEND;
748 749 750
    else
        logmode |= O_TRUNC;

751
    if ((fd = open(logfile, logmode, S_IRUSR | S_IWUSR)) < 0) {
752
        virReportSystemError(errno,
753 754
                             _("failed to create logfile %s"),
                             logfile);
755 756
        return -1;
    }
757
    if (virSetCloseExec(fd) < 0) {
758
        virReportSystemError(errno, "%s",
759
                             _("Unable to set VM logfile close-on-exec flag"));
760 761 762 763 764 765 766
        close(fd);
        return -1;
    }
    return fd;
}


767
static int
768
qemudLogReadFD(const char* logDir, const char* name, off_t pos)
769 770 771 772 773 774 775
{
    char logfile[PATH_MAX];
    mode_t logmode = O_RDONLY;
    int ret, fd = -1;

    if ((ret = snprintf(logfile, sizeof(logfile), "%s/%s.log", logDir, name))
        < 0 || ret >= sizeof(logfile)) {
776 777 778
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("failed to build logfile name %s/%s.log"),
                        logDir, name);
779 780 781 782 783
        return -1;
    }


    if ((fd = open(logfile, logmode)) < 0) {
784
        virReportSystemError(errno,
785 786
                             _("failed to create logfile %s"),
                             logfile);
787 788
        return -1;
    }
789
    if (virSetCloseExec(fd) < 0) {
790
        virReportSystemError(errno, "%s",
791
                             _("Unable to set VM logfile close-on-exec flag"));
792 793 794
        close(fd);
        return -1;
    }
795
    if (pos < 0 || lseek(fd, pos, SEEK_SET) < 0) {
796
      virReportSystemError(pos < 0 ? 0 : errno,
797 798
                             _("Unable to seek to %lld in %s"),
                             (long long) pos, logfile);
799 800 801 802 803 804
        close(fd);
    }
    return fd;
}


805 806 807 808 809 810 811 812 813
struct qemuAutostartData {
    struct qemud_driver *driver;
    virConnectPtr conn;
};
static void
qemuAutostartDomain(void *payload, const char *name ATTRIBUTE_UNUSED, void *opaque)
{
    virDomainObjPtr vm = payload;
    struct qemuAutostartData *data = opaque;
814
    virErrorPtr err;
815 816

    virDomainObjLock(vm);
817 818 819 820 821 822 823 824 825
    virResetLastError();
    if (qemuDomainObjBeginJobWithDriver(data->driver, vm) < 0) {
        err = virGetLastError();
        VIR_ERROR(_("Failed to start job on VM '%s': %s"),
                  vm->def->name,
                  err ? err->message : _("unknown error"));
    } else {
        if (vm->autostart &&
            !virDomainObjIsActive(vm) &&
826
            qemudDomainObjStart(data->conn, data->driver, vm, false) < 0) {
827
            err = virGetLastError();
828
            VIR_ERROR(_("Failed to autostart VM '%s': %s"),
829
                      vm->def->name,
830
                      err ? err->message : _("unknown error"));
831
        }
832 833 834

        if (qemuDomainObjEndJob(vm) == 0)
            vm = NULL;
835
    }
836 837 838

    if (vm)
        virDomainObjUnlock(vm);
839 840
}

841 842
static void
qemudAutostartConfigs(struct qemud_driver *driver) {
843 844 845 846 847
    /* XXX: Figure out a better way todo this. The domain
     * startup code needs a connection handle in order
     * to lookup the bridge associated with a virtual
     * network
     */
848 849 850
    virConnectPtr conn = virConnectOpen(driver->privileged ?
                                        "qemu:///system" :
                                        "qemu:///session");
851
    /* Ignoring NULL conn which is mostly harmless here */
852
    struct qemuAutostartData data = { driver, conn };
853

854
    qemuDriverLock(driver);
855
    virHashForEach(driver->domains.objs, qemuAutostartDomain, &data);
856
    qemuDriverUnlock(driver);
857

858 859
    if (conn)
        virConnectClose(conn);
860 861
}

862 863 864 865 866 867 868 869 870

/**
 * qemudRemoveDomainStatus
 *
 * remove all state files of a domain from statedir
 *
 * Returns 0 on success
 */
static int
871
qemudRemoveDomainStatus(struct qemud_driver *driver,
872 873
                        virDomainObjPtr vm)
{
874
    char ebuf[1024];
875 876 877
    char *file = NULL;

    if (virAsprintf(&file, "%s/%s.xml", driver->stateDir, vm->def->name) < 0) {
878
        virReportOOMError();
D
Daniel Veillard 已提交
879
        return(-1);
880 881
    }

882
    if (unlink(file) < 0 && errno != ENOENT && errno != ENOTDIR)
883
        VIR_WARN("Failed to remove domain XML for %s: %s",
D
Daniel Veillard 已提交
884 885 886
                 vm->def->name, virStrerror(errno, ebuf, sizeof(ebuf)));
    VIR_FREE(file);

887
    if (virFileDeletePid(driver->stateDir, vm->def->name) != 0)
888
        VIR_WARN("Failed to remove PID file for %s: %s",
889
                 vm->def->name, virStrerror(errno, ebuf, sizeof(ebuf)));
890

D
Daniel Veillard 已提交
891

892
    return 0;
893 894
}

895 896 897 898 899 900 901 902 903 904 905 906 907 908

/*
 * This is a callback registered with a qemuMonitorPtr  instance,
 * and to be invoked when the monitor console hits an end of file
 * condition, or error, thus indicating VM shutdown should be
 * performed
 */
static void
qemuHandleMonitorEOF(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                     virDomainObjPtr vm,
                     int hasError) {
    struct qemud_driver *driver = qemu_driver;
    virDomainEventPtr event = NULL;

909
    VIR_DEBUG("Received EOF on %p '%s'", vm, vm->def->name);
910 911 912 913 914 915 916 917
    virDomainObjLock(vm);

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     hasError ?
                                     VIR_DOMAIN_EVENT_STOPPED_FAILED :
                                     VIR_DOMAIN_EVENT_STOPPED_SHUTDOWN);

918
    qemudShutdownVMDaemon(driver, vm, 0);
919 920 921 922 923 924 925 926 927 928 929 930 931
    if (!vm->persistent)
        virDomainRemoveInactive(&driver->domains, vm);
    else
        virDomainObjUnlock(vm);

    if (event) {
        qemuDriverLock(driver);
        qemuDomainEventQueue(driver, event);
        qemuDriverUnlock(driver);
    }
}


932 933 934
static virDomainDiskDefPtr
findDomainDiskByPath(virDomainObjPtr vm,
                     const char *path)
935 936 937 938 939 940 941
{
    int i;

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

        disk = vm->def->disks[i];
942 943
        if (disk->src != NULL && STREQ(disk->src, path))
            return disk;
944
    }
945 946 947 948

    qemuReportError(VIR_ERR_INTERNAL_ERROR,
                    _("no disk found with path %s"),
                    path);
949 950 951
    return NULL;
}

952 953 954 955 956 957
static virDomainDiskDefPtr
findDomainDiskByAlias(virDomainObjPtr vm,
                      const char *alias)
{
    int i;

958 959 960
    if (STRPREFIX(alias, QEMU_DRIVE_HOST_PREFIX))
        alias += strlen(QEMU_DRIVE_HOST_PREFIX);

961 962 963 964 965 966 967 968 969 970 971 972 973 974
    for (i = 0; i < vm->def->ndisks; i++) {
        virDomainDiskDefPtr disk;

        disk = vm->def->disks[i];
        if (disk->info.alias != NULL && STREQ(disk->info.alias, alias))
            return disk;
    }

    qemuReportError(VIR_ERR_INTERNAL_ERROR,
                    _("no disk found with alias %s"),
                    alias);
    return NULL;
}

975
static int
976 977 978 979
getVolumeQcowPassphrase(virConnectPtr conn,
                        virDomainDiskDefPtr disk,
                        char **secretRet,
                        size_t *secretLen)
980 981 982 983 984
{
    virSecretPtr secret;
    char *passphrase;
    unsigned char *data;
    size_t size;
985
    int ret = -1;
986
    virStorageEncryptionPtr enc;
987

988 989 990 991 992 993 994
    if (!disk->encryption) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("disk %s does not have any encryption information"),
                        disk->src);
        return -1;
    }
    enc = disk->encryption;
995 996

    if (!conn) {
997
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
998
                        "%s", _("cannot find secrets without a connection"));
999
        goto cleanup;
1000 1001 1002 1003 1004
    }

    if (conn->secretDriver == NULL ||
        conn->secretDriver->lookupByUUID == NULL ||
        conn->secretDriver->getValue == NULL) {
1005 1006
        qemuReportError(VIR_ERR_NO_SUPPORT, "%s",
                        _("secret storage not supported"));
1007
        goto cleanup;
1008 1009 1010 1011 1012 1013
    }

    if (enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_QCOW ||
        enc->nsecrets != 1 ||
        enc->secrets[0]->type !=
        VIR_STORAGE_ENCRYPTION_SECRET_TYPE_PASSPHRASE) {
1014
        qemuReportError(VIR_ERR_INVALID_DOMAIN,
1015
                        _("invalid <encryption> for volume %s"), disk->src);
1016
        goto cleanup;
1017 1018 1019 1020 1021
    }

    secret = conn->secretDriver->lookupByUUID(conn,
                                              enc->secrets[0]->uuid);
    if (secret == NULL)
1022
        goto cleanup;
1023 1024 1025 1026
    data = conn->secretDriver->getValue(secret, &size,
                                        VIR_SECRET_GET_VALUE_INTERNAL_CALL);
    virUnrefSecret(secret);
    if (data == NULL)
1027
        goto cleanup;
1028 1029 1030 1031

    if (memchr(data, '\0', size) != NULL) {
        memset(data, 0, size);
        VIR_FREE(data);
1032 1033
        qemuReportError(VIR_ERR_INVALID_SECRET,
                        _("format='qcow' passphrase for %s must not contain a "
1034
                          "'\\0'"), disk->src);
1035
        goto cleanup;
1036 1037 1038 1039 1040
    }

    if (VIR_ALLOC_N(passphrase, size + 1) < 0) {
        memset(data, 0, size);
        VIR_FREE(data);
1041
        virReportOOMError();
1042
        goto cleanup;
1043 1044 1045 1046 1047 1048 1049 1050 1051 1052
    }
    memcpy(passphrase, data, size);
    passphrase[size] = '\0';

    memset(data, 0, size);
    VIR_FREE(data);

    *secretRet = passphrase;
    *secretLen = size;

1053 1054 1055
    ret = 0;

cleanup:
1056 1057
    return ret;
}
1058

1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079
static int
findVolumeQcowPassphrase(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                         virConnectPtr conn,
                         virDomainObjPtr vm,
                         const char *path,
                         char **secretRet,
                         size_t *secretLen)
{
    virDomainDiskDefPtr disk;
    int ret = -1;

    virDomainObjLock(vm);
    disk = findDomainDiskByPath(vm, path);

    if (!disk)
        goto cleanup;

    ret = getVolumeQcowPassphrase(conn, disk, secretRet, secretLen);

cleanup:
    virDomainObjUnlock(vm);
1080
    return ret;
1081 1082
}

1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104

static int
qemuHandleDomainReset(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                      virDomainObjPtr vm)
{
    struct qemud_driver *driver = qemu_driver;
    virDomainEventPtr event;

    virDomainObjLock(vm);
    event = virDomainEventRebootNewFromObj(vm);
    virDomainObjUnlock(vm);

    if (event) {
        qemuDriverLock(driver);
        qemuDomainEventQueue(driver, event);
        qemuDriverUnlock(driver);
    }

    return 0;
}


1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136
static int
qemuHandleDomainStop(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                     virDomainObjPtr vm)
{
    struct qemud_driver *driver = qemu_driver;
    virDomainEventPtr event = NULL;

    virDomainObjLock(vm);
    if (vm->state == VIR_DOMAIN_RUNNING) {
        VIR_DEBUG("Transitioned guest %s to paused state due to unknown event", vm->def->name);

        vm->state = VIR_DOMAIN_PAUSED;
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_SUSPENDED,
                                         VIR_DOMAIN_EVENT_SUSPENDED_PAUSED);

        if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
            VIR_WARN("Unable to save status on vm %s after IO error", vm->def->name);
    }
    virDomainObjUnlock(vm);

    if (event) {
        qemuDriverLock(driver);
        if (event)
            qemuDomainEventQueue(driver, event);
        qemuDriverUnlock(driver);
    }

    return 0;
}


1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165
static int
qemuHandleDomainRTCChange(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                          virDomainObjPtr vm,
                          long long offset)
{
    struct qemud_driver *driver = qemu_driver;
    virDomainEventPtr event;

    virDomainObjLock(vm);
    event = virDomainEventRTCChangeNewFromObj(vm, offset);

    if (vm->def->clock.offset == VIR_DOMAIN_CLOCK_OFFSET_VARIABLE)
        vm->def->clock.data.adjustment = offset;

    if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
        VIR_WARN0("unable to save domain status with RTC change");

    virDomainObjUnlock(vm);

    if (event) {
        qemuDriverLock(driver);
        qemuDomainEventQueue(driver, event);
        qemuDriverUnlock(driver);
    }

    return 0;
}


1166 1167 1168 1169 1170 1171
static int
qemuHandleDomainWatchdog(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                         virDomainObjPtr vm,
                         int action)
{
    struct qemud_driver *driver = qemu_driver;
1172 1173
    virDomainEventPtr watchdogEvent = NULL;
    virDomainEventPtr lifecycleEvent = NULL;
1174 1175

    virDomainObjLock(vm);
1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189
    watchdogEvent = virDomainEventWatchdogNewFromObj(vm, action);

    if (action == VIR_DOMAIN_EVENT_WATCHDOG_PAUSE &&
        vm->state == VIR_DOMAIN_RUNNING) {
        VIR_DEBUG("Transitioned guest %s to paused state due to watchdog", vm->def->name);

        vm->state = VIR_DOMAIN_PAUSED;
        lifecycleEvent = virDomainEventNewFromObj(vm,
                                                  VIR_DOMAIN_EVENT_SUSPENDED,
                                                  VIR_DOMAIN_EVENT_SUSPENDED_WATCHDOG);

        if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
            VIR_WARN("Unable to save status on vm %s after IO error", vm->def->name);
    }
1190 1191
    virDomainObjUnlock(vm);

1192
    if (watchdogEvent || lifecycleEvent) {
1193
        qemuDriverLock(driver);
1194 1195 1196 1197
        if (watchdogEvent)
            qemuDomainEventQueue(driver, watchdogEvent);
        if (lifecycleEvent)
            qemuDomainEventQueue(driver, lifecycleEvent);
1198 1199 1200 1201 1202 1203 1204
        qemuDriverUnlock(driver);
    }

    return 0;
}


1205 1206 1207 1208
static int
qemuHandleDomainIOError(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                        virDomainObjPtr vm,
                        const char *diskAlias,
1209 1210
                        int action,
                        const char *reason)
1211 1212
{
    struct qemud_driver *driver = qemu_driver;
1213
    virDomainEventPtr ioErrorEvent = NULL;
1214
    virDomainEventPtr ioErrorEvent2 = NULL;
1215
    virDomainEventPtr lifecycleEvent = NULL;
1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230
    const char *srcPath;
    const char *devAlias;
    virDomainDiskDefPtr disk;

    virDomainObjLock(vm);
    disk = findDomainDiskByAlias(vm, diskAlias);

    if (disk) {
        srcPath = disk->src;
        devAlias = disk->info.alias;
    } else {
        srcPath = "";
        devAlias = "";
    }

1231
    ioErrorEvent = virDomainEventIOErrorNewFromObj(vm, srcPath, devAlias, action);
1232
    ioErrorEvent2 = virDomainEventIOErrorReasonNewFromObj(vm, srcPath, devAlias, action, reason);
1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245

    if (action == VIR_DOMAIN_EVENT_IO_ERROR_PAUSE &&
        vm->state == VIR_DOMAIN_RUNNING) {
        VIR_DEBUG("Transitioned guest %s to paused state due to IO error", vm->def->name);

        vm->state = VIR_DOMAIN_PAUSED;
        lifecycleEvent = virDomainEventNewFromObj(vm,
                                                  VIR_DOMAIN_EVENT_SUSPENDED,
                                                  VIR_DOMAIN_EVENT_SUSPENDED_IOERROR);

        if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
            VIR_WARN("Unable to save status on vm %s after IO error", vm->def->name);
    }
1246 1247
    virDomainObjUnlock(vm);

1248
    if (ioErrorEvent || ioErrorEvent2 || lifecycleEvent) {
1249
        qemuDriverLock(driver);
1250 1251
        if (ioErrorEvent)
            qemuDomainEventQueue(driver, ioErrorEvent);
1252 1253
        if (ioErrorEvent2)
            qemuDomainEventQueue(driver, ioErrorEvent2);
1254 1255
        if (lifecycleEvent)
            qemuDomainEventQueue(driver, lifecycleEvent);
1256 1257 1258 1259 1260 1261 1262
        qemuDriverUnlock(driver);
    }

    return 0;
}


1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354
static int
qemuHandleDomainGraphics(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                         virDomainObjPtr vm,
                         int phase,
                         int localFamily,
                         const char *localNode,
                         const char *localService,
                         int remoteFamily,
                         const char *remoteNode,
                         const char *remoteService,
                         const char *authScheme,
                         const char *x509dname,
                         const char *saslUsername)
{
    struct qemud_driver *driver = qemu_driver;
    virDomainEventPtr event;
    virDomainEventGraphicsAddressPtr localAddr = NULL;
    virDomainEventGraphicsAddressPtr remoteAddr = NULL;
    virDomainEventGraphicsSubjectPtr subject = NULL;
    int i;

    virDomainObjLock(vm);

    if (VIR_ALLOC(localAddr) < 0)
        goto no_memory;
    localAddr->family = localFamily;
    if (!(localAddr->service = strdup(localService)) ||
        !(localAddr->node = strdup(localNode)))
        goto no_memory;

    if (VIR_ALLOC(remoteAddr) < 0)
        goto no_memory;
    remoteAddr->family = remoteFamily;
    if (!(remoteAddr->service = strdup(remoteService)) ||
        !(remoteAddr->node = strdup(remoteNode)))
        goto no_memory;

    if (VIR_ALLOC(subject) < 0)
        goto no_memory;
    if (x509dname) {
        if (VIR_REALLOC_N(subject->identities, subject->nidentity+1) < 0)
            goto no_memory;
        if (!(subject->identities[subject->nidentity].type = strdup("x509dname")) ||
            !(subject->identities[subject->nidentity].name = strdup(x509dname)))
            goto no_memory;
        subject->nidentity++;
    }
    if (saslUsername) {
        if (VIR_REALLOC_N(subject->identities, subject->nidentity+1) < 0)
            goto no_memory;
        if (!(subject->identities[subject->nidentity].type = strdup("saslUsername")) ||
            !(subject->identities[subject->nidentity].name = strdup(saslUsername)))
            goto no_memory;
        subject->nidentity++;
    }

    event = virDomainEventGraphicsNewFromObj(vm, phase, localAddr, remoteAddr, authScheme, subject);
    virDomainObjUnlock(vm);

    if (event) {
        qemuDriverLock(driver);
        qemuDomainEventQueue(driver, event);
        qemuDriverUnlock(driver);
    }

    return 0;

no_memory:
    virReportOOMError();
    if (localAddr) {
        VIR_FREE(localAddr->service);
        VIR_FREE(localAddr->node);
        VIR_FREE(localAddr);
    }
    if (remoteAddr) {
        VIR_FREE(remoteAddr->service);
        VIR_FREE(remoteAddr->node);
        VIR_FREE(remoteAddr);
    }
    if (subject) {
        for (i = 0 ; i < subject->nidentity ; i++) {
            VIR_FREE(subject->identities[i].type);
            VIR_FREE(subject->identities[i].name);
        }
        VIR_FREE(subject->identities);
        VIR_FREE(subject);
    }

    return -1;
}


1355 1356 1357 1358 1359 1360 1361 1362 1363
static void qemuHandleMonitorDestroy(qemuMonitorPtr mon,
                                     virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    if (priv->mon == mon)
        priv->mon = NULL;
    virDomainObjUnref(vm);
}

1364
static qemuMonitorCallbacks monitorCallbacks = {
1365
    .destroy = qemuHandleMonitorDestroy,
1366 1367
    .eofNotify = qemuHandleMonitorEOF,
    .diskSecretLookup = findVolumeQcowPassphrase,
1368
    .domainStop = qemuHandleDomainStop,
1369
    .domainReset = qemuHandleDomainReset,
1370
    .domainRTCChange = qemuHandleDomainRTCChange,
1371
    .domainWatchdog = qemuHandleDomainWatchdog,
1372
    .domainIOError = qemuHandleDomainIOError,
1373
    .domainGraphics = qemuHandleDomainGraphics,
1374 1375
};

1376
static int
1377
qemuConnectMonitor(struct qemud_driver *driver, virDomainObjPtr vm)
1378
{
1379
    qemuDomainObjPrivatePtr priv = vm->privateData;
1380
    int ret = -1;
1381

1382 1383 1384 1385 1386 1387
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecuritySocketLabel &&
        driver->securityDriver->domainSetSecuritySocketLabel
          (driver->securityDriver,vm) < 0) {
        VIR_ERROR(_("Failed to set security context for monitor for %s"),
                  vm->def->name);
1388 1389 1390
        goto error;
    }

1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401
    /* Hold an extra reference because we can't allow 'vm' to be
     * deleted while the monitor is active */
    virDomainObjRef(vm);

    priv->mon = qemuMonitorOpen(vm,
                                priv->monConfig,
                                priv->monJSON,
                                &monitorCallbacks);

    if (priv->mon == NULL)
        virDomainObjUnref(vm);
1402

1403 1404 1405 1406 1407 1408
    if (driver->securityDriver &&
        driver->securityDriver->domainClearSecuritySocketLabel &&
        driver->securityDriver->domainClearSecuritySocketLabel
          (driver->securityDriver,vm) < 0) {
        VIR_ERROR(_("Failed to clear security context for monitor for %s"),
                  vm->def->name);
1409
        goto error;
1410
    }
1411

1412 1413 1414 1415 1416 1417
    if (priv->mon == NULL) {
        VIR_INFO("Failed to connect monitor for %s", vm->def->name);
        goto error;
    }


1418 1419 1420 1421
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
    ret = qemuMonitorSetCapabilities(priv->mon);
    qemuDomainObjExitMonitorWithDriver(driver, vm);

1422
error:
1423
    if (ret < 0)
1424 1425 1426
        qemuMonitorClose(priv->mon);

    return ret;
1427
}
1428

1429 1430 1431 1432
struct virReconnectDomainData {
    virConnectPtr conn;
    struct qemud_driver *driver;
};
1433 1434 1435
/*
 * Open an existing VM's monitor, re-detect VCPU threads
 * and re-reserve the security labels in use
1436
 */
1437 1438
static void
qemuReconnectDomain(void *payload, const char *name ATTRIBUTE_UNUSED, void *opaque)
1439
{
1440
    virDomainObjPtr obj = payload;
1441 1442
    struct virReconnectDomainData *data = opaque;
    struct qemud_driver *driver = data->driver;
1443
    qemuDomainObjPrivatePtr priv;
1444
    unsigned long long qemuCmdFlags;
1445
    virConnectPtr conn = data->conn;
1446 1447

    virDomainObjLock(obj);
1448

1449 1450
    VIR_DEBUG("Reconnect monitor to %p '%s'", obj, obj->def->name);

1451 1452
    priv = obj->privateData;

1453
    /* XXX check PID liveliness & EXE path */
1454
    if (qemuConnectMonitor(driver, obj) < 0)
1455
        goto error;
1456

1457 1458 1459 1460
    if (qemuUpdateActivePciHostdevs(driver, obj->def) < 0) {
        goto error;
    }

1461 1462 1463 1464 1465 1466
    /* XXX we should be persisting the original flags in the XML
     * not re-detecting them, since the binary may have changed
     * since launch time */
    if (qemudExtractVersionInfo(obj->def->emulator,
                                NULL,
                                &qemuCmdFlags) >= 0 &&
1467
        (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)) {
1468 1469
        priv->persistentAddrs = 1;

1470 1471 1472 1473
        if (!(priv->pciaddrs = qemuDomainPCIAddressSetCreate(obj->def)) ||
            qemuAssignDevicePCISlots(obj->def, priv->pciaddrs) < 0)
            goto error;
    }
1474

1475
    if (driver->securityDriver &&
1476
        driver->securityDriver->domainReserveSecurityLabel &&
1477 1478
        driver->securityDriver->domainReserveSecurityLabel(driver->securityDriver,
                                                           obj) < 0)
1479
        goto error;
1480

1481 1482 1483
    if (qemudVMFiltersInstantiate(conn, obj->def))
        goto error;

1484 1485
    if (obj->def->id >= driver->nextvmid)
        driver->nextvmid = obj->def->id + 1;
1486

1487 1488
    virDomainObjUnlock(obj);
    return;
1489

1490
error:
1491 1492 1493
    /* We can't get the monitor back, so must kill the VM
     * to remove danger of it ending up running twice if
     * user tries to start it again later */
1494
    qemudShutdownVMDaemon(driver, obj, 0);
1495 1496 1497 1498
    if (!obj->persistent)
        virDomainRemoveInactive(&driver->domains, obj);
    else
        virDomainObjUnlock(obj);
1499
}
1500

1501
/**
1502
 * qemudReconnectDomains
1503 1504 1505 1506 1507
 *
 * Try to re-open the resources for live VMs that we care
 * about.
 */
static void
1508
qemuReconnectDomains(virConnectPtr conn, struct qemud_driver *driver)
1509
{
1510 1511
    struct virReconnectDomainData data = {conn, driver};
    virHashForEach(driver->domains.objs, qemuReconnectDomain, &data);
1512 1513
}

1514

1515 1516 1517 1518 1519 1520
static int
qemudSecurityInit(struct qemud_driver *qemud_drv)
{
    int ret;
    virSecurityDriverPtr security_drv;

1521 1522 1523
    qemuSecurityStackedSetDriver(qemud_drv);
    qemuSecurityDACSetDriver(qemud_drv);

1524
    ret = virSecurityDriverStartup(&security_drv,
1525 1526
                                   qemud_drv->securityDriverName,
                                   qemud_drv->allowDiskFormatProbing);
1527 1528 1529 1530
    if (ret == -1) {
        VIR_ERROR0(_("Failed to start security driver"));
        return -1;
    }
1531 1532 1533

    /* No primary security driver wanted to be enabled: just setup
     * the DAC driver on its own */
1534
    if (ret == -2) {
1535
        qemud_drv->securityDriver = &qemuDACSecurityDriver;
1536
        VIR_INFO0(_("No security driver available"));
1537 1538 1539 1540 1541
    } else {
        qemud_drv->securityPrimaryDriver = security_drv;
        qemud_drv->securitySecondaryDriver = &qemuDACSecurityDriver;
        qemud_drv->securityDriver = &qemuStackedSecurityDriver;
        VIR_INFO("Initialized security driver %s", security_drv->name);
1542 1543
    }

1544
    return 0;
1545
}
1546 1547


1548 1549
static virCapsPtr
qemuCreateCapabilities(virCapsPtr oldcaps,
1550
                       struct qemud_driver *driver)
1551 1552 1553 1554 1555
{
    virCapsPtr caps;

    /* Basic host arch / guest machine capabilities */
    if (!(caps = qemudCapsInit(oldcaps))) {
1556
        virReportOOMError();
1557 1558 1559
        return NULL;
    }

1560 1561 1562 1563 1564 1565 1566 1567
    if (driver->allowDiskFormatProbing) {
        caps->defaultDiskDriverName = NULL;
        caps->defaultDiskDriverType = NULL;
    } else {
        caps->defaultDiskDriverName = "qemu";
        caps->defaultDiskDriverType = "raw";
    }

1568 1569 1570 1571 1572 1573
    /* Domain XML parser hooks */
    caps->privateDataAllocFunc = qemuDomainObjPrivateAlloc;
    caps->privateDataFreeFunc = qemuDomainObjPrivateFree;
    caps->privateDataXMLFormat = qemuDomainObjPrivateXMLFormat;
    caps->privateDataXMLParse = qemuDomainObjPrivateXMLParse;

1574 1575 1576 1577 1578
    if (virGetHostUUID(caps->host.host_uuid)) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                         "%s", _("cannot get the host uuid"));
        goto err_exit;
    }
1579

1580 1581 1582 1583 1584 1585
    /* Domain Namespace XML parser hooks */
    caps->ns.parse = qemuDomainDefNamespaceParse;
    caps->ns.free = qemuDomainDefNamespaceFree;
    caps->ns.format = qemuDomainDefNamespaceFormatXML;
    caps->ns.href = qemuDomainDefNamespaceHref;

1586
    /* Security driver data */
1587
    if (driver->securityPrimaryDriver) {
1588 1589
        const char *doi, *model;

1590 1591
        doi = virSecurityDriverGetDOI(driver->securityPrimaryDriver);
        model = virSecurityDriverGetModel(driver->securityPrimaryDriver);
1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604

        if (!(caps->host.secModel.model = strdup(model)))
            goto no_memory;
        if (!(caps->host.secModel.doi = strdup(doi)))
            goto no_memory;

        VIR_DEBUG("Initialized caps for security driver \"%s\" with "
                  "DOI \"%s\"", model, doi);
    }

    return caps;

no_memory:
1605
    virReportOOMError();
1606
err_exit:
1607 1608 1609
    virCapabilitiesFree(caps);
    return NULL;
}
1610

C
Chris Lalancette 已提交
1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627
static void qemuDomainSnapshotLoad(void *payload,
                                   const char *name ATTRIBUTE_UNUSED,
                                   void *data)
{
    virDomainObjPtr vm = (virDomainObjPtr)payload;
    char *baseDir = (char *)data;
    char *snapDir = NULL;
    DIR *dir = NULL;
    struct dirent *entry;
    char *xmlStr;
    int ret;
    char *fullpath;
    virDomainSnapshotDefPtr def = NULL;
    char ebuf[1024];

    virDomainObjLock(vm);
    if (virAsprintf(&snapDir, "%s/%s", baseDir, vm->def->name) < 0) {
1628
        VIR_ERROR(_("Failed to allocate memory for snapshot directory for domain %s"),
C
Chris Lalancette 已提交
1629 1630 1631 1632 1633 1634 1635 1636 1637
                   vm->def->name);
        goto cleanup;
    }

    VIR_INFO("Scanning for snapshots for domain %s in %s", vm->def->name,
             snapDir);

    if (!(dir = opendir(snapDir))) {
        if (errno != ENOENT)
1638
            VIR_ERROR(_("Failed to open snapshot directory %s for domain %s: %s"),
C
Chris Lalancette 已提交
1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652
                      snapDir, vm->def->name,
                      virStrerror(errno, ebuf, sizeof(ebuf)));
        goto cleanup;
    }

    while ((entry = readdir(dir))) {
        if (entry->d_name[0] == '.')
            continue;

        /* NB: ignoring errors, so one malformed config doesn't
           kill the whole process */
        VIR_INFO("Loading snapshot file '%s'", entry->d_name);

        if (virAsprintf(&fullpath, "%s/%s", snapDir, entry->d_name) < 0) {
1653
            VIR_ERROR0(_("Failed to allocate memory for path"));
C
Chris Lalancette 已提交
1654 1655 1656 1657 1658 1659
            continue;
        }

        ret = virFileReadAll(fullpath, 1024*1024*1, &xmlStr);
        if (ret < 0) {
            /* Nothing we can do here, skip this one */
1660
            VIR_ERROR(_("Failed to read snapshot file %s: %s"), fullpath,
C
Chris Lalancette 已提交
1661
                      virStrerror(errno, ebuf, sizeof(ebuf)));
1662
            VIR_FREE(fullpath);
C
Chris Lalancette 已提交
1663 1664 1665 1666 1667 1668
            continue;
        }

        def = virDomainSnapshotDefParseString(xmlStr, 0);
        if (def == NULL) {
            /* Nothing we can do here, skip this one */
1669
            VIR_ERROR(_("Failed to parse snapshot XML from file '%s'"), fullpath);
1670
            VIR_FREE(fullpath);
C
Chris Lalancette 已提交
1671 1672 1673 1674
            VIR_FREE(xmlStr);
            continue;
        }

1675
        virDomainSnapshotAssignDef(&vm->snapshots, def);
C
Chris Lalancette 已提交
1676

1677
        VIR_FREE(fullpath);
C
Chris Lalancette 已提交
1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698
        VIR_FREE(xmlStr);
    }

    /* FIXME: qemu keeps internal track of snapshots.  We can get access
     * to this info via the "info snapshots" monitor command for running
     * domains, or via "qemu-img snapshot -l" for shutoff domains.  It would
     * be nice to update our internal state based on that, but there is a
     * a problem.  qemu doesn't track all of the same metadata that we do.
     * In particular we wouldn't be able to fill in the <parent>, which is
     * pretty important in our metadata.
     */

    virResetLastError();

cleanup:
    if (dir)
        closedir(dir);
    VIR_FREE(snapDir);
    virDomainObjUnlock(vm);
}

1699 1700 1701 1702 1703 1704
/**
 * qemudStartup:
 *
 * Initialization function for the QEmu daemon
 */
static int
1705
qemudStartup(int privileged) {
1706
    char *base = NULL;
D
Daniel P. Berrange 已提交
1707
    char driverConf[PATH_MAX];
1708
    int rc;
1709
    virConnectPtr conn = NULL;
1710

1711
    if (VIR_ALLOC(qemu_driver) < 0)
1712 1713
        return -1;

1714
    if (virMutexInit(&qemu_driver->lock) < 0) {
1715
        VIR_ERROR0(_("cannot initialize mutex"));
1716 1717 1718
        VIR_FREE(qemu_driver);
        return -1;
    }
1719
    qemuDriverLock(qemu_driver);
1720
    qemu_driver->privileged = privileged;
1721

1722 1723 1724
    /* Don't have a dom0 so start from 1 */
    qemu_driver->nextvmid = 1;

1725 1726 1727
    if (virDomainObjListInit(&qemu_driver->domains) < 0)
        goto out_of_memory;

1728
    /* Init callback list */
1729
    if (VIR_ALLOC(qemu_driver->domainEventCallbacks) < 0)
1730
        goto out_of_memory;
1731 1732 1733 1734 1735 1736
    if (!(qemu_driver->domainEventQueue = virDomainEventQueueNew()))
        goto out_of_memory;

    if ((qemu_driver->domainEventTimer =
         virEventAddTimeout(-1, qemuDomainEventFlush, qemu_driver, NULL)) < 0)
        goto error;
1737

1738 1739 1740 1741 1742
    /* Allocate bitmap for vnc port reservation */
    if ((qemu_driver->reservedVNCPorts =
         virBitmapAlloc(QEMU_VNC_PORT_MAX - QEMU_VNC_PORT_MIN)) == NULL)
        goto out_of_memory;

1743
    if (privileged) {
1744 1745
        if (virAsprintf(&qemu_driver->logDir,
                        "%s/log/libvirt/qemu", LOCAL_STATE_DIR) == -1)
1746
            goto out_of_memory;
1747

D
Daniel P. Berrange 已提交
1748
        if ((base = strdup (SYSCONF_DIR "/libvirt")) == NULL)
1749
            goto out_of_memory;
1750 1751

        if (virAsprintf(&qemu_driver->stateDir,
1752
                      "%s/run/libvirt/qemu", LOCAL_STATE_DIR) == -1)
1753
            goto out_of_memory;
1754 1755 1756 1757 1758 1759 1760 1761

        if (virAsprintf(&qemu_driver->libDir,
                      "%s/lib/libvirt/qemu", LOCAL_STATE_DIR) == -1)
            goto out_of_memory;

        if (virAsprintf(&qemu_driver->cacheDir,
                      "%s/cache/libvirt/qemu", LOCAL_STATE_DIR) == -1)
            goto out_of_memory;
1762 1763 1764
        if (virAsprintf(&qemu_driver->saveDir,
                      "%s/lib/libvirt/qemu/save/", LOCAL_STATE_DIR) == -1)
            goto out_of_memory;
C
Chris Lalancette 已提交
1765 1766 1767
        if (virAsprintf(&qemu_driver->snapshotDir,
                        "%s/lib/libvirt/qemu/snapshot", LOCAL_STATE_DIR) == -1)
            goto out_of_memory;
1768
    } else {
1769
        uid_t uid = geteuid();
1770
        char *userdir = virGetUserDirectory(uid);
1771
        if (!userdir)
1772
            goto error;
1773

1774
        if (virAsprintf(&qemu_driver->logDir,
1775 1776
                        "%s/.libvirt/qemu/log", userdir) == -1) {
            VIR_FREE(userdir);
1777
            goto out_of_memory;
1778
        }
1779

1780 1781
        if (virAsprintf(&base, "%s/.libvirt", userdir) == -1) {
            VIR_FREE(userdir);
1782
            goto out_of_memory;
1783 1784
        }
        VIR_FREE(userdir);
1785 1786 1787

        if (virAsprintf(&qemu_driver->stateDir, "%s/qemu/run", base) == -1)
            goto out_of_memory;
1788 1789 1790 1791
        if (virAsprintf(&qemu_driver->libDir, "%s/qemu/lib", base) == -1)
            goto out_of_memory;
        if (virAsprintf(&qemu_driver->cacheDir, "%s/qemu/cache", base) == -1)
            goto out_of_memory;
1792 1793
        if (virAsprintf(&qemu_driver->saveDir, "%s/qemu/save", base) == -1)
            goto out_of_memory;
C
Chris Lalancette 已提交
1794 1795
        if (virAsprintf(&qemu_driver->snapshotDir, "%s/qemu/snapshot", base) == -1)
            goto out_of_memory;
1796 1797
    }

L
Laine Stump 已提交
1798
    if (virFileMakePath(qemu_driver->stateDir) != 0) {
1799
        char ebuf[1024];
1800
        VIR_ERROR(_("Failed to create state dir '%s': %s"),
1801
                  qemu_driver->stateDir, virStrerror(errno, ebuf, sizeof ebuf));
1802
        goto error;
1803
    }
L
Laine Stump 已提交
1804
    if (virFileMakePath(qemu_driver->libDir) != 0) {
1805
        char ebuf[1024];
1806
        VIR_ERROR(_("Failed to create lib dir '%s': %s"),
1807 1808 1809
                  qemu_driver->libDir, virStrerror(errno, ebuf, sizeof ebuf));
        goto error;
    }
L
Laine Stump 已提交
1810
    if (virFileMakePath(qemu_driver->cacheDir) != 0) {
1811
        char ebuf[1024];
1812
        VIR_ERROR(_("Failed to create cache dir '%s': %s"),
1813 1814 1815
                  qemu_driver->cacheDir, virStrerror(errno, ebuf, sizeof ebuf));
        goto error;
    }
1816 1817 1818 1819 1820 1821
    if (virFileMakePath(qemu_driver->saveDir) != 0) {
        char ebuf[1024];
        VIR_ERROR(_("Failed to create save dir '%s': %s"),
                  qemu_driver->saveDir, virStrerror(errno, ebuf, sizeof ebuf));
        goto error;
    }
C
Chris Lalancette 已提交
1822 1823 1824 1825 1826 1827
    if (virFileMakePath(qemu_driver->snapshotDir) != 0) {
        char ebuf[1024];
        VIR_ERROR(_("Failed to create save dir '%s': %s"),
                  qemu_driver->snapshotDir, virStrerror(errno, ebuf, sizeof ebuf));
        goto error;
    }
1828 1829 1830 1831

    /* Configuration paths are either ~/.libvirt/qemu/... (session) or
     * /etc/libvirt/qemu/... (system).
     */
D
Daniel P. Berrange 已提交
1832
    if (snprintf (driverConf, sizeof(driverConf), "%s/qemu.conf", base) == -1)
1833
        goto out_of_memory;
D
Daniel P. Berrange 已提交
1834
    driverConf[sizeof(driverConf)-1] = '\0';
1835

1836
    if (virAsprintf(&qemu_driver->configDir, "%s/qemu", base) == -1)
1837 1838
        goto out_of_memory;

1839
    if (virAsprintf(&qemu_driver->autostartDir, "%s/qemu/autostart", base) == -1)
1840 1841
        goto out_of_memory;

1842
    VIR_FREE(base);
1843

1844 1845 1846 1847 1848 1849 1850
    rc = virCgroupForDriver("qemu", &qemu_driver->cgroup, privileged, 1);
    if (rc < 0) {
        char buf[1024];
        VIR_WARN("Unable to create cgroup for driver: %s",
                 virStrerror(-rc, buf, sizeof(buf)));
    }

1851 1852 1853 1854
    if (qemudLoadDriverConfig(qemu_driver, driverConf) < 0) {
        goto error;
    }

1855 1856
    if (qemudSecurityInit(qemu_driver) < 0)
        goto error;
D
Daniel P. Berrange 已提交
1857

1858
    if ((qemu_driver->caps = qemuCreateCapabilities(NULL,
1859
                                                    qemu_driver)) == NULL)
1860
        goto error;
1861

1862
    if ((qemu_driver->activePciHostdevs = pciDeviceListNew()) == NULL)
1863 1864
        goto error;

1865 1866
    if (privileged) {
        if (chown(qemu_driver->libDir, qemu_driver->user, qemu_driver->group) < 0) {
1867
            virReportSystemError(errno,
1868 1869 1870 1871 1872
                                 _("unable to set ownership of '%s' to user %d:%d"),
                                 qemu_driver->libDir, qemu_driver->user, qemu_driver->group);
            goto error;
        }
        if (chown(qemu_driver->cacheDir, qemu_driver->user, qemu_driver->group) < 0) {
1873
            virReportSystemError(errno,
1874 1875 1876 1877
                                 _("unable to set ownership of '%s' to %d:%d"),
                                 qemu_driver->cacheDir, qemu_driver->user, qemu_driver->group);
            goto error;
        }
1878 1879 1880 1881 1882 1883
        if (chown(qemu_driver->saveDir, qemu_driver->user, qemu_driver->group) < 0) {
            virReportSystemError(errno,
                                 _("unable to set ownership of '%s' to %d:%d"),
                                 qemu_driver->saveDir, qemu_driver->user, qemu_driver->group);
            goto error;
        }
C
Chris Lalancette 已提交
1884 1885 1886 1887 1888 1889
        if (chown(qemu_driver->snapshotDir, qemu_driver->user, qemu_driver->group) < 0) {
            virReportSystemError(errno,
                                 _("unable to set ownership of '%s' to %d:%d"),
                                 qemu_driver->snapshotDir, qemu_driver->user, qemu_driver->group);
            goto error;
        }
1890 1891
    }

1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905
    /* If hugetlbfs is present, then we need to create a sub-directory within
     * it, since we can't assume the root mount point has permissions that
     * will let our spawned QEMU instances use it.
     *
     * NB the check for '/', since user may config "" to disable hugepages
     * even when mounted
     */
    if (qemu_driver->hugetlbfs_mount &&
        qemu_driver->hugetlbfs_mount[0] == '/') {
        char *mempath = NULL;
        if (virAsprintf(&mempath, "%s/libvirt/qemu", qemu_driver->hugetlbfs_mount) < 0)
            goto out_of_memory;

        if ((rc = virFileMakePath(mempath)) != 0) {
1906
            virReportSystemError(rc,
1907 1908 1909 1910 1911 1912
                                 _("unable to create hugepage path %s"), mempath);
            VIR_FREE(mempath);
            goto error;
        }
        if (qemu_driver->privileged &&
            chown(mempath, qemu_driver->user, qemu_driver->group) < 0) {
1913
            virReportSystemError(errno,
1914 1915 1916 1917 1918 1919 1920 1921 1922
                                 _("unable to set ownership on %s to %d:%d"),
                                 mempath, qemu_driver->user, qemu_driver->group);
            VIR_FREE(mempath);
            goto error;
        }

        qemu_driver->hugepage_path = mempath;
    }

1923
    /* Get all the running persistent or transient configs first */
1924
    if (virDomainLoadAllConfigs(qemu_driver->caps,
1925 1926 1927 1928 1929 1930
                                &qemu_driver->domains,
                                qemu_driver->stateDir,
                                NULL,
                                1, NULL, NULL) < 0)
        goto error;

1931 1932 1933 1934 1935
    conn = virConnectOpen(qemu_driver->privileged ?
                          "qemu:///system" :
                          "qemu:///session");

    qemuReconnectDomains(conn, qemu_driver);
1936 1937

    /* Then inactive persistent configs */
1938
    if (virDomainLoadAllConfigs(qemu_driver->caps,
1939 1940
                                &qemu_driver->domains,
                                qemu_driver->configDir,
1941
                                qemu_driver->autostartDir,
1942
                                0, NULL, NULL) < 0)
1943
        goto error;
C
Chris Lalancette 已提交
1944 1945 1946 1947 1948


    virHashForEach(qemu_driver->domains.objs, qemuDomainSnapshotLoad,
                   qemu_driver->snapshotDir);

1949 1950
    qemuDriverUnlock(qemu_driver);

1951 1952
    qemudAutostartConfigs(qemu_driver);

1953 1954
    if (conn)
        virConnectClose(conn);
1955

1956 1957
    return 0;

1958
out_of_memory:
1959
    virReportOOMError();
1960 1961 1962
error:
    if (qemu_driver)
        qemuDriverUnlock(qemu_driver);
1963 1964
    if (conn)
        virConnectClose(conn);
1965
    VIR_FREE(base);
1966
    qemudShutdown();
1967 1968 1969
    return -1;
}

1970 1971 1972 1973
static void qemudNotifyLoadDomain(virDomainObjPtr vm, int newVM, void *opaque)
{
    struct qemud_driver *driver = opaque;

1974 1975 1976 1977 1978 1979 1980 1981
    if (newVM) {
        virDomainEventPtr event =
            virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_DEFINED,
                                     VIR_DOMAIN_EVENT_DEFINED_ADDED);
        if (event)
            qemuDomainEventQueue(driver, event);
    }
1982 1983
}

1984 1985 1986 1987 1988 1989 1990 1991
/**
 * qemudReload:
 *
 * Function to restart the QEmu daemon, it will recheck the configuration
 * files and update its state and the networking
 */
static int
qemudReload(void) {
1992 1993 1994
    if (!qemu_driver)
        return 0;

1995
    qemuDriverLock(qemu_driver);
1996
    virDomainLoadAllConfigs(qemu_driver->caps,
1997 1998
                            &qemu_driver->domains,
                            qemu_driver->configDir,
1999
                            qemu_driver->autostartDir,
2000
                            0, qemudNotifyLoadDomain, qemu_driver);
2001
    qemuDriverUnlock(qemu_driver);
2002

2003
    qemudAutostartConfigs(qemu_driver);
2004 2005

    return 0;
2006 2007
}

2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
/**
 * qemudActive:
 *
 * Checks if the QEmu daemon is active, i.e. has an active domain or
 * an active network
 *
 * Returns 1 if active, 0 otherwise
 */
static int
qemudActive(void) {
2018
    int active = 0;
2019

2020 2021 2022
    if (!qemu_driver)
        return 0;

2023
    /* XXX having to iterate here is not great because it requires many locks */
2024
    qemuDriverLock(qemu_driver);
2025
    active = virDomainObjListNumOfDomains(&qemu_driver->domains, 1);
2026 2027
    qemuDriverUnlock(qemu_driver);
    return active;
2028 2029
}

2030 2031 2032 2033 2034 2035 2036
/**
 * qemudShutdown:
 *
 * Shutdown the QEmu daemon, it will stop all active domains and networks
 */
static int
qemudShutdown(void) {
2037
    int i;
2038

2039
    if (!qemu_driver)
2040
        return -1;
2041

2042
    qemuDriverLock(qemu_driver);
2043
    pciDeviceListFree(qemu_driver->activePciHostdevs);
2044 2045
    virCapabilitiesFree(qemu_driver->caps);

2046
    virDomainObjListDeinit(&qemu_driver->domains);
2047
    virBitmapFree(qemu_driver->reservedVNCPorts);
2048

2049
    VIR_FREE(qemu_driver->securityDriverName);
2050
    VIR_FREE(qemu_driver->logDir);
2051 2052
    VIR_FREE(qemu_driver->configDir);
    VIR_FREE(qemu_driver->autostartDir);
2053
    VIR_FREE(qemu_driver->stateDir);
2054 2055
    VIR_FREE(qemu_driver->libDir);
    VIR_FREE(qemu_driver->cacheDir);
2056
    VIR_FREE(qemu_driver->saveDir);
C
Chris Lalancette 已提交
2057
    VIR_FREE(qemu_driver->snapshotDir);
2058
    VIR_FREE(qemu_driver->vncTLSx509certdir);
J
Jim Meyering 已提交
2059
    VIR_FREE(qemu_driver->vncListen);
2060
    VIR_FREE(qemu_driver->vncPassword);
2061
    VIR_FREE(qemu_driver->vncSASLdir);
2062
    VIR_FREE(qemu_driver->saveImageFormat);
2063 2064
    VIR_FREE(qemu_driver->hugetlbfs_mount);
    VIR_FREE(qemu_driver->hugepage_path);
D
Daniel P. Berrange 已提交
2065

2066 2067 2068 2069 2070 2071
    if (qemu_driver->cgroupDeviceACL) {
        for (i = 0 ; qemu_driver->cgroupDeviceACL[i] != NULL ; i++)
            VIR_FREE(qemu_driver->cgroupDeviceACL[i]);
        VIR_FREE(qemu_driver->cgroupDeviceACL);
    }

2072 2073
    /* Free domain callback list */
    virDomainEventCallbackListFree(qemu_driver->domainEventCallbacks);
2074 2075 2076 2077
    virDomainEventQueueFree(qemu_driver->domainEventQueue);

    if (qemu_driver->domainEventTimer != -1)
        virEventRemoveTimeout(qemu_driver->domainEventTimer);
2078

2079 2080 2081
    if (qemu_driver->brctl)
        brShutdown(qemu_driver->brctl);

2082 2083
    virCgroupFree(&qemu_driver->cgroup);

2084
    qemuDriverUnlock(qemu_driver);
2085
    virMutexDestroy(&qemu_driver->lock);
2086
    VIR_FREE(qemu_driver);
2087 2088

    return 0;
2089 2090
}

2091
typedef int qemuLogHandleOutput(virDomainObjPtr vm,
2092 2093
                                const char *output,
                                int fd);
2094 2095 2096 2097 2098

/*
 * Returns -1 for error, 0 on success
 */
static int
2099
qemudReadLogOutput(virDomainObjPtr vm,
2100 2101
                   int fd,
                   char *buf,
G
Guido Günther 已提交
2102
                   size_t buflen,
2103
                   qemuLogHandleOutput func,
2104 2105 2106
                   const char *what,
                   int timeout)
{
2107
    int retries = (timeout*10);
2108
    int got = 0;
2109 2110 2111
    buf[0] = '\0';

    while (retries) {
2112
        ssize_t func_ret, ret;
2113
        int isdead = 0;
G
Guido Günther 已提交
2114

2115
        func_ret = func(vm, buf, fd);
2116

2117 2118
        if (kill(vm->pid, 0) == -1 && errno == ESRCH)
            isdead = 1;
2119

2120 2121
        /* Any failures should be detected before we read the log, so we
         * always have something useful to report on failure. */
2122 2123
        ret = saferead(fd, buf+got, buflen-got-1);
        if (ret < 0) {
2124
            virReportSystemError(errno,
2125 2126 2127 2128 2129
                                 _("Failure while reading %s log output"),
                                 what);
            return -1;
        }

2130 2131 2132
        got += ret;
        buf[got] = '\0';
        if (got == buflen-1) {
2133
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
2134 2135
                            _("Out of space while reading %s log output: %s"),
                            what, buf);
2136 2137 2138 2139
            return -1;
        }

        if (isdead) {
2140
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
2141 2142
                            _("Process exited while reading %s log output: %s"),
                            what, buf);
2143 2144 2145
            return -1;
        }

2146 2147
        if (func_ret <= 0)
            return func_ret;
2148 2149 2150 2151

        usleep(100*1000);
        retries--;
    }
2152

2153
    qemuReportError(VIR_ERR_INTERNAL_ERROR,
2154 2155
                    _("Timed out while reading %s log output: %s"),
                    what, buf);
2156 2157 2158
    return -1;
}

2159

2160 2161 2162 2163 2164 2165 2166 2167
/*
 * Look at a chunk of data from the QEMU stdout logs and try to
 * find a TTY device, as indicated by a line like
 *
 * char device redirected to /dev/pts/3
 *
 * Returns -1 for error, 0 success, 1 continue reading
 */
2168
static int
2169
qemudExtractTTYPath(const char *haystack,
2170 2171
                    size_t *offset,
                    char **path)
2172
{
2173
    static const char needle[] = "char device redirected to";
2174
    char *tmp, *dev;
2175

2176
    VIR_FREE(*path);
2177
    /* First look for our magic string */
2178 2179 2180 2181 2182
    if (!(tmp = strstr(haystack + *offset, needle))) {
        return 1;
    }
    tmp += sizeof(needle);
    dev = tmp;
2183

2184 2185 2186 2187 2188
    /*
     * And look for first whitespace character and nul terminate
     * to mark end of the pty path
     */
    while (*tmp) {
2189
        if (c_isspace(*tmp)) {
2190 2191
            *path = strndup(dev, tmp-dev);
            if (*path == NULL) {
2192
                virReportOOMError();
2193 2194
                return -1;
            }
2195

2196
            /* ... now further update offset till we get EOL */
2197
            *offset = tmp - haystack;
2198 2199
            return 0;
        }
2200
        tmp++;
2201 2202 2203 2204 2205
    }

    /*
     * We found a path, but didn't find any whitespace,
     * so it must be still incomplete - we should at
2206 2207
     * least see a \n - indicate that we want to carry
     * on trying again
2208
     */
2209
    return 1;
2210 2211
}

2212
static int
2213
qemudFindCharDevicePTYsMonitor(virDomainObjPtr vm,
2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228
                               virHashTablePtr paths)
{
    int i;

#define LOOKUP_PTYS(array, arraylen, idprefix)                            \
    for (i = 0 ; i < (arraylen) ; i++) {                                  \
        virDomainChrDefPtr chr = (array)[i];                              \
        if (chr->type == VIR_DOMAIN_CHR_TYPE_PTY) {                       \
            char id[16];                                                  \
                                                                          \
            if (snprintf(id, sizeof(id), idprefix "%i", i) >= sizeof(id)) \
                return -1;                                                \
                                                                          \
            const char *path = (const char *) virHashLookup(paths, id);   \
            if (path == NULL) {                                           \
2229 2230 2231
                if (chr->data.file.path == NULL) {                        \
                    /* neither the log output nor 'info chardev' had a */ \
                    /* pty path for this chardev, report an error */      \
2232 2233 2234
                    qemuReportError(VIR_ERR_INTERNAL_ERROR,               \
                                    _("no assigned pty for device %s"), id); \
                    return -1;                                            \
2235 2236 2237 2238 2239
                } else {                                                  \
                    /* 'info chardev' had no pty path for this chardev, */\
                    /* but the log output had, so we're fine */           \
                    continue;                                             \
                }                                                         \
2240 2241
            }                                                             \
                                                                          \
2242
            VIR_FREE(chr->data.file.path);                                \
2243
            chr->data.file.path = strdup(path);                           \
2244 2245
                                                                          \
            if (chr->data.file.path == NULL) {                            \
2246
                virReportOOMError();                                      \
2247 2248
                return -1;                                                \
            }                                                             \
2249 2250 2251 2252 2253 2254
        }                                                                 \
    }

    LOOKUP_PTYS(vm->def->serials,   vm->def->nserials,   "serial");
    LOOKUP_PTYS(vm->def->parallels, vm->def->nparallels, "parallel");
    LOOKUP_PTYS(vm->def->channels,  vm->def->nchannels,  "channel");
C
Cole Robinson 已提交
2255 2256
    if (vm->def->console)
        LOOKUP_PTYS(&vm->def->console, 1,  "console");
2257
#undef LOOKUP_PTYS
2258 2259 2260 2261

    return 0;
}

2262
static int
2263
qemudFindCharDevicePTYs(virDomainObjPtr vm,
2264 2265
                        const char *output,
                        int fd ATTRIBUTE_UNUSED)
2266
{
2267
    size_t offset = 0;
2268
    int ret, i;
2269 2270

    /* The order in which QEMU prints out the PTY paths is
2271 2272
       the order in which it procsses its serial and parallel
       device args. This code must match that ordering.... */
2273

2274
    /* first comes the serial devices */
2275 2276
    for (i = 0 ; i < vm->def->nserials ; i++) {
        virDomainChrDefPtr chr = vm->def->serials[i];
2277
        if (chr->type == VIR_DOMAIN_CHR_TYPE_PTY) {
2278
            if ((ret = qemudExtractTTYPath(output, &offset,
2279
                                           &chr->data.file.path)) != 0)
2280
                return ret;
2281 2282 2283
        }
    }

2284
    /* then the parallel devices */
2285 2286
    for (i = 0 ; i < vm->def->nparallels ; i++) {
        virDomainChrDefPtr chr = vm->def->parallels[i];
2287
        if (chr->type == VIR_DOMAIN_CHR_TYPE_PTY) {
2288
            if ((ret = qemudExtractTTYPath(output, &offset,
2289
                                           &chr->data.file.path)) != 0)
2290
                return ret;
2291 2292 2293
        }
    }

2294 2295 2296 2297
    /* then the channel devices */
    for (i = 0 ; i < vm->def->nchannels ; i++) {
        virDomainChrDefPtr chr = vm->def->channels[i];
        if (chr->type == VIR_DOMAIN_CHR_TYPE_PTY) {
2298
            if ((ret = qemudExtractTTYPath(output, &offset,
2299 2300 2301 2302 2303
                                           &chr->data.file.path)) != 0)
                return ret;
        }
    }

2304
    return 0;
2305 2306
}

2307 2308 2309 2310 2311
static void qemudFreePtyPath(void *payload, const char *name ATTRIBUTE_UNUSED)
{
    VIR_FREE(payload);
}

2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325
static void
qemuReadLogFD(int logfd, char *buf, int maxlen, int off)
{
    int ret;
    char *tmpbuf = buf + off;

    ret = saferead(logfd, tmpbuf, maxlen - off - 1);
    if (ret < 0) {
        ret = 0;
    }

    tmpbuf[ret] = '\0';
}

2326
static int
2327
qemudWaitForMonitor(struct qemud_driver* driver,
2328
                    virDomainObjPtr vm, off_t pos)
2329
{
2330
    char buf[4096] = ""; /* Plenty of space to get startup greeting */
2331
    int logfd;
2332
    int ret = -1;
2333
    virHashTablePtr paths = NULL;
2334

2335
    if ((logfd = qemudLogReadFD(driver->logDir, vm->def->name, pos)) < 0)
2336
        return -1;
2337

2338 2339 2340 2341
    if (qemudReadLogOutput(vm, logfd, buf, sizeof(buf),
                           qemudFindCharDevicePTYs,
                           "console", 30) < 0)
        goto closelog;
2342

2343
    VIR_DEBUG("Connect monitor to %p '%s'", vm, vm->def->name);
2344 2345 2346
    if (qemuConnectMonitor(driver, vm) < 0) {
        goto cleanup;
    }
2347

2348 2349 2350 2351
    /* Try to get the pty path mappings again via the monitor. This is much more
     * reliable if it's available.
     * Note that the monitor itself can be on a pty, so we still need to try the
     * log output method. */
2352
    paths = virHashCreate(0);
2353
    if (paths == NULL) {
2354
        virReportOOMError();
2355 2356 2357
        goto cleanup;
    }

2358
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
2359 2360
    qemuDomainObjPrivatePtr priv = vm->privateData;
    ret = qemuMonitorGetPtyPaths(priv->mon, paths);
2361
    qemuDomainObjExitMonitorWithDriver(driver, vm);
2362 2363 2364

    VIR_DEBUG("qemuMonitorGetPtyPaths returned %i", ret);
    if (ret == 0) {
2365
        ret = qemudFindCharDevicePTYsMonitor(vm, paths);
2366 2367 2368 2369 2370 2371 2372
    }

cleanup:
    if (paths) {
        virHashFree(paths, qemudFreePtyPath);
    }

2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385
    if (kill(vm->pid, 0) == -1 && errno == ESRCH) {
        /* VM is dead, any other error raised in the interim is probably
         * not as important as the qemu cmdline output */
        qemuReadLogFD(logfd, buf, sizeof(buf), strlen(buf));
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("process exited while connecting to monitor: %s"),
                        buf);
        ret = -1;
    }

closelog:
    if (close(logfd) < 0) {
        char ebuf[4096];
2386
        VIR_WARN("Unable to close logfile: %s",
2387 2388 2389
                 virStrerror(errno, ebuf, sizeof ebuf));
    }

2390
    return ret;
2391 2392
}

2393
static int
2394
qemuDetectVcpuPIDs(struct qemud_driver *driver,
2395 2396 2397
                   virDomainObjPtr vm) {
    pid_t *cpupids = NULL;
    int ncpupids;
2398
    qemuDomainObjPrivatePtr priv = vm->privateData;
2399

2400
    if (vm->def->virtType != VIR_DOMAIN_VIRT_KVM) {
2401 2402
        priv->nvcpupids = 1;
        if (VIR_ALLOC_N(priv->vcpupids, priv->nvcpupids) < 0) {
2403
            virReportOOMError();
2404 2405
            return -1;
        }
2406
        priv->vcpupids[0] = vm->pid;
2407 2408 2409
        return 0;
    }

2410
    /* What follows is now all KVM specific */
2411

2412
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
2413
    if ((ncpupids = qemuMonitorGetCPUInfo(priv->mon, &cpupids)) < 0) {
2414
        qemuDomainObjExitMonitorWithDriver(driver, vm);
2415
        return -1;
2416
    }
2417
    qemuDomainObjExitMonitorWithDriver(driver, vm);
2418

2419 2420 2421
    /* Treat failure to get VCPU<->PID mapping as non-fatal */
    if (ncpupids == 0)
        return 0;
2422

2423
    if (ncpupids != vm->def->vcpus) {
2424 2425 2426
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("got wrong number of vCPU pids from QEMU monitor. got %d, wanted %d"),
                        ncpupids, (int)vm->def->vcpus);
2427 2428 2429
        VIR_FREE(cpupids);
        return -1;
    }
2430

2431 2432
    priv->nvcpupids = ncpupids;
    priv->vcpupids = cpupids;
2433 2434 2435
    return 0;
}

2436 2437 2438
/*
 * To be run between fork/exec of QEMU only
 */
2439
static int
2440 2441
qemudInitCpuAffinity(virDomainObjPtr vm)
{
2442
    int i, hostcpus, maxcpu = QEMUD_CPUMASK_LEN;
2443
    virNodeInfo nodeinfo;
2444 2445
    unsigned char *cpumap;
    int cpumaplen;
2446 2447

    DEBUG0("Setting CPU affinity");
2448

2449
    if (nodeGetInfo(NULL, &nodeinfo) < 0)
2450 2451 2452 2453
        return -1;

    /* setaffinity fails if you set bits for CPUs which
     * aren't present, so we have to limit ourselves */
2454 2455 2456
    hostcpus = VIR_NODEINFO_MAXCPUS(nodeinfo);
    if (maxcpu > hostcpus)
        maxcpu = hostcpus;
2457

2458 2459
    cpumaplen = VIR_CPU_MAPLEN(maxcpu);
    if (VIR_ALLOC_N(cpumap, cpumaplen) < 0) {
2460
        virReportOOMError();
2461 2462 2463
        return -1;
    }

D
Daniel P. Berrange 已提交
2464
    if (vm->def->cpumask) {
2465 2466 2467
        /* XXX why don't we keep 'cpumask' in the libvirt cpumap
         * format to start with ?!?! */
        for (i = 0 ; i < maxcpu && i < vm->def->cpumasklen ; i++)
D
Daniel P. Berrange 已提交
2468
            if (vm->def->cpumask[i])
2469
                VIR_USE_CPU(cpumap, i);
D
Daniel P. Berrange 已提交
2470
    } else {
2471 2472 2473 2474
        /* You may think this is redundant, but we can't assume libvirtd
         * itself is running on all pCPUs, so we need to explicitly set
         * the spawned QEMU instance to all pCPUs if no map is given in
         * its config file */
D
Daniel P. Berrange 已提交
2475
        for (i = 0 ; i < maxcpu ; i++)
2476
            VIR_USE_CPU(cpumap, i);
D
Daniel P. Berrange 已提交
2477
    }
2478

2479 2480 2481 2482 2483 2484 2485 2486
    /* We are pressuming we are running between fork/exec of QEMU
     * so use '0' to indicate our own process ID. No threads are
     * running at this point
     */
    if (virProcessInfoSetAffinity(0, /* Self */
                                  cpumap, cpumaplen, maxcpu) < 0) {
        VIR_FREE(cpumap);
        return -1;
2487
    }
2488
    VIR_FREE(cpumap);
2489 2490 2491 2492 2493

    return 0;
}


2494
static int
2495 2496 2497 2498
qemuInitPasswords(virConnectPtr conn,
                  struct qemud_driver *driver,
                  virDomainObjPtr vm,
                  unsigned long long qemuCmdFlags) {
2499
    int ret = 0;
2500
    qemuDomainObjPrivatePtr priv = vm->privateData;
2501

2502 2503 2504
    if ((vm->def->ngraphics == 1) &&
        vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
        (vm->def->graphics[0]->data.vnc.passwd || driver->vncPassword)) {
2505

2506
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
2507
        ret = qemuMonitorSetVNCPassword(priv->mon,
2508 2509 2510
                                        vm->def->graphics[0]->data.vnc.passwd ?
                                        vm->def->graphics[0]->data.vnc.passwd :
                                        driver->vncPassword);
2511
        qemuDomainObjExitMonitorWithDriver(driver, vm);
2512 2513
    }

2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536
    if (ret < 0)
        goto cleanup;

    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        int i;

        for (i = 0 ; i < vm->def->ndisks ; i++) {
            char *secret;
            size_t secretLen;

            if (!vm->def->disks[i]->encryption ||
                !vm->def->disks[i]->src)
                continue;

            if (getVolumeQcowPassphrase(conn,
                                        vm->def->disks[i],
                                        &secret, &secretLen) < 0)
                goto cleanup;

            qemuDomainObjEnterMonitorWithDriver(driver, vm);
            ret = qemuMonitorSetDrivePassphrase(priv->mon,
                                                vm->def->disks[i]->info.alias,
                                                secret);
2537
            VIR_FREE(secret);
2538 2539 2540 2541 2542 2543 2544
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            if (ret < 0)
                goto cleanup;
        }
    }

cleanup:
2545
    return ret;
2546 2547 2548
}


2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560
#define QEMU_PCI_VENDOR_INTEL     0x8086
#define QEMU_PCI_VENDOR_LSI_LOGIC 0x1000
#define QEMU_PCI_VENDOR_REDHAT    0x1af4
#define QEMU_PCI_VENDOR_CIRRUS    0x1013
#define QEMU_PCI_VENDOR_REALTEK   0x10ec
#define QEMU_PCI_VENDOR_AMD       0x1022
#define QEMU_PCI_VENDOR_ENSONIQ   0x1274
#define QEMU_PCI_VENDOR_VMWARE    0x15ad
#define QEMU_PCI_VENDOR_QEMU      0x1234

#define QEMU_PCI_PRODUCT_DISK_VIRTIO 0x1001

2561 2562
#define QEMU_PCI_PRODUCT_BALLOON_VIRTIO 0x1002

2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770
#define QEMU_PCI_PRODUCT_NIC_NE2K     0x8029
#define QEMU_PCI_PRODUCT_NIC_PCNET    0x2000
#define QEMU_PCI_PRODUCT_NIC_RTL8139  0x8139
#define QEMU_PCI_PRODUCT_NIC_E1000    0x100E
#define QEMU_PCI_PRODUCT_NIC_VIRTIO   0x1000

#define QEMU_PCI_PRODUCT_VGA_CIRRUS 0x00b8
#define QEMU_PCI_PRODUCT_VGA_VMWARE 0x0405
#define QEMU_PCI_PRODUCT_VGA_STDVGA 0x1111

#define QEMU_PCI_PRODUCT_AUDIO_AC97    0x2415
#define QEMU_PCI_PRODUCT_AUDIO_ES1370  0x5000

#define QEMU_PCI_PRODUCT_CONTROLLER_PIIX 0x7010
#define QEMU_PCI_PRODUCT_CONTROLLER_LSI  0x0012

#define QEMU_PCI_PRODUCT_WATCHDOG_I63000ESB 0x25ab

static int
qemuAssignNextPCIAddress(virDomainDeviceInfo *info,
                         int vendor,
                         int product,
                         qemuMonitorPCIAddress *addrs,
                         int naddrs)
{
    int found = 0;
    int i;

    VIR_DEBUG("Look for %x:%x out of %d", vendor, product, naddrs);

    for (i = 0 ; (i < naddrs) && !found; i++) {
        VIR_DEBUG("Maybe %x:%x", addrs[i].vendor, addrs[i].product);
        if (addrs[i].vendor == vendor &&
            addrs[i].product == product) {
            VIR_DEBUG("Match %d", i);
            found = 1;
            break;
        }
    }
    if (!found) {
        return -1;
    }

    /* Blank it out so this device isn't matched again */
    addrs[i].vendor = 0;
    addrs[i].product = 0;

    if (info->type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE)
        info->type = VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI;

    if (info->type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI) {
        info->addr.pci.domain = addrs[i].addr.domain;
        info->addr.pci.bus = addrs[i].addr.bus;
        info->addr.pci.slot = addrs[i].addr.slot;
        info->addr.pci.function = addrs[i].addr.function;
    }

    return 0;
}

static int
qemuGetPCIDiskVendorProduct(virDomainDiskDefPtr def,
                            unsigned *vendor,
                            unsigned *product)
{
    switch (def->bus) {
    case VIR_DOMAIN_DISK_BUS_VIRTIO:
        *vendor = QEMU_PCI_VENDOR_REDHAT;
        *product = QEMU_PCI_PRODUCT_DISK_VIRTIO;
        break;

    default:
        return -1;
    }

    return 0;
}

static int
qemuGetPCINetVendorProduct(virDomainNetDefPtr def,
                            unsigned *vendor,
                            unsigned *product)
{
    if (!def->model)
        return -1;

    if (STREQ(def->model, "ne2k_pci")) {
        *vendor = QEMU_PCI_VENDOR_REALTEK;
        *product = QEMU_PCI_PRODUCT_NIC_NE2K;
    } else if (STREQ(def->model, "pcnet")) {
        *vendor = QEMU_PCI_VENDOR_AMD;
        *product = QEMU_PCI_PRODUCT_NIC_PCNET;
    } else if (STREQ(def->model, "rtl8139")) {
        *vendor = QEMU_PCI_VENDOR_REALTEK;
        *product = QEMU_PCI_PRODUCT_NIC_RTL8139;
    } else if (STREQ(def->model, "e1000")) {
        *vendor = QEMU_PCI_VENDOR_INTEL;
        *product = QEMU_PCI_PRODUCT_NIC_E1000;
    } else if (STREQ(def->model, "virtio")) {
        *vendor = QEMU_PCI_VENDOR_REDHAT;
        *product = QEMU_PCI_PRODUCT_NIC_VIRTIO;
    } else {
        VIR_INFO("Unexpected NIC model %s, cannot get PCI address",
                 def->model);
        return -1;
    }
    return 0;
}

static int
qemuGetPCIControllerVendorProduct(virDomainControllerDefPtr def,
                                  unsigned *vendor,
                                  unsigned *product)
{
    switch (def->type) {
    case VIR_DOMAIN_CONTROLLER_TYPE_SCSI:
        *vendor = QEMU_PCI_VENDOR_LSI_LOGIC;
        *product = QEMU_PCI_PRODUCT_CONTROLLER_LSI;
        break;

    case VIR_DOMAIN_CONTROLLER_TYPE_FDC:
        /* XXX we could put in the ISA bridge address, but
           that's not technically the FDC's address */
        return -1;

    case VIR_DOMAIN_CONTROLLER_TYPE_IDE:
        *vendor = QEMU_PCI_VENDOR_INTEL;
        *product = QEMU_PCI_PRODUCT_CONTROLLER_PIIX;
        break;

    default:
        VIR_INFO("Unexpected controller type %s, cannot get PCI address",
                 virDomainControllerTypeToString(def->type));
        return -1;
    }

    return 0;
}

static int
qemuGetPCIVideoVendorProduct(virDomainVideoDefPtr def,
                             unsigned *vendor,
                             unsigned *product)
{
    switch (def->type) {
    case VIR_DOMAIN_VIDEO_TYPE_CIRRUS:
        *vendor = QEMU_PCI_VENDOR_CIRRUS;
        *product = QEMU_PCI_PRODUCT_VGA_CIRRUS;
        break;

    case VIR_DOMAIN_VIDEO_TYPE_VGA:
        *vendor = QEMU_PCI_VENDOR_QEMU;
        *product = QEMU_PCI_PRODUCT_VGA_STDVGA;
        break;

    case VIR_DOMAIN_VIDEO_TYPE_VMVGA:
        *vendor = QEMU_PCI_VENDOR_VMWARE;
        *product = QEMU_PCI_PRODUCT_VGA_VMWARE;
        break;

    default:
        return -1;
    }
    return 0;
}

static int
qemuGetPCISoundVendorProduct(virDomainSoundDefPtr def,
                             unsigned *vendor,
                             unsigned *product)
{
    switch (def->model) {
    case VIR_DOMAIN_SOUND_MODEL_ES1370:
        *vendor = QEMU_PCI_VENDOR_ENSONIQ;
        *product = QEMU_PCI_PRODUCT_AUDIO_ES1370;
        break;

    case VIR_DOMAIN_SOUND_MODEL_AC97:
        *vendor = QEMU_PCI_VENDOR_INTEL;
        *product = QEMU_PCI_PRODUCT_AUDIO_AC97;
        break;

    default:
        return -1;
    }

    return 0;
}

static int
qemuGetPCIWatchdogVendorProduct(virDomainWatchdogDefPtr def,
                                unsigned *vendor,
                                unsigned *product)
{
    switch (def->model) {
    case VIR_DOMAIN_WATCHDOG_MODEL_I6300ESB:
        *vendor = QEMU_PCI_VENDOR_INTEL;
        *product = QEMU_PCI_PRODUCT_WATCHDOG_I63000ESB;
        break;

    default:
        return -1;
    }

    return 0;
}


2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789
static int
qemuGetPCIMemballoonVendorProduct(virDomainMemballoonDefPtr def,
                                  unsigned *vendor,
                                  unsigned *product)
{
    switch (def->model) {
    case VIR_DOMAIN_MEMBALLOON_MODEL_VIRTIO:
        *vendor = QEMU_PCI_VENDOR_REDHAT;
        *product = QEMU_PCI_PRODUCT_BALLOON_VIRTIO;
        break;

    default:
        return -1;
    }

    return 0;
}


2790 2791 2792 2793 2794 2795 2796 2797 2798
/*
 * This entire method assumes that PCI devices in 'info pci'
 * match ordering of devices specified on the command line
 * wrt to devices of matching vendor+product
 *
 * XXXX this might not be a valid assumption if we assign
 * some static addrs on CLI. Have to check that...
 */
static int
2799
qemuDetectPCIAddresses(virDomainObjPtr vm,
2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816
                       qemuMonitorPCIAddress *addrs,
                       int naddrs)
{
    unsigned int vendor = 0, product = 0;
    int i;

    /* XXX should all these vendor/product IDs be kept in the
     * actual device data structure instead ?
     */

    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (qemuGetPCIDiskVendorProduct(vm->def->disks[i], &vendor, &product) < 0)
            continue;

        if (qemuAssignNextPCIAddress(&(vm->def->disks[i]->info),
                                     vendor, product,
                                     addrs, naddrs) < 0) {
2817 2818 2819
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for VirtIO disk %s"),
                            vm->def->disks[i]->dst);
2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830
            return -1;
        }
    }

    for (i = 0 ; i < vm->def->nnets ; i++) {
        if (qemuGetPCINetVendorProduct(vm->def->nets[i], &vendor, &product) < 0)
            continue;

        if (qemuAssignNextPCIAddress(&(vm->def->nets[i]->info),
                                     vendor, product,
                                     addrs,  naddrs) < 0) {
2831 2832 2833
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for %s NIC"),
                            vm->def->nets[i]->model);
2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844
            return -1;
        }
    }

    for (i = 0 ; i < vm->def->ncontrollers ; i++) {
        if (qemuGetPCIControllerVendorProduct(vm->def->controllers[i], &vendor, &product) < 0)
            continue;

        if (qemuAssignNextPCIAddress(&(vm->def->controllers[i]->info),
                                     vendor, product,
                                     addrs,  naddrs) < 0) {
2845 2846 2847
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for controller %s"),
                            virDomainControllerTypeToString(vm->def->controllers[i]->type));
2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858
            return -1;
        }
    }

    for (i = 0 ; i < vm->def->nvideos ; i++) {
        if (qemuGetPCIVideoVendorProduct(vm->def->videos[i], &vendor, &product) < 0)
            continue;

        if (qemuAssignNextPCIAddress(&(vm->def->videos[i]->info),
                                     vendor, product,
                                     addrs,  naddrs) < 0) {
2859 2860 2861
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for video adapter %s"),
                            virDomainVideoTypeToString(vm->def->videos[i]->type));
2862 2863 2864 2865 2866 2867 2868 2869 2870
            return -1;
        }
    }

    for (i = 0 ; i < vm->def->nsounds ; i++) {
        if (qemuGetPCISoundVendorProduct(vm->def->sounds[i], &vendor, &product) < 0)
            continue;

        if (qemuAssignNextPCIAddress(&(vm->def->sounds[i]->info),
2871
                                    vendor, product,
2872
                                     addrs,  naddrs) < 0) {
2873 2874 2875
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for sound adapter %s"),
                            virDomainSoundModelTypeToString(vm->def->sounds[i]->model));
2876 2877 2878 2879 2880 2881 2882 2883 2884 2885
            return -1;
        }
    }


    if (vm->def->watchdog &&
        qemuGetPCIWatchdogVendorProduct(vm->def->watchdog, &vendor, &product) == 0) {
        if (qemuAssignNextPCIAddress(&(vm->def->watchdog->info),
                                     vendor, product,
                                     addrs,  naddrs) < 0) {
2886 2887 2888
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for watchdog %s"),
                            virDomainWatchdogModelTypeToString(vm->def->watchdog->model));
2889 2890 2891 2892
            return -1;
        }
    }

2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904
    if (vm->def->memballoon &&
        qemuGetPCIMemballoonVendorProduct(vm->def->memballoon, &vendor, &product) == 0) {
        if (qemuAssignNextPCIAddress(&(vm->def->memballoon->info),
                                     vendor, product,
                                     addrs, naddrs) < 0) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for balloon %s"),
                            virDomainMemballoonModelTypeToString(vm->def->memballoon->model));
            return -1;
        }
    }

2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930
    /* XXX console (virtio) */


    /* ... and now things we don't have in our xml */

    /* XXX USB controller ? */

    /* XXX what about other PCI devices (ie bridges) */

    return 0;
}

static int
qemuInitPCIAddresses(struct qemud_driver *driver,
                     virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int naddrs;
    int ret;
    qemuMonitorPCIAddress *addrs = NULL;

    qemuDomainObjEnterMonitorWithDriver(driver, vm);
    naddrs = qemuMonitorGetAllPCIAddresses(priv->mon,
                                           &addrs);
    qemuDomainObjExitMonitorWithDriver(driver, vm);

2931
    ret = qemuDetectPCIAddresses(vm, addrs, naddrs);
2932 2933 2934 2935 2936 2937

    VIR_FREE(addrs);

    return ret;
}

2938
static int qemudNextFreeVNCPort(struct qemud_driver *driver) {
2939 2940
    int i;

2941
    for (i = QEMU_VNC_PORT_MIN; i < QEMU_VNC_PORT_MAX; i++) {
2942 2943 2944
        int fd;
        int reuse = 1;
        struct sockaddr_in addr;
2945 2946 2947 2948 2949 2950 2951 2952 2953
        bool used = false;

        if (virBitmapGetBit(driver->reservedVNCPorts,
                            i - QEMU_VNC_PORT_MIN, &used) < 0)
            VIR_DEBUG("virBitmapGetBit failed on bit %d", i - QEMU_VNC_PORT_MIN);

        if (used)
            continue;

2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968
        addr.sin_family = AF_INET;
        addr.sin_port = htons(i);
        addr.sin_addr.s_addr = htonl(INADDR_ANY);
        fd = socket(PF_INET, SOCK_STREAM, 0);
        if (fd < 0)
            return -1;

        if (setsockopt(fd, SOL_SOCKET, SO_REUSEADDR, (void*)&reuse, sizeof(reuse)) < 0) {
            close(fd);
            break;
        }

        if (bind(fd, (struct sockaddr*)&addr, sizeof(addr)) == 0) {
            /* Not in use, lets grab it */
            close(fd);
2969 2970 2971 2972 2973 2974
            /* Add port to bitmap of reserved ports */
            if (virBitmapSetBit(driver->reservedVNCPorts,
                                i - QEMU_VNC_PORT_MIN) < 0) {
                VIR_DEBUG("virBitmapSetBit failed on bit %d",
                          i - QEMU_VNC_PORT_MIN);
            }
2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988
            return i;
        }
        close(fd);

        if (errno == EADDRINUSE) {
            /* In use, try next */
            continue;
        }
        /* Some other bad failure, get out.. */
        break;
    }
    return -1;
}

2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026

static int
qemuAssignPCIAddresses(virDomainDefPtr def)
{
    int ret = -1;
    unsigned long long qemuCmdFlags = 0;
    qemuDomainPCIAddressSetPtr addrs = NULL;
    struct stat sb;

    if (stat(def->emulator, &sb) < 0) {
        virReportSystemError(errno,
                             _("Cannot find QEMU binary %s"),
                             def->emulator);
        goto cleanup;
    }

    if (qemudExtractVersionInfo(def->emulator,
                                NULL,
                                &qemuCmdFlags) < 0)
        goto cleanup;

    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (!(addrs = qemuDomainPCIAddressSetCreate(def)))
            goto cleanup;

        if (qemuAssignDevicePCISlots(def, addrs) < 0)
            goto cleanup;
    }

    ret = 0;

cleanup:
    qemuDomainPCIAddressSetFree(addrs);

    return ret;
}


3027
static pciDeviceList *
3028
qemuGetPciHostDeviceList(virDomainHostdevDefPtr *hostdevs, int nhostdevs)
3029 3030
{
    pciDeviceList *list;
3031 3032
    int i;

3033
    if (!(list = pciDeviceListNew()))
3034
        return NULL;
3035

3036 3037
    for (i = 0 ; i < nhostdevs ; i++) {
        virDomainHostdevDefPtr hostdev = hostdevs[i];
3038 3039 3040 3041 3042 3043 3044
        pciDevice *dev;

        if (hostdev->mode != VIR_DOMAIN_HOSTDEV_MODE_SUBSYS)
            continue;
        if (hostdev->source.subsys.type != VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI)
            continue;

3045
        dev = pciGetDevice(hostdev->source.subsys.u.pci.domain,
3046 3047 3048
                           hostdev->source.subsys.u.pci.bus,
                           hostdev->source.subsys.u.pci.slot,
                           hostdev->source.subsys.u.pci.function);
3049
        if (!dev) {
3050
            pciDeviceListFree(list);
3051 3052
            return NULL;
        }
3053

3054 3055 3056
        if (pciDeviceListAdd(list, dev) < 0) {
            pciFreeDevice(dev);
            pciDeviceListFree(list);
3057
            return NULL;
3058 3059
        }

3060
        pciDeviceSetManaged(dev, hostdev->managed);
3061 3062
    }

3063 3064 3065 3066
    return list;
}

static int
3067 3068 3069 3070
qemuUpdateActivePciHostdevs(struct qemud_driver *driver,
                            virDomainDefPtr def)
{
    pciDeviceList *pcidevs;
3071
    int i;
3072
    int ret = -1;
3073 3074 3075 3076

    if (!def->nhostdevs)
        return 0;

3077
    if (!(pcidevs = qemuGetPciHostDeviceList(def->hostdevs, def->nhostdevs)))
3078 3079
        return -1;

3080 3081
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
3082 3083 3084
        pciDeviceListSteal(pcidevs, dev);
        if (pciDeviceListAdd(driver->activePciHostdevs, dev) < 0) {
            pciFreeDevice(dev);
3085
            goto cleanup;
3086 3087 3088
        }
    }

3089 3090 3091
    ret = 0;

cleanup:
3092
    pciDeviceListFree(pcidevs);
3093 3094 3095
    return ret;
}

3096

3097
static int
3098 3099 3100
qemuPrepareHostdevPCIDevices(struct qemud_driver *driver,
                             virDomainHostdevDefPtr *hostdevs,
                             int nhostdevs)
3101 3102 3103
{
    pciDeviceList *pcidevs;
    int i;
3104
    int ret = -1;
3105

3106
    if (!(pcidevs = qemuGetPciHostDeviceList(hostdevs, nhostdevs)))
3107 3108
        return -1;

3109
    /* We have to use 3 loops here. *All* devices must
3110 3111
     * be detached before we reset any of them, because
     * in some cases you have to reset the whole PCI,
3112 3113
     * which impacts all devices on it. Also, all devices
     * must be reset before being marked as active.
3114 3115 3116 3117 3118 3119 3120
     */

    /* XXX validate that non-managed device isn't in use, eg
     * by checking that device is either un-bound, or bound
     * to pci-stub.ko
     */

3121 3122
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
3123
        if (!pciDeviceIsAssignable(dev, !driver->relaxedACS))
3124 3125
            goto cleanup;

3126
        if (pciDeviceGetManaged(dev) &&
3127
            pciDettachDevice(dev, driver->activePciHostdevs) < 0)
3128 3129
            goto cleanup;
    }
3130 3131 3132

    /* Now that all the PCI hostdevs have be dettached, we can safely
     * reset them */
3133 3134
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
3135
        if (pciResetDevice(dev, driver->activePciHostdevs, pcidevs) < 0)
3136 3137
            goto cleanup;
    }
3138

3139
    /* Now mark all the devices as active */
3140 3141
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
3142 3143 3144
        pciDeviceListSteal(pcidevs, dev);
        if (pciDeviceListAdd(driver->activePciHostdevs, dev) < 0) {
            pciFreeDevice(dev);
3145 3146
            goto cleanup;
        }
3147 3148
    }

3149
    ret = 0;
3150

3151
cleanup:
3152
    pciDeviceListFree(pcidevs);
3153
    return ret;
3154 3155
}

3156 3157 3158 3159 3160 3161 3162
static int
qemuPrepareHostPCIDevices(struct qemud_driver *driver,
                          virDomainDefPtr def)
{
    return qemuPrepareHostdevPCIDevices(driver, def->hostdevs, def->nhostdevs);
}

3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173

static int
qemuPrepareHostUSBDevices(struct qemud_driver *driver ATTRIBUTE_UNUSED,
                          virDomainDefPtr def)
{
    int i;
    for (i = 0 ; i < def->nhostdevs ; i++) {
        virDomainHostdevDefPtr hostdev = def->hostdevs[i];

        if (hostdev->mode != VIR_DOMAIN_HOSTDEV_MODE_SUBSYS)
            continue;
3174
        if (hostdev->source.subsys.type != VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB)
3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212
            continue;

        /* Resolve a vendor/product to bus/device */
        if (hostdev->source.subsys.u.usb.vendor) {
            usbDevice *usb
                = usbFindDevice(hostdev->source.subsys.u.usb.vendor,
                                hostdev->source.subsys.u.usb.product);

            if (!usb)
                return -1;

            hostdev->source.subsys.u.usb.bus = usbDeviceGetBus(usb);
            hostdev->source.subsys.u.usb.device = usbDeviceGetDevno(usb);

            usbFreeDevice(usb);
        }
    }

    return 0;
}

static int
qemuPrepareHostDevices(struct qemud_driver *driver,
                       virDomainDefPtr def)
{
    if (!def->nhostdevs)
        return 0;

    if (qemuPrepareHostPCIDevices(driver, def) < 0)
        return -1;

    if (qemuPrepareHostUSBDevices(driver, def) < 0)
        return -1;

    return 0;
}


3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234
static int
qemuPrepareChardevDevice(virDomainDefPtr def ATTRIBUTE_UNUSED,
                         virDomainChrDefPtr dev,
                         void *opaque ATTRIBUTE_UNUSED)
{
    int fd;
    if (dev->type != VIR_DOMAIN_CHR_TYPE_FILE)
        return 0;

    if ((fd = open(dev->data.file.path, O_CREAT | O_APPEND, S_IRUSR|S_IWUSR)) < 0) {
        virReportSystemError(errno,
                             _("Unable to pre-create chardev file '%s'"),
                             dev->data.file.path);
        return -1;
    }

    close(fd);

    return 0;
}


3235
static void
3236
qemudReattachPciDevice(pciDevice *dev, struct qemud_driver *driver)
3237 3238 3239
{
    int retries = 100;

3240 3241 3242 3243 3244 3245
    while (pciWaitForDeviceCleanup(dev, "kvm_assigned_device")
           && retries) {
        usleep(100*1000);
        retries--;
    }

3246
    if (pciDeviceGetManaged(dev)) {
3247
        if (pciReAttachDevice(dev, driver->activePciHostdevs) < 0) {
3248 3249
            virErrorPtr err = virGetLastError();
            VIR_ERROR(_("Failed to re-attach PCI device: %s"),
3250
                      err ? err->message : _("unknown error"));
3251 3252 3253 3254 3255
            virResetError(err);
        }
    }
}

3256
static void
3257 3258 3259
qemuDomainReAttachHostdevDevices(struct qemud_driver *driver,
                              virDomainHostdevDefPtr *hostdevs,
                              int nhostdevs)
3260
{
3261
    pciDeviceList *pcidevs;
3262 3263
    int i;

3264
    if (!(pcidevs = qemuGetPciHostDeviceList(hostdevs, nhostdevs))) {
3265
        virErrorPtr err = virGetLastError();
3266
        VIR_ERROR(_("Failed to allocate pciDeviceList: %s"),
3267
                  err ? err->message : _("unknown error"));
3268 3269
        virResetError(err);
        return;
3270 3271
    }

3272 3273
    /* Again 3 loops; mark all devices as inactive before reset
     * them and reset all the devices before re-attach */
3274

3275 3276
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
3277
        pciDeviceListDel(driver->activePciHostdevs, dev);
3278
    }
3279

3280 3281
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
3282
        if (pciResetDevice(dev, driver->activePciHostdevs, pcidevs) < 0) {
3283
            virErrorPtr err = virGetLastError();
3284
            VIR_ERROR(_("Failed to reset PCI device: %s"),
3285
                      err ? err->message : _("unknown error"));
3286 3287
            virResetError(err);
        }
3288
    }
3289

3290 3291
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
3292
        qemudReattachPciDevice(dev, driver);
3293
    }
3294

3295
    pciDeviceListFree(pcidevs);
3296 3297
}

3298 3299 3300 3301 3302 3303 3304 3305 3306 3307
static void
qemuDomainReAttachHostDevices(struct qemud_driver *driver,
                              virDomainDefPtr def)
{
    if (!def->nhostdevs)
        return;

    qemuDomainReAttachHostdevDevices(driver, def->hostdevs, def->nhostdevs);
}

3308 3309 3310 3311 3312 3313 3314 3315 3316 3317
static const char *const defaultDeviceACL[] = {
    "/dev/null", "/dev/full", "/dev/zero",
    "/dev/random", "/dev/urandom",
    "/dev/ptmx", "/dev/kvm", "/dev/kqemu",
    "/dev/rtc", "/dev/hpet", "/dev/net/tun",
    NULL,
};
#define DEVICE_PTY_MAJOR 136
#define DEVICE_SND_MAJOR 116

3318

3319 3320 3321 3322 3323 3324 3325
static int qemuSetupDiskPathAllow(virDomainDiskDefPtr disk ATTRIBUTE_UNUSED,
                                  const char *path,
                                  size_t depth ATTRIBUTE_UNUSED,
                                  void *opaque)
{
    virCgroupPtr cgroup = opaque;
    int rc;
3326

3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340
    VIR_DEBUG("Process path %s for disk", path);
    /* XXX RO vs RW */
    rc = virCgroupAllowDevicePath(cgroup, path);
    if (rc != 0) {
        /* Get this for non-block devices */
        if (rc == -EINVAL) {
            VIR_DEBUG("Ignoring EINVAL for %s", path);
        } else if (rc == -EACCES) { /* Get this for root squash NFS */
            VIR_DEBUG("Ignoring EACCES for %s", path);
        } else {
            virReportSystemError(-rc,
                                 _("Unable to allow access for disk path %s"),
                                 path);
            return -1;
3341
        }
3342
    }
3343 3344
    return 0;
}
3345 3346


3347 3348
static int qemuSetupDiskCgroup(struct qemud_driver *driver,
                               virCgroupPtr cgroup,
3349 3350 3351
                               virDomainDiskDefPtr disk)
{
    return virDomainDiskDefForeachPath(disk,
3352
                                       driver->allowDiskFormatProbing,
3353 3354 3355
                                       true,
                                       qemuSetupDiskPathAllow,
                                       cgroup);
3356 3357 3358
}


3359 3360 3361 3362
static int qemuTeardownDiskPathDeny(virDomainDiskDefPtr disk ATTRIBUTE_UNUSED,
                                    const char *path,
                                    size_t depth ATTRIBUTE_UNUSED,
                                    void *opaque)
3363
{
3364 3365
    virCgroupPtr cgroup = opaque;
    int rc;
3366

3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380
    VIR_DEBUG("Process path %s for disk", path);
    /* XXX RO vs RW */
    rc = virCgroupDenyDevicePath(cgroup, path);
    if (rc != 0) {
        /* Get this for non-block devices */
        if (rc == -EINVAL) {
            VIR_DEBUG("Ignoring EINVAL for %s", path);
        } else if (rc == -EACCES) { /* Get this for root squash NFS */
            VIR_DEBUG("Ignoring EACCES for %s", path);
        } else {
            virReportSystemError(-rc,
                                 _("Unable to allow access for disk path %s"),
                                 path);
            return -1;
3381
        }
3382
    }
3383 3384
    return 0;
}
3385 3386


3387 3388
static int qemuTeardownDiskCgroup(struct qemud_driver *driver,
                                  virCgroupPtr cgroup,
3389 3390 3391
                                  virDomainDiskDefPtr disk)
{
    return virDomainDiskDefForeachPath(disk,
3392
                                       driver->allowDiskFormatProbing,
3393 3394 3395
                                       true,
                                       qemuTeardownDiskPathDeny,
                                       cgroup);
3396 3397 3398
}


3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422
static int qemuSetupChardevCgroup(virDomainDefPtr def,
                                  virDomainChrDefPtr dev,
                                  void *opaque)
{
    virCgroupPtr cgroup = opaque;
    int rc;

    if (dev->type != VIR_DOMAIN_CHR_TYPE_DEV)
        return 0;


    VIR_DEBUG("Process path '%s' for disk", dev->data.file.path);
    rc = virCgroupAllowDevicePath(cgroup, dev->data.file.path);
    if (rc != 0) {
        virReportSystemError(-rc,
                             _("Unable to allow device %s for %s"),
                             dev->data.file.path, def->name);
        return -1;
    }

    return 0;
}


3423
static int qemuSetupCgroup(struct qemud_driver *driver,
3424 3425 3426 3427
                           virDomainObjPtr vm)
{
    virCgroupPtr cgroup = NULL;
    int rc;
3428
    unsigned int i;
3429 3430 3431 3432
    const char *const *deviceACL =
        driver->cgroupDeviceACL ?
        (const char *const *)driver->cgroupDeviceACL :
        defaultDeviceACL;
3433 3434 3435 3436 3437 3438

    if (driver->cgroup == NULL)
        return 0; /* Not supported, so claim success */

    rc = virCgroupForDomain(driver->cgroup, vm->def->name, &cgroup, 1);
    if (rc != 0) {
3439
        virReportSystemError(-rc,
3440 3441 3442 3443 3444
                             _("Unable to create cgroup for %s"),
                             vm->def->name);
        goto cleanup;
    }

3445 3446
    if (qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_DEVICES)) {
        rc = virCgroupDenyAllDevices(cgroup);
3447
        if (rc != 0) {
3448 3449 3450 3451 3452
            if (rc == -EPERM) {
                VIR_WARN0("Group devices ACL is not accessible, disabling whitelisting");
                goto done;
            }

3453
            virReportSystemError(-rc,
3454
                                 _("Unable to deny all devices for %s"), vm->def->name);
3455 3456 3457
            goto cleanup;
        }

3458
        for (i = 0; i < vm->def->ndisks ; i++) {
3459
            if (qemuSetupDiskCgroup(driver, cgroup, vm->def->disks[i]) < 0)
3460 3461
                goto cleanup;
        }
3462

3463
        rc = virCgroupAllowDeviceMajor(cgroup, 'c', DEVICE_PTY_MAJOR);
3464
        if (rc != 0) {
3465
            virReportSystemError(-rc, "%s",
3466
                                 _("unable to allow /dev/pts/ devices"));
3467 3468 3469
            goto cleanup;
        }

3470 3471 3472
        if (vm->def->nsounds) {
            rc = virCgroupAllowDeviceMajor(cgroup, 'c', DEVICE_SND_MAJOR);
            if (rc != 0) {
3473
                virReportSystemError(-rc, "%s",
3474 3475 3476 3477 3478 3479 3480 3481 3482 3483
                                     _("unable to allow /dev/snd/ devices"));
                goto cleanup;
            }
        }

        for (i = 0; deviceACL[i] != NULL ; i++) {
            rc = virCgroupAllowDevicePath(cgroup,
                                          deviceACL[i]);
            if (rc < 0 &&
                rc != -ENOENT) {
3484
                virReportSystemError(-rc,
3485 3486 3487 3488
                                     _("unable to allow device %s"),
                                     deviceACL[i]);
                goto cleanup;
            }
3489
        }
3490 3491 3492 3493 3494 3495

        if (virDomainChrDefForeach(vm->def,
                                   true,
                                   qemuSetupChardevCgroup,
                                   cgroup) < 0)
            goto cleanup;
3496 3497 3498
    }

done:
3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510
    virCgroupFree(&cgroup);
    return 0;

cleanup:
    if (cgroup) {
        virCgroupRemove(cgroup);
        virCgroupFree(&cgroup);
    }
    return -1;
}


3511
static int qemuRemoveCgroup(struct qemud_driver *driver,
3512 3513
                            virDomainObjPtr vm,
                            int quiet)
3514 3515 3516 3517 3518 3519 3520 3521 3522
{
    virCgroupPtr cgroup;
    int rc;

    if (driver->cgroup == NULL)
        return 0; /* Not supported, so claim success */

    rc = virCgroupForDomain(driver->cgroup, vm->def->name, &cgroup, 0);
    if (rc != 0) {
3523
        if (!quiet)
3524 3525 3526
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("Unable to find cgroup for %s\n"),
                            vm->def->name);
3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546
        return rc;
    }

    rc = virCgroupRemove(cgroup);
    virCgroupFree(&cgroup);
    return rc;
}

static int qemuAddToCgroup(struct qemud_driver *driver,
                           virDomainDefPtr def)
{
    virCgroupPtr cgroup = NULL;
    int ret = -1;
    int rc;

    if (driver->cgroup == NULL)
        return 0; /* Not supported, so claim success */

    rc = virCgroupForDomain(driver->cgroup, def->name, &cgroup, 0);
    if (rc != 0) {
3547
        virReportSystemError(-rc,
3548 3549 3550 3551 3552 3553 3554
                             _("unable to find cgroup for domain %s"),
                             def->name);
        goto cleanup;
    }

    rc = virCgroupAddTask(cgroup, getpid());
    if (rc != 0) {
3555
        virReportSystemError(-rc,
3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572
                             _("unable to add domain %s task %d to cgroup"),
                             def->name, getpid());
        goto cleanup;
    }

    ret = 0;

cleanup:
    virCgroupFree(&cgroup);
    return ret;
}


struct qemudHookData {
    virConnectPtr conn;
    virDomainObjPtr vm;
    struct qemud_driver *driver;
3573 3574 3575
};

static int qemudSecurityHook(void *data) {
3576 3577
    struct qemudHookData *h = data;

3578 3579 3580
    /* This must take place before exec(), so that all QEMU
     * memory allocation is on the correct NUMA node
     */
3581
    if (qemuAddToCgroup(h->driver, h->vm->def) < 0)
3582 3583
        return -1;

3584 3585 3586
    /* This must be done after cgroup placement to avoid resetting CPU
     * affinity */
    if (qemudInitCpuAffinity(h->vm) < 0)
3587
        return -1;
3588

3589 3590
    if (h->driver->securityDriver &&
        h->driver->securityDriver->domainSetSecurityProcessLabel &&
3591
        h->driver->securityDriver->domainSetSecurityProcessLabel(h->driver->securityDriver, h->vm) < 0)
3592 3593 3594
        return -1;

    return 0;
3595 3596
}

3597
static int
3598
qemuPrepareMonitorChr(struct qemud_driver *driver,
3599
                      virDomainChrDefPtr monConfig,
3600 3601
                      const char *vm)
{
3602
    monConfig->deviceType = VIR_DOMAIN_CHR_DEVICE_TYPE_MONITOR;
3603

3604 3605
    monConfig->type = VIR_DOMAIN_CHR_TYPE_UNIX;
    monConfig->data.nix.listen = 1;
3606

D
Daniel P. Berrange 已提交
3607
    if (!(monConfig->info.alias = strdup("monitor"))) {
3608
        virReportOOMError();
D
Daniel P. Berrange 已提交
3609 3610 3611
        return -1;
    }

3612
    if (virAsprintf(&monConfig->data.nix.path, "%s/%s.monitor",
3613
                    driver->libDir, vm) < 0) {
3614
        virReportOOMError();
3615 3616 3617 3618 3619 3620
        return -1;
    }

    return 0;
}

C
Chris Lalancette 已提交
3621 3622 3623 3624 3625
static int qemuDomainSnapshotSetActive(virDomainObjPtr vm,
                                       char *snapshotDir);
static int qemuDomainSnapshotSetInactive(virDomainObjPtr vm,
                                         char *snapshotDir);

3626 3627
static int qemudStartVMDaemon(virConnectPtr conn,
                              struct qemud_driver *driver,
3628
                              virDomainObjPtr vm,
3629
                              const char *migrateFrom,
3630
                              bool start_paused,
3631 3632
                              int stdin_fd,
                              const char *stdin_path) {
3633
    const char **argv = NULL, **tmp;
3634
    const char **progenv = NULL;
3635
    int i, ret, runflags;
3636
    struct stat sb;
3637 3638
    int *vmfds = NULL;
    int nvmfds = 0;
3639
    unsigned long long qemuCmdFlags;
3640
    fd_set keepfd;
3641
    const char *emulator;
G
Guido Günther 已提交
3642
    pid_t child;
3643
    int pos = -1;
3644
    char ebuf[1024];
3645
    char *pidfile = NULL;
3646
    int logfile = -1;
3647
    qemuDomainObjPrivatePtr priv = vm->privateData;
3648

3649
    struct qemudHookData hookData;
3650 3651 3652 3653
    hookData.conn = conn;
    hookData.vm = vm;
    hookData.driver = driver;

3654
    FD_ZERO(&keepfd);
3655

3656 3657
    DEBUG0("Beginning VM startup process");

D
Daniel P. Berrange 已提交
3658
    if (virDomainObjIsActive(vm)) {
3659 3660
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("VM is already active"));
3661 3662 3663
        return -1;
    }

3664 3665 3666 3667 3668
    /* Must be run before security labelling */
    DEBUG0("Preparing host devices");
    if (qemuPrepareHostDevices(driver, vm->def) < 0)
        goto cleanup;

3669 3670 3671 3672 3673 3674 3675
    DEBUG0("Preparing chr devices");
    if (virDomainChrDefForeach(vm->def,
                               true,
                               qemuPrepareChardevDevice,
                               NULL) < 0)
        goto cleanup;

3676 3677
    /* If you are using a SecurityDriver with dynamic labelling,
       then generate a security label for isolation */
3678
    DEBUG0("Generating domain security label (if required)");
3679
    if (driver->securityDriver &&
3680
        driver->securityDriver->domainGenSecurityLabel &&
3681 3682
        driver->securityDriver->domainGenSecurityLabel(driver->securityDriver,
                                                       vm) < 0)
3683
        goto cleanup;
3684

3685
    DEBUG0("Generating setting domain security labels (if required)");
3686 3687
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityAllLabel &&
3688 3689
        driver->securityDriver->domainSetSecurityAllLabel(driver->securityDriver,
                                                          vm, stdin_path) < 0) {
3690
        if (stdin_path && virStorageFileIsSharedFS(stdin_path) != 1)
3691 3692
            goto cleanup;
    }
3693

3694 3695 3696
    /* Ensure no historical cgroup for this VM is lying around bogus
     * settings */
    DEBUG0("Ensuring no historical cgroup is lying around");
3697
    qemuRemoveCgroup(driver, vm, 1);
3698

3699 3700 3701
    if ((vm->def->ngraphics == 1) &&
        vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
        vm->def->graphics[0]->data.vnc.autoport) {
3702
        DEBUG0("Determining VNC port");
3703
        int port = qemudNextFreeVNCPort(driver);
3704
        if (port < 0) {
3705 3706
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            "%s", _("Unable to find an unused VNC port"));
3707
            goto cleanup;
3708
        }
3709
        vm->def->graphics[0]->data.vnc.port = port;
3710
    }
3711

L
Laine Stump 已提交
3712
    if (virFileMakePath(driver->logDir) != 0) {
3713
        virReportSystemError(errno,
3714 3715
                             _("cannot create log directory %s"),
                             driver->logDir);
3716
        goto cleanup;
3717 3718
    }

3719
    DEBUG0("Creating domain log file");
3720
    if ((logfile = qemudLogFD(driver, vm->def->name)) < 0)
3721
        goto cleanup;
3722

3723 3724
    emulator = vm->def->emulator;

3725 3726 3727 3728
    /* Make sure the binary we are about to try exec'ing exists.
     * Technically we could catch the exec() failure, but that's
     * in a sub-process so its hard to feed back a useful error
     */
3729
    if (stat(emulator, &sb) < 0) {
3730
        virReportSystemError(errno,
3731 3732
                             _("Cannot find QEMU binary %s"),
                             emulator);
3733
        goto cleanup;
3734 3735
    }

3736
    DEBUG0("Determing emulator version");
3737
    if (qemudExtractVersionInfo(emulator,
3738
                                NULL,
3739
                                &qemuCmdFlags) < 0)
3740
        goto cleanup;
3741

3742
    DEBUG0("Setting up domain cgroup (if required)");
3743
    if (qemuSetupCgroup(driver, vm) < 0)
3744 3745
        goto cleanup;

3746
    if (VIR_ALLOC(priv->monConfig) < 0) {
3747
        virReportOOMError();
3748 3749 3750
        goto cleanup;
    }

3751
    DEBUG0("Preparing monitor state");
3752
    if (qemuPrepareMonitorChr(driver, priv->monConfig, vm->def->name) < 0)
3753
        goto cleanup;
3754

D
Daniel P. Berrange 已提交
3755 3756 3757
#if HAVE_YAJL
    if (qemuCmdFlags & QEMUD_CMD_FLAG_MONITOR_JSON)
        priv->monJSON = 1;
3758
    else
D
Daniel P. Berrange 已提交
3759
#endif
3760
        priv->monJSON = 0;
D
Daniel P. Berrange 已提交
3761

3762 3763
    priv->monitor_warned = 0;

D
Daniel P. Berrange 已提交
3764
    if ((ret = virFileDeletePid(driver->stateDir, vm->def->name)) != 0) {
3765
        virReportSystemError(ret,
D
Daniel P. Berrange 已提交
3766 3767 3768 3769 3770
                             _("Cannot remove stale PID file for %s"),
                             vm->def->name);
        goto cleanup;
    }

3771
    if (!(pidfile = virFilePid(driver->stateDir, vm->def->name))) {
3772
        virReportSystemError(errno,
3773 3774 3775
                             "%s", _("Failed to build pidfile path."));
        goto cleanup;
    }
D
Daniel P. Berrange 已提交
3776

3777
    /*
M
Matthew Booth 已提交
3778
     * Normally PCI addresses are assigned in the virDomainCreate
3779 3780 3781 3782 3783
     * or virDomainDefine methods. We might still need to assign
     * some here to cope with the question of upgrades. Regardless
     * we also need to populate the PCi address set cache for later
     * use in hotplug
     */
3784
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
3785
        DEBUG0("Assigning domain PCI addresses");
3786
        /* Populate cache with current addresses */
3787 3788 3789 3790 3791 3792 3793
        if (priv->pciaddrs) {
            qemuDomainPCIAddressSetFree(priv->pciaddrs);
            priv->pciaddrs = NULL;
        }
        if (!(priv->pciaddrs = qemuDomainPCIAddressSetCreate(vm->def)))
            goto cleanup;

3794 3795

        /* Assign any remaining addresses */
3796 3797
        if (qemuAssignDevicePCISlots(vm->def, priv->pciaddrs) < 0)
            goto cleanup;
3798 3799 3800 3801

        priv->persistentAddrs = 1;
    } else {
        priv->persistentAddrs = 0;
3802 3803
    }

3804
    DEBUG0("Building emulator command line");
3805
    vm->def->id = driver->nextvmid++;
3806
    if (qemudBuildCommandLine(conn, driver, vm->def, priv->monConfig,
D
Daniel P. Berrange 已提交
3807
                              priv->monJSON, qemuCmdFlags, &argv, &progenv,
3808
                              &vmfds, &nvmfds, migrateFrom,
C
Chris Lalancette 已提交
3809 3810 3811 3812
                              vm->current_snapshot) < 0)
        goto cleanup;

    if (qemuDomainSnapshotSetInactive(vm, driver->snapshotDir) < 0)
3813
        goto cleanup;
3814

3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830
    /* now that we know it is about to start call the hook if present */
    if (virHookPresent(VIR_HOOK_DRIVER_QEMU)) {
        char *xml = virDomainDefFormat(vm->def, 0);
        int hookret;

        hookret = virHookCall(VIR_HOOK_DRIVER_QEMU, vm->def->name,
                    VIR_HOOK_QEMU_OP_START, VIR_HOOK_SUBOP_BEGIN, NULL, xml);
        VIR_FREE(xml);

        /*
         * If the script raised an error abort the launch
         */
        if (hookret < 0)
            goto cleanup;
    }

3831 3832
    tmp = progenv;
    while (*tmp) {
3833
        if (safewrite(logfile, *tmp, strlen(*tmp)) < 0)
3834
            VIR_WARN("Unable to write envv to logfile: %s",
3835
                     virStrerror(errno, ebuf, sizeof ebuf));
3836
        if (safewrite(logfile, " ", 1) < 0)
3837
            VIR_WARN("Unable to write envv to logfile: %s",
3838
                     virStrerror(errno, ebuf, sizeof ebuf));
3839 3840
        tmp++;
    }
3841 3842
    tmp = argv;
    while (*tmp) {
3843
        if (safewrite(logfile, *tmp, strlen(*tmp)) < 0)
3844
            VIR_WARN("Unable to write argv to logfile: %s",
3845
                     virStrerror(errno, ebuf, sizeof ebuf));
3846
        if (safewrite(logfile, " ", 1) < 0)
3847
            VIR_WARN("Unable to write argv to logfile: %s",
3848
                     virStrerror(errno, ebuf, sizeof ebuf));
3849 3850
        tmp++;
    }
3851
    if (safewrite(logfile, "\n", 1) < 0)
3852
        VIR_WARN("Unable to write argv to logfile: %s",
3853
                 virStrerror(errno, ebuf, sizeof ebuf));
3854

3855
    if ((pos = lseek(logfile, 0, SEEK_END)) < 0)
3856
        VIR_WARN("Unable to seek to end of logfile: %s",
3857
                 virStrerror(errno, ebuf, sizeof ebuf));
3858

3859 3860
    for (i = 0 ; i < nvmfds ; i++)
        FD_SET(vmfds[i], &keepfd);
3861

3862 3863 3864 3865 3866 3867 3868
    VIR_DEBUG("Clear emulator capabilities: %d",
              driver->clearEmulatorCapabilities);
    runflags = VIR_EXEC_NONBLOCK;
    if (driver->clearEmulatorCapabilities) {
        runflags |= VIR_EXEC_CLEAR_CAPS;
    }

3869
    ret = virExecDaemonize(argv, progenv, &keepfd, &child,
3870
                           stdin_fd, &logfile, &logfile,
3871
                           runflags,
3872 3873 3874
                           qemudSecurityHook, &hookData,
                           pidfile);
    VIR_FREE(pidfile);
G
Guido Günther 已提交
3875 3876 3877

    /* wait for qemu process to to show up */
    if (ret == 0) {
3878
        if (virFileReadPid(driver->stateDir, vm->def->name, &vm->pid)) {
3879 3880
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("Domain %s didn't show up\n"), vm->def->name);
3881
            ret = -1;
G
Guido Günther 已提交
3882
        }
3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893
    } else if (ret == -2) {
        /* The virExec process that launches the daemon failed. Pending on
         * when it failed (we can't determine for sure), there may be
         * extra info in the domain log (if the hook failed for example).
         *
         * Pretend like things succeeded, and let 'WaitForMonitor' report
         * the log contents for us.
         */
        vm->pid = child;
        ret = 0;
    }
3894

3895 3896 3897
    if (migrateFrom)
        start_paused = true;
    vm->state = start_paused ? VIR_DOMAIN_PAUSED : VIR_DOMAIN_RUNNING;
3898

3899
    for (i = 0 ; argv[i] ; i++)
3900 3901
        VIR_FREE(argv[i]);
    VIR_FREE(argv);
3902

3903 3904 3905 3906
    for (i = 0 ; progenv[i] ; i++)
        VIR_FREE(progenv[i]);
    VIR_FREE(progenv);

3907
    if (ret == -1) /* The VM failed to start; tear filters before taps */
3908
        virDomainConfVMNWFilterTeardown(vm);
3909

3910 3911 3912
    if (vmfds) {
        for (i = 0 ; i < nvmfds ; i++) {
            close(vmfds[i]);
3913
        }
3914
        VIR_FREE(vmfds);
3915 3916
    }

3917
    if (ret == -1) /* The VM failed to start */
3918 3919
        goto cleanup;

3920
    DEBUG0("Waiting for monitor to show up");
3921
    if (qemudWaitForMonitor(driver, vm, pos) < 0)
3922
        goto cleanup;
3923

3924
    DEBUG0("Detecting VCPU PIDs");
3925
    if (qemuDetectVcpuPIDs(driver, vm) < 0)
3926
        goto cleanup;
3927

3928
    DEBUG0("Setting any required VM passwords");
3929
    if (qemuInitPasswords(conn, driver, vm, qemuCmdFlags) < 0)
3930
        goto cleanup;
3931

D
Daniel P. Berrange 已提交
3932 3933 3934
    /* If we have -device, then addresses are assigned explicitly.
     * If not, then we have to detect dynamic ones here */
    if (!(qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)) {
3935
        DEBUG0("Determining domain device PCI addresses");
D
Daniel P. Berrange 已提交
3936
        if (qemuInitPCIAddresses(driver, vm) < 0)
3937
            goto cleanup;
D
Daniel P. Berrange 已提交
3938
    }
3939

3940
    DEBUG0("Setting initial memory amount");
3941
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
3942
    if (qemuMonitorSetBalloon(priv->mon, vm->def->memory) < 0) {
3943
        qemuDomainObjExitMonitorWithDriver(driver, vm);
3944
        goto cleanup;
3945
    }
3946

3947
    if (!start_paused) {
3948
        DEBUG0("Starting domain CPUs");
3949 3950 3951
        /* Allow the CPUS to start executing */
        if (qemuMonitorStartCPUs(priv->mon, conn) < 0) {
            if (virGetLastError() == NULL)
3952 3953
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                "%s", _("resume operation failed"));
3954
            qemuDomainObjExitMonitorWithDriver(driver, vm);
3955
            goto cleanup;
3956 3957
        }
    }
3958
    qemuDomainObjExitMonitorWithDriver(driver, vm);
3959

3960

3961
    DEBUG0("Writing domain status to disk");
3962
    if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
3963
        goto cleanup;
3964

3965 3966 3967
    if (logfile != -1)
        close(logfile);

3968
    return 0;
3969 3970

cleanup:
3971 3972 3973
    /* We jump here if we failed to start the VM for any reason, or
     * if we failed to initialize the now running VM. kill it off and
     * pretend we never started it */
3974
    qemudShutdownVMDaemon(driver, vm, 0);
3975 3976 3977 3978 3979

    if (logfile != -1)
        close(logfile);

    return -1;
3980 3981 3982
}


3983
static void qemudShutdownVMDaemon(struct qemud_driver *driver,
3984 3985
                                  virDomainObjPtr vm,
                                  int migrated) {
D
Daniel P. Berrange 已提交
3986
    int ret;
3987
    int retries = 0;
3988
    qemuDomainObjPrivatePtr priv = vm->privateData;
3989
    virErrorPtr orig_err;
3990 3991
    virDomainDefPtr def;
    int i;
D
Daniel P. Berrange 已提交
3992

3993 3994
    VIR_DEBUG("Shutting down VM '%s' pid=%d migrated=%d",
              vm->def->name, vm->pid, migrated);
3995

3996 3997 3998 3999
    /* This method is routinely used in clean up paths. Disable error
     * reporting so we don't squash a legit error. */
    orig_err = virSaveLastError();

4000
    virDomainConfVMNWFilterTeardown(vm);
S
Stefan Berger 已提交
4001

4002
    if (driver->macFilter) {
4003
        def = vm->def;
4004 4005 4006 4007
        for (i = 0 ; i < def->nnets ; i++) {
            virDomainNetDefPtr net = def->nets[i];
            if (net->ifname == NULL)
                continue;
4008
            if ((errno = networkDisallowMacOnPort(driver, net->ifname,
4009
                                                  net->mac))) {
4010
                virReportSystemError(errno,
4011 4012 4013 4014 4015 4016
             _("failed to remove ebtables rule to allow MAC address on  '%s'"),
                                     net->ifname);
            }
        }
    }

4017
    /* This will safely handle a non-running guest with pid=0 or pid=-1*/
G
Guido Günther 已提交
4018 4019
    if (virKillProcess(vm->pid, 0) == 0 &&
        virKillProcess(vm->pid, SIGTERM) < 0)
4020
        virReportSystemError(errno,
4021 4022
                             _("Failed to send SIGTERM to %s (%d)"),
                             vm->def->name, vm->pid);
4023

4024 4025
    if (priv->mon)
        qemuMonitorClose(priv->mon);
4026

4027 4028 4029 4030 4031
    if (priv->monConfig) {
        if (priv->monConfig->type == VIR_DOMAIN_CHR_TYPE_UNIX)
            unlink(priv->monConfig->data.nix.path);
        virDomainChrDefFree(priv->monConfig);
        priv->monConfig = NULL;
4032 4033
    }

G
Guido Günther 已提交
4034 4035
    /* shut it off for sure */
    virKillProcess(vm->pid, SIGKILL);
4036

4037 4038 4039 4040 4041 4042 4043 4044 4045 4046
    /* now that we know it's stopped call the hook if present */
    if (virHookPresent(VIR_HOOK_DRIVER_QEMU)) {
        char *xml = virDomainDefFormat(vm->def, 0);

        /* we can't stop the operation even if the script raised an error */
        virHookCall(VIR_HOOK_DRIVER_QEMU, vm->def->name,
                    VIR_HOOK_QEMU_OP_STOPPED, VIR_HOOK_SUBOP_END, NULL, xml);
        VIR_FREE(xml);
    }

4047
    /* Reset Security Labels */
4048
    if (driver->securityDriver &&
4049
        driver->securityDriver->domainRestoreSecurityAllLabel)
4050 4051
        driver->securityDriver->domainRestoreSecurityAllLabel(driver->securityDriver,
                                                              vm, migrated);
4052 4053
    if (driver->securityDriver &&
        driver->securityDriver->domainReleaseSecurityLabel)
4054 4055
        driver->securityDriver->domainReleaseSecurityLabel(driver->securityDriver,
                                                           vm);
4056

4057 4058 4059 4060 4061 4062 4063
    /* Clear out dynamically assigned labels */
    if (vm->def->seclabel.type == VIR_DOMAIN_SECLABEL_DYNAMIC) {
        VIR_FREE(vm->def->seclabel.model);
        VIR_FREE(vm->def->seclabel.label);
        VIR_FREE(vm->def->seclabel.imagelabel);
    }

D
Daniel P. Berrange 已提交
4064
    virDomainDefClearDeviceAliases(vm->def);
4065 4066 4067 4068 4069
    if (!priv->persistentAddrs) {
        virDomainDefClearPCIAddresses(vm->def);
        qemuDomainPCIAddressSetFree(priv->pciaddrs);
        priv->pciaddrs = NULL;
    }
4070

4071
    qemuDomainReAttachHostDevices(driver, vm->def);
4072

S
Stefan Berger 已提交
4073 4074 4075 4076
#if WITH_MACVTAP
    def = vm->def;
    for (i = 0; i < def->nnets; i++) {
        virDomainNetDefPtr net = def->nets[i];
4077
        if (net->type == VIR_DOMAIN_NET_TYPE_DIRECT) {
4078
            delMacvtap(net->ifname, net->mac, net->data.direct.linkdev,
4079
                       &net->data.direct.virtPortProfile);
4080 4081
            VIR_FREE(net->ifname);
        }
S
Stefan Berger 已提交
4082 4083 4084
    }
#endif

4085
retry:
4086
    if ((ret = qemuRemoveCgroup(driver, vm, 0)) < 0) {
4087 4088 4089 4090 4091 4092 4093 4094
        if (ret == -EBUSY && (retries++ < 5)) {
            usleep(200*1000);
            goto retry;
        }
        VIR_WARN("Failed to remove cgroup for %s",
                 vm->def->name);
    }

4095
    qemudRemoveDomainStatus(driver, vm);
D
Daniel P. Berrange 已提交
4096

4097 4098 4099 4100 4101 4102
    /* Remove VNC port from port reservation bitmap, but only if it was
       reserved by the driver (autoport=yes)
    */
    if ((vm->def->ngraphics == 1) &&
        vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
        vm->def->graphics[0]->data.vnc.autoport &&
4103
        vm->def->graphics[0]->data.vnc.port >= QEMU_VNC_PORT_MIN) {
4104 4105 4106 4107 4108 4109 4110 4111
        if (virBitmapClearBit(driver->reservedVNCPorts,
                              vm->def->graphics[0]->data.vnc.port - \
                              QEMU_VNC_PORT_MIN) < 0) {
            VIR_DEBUG("virBitmapClearBit failed on bit %d",
                      vm->def->graphics[0]->data.vnc.port - QEMU_VNC_PORT_MIN);
       }
    }

4112
    vm->pid = -1;
4113
    vm->def->id = -1;
4114
    vm->state = VIR_DOMAIN_SHUTOFF;
4115 4116
    VIR_FREE(priv->vcpupids);
    priv->nvcpupids = 0;
4117 4118

    if (vm->newDef) {
4119
        virDomainDefFree(vm->def);
4120
        vm->def = vm->newDef;
4121
        vm->def->id = -1;
4122 4123
        vm->newDef = NULL;
    }
4124 4125 4126 4127 4128

    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
4129 4130 4131
}


4132
static virDrvOpenStatus qemudOpen(virConnectPtr conn,
4133
                                  virConnectAuthPtr auth ATTRIBUTE_UNUSED,
4134
                                  int flags ATTRIBUTE_UNUSED) {
4135
    if (conn->uri == NULL) {
4136 4137 4138
        if (qemu_driver == NULL)
            return VIR_DRV_OPEN_DECLINED;

4139
        conn->uri = xmlParseURI(qemu_driver->privileged ?
4140 4141
                                "qemu:///system" :
                                "qemu:///session");
4142
        if (!conn->uri) {
4143
            virReportOOMError();
4144 4145
            return VIR_DRV_OPEN_ERROR;
        }
4146 4147 4148 4149 4150 4151 4152 4153 4154 4155
    } else {
        /* If URI isn't 'qemu' its definitely not for us */
        if (conn->uri->scheme == NULL ||
            STRNEQ(conn->uri->scheme, "qemu"))
            return VIR_DRV_OPEN_DECLINED;

        /* Allow remote driver to deal with URIs with hostname server */
        if (conn->uri->server != NULL)
            return VIR_DRV_OPEN_DECLINED;

4156
        if (qemu_driver == NULL) {
4157 4158
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("qemu state driver is not active"));
4159 4160 4161
            return VIR_DRV_OPEN_ERROR;
        }

4162
        if (conn->uri->path == NULL) {
4163 4164 4165 4166 4167
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("no QEMU URI path given, try %s"),
                            qemu_driver->privileged
                            ? "qemu:///system"
                            : "qemu:///session");
4168 4169 4170
                return VIR_DRV_OPEN_ERROR;
        }

4171
        if (qemu_driver->privileged) {
4172 4173
            if (STRNEQ (conn->uri->path, "/system") &&
                STRNEQ (conn->uri->path, "/session")) {
4174 4175 4176
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("unexpected QEMU URI path '%s', try qemu:///system"),
                                conn->uri->path);
4177 4178 4179 4180
                return VIR_DRV_OPEN_ERROR;
            }
        } else {
            if (STRNEQ (conn->uri->path, "/session")) {
4181 4182 4183
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("unexpected QEMU URI path '%s', try qemu:///session"),
                                conn->uri->path);
4184 4185 4186
                return VIR_DRV_OPEN_ERROR;
            }
        }
4187 4188 4189 4190 4191 4192 4193
    }
    conn->privateData = qemu_driver;

    return VIR_DRV_OPEN_SUCCESS;
}

static int qemudClose(virConnectPtr conn) {
4194
    struct qemud_driver *driver = conn->privateData;
4195 4196

    /* Get rid of callbacks registered for this conn */
4197
    qemuDriverLock(driver);
4198
    virDomainEventCallbackListRemoveConn(conn, driver->domainEventCallbacks);
4199
    qemuDriverUnlock(driver);
4200 4201 4202 4203 4204 4205

    conn->privateData = NULL;

    return 0;
}

D
Daniel Veillard 已提交
4206 4207 4208 4209 4210
/* Which features are supported by this driver? */
static int
qemudSupportsFeature (virConnectPtr conn ATTRIBUTE_UNUSED, int feature)
{
    switch (feature) {
4211 4212 4213 4214 4215
    case VIR_DRV_FEATURE_MIGRATION_V2:
    case VIR_DRV_FEATURE_MIGRATION_P2P:
        return 1;
    default:
        return 0;
D
Daniel Veillard 已提交
4216 4217 4218
    }
}

4219
static const char *qemudGetType(virConnectPtr conn ATTRIBUTE_UNUSED) {
4220
    return "QEMU";
4221 4222
}

4223

4224 4225 4226 4227 4228 4229 4230 4231 4232 4233 4234 4235 4236
static int qemuIsSecure(virConnectPtr conn ATTRIBUTE_UNUSED)
{
    /* Trivially secure, since always inside the daemon */
    return 1;
}

static int qemuIsEncrypted(virConnectPtr conn ATTRIBUTE_UNUSED)
{
    /* Not encrypted, but remote driver takes care of that */
    return 0;
}


4237 4238 4239 4240
static int kvmGetMaxVCPUs(void) {
    int maxvcpus = 1;

    int r, fd;
4241

4242 4243
    fd = open(KVM_DEVICE, O_RDONLY);
    if (fd < 0) {
4244
        virReportSystemError(errno, _("Unable to open %s"), KVM_DEVICE);
4245
        return -1;
4246 4247 4248 4249 4250 4251 4252 4253 4254 4255 4256
    }

    r = ioctl(fd, KVM_CHECK_EXTENSION, KVM_CAP_NR_VCPUS);
    if (r > 0)
        maxvcpus = r;

    close(fd);
    return maxvcpus;
}


4257
static int qemudGetMaxVCPUs(virConnectPtr conn ATTRIBUTE_UNUSED, const char *type) {
4258 4259 4260
    if (!type)
        return 16;

4261
    if (STRCASEEQ(type, "qemu"))
4262 4263
        return 16;

4264
    if (STRCASEEQ(type, "kvm"))
4265
        return kvmGetMaxVCPUs();
4266

4267
    if (STRCASEEQ(type, "kqemu"))
4268
        return 1;
4269

4270 4271
    qemuReportError(VIR_ERR_INVALID_ARG,
                    _("unknown type '%s'"), type);
4272 4273 4274
    return -1;
}

4275

4276
static char *qemudGetCapabilities(virConnectPtr conn) {
4277
    struct qemud_driver *driver = conn->privateData;
4278
    virCapsPtr caps = NULL;
4279
    char *xml = NULL;
4280

4281
    qemuDriverLock(driver);
4282

4283
    if ((caps = qemuCreateCapabilities(qemu_driver->caps,
4284
                                       qemu_driver)) == NULL) {
4285 4286 4287
        virCapabilitiesFree(caps);
        goto cleanup;
    }
4288

4289
    virCapabilitiesFree(qemu_driver->caps);
4290 4291 4292
    qemu_driver->caps = caps;

    if ((xml = virCapabilitiesFormatXML(driver->caps)) == NULL)
4293
        virReportOOMError();
4294 4295

cleanup:
4296
    qemuDriverUnlock(driver);
4297

4298
    return xml;
4299 4300 4301
}


4302
static int qemudGetProcessInfo(unsigned long long *cpuTime, int *lastCpu, int pid, int tid) {
D
Daniel P. Berrange 已提交
4303 4304
    char proc[PATH_MAX];
    FILE *pidinfo;
4305
    unsigned long long usertime, systime;
4306 4307
    int cpu;
    int ret;
D
Daniel P. Berrange 已提交
4308

4309 4310 4311 4312 4313 4314
    if (tid)
        ret = snprintf(proc, sizeof(proc), "/proc/%d/task/%d/stat", pid, tid);
    else
        ret = snprintf(proc, sizeof(proc), "/proc/%d/stat", pid);
    if (ret >= (int)sizeof(proc)) {
        errno = E2BIG;
D
Daniel P. Berrange 已提交
4315 4316 4317 4318
        return -1;
    }

    if (!(pidinfo = fopen(proc, "r"))) {
4319
        /*printf("cannot read pid info");*/
D
Daniel P. Berrange 已提交
4320
        /* VM probably shut down, so fake 0 */
4321 4322 4323 4324
        if (cpuTime)
            *cpuTime = 0;
        if (lastCpu)
            *lastCpu = 0;
D
Daniel P. Berrange 已提交
4325 4326 4327
        return 0;
    }

4328 4329 4330 4331 4332 4333 4334 4335 4336 4337
    /* See 'man proc' for information about what all these fields are. We're
     * only interested in a very few of them */
    if (fscanf(pidinfo,
               /* pid -> stime */
               "%*d %*s %*c %*d %*d %*d %*d %*d %*u %*u %*u %*u %*u %llu %llu"
               /* cutime -> endcode */
               "%*d %*d %*d %*d %*d %*u %*u %*d %*u %*u %*u %*u"
               /* startstack -> processor */
               "%*u %*u %*u %*u %*u %*u %*u %*u %*u %*u %*d %d",
               &usertime, &systime, &cpu) != 3) {
4338
        fclose(pidinfo);
4339 4340
        VIR_WARN0("cannot parse process status data");
        errno = -EINVAL;
D
Daniel P. Berrange 已提交
4341 4342 4343 4344 4345 4346 4347 4348
        return -1;
    }

    /* We got jiffies
     * We want nanoseconds
     * _SC_CLK_TCK is jiffies per second
     * So calulate thus....
     */
4349 4350 4351 4352 4353
    if (cpuTime)
        *cpuTime = 1000ull * 1000ull * 1000ull * (usertime + systime) / (unsigned long long)sysconf(_SC_CLK_TCK);
    if (lastCpu)
        *lastCpu = cpu;

D
Daniel P. Berrange 已提交
4354

4355 4356
    VIR_DEBUG("Got status for %d/%d user=%llu sys=%llu cpu=%d",
              pid, tid, usertime, systime, cpu);
D
Daniel P. Berrange 已提交
4357 4358 4359 4360 4361 4362 4363

    fclose(pidinfo);

    return 0;
}


4364
static virDomainPtr qemudDomainLookupByID(virConnectPtr conn,
4365
                                          int id) {
4366 4367 4368 4369
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;

4370
    qemuDriverLock(driver);
4371
    vm  = virDomainFindByID(&driver->domains, id);
4372
    qemuDriverUnlock(driver);
4373 4374

    if (!vm) {
4375 4376
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching id %d"), id);
4377
        goto cleanup;
4378 4379
    }

4380
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
4381
    if (dom) dom->id = vm->def->id;
4382 4383

cleanup:
4384 4385
    if (vm)
        virDomainObjUnlock(vm);
4386 4387
    return dom;
}
4388

4389
static virDomainPtr qemudDomainLookupByUUID(virConnectPtr conn,
4390
                                            const unsigned char *uuid) {
4391 4392 4393
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
4394

4395
    qemuDriverLock(driver);
4396
    vm = virDomainFindByUUID(&driver->domains, uuid);
4397 4398
    qemuDriverUnlock(driver);

4399
    if (!vm) {
4400 4401
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(uuid, uuidstr);
4402 4403
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4404
        goto cleanup;
4405 4406
    }

4407
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
4408
    if (dom) dom->id = vm->def->id;
4409 4410

cleanup:
4411 4412
    if (vm)
        virDomainObjUnlock(vm);
4413 4414
    return dom;
}
4415

4416
static virDomainPtr qemudDomainLookupByName(virConnectPtr conn,
4417
                                            const char *name) {
4418 4419 4420
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
4421

4422
    qemuDriverLock(driver);
4423
    vm = virDomainFindByName(&driver->domains, name);
4424 4425
    qemuDriverUnlock(driver);

4426
    if (!vm) {
4427 4428
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching name '%s'"), name);
4429
        goto cleanup;
4430 4431
    }

4432
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
4433
    if (dom) dom->id = vm->def->id;
4434 4435

cleanup:
4436 4437
    if (vm)
        virDomainObjUnlock(vm);
4438 4439 4440
    return dom;
}

4441 4442 4443 4444 4445 4446 4447 4448 4449 4450 4451

static int qemuDomainIsActive(virDomainPtr dom)
{
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr obj;
    int ret = -1;

    qemuDriverLock(driver);
    obj = virDomainFindByUUID(&driver->domains, dom->uuid);
    qemuDriverUnlock(driver);
    if (!obj) {
4452
        qemuReportError(VIR_ERR_NO_DOMAIN, NULL);
4453 4454 4455 4456 4457 4458 4459 4460 4461 4462 4463 4464 4465 4466 4467 4468 4469 4470 4471 4472
        goto cleanup;
    }
    ret = virDomainObjIsActive(obj);

cleanup:
    if (obj)
        virDomainObjUnlock(obj);
    return ret;
}

static int qemuDomainIsPersistent(virDomainPtr dom)
{
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr obj;
    int ret = -1;

    qemuDriverLock(driver);
    obj = virDomainFindByUUID(&driver->domains, dom->uuid);
    qemuDriverUnlock(driver);
    if (!obj) {
4473
        qemuReportError(VIR_ERR_NO_DOMAIN, NULL);
4474 4475 4476 4477 4478 4479 4480 4481 4482 4483 4484
        goto cleanup;
    }
    ret = obj->persistent;

cleanup:
    if (obj)
        virDomainObjUnlock(obj);
    return ret;
}


4485
static int qemudGetVersion(virConnectPtr conn, unsigned long *version) {
4486 4487 4488
    struct qemud_driver *driver = conn->privateData;
    int ret = -1;

4489
    qemuDriverLock(driver);
4490
    if (qemudExtractVersion(driver) < 0)
4491
        goto cleanup;
4492

4493
    *version = qemu_driver->qemuVersion;
4494 4495 4496
    ret = 0;

cleanup:
4497
    qemuDriverUnlock(driver);
4498
    return ret;
D
Daniel P. Berrange 已提交
4499 4500
}

4501
static int qemudListDomains(virConnectPtr conn, int *ids, int nids) {
4502
    struct qemud_driver *driver = conn->privateData;
4503
    int n;
4504

4505
    qemuDriverLock(driver);
4506
    n = virDomainObjListGetActiveIDs(&driver->domains, ids, nids);
4507
    qemuDriverUnlock(driver);
4508

4509
    return n;
D
Daniel P. Berrange 已提交
4510
}
4511

4512
static int qemudNumDomains(virConnectPtr conn) {
4513
    struct qemud_driver *driver = conn->privateData;
4514
    int n;
4515

4516
    qemuDriverLock(driver);
4517
    n = virDomainObjListNumOfDomains(&driver->domains, 1);
4518
    qemuDriverUnlock(driver);
4519

4520
    return n;
D
Daniel P. Berrange 已提交
4521
}
4522

4523
static virDomainPtr qemudDomainCreate(virConnectPtr conn, const char *xml,
4524
                                      unsigned int flags) {
4525
    struct qemud_driver *driver = conn->privateData;
4526
    virDomainDefPtr def;
4527
    virDomainObjPtr vm = NULL;
4528
    virDomainPtr dom = NULL;
4529
    virDomainEventPtr event = NULL;
D
Daniel P. Berrange 已提交
4530

4531
    virCheckFlags(VIR_DOMAIN_START_PAUSED, NULL);
4532

4533
    qemuDriverLock(driver);
4534
    if (!(def = virDomainDefParseString(driver->caps, xml,
4535
                                        VIR_DOMAIN_XML_INACTIVE)))
4536
        goto cleanup;
4537

4538
    if (virSecurityDriverVerify(def) < 0)
4539 4540
        goto cleanup;

4541 4542
    if (virDomainObjIsDuplicate(&driver->domains, def, 1) < 0)
        goto cleanup;
4543

4544 4545 4546 4547 4548 4549
    if (qemudCanonicalizeMachine(driver, def) < 0)
        goto cleanup;

    if (qemuAssignPCIAddresses(def) < 0)
        goto cleanup;

4550
    if (!(vm = virDomainAssignDef(driver->caps,
4551
                                  &driver->domains,
4552
                                  def, false)))
4553 4554 4555
        goto cleanup;

    def = NULL;
D
Daniel P. Berrange 已提交
4556

4557 4558 4559
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup; /* XXXX free the 'vm' we created ? */

4560 4561 4562
    if (qemudStartVMDaemon(conn, driver, vm, NULL,
                           (flags & VIR_DOMAIN_START_PAUSED) != 0,
                           -1, NULL) < 0) {
4563 4564 4565
        if (qemuDomainObjEndJob(vm) > 0)
            virDomainRemoveInactive(&driver->domains,
                                    vm);
4566
        vm = NULL;
4567
        goto cleanup;
D
Daniel P. Berrange 已提交
4568
    }
4569 4570 4571 4572

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_BOOTED);
D
Daniel P. Berrange 已提交
4573

4574
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
4575
    if (dom) dom->id = vm->def->id;
4576

4577 4578 4579
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4580

4581 4582
cleanup:
    virDomainDefFree(def);
4583 4584
    if (vm)
        virDomainObjUnlock(vm);
4585 4586
    if (event)
        qemuDomainEventQueue(driver, event);
4587
    qemuDriverUnlock(driver);
4588
    return dom;
D
Daniel P. Berrange 已提交
4589 4590 4591
}


4592
static int qemudDomainSuspend(virDomainPtr dom) {
4593 4594 4595
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4596
    virDomainEventPtr event = NULL;
4597
    qemuDomainObjPrivatePtr priv;
4598

4599
    qemuDriverLock(driver);
4600
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4601

D
Daniel P. Berrange 已提交
4602
    if (!vm) {
4603 4604
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4605 4606
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4607
        goto cleanup;
D
Daniel P. Berrange 已提交
4608
    }
D
Daniel P. Berrange 已提交
4609
    if (!virDomainObjIsActive(vm)) {
4610 4611
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
4612
        goto cleanup;
D
Daniel P. Berrange 已提交
4613
    }
4614 4615 4616

    priv = vm->privateData;

4617
    if (priv->jobActive == QEMU_JOB_MIGRATION_OUT) {
4618 4619 4620 4621 4622 4623 4624 4625 4626 4627 4628 4629 4630 4631
        if (vm->state != VIR_DOMAIN_PAUSED) {
            VIR_DEBUG("Requesting domain pause on %s",
                      vm->def->name);
            priv->jobSignals |= QEMU_JOB_SIGNAL_SUSPEND;
        }
        ret = 0;
        goto cleanup;
    } else {
        if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
            goto cleanup;

        if (!virDomainObjIsActive(vm)) {
            qemuReportError(VIR_ERR_OPERATION_INVALID,
                            "%s", _("domain is not running"));
4632
            goto endjob;
4633
        }
4634 4635
        if (vm->state != VIR_DOMAIN_PAUSED) {
            int rc;
4636
            int state = vm->state;
4637

4638
            vm->state = VIR_DOMAIN_PAUSED;
4639 4640 4641
            qemuDomainObjEnterMonitorWithDriver(driver, vm);
            rc = qemuMonitorStopCPUs(priv->mon);
            qemuDomainObjExitMonitorWithDriver(driver, vm);
4642 4643
            if (rc < 0) {
                vm->state = state;
4644
                goto endjob;
4645
            }
4646 4647 4648 4649 4650 4651 4652
            event = virDomainEventNewFromObj(vm,
                                             VIR_DOMAIN_EVENT_SUSPENDED,
                                             VIR_DOMAIN_EVENT_SUSPENDED_PAUSED);
        }
        if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
            goto endjob;
        ret = 0;
D
Daniel P. Berrange 已提交
4653
    }
4654

4655
endjob:
4656 4657
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4658

4659
cleanup:
4660 4661
    if (vm)
        virDomainObjUnlock(vm);
4662

4663
    if (event)
4664
        qemuDomainEventQueue(driver, event);
4665
    qemuDriverUnlock(driver);
4666
    return ret;
D
Daniel P. Berrange 已提交
4667 4668 4669
}


4670
static int qemudDomainResume(virDomainPtr dom) {
4671 4672 4673
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4674
    virDomainEventPtr event = NULL;
4675

4676
    qemuDriverLock(driver);
4677
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4678

D
Daniel P. Berrange 已提交
4679
    if (!vm) {
4680 4681
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4682 4683
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4684
        goto cleanup;
D
Daniel P. Berrange 已提交
4685
    }
4686 4687 4688 4689

    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
4690
    if (!virDomainObjIsActive(vm)) {
4691 4692
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
4693
        goto endjob;
D
Daniel P. Berrange 已提交
4694
    }
4695
    if (vm->state == VIR_DOMAIN_PAUSED) {
4696
        qemuDomainObjPrivatePtr priv = vm->privateData;
4697
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
4698
        if (qemuMonitorStartCPUs(priv->mon, dom->conn) < 0) {
4699
            qemuDomainObjExitMonitorWithDriver(driver, vm);
4700
            if (virGetLastError() == NULL)
4701 4702
                qemuReportError(VIR_ERR_OPERATION_FAILED,
                                "%s", _("resume operation failed"));
4703
            goto endjob;
4704
        }
4705
        qemuDomainObjExitMonitorWithDriver(driver, vm);
4706
        vm->state = VIR_DOMAIN_RUNNING;
4707 4708 4709
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_UNPAUSED);
D
Daniel P. Berrange 已提交
4710
    }
4711
    if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
4712
        goto endjob;
4713 4714
    ret = 0;

4715
endjob:
4716 4717
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4718

4719
cleanup:
4720 4721
    if (vm)
        virDomainObjUnlock(vm);
4722
    if (event)
4723
        qemuDomainEventQueue(driver, event);
4724
    qemuDriverUnlock(driver);
4725
    return ret;
D
Daniel P. Berrange 已提交
4726 4727 4728
}


4729
static int qemudDomainShutdown(virDomainPtr dom) {
4730 4731 4732
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4733

4734
    qemuDriverLock(driver);
4735
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4736 4737
    qemuDriverUnlock(driver);

4738
    if (!vm) {
4739 4740
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4741 4742
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4743
        goto cleanup;
4744 4745
    }

4746 4747 4748
    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
4749
    if (!virDomainObjIsActive(vm)) {
4750 4751
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
4752
        goto endjob;
4753 4754
    }

4755
    qemuDomainObjPrivatePtr priv = vm->privateData;
4756 4757 4758
    qemuDomainObjEnterMonitor(vm);
    ret = qemuMonitorSystemPowerdown(priv->mon);
    qemuDomainObjExitMonitor(vm);
4759

4760
endjob:
4761 4762
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4763

4764
cleanup:
4765 4766
    if (vm)
        virDomainObjUnlock(vm);
4767
    return ret;
4768 4769 4770
}


4771
static int qemudDomainDestroy(virDomainPtr dom) {
4772 4773 4774
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4775
    virDomainEventPtr event = NULL;
4776

4777
    qemuDriverLock(driver);
4778
    vm  = virDomainFindByUUID(&driver->domains, dom->uuid);
D
Daniel P. Berrange 已提交
4779
    if (!vm) {
4780 4781
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4782 4783
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4784
        goto cleanup;
D
Daniel P. Berrange 已提交
4785
    }
4786 4787 4788 4789

    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
4790
    if (!virDomainObjIsActive(vm)) {
4791 4792
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
4793
        goto endjob;
4794
    }
4795

4796
    qemudShutdownVMDaemon(driver, vm, 0);
4797 4798 4799
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_DESTROYED);
4800
    if (!vm->persistent) {
4801 4802 4803
        if (qemuDomainObjEndJob(vm) > 0)
            virDomainRemoveInactive(&driver->domains,
                                    vm);
4804 4805
        vm = NULL;
    }
4806 4807
    ret = 0;

4808
endjob:
4809 4810 4811
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4812

4813
cleanup:
4814 4815
    if (vm)
        virDomainObjUnlock(vm);
4816 4817
    if (event)
        qemuDomainEventQueue(driver, event);
4818
    qemuDriverUnlock(driver);
4819
    return ret;
D
Daniel P. Berrange 已提交
4820 4821 4822
}


4823
static char *qemudDomainGetOSType(virDomainPtr dom) {
4824 4825 4826
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    char *type = NULL;
4827

4828
    qemuDriverLock(driver);
4829
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4830
    qemuDriverUnlock(driver);
4831
    if (!vm) {
4832 4833
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4834 4835
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4836
        goto cleanup;
4837 4838
    }

4839
    if (!(type = strdup(vm->def->os.type)))
4840
        virReportOOMError();
4841 4842

cleanup:
4843 4844
    if (vm)
        virDomainObjUnlock(vm);
4845 4846 4847
    return type;
}

4848 4849
/* Returns max memory in kb, 0 if error */
static unsigned long qemudDomainGetMaxMemory(virDomainPtr dom) {
4850 4851 4852
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    unsigned long ret = 0;
4853

4854
    qemuDriverLock(driver);
4855
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4856 4857
    qemuDriverUnlock(driver);

4858
    if (!vm) {
4859 4860
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4861 4862
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4863
        goto cleanup;
4864 4865
    }

4866 4867 4868
    ret = vm->def->maxmem;

cleanup:
4869 4870
    if (vm)
        virDomainObjUnlock(vm);
4871
    return ret;
4872 4873 4874
}

static int qemudDomainSetMemory(virDomainPtr dom, unsigned long newmem) {
4875
    struct qemud_driver *driver = dom->conn->privateData;
4876
    qemuDomainObjPrivatePtr priv;
4877
    virDomainObjPtr vm;
4878
    int ret = -1, r;
4879

4880
    qemuDriverLock(driver);
4881
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4882
    qemuDriverUnlock(driver);
4883
    if (!vm) {
4884 4885
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4886 4887
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4888
        goto cleanup;
4889 4890
    }

4891 4892 4893 4894 4895 4896
    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
        goto cleanup;
    }

4897
    if (newmem > vm->def->maxmem) {
4898 4899
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("cannot set memory higher than max memory"));
4900
        goto cleanup;
4901 4902
    }

4903 4904 4905
    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

4906 4907 4908 4909 4910 4911
    priv = vm->privateData;
    qemuDomainObjEnterMonitor(vm);
    r = qemuMonitorSetBalloon(priv->mon, newmem);
    qemuDomainObjExitMonitor(vm);
    if (r < 0)
        goto endjob;
4912

4913 4914 4915 4916 4917
    /* Lack of balloon support is a fatal error */
    if (r == 0) {
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        "%s", _("cannot set memory of an active domain"));
        goto endjob;
4918
    }
4919

4920
    ret = 0;
4921
endjob:
4922 4923
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4924

4925
cleanup:
4926 4927
    if (vm)
        virDomainObjUnlock(vm);
4928
    return ret;
4929 4930
}

4931
static int qemudDomainGetInfo(virDomainPtr dom,
4932
                              virDomainInfoPtr info) {
4933 4934 4935
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4936 4937
    int err;
    unsigned long balloon;
4938

4939
    qemuDriverLock(driver);
4940
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4941
    qemuDriverUnlock(driver);
D
Daniel P. Berrange 已提交
4942
    if (!vm) {
4943 4944
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4945 4946
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4947
        goto cleanup;
D
Daniel P. Berrange 已提交
4948 4949
    }

4950
    info->state = vm->state;
D
Daniel P. Berrange 已提交
4951

D
Daniel P. Berrange 已提交
4952
    if (!virDomainObjIsActive(vm)) {
4953
        info->cpuTime = 0;
D
Daniel P. Berrange 已提交
4954
    } else {
4955
        if (qemudGetProcessInfo(&(info->cpuTime), NULL, vm->pid, 0) < 0) {
4956
            qemuReportError(VIR_ERR_OPERATION_FAILED, ("cannot read cputime for domain"));
4957
            goto cleanup;
D
Daniel P. Berrange 已提交
4958 4959 4960
        }
    }

4961
    info->maxMem = vm->def->maxmem;
4962

D
Daniel P. Berrange 已提交
4963
    if (virDomainObjIsActive(vm)) {
4964
        qemuDomainObjPrivatePtr priv = vm->privateData;
4965 4966 4967 4968 4969

        if ((vm->def->memballoon != NULL) &&
            (vm->def->memballoon->model == VIR_DOMAIN_MEMBALLOON_MODEL_NONE)) {
            info->memory = vm->def->maxmem;
        } else if (!priv->jobActive) {
4970 4971
            if (qemuDomainObjBeginJob(vm) < 0)
                goto cleanup;
4972

4973 4974 4975 4976
            qemuDomainObjEnterMonitor(vm);
            err = qemuMonitorGetBalloonInfo(priv->mon, &balloon);
            qemuDomainObjExitMonitor(vm);
            if (err < 0) {
4977 4978
                if (qemuDomainObjEndJob(vm) == 0)
                    vm = NULL;
4979 4980 4981 4982 4983 4984 4985 4986 4987
                goto cleanup;
            }

            if (err == 0)
                /* Balloon not supported, so maxmem is always the allocation */
                info->memory = vm->def->maxmem;
            else
                info->memory = balloon;

4988 4989 4990 4991
            if (qemuDomainObjEndJob(vm) == 0) {
                vm = NULL;
                goto cleanup;
            }
4992 4993 4994
        } else {
            info->memory = vm->def->memory;
        }
4995 4996 4997 4998
    } else {
        info->memory = vm->def->memory;
    }

4999
    info->nrVirtCpu = vm->def->vcpus;
5000 5001 5002
    ret = 0;

cleanup:
5003 5004
    if (vm)
        virDomainObjUnlock(vm);
5005
    return ret;
D
Daniel P. Berrange 已提交
5006 5007 5008
}


5009 5010 5011 5012 5013 5014 5015 5016
/** qemuDomainMigrateOffline:
 * Pause domain for non-live migration.
 */
static int
qemuDomainMigrateOffline(struct qemud_driver *driver,
                         virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
5017
    int state = vm->state;
5018 5019
    int ret;

5020
    vm->state = VIR_DOMAIN_PAUSED;
5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
    ret = qemuMonitorStopCPUs(priv->mon);
    qemuDomainObjExitMonitorWithDriver(driver, vm);

    if (ret == 0) {
        virDomainEventPtr event;

        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_SUSPENDED,
                                         VIR_DOMAIN_EVENT_SUSPENDED_MIGRATED);
        if (event)
            qemuDomainEventQueue(driver, event);
5033 5034
    } else
        vm->state = state;
5035 5036 5037 5038 5039

    return ret;
}


5040 5041 5042 5043 5044 5045 5046 5047 5048 5049 5050 5051 5052 5053
static int
qemuDomainWaitForMigrationComplete(struct qemud_driver *driver, virDomainObjPtr vm)
{
    int ret = -1;
    int status;
    unsigned long long memProcessed;
    unsigned long long memRemaining;
    unsigned long long memTotal;
    qemuDomainObjPrivatePtr priv = vm->privateData;

    priv->jobInfo.type = VIR_DOMAIN_JOB_UNBOUNDED;

    while (priv->jobInfo.type == VIR_DOMAIN_JOB_UNBOUNDED) {
        /* Poll every 50ms for progress & to allow cancellation */
5054
        struct timespec ts = { .tv_sec = 0, .tv_nsec = 50 * 1000 * 1000ull };
5055 5056 5057
        struct timeval now;
        int rc;

5058 5059 5060 5061 5062 5063
        if (!virDomainObjIsActive(vm)) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("guest unexpectedly quit during migration"));
            goto cleanup;
        }

5064 5065
        if (priv->jobSignals & QEMU_JOB_SIGNAL_CANCEL) {
            priv->jobSignals ^= QEMU_JOB_SIGNAL_CANCEL;
5066 5067 5068 5069 5070 5071 5072
            VIR_DEBUG0("Cancelling migration at client request");
            qemuDomainObjEnterMonitorWithDriver(driver, vm);
            rc = qemuMonitorMigrateCancel(priv->mon);
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            if (rc < 0) {
                VIR_WARN0("Unable to cancel migration");
            }
5073 5074 5075 5076 5077
        } else if (priv->jobSignals & QEMU_JOB_SIGNAL_SUSPEND) {
            priv->jobSignals ^= QEMU_JOB_SIGNAL_SUSPEND;
            VIR_DEBUG0("Pausing domain for non-live migration");
            if (qemuDomainMigrateOffline(driver, vm) < 0)
                VIR_WARN0("Unable to pause domain");
5078 5079 5080 5081 5082 5083 5084 5085 5086 5087 5088
        } else if (priv->jobSignals & QEMU_JOB_SIGNAL_MIGRATE_DOWNTIME) {
            unsigned long long ms = priv->jobSignalsData.migrateDowntime;

            priv->jobSignals ^= QEMU_JOB_SIGNAL_MIGRATE_DOWNTIME;
            priv->jobSignalsData.migrateDowntime = 0;
            VIR_DEBUG("Setting migration downtime to %llums", ms);
            qemuDomainObjEnterMonitorWithDriver(driver, vm);
            rc = qemuMonitorSetMigrationDowntime(priv->mon, ms);
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            if (rc < 0)
                VIR_WARN0("Unable to set migration downtime");
5089 5090
        }

5091 5092 5093 5094 5095 5096 5097 5098 5099
        /* Repeat check because the job signals might have caused
         * guest to die
         */
        if (!virDomainObjIsActive(vm)) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("guest unexpectedly quit during migration"));
            goto cleanup;
        }

5100 5101 5102 5103 5104 5105 5106 5107 5108 5109 5110 5111 5112 5113 5114 5115 5116 5117 5118 5119 5120 5121 5122 5123 5124 5125 5126 5127 5128 5129 5130 5131 5132 5133 5134 5135 5136 5137 5138 5139 5140 5141 5142 5143 5144 5145 5146 5147 5148 5149 5150 5151 5152 5153 5154 5155 5156 5157 5158 5159 5160 5161 5162 5163 5164 5165 5166 5167 5168 5169 5170 5171
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
        rc = qemuMonitorGetMigrationStatus(priv->mon,
                                           &status,
                                           &memProcessed,
                                           &memRemaining,
                                           &memTotal);
        qemuDomainObjExitMonitorWithDriver(driver, vm);

        if (rc < 0) {
            priv->jobInfo.type = VIR_DOMAIN_JOB_FAILED;
            goto cleanup;
        }

        if (gettimeofday(&now, NULL) < 0) {
            priv->jobInfo.type = VIR_DOMAIN_JOB_FAILED;
            virReportSystemError(errno, "%s",
                                 _("cannot get time of day"));
            goto cleanup;
        }
        priv->jobInfo.timeElapsed =
            ((now.tv_sec * 1000ull) + (now.tv_usec / 1000)) -
            priv->jobStart;

        switch (status) {
        case QEMU_MONITOR_MIGRATION_STATUS_INACTIVE:
            priv->jobInfo.type = VIR_DOMAIN_JOB_NONE;
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            "%s", _("Migration is not active"));
            break;

        case QEMU_MONITOR_MIGRATION_STATUS_ACTIVE:
            priv->jobInfo.dataTotal = memTotal;
            priv->jobInfo.dataRemaining = memRemaining;
            priv->jobInfo.dataProcessed = memProcessed;

            priv->jobInfo.memTotal = memTotal;
            priv->jobInfo.memRemaining = memRemaining;
            priv->jobInfo.memProcessed = memProcessed;
            break;

        case QEMU_MONITOR_MIGRATION_STATUS_COMPLETED:
            priv->jobInfo.type = VIR_DOMAIN_JOB_COMPLETED;
            ret = 0;
            break;

        case QEMU_MONITOR_MIGRATION_STATUS_ERROR:
            priv->jobInfo.type = VIR_DOMAIN_JOB_FAILED;
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            "%s", _("Migration unexpectedly failed"));
            break;

        case QEMU_MONITOR_MIGRATION_STATUS_CANCELLED:
            priv->jobInfo.type = VIR_DOMAIN_JOB_CANCELLED;
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            "%s", _("Migration was cancelled by client"));
            break;
        }

        virDomainObjUnlock(vm);
        qemuDriverUnlock(driver);

        nanosleep(&ts, NULL);

        qemuDriverLock(driver);
        virDomainObjLock(vm);
    }

cleanup:
    return ret;
}


5172
#define QEMUD_SAVE_MAGIC "LibvirtQemudSave"
5173 5174 5175
#define QEMUD_SAVE_VERSION 2

enum qemud_save_formats {
5176 5177 5178
    QEMUD_SAVE_FORMAT_RAW = 0,
    QEMUD_SAVE_FORMAT_GZIP = 1,
    QEMUD_SAVE_FORMAT_BZIP2 = 2,
5179 5180
    /*
     * Deprecated by xz and never used as part of a release
5181
     * QEMUD_SAVE_FORMAT_LZMA
5182 5183
     */
    QEMUD_SAVE_FORMAT_XZ = 3,
5184
    QEMUD_SAVE_FORMAT_LZOP = 4,
5185 5186 5187
    /* Note: add new members only at the end.
       These values are used in the on-disk format.
       Do not change or re-use numbers. */
5188 5189

    QEMUD_SAVE_FORMAT_LAST
5190
};
5191

5192 5193 5194 5195 5196
VIR_ENUM_DECL(qemudSaveCompression)
VIR_ENUM_IMPL(qemudSaveCompression, QEMUD_SAVE_FORMAT_LAST,
              "raw",
              "gzip",
              "bzip2",
5197 5198
              "xz",
              "lzop")
5199

5200 5201 5202 5203 5204
struct qemud_save_header {
    char magic[sizeof(QEMUD_SAVE_MAGIC)-1];
    int version;
    int xml_len;
    int was_running;
5205 5206
    int compressed;
    int unused[15];
5207 5208
};

5209 5210 5211 5212 5213 5214 5215
struct fileOpHookData {
    virDomainPtr dom;
    const char *path;
    char *xml;
    struct qemud_save_header *header;
};

5216
/* return -errno on failure, or 0 on success */
5217 5218 5219 5220 5221
static int qemudDomainSaveFileOpHook(int fd, void *data) {
    struct fileOpHookData *hdata = data;
    int ret = 0;

    if (safewrite(fd, hdata->header, sizeof(*hdata->header)) != sizeof(*hdata->header)) {
5222
        ret = -errno;
5223
        qemuReportError(VIR_ERR_OPERATION_FAILED,
5224 5225
                        _("failed to write header to domain save file '%s'"),
                        hdata->path);
5226 5227 5228 5229
        goto endjob;
    }

    if (safewrite(fd, hdata->xml, hdata->header->xml_len) != hdata->header->xml_len) {
5230
        ret = -errno;
5231 5232 5233 5234 5235 5236 5237 5238 5239
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                         _("failed to write xml to '%s'"), hdata->path);
        goto endjob;
    }
endjob:
    return ret;
}


5240 5241
static int qemudDomainSaveFlag(virDomainPtr dom, const char *path,
                               int compressed)
5242
{
5243
    struct qemud_driver *driver = dom->conn->privateData;
5244
    virDomainObjPtr vm = NULL;
5245
    char *xml = NULL;
5246
    struct qemud_save_header header;
5247 5248
    struct fileOpHookData hdata;
    int bypassSecurityDriver = 0;
5249
    int ret = -1;
5250
    int rc;
5251
    virDomainEventPtr event = NULL;
5252
    qemuDomainObjPrivatePtr priv;
5253 5254
    struct stat sb;
    int is_reg = 0;
5255
    unsigned long long offset;
5256
    virCgroupPtr cgroup = NULL;
5257 5258 5259 5260 5261

    memset(&header, 0, sizeof(header));
    memcpy(header.magic, QEMUD_SAVE_MAGIC, sizeof(header.magic));
    header.version = QEMUD_SAVE_VERSION;

5262
    qemuDriverLock(driver);
5263 5264

    header.compressed = compressed;
5265

5266
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5267

D
Daniel P. Berrange 已提交
5268
    if (!vm) {
5269 5270
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
5271 5272
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
5273
        goto cleanup;
D
Daniel P. Berrange 已提交
5274
    }
5275
    priv = vm->privateData;
5276

5277 5278 5279
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
5280
    if (!virDomainObjIsActive(vm)) {
5281 5282
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
5283
        goto endjob;
D
Daniel P. Berrange 已提交
5284
    }
5285

5286 5287 5288
    memset(&priv->jobInfo, 0, sizeof(priv->jobInfo));
    priv->jobInfo.type = VIR_DOMAIN_JOB_UNBOUNDED;

5289 5290 5291
    /* Pause */
    if (vm->state == VIR_DOMAIN_RUNNING) {
        header.was_running = 1;
5292
        vm->state = VIR_DOMAIN_PAUSED;
5293
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
5294
        if (qemuMonitorStopCPUs(priv->mon) < 0) {
5295
            qemuDomainObjExitMonitorWithDriver(driver, vm);
5296
            vm->state = VIR_DOMAIN_RUNNING;
5297
            goto endjob;
5298
        }
5299
        qemuDomainObjExitMonitorWithDriver(driver, vm);
5300 5301 5302 5303 5304 5305

        if (!virDomainObjIsActive(vm)) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("guest unexpectedly quit"));
            goto endjob;
        }
5306 5307 5308
    }

    /* Get XML for the domain */
5309
    xml = virDomainDefFormat(vm->def, VIR_DOMAIN_XML_SECURE);
5310
    if (!xml) {
5311 5312
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to get domain xml"));
5313
        goto endjob;
5314 5315 5316
    }
    header.xml_len = strlen(xml) + 1;

5317 5318 5319 5320 5321 5322 5323 5324 5325 5326 5327 5328 5329
    /* path might be a pre-existing block dev, in which case
     * we need to skip the create step, and also avoid unlink
     * in the failure case */
    if (stat(path, &sb) < 0) {
        /* Avoid throwing an error here, since it is possible
         * that with NFS we can't actually stat() the file.
         * The subsequent codepaths will still raise an error
         * if a truely fatal problem is hit */
        is_reg = 1;
    } else {
        is_reg = S_ISREG(sb.st_mode);
    }

5330 5331 5332 5333 5334
    offset = sizeof(header) + header.xml_len;

    /* Due to way we append QEMU state on our header with dd,
     * we need to ensure there's a 512 byte boundary. Unfortunately
     * we don't have an explicit offset in the header, so we fake
5335 5336
     * it by padding the XML string with NULLs.
     */
5337 5338 5339 5340 5341 5342 5343 5344 5345 5346 5347 5348 5349
    if (offset % QEMU_MONITOR_MIGRATE_TO_FILE_BS) {
        unsigned long long pad =
            QEMU_MONITOR_MIGRATE_TO_FILE_BS -
            (offset % QEMU_MONITOR_MIGRATE_TO_FILE_BS);

        if (VIR_REALLOC_N(xml, header.xml_len + pad) < 0) {
            virReportOOMError();
            goto endjob;
        }
        memset(xml + header.xml_len, 0, pad);
        offset += pad;
        header.xml_len += pad;
    }
5350

5351 5352 5353 5354 5355 5356
    /* Setup hook data needed by virFileOperation hook function */
    hdata.dom = dom;
    hdata.path = path;
    hdata.xml = xml;
    hdata.header = &header;

5357 5358
    /* Write header to file, followed by XML */

5359
    /* First try creating the file as root */
5360 5361 5362 5363
    if (!is_reg) {
        int fd = open(path, O_WRONLY | O_TRUNC);
        if (fd < 0) {
            virReportSystemError(errno, _("unable to open %s"), path);
5364 5365
            goto endjob;
        }
5366
        if (qemudDomainSaveFileOpHook(fd, &hdata) < 0) {
5367 5368 5369 5370 5371
            close(fd);
            goto endjob;
        }
        if (close(fd) < 0) {
            virReportSystemError(errno, _("unable to close %s"), path);
5372 5373
            goto endjob;
        }
5374 5375 5376 5377 5378
    } else {
        if ((rc = virFileOperation(path, O_CREAT|O_TRUNC|O_WRONLY,
                                  S_IRUSR|S_IWUSR,
                                  getuid(), getgid(),
                                  qemudDomainSaveFileOpHook, &hdata,
5379
                                  0)) < 0) {
5380 5381 5382 5383 5384 5385 5386
            /* If we failed as root, and the error was permission-denied
               (EACCES), assume it's on a network-connected share where
               root access is restricted (eg, root-squashed NFS). If the
               qemu user (driver->user) is non-root, just set a flag to
               bypass security driver shenanigans, and retry the operation
               after doing setuid to qemu user */

5387
            if ((rc != -EACCES) ||
5388
                driver->user == getuid()) {
5389
                virReportSystemError(-rc, _("Failed to create domain save file '%s'"),
5390 5391 5392
                                     path);
                goto endjob;
            }
5393

5394
            /* On Linux we can also verify the FS-type of the directory. */
5395 5396 5397 5398 5399 5400 5401 5402 5403 5404 5405 5406 5407 5408 5409 5410 5411 5412
            switch (virStorageFileIsSharedFS(path)) {
                case 1:
                   /* it was on a network share, so we'll continue
                    * as outlined above
                    */
                   break;

                case -1:
                   virReportSystemError(errno,
                                        _("Failed to create domain save file "
                                          "'%s': couldn't determine fs type"),
                                        path);
                   goto endjob;
                   break;

                case 0:
                default:
                   /* local file - log the error returned by virFileOperation */
5413
                   virReportSystemError(-rc,
5414 5415 5416 5417
                                        _("Failed to create domain save file '%s'"),
                                        path);
                   goto endjob;
                   break;
5418 5419 5420

            }

5421
            /* Retry creating the file as driver->user */
5422

5423 5424 5425 5426
            if ((rc = virFileOperation(path, O_CREAT|O_TRUNC|O_WRONLY,
                                       S_IRUSR|S_IWUSR|S_IRGRP|S_IWGRP,
                                       driver->user, driver->group,
                                       qemudDomainSaveFileOpHook, &hdata,
5427 5428
                                       VIR_FILE_OP_AS_UID)) < 0) {
                virReportSystemError(-rc, _("Error from child process creating '%s'"),
5429
                                 path);
5430 5431
                goto endjob;
            }
5432

5433 5434 5435
            /* Since we had to setuid to create the file, and the fstype
               is NFS, we assume it's a root-squashing NFS share, and that
               the security driver stuff would have failed anyway */
5436

5437 5438
            bypassSecurityDriver = 1;
        }
5439
    }
5440

5441

5442 5443 5444 5445 5446 5447 5448 5449 5450 5451 5452 5453 5454 5455 5456 5457 5458
    if (!is_reg &&
        qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_DEVICES)) {
        if (virCgroupForDomain(driver->cgroup, vm->def->name, &cgroup, 0) != 0) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("Unable to find cgroup for %s\n"),
                            vm->def->name);
            goto endjob;
        }
        rc = virCgroupAllowDevicePath(cgroup, path);
        if (rc != 0) {
            virReportSystemError(-rc,
                                 _("Unable to allow device %s for %s"),
                                 path, vm->def->name);
            goto endjob;
        }
    }

5459 5460
    if ((!bypassSecurityDriver) &&
        driver->securityDriver &&
5461
        driver->securityDriver->domainSetSavedStateLabel &&
5462 5463
        driver->securityDriver->domainSetSavedStateLabel(driver->securityDriver,
                                                         vm, path) == -1)
5464 5465
        goto endjob;

5466 5467
    if (header.compressed == QEMUD_SAVE_FORMAT_RAW) {
        const char *args[] = { "cat", NULL };
M
Matthias Bolte 已提交
5468
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
5469 5470 5471
        rc = qemuMonitorMigrateToFile(priv->mon,
                                      QEMU_MONITOR_MIGRATE_BACKGROUND,
                                      args, path, offset);
M
Matthias Bolte 已提交
5472
        qemuDomainObjExitMonitorWithDriver(driver, vm);
5473
    } else {
5474
        const char *prog = qemudSaveCompressionTypeToString(header.compressed);
5475 5476 5477 5478 5479
        const char *args[] = {
            prog,
            "-c",
            NULL
        };
M
Matthias Bolte 已提交
5480
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
5481 5482 5483
        rc = qemuMonitorMigrateToFile(priv->mon,
                                      QEMU_MONITOR_MIGRATE_BACKGROUND,
                                      args, path, offset);
M
Matthias Bolte 已提交
5484
        qemuDomainObjExitMonitorWithDriver(driver, vm);
5485 5486
    }

5487 5488 5489 5490 5491
    if (rc < 0)
        goto endjob;

    rc = qemuDomainWaitForMigrationComplete(driver, vm);

5492
    if (rc < 0)
5493
        goto endjob;
5494

5495 5496
    if ((!bypassSecurityDriver) &&
        driver->securityDriver &&
5497
        driver->securityDriver->domainRestoreSavedStateLabel &&
5498 5499
        driver->securityDriver->domainRestoreSavedStateLabel(driver->securityDriver,
                                                             vm, path) == -1)
5500
        VIR_WARN("failed to restore save state label on %s", path);
5501

5502 5503
    if (cgroup != NULL) {
        rc = virCgroupDenyDevicePath(cgroup, path);
5504 5505 5506
        if (rc != 0)
            VIR_WARN("Unable to deny device %s for %s %d",
                     path, vm->def->name, rc);
5507 5508
    }

5509 5510
    ret = 0;

5511
    /* Shut it down */
5512
    qemudShutdownVMDaemon(driver, vm, 0);
5513 5514 5515
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_SAVED);
5516
    if (!vm->persistent) {
5517 5518 5519
        if (qemuDomainObjEndJob(vm) > 0)
            virDomainRemoveInactive(&driver->domains,
                                    vm);
5520 5521
        vm = NULL;
    }
5522

5523
endjob:
5524
    if (vm) {
5525
        if (ret != 0) {
5526
            if (header.was_running && virDomainObjIsActive(vm)) {
5527 5528 5529 5530 5531 5532 5533 5534
                qemuDomainObjEnterMonitorWithDriver(driver, vm);
                rc = qemuMonitorStartCPUs(priv->mon, dom->conn);
                qemuDomainObjExitMonitorWithDriver(driver, vm);
                if (rc < 0)
                    VIR_WARN0("Unable to resume guest CPUs after save failure");
                else
                    vm->state = VIR_DOMAIN_RUNNING;
            }
5535

5536 5537 5538 5539 5540
            if (cgroup != NULL) {
                rc = virCgroupDenyDevicePath(cgroup, path);
                if (rc != 0)
                    VIR_WARN("Unable to deny device %s for %s: %d",
                             path, vm->def->name, rc);
5541
            }
5542 5543 5544 5545

            if ((!bypassSecurityDriver) &&
                driver->securityDriver &&
                driver->securityDriver->domainRestoreSavedStateLabel &&
5546 5547
                driver->securityDriver->domainRestoreSavedStateLabel(driver->securityDriver,
                                                                     vm, path) == -1)
5548
                VIR_WARN("failed to restore save state label on %s", path);
5549 5550
        }

5551
        if (qemuDomainObjEndJob(vm) == 0)
5552
            vm = NULL;
5553
    }
5554

5555 5556
cleanup:
    VIR_FREE(xml);
5557
    if (ret != 0 && is_reg)
5558
        unlink(path);
5559 5560
    if (vm)
        virDomainObjUnlock(vm);
5561 5562
    if (event)
        qemuDomainEventQueue(driver, event);
5563
    virCgroupFree(&cgroup);
5564
    qemuDriverUnlock(driver);
5565
    return ret;
D
Daniel P. Berrange 已提交
5566 5567
}

5568 5569 5570 5571 5572 5573 5574 5575 5576 5577 5578 5579 5580 5581 5582 5583 5584 5585 5586 5587 5588 5589 5590 5591 5592 5593 5594 5595 5596 5597 5598 5599 5600 5601 5602 5603 5604 5605 5606 5607 5608 5609
static int qemudDomainSave(virDomainPtr dom, const char *path)
{
    struct qemud_driver *driver = dom->conn->privateData;
    int compressed;

    /* Hm, is this safe against qemudReload? */
    if (driver->saveImageFormat == NULL)
        compressed = QEMUD_SAVE_FORMAT_RAW;
    else {
        compressed = qemudSaveCompressionTypeFromString(driver->saveImageFormat);
        if (compressed < 0) {
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            "%s", _("Invalid save image format specified "
                                    "in configuration file"));
            return -1;
        }
    }

    return qemudDomainSaveFlag(dom, path, compressed);
}

static char *
qemuDomainManagedSavePath(struct qemud_driver *driver, virDomainObjPtr vm) {
    char *ret;

    if (virAsprintf(&ret, "%s/%s.save", driver->saveDir, vm->def->name) < 0) {
        virReportOOMError();
        return(NULL);
    }

    return(ret);
}

static int
qemuDomainManagedSave(virDomainPtr dom, unsigned int flags)
{
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm = NULL;
    char *name = NULL;
    int ret = -1;
    int compressed;

5610
    virCheckFlags(0, -1);
5611 5612 5613 5614 5615 5616 5617 5618 5619 5620 5621 5622 5623 5624 5625 5626 5627 5628 5629 5630 5631 5632 5633 5634 5635 5636 5637 5638 5639 5640 5641 5642 5643 5644 5645 5646 5647 5648 5649 5650 5651 5652 5653 5654 5655 5656 5657 5658 5659 5660 5661 5662 5663 5664 5665 5666 5667 5668 5669 5670

    qemuDriverLock(driver);
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
        goto error;
    }

    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
        goto error;
    }

    name = qemuDomainManagedSavePath(driver, vm);
    if (name == NULL)
        goto error;

    VIR_DEBUG("Saving state to %s", name);

    /* FIXME: we should take the flags parameter, and use bits out
     * of there to control whether we are compressing or not
     */
    compressed = QEMUD_SAVE_FORMAT_RAW;

    /* we have to drop our locks here because qemudDomainSaveFlag is
     * going to pick them back up.  Unfortunately it opens a race window
     * between us dropping and qemudDomainSaveFlag picking it back up, but
     * if we want to allow other operations to be able to happen while
     * qemuDomainSaveFlag is running (possibly for a long time), I'm not
     * sure if there is a better solution
     */
    virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);

    ret = qemudDomainSaveFlag(dom, name, compressed);

cleanup:
    VIR_FREE(name);

    return ret;

error:
    if (vm)
        virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);
    goto cleanup;
}

static int
qemuDomainHasManagedSaveImage(virDomainPtr dom, unsigned int flags)
{
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm = NULL;
    int ret = -1;
    char *name = NULL;

5671
    virCheckFlags(0, -1);
5672 5673 5674 5675 5676 5677 5678 5679 5680 5681 5682 5683 5684 5685 5686 5687 5688 5689 5690 5691 5692 5693 5694 5695 5696 5697 5698 5699 5700 5701 5702 5703 5704

    qemuDriverLock(driver);
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
        goto cleanup;
    }

    name = qemuDomainManagedSavePath(driver, vm);
    if (name == NULL)
        goto cleanup;

    ret = virFileExists(name);

cleanup:
    VIR_FREE(name);
    if (vm)
        virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);
    return ret;
}

static int
qemuDomainManagedSaveRemove(virDomainPtr dom, unsigned int flags)
{
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm = NULL;
    int ret = -1;
    char *name = NULL;

5705
    virCheckFlags(0, -1);
5706 5707 5708 5709 5710 5711 5712 5713 5714 5715 5716 5717 5718 5719 5720 5721 5722 5723 5724 5725 5726 5727 5728 5729

    qemuDriverLock(driver);
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
        goto cleanup;
    }

    name = qemuDomainManagedSavePath(driver, vm);
    if (name == NULL)
        goto cleanup;

    ret = unlink(name);

cleanup:
    VIR_FREE(name);
    if (vm)
        virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);
    return ret;
}
D
Daniel P. Berrange 已提交
5730

P
Paolo Bonzini 已提交
5731 5732 5733 5734 5735 5736
static int qemudDomainCoreDump(virDomainPtr dom,
                               const char *path,
                               int flags ATTRIBUTE_UNUSED) {
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int resume = 0, paused = 0;
5737
    int ret = -1, fd = -1;
5738
    virDomainEventPtr event = NULL;
5739 5740 5741 5742
    const char *args[] = {
        "cat",
        NULL,
    };
5743
    qemuDomainObjPrivatePtr priv;
P
Paolo Bonzini 已提交
5744 5745 5746 5747 5748 5749 5750

    qemuDriverLock(driver);
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);

    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
5751 5752
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
P
Paolo Bonzini 已提交
5753 5754
        goto cleanup;
    }
5755
    priv = vm->privateData;
P
Paolo Bonzini 已提交
5756

5757
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
5758 5759
        goto cleanup;

D
Daniel P. Berrange 已提交
5760
    if (!virDomainObjIsActive(vm)) {
5761 5762
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
5763
        goto endjob;
P
Paolo Bonzini 已提交
5764 5765
    }

5766 5767
    /* Create an empty file with appropriate ownership.  */
    if ((fd = open(path, O_CREAT|O_TRUNC|O_WRONLY, S_IRUSR|S_IWUSR)) < 0) {
5768 5769
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("failed to create '%s'"), path);
5770 5771 5772 5773
        goto endjob;
    }

    if (close(fd) < 0) {
5774
        virReportSystemError(errno,
5775 5776 5777 5778 5779 5780 5781
                             _("unable to save file %s"),
                             path);
        goto endjob;
    }

    if (driver->securityDriver &&
        driver->securityDriver->domainSetSavedStateLabel &&
5782 5783
        driver->securityDriver->domainSetSavedStateLabel(driver->securityDriver,
                                                         vm, path) == -1)
5784 5785
        goto endjob;

P
Paolo Bonzini 已提交
5786 5787
    /* Migrate will always stop the VM, so the resume condition is
       independent of whether the stop command is issued.  */
P
Paolo Bonzini 已提交
5788 5789 5790
    resume = (vm->state == VIR_DOMAIN_RUNNING);

    /* Pause domain for non-live dump */
P
Paolo Bonzini 已提交
5791
    if (!(flags & VIR_DUMP_LIVE) && vm->state == VIR_DOMAIN_RUNNING) {
5792
        vm->state = VIR_DOMAIN_PAUSED;
5793
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
5794
        if (qemuMonitorStopCPUs(priv->mon) < 0) {
5795
            qemuDomainObjExitMonitorWithDriver(driver, vm);
5796
            vm->state = VIR_DOMAIN_RUNNING;
5797
            goto endjob;
5798
        }
5799
        qemuDomainObjExitMonitorWithDriver(driver, vm);
P
Paolo Bonzini 已提交
5800
        paused = 1;
5801 5802 5803 5804 5805 5806

        if (!virDomainObjIsActive(vm)) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("guest unexpectedly quit"));
            goto endjob;
        }
P
Paolo Bonzini 已提交
5807 5808
    }

5809
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
5810 5811 5812
    ret = qemuMonitorMigrateToFile(priv->mon,
                                   QEMU_MONITOR_MIGRATE_BACKGROUND,
                                   args, path, 0);
5813
    qemuDomainObjExitMonitorWithDriver(driver, vm);
5814 5815 5816 5817 5818 5819 5820 5821 5822
    if (ret < 0)
        goto endjob;

    ret = qemuDomainWaitForMigrationComplete(driver, vm);

    if (ret < 0)
        goto endjob;

    paused = 1;
5823 5824 5825

    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSavedStateLabel &&
5826 5827
        driver->securityDriver->domainRestoreSavedStateLabel(driver->securityDriver,
                                                             vm, path) == -1)
5828 5829 5830
        goto endjob;

endjob:
5831
    if ((ret == 0) && (flags & VIR_DUMP_CRASH)) {
5832
        qemudShutdownVMDaemon(driver, vm, 0);
5833 5834 5835 5836 5837
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         VIR_DOMAIN_EVENT_STOPPED_CRASHED);
    }

P
Paolo Bonzini 已提交
5838 5839 5840
    /* Since the monitor is always attached to a pty for libvirt, it
       will support synchronous operations so we always get here after
       the migration is complete.  */
5841
    else if (resume && paused && virDomainObjIsActive(vm)) {
5842
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
5843
        if (qemuMonitorStartCPUs(priv->mon, dom->conn) < 0) {
5844
            if (virGetLastError() == NULL)
5845 5846
                qemuReportError(VIR_ERR_OPERATION_FAILED,
                                "%s", _("resuming after dump failed"));
P
Paolo Bonzini 已提交
5847
        }
5848
        qemuDomainObjExitMonitorWithDriver(driver, vm);
5849
        vm->state = VIR_DOMAIN_RUNNING;
P
Paolo Bonzini 已提交
5850
    }
5851

5852 5853
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
5854
    else if ((ret == 0) && (flags & VIR_DUMP_CRASH) && !vm->persistent) {
5855 5856 5857 5858
        virDomainRemoveInactive(&driver->domains,
                                vm);
        vm = NULL;
    }
5859 5860

cleanup:
5861 5862
    if (ret != 0)
        unlink(path);
P
Paolo Bonzini 已提交
5863 5864
    if (vm)
        virDomainObjUnlock(vm);
5865 5866
    if (event)
        qemuDomainEventQueue(driver, event);
5867
    qemuDriverUnlock(driver);
P
Paolo Bonzini 已提交
5868 5869 5870 5871
    return ret;
}


5872 5873 5874 5875 5876 5877
static int qemudDomainHotplugVcpus(virDomainObjPtr vm, unsigned int nvcpus)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int i, rc;
    int ret = -1;

5878 5879
    qemuDomainObjEnterMonitor(vm);

5880 5881 5882 5883 5884 5885 5886 5887 5888 5889 5890 5891 5892 5893 5894 5895 5896 5897 5898 5899 5900 5901 5902 5903 5904 5905 5906 5907 5908 5909
    /* We need different branches here, because we want to offline
     * in reverse order to onlining, so any partial fail leaves us in a
     * reasonably sensible state */
    if (nvcpus > vm->def->vcpus) {
        for (i = vm->def->vcpus ; i < nvcpus ; i++) {
            /* Online new CPU */
            rc = qemuMonitorSetCPU(priv->mon, i, 1);
            if (rc == 0)
                goto unsupported;
            if (rc < 0)
                goto cleanup;

            vm->def->vcpus++;
        }
    } else {
        for (i = vm->def->vcpus - 1 ; i >= nvcpus ; i--) {
            /* Offline old CPU */
            rc = qemuMonitorSetCPU(priv->mon, i, 0);
            if (rc == 0)
                goto unsupported;
            if (rc < 0)
                goto cleanup;

            vm->def->vcpus--;
        }
    }

    ret = 0;

cleanup:
5910
    qemuDomainObjExitMonitor(vm);
5911 5912 5913 5914 5915 5916 5917 5918 5919 5920
    return ret;

unsupported:
    qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                    _("cannot change vcpu count of this domain"));
    goto cleanup;
}


static int qemudDomainSetVcpus(virDomainPtr dom, unsigned int nvcpus) {
5921 5922
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
5923 5924
    const char * type;
    int max;
5925
    int ret = -1;
5926

5927
    qemuDriverLock(driver);
5928
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5929 5930
    qemuDriverUnlock(driver);

5931
    if (!vm) {
5932 5933
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
5934 5935
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
5936
        goto cleanup;
5937 5938
    }

5939 5940 5941
    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

5942 5943 5944
    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                         "%s", _("domain is not running"));
5945
        goto endjob;
5946 5947
    }

5948 5949 5950 5951 5952 5953 5954 5955 5956 5957 5958 5959 5960 5961 5962 5963 5964 5965 5966 5967 5968 5969 5970 5971 5972
    if (!(type = virDomainVirtTypeToString(vm->def->virtType))) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("unknown virt type in domain definition '%d'"),
                        vm->def->virtType);
        goto endjob;
    }

    if ((max = qemudGetMaxVCPUs(NULL, type)) < 0) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("could not determine max vcpus for the domain"));
        goto endjob;
    }

    if (nvcpus > max) {
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("requested vcpus is greater than max allowable"
                          " vcpus for the domain: %d > %d"), nvcpus, max);
        goto endjob;
    }

    ret = qemudDomainHotplugVcpus(vm, nvcpus);

endjob:
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
5973

5974
cleanup:
5975 5976
    if (vm)
        virDomainObjUnlock(vm);
5977
    return ret;
5978 5979
}

5980 5981 5982 5983 5984 5985

static int
qemudDomainPinVcpu(virDomainPtr dom,
                   unsigned int vcpu,
                   unsigned char *cpumap,
                   int maplen) {
5986 5987
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
5988
    int maxcpu, hostcpus;
5989
    virNodeInfo nodeinfo;
5990
    int ret = -1;
5991
    qemuDomainObjPrivatePtr priv;
5992

5993
    qemuDriverLock(driver);
5994
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5995 5996
    qemuDriverUnlock(driver);

5997 5998 5999
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
6000 6001
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
6002 6003 6004
        goto cleanup;
    }

D
Daniel P. Berrange 已提交
6005
    if (!virDomainObjIsActive(vm)) {
6006 6007
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s",_("cannot pin vcpus on an inactive domain"));
6008
        goto cleanup;
6009 6010
    }

6011 6012 6013
    priv = vm->privateData;

    if (vcpu > (priv->nvcpupids-1)) {
6014 6015 6016
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("vcpu number out of range %d > %d"),
                        vcpu, priv->nvcpupids);
6017
        goto cleanup;
6018 6019
    }

6020
    if (nodeGetInfo(dom->conn, &nodeinfo) < 0)
6021
        goto cleanup;
6022

6023
    hostcpus = VIR_NODEINFO_MAXCPUS(nodeinfo);
6024
    maxcpu = maplen * 8;
6025 6026
    if (maxcpu > hostcpus)
        maxcpu = hostcpus;
6027

6028 6029
    if (priv->vcpupids != NULL) {
        if (virProcessInfoSetAffinity(priv->vcpupids[vcpu],
6030
                                      cpumap, maplen, maxcpu) < 0)
6031
            goto cleanup;
6032
    } else {
6033 6034
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        "%s", _("cpu affinity is not supported"));
6035
        goto cleanup;
6036
    }
6037
    ret = 0;
6038

6039
cleanup:
6040 6041
    if (vm)
        virDomainObjUnlock(vm);
6042
    return ret;
6043 6044 6045 6046 6047 6048 6049 6050
}

static int
qemudDomainGetVcpus(virDomainPtr dom,
                    virVcpuInfoPtr info,
                    int maxinfo,
                    unsigned char *cpumaps,
                    int maplen) {
6051 6052
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
6053
    virNodeInfo nodeinfo;
6054
    int i, v, maxcpu, hostcpus;
6055
    int ret = -1;
6056
    qemuDomainObjPrivatePtr priv;
6057

6058
    qemuDriverLock(driver);
6059
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
6060 6061
    qemuDriverUnlock(driver);

6062 6063 6064
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
6065 6066
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
6067 6068 6069
        goto cleanup;
    }

D
Daniel P. Berrange 已提交
6070
    if (!virDomainObjIsActive(vm)) {
6071 6072 6073
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s",
                        _("cannot list vcpu pinning for an inactive domain"));
6074
        goto cleanup;
6075 6076
    }

6077 6078
    priv = vm->privateData;

6079
    if (nodeGetInfo(dom->conn, &nodeinfo) < 0)
6080
        goto cleanup;
6081

6082
    hostcpus = VIR_NODEINFO_MAXCPUS(nodeinfo);
6083
    maxcpu = maplen * 8;
6084 6085
    if (maxcpu > hostcpus)
        maxcpu = hostcpus;
6086 6087

    /* Clamp to actual number of vcpus */
6088 6089
    if (maxinfo > priv->nvcpupids)
        maxinfo = priv->nvcpupids;
6090

6091 6092 6093 6094 6095 6096
    if (maxinfo >= 1) {
        if (info != NULL) {
            memset(info, 0, sizeof(*info) * maxinfo);
            for (i = 0 ; i < maxinfo ; i++) {
                info[i].number = i;
                info[i].state = VIR_VCPU_RUNNING;
6097

6098
                if (priv->vcpupids != NULL &&
6099 6100 6101
                    qemudGetProcessInfo(&(info[i].cpuTime),
                                        &(info[i].cpu),
                                        vm->pid,
6102
                                        priv->vcpupids[i]) < 0) {
6103
                    virReportSystemError(errno, "%s",
6104 6105 6106
                                         _("cannot get vCPU placement & pCPU time"));
                    goto cleanup;
                }
6107
            }
6108 6109
        }

6110 6111
        if (cpumaps != NULL) {
            memset(cpumaps, 0, maplen * maxinfo);
6112
            if (priv->vcpupids != NULL) {
6113 6114 6115
                for (v = 0 ; v < maxinfo ; v++) {
                    unsigned char *cpumap = VIR_GET_CPUMAP(cpumaps, maplen, v);

6116
                    if (virProcessInfoGetAffinity(priv->vcpupids[v],
6117
                                                  cpumap, maplen, maxcpu) < 0)
6118
                        goto cleanup;
6119
                }
6120
            } else {
6121 6122
                qemuReportError(VIR_ERR_NO_SUPPORT,
                                "%s", _("cpu affinity is not available"));
6123
                goto cleanup;
6124 6125 6126
            }
        }
    }
6127
    ret = maxinfo;
6128

6129
cleanup:
6130 6131
    if (vm)
        virDomainObjUnlock(vm);
6132
    return ret;
6133 6134 6135
}


6136
static int qemudDomainGetMaxVcpus(virDomainPtr dom) {
6137 6138
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
6139
    const char *type;
6140
    int ret = -1;
6141

6142
    qemuDriverLock(driver);
6143
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
6144 6145
    qemuDriverUnlock(driver);

6146
    if (!vm) {
6147 6148
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
6149 6150
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
6151
        goto cleanup;
6152 6153
    }

6154
    if (!(type = virDomainVirtTypeToString(vm->def->virtType))) {
6155 6156 6157
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("unknown virt type in domain definition '%d'"),
                        vm->def->virtType);
6158
        goto cleanup;
6159 6160
    }

6161
    ret = qemudGetMaxVCPUs(NULL, type);
6162

6163
cleanup:
6164 6165
    if (vm)
        virDomainObjUnlock(vm);
6166 6167 6168
    return ret;
}

6169 6170 6171 6172 6173 6174 6175 6176 6177
static int qemudDomainGetSecurityLabel(virDomainPtr dom, virSecurityLabelPtr seclabel)
{
    struct qemud_driver *driver = (struct qemud_driver *)dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;

    qemuDriverLock(driver);
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);

6178 6179
    memset(seclabel, 0, sizeof(*seclabel));

6180 6181 6182
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
6183 6184
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
6185 6186 6187
        goto cleanup;
    }

6188
    if (!virDomainVirtTypeToString(vm->def->virtType)) {
6189 6190 6191
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("unknown virt type in domain definition '%d'"),
                        vm->def->virtType);
6192 6193 6194 6195 6196 6197 6198 6199 6200 6201 6202 6203 6204 6205 6206 6207 6208
        goto cleanup;
    }

    /*
     * Theoretically, the pid can be replaced during this operation and
     * return the label of a different process.  If atomicity is needed,
     * further validation will be required.
     *
     * Comment from Dan Berrange:
     *
     *   Well the PID as stored in the virDomainObjPtr can't be changed
     *   because you've got a locked object.  The OS level PID could have
     *   exited, though and in extreme circumstances have cycled through all
     *   PIDs back to ours. We could sanity check that our PID still exists
     *   after reading the label, by checking that our FD connecting to the
     *   QEMU monitor hasn't seen SIGHUP/ERR on poll().
     */
D
Daniel P. Berrange 已提交
6209
    if (virDomainObjIsActive(vm)) {
6210 6211 6212 6213 6214 6215 6216
        if (driver->securityDriver &&
            driver->securityDriver->domainGetSecurityProcessLabel &&
            driver->securityDriver->domainGetSecurityProcessLabel(driver->securityDriver,
                                                                  vm, seclabel) < 0) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            "%s", _("Failed to get security label"));
            goto cleanup;
6217 6218 6219 6220 6221 6222 6223 6224
        }
    }

    ret = 0;

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
6225
    qemuDriverUnlock(driver);
6226 6227 6228
    return ret;
}

6229 6230
static int qemudNodeGetSecurityModel(virConnectPtr conn,
                                     virSecurityModelPtr secmodel)
6231 6232 6233
{
    struct qemud_driver *driver = (struct qemud_driver *)conn->privateData;
    char *p;
6234
    int ret = 0;
6235

6236
    qemuDriverLock(driver);
6237
    if (!driver->securityPrimaryDriver) {
6238
        memset(secmodel, 0, sizeof (*secmodel));
6239 6240
        goto cleanup;
    }
6241

6242 6243
    p = driver->caps->host.secModel.model;
    if (strlen(p) >= VIR_SECURITY_MODEL_BUFLEN-1) {
6244 6245 6246
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("security model string exceeds max %d bytes"),
                        VIR_SECURITY_MODEL_BUFLEN-1);
6247 6248
        ret = -1;
        goto cleanup;
6249 6250 6251 6252 6253
    }
    strcpy(secmodel->model, p);

    p = driver->caps->host.secModel.doi;
    if (strlen(p) >= VIR_SECURITY_DOI_BUFLEN-1) {
6254 6255 6256
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("security DOI string exceeds max %d bytes"),
                        VIR_SECURITY_DOI_BUFLEN-1);
6257 6258
        ret = -1;
        goto cleanup;
6259 6260
    }
    strcpy(secmodel->doi, p);
6261 6262 6263 6264

cleanup:
    qemuDriverUnlock(driver);
    return ret;
6265 6266
}

6267 6268 6269 6270 6271 6272 6273 6274 6275 6276 6277 6278 6279 6280 6281 6282 6283 6284 6285 6286 6287 6288 6289 6290 6291 6292 6293 6294 6295 6296 6297 6298 6299 6300 6301 6302 6303 6304 6305 6306 6307 6308 6309 6310 6311 6312 6313 6314 6315 6316 6317 6318 6319 6320 6321 6322 6323 6324 6325 6326 6327 6328 6329 6330 6331 6332 6333 6334 6335 6336 6337 6338 6339 6340 6341 6342 6343 6344 6345 6346 6347 6348 6349 6350 6351 6352 6353 6354 6355 6356 6357 6358 6359 6360 6361 6362 6363 6364 6365 6366 6367 6368 6369 6370 6371 6372 6373 6374 6375 6376 6377 6378 6379 6380 6381 6382 6383 6384 6385 6386 6387 6388 6389 6390 6391 6392 6393 6394 6395 6396 6397 6398
/* qemudOpenAsUID() - pipe/fork/setuid/open a file, and return the
   pipe fd to caller, so that it can read from the file. Also return
   the pid of the child process, so the caller can wait for it to exit
   after it's finished reading (to avoid a zombie, if nothing
   else). */

static int qemudOpenAsUID(const char *path, uid_t uid, pid_t *child_pid) {
    int pipefd[2];
    int fd = -1;

    *child_pid = -1;

    if (pipe(pipefd) < 0) {
        virReportSystemError(errno,
                             _("failed to create pipe to read '%s'"),
                             path);
        pipefd[0] = pipefd[1] = -1;
        goto parent_cleanup;
    }

    int forkRet = virFork(child_pid);

    if (*child_pid < 0) {
        virReportSystemError(errno,
                             _("failed to fork child to read '%s'"),
                             path);
        goto parent_cleanup;
    }

    if (*child_pid > 0) {

        /* parent */

        /* parent doesn't need the write side of the pipe */
        close(pipefd[1]);
        pipefd[1] = -1;

        if (forkRet < 0) {
            virReportSystemError(errno,
                                 _("failed in parent after forking child to read '%s'"),
                                 path);
            goto parent_cleanup;
        }
        /* caller gets the read side of the pipe */
        fd = pipefd[0];
        pipefd[0] = -1;
parent_cleanup:
        if (pipefd[0] != -1)
            close(pipefd[0]);
        if (pipefd[1] != -1)
            close(pipefd[1]);
        if ((fd < 0) && (*child_pid > 0)) {
            /* a child process was started and subsequently an error
               occurred in the parent, so we need to wait for it to
               exit, but its status is inconsequential. */
            while ((waitpid(*child_pid, NULL, 0) == -1)
                   && (errno == EINTR)) {
                /* empty */
            }
            *child_pid = -1;
        }
        return fd;
    }

    /* child */

    /* setuid to the qemu user, then open the file, read it,
       and stuff it into the pipe for the parent process to
       read */
    int exit_code;
    char *buf = NULL;
    size_t bufsize = 1024 * 1024;
    int bytesread;

    /* child doesn't need the read side of the pipe */
    close(pipefd[0]);

    if (forkRet < 0) {
        exit_code = errno;
        virReportSystemError(errno,
                             _("failed in child after forking to read '%s'"),
                             path);
        goto child_cleanup;
    }

    if (setuid(uid) != 0) {
        exit_code = errno;
        virReportSystemError(errno,
                             _("cannot setuid(%d) to read '%s'"),
                             uid, path);
        goto child_cleanup;
    }
    if ((fd = open(path, O_RDONLY)) < 0) {
        exit_code = errno;
        virReportSystemError(errno,
                             _("cannot open '%s' as uid %d"),
                             path, uid);
        goto child_cleanup;
    }
    if (VIR_ALLOC_N(buf, bufsize) < 0) {
        exit_code = ENOMEM;
        virReportOOMError();
        goto child_cleanup;
    }

    /* read from fd and write to pipefd[1] until EOF */
    do {
        if ((bytesread = saferead(fd, buf, bufsize)) < 0) {
            exit_code = errno;
            virReportSystemError(errno,
                                 _("child failed reading from '%s'"),
                                 path);
            goto child_cleanup;
        }
        if (safewrite(pipefd[1], buf, bytesread) != bytesread) {
            exit_code = errno;
            virReportSystemError(errno, "%s",
                                 _("child failed writing to pipe"));
            goto child_cleanup;
        }
    } while (bytesread > 0);
    exit_code = 0;

child_cleanup:
    VIR_FREE(buf);
    if (fd != -1)
        close(fd);
    if (pipefd[1] != -1)
        close(pipefd[1]);
    _exit(exit_code);
}

J
Jiri Denemark 已提交
6399 6400 6401 6402 6403 6404 6405 6406 6407 6408 6409 6410 6411 6412 6413 6414 6415 6416 6417 6418 6419 6420 6421 6422 6423 6424 6425 6426
static int qemudDomainSaveImageClose(int fd, pid_t read_pid, int *status)
{
    int ret = 0;

    if (fd != -1)
        close(fd);

    if (read_pid != -1) {
        /* reap the process that read the file */
        while ((ret = waitpid(read_pid, status, 0)) == -1
               && errno == EINTR) {
            /* empty */
        }
    } else if (status) {
        *status = 0;
    }

    return ret;
}

static int ATTRIBUTE_NONNULL(3) ATTRIBUTE_NONNULL(4) ATTRIBUTE_NONNULL(5)
qemudDomainSaveImageOpen(struct qemud_driver *driver,
                                    const char *path,
                                    virDomainDefPtr *ret_def,
                                    struct qemud_save_header *ret_header,
                                    pid_t *ret_read_pid)
{
    int fd;
6427
    pid_t read_pid = -1;
6428
    struct qemud_save_header header;
J
Jiri Denemark 已提交
6429 6430
    char *xml = NULL;
    virDomainDefPtr def = NULL;
6431 6432

    if ((fd = open(path, O_RDONLY)) < 0) {
6433 6434 6435
        if ((driver->user == 0) || (getuid() != 0)) {
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            "%s", _("cannot read domain image"));
J
Jiri Denemark 已提交
6436
            goto error;
6437 6438 6439 6440 6441 6442 6443 6444
        }

        /* Opening as root failed, but qemu runs as a different user
           that might have better luck. Create a pipe, then fork a
           child process to run as the qemu user, which will hopefully
           have the necessary authority to read the file. */
        if ((fd = qemudOpenAsUID(path, driver->user, &read_pid)) < 0) {
            /* error already reported */
J
Jiri Denemark 已提交
6445
            goto error;
6446
        }
6447 6448 6449
    }

    if (saferead(fd, &header, sizeof(header)) != sizeof(header)) {
6450 6451
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to read qemu header"));
J
Jiri Denemark 已提交
6452
        goto error;
6453 6454 6455
    }

    if (memcmp(header.magic, QEMUD_SAVE_MAGIC, sizeof(header.magic)) != 0) {
6456 6457
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("image magic is incorrect"));
J
Jiri Denemark 已提交
6458
        goto error;
6459 6460 6461
    }

    if (header.version > QEMUD_SAVE_VERSION) {
6462 6463 6464
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("image version is not supported (%d > %d)"),
                        header.version, QEMUD_SAVE_VERSION);
J
Jiri Denemark 已提交
6465
        goto error;
6466 6467
    }

6468 6469 6470
    if (header.xml_len <= 0) {
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("invalid XML length: %d"), header.xml_len);
J
Jiri Denemark 已提交
6471
        goto error;
6472 6473
    }

6474 6475
    if (VIR_ALLOC_N(xml, header.xml_len) < 0) {
        virReportOOMError();
J
Jiri Denemark 已提交
6476
        goto error;
6477 6478 6479
    }

    if (saferead(fd, xml, header.xml_len) != header.xml_len) {
6480 6481
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to read XML"));
J
Jiri Denemark 已提交
6482
        goto error;
6483 6484 6485
    }

    /* Create a domain from this XML */
6486
    if (!(def = virDomainDefParseString(driver->caps, xml,
6487
                                        VIR_DOMAIN_XML_INACTIVE))) {
6488 6489
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to parse XML"));
J
Jiri Denemark 已提交
6490
        goto error;
6491 6492
    }

J
Jiri Denemark 已提交
6493
    VIR_FREE(xml);
6494

J
Jiri Denemark 已提交
6495 6496 6497
    *ret_def = def;
    *ret_header = header;
    *ret_read_pid = read_pid;
6498

J
Jiri Denemark 已提交
6499
    return fd;
6500

J
Jiri Denemark 已提交
6501 6502 6503 6504 6505 6506 6507 6508 6509 6510 6511 6512 6513 6514 6515 6516 6517 6518 6519 6520 6521 6522 6523 6524 6525 6526
error:
    virDomainDefFree(def);
    VIR_FREE(xml);
    qemudDomainSaveImageClose(fd, read_pid, NULL);

    return -1;
}

static int ATTRIBUTE_NONNULL(6)
qemudDomainSaveImageStartVM(virConnectPtr conn,
                            struct qemud_driver *driver,
                            virDomainObjPtr vm,
                            int fd,
                            pid_t read_pid,
                            const struct qemud_save_header *header,
                            const char *path)
{
    int ret = -1;
    virDomainEventPtr event;
    int intermediatefd = -1;
    pid_t intermediate_pid = -1;
    int childstat;
    int wait_ret;
    int status;

    if (header->version == 2) {
6527
        const char *intermediate_argv[3] = { NULL, "-dc", NULL };
J
Jiri Denemark 已提交
6528
        const char *prog = qemudSaveCompressionTypeToString(header->compressed);
6529
        if (prog == NULL) {
6530 6531
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("Invalid compressed save format %d"),
J
Jiri Denemark 已提交
6532 6533
                            header->compressed);
            goto out;
6534
        }
6535

J
Jiri Denemark 已提交
6536
        if (header->compressed != QEMUD_SAVE_FORMAT_RAW) {
6537
            intermediate_argv[0] = prog;
6538 6539
            intermediatefd = fd;
            fd = -1;
6540
            if (virExec(intermediate_argv, NULL, NULL,
6541
                        &intermediate_pid, intermediatefd, &fd, NULL, 0) < 0) {
6542 6543 6544
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("Failed to start decompression binary %s"),
                                intermediate_argv[0]);
J
Jiri Denemark 已提交
6545
                goto out;
6546 6547 6548
            }
        }
    }
J
Jiri Denemark 已提交
6549

6550
    /* Set the migration source and start it up. */
6551
    ret = qemudStartVMDaemon(conn, driver, vm, "stdio", true, fd, path);
J
Jiri Denemark 已提交
6552

6553 6554 6555
    if (intermediate_pid != -1) {
        /* Wait for intermediate process to exit */
        while (waitpid(intermediate_pid, &childstat, 0) == -1 &&
J
Jiri Denemark 已提交
6556 6557 6558
               errno == EINTR) {
            /* empty */
        }
6559 6560 6561
    }
    if (intermediatefd != -1)
        close(intermediatefd);
J
Jiri Denemark 已提交
6562 6563

    wait_ret = qemudDomainSaveImageClose(fd, read_pid, &status);
6564
    fd = -1;
6565 6566 6567 6568 6569 6570 6571 6572 6573 6574 6575 6576 6577 6578 6579 6580 6581 6582 6583 6584 6585
    if (read_pid != -1) {
        if (wait_ret == -1) {
            virReportSystemError(errno,
                                 _("failed to wait for process reading '%s'"),
                                 path);
            ret = -1;
        } else if (!WIFEXITED(status)) {
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("child process exited abnormally reading '%s'"),
                            path);
            ret = -1;
        } else {
            int exit_status = WEXITSTATUS(status);
            if (exit_status != 0) {
                virReportSystemError(exit_status,
                                     _("child process returned error reading '%s'"),
                                     path);
                ret = -1;
            }
        }
    }
J
Jiri Denemark 已提交
6586 6587 6588

    if (ret < 0)
        goto out;
6589

6590 6591 6592
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_RESTORED);
J
Jiri Denemark 已提交
6593 6594 6595
    if (event)
        qemuDomainEventQueue(driver, event);

6596

6597
    /* If it was running before, resume it now. */
J
Jiri Denemark 已提交
6598
    if (header->was_running) {
6599
        qemuDomainObjPrivatePtr priv = vm->privateData;
6600
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
6601
        if (qemuMonitorStartCPUs(priv->mon, conn) < 0) {
6602
            if (virGetLastError() == NULL)
6603 6604
                qemuReportError(VIR_ERR_OPERATION_FAILED,
                                "%s", _("failed to resume domain"));
6605
            qemuDomainObjExitMonitorWithDriver(driver,vm);
J
Jiri Denemark 已提交
6606
            goto out;
6607
        }
6608
        qemuDomainObjExitMonitorWithDriver(driver, vm);
6609
        vm->state = VIR_DOMAIN_RUNNING;
6610 6611
        if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0) {
            VIR_WARN("Failed to save status on vm %s", vm->def->name);
J
Jiri Denemark 已提交
6612
            goto out;
6613
        }
6614
    }
J
Jiri Denemark 已提交
6615

6616
    ret = 0;
6617

J
Jiri Denemark 已提交
6618
out:
6619 6620
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSavedStateLabel &&
6621 6622
        driver->securityDriver->domainRestoreSavedStateLabel(driver->securityDriver,
                                                             vm, path) == -1)
6623 6624
        VIR_WARN("failed to restore save state label on %s", path);

J
Jiri Denemark 已提交
6625 6626 6627 6628 6629 6630 6631 6632 6633 6634 6635 6636 6637 6638 6639 6640 6641 6642 6643 6644 6645 6646 6647 6648 6649
    return ret;
}

static int qemudDomainRestore(virConnectPtr conn,
                              const char *path) {
    struct qemud_driver *driver = conn->privateData;
    virDomainDefPtr def = NULL;
    virDomainObjPtr vm = NULL;
    int fd = -1;
    pid_t read_pid = -1;
    int ret = -1;
    struct qemud_save_header header;

    qemuDriverLock(driver);

    fd = qemudDomainSaveImageOpen(driver, path, &def, &header, &read_pid);
    if (fd < 0)
        goto cleanup;

    if (virDomainObjIsDuplicate(&driver->domains, def, 1) < 0)
        goto cleanup;

    if (!(vm = virDomainAssignDef(driver->caps,
                                  &driver->domains,
                                  def, true))) {
6650
        /* virDomainAssignDef already set the error */
J
Jiri Denemark 已提交
6651 6652 6653 6654 6655 6656 6657 6658 6659 6660 6661
        goto cleanup;
    }
    def = NULL;

    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

    ret = qemudDomainSaveImageStartVM(conn, driver, vm, fd,
                                      read_pid, &header, path);

    if (qemuDomainObjEndJob(vm) == 0)
6662
        vm = NULL;
J
Jiri Denemark 已提交
6663 6664 6665 6666
    else if (ret < 0 && !vm->persistent) {
        virDomainRemoveInactive(&driver->domains, vm);
        vm = NULL;
    }
6667

6668 6669
cleanup:
    virDomainDefFree(def);
J
Jiri Denemark 已提交
6670
    qemudDomainSaveImageClose(fd, read_pid, NULL);
6671 6672 6673
    if (vm)
        virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);
6674
    return ret;
D
Daniel P. Berrange 已提交
6675 6676
}

J
Jiri Denemark 已提交
6677 6678 6679 6680 6681 6682 6683 6684 6685 6686 6687 6688 6689 6690 6691 6692 6693 6694 6695 6696 6697 6698 6699 6700 6701 6702 6703 6704 6705 6706 6707 6708 6709 6710 6711 6712 6713 6714 6715 6716 6717
static int qemudDomainObjRestore(virConnectPtr conn,
                                 struct qemud_driver *driver,
                                 virDomainObjPtr vm,
                                 const char *path)
{
    virDomainDefPtr def = NULL;
    int fd = -1;
    pid_t read_pid = -1;
    int ret = -1;
    struct qemud_save_header header;

    fd = qemudDomainSaveImageOpen(driver, path, &def, &header, &read_pid);
    if (fd < 0)
        goto cleanup;

    if (STRNEQ(vm->def->name, def->name) ||
        memcmp(vm->def->uuid, def->uuid, VIR_UUID_BUFLEN)) {
        char vm_uuidstr[VIR_UUID_STRING_BUFLEN];
        char def_uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(vm->def->uuid, vm_uuidstr);
        virUUIDFormat(def->uuid, def_uuidstr);
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("cannot restore domain '%s' uuid %s from a file"
                          " which belongs to domain '%s' uuid %s"),
                        vm->def->name, vm_uuidstr,
                        def->name, def_uuidstr);
        goto cleanup;
    }

    virDomainObjAssignDef(vm, def, true);
    def = NULL;

    ret = qemudDomainSaveImageStartVM(conn, driver, vm, fd,
                                      read_pid, &header, path);

cleanup:
    virDomainDefFree(def);
    qemudDomainSaveImageClose(fd, read_pid, NULL);
    return ret;
}

D
Daniel P. Berrange 已提交
6718

6719 6720 6721 6722 6723 6724 6725 6726 6727 6728 6729 6730 6731 6732 6733 6734 6735 6736 6737 6738 6739 6740 6741 6742 6743 6744 6745 6746 6747 6748 6749 6750 6751 6752 6753 6754 6755 6756
static char *qemudVMDumpXML(struct qemud_driver *driver,
                            virDomainObjPtr vm,
                            int flags)
{
    char *ret = NULL;
    virCPUDefPtr cpu = NULL;
    virDomainDefPtr def;
    virCPUDefPtr def_cpu;

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


6757
static char *qemudDomainDumpXML(virDomainPtr dom,
6758
                                int flags) {
6759 6760 6761
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    char *ret = NULL;
6762 6763
    unsigned long balloon;
    int err;
6764

6765
    qemuDriverLock(driver);
6766
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
6767

D
Daniel P. Berrange 已提交
6768
    if (!vm) {
6769 6770
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
6771 6772
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
6773
        goto cleanup;
D
Daniel P. Berrange 已提交
6774 6775
    }

6776 6777 6778 6779
    /* Refresh current memory based on balloon info if supported */
    if ((vm->def->memballoon != NULL) &&
        (vm->def->memballoon->model != VIR_DOMAIN_MEMBALLOON_MODEL_NONE) &&
        (virDomainObjIsActive(vm))) {
6780
        qemuDomainObjPrivatePtr priv = vm->privateData;
6781 6782 6783
        /* Don't delay if someone's using the monitor, just use
         * existing most recent data instead */
        if (!priv->jobActive) {
6784
            if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
6785 6786
                goto cleanup;

6787
            qemuDomainObjEnterMonitorWithDriver(driver, vm);
6788
            err = qemuMonitorGetBalloonInfo(priv->mon, &balloon);
6789
            qemuDomainObjExitMonitorWithDriver(driver, vm);
6790 6791 6792 6793
            if (qemuDomainObjEndJob(vm) == 0) {
                vm = NULL;
                goto cleanup;
            }
6794 6795 6796 6797 6798 6799
            if (err < 0)
                goto cleanup;
            if (err > 0)
                vm->def->memory = balloon;
            /* err == 0 indicates no balloon support, so ignore it */
        }
6800
    }
6801

6802
    ret = qemudVMDumpXML(driver, vm, flags);
6803 6804

cleanup:
6805 6806
    if (vm)
        virDomainObjUnlock(vm);
6807
    qemuDriverUnlock(driver);
6808
    return ret;
D
Daniel P. Berrange 已提交
6809 6810 6811
}


6812 6813 6814 6815
static char *qemuDomainXMLFromNative(virConnectPtr conn,
                                     const char *format,
                                     const char *config,
                                     unsigned int flags ATTRIBUTE_UNUSED) {
6816
    struct qemud_driver *driver = conn->privateData;
6817 6818 6819 6820
    virDomainDefPtr def = NULL;
    char *xml = NULL;

    if (STRNEQ(format, QEMU_CONFIG_FORMAT_ARGV)) {
6821 6822
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("unsupported config type %s"), format);
6823 6824 6825
        goto cleanup;
    }

6826
    qemuDriverLock(driver);
6827
    def = qemuParseCommandLineString(driver->caps, config);
6828
    qemuDriverUnlock(driver);
6829 6830 6831
    if (!def)
        goto cleanup;

6832
    xml = virDomainDefFormat(def, VIR_DOMAIN_XML_INACTIVE);
6833 6834 6835 6836 6837 6838

cleanup:
    virDomainDefFree(def);
    return xml;
}

6839 6840 6841 6842 6843 6844
static char *qemuDomainXMLToNative(virConnectPtr conn,
                                   const char *format,
                                   const char *xmlData,
                                   unsigned int flags ATTRIBUTE_UNUSED) {
    struct qemud_driver *driver = conn->privateData;
    virDomainDefPtr def = NULL;
6845
    virDomainChrDef monConfig;
6846
    const char *emulator;
6847
    unsigned long long qemuCmdFlags;
6848 6849 6850 6851 6852 6853 6854 6855
    struct stat sb;
    const char **retargv = NULL;
    const char **retenv = NULL;
    const char **tmp;
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    char *ret = NULL;
    int i;

6856 6857
    qemuDriverLock(driver);

6858
    if (STRNEQ(format, QEMU_CONFIG_FORMAT_ARGV)) {
6859 6860
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("unsupported config type %s"), format);
6861 6862 6863
        goto cleanup;
    }

6864
    def = virDomainDefParseString(driver->caps, xmlData, 0);
6865 6866 6867 6868 6869 6870 6871 6872 6873 6874 6875 6876 6877 6878 6879 6880 6881 6882 6883 6884 6885 6886 6887 6888 6889 6890 6891 6892 6893 6894 6895 6896 6897 6898 6899 6900 6901 6902 6903 6904 6905 6906 6907 6908
    if (!def)
        goto cleanup;

    /* Since we're just exporting args, we can't do bridge/network
     * setups, since libvirt will normally create TAP devices
     * directly. We convert those configs into generic 'ethernet'
     * config and assume the user has suitable 'ifup-qemu' scripts
     */
    for (i = 0 ; i < def->nnets ; i++) {
        virDomainNetDefPtr net = def->nets[i];
        if (net->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
            VIR_FREE(net->data.network.name);

            memset(net, 0, sizeof *net);

            net->type = VIR_DOMAIN_NET_TYPE_ETHERNET;
            net->data.ethernet.dev = NULL;
            net->data.ethernet.script = NULL;
            net->data.ethernet.ipaddr = NULL;
        } else if (net->type == VIR_DOMAIN_NET_TYPE_BRIDGE) {
            char *brname = net->data.bridge.brname;
            char *script = net->data.bridge.script;
            char *ipaddr = net->data.bridge.ipaddr;

            memset(net, 0, sizeof *net);

            net->type = VIR_DOMAIN_NET_TYPE_ETHERNET;
            net->data.ethernet.dev = brname;
            net->data.ethernet.script = script;
            net->data.ethernet.ipaddr = ipaddr;
        }
    }
    for (i = 0 ; i < def->ngraphics ; i++) {
        if (def->graphics[i]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
            def->graphics[i]->data.vnc.autoport)
            def->graphics[i]->data.vnc.port = 5900;
    }
    emulator = def->emulator;

    /* Make sure the binary we are about to try exec'ing exists.
     * Technically we could catch the exec() failure, but that's
     * in a sub-process so its hard to feed back a useful error
     */
    if (stat(emulator, &sb) < 0) {
6909
        virReportSystemError(errno,
6910 6911 6912 6913 6914 6915 6916 6917
                             _("Cannot find QEMU binary %s"),
                             emulator);
        goto cleanup;
    }

    if (qemudExtractVersionInfo(emulator,
                                NULL,
                                &qemuCmdFlags) < 0) {
6918 6919 6920
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Cannot determine QEMU argv syntax %s"),
                        emulator);
6921 6922 6923
        goto cleanup;
    }

6924
    if (qemuPrepareMonitorChr(driver, &monConfig, def->name) < 0)
6925
        goto cleanup;
6926 6927

    if (qemudBuildCommandLine(conn, driver, def,
D
Daniel P. Berrange 已提交
6928
                              &monConfig, 0, qemuCmdFlags,
6929 6930
                              &retargv, &retenv,
                              NULL, NULL, /* Don't want it to create TAP devices */
C
Chris Lalancette 已提交
6931
                              NULL, NULL) < 0) {
6932 6933 6934 6935 6936 6937 6938 6939 6940 6941 6942 6943 6944 6945 6946 6947
        goto cleanup;
    }

    tmp = retenv;
    while (*tmp) {
        virBufferAdd(&buf, *tmp, strlen(*tmp));
        virBufferAddLit(&buf, " ");
        tmp++;
    }
    tmp = retargv;
    while (*tmp) {
        virBufferAdd(&buf, *tmp, strlen(*tmp));
        virBufferAddLit(&buf, " ");
        tmp++;
    }

6948 6949
    if (virBufferError(&buf)) {
        virBufferFreeAndReset(&buf);
6950
        virReportOOMError();
6951
        goto cleanup;
6952
    }
6953 6954 6955 6956

    ret = virBufferContentAndReset(&buf);

cleanup:
6957
    qemuDriverUnlock(driver);
6958 6959 6960 6961 6962 6963 6964 6965 6966 6967 6968 6969 6970
    for (tmp = retargv ; tmp && *tmp ; tmp++)
        VIR_FREE(*tmp);
    VIR_FREE(retargv);

    for (tmp = retenv ; tmp && *tmp ; tmp++)
        VIR_FREE(*tmp);
    VIR_FREE(retenv);

    virDomainDefFree(def);
    return ret;
}


6971
static int qemudListDefinedDomains(virConnectPtr conn,
6972
                            char **const names, int nnames) {
6973
    struct qemud_driver *driver = conn->privateData;
6974
    int n;
6975

6976
    qemuDriverLock(driver);
6977
    n = virDomainObjListGetInactiveNames(&driver->domains, names, nnames);
6978
    qemuDriverUnlock(driver);
6979
    return n;
D
Daniel P. Berrange 已提交
6980 6981
}

6982
static int qemudNumDefinedDomains(virConnectPtr conn) {
6983
    struct qemud_driver *driver = conn->privateData;
6984
    int n;
6985

6986
    qemuDriverLock(driver);
6987
    n = virDomainObjListNumOfDomains(&driver->domains, 0);
6988
    qemuDriverUnlock(driver);
6989

6990
    return n;
D
Daniel P. Berrange 已提交
6991 6992 6993
}


J
Jiri Denemark 已提交
6994 6995
static int qemudDomainObjStart(virConnectPtr conn,
                               struct qemud_driver *driver,
6996 6997
                               virDomainObjPtr vm,
                               bool start_paused)
J
Jiri Denemark 已提交
6998 6999 7000 7001 7002 7003 7004 7005 7006 7007 7008 7009 7010 7011 7012 7013 7014 7015 7016 7017
{
    int ret = -1;
    char *managed_save;

    /*
     * If there is a managed saved state restore it instead of starting
     * from scratch. In any case the old state is removed.
     */
    managed_save = qemuDomainManagedSavePath(driver, vm);
    if ((managed_save) && (virFileExists(managed_save))) {
        ret = qemudDomainObjRestore(conn, driver, vm, managed_save);

        if (unlink(managed_save) < 0) {
            VIR_WARN("Failed to remove the managed state %s", managed_save);
        }

        if (ret == 0)
            goto cleanup;
    }

7018
    ret = qemudStartVMDaemon(conn, driver, vm, NULL, start_paused, -1, NULL);
J
Jiri Denemark 已提交
7019 7020 7021 7022 7023 7024 7025 7026 7027 7028 7029 7030 7031 7032
    if (ret != -1) {
        virDomainEventPtr event =
            virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_BOOTED);
        if (event)
            qemuDomainEventQueue(driver, event);
    }

cleanup:
    VIR_FREE(managed_save);
    return ret;
}

7033 7034 7035
static int
qemudDomainStartWithFlags(virDomainPtr dom, unsigned int flags)
{
7036 7037 7038
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
7039

7040
    virCheckFlags(VIR_DOMAIN_START_PAUSED, -1);
7041

7042
    qemuDriverLock(driver);
7043
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
7044

7045
    if (!vm) {
7046 7047
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
7048 7049
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
7050
        goto cleanup;
7051 7052
    }

7053 7054 7055 7056
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

    if (virDomainObjIsActive(vm)) {
7057 7058
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is already running"));
7059 7060 7061
        goto endjob;
    }

7062 7063
    ret = qemudDomainObjStart(dom->conn, driver, vm,
                              (flags & VIR_DOMAIN_START_PAUSED) != 0);
7064

7065
endjob:
7066 7067
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
7068

7069
cleanup:
7070 7071
    if (vm)
        virDomainObjUnlock(vm);
7072
    qemuDriverUnlock(driver);
7073
    return ret;
D
Daniel P. Berrange 已提交
7074 7075
}

7076 7077 7078 7079 7080 7081
static int
qemudDomainStart(virDomainPtr dom)
{
    return qemudDomainStartWithFlags(dom, 0);
}

7082 7083 7084 7085 7086 7087 7088 7089 7090 7091 7092 7093 7094 7095 7096
static int
qemudCanonicalizeMachineFromInfo(virDomainDefPtr def,
                                 virCapsGuestDomainInfoPtr info,
                                 char **canonical)
{
    int i;

    *canonical = NULL;

    for (i = 0; i < info->nmachines; i++) {
        virCapsGuestMachinePtr machine = info->machines[i];

        if (!machine->canonical)
            continue;

7097
        if (STRNEQ(def->os.machine, machine->name))
7098 7099 7100
            continue;

        if (!(*canonical = strdup(machine->canonical))) {
7101
            virReportOOMError();
7102 7103 7104 7105 7106 7107 7108 7109 7110
            return -1;
        }

        break;
    }

    return 0;
}

7111 7112 7113 7114 7115 7116 7117
static int
qemudCanonicalizeMachineDirect(virDomainDefPtr def, char **canonical)
{
    virCapsGuestMachinePtr *machines = NULL;
    int i, nmachines = 0;

    if (qemudProbeMachineTypes(def->emulator, &machines, &nmachines) < 0) {
7118
        virReportOOMError();
7119 7120 7121 7122 7123 7124 7125
        return -1;
    }

    for (i = 0; i < nmachines; i++) {
        if (!machines[i]->canonical)
            continue;

7126
        if (STRNEQ(def->os.machine, machines[i]->name))
7127 7128 7129 7130 7131 7132 7133 7134 7135 7136 7137 7138
            continue;

        *canonical = machines[i]->canonical;
        machines[i]->canonical = NULL;
        break;
    }

    virCapabilitiesFreeMachines(machines, nmachines);

    return 0;
}

7139 7140
int
qemudCanonicalizeMachine(struct qemud_driver *driver, virDomainDefPtr def)
7141 7142 7143 7144 7145 7146
{
    char *canonical = NULL;
    int i;

    for (i = 0; i < driver->caps->nguests; i++) {
        virCapsGuestPtr guest = driver->caps->guests[i];
7147
        virCapsGuestDomainInfoPtr info;
7148 7149 7150
        int j;

        for (j = 0; j < guest->arch.ndomains; j++) {
7151
            info = &guest->arch.domains[j]->info;
7152

7153 7154 7155 7156 7157 7158 7159 7160 7161
            if (!info->emulator || !STREQ(info->emulator, def->emulator))
                continue;

            if (!info->nmachines)
                info = &guest->arch.defaultInfo;

            if (qemudCanonicalizeMachineFromInfo(def, info, &canonical) < 0)
                return -1;
            goto out;
7162 7163
        }

7164 7165 7166 7167
        info = &guest->arch.defaultInfo;

        if (info->emulator && STREQ(info->emulator, def->emulator)) {
            if (qemudCanonicalizeMachineFromInfo(def, info, &canonical) < 0)
7168 7169 7170 7171
                return -1;
            goto out;
        }
    }
7172 7173 7174 7175

    if (qemudCanonicalizeMachineDirect(def, &canonical) < 0)
        return -1;

7176 7177 7178 7179 7180 7181 7182
out:
    if (canonical) {
        VIR_FREE(def->os.machine);
        def->os.machine = canonical;
    }
    return 0;
}
D
Daniel P. Berrange 已提交
7183

7184
static virDomainPtr qemudDomainDefine(virConnectPtr conn, const char *xml) {
7185
    struct qemud_driver *driver = conn->privateData;
7186
    virDomainDefPtr def;
7187
    virDomainObjPtr vm = NULL;
7188
    virDomainPtr dom = NULL;
7189
    virDomainEventPtr event = NULL;
7190
    int dupVM;
7191

7192
    qemuDriverLock(driver);
7193
    if (!(def = virDomainDefParseString(driver->caps, xml,
7194
                                        VIR_DOMAIN_XML_INACTIVE)))
7195
        goto cleanup;
7196

7197
    if (virSecurityDriverVerify(def) < 0)
7198 7199
        goto cleanup;

7200 7201
    if ((dupVM = virDomainObjIsDuplicate(&driver->domains, def, 0)) < 0)
        goto cleanup;
7202

7203
    if (qemudCanonicalizeMachine(driver, def) < 0)
7204 7205
        goto cleanup;

7206 7207 7208
    if (qemuAssignPCIAddresses(def) < 0)
        goto cleanup;

7209
    if (!(vm = virDomainAssignDef(driver->caps,
7210
                                  &driver->domains,
7211
                                  def, false))) {
7212
        goto cleanup;
7213
    }
7214
    def = NULL;
7215
    vm->persistent = 1;
7216

7217
    if (virDomainSaveConfig(driver->configDir,
7218
                            vm->newDef ? vm->newDef : vm->def) < 0) {
7219 7220
        virDomainRemoveInactive(&driver->domains,
                                vm);
7221
        vm = NULL;
7222
        goto cleanup;
7223 7224
    }

7225 7226
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_DEFINED,
7227
                                     !dupVM ?
7228 7229
                                     VIR_DOMAIN_EVENT_DEFINED_ADDED :
                                     VIR_DOMAIN_EVENT_DEFINED_UPDATED);
7230

7231
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
7232
    if (dom) dom->id = vm->def->id;
7233 7234

cleanup:
7235
    virDomainDefFree(def);
7236 7237
    if (vm)
        virDomainObjUnlock(vm);
7238 7239
    if (event)
        qemuDomainEventQueue(driver, event);
7240
    qemuDriverUnlock(driver);
7241
    return dom;
D
Daniel P. Berrange 已提交
7242 7243
}

7244
static int qemudDomainUndefine(virDomainPtr dom) {
7245 7246
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
7247
    virDomainEventPtr event = NULL;
7248
    int ret = -1;
D
Daniel P. Berrange 已提交
7249

7250
    qemuDriverLock(driver);
7251
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
7252

D
Daniel P. Berrange 已提交
7253
    if (!vm) {
7254 7255
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
7256 7257
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
7258
        goto cleanup;
D
Daniel P. Berrange 已提交
7259 7260
    }

D
Daniel P. Berrange 已提交
7261
    if (virDomainObjIsActive(vm)) {
7262 7263
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot delete active domain"));
7264
        goto cleanup;
D
Daniel P. Berrange 已提交
7265 7266
    }

7267
    if (!vm->persistent) {
7268 7269
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("cannot undefine transient domain"));
7270
        goto cleanup;
7271 7272
    }

7273
    if (virDomainDeleteConfig(driver->configDir, driver->autostartDir, vm) < 0)
7274
        goto cleanup;
D
Daniel P. Berrange 已提交
7275

7276 7277 7278
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_UNDEFINED,
                                     VIR_DOMAIN_EVENT_UNDEFINED_REMOVED);
7279

7280 7281
    virDomainRemoveInactive(&driver->domains,
                            vm);
7282
    vm = NULL;
7283
    ret = 0;
D
Daniel P. Berrange 已提交
7284

7285
cleanup:
7286 7287
    if (vm)
        virDomainObjUnlock(vm);
7288 7289
    if (event)
        qemuDomainEventQueue(driver, event);
7290
    qemuDriverUnlock(driver);
7291
    return ret;
D
Daniel P. Berrange 已提交
7292 7293
}

7294

7295
static int qemudDomainChangeEjectableMedia(struct qemud_driver *driver,
7296
                                           virDomainObjPtr vm,
7297 7298
                                           virDomainDiskDefPtr disk,
                                           unsigned long long qemuCmdFlags)
7299
{
7300
    virDomainDiskDefPtr origdisk = NULL;
7301
    int i;
7302
    int ret;
7303
    char *driveAlias = NULL;
7304

7305
    origdisk = NULL;
7306
    for (i = 0 ; i < vm->def->ndisks ; i++) {
7307 7308
        if (vm->def->disks[i]->bus == disk->bus &&
            STREQ(vm->def->disks[i]->dst, disk->dst)) {
7309
            origdisk = vm->def->disks[i];
7310
            break;
7311
        }
7312 7313 7314
    }

    if (!origdisk) {
7315 7316 7317 7318
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("No device with bus '%s' and target '%s'"),
                        virDomainDiskBusTypeToString(disk->bus),
                        disk->dst);
7319 7320 7321
        return -1;
    }

7322
    if (!origdisk->info.alias) {
7323 7324
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("missing disk device alias name for %s"), origdisk->dst);
7325 7326
        return -1;
    }
7327

7328 7329
    if (origdisk->device != VIR_DOMAIN_DISK_DEVICE_FLOPPY &&
        origdisk->device != VIR_DOMAIN_DISK_DEVICE_CDROM) {
7330 7331 7332
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Removable media not supported for %s device"),
                        virDomainDiskDeviceTypeToString(disk->device));
7333
        return -1;
7334
    }
7335

7336 7337
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityImageLabel &&
7338 7339
        driver->securityDriver->domainSetSecurityImageLabel(driver->securityDriver,
                                                            vm, disk) < 0)
7340 7341
        return -1;

7342 7343 7344
    if (!(driveAlias = qemuDeviceDriveHostAlias(origdisk, qemuCmdFlags)))
        goto error;

7345
    qemuDomainObjPrivatePtr priv = vm->privateData;
7346
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7347
    if (disk->src) {
7348
        const char *format = NULL;
7349 7350 7351
        if (disk->type != VIR_DOMAIN_DISK_TYPE_DIR) {
            if (disk->driverType)
                format = disk->driverType;
7352 7353 7354
            else if (origdisk->driverType)
                format = origdisk->driverType;
        }
7355
        ret = qemuMonitorChangeMedia(priv->mon,
7356
                                     driveAlias,
7357
                                     disk->src, format);
7358
    } else {
7359
        ret = qemuMonitorEjectMedia(priv->mon, driveAlias);
7360
    }
7361
    qemuDomainObjExitMonitorWithDriver(driver, vm);
7362

7363 7364 7365 7366 7367
    if (ret < 0)
        goto error;

    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
7368 7369
        driver->securityDriver->domainRestoreSecurityImageLabel(driver->securityDriver,
                                                                vm, origdisk) < 0)
7370 7371 7372
        VIR_WARN("Unable to restore security label on ejected image %s", origdisk->src);

    VIR_FREE(origdisk->src);
7373 7374 7375 7376
    origdisk->src = disk->src;
    disk->src = NULL;
    origdisk->type = disk->type;

7377 7378
    VIR_FREE(driveAlias);

7379
    virDomainDiskDefFree(disk);
7380

7381
    return ret;
7382 7383

error:
7384
    VIR_FREE(driveAlias);
7385 7386
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
7387 7388
        driver->securityDriver->domainRestoreSecurityImageLabel(driver->securityDriver,
                                                                vm, disk) < 0)
7389
        VIR_WARN("Unable to restore security label on new media %s", disk->src);
7390
    return -1;
7391 7392
}

7393

7394
static int qemudDomainAttachPciDiskDevice(struct qemud_driver *driver,
7395
                                          virDomainObjPtr vm,
7396
                                          virDomainDiskDefPtr disk,
7397
                                          unsigned long long qemuCmdFlags)
7398
{
7399
    int i, ret;
7400
    const char* type = virDomainDiskBusTypeToString(disk->bus);
7401
    qemuDomainObjPrivatePtr priv = vm->privateData;
7402 7403
    char *devstr = NULL;
    char *drivestr = NULL;
7404 7405

    for (i = 0 ; i < vm->def->ndisks ; i++) {
7406
        if (STREQ(vm->def->disks[i]->dst, disk->dst)) {
7407 7408
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("target %s already exists"), disk->dst);
7409 7410 7411 7412
            return -1;
        }
    }

7413 7414
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityImageLabel &&
7415 7416
        driver->securityDriver->domainSetSecurityImageLabel(driver->securityDriver,
                                                            vm, disk) < 0)
7417 7418
        return -1;

7419
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
7420 7421
        if (qemuDomainPCIAddressEnsureAddr(priv->pciaddrs, &disk->info) < 0)
            goto error;
7422 7423
        if (qemuAssignDeviceDiskAlias(disk, qemuCmdFlags) < 0)
            goto error;
7424

7425 7426 7427
        if (!(drivestr = qemuBuildDriveStr(disk, 0, qemuCmdFlags)))
            goto error;

7428
        if (!(devstr = qemuBuildDriveDevStr(disk)))
7429 7430 7431
            goto error;
    }

7432
    if (VIR_REALLOC_N(vm->def->disks, vm->def->ndisks+1) < 0) {
7433
        virReportOOMError();
7434
        goto error;
7435 7436
    }

7437
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7438 7439
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        ret = qemuMonitorAddDrive(priv->mon, drivestr);
7440 7441 7442
        if (ret == 0) {
            ret = qemuMonitorAddDevice(priv->mon, devstr);
            if (ret < 0) {
7443
                VIR_WARN("qemuMonitorAddDevice failed on %s (%s)",
7444 7445 7446 7447 7448
                         drivestr, devstr);
                /* XXX should call 'drive_del' on error but this does not
                   exist yet */
            }
        }
7449 7450 7451 7452 7453 7454 7455 7456 7457 7458 7459
    } else {
        virDomainDevicePCIAddress guestAddr;
        ret = qemuMonitorAddPCIDisk(priv->mon,
                                    disk->src,
                                    type,
                                    &guestAddr);
        if (ret == 0) {
            disk->info.type = VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI;
            memcpy(&disk->info.addr.pci, &guestAddr, sizeof(guestAddr));
        }
    }
7460
    qemuDomainObjExitMonitorWithDriver(driver, vm);
7461

7462 7463
    if (ret < 0)
        goto error;
7464

7465
    virDomainDiskInsertPreAlloced(vm->def, disk);
7466

7467 7468 7469
    VIR_FREE(devstr);
    VIR_FREE(drivestr);

7470 7471 7472
    return 0;

error:
7473 7474 7475
    VIR_FREE(devstr);
    VIR_FREE(drivestr);

7476 7477 7478 7479 7480
    if ((qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) &&
        (disk->info.type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI) &&
        qemuDomainPCIAddressReleaseAddr(priv->pciaddrs, &disk->info) < 0)
        VIR_WARN("Unable to release PCI address on %s", disk->src);

7481 7482
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
7483 7484
        driver->securityDriver->domainRestoreSecurityImageLabel(driver->securityDriver,
                                                                vm, disk) < 0)
7485
        VIR_WARN("Unable to restore security label on %s", disk->src);
7486 7487

    return -1;
7488
}
7489

7490

7491
static int qemudDomainAttachPciControllerDevice(struct qemud_driver *driver,
7492
                                                virDomainObjPtr vm,
7493
                                                virDomainControllerDefPtr controller,
7494
                                                unsigned long long qemuCmdFlags)
7495
{
7496 7497
    int i;
    int ret = -1;
7498
    const char* type = virDomainControllerTypeToString(controller->type);
7499
    char *devstr = NULL;
7500 7501 7502
    qemuDomainObjPrivatePtr priv = vm->privateData;

    for (i = 0 ; i < vm->def->ncontrollers ; i++) {
7503 7504
        if ((vm->def->controllers[i]->type == controller->type) &&
            (vm->def->controllers[i]->idx == controller->idx)) {
7505 7506 7507
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("target %s:%d already exists"),
                            type, controller->idx);
7508 7509 7510 7511
            return -1;
        }
    }

7512 7513 7514 7515 7516 7517
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuDomainPCIAddressEnsureAddr(priv->pciaddrs, &controller->info) < 0)
            goto cleanup;
        if (qemuAssignDeviceControllerAlias(controller) < 0)
            goto cleanup;
    }
7518

7519
    if (!(devstr = qemuBuildControllerDevStr(controller))) {
7520
        virReportOOMError();
7521 7522 7523
        goto cleanup;
    }

7524
    if (VIR_REALLOC_N(vm->def->controllers, vm->def->ncontrollers+1) < 0) {
7525
        virReportOOMError();
7526
        goto cleanup;
7527 7528 7529
    }

    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7530 7531 7532 7533 7534 7535 7536
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        ret = qemuMonitorAddDevice(priv->mon, devstr);
    } else {
        ret = qemuMonitorAttachPCIDiskController(priv->mon,
                                                 type,
                                                 &controller->info.addr.pci);
    }
7537 7538 7539
    qemuDomainObjExitMonitorWithDriver(driver, vm);

    if (ret == 0) {
7540 7541
        controller->info.type = VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI;
        virDomainControllerInsertPreAlloced(vm->def, controller);
7542 7543
    }

7544
cleanup:
7545 7546 7547 7548 7549 7550
    if ((ret != 0) &&
        (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) &&
        (controller->info.type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI) &&
        qemuDomainPCIAddressReleaseAddr(priv->pciaddrs, &controller->info) < 0)
        VIR_WARN0("Unable to release PCI address on controller");

7551
    VIR_FREE(devstr);
7552 7553 7554
    return ret;
}

7555

7556
static virDomainControllerDefPtr
7557
qemuDomainFindOrCreateSCSIDiskController(struct qemud_driver *driver,
7558
                                         virDomainObjPtr vm,
7559
                                         int controller,
7560
                                         unsigned long long qemuCmdFlags)
7561 7562 7563 7564 7565 7566 7567 7568 7569 7570 7571 7572 7573
{
    int i;
    virDomainControllerDefPtr cont;
    for (i = 0 ; i < vm->def->ncontrollers ; i++) {
        cont = vm->def->controllers[i];

        if (cont->type != VIR_DOMAIN_CONTROLLER_TYPE_SCSI)
            continue;

        if (cont->idx == controller)
            return cont;
    }

W
Wolfgang Mauerer 已提交
7574
    /* No SCSI controller present, for backward compatibility we
7575 7576
     * now hotplug a controller */
    if (VIR_ALLOC(cont) < 0) {
7577
        virReportOOMError();
7578 7579 7580 7581
        return NULL;
    }
    cont->type = VIR_DOMAIN_CONTROLLER_TYPE_SCSI;
    cont->idx = 0;
7582
    cont->model = -1;
7583 7584

    VIR_INFO0("No SCSI controller present, hotplugging one");
7585
    if (qemudDomainAttachPciControllerDevice(driver,
7586
                                             vm, cont, qemuCmdFlags) < 0) {
7587 7588 7589
        VIR_FREE(cont);
        return NULL;
    }
7590 7591 7592 7593 7594 7595 7596 7597 7598

    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("guest unexpectedly quit"));
        /* cont doesn't need freeing here, since the reference
         * now held in def->controllers */
        return NULL;
    }

7599 7600 7601
    return cont;
}

7602

7603
static int qemudDomainAttachSCSIDisk(struct qemud_driver *driver,
7604
                                     virDomainObjPtr vm,
7605
                                     virDomainDiskDefPtr disk,
7606
                                     unsigned long long qemuCmdFlags)
7607 7608 7609
{
    int i;
    qemuDomainObjPrivatePtr priv = vm->privateData;
7610
    virDomainControllerDefPtr cont = NULL;
7611
    char *drivestr = NULL;
7612
    char *devstr = NULL;
7613 7614 7615
    int ret = -1;

    for (i = 0 ; i < vm->def->ndisks ; i++) {
7616
        if (STREQ(vm->def->disks[i]->dst, disk->dst)) {
7617 7618
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("target %s already exists"), disk->dst);
7619
            return -1;
7620 7621 7622
        }
    }

7623 7624 7625

    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityImageLabel &&
7626 7627
        driver->securityDriver->domainSetSecurityImageLabel(driver->securityDriver,
                                                            vm, disk) < 0)
7628 7629
        return -1;

7630
    /* We should have an address already, so make sure */
7631
    if (disk->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_DRIVE) {
7632 7633 7634
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("unexpected disk address type %s"),
                        virDomainDeviceAddressTypeToString(disk->info.type));
7635
        goto error;
7636 7637
    }

7638 7639 7640
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuAssignDeviceDiskAlias(disk, qemuCmdFlags) < 0)
            goto error;
7641
        if (!(devstr = qemuBuildDriveDevStr(disk)))
7642 7643 7644 7645 7646 7647
            goto error;
    }

    if (!(drivestr = qemuBuildDriveStr(disk, 0, qemuCmdFlags)))
        goto error;

7648
    for (i = 0 ; i <= disk->info.addr.drive.controller ; i++) {
7649
        cont = qemuDomainFindOrCreateSCSIDiskController(driver, vm, i, qemuCmdFlags);
7650
        if (!cont)
7651
            goto error;
7652 7653
    }

7654 7655 7656 7657 7658
    /* Tell clang that "cont" is non-NULL.
       This is because disk->info.addr.driver.controller is unsigned,
       and hence the above loop must iterate at least once.  */
    sa_assert (cont);

7659
    if (cont->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI) {
7660 7661
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("SCSI controller %d was missing its PCI address"), cont->idx);
7662
        goto error;
7663 7664 7665
    }

    if (VIR_REALLOC_N(vm->def->disks, vm->def->ndisks+1) < 0) {
7666
        virReportOOMError();
7667
        goto error;
7668 7669 7670
    }

    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7671
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
7672 7673 7674 7675
        ret = qemuMonitorAddDrive(priv->mon, drivestr);
        if (ret == 0) {
            ret = qemuMonitorAddDevice(priv->mon, devstr);
            if (ret < 0) {
7676
                VIR_WARN("qemuMonitorAddDevice failed on %s (%s)",
7677 7678 7679 7680 7681
                         drivestr, devstr);
                /* XXX should call 'drive_del' on error but this does not
                   exist yet */
            }
        }
7682 7683 7684 7685 7686 7687 7688 7689 7690 7691 7692 7693 7694
    } else {
        virDomainDeviceDriveAddress driveAddr;
        ret = qemuMonitorAttachDrive(priv->mon,
                                     drivestr,
                                     &cont->info.addr.pci,
                                     &driveAddr);
        if (ret == 0) {
            /* XXX we should probably validate that the addr matches
             * our existing defined addr instead of overwriting */
            disk->info.type = VIR_DOMAIN_DEVICE_ADDRESS_TYPE_DRIVE;
            memcpy(&disk->info.addr.drive, &driveAddr, sizeof(driveAddr));
        }
    }
7695 7696
    qemuDomainObjExitMonitorWithDriver(driver, vm);

7697 7698
    if (ret < 0)
        goto error;
7699

7700
    virDomainDiskInsertPreAlloced(vm->def, disk);
7701 7702

    VIR_FREE(devstr);
7703
    VIR_FREE(drivestr);
7704 7705 7706 7707

    return 0;

error:
7708
    VIR_FREE(devstr);
7709
    VIR_FREE(drivestr);
7710

7711 7712
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
7713 7714
        driver->securityDriver->domainRestoreSecurityImageLabel(driver->securityDriver,
                                                                vm, disk) < 0)
7715
        VIR_WARN("Unable to restore security label on %s", disk->src);
7716 7717

    return -1;
7718 7719
}

7720

7721
static int qemudDomainAttachUsbMassstorageDevice(struct qemud_driver *driver,
7722
                                                 virDomainObjPtr vm,
7723
                                                 virDomainDiskDefPtr disk,
7724
                                                 unsigned long long qemuCmdFlags)
7725
{
7726
    qemuDomainObjPrivatePtr priv = vm->privateData;
7727
    int i, ret;
7728 7729
    char *drivestr = NULL;
    char *devstr = NULL;
7730

7731
    for (i = 0 ; i < vm->def->ndisks ; i++) {
7732
        if (STREQ(vm->def->disks[i]->dst, disk->dst)) {
7733 7734
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("target %s already exists"), disk->dst);
7735 7736 7737 7738
            return -1;
        }
    }

7739 7740
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityImageLabel &&
7741 7742
        driver->securityDriver->domainSetSecurityImageLabel(driver->securityDriver,
                                                            vm, disk) < 0)
7743 7744
        return -1;

7745
    if (!disk->src) {
7746 7747
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("disk source path is missing"));
7748
        goto error;
7749 7750
    }

7751
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
7752 7753
        if (qemuAssignDeviceDiskAlias(disk, qemuCmdFlags) < 0)
            goto error;
7754 7755
        if (!(drivestr = qemuBuildDriveStr(disk, 0, qemuCmdFlags)))
            goto error;
7756
        if (!(devstr = qemuBuildDriveDevStr(disk)))
7757 7758 7759
            goto error;
    }

7760
    if (VIR_REALLOC_N(vm->def->disks, vm->def->ndisks+1) < 0) {
7761
        virReportOOMError();
7762
        goto error;
7763 7764
    }

7765
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7766
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
7767 7768 7769 7770
        ret = qemuMonitorAddDrive(priv->mon, drivestr);
        if (ret == 0) {
            ret = qemuMonitorAddDevice(priv->mon, devstr);
            if (ret < 0) {
7771
                VIR_WARN("qemuMonitorAddDevice failed on %s (%s)",
7772 7773 7774 7775 7776
                         drivestr, devstr);
                /* XXX should call 'drive_del' on error but this does not
                   exist yet */
            }
        }
7777 7778 7779
    } else {
        ret = qemuMonitorAddUSBDisk(priv->mon, disk->src);
    }
7780
    qemuDomainObjExitMonitorWithDriver(driver, vm);
7781

7782 7783 7784
    if (ret < 0)
        goto error;

7785
    virDomainDiskInsertPreAlloced(vm->def, disk);
7786

7787 7788 7789
    VIR_FREE(devstr);
    VIR_FREE(drivestr);

7790 7791 7792
    return 0;

error:
7793 7794 7795
    VIR_FREE(devstr);
    VIR_FREE(drivestr);

7796 7797
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
7798 7799
        driver->securityDriver->domainRestoreSecurityImageLabel(driver->securityDriver,
                                                                vm, disk) < 0)
7800
        VIR_WARN("Unable to restore security label on %s", disk->src);
7801 7802

    return -1;
7803 7804
}

7805

7806
/* XXX conn required for network -> bridge resolution */
M
Mark McLoughlin 已提交
7807
static int qemudDomainAttachNetDevice(virConnectPtr conn,
7808
                                      struct qemud_driver *driver,
M
Mark McLoughlin 已提交
7809
                                      virDomainObjPtr vm,
7810
                                      virDomainNetDefPtr net,
7811
                                      unsigned long long qemuCmdFlags)
M
Mark McLoughlin 已提交
7812
{
7813
    qemuDomainObjPrivatePtr priv = vm->privateData;
7814
    char *tapfd_name = NULL;
7815
    int tapfd = -1;
7816
    char *nicstr = NULL;
7817
    char *netstr = NULL;
7818
    int ret = -1;
7819
    virDomainDevicePCIAddress guestAddr;
7820
    int vlan;
M
Mark McLoughlin 已提交
7821 7822

    if (!(qemuCmdFlags & QEMUD_CMD_FLAG_HOST_NET_ADD)) {
7823
        qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
7824
                        _("installed qemu version does not support host_net_add"));
M
Mark McLoughlin 已提交
7825 7826 7827 7828 7829
        return -1;
    }

    if (net->type == VIR_DOMAIN_NET_TYPE_BRIDGE ||
        net->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
7830
        if (priv->monConfig->type != VIR_DOMAIN_CHR_TYPE_UNIX) {
7831
            qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
7832 7833 7834
                            _("network device type '%s' cannot be attached: "
                              "qemu is not using a unix socket monitor"),
                            virDomainNetTypeToString(net->type));
7835 7836 7837 7838 7839
            return -1;
        }

        if ((tapfd = qemudNetworkIfaceConnect(conn, driver, net, qemuCmdFlags)) < 0)
            return -1;
7840 7841
    } else if (net->type == VIR_DOMAIN_NET_TYPE_DIRECT) {
        if (priv->monConfig->type != VIR_DOMAIN_CHR_TYPE_UNIX) {
7842
            qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
7843 7844 7845 7846 7847 7848
                            _("network device type '%s' cannot be attached: "
                            "qemu is not using a unix socket monitor"),
                            virDomainNetTypeToString(net->type));
            return -1;
        }

S
Stefan Berger 已提交
7849
        if ((tapfd = qemudPhysIfaceConnect(conn, driver, net,
7850 7851
                                           qemuCmdFlags,
                                           vm->def->uuid)) < 0)
7852
            return -1;
M
Mark McLoughlin 已提交
7853 7854
    }

7855 7856
    if (VIR_REALLOC_N(vm->def->nets, vm->def->nnets+1) < 0)
        goto no_memory;
M
Mark McLoughlin 已提交
7857

7858 7859 7860 7861 7862
    if ((qemuCmdFlags & QEMUD_CMD_FLAG_NET_NAME) ||
        (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)) {
        if (qemuAssignDeviceNetAlias(vm->def, net, -1) < 0)
            goto cleanup;
    }
M
Mark McLoughlin 已提交
7863

7864 7865 7866 7867
    if ((qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) &&
        qemuDomainPCIAddressEnsureAddr(priv->pciaddrs, &net->info) < 0)
        goto cleanup;

7868 7869 7870 7871 7872
    if ((qemuCmdFlags & QEMUD_CMD_FLAG_NETDEV) &&
        (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)) {
        vlan = -1;
    } else {
        vlan = qemuDomainNetVLAN(net);
M
Mark McLoughlin 已提交
7873

7874
        if (vlan < 0) {
7875
            qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
7876 7877 7878
                            _("Unable to attach network devices without vlan"));
            goto cleanup;
        }
7879 7880
    }

7881
    if (tapfd != -1) {
7882
        if (virAsprintf(&tapfd_name, "fd-%s", net->info.alias) < 0)
7883 7884
            goto no_memory;

7885
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
7886
        if (qemuMonitorSendFileHandle(priv->mon, tapfd_name, tapfd) < 0) {
7887
            qemuDomainObjExitMonitorWithDriver(driver, vm);
7888
            goto cleanup;
7889
        }
7890
        qemuDomainObjExitMonitorWithDriver(driver, vm);
7891 7892 7893 7894 7895 7896

        if (!virDomainObjIsActive(vm)) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("guest unexpectedly quit"));
            goto cleanup;
        }
7897 7898
    }

7899
    /* FIXME - need to support vhost-net here (5th arg) */
7900 7901 7902 7903 7904 7905 7906 7907 7908 7909
    if ((qemuCmdFlags & QEMUD_CMD_FLAG_NETDEV) &&
        (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)) {
        if (!(netstr = qemuBuildHostNetStr(net, ',',
                                           -1, tapfd_name, 0)))
            goto try_tapfd_close;
    } else {
        if (!(netstr = qemuBuildHostNetStr(net, ' ',
                                           vlan, tapfd_name, 0)))
            goto try_tapfd_close;
    }
M
Mark McLoughlin 已提交
7910

7911
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7912 7913 7914 7915 7916 7917 7918 7919 7920 7921 7922
    if ((qemuCmdFlags & QEMUD_CMD_FLAG_NETDEV) &&
        (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)) {
        if (qemuMonitorAddNetdev(priv->mon, netstr) < 0) {
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            goto try_tapfd_close;
        }
    } else {
        if (qemuMonitorAddHostNetwork(priv->mon, netstr) < 0) {
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            goto try_tapfd_close;
        }
7923
    }
7924
    qemuDomainObjExitMonitorWithDriver(driver, vm);
M
Mark McLoughlin 已提交
7925

7926 7927 7928
    if (tapfd != -1)
        close(tapfd);
    tapfd = -1;
M
Mark McLoughlin 已提交
7929

7930 7931 7932 7933 7934 7935
    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("guest unexpectedly quit"));
        goto cleanup;
    }

7936 7937 7938 7939 7940 7941 7942
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (!(nicstr = qemuBuildNicDevStr(net, vlan)))
            goto try_remove;
    } else {
        if (!(nicstr = qemuBuildNicStr(net, NULL, vlan)))
            goto try_remove;
    }
M
Mark McLoughlin 已提交
7943

7944
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7945 7946 7947 7948 7949 7950 7951 7952 7953 7954 7955 7956 7957
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuMonitorAddDevice(priv->mon, nicstr) < 0) {
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            goto try_remove;
        }
    } else {
        if (qemuMonitorAddPCINetwork(priv->mon, nicstr,
                                     &guestAddr) < 0) {
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            goto try_remove;
        }
        net->info.type = VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI;
        memcpy(&net->info.addr.pci, &guestAddr, sizeof(guestAddr));
7958
    }
7959
    qemuDomainObjExitMonitorWithDriver(driver, vm);
7960

7961
    ret = 0;
M
Mark McLoughlin 已提交
7962 7963 7964

    vm->def->nets[vm->def->nnets++] = net;

7965
cleanup:
7966 7967 7968 7969 7970 7971
    if ((ret != 0) &&
        (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) &&
        (net->info.type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI) &&
        qemuDomainPCIAddressReleaseAddr(priv->pciaddrs, &net->info) < 0)
        VIR_WARN0("Unable to release PCI address on NIC");

7972
    if (ret != 0)
7973
        virDomainConfNWFilterTeardown(net);
S
Stefan Berger 已提交
7974

7975 7976 7977 7978 7979
    VIR_FREE(nicstr);
    VIR_FREE(netstr);
    VIR_FREE(tapfd_name);
    if (tapfd != -1)
        close(tapfd);
7980

7981
    return ret;
7982

7983
try_remove:
7984
    if (!virDomainObjIsActive(vm))
7985 7986
        goto cleanup;

7987
    if (vlan < 0) {
7988 7989 7990 7991 7992 7993 7994
        if ((qemuCmdFlags & QEMUD_CMD_FLAG_NETDEV) &&
            (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)) {
            char *netdev_name;
            if (virAsprintf(&netdev_name, "host%s", net->info.alias) < 0)
                goto no_memory;
            qemuDomainObjEnterMonitorWithDriver(driver, vm);
            if (qemuMonitorRemoveNetdev(priv->mon, netdev_name) < 0)
7995
                VIR_WARN("Failed to remove network backend for netdev %s",
7996 7997 7998 7999
                         netdev_name);
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            VIR_FREE(netdev_name);
        } else {
8000
            VIR_WARN0("Unable to remove network backend");
8001
        }
8002 8003 8004 8005
    } else {
        char *hostnet_name;
        if (virAsprintf(&hostnet_name, "host%s", net->info.alias) < 0)
            goto no_memory;
8006
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
8007
        if (qemuMonitorRemoveHostNetwork(priv->mon, vlan, hostnet_name) < 0)
8008
            VIR_WARN("Failed to remove network backend for vlan %d, net %s",
8009
                     vlan, hostnet_name);
8010
        qemuDomainObjExitMonitorWithDriver(driver, vm);
8011
        VIR_FREE(hostnet_name);
8012
    }
8013
    goto cleanup;
8014

8015
try_tapfd_close:
8016
    if (!virDomainObjIsActive(vm))
8017 8018
        goto cleanup;

8019
    if (tapfd_name) {
8020
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
8021
        if (qemuMonitorCloseFileHandle(priv->mon, tapfd_name) < 0)
8022
            VIR_WARN("Failed to close tapfd with '%s'", tapfd_name);
8023
        qemuDomainObjExitMonitorWithDriver(driver, vm);
8024
    }
8025

8026 8027
    goto cleanup;

8028
no_memory:
8029
    virReportOOMError();
8030
    goto cleanup;
M
Mark McLoughlin 已提交
8031 8032
}

8033

8034
static int qemudDomainAttachHostPciDevice(struct qemud_driver *driver,
8035
                                          virDomainObjPtr vm,
8036
                                          virDomainHostdevDefPtr hostdev,
8037
                                          unsigned long long qemuCmdFlags)
8038
{
8039
    qemuDomainObjPrivatePtr priv = vm->privateData;
8040
    int ret;
8041
    char *devstr = NULL;
8042 8043
    int configfd = -1;
    char *configfd_name = NULL;
8044 8045

    if (VIR_REALLOC_N(vm->def->hostdevs, vm->def->nhostdevs+1) < 0) {
8046
        virReportOOMError();
8047 8048 8049
        return -1;
    }

8050
    if (qemuPrepareHostdevPCIDevices(driver, &hostdev, 1) < 0)
8051
        return -1;
8052

8053
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
8054 8055
        if (qemuAssignDeviceHostdevAlias(vm->def, hostdev, -1) < 0)
            goto error;
8056 8057
        if (qemuDomainPCIAddressEnsureAddr(priv->pciaddrs, &hostdev->info) < 0)
            goto error;
8058 8059 8060 8061 8062 8063 8064 8065 8066 8067 8068 8069 8070 8071 8072 8073 8074 8075 8076 8077 8078 8079 8080 8081
        if (qemuCmdFlags & QEMUD_CMD_FLAG_PCI_CONFIGFD) {
            configfd = qemudOpenPCIConfig(hostdev);
            if (configfd >= 0) {
                if (virAsprintf(&configfd_name, "fd-%s",
                                hostdev->info.alias) < 0) {
                    virReportOOMError();
                    goto error;
                }

                qemuDomainObjEnterMonitorWithDriver(driver, vm);
                if (qemuMonitorSendFileHandle(priv->mon, configfd_name,
                                              configfd) < 0) {
                    qemuDomainObjExitMonitorWithDriver(driver, vm);
                    goto error;
                }
                qemuDomainObjExitMonitorWithDriver(driver, vm);
            }
        }

        if (!virDomainObjIsActive(vm)) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("guest unexpectedly quit during hotplug"));
            goto error;
        }
8082

8083
        if (!(devstr = qemuBuildPCIHostdevDevStr(hostdev, configfd_name)))
8084
            goto error;
8085

8086
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
8087
        ret = qemuMonitorAddDevice(priv->mon, devstr);
8088 8089 8090 8091 8092
        qemuDomainObjExitMonitorWithDriver(driver, vm);
    } else {
        virDomainDevicePCIAddress guestAddr;

        qemuDomainObjEnterMonitorWithDriver(driver, vm);
8093 8094 8095
        ret = qemuMonitorAddPCIHostDevice(priv->mon,
                                          &hostdev->source.subsys.u.pci,
                                          &guestAddr);
8096 8097 8098 8099 8100
        qemuDomainObjExitMonitorWithDriver(driver, vm);

        hostdev->info.type = VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI;
        memcpy(&hostdev->info.addr.pci, &guestAddr, sizeof(guestAddr));
    }
8101
    if (ret < 0)
8102
        goto error;
8103 8104 8105

    vm->def->hostdevs[vm->def->nhostdevs++] = hostdev;

8106
    VIR_FREE(devstr);
8107 8108 8109
    VIR_FREE(configfd_name);
    if (configfd >= 0)
        close(configfd);
8110

8111
    return 0;
8112 8113

error:
8114 8115 8116 8117 8118
    if ((qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) &&
        (hostdev->info.type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI) &&
        qemuDomainPCIAddressReleaseAddr(priv->pciaddrs, &hostdev->info) < 0)
        VIR_WARN0("Unable to release PCI address on host device");

8119
    qemuDomainReAttachHostdevDevices(driver, &hostdev, 1);
8120 8121

    VIR_FREE(devstr);
8122 8123 8124
    VIR_FREE(configfd_name);
    if (configfd >= 0)
        close(configfd);
8125 8126

    return -1;
8127 8128
}

8129

8130
static int qemudDomainAttachHostUsbDevice(struct qemud_driver *driver,
M
Mark McLoughlin 已提交
8131
                                          virDomainObjPtr vm,
8132
                                          virDomainHostdevDefPtr hostdev,
8133
                                          unsigned long long qemuCmdFlags)
8134 8135
{
    int ret;
8136
    qemuDomainObjPrivatePtr priv = vm->privateData;
8137 8138
    char *devstr = NULL;

8139 8140 8141 8142 8143 8144
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuAssignDeviceHostdevAlias(vm->def, hostdev, -1) < 0)
            goto error;
        if (!(devstr = qemuBuildUSBHostdevDevStr(hostdev)))
            goto error;
    }
8145

8146
    if (VIR_REALLOC_N(vm->def->hostdevs, vm->def->nhostdevs+1) < 0) {
8147
        virReportOOMError();
8148
        goto error;
8149
    }
8150

8151
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
8152 8153 8154
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)
        ret = qemuMonitorAddDevice(priv->mon, devstr);
    else
8155
        ret = qemuMonitorAddUSBDeviceExact(priv->mon,
8156 8157
                                           hostdev->source.subsys.u.usb.bus,
                                           hostdev->source.subsys.u.usb.device);
8158
    qemuDomainObjExitMonitorWithDriver(driver, vm);
8159 8160 8161 8162
    if (ret < 0)
        goto error;

    vm->def->hostdevs[vm->def->nhostdevs++] = hostdev;
8163

8164
    VIR_FREE(devstr);
8165

8166 8167 8168 8169 8170
    return 0;

error:
    VIR_FREE(devstr);
    return -1;
8171 8172
}

8173

8174
static int qemudDomainAttachHostDevice(struct qemud_driver *driver,
M
Mark McLoughlin 已提交
8175
                                       virDomainObjPtr vm,
8176
                                       virDomainHostdevDefPtr hostdev,
8177
                                       unsigned long long qemuCmdFlags)
M
Mark McLoughlin 已提交
8178 8179
{
    if (hostdev->mode != VIR_DOMAIN_HOSTDEV_MODE_SUBSYS) {
8180
        qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
8181 8182
                        _("hostdev mode '%s' not supported"),
                        virDomainHostdevModeTypeToString(hostdev->mode));
M
Mark McLoughlin 已提交
8183 8184 8185
        return -1;
    }

8186 8187 8188 8189 8190 8191 8192 8193 8194 8195 8196 8197 8198 8199 8200 8201 8202
    /* Resolve USB product/vendor to bus/device */
    if (hostdev->source.subsys.type == VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB &&
        hostdev->source.subsys.u.usb.vendor) {
        usbDevice *usb
            = usbFindDevice(hostdev->source.subsys.u.usb.vendor,
                            hostdev->source.subsys.u.usb.product);

        if (!usb)
            return -1;

        hostdev->source.subsys.u.usb.bus = usbDeviceGetBus(usb);
        hostdev->source.subsys.u.usb.device = usbDeviceGetDevno(usb);

        usbFreeDevice(usb);
    }


8203
    if (driver->securityDriver &&
8204
        driver->securityDriver->domainSetSecurityHostdevLabel &&
8205 8206
        driver->securityDriver->domainSetSecurityHostdevLabel(driver->securityDriver,
                                                              vm, hostdev) < 0)
8207
        return -1;
M
Mark McLoughlin 已提交
8208 8209

    switch (hostdev->source.subsys.type) {
8210
    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI:
8211
        if (qemudDomainAttachHostPciDevice(driver, vm,
8212
                                           hostdev, qemuCmdFlags) < 0)
8213 8214 8215
            goto error;
        break;

M
Mark McLoughlin 已提交
8216
    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB:
8217
        if (qemudDomainAttachHostUsbDevice(driver, vm,
8218
                                           hostdev, qemuCmdFlags) < 0)
8219 8220 8221
            goto error;
        break;

M
Mark McLoughlin 已提交
8222
    default:
8223
        qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
8224 8225
                        _("hostdev subsys type '%s' not supported"),
                        virDomainHostdevSubsysTypeToString(hostdev->source.subsys.type));
8226
        goto error;
M
Mark McLoughlin 已提交
8227
    }
8228 8229 8230 8231 8232 8233

    return 0;

error:
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityHostdevLabel &&
8234 8235
        driver->securityDriver->domainRestoreSecurityHostdevLabel(driver->securityDriver,
                                                                  vm, hostdev) < 0)
8236 8237 8238
        VIR_WARN0("Unable to restore host device labelling on hotplug fail");

    return -1;
M
Mark McLoughlin 已提交
8239 8240
}

8241

8242
static int qemudDomainAttachDevice(virDomainPtr dom,
8243 8244
                                   const char *xml)
{
8245 8246 8247
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    virDomainDeviceDefPtr dev = NULL;
8248
    unsigned long long qemuCmdFlags;
8249
    virCgroupPtr cgroup = NULL;
8250
    int ret = -1;
8251

8252
    qemuDriverLock(driver);
8253
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
8254
    if (!vm) {
8255 8256
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
8257 8258
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
8259
        goto cleanup;
8260 8261
    }

8262 8263 8264
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
8265
    if (!virDomainObjIsActive(vm)) {
8266 8267
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot attach device on inactive domain"));
8268
        goto endjob;
8269 8270
    }

8271
    dev = virDomainDeviceDefParse(driver->caps, vm->def, xml,
8272
                                  VIR_DOMAIN_XML_INACTIVE);
8273
    if (dev == NULL)
8274
        goto endjob;
8275

8276 8277 8278
    if (qemudExtractVersionInfo(vm->def->emulator,
                                NULL,
                                &qemuCmdFlags) < 0)
8279
        goto endjob;
8280

8281
    if (dev->type == VIR_DOMAIN_DEVICE_DISK) {
8282
        if (qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_DEVICES)) {
8283
            if (virCgroupForDomain(driver->cgroup, vm->def->name, &cgroup, 0) !=0 ) {
8284 8285 8286
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("Unable to find cgroup for %s\n"),
                                vm->def->name);
8287
                goto endjob;
8288
            }
8289
            if (qemuSetupDiskCgroup(driver, cgroup, dev->data.disk) < 0)
8290
                goto endjob;
8291 8292
        }

8293
        switch (dev->data.disk->device) {
8294 8295
        case VIR_DOMAIN_DISK_DEVICE_CDROM:
        case VIR_DOMAIN_DISK_DEVICE_FLOPPY:
8296 8297 8298
            ret = qemudDomainChangeEjectableMedia(driver, vm,
                                                  dev->data.disk,
                                                  qemuCmdFlags);
8299 8300
            if (ret == 0)
                dev->data.disk = NULL;
8301
            break;
8302

8303 8304
        case VIR_DOMAIN_DISK_DEVICE_DISK:
            if (dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_USB) {
8305
                ret = qemudDomainAttachUsbMassstorageDevice(driver, vm,
8306
                                                            dev->data.disk, qemuCmdFlags);
8307 8308
                if (ret == 0)
                    dev->data.disk = NULL;
8309
            } else if (dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_VIRTIO) {
8310
                ret = qemudDomainAttachPciDiskDevice(driver, vm,
8311
                                                     dev->data.disk, qemuCmdFlags);
8312 8313
                if (ret == 0)
                    dev->data.disk = NULL;
8314
            } else if (dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_SCSI) {
8315
                ret = qemudDomainAttachSCSIDisk(driver, vm,
8316
                                                dev->data.disk, qemuCmdFlags);
8317 8318
                if (ret == 0)
                    dev->data.disk = NULL;
8319
            } else {
8320
                qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
8321 8322
                                _("disk bus '%s' cannot be hotplugged."),
                                virDomainDiskBusTypeToString(dev->data.disk->bus));
8323
                /* fallthrough */
8324 8325
            }
            break;
8326

8327
        default:
8328
            qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
8329 8330
                            _("disk device type '%s' cannot be hotplugged"),
                            virDomainDiskDeviceTypeToString(dev->data.disk->device));
8331 8332
            /* Fallthrough */
        }
8333
        if (ret != 0 && cgroup) {
8334
            if (qemuTeardownDiskCgroup(driver, cgroup, dev->data.disk) < 0)
8335 8336
                VIR_WARN("Failed to teardown cgroup for disk path %s",
                         NULLSTR(dev->data.disk->src));
8337
        }
8338 8339
    } else if (dev->type == VIR_DOMAIN_DEVICE_CONTROLLER) {
        if (dev->data.controller->type == VIR_DOMAIN_CONTROLLER_TYPE_SCSI) {
8340
            ret = qemudDomainAttachPciControllerDevice(driver, vm,
8341
                                                       dev->data.controller, qemuCmdFlags);
8342 8343
            if (ret == 0)
                dev->data.controller = NULL;
8344
        } else {
8345
            qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
8346 8347
                            _("disk controller bus '%s' cannot be hotplugged."),
                            virDomainControllerTypeToString(dev->data.controller->type));
8348 8349
            /* fallthrough */
        }
M
Mark McLoughlin 已提交
8350
    } else if (dev->type == VIR_DOMAIN_DEVICE_NET) {
8351 8352
        ret = qemudDomainAttachNetDevice(dom->conn, driver, vm,
                                         dev->data.net, qemuCmdFlags);
8353 8354
        if (ret == 0)
            dev->data.net = NULL;
M
Mark McLoughlin 已提交
8355
    } else if (dev->type == VIR_DOMAIN_DEVICE_HOSTDEV) {
8356
        ret = qemudDomainAttachHostDevice(driver, vm,
8357
                                          dev->data.hostdev, qemuCmdFlags);
8358 8359
        if (ret == 0)
            dev->data.hostdev = NULL;
8360
    } else {
8361
        qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
8362 8363
                        _("device type '%s' cannot be attached"),
                        virDomainDeviceTypeToString(dev->type));
8364
        goto endjob;
8365 8366
    }

8367
    if (!ret && virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
8368 8369
        ret = -1;

8370
endjob:
8371 8372
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
8373

8374
cleanup:
8375 8376 8377
    if (cgroup)
        virCgroupFree(&cgroup);

8378
    virDomainDeviceDefFree(dev);
8379 8380
    if (vm)
        virDomainObjUnlock(vm);
8381
    qemuDriverUnlock(driver);
8382 8383 8384
    return ret;
}

8385 8386 8387 8388
static int qemudDomainAttachDeviceFlags(virDomainPtr dom,
                                        const char *xml,
                                        unsigned int flags) {
    if (flags & VIR_DOMAIN_DEVICE_MODIFY_CONFIG) {
8389 8390
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot modify the persistent configuration of a domain"));
8391 8392 8393 8394 8395 8396
        return -1;
    }

    return qemudDomainAttachDevice(dom, xml);
}

8397

8398 8399 8400 8401 8402 8403 8404 8405 8406 8407 8408 8409 8410 8411 8412 8413 8414 8415 8416 8417 8418 8419 8420 8421 8422 8423 8424 8425 8426 8427 8428 8429 8430 8431 8432 8433 8434 8435 8436 8437 8438 8439 8440 8441 8442 8443 8444 8445 8446 8447 8448 8449 8450 8451 8452 8453 8454 8455 8456 8457 8458 8459 8460 8461 8462 8463 8464 8465 8466 8467 8468 8469 8470 8471 8472 8473 8474
static virDomainGraphicsDefPtr qemuDomainFindGraphics(virDomainObjPtr vm,
                                                      virDomainGraphicsDefPtr dev)
{
    int i;

    for (i = 0 ; i < vm->def->ngraphics ; i++) {
        if (vm->def->graphics[i]->type == dev->type)
            return vm->def->graphics[i];
    }

    return NULL;
}


static int
qemuDomainChangeGraphics(struct qemud_driver *driver,
                         virDomainObjPtr vm,
                         virDomainGraphicsDefPtr dev)
{
    virDomainGraphicsDefPtr olddev = qemuDomainFindGraphics(vm, dev);
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int ret = -1;

    if (!olddev) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("cannot find existing graphics device to modify"));
        return -1;
    }

    switch (dev->type) {
    case VIR_DOMAIN_GRAPHICS_TYPE_VNC:
        if ((olddev->data.vnc.autoport != dev->data.vnc.autoport) ||
            (!dev->data.vnc.autoport && (olddev->data.vnc.port != dev->data.vnc.port))) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("cannot change port settings on vnc graphics"));
            return -1;
        }
        if (STRNEQ_NULLABLE(olddev->data.vnc.listenAddr, dev->data.vnc.listenAddr)) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("cannot change listen address setting on vnc graphics"));
            return -1;
        }
        if (STRNEQ_NULLABLE(olddev->data.vnc.keymap, dev->data.vnc.keymap)) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("cannot change keymap setting on vnc graphics"));
            return -1;
        }

        if (STRNEQ_NULLABLE(olddev->data.vnc.passwd, dev->data.vnc.passwd)) {
            VIR_DEBUG("Updating password on VNC server %p %p", dev->data.vnc.passwd, driver->vncPassword);
            qemuDomainObjEnterMonitorWithDriver(driver, vm);
            ret = qemuMonitorSetVNCPassword(priv->mon,
                                            dev->data.vnc.passwd ?
                                            dev->data.vnc.passwd :
                                            driver->vncPassword);
            qemuDomainObjExitMonitorWithDriver(driver, vm);

            /* Steal the new dev's  char * reference */
            VIR_FREE(olddev->data.vnc.passwd);
            olddev->data.vnc.passwd = dev->data.vnc.passwd;
            dev->data.vnc.passwd = NULL;
        } else {
            ret = 0;
        }
        break;

    default:
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("unable to change config on '%s' graphics type"),
                        virDomainGraphicsTypeToString(dev->type));
        break;
    }

    return ret;
}


8475 8476 8477 8478 8479 8480 8481 8482 8483 8484 8485
static int qemuDomainUpdateDeviceFlags(virDomainPtr dom,
                                       const char *xml,
                                       unsigned int flags)
{
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    virDomainDeviceDefPtr dev = NULL;
    unsigned long long qemuCmdFlags;
    virCgroupPtr cgroup = NULL;
    int ret = -1;

8486 8487 8488 8489
    virCheckFlags(VIR_DOMAIN_DEVICE_MODIFY_CURRENT |
                  VIR_DOMAIN_DEVICE_MODIFY_LIVE |
                  VIR_DOMAIN_DEVICE_MODIFY_CONFIG, -1);

8490 8491 8492 8493 8494 8495 8496 8497 8498 8499 8500 8501 8502 8503 8504 8505 8506 8507 8508 8509 8510 8511 8512 8513 8514 8515 8516 8517 8518 8519 8520 8521 8522 8523 8524 8525 8526 8527 8528 8529 8530 8531 8532 8533
    if (flags & VIR_DOMAIN_DEVICE_MODIFY_CONFIG) {
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot modify the persistent configuration of a domain"));
        return -1;
    }

    qemuDriverLock(driver);
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
        goto cleanup;
    }

    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot attach device on inactive domain"));
        goto endjob;
    }

    dev = virDomainDeviceDefParse(driver->caps, vm->def, xml,
                                  VIR_DOMAIN_XML_INACTIVE);
    if (dev == NULL)
        goto endjob;

    if (qemudExtractVersionInfo(vm->def->emulator,
                                NULL,
                                &qemuCmdFlags) < 0)
        goto endjob;

    switch (dev->type) {
    case VIR_DOMAIN_DEVICE_DISK:
        if (qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_DEVICES)) {
            if (virCgroupForDomain(driver->cgroup, vm->def->name, &cgroup, 0) !=0 ) {
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("Unable to find cgroup for %s\n"),
                                vm->def->name);
                goto endjob;
            }
8534
            if (qemuSetupDiskCgroup(driver, cgroup, dev->data.disk) < 0)
8535 8536 8537 8538 8539 8540
                goto endjob;
        }

        switch (dev->data.disk->device) {
        case VIR_DOMAIN_DISK_DEVICE_CDROM:
        case VIR_DOMAIN_DISK_DEVICE_FLOPPY:
8541 8542 8543
            ret = qemudDomainChangeEjectableMedia(driver, vm,
                                                  dev->data.disk,
                                                  qemuCmdFlags);
8544 8545 8546 8547 8548 8549
            if (ret == 0)
                dev->data.disk = NULL;
            break;


        default:
8550
            qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
8551 8552 8553 8554 8555 8556
                            _("disk bus '%s' cannot be updated."),
                            virDomainDiskBusTypeToString(dev->data.disk->bus));
            break;
        }

        if (ret != 0 && cgroup) {
8557
            if (qemuTeardownDiskCgroup(driver, cgroup, dev->data.disk) < 0)
8558 8559
                VIR_WARN("Failed to teardown cgroup for disk path %s",
                         NULLSTR(dev->data.disk->src));
8560 8561 8562
        }
        break;

8563 8564 8565 8566
    case VIR_DOMAIN_DEVICE_GRAPHICS:
        ret = qemuDomainChangeGraphics(driver, vm, dev->data.graphics);
        break;

8567
    default:
8568
        qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
8569 8570 8571 8572 8573 8574 8575 8576 8577 8578 8579 8580 8581 8582 8583 8584 8585 8586 8587 8588 8589 8590 8591 8592
                        _("disk device type '%s' cannot be updated"),
                        virDomainDiskDeviceTypeToString(dev->data.disk->device));
        break;
    }

    if (!ret && virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
        ret = -1;

endjob:
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;

cleanup:
    if (cgroup)
        virCgroupFree(&cgroup);

    virDomainDeviceDefFree(dev);
    if (vm)
        virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);
    return ret;
}


E
Eric Blake 已提交
8593
static inline int qemudFindDisk(virDomainDefPtr def, const char *dst)
W
Wolfgang Mauerer 已提交
8594 8595 8596 8597 8598 8599 8600 8601 8602 8603 8604 8605
{
    int i;

    for (i = 0 ; i < def->ndisks ; i++) {
        if (STREQ(def->disks[i]->dst, dst)) {
            return i;
        }
    }

    return -1;
}

E
Eric Blake 已提交
8606
static inline void qemudShrinkDisks(virDomainDefPtr def, size_t i)
W
Wolfgang Mauerer 已提交
8607 8608 8609 8610 8611 8612 8613 8614 8615 8616 8617 8618 8619 8620 8621 8622
{
    if (def->ndisks > 1) {
        memmove(def->disks + i,
                def->disks + i + 1,
                sizeof(*def->disks) *
                (def->ndisks - (i + 1)));
        def->ndisks--;
        if (VIR_REALLOC_N(def->disks, def->ndisks) < 0) {
            /* ignore, harmless */
        }
    } else {
        VIR_FREE(def->disks);
        def->ndisks = 0;
    }
}

8623
static int qemudDomainDetachPciDiskDevice(struct qemud_driver *driver,
8624
                                          virDomainObjPtr vm,
8625 8626
                                          virDomainDeviceDefPtr dev,
                                          unsigned long long qemuCmdFlags)
8627 8628 8629
{
    int i, ret = -1;
    virDomainDiskDefPtr detach = NULL;
8630
    qemuDomainObjPrivatePtr priv = vm->privateData;
8631
    virCgroupPtr cgroup = NULL;
8632

W
Wolfgang Mauerer 已提交
8633
    i = qemudFindDisk(vm->def, dev->data.disk->dst);
8634

W
Wolfgang Mauerer 已提交
8635
    if (i < 0) {
8636 8637
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("disk %s not found"), dev->data.disk->dst);
8638
        goto cleanup;
8639 8640
    }

W
Wolfgang Mauerer 已提交
8641 8642
    detach = vm->def->disks[i];

8643 8644 8645 8646 8647 8648 8649 8650 8651
    if (qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_DEVICES)) {
        if (virCgroupForDomain(driver->cgroup, vm->def->name, &cgroup, 0) != 0) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("Unable to find cgroup for %s\n"),
                            vm->def->name);
            goto cleanup;
        }
    }

8652 8653
    if (!virDomainDeviceAddressIsValid(&detach->info,
                                       VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI)) {
8654 8655
        qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
                        _("device cannot be detached without a PCI address"));
8656
        goto cleanup;
8657 8658
    }

8659
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
8660 8661 8662 8663 8664 8665 8666 8667 8668 8669 8670
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuMonitorDelDevice(priv->mon, detach->info.alias) < 0) {
            qemuDomainObjExitMonitor(vm);
            goto cleanup;
        }
    } else {
        if (qemuMonitorRemovePCIDevice(priv->mon,
                                       &detach->info.addr.pci) < 0) {
            qemuDomainObjExitMonitor(vm);
            goto cleanup;
        }
8671
    }
8672
    qemuDomainObjExitMonitorWithDriver(driver, vm);
8673

W
Wolfgang Mauerer 已提交
8674 8675
    qemudShrinkDisks(vm->def, i);

8676 8677
    virDomainDiskDefFree(detach);

8678 8679
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
8680 8681
        driver->securityDriver->domainRestoreSecurityImageLabel(driver->securityDriver,
                                                                vm, dev->data.disk) < 0)
8682 8683
        VIR_WARN("Unable to restore security label on %s", dev->data.disk->src);

8684
    if (cgroup != NULL) {
8685
        if (qemuTeardownDiskCgroup(driver, cgroup, dev->data.disk) < 0)
8686 8687 8688 8689
            VIR_WARN("Failed to teardown cgroup for disk path %s",
                     NULLSTR(dev->data.disk->src));
    }

8690 8691 8692 8693 8694 8695 8696 8697 8698 8699 8700 8701 8702 8703
    ret = 0;

cleanup:
    return ret;
}

static int qemudDomainDetachSCSIDiskDevice(struct qemud_driver *driver,
                                           virDomainObjPtr vm,
                                           virDomainDeviceDefPtr dev,
                                           unsigned long long qemuCmdFlags)
{
    int i, ret = -1;
    virDomainDiskDefPtr detach = NULL;
    qemuDomainObjPrivatePtr priv = vm->privateData;
8704
    virCgroupPtr cgroup = NULL;
8705 8706 8707 8708 8709 8710 8711 8712 8713 8714

    i = qemudFindDisk(vm->def, dev->data.disk->dst);

    if (i < 0) {
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("disk %s not found"), dev->data.disk->dst);
        goto cleanup;
    }

    if (!(qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)) {
8715
        qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
8716 8717 8718 8719 8720 8721
                        _("Underlying qemu does not support SCSI disk removal"));
        goto cleanup;
    }

    detach = vm->def->disks[i];

8722 8723 8724 8725 8726 8727 8728 8729 8730
    if (qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_DEVICES)) {
        if (virCgroupForDomain(driver->cgroup, vm->def->name, &cgroup, 0) != 0) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("Unable to find cgroup for %s\n"),
                            vm->def->name);
            goto cleanup;
        }
    }

8731 8732 8733 8734 8735 8736 8737 8738 8739
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
    if (qemuMonitorDelDevice(priv->mon, detach->info.alias) < 0) {
        qemuDomainObjExitMonitor(vm);
        goto cleanup;
    }
    qemuDomainObjExitMonitorWithDriver(driver, vm);

    qemudShrinkDisks(vm->def, i);

8740
    virDomainDiskDefFree(detach);
8741

8742 8743
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
8744 8745
        driver->securityDriver->domainRestoreSecurityImageLabel(driver->securityDriver,
                                                                vm, dev->data.disk) < 0)
8746 8747
        VIR_WARN("Unable to restore security label on %s", dev->data.disk->src);

8748
    if (cgroup != NULL) {
8749
        if (qemuTeardownDiskCgroup(driver, cgroup, dev->data.disk) < 0)
8750 8751 8752 8753
            VIR_WARN("Failed to teardown cgroup for disk path %s",
                     NULLSTR(dev->data.disk->src));
    }

8754
    ret = 0;
8755 8756

cleanup:
8757
    virCgroupFree(&cgroup);
8758 8759 8760
    return ret;
}

8761
static int qemudDomainDetachPciControllerDevice(struct qemud_driver *driver,
8762
                                                virDomainObjPtr vm,
8763 8764
                                                virDomainDeviceDefPtr dev,
                                                unsigned long long qemuCmdFlags)
8765 8766 8767 8768 8769 8770 8771 8772 8773 8774 8775 8776 8777 8778
{
    int i, ret = -1;
    virDomainControllerDefPtr detach = NULL;
    qemuDomainObjPrivatePtr priv = vm->privateData;

    for (i = 0 ; i < vm->def->ncontrollers ; i++) {
        if ((vm->def->controllers[i]->type == dev->data.controller->type) &&
            (vm->def->controllers[i]->idx == dev->data.controller->idx)) {
            detach = vm->def->controllers[i];
            break;
        }
    }

    if (!detach) {
8779 8780 8781 8782
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("disk controller %s:%d not found"),
                        virDomainControllerTypeToString(dev->data.controller->type),
                        dev->data.controller->idx);
8783 8784 8785 8786 8787
        goto cleanup;
    }

    if (!virDomainDeviceAddressIsValid(&detach->info,
                                       VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI)) {
8788 8789
        qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
                        _("device cannot be detached without a PCI address"));
8790 8791 8792
        goto cleanup;
    }

8793 8794 8795 8796 8797
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuAssignDeviceControllerAlias(detach) < 0)
            goto cleanup;
    }

8798
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
8799 8800 8801 8802 8803 8804 8805 8806 8807 8808 8809
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuMonitorDelDevice(priv->mon, detach->info.alias)) {
            qemuDomainObjExitMonitor(vm);
            goto cleanup;
        }
    } else {
        if (qemuMonitorRemovePCIDevice(priv->mon,
                                       &detach->info.addr.pci) < 0) {
            qemuDomainObjExitMonitor(vm);
            goto cleanup;
        }
8810 8811 8812 8813 8814 8815 8816 8817 8818 8819 8820 8821 8822 8823 8824 8825
    }
    qemuDomainObjExitMonitorWithDriver(driver, vm);

    if (vm->def->ncontrollers > 1) {
        memmove(vm->def->controllers + i,
                vm->def->controllers + i + 1,
                sizeof(*vm->def->controllers) *
                (vm->def->ncontrollers - (i + 1)));
        vm->def->ncontrollers--;
        if (VIR_REALLOC_N(vm->def->controllers, vm->def->ncontrollers) < 0) {
            /* ignore, harmless */
        }
    } else {
        VIR_FREE(vm->def->controllers);
        vm->def->ncontrollers = 0;
    }
8826

8827 8828
    if ((qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) &&
        qemuDomainPCIAddressReleaseAddr(priv->pciaddrs, &detach->info) < 0)
8829 8830
        VIR_WARN0("Unable to release PCI address on controller");

8831 8832 8833 8834 8835 8836 8837 8838
    virDomainControllerDefFree(detach);

    ret = 0;

cleanup:
    return ret;
}

8839
static int
8840
qemudDomainDetachNetDevice(struct qemud_driver *driver,
8841
                           virDomainObjPtr vm,
8842 8843
                           virDomainDeviceDefPtr dev,
                           unsigned long long qemuCmdFlags)
8844 8845 8846
{
    int i, ret = -1;
    virDomainNetDefPtr detach = NULL;
8847
    qemuDomainObjPrivatePtr priv = vm->privateData;
8848 8849
    int vlan;
    char *hostnet_name = NULL;
8850 8851 8852 8853 8854 8855 8856 8857 8858 8859 8860

    for (i = 0 ; i < vm->def->nnets ; i++) {
        virDomainNetDefPtr net = vm->def->nets[i];

        if (!memcmp(net->mac, dev->data.net->mac,  sizeof(net->mac))) {
            detach = net;
            break;
        }
    }

    if (!detach) {
8861 8862 8863 8864 8865
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("network device %02x:%02x:%02x:%02x:%02x:%02x not found"),
                        dev->data.net->mac[0], dev->data.net->mac[1],
                        dev->data.net->mac[2], dev->data.net->mac[3],
                        dev->data.net->mac[4], dev->data.net->mac[5]);
8866 8867 8868
        goto cleanup;
    }

8869 8870
    if (!virDomainDeviceAddressIsValid(&detach->info,
                                       VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI)) {
8871 8872
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("device cannot be detached without a PCI address"));
8873 8874 8875
        goto cleanup;
    }

8876
    if ((vlan = qemuDomainNetVLAN(detach)) < 0) {
8877 8878
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("unable to determine original VLAN"));
8879 8880 8881 8882
        goto cleanup;
    }

    if (virAsprintf(&hostnet_name, "host%s", detach->info.alias) < 0) {
8883
        virReportOOMError();
8884 8885 8886
        goto cleanup;
    }

8887
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
8888 8889 8890 8891 8892 8893 8894 8895 8896 8897 8898
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuMonitorDelDevice(priv->mon, detach->info.alias) < 0) {
            qemuDomainObjExitMonitor(vm);
            goto cleanup;
        }
    } else {
        if (qemuMonitorRemovePCIDevice(priv->mon,
                                       &detach->info.addr.pci) < 0) {
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            goto cleanup;
        }
8899
    }
8900

8901 8902 8903 8904 8905 8906 8907 8908 8909 8910 8911
    if ((qemuCmdFlags & QEMUD_CMD_FLAG_NETDEV) &&
        (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)) {
        if (qemuMonitorRemoveNetdev(priv->mon, hostnet_name) < 0) {
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            goto cleanup;
        }
    } else {
        if (qemuMonitorRemoveHostNetwork(priv->mon, vlan, hostnet_name) < 0) {
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            goto cleanup;
        }
8912
    }
8913
    qemuDomainObjExitMonitorWithDriver(driver, vm);
8914

8915
    virDomainConfNWFilterTeardown(detach);
8916

8917
#if WITH_MACVTAP
8918
    if (detach->type == VIR_DOMAIN_NET_TYPE_DIRECT) {
8919
        delMacvtap(detach->ifname, detach->mac, detach->data.direct.linkdev,
8920
                   &detach->data.direct.virtPortProfile);
8921 8922
        VIR_FREE(detach->ifname);
    }
8923 8924
#endif

8925
    if ((driver->macFilter) && (detach->ifname != NULL)) {
8926
        if ((errno = networkDisallowMacOnPort(driver,
8927 8928
                                              detach->ifname,
                                              detach->mac))) {
8929
            virReportSystemError(errno,
8930 8931 8932 8933 8934
             _("failed to remove ebtables rule on  '%s'"),
                                 detach->ifname);
        }
    }

8935 8936 8937 8938 8939 8940 8941 8942 8943 8944
    if (vm->def->nnets > 1) {
        memmove(vm->def->nets + i,
                vm->def->nets + i + 1,
                sizeof(*vm->def->nets) *
                (vm->def->nnets - (i + 1)));
        vm->def->nnets--;
        if (VIR_REALLOC_N(vm->def->nets, vm->def->nnets) < 0) {
            /* ignore, harmless */
        }
    } else {
8945
        VIR_FREE(vm->def->nets);
8946
        vm->def->nnets = 0;
8947
    }
8948
    virDomainNetDefFree(detach);
8949

8950 8951 8952
    ret = 0;

cleanup:
8953
    VIR_FREE(hostnet_name);
8954 8955 8956
    return ret;
}

8957
static int qemudDomainDetachHostPciDevice(struct qemud_driver *driver,
8958
                                          virDomainObjPtr vm,
8959 8960
                                          virDomainDeviceDefPtr dev,
                                          unsigned long long qemuCmdFlags)
8961
{
8962
    virDomainHostdevDefPtr detach = NULL;
8963
    qemuDomainObjPrivatePtr priv = vm->privateData;
8964
    int i, ret;
8965
    pciDevice *pci;
8966 8967

    for (i = 0 ; i < vm->def->nhostdevs ; i++) {
8968 8969 8970 8971
        if (vm->def->hostdevs[i]->mode != VIR_DOMAIN_HOSTDEV_MODE_SUBSYS ||
            vm->def->hostdevs[i]->source.subsys.type != VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI)
            continue;

8972 8973 8974 8975 8976 8977 8978 8979 8980 8981 8982 8983 8984 8985 8986
        unsigned domain   = vm->def->hostdevs[i]->source.subsys.u.pci.domain;
        unsigned bus      = vm->def->hostdevs[i]->source.subsys.u.pci.bus;
        unsigned slot     = vm->def->hostdevs[i]->source.subsys.u.pci.slot;
        unsigned function = vm->def->hostdevs[i]->source.subsys.u.pci.function;

        if (dev->data.hostdev->source.subsys.u.pci.domain   == domain &&
            dev->data.hostdev->source.subsys.u.pci.bus      == bus &&
            dev->data.hostdev->source.subsys.u.pci.slot     == slot &&
            dev->data.hostdev->source.subsys.u.pci.function == function) {
            detach = vm->def->hostdevs[i];
            break;
        }
    }

    if (!detach) {
8987 8988 8989 8990 8991 8992
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("host pci device %.4x:%.2x:%.2x.%.1x not found"),
                        dev->data.hostdev->source.subsys.u.pci.domain,
                        dev->data.hostdev->source.subsys.u.pci.bus,
                        dev->data.hostdev->source.subsys.u.pci.slot,
                        dev->data.hostdev->source.subsys.u.pci.function);
8993 8994 8995
        return -1;
    }

8996 8997
    if (!virDomainDeviceAddressIsValid(&detach->info,
                                       VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI)) {
8998 8999
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("device cannot be detached without a PCI address"));
9000 9001 9002
        return -1;
    }

9003
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
9004 9005 9006 9007 9008 9009 9010 9011 9012 9013 9014
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuMonitorDelDevice(priv->mon, detach->info.alias) < 0) {
            qemuDomainObjExitMonitor(vm);
            return -1;
        }
    } else {
        if (qemuMonitorRemovePCIDevice(priv->mon,
                                       &detach->info.addr.pci) < 0) {
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            return -1;
        }
9015
    }
9016
    qemuDomainObjExitMonitorWithDriver(driver, vm);
9017 9018 9019

    ret = 0;

9020
    pci = pciGetDevice(detach->source.subsys.u.pci.domain,
9021 9022 9023 9024 9025 9026
                       detach->source.subsys.u.pci.bus,
                       detach->source.subsys.u.pci.slot,
                       detach->source.subsys.u.pci.function);
    if (!pci)
        ret = -1;
    else {
9027
        pciDeviceSetManaged(pci, detach->managed);
9028
        pciDeviceListDel(driver->activePciHostdevs, pci);
9029
        if (pciResetDevice(pci, driver->activePciHostdevs, NULL) < 0)
9030
            ret = -1;
9031
        qemudReattachPciDevice(pci, driver);
9032
        pciFreeDevice(pci);
9033 9034
    }

9035 9036
    if ((qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) &&
        qemuDomainPCIAddressReleaseAddr(priv->pciaddrs, &detach->info) < 0)
9037 9038
        VIR_WARN0("Unable to release PCI address on controller");

9039 9040 9041 9042 9043 9044 9045 9046 9047 9048 9049 9050
    if (vm->def->nhostdevs > 1) {
        memmove(vm->def->hostdevs + i,
                vm->def->hostdevs + i + 1,
                sizeof(*vm->def->hostdevs) *
                (vm->def->nhostdevs - (i + 1)));
        vm->def->nhostdevs--;
        if (VIR_REALLOC_N(vm->def->hostdevs, vm->def->nhostdevs) < 0) {
            /* ignore, harmless */
        }
    } else {
        VIR_FREE(vm->def->hostdevs);
        vm->def->nhostdevs = 0;
9051
    }
9052
    virDomainHostdevDefFree(detach);
9053 9054 9055 9056

    return ret;
}

9057 9058 9059 9060 9061 9062 9063 9064 9065 9066 9067 9068 9069 9070 9071 9072
static int qemudDomainDetachHostUsbDevice(struct qemud_driver *driver,
                                          virDomainObjPtr vm,
                                          virDomainDeviceDefPtr dev,
                                          unsigned long long qemuCmdFlags)
{
    virDomainHostdevDefPtr detach = NULL;
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int i, ret;

    for (i = 0 ; i < vm->def->nhostdevs ; i++) {
        if (vm->def->hostdevs[i]->mode != VIR_DOMAIN_HOSTDEV_MODE_SUBSYS ||
            vm->def->hostdevs[i]->source.subsys.type != VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB)
            continue;

        unsigned bus = vm->def->hostdevs[i]->source.subsys.u.usb.bus;
        unsigned device = vm->def->hostdevs[i]->source.subsys.u.usb.device;
9073 9074 9075 9076 9077 9078 9079 9080 9081 9082 9083 9084 9085 9086 9087 9088
        unsigned product = vm->def->hostdevs[i]->source.subsys.u.usb.product;
        unsigned vendor = vm->def->hostdevs[i]->source.subsys.u.usb.vendor;

        if (dev->data.hostdev->source.subsys.u.usb.bus &&
            dev->data.hostdev->source.subsys.u.usb.device) {
            if (dev->data.hostdev->source.subsys.u.usb.bus == bus &&
                dev->data.hostdev->source.subsys.u.usb.device == device) {
                detach = vm->def->hostdevs[i];
                break;
            }
        } else {
            if (dev->data.hostdev->source.subsys.u.usb.product == product &&
                dev->data.hostdev->source.subsys.u.usb.vendor == vendor) {
                detach = vm->def->hostdevs[i];
                break;
            }
9089 9090 9091 9092 9093 9094 9095 9096 9097 9098 9099 9100 9101 9102 9103 9104 9105 9106 9107 9108 9109 9110 9111 9112 9113 9114 9115 9116 9117 9118 9119 9120 9121 9122 9123 9124 9125 9126 9127 9128 9129 9130 9131 9132 9133 9134 9135 9136 9137 9138
        }
    }

    if (!detach) {
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("host usb device %03d.%03d not found"),
                        dev->data.hostdev->source.subsys.u.usb.bus,
                        dev->data.hostdev->source.subsys.u.usb.device);
        return -1;
    }

    if (!detach->info.alias) {
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("device cannot be detached without a device alias"));
        return -1;
    }

    if (!(qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)) {
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("device cannot be detached with this QEMU version"));
        return -1;
    }

    qemuDomainObjEnterMonitorWithDriver(driver, vm);
    if (qemuMonitorDelDevice(priv->mon, detach->info.alias) < 0) {
        qemuDomainObjExitMonitorWithDriver(driver, vm);
        return -1;
    }
    qemuDomainObjExitMonitorWithDriver(driver, vm);

    ret = 0;

    if (vm->def->nhostdevs > 1) {
        memmove(vm->def->hostdevs + i,
                vm->def->hostdevs + i + 1,
                sizeof(*vm->def->hostdevs) *
                (vm->def->nhostdevs - (i + 1)));
        vm->def->nhostdevs--;
        if (VIR_REALLOC_N(vm->def->hostdevs, vm->def->nhostdevs) < 0) {
            /* ignore, harmless */
        }
    } else {
        VIR_FREE(vm->def->hostdevs);
        vm->def->nhostdevs = 0;
    }
    virDomainHostdevDefFree(detach);

    return ret;
}

9139
static int qemudDomainDetachHostDevice(struct qemud_driver *driver,
9140
                                       virDomainObjPtr vm,
9141 9142
                                       virDomainDeviceDefPtr dev,
                                       unsigned long long qemuCmdFlags)
9143 9144 9145 9146 9147
{
    virDomainHostdevDefPtr hostdev = dev->data.hostdev;
    int ret;

    if (hostdev->mode != VIR_DOMAIN_HOSTDEV_MODE_SUBSYS) {
9148
        qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
9149 9150
                        _("hostdev mode '%s' not supported"),
                        virDomainHostdevModeTypeToString(hostdev->mode));
9151 9152 9153 9154 9155
        return -1;
    }

    switch (hostdev->source.subsys.type) {
    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI:
9156
        ret = qemudDomainDetachHostPciDevice(driver, vm, dev, qemuCmdFlags);
9157
        break;
9158 9159 9160
    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB:
        ret = qemudDomainDetachHostUsbDevice(driver, vm, dev, qemuCmdFlags);
        break;
9161
    default:
9162
        qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
9163 9164
                        _("hostdev subsys type '%s' not supported"),
                        virDomainHostdevSubsysTypeToString(hostdev->source.subsys.type));
9165 9166 9167
        return -1;
    }

9168
    if (driver->securityDriver &&
9169
        driver->securityDriver->domainRestoreSecurityHostdevLabel &&
9170 9171
        driver->securityDriver->domainRestoreSecurityHostdevLabel(driver->securityDriver,
                                                                  vm, dev->data.hostdev) < 0)
9172
        VIR_WARN0("Failed to restore host device labelling");
9173

9174 9175 9176
    return ret;
}

9177 9178
static int qemudDomainDetachDevice(virDomainPtr dom,
                                   const char *xml) {
9179 9180
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
9181
    unsigned long long qemuCmdFlags;
9182 9183
    virDomainDeviceDefPtr dev = NULL;
    int ret = -1;
9184

9185
    qemuDriverLock(driver);
9186
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
9187
    if (!vm) {
9188 9189
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
9190 9191
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
9192
        goto cleanup;
9193 9194
    }

9195 9196 9197
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
9198
    if (!virDomainObjIsActive(vm)) {
9199 9200
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot detach device on inactive domain"));
9201
        goto endjob;
9202 9203
    }

9204
    dev = virDomainDeviceDefParse(driver->caps, vm->def, xml,
9205
                                  VIR_DOMAIN_XML_INACTIVE);
9206
    if (dev == NULL)
9207
        goto endjob;
9208

9209 9210 9211 9212
    if (qemudExtractVersionInfo(vm->def->emulator,
                                NULL,
                                &qemuCmdFlags) < 0)
        goto endjob;
9213 9214

    if (dev->type == VIR_DOMAIN_DEVICE_DISK &&
9215 9216 9217 9218 9219 9220 9221 9222 9223
        dev->data.disk->device == VIR_DOMAIN_DISK_DEVICE_DISK) {
        if (dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_VIRTIO) {
            ret = qemudDomainDetachPciDiskDevice(driver, vm, dev, qemuCmdFlags);
        }
        else if (dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_SCSI) {
            ret = qemudDomainDetachSCSIDiskDevice(driver, vm, dev,
                                                  qemuCmdFlags);
        }
        else {
9224
            qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
9225 9226
                            _("This type of disk cannot be hot unplugged"));
        }
9227
    } else if (dev->type == VIR_DOMAIN_DEVICE_NET) {
9228
        ret = qemudDomainDetachNetDevice(driver, vm, dev, qemuCmdFlags);
9229 9230
    } else if (dev->type == VIR_DOMAIN_DEVICE_CONTROLLER) {
        if (dev->data.controller->type == VIR_DOMAIN_CONTROLLER_TYPE_SCSI) {
9231 9232
            ret = qemudDomainDetachPciControllerDevice(driver, vm, dev,
                                                       qemuCmdFlags);
9233
        } else {
9234
            qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
9235 9236
                            _("disk controller bus '%s' cannot be hotunplugged."),
                            virDomainControllerTypeToString(dev->data.controller->type));
9237 9238
            /* fallthrough */
        }
9239
    } else if (dev->type == VIR_DOMAIN_DEVICE_HOSTDEV) {
9240
        ret = qemudDomainDetachHostDevice(driver, vm, dev, qemuCmdFlags);
9241
    } else {
9242
        qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
9243
                        "%s", _("This type of device cannot be hot unplugged"));
9244
    }
9245

9246
    if (!ret && virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
9247 9248
        ret = -1;

9249
endjob:
9250 9251
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
9252

9253 9254
cleanup:
    virDomainDeviceDefFree(dev);
9255 9256
    if (vm)
        virDomainObjUnlock(vm);
9257
    qemuDriverUnlock(driver);
9258 9259 9260
    return ret;
}

9261 9262 9263 9264
static int qemudDomainDetachDeviceFlags(virDomainPtr dom,
                                        const char *xml,
                                        unsigned int flags) {
    if (flags & VIR_DOMAIN_DEVICE_MODIFY_CONFIG) {
9265 9266
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot modify the persistent configuration of a domain"));
9267 9268 9269 9270 9271 9272
        return -1;
    }

    return qemudDomainDetachDevice(dom, xml);
}

9273
static int qemudDomainGetAutostart(virDomainPtr dom,
9274
                                   int *autostart) {
9275 9276 9277
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
9278

9279
    qemuDriverLock(driver);
9280
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
9281 9282
    qemuDriverUnlock(driver);

9283
    if (!vm) {
9284 9285
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
9286 9287
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
9288
        goto cleanup;
9289 9290 9291
    }

    *autostart = vm->autostart;
9292
    ret = 0;
9293

9294
cleanup:
9295 9296
    if (vm)
        virDomainObjUnlock(vm);
9297
    return ret;
9298 9299
}

9300
static int qemudDomainSetAutostart(virDomainPtr dom,
9301
                                   int autostart) {
9302 9303
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
9304 9305
    char *configFile = NULL, *autostartLink = NULL;
    int ret = -1;
9306

9307
    qemuDriverLock(driver);
9308
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
9309

9310
    if (!vm) {
9311 9312
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
9313 9314
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
9315
        goto cleanup;
9316 9317
    }

9318
    if (!vm->persistent) {
9319 9320
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("cannot set autostart for transient domain"));
9321
        goto cleanup;
9322 9323
    }

9324 9325
    autostart = (autostart != 0);

9326
    if (vm->autostart != autostart) {
9327
        if ((configFile = virDomainConfigFile(driver->configDir, vm->def->name)) == NULL)
9328
            goto cleanup;
9329
        if ((autostartLink = virDomainConfigFile(driver->autostartDir, vm->def->name)) == NULL)
9330
            goto cleanup;
9331

9332 9333
        if (autostart) {
            int err;
9334

9335
            if ((err = virFileMakePath(driver->autostartDir))) {
9336
                virReportSystemError(err,
9337 9338
                                     _("cannot create autostart directory %s"),
                                     driver->autostartDir);
9339 9340
                goto cleanup;
            }
9341

9342
            if (symlink(configFile, autostartLink) < 0) {
9343
                virReportSystemError(errno,
9344 9345
                                     _("Failed to create symlink '%s to '%s'"),
                                     autostartLink, configFile);
9346 9347 9348 9349
                goto cleanup;
            }
        } else {
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
9350
                virReportSystemError(errno,
9351 9352
                                     _("Failed to delete symlink '%s'"),
                                     autostartLink);
9353 9354
                goto cleanup;
            }
9355 9356
        }

9357
        vm->autostart = autostart;
9358
    }
9359
    ret = 0;
9360

9361 9362 9363
cleanup:
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
9364 9365
    if (vm)
        virDomainObjUnlock(vm);
9366
    qemuDriverUnlock(driver);
9367
    return ret;
9368 9369
}

9370 9371 9372 9373 9374

static char *qemuGetSchedulerType(virDomainPtr dom,
                                  int *nparams)
{
    struct qemud_driver *driver = dom->conn->privateData;
9375
    char *ret = NULL;
9376

9377
    qemuDriverLock(driver);
9378
    if (!qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_CPU)) {
9379 9380
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        __FUNCTION__);
9381
        goto cleanup;
9382 9383 9384 9385 9386 9387 9388
    }

    if (nparams)
        *nparams = 1;

    ret = strdup("posix");
    if (!ret)
9389
        virReportOOMError();
9390 9391 9392

cleanup:
    qemuDriverUnlock(driver);
9393 9394 9395 9396 9397 9398 9399 9400 9401 9402 9403 9404 9405
    return ret;
}

static int qemuSetSchedulerParameters(virDomainPtr dom,
                                      virSchedParameterPtr params,
                                      int nparams)
{
    struct qemud_driver *driver = dom->conn->privateData;
    int i;
    virCgroupPtr group = NULL;
    virDomainObjPtr vm = NULL;
    int ret = -1;

9406
    qemuDriverLock(driver);
9407
    if (!qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_CPU)) {
9408 9409
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        __FUNCTION__);
9410
        goto cleanup;
9411 9412 9413 9414 9415
    }

    vm = virDomainFindByUUID(&driver->domains, dom->uuid);

    if (vm == NULL) {
9416 9417
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("No such domain %s"), dom->uuid);
9418 9419 9420 9421
        goto cleanup;
    }

    if (virCgroupForDomain(driver->cgroup, vm->def->name, &group, 0) != 0) {
9422 9423
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("cannot find cgroup for domain %s"), vm->def->name);
9424 9425 9426 9427 9428 9429 9430 9431 9432
        goto cleanup;
    }

    for (i = 0; i < nparams; i++) {
        virSchedParameterPtr param = &params[i];

        if (STREQ(param->field, "cpu_shares")) {
            int rc;
            if (param->type != VIR_DOMAIN_SCHED_FIELD_ULLONG) {
9433 9434
                qemuReportError(VIR_ERR_INVALID_ARG, "%s",
                                _("invalid type for cpu_shares tunable, expected a 'ullong'"));
9435 9436 9437 9438 9439
                goto cleanup;
            }

            rc = virCgroupSetCpuShares(group, params[i].value.ul);
            if (rc != 0) {
9440
                virReportSystemError(-rc, "%s",
9441 9442 9443 9444
                                     _("unable to set cpu shares tunable"));
                goto cleanup;
            }
        } else {
9445 9446
            qemuReportError(VIR_ERR_INVALID_ARG,
                            _("Invalid parameter `%s'"), param->field);
9447 9448 9449 9450 9451 9452 9453 9454 9455
            goto cleanup;
        }
    }
    ret = 0;

cleanup:
    virCgroupFree(&group);
    if (vm)
        virDomainObjUnlock(vm);
9456
    qemuDriverUnlock(driver);
9457 9458 9459 9460 9461 9462 9463 9464 9465 9466 9467 9468 9469 9470
    return ret;
}

static int qemuGetSchedulerParameters(virDomainPtr dom,
                                      virSchedParameterPtr params,
                                      int *nparams)
{
    struct qemud_driver *driver = dom->conn->privateData;
    virCgroupPtr group = NULL;
    virDomainObjPtr vm = NULL;
    unsigned long long val;
    int ret = -1;
    int rc;

9471
    qemuDriverLock(driver);
9472
    if (!qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_CPU)) {
9473 9474
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        __FUNCTION__);
9475
        goto cleanup;
9476 9477 9478
    }

    if ((*nparams) != 1) {
9479 9480
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("Invalid parameter count"));
9481
        goto cleanup;
9482 9483 9484 9485 9486
    }

    vm = virDomainFindByUUID(&driver->domains, dom->uuid);

    if (vm == NULL) {
9487 9488
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("No such domain %s"), dom->uuid);
9489 9490 9491 9492
        goto cleanup;
    }

    if (virCgroupForDomain(driver->cgroup, vm->def->name, &group, 0) != 0) {
9493 9494
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("cannot find cgroup for domain %s"), vm->def->name);
9495 9496 9497 9498 9499
        goto cleanup;
    }

    rc = virCgroupGetCpuShares(group, &val);
    if (rc != 0) {
9500
        virReportSystemError(-rc, "%s",
9501 9502 9503 9504 9505
                             _("unable to get cpu shares tunable"));
        goto cleanup;
    }
    params[0].value.ul = val;
    params[0].type = VIR_DOMAIN_SCHED_FIELD_ULLONG;
C
Chris Lalancette 已提交
9506
    if (virStrcpyStatic(params[0].field, "cpu_shares") == NULL) {
9507 9508
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("Field cpu_shares too long for destination"));
C
Chris Lalancette 已提交
9509 9510
        goto cleanup;
    }
9511 9512 9513 9514 9515 9516 9517

    ret = 0;

cleanup:
    virCgroupFree(&group);
    if (vm)
        virDomainObjUnlock(vm);
9518
    qemuDriverUnlock(driver);
9519 9520 9521 9522
    return ret;
}


9523 9524 9525 9526 9527 9528 9529 9530 9531
/* This uses the 'info blockstats' monitor command which was
 * integrated into both qemu & kvm in late 2007.  If the command is
 * not supported we detect this and return the appropriate error.
 */
static int
qemudDomainBlockStats (virDomainPtr dom,
                       const char *path,
                       struct _virDomainBlockStats *stats)
{
9532
    struct qemud_driver *driver = dom->conn->privateData;
9533
    int i, ret = -1;
9534
    virDomainObjPtr vm;
9535
    virDomainDiskDefPtr disk = NULL;
9536

9537
    qemuDriverLock(driver);
9538
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
9539
    qemuDriverUnlock(driver);
9540
    if (!vm) {
9541 9542
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
9543 9544
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
9545
        goto cleanup;
9546
    }
9547 9548 9549 9550

    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
9551
    if (!virDomainObjIsActive (vm)) {
9552 9553
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
9554
        goto endjob;
9555 9556
    }

9557 9558 9559 9560 9561 9562 9563 9564
    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (STREQ(path, vm->def->disks[i]->dst)) {
            disk = vm->def->disks[i];
            break;
        }
    }

    if (!disk) {
9565 9566
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("invalid path: %s"), path);
9567
        goto endjob;
9568 9569
    }

9570
    if (!disk->info.alias) {
9571 9572
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("missing disk device alias name for %s"), disk->dst);
9573
        goto endjob;
9574
    }
9575

9576
    qemuDomainObjPrivatePtr priv = vm->privateData;
9577 9578
    qemuDomainObjEnterMonitor(vm);
    ret = qemuMonitorGetBlockStatsInfo(priv->mon,
9579
                                       disk->info.alias,
9580 9581 9582 9583 9584 9585
                                       &stats->rd_req,
                                       &stats->rd_bytes,
                                       &stats->wr_req,
                                       &stats->wr_bytes,
                                       &stats->errs);
    qemuDomainObjExitMonitor(vm);
9586

9587
endjob:
9588 9589
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
9590

9591
cleanup:
9592 9593
    if (vm)
        virDomainObjUnlock(vm);
9594
    return ret;
9595 9596
}

9597
#ifdef __linux__
9598 9599 9600 9601 9602
static int
qemudDomainInterfaceStats (virDomainPtr dom,
                           const char *path,
                           struct _virDomainInterfaceStats *stats)
{
9603 9604
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
9605
    int i;
9606
    int ret = -1;
9607

9608
    qemuDriverLock(driver);
9609
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
9610 9611
    qemuDriverUnlock(driver);

9612
    if (!vm) {
9613 9614
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
9615 9616
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
9617
        goto cleanup;
9618 9619
    }

D
Daniel P. Berrange 已提交
9620
    if (!virDomainObjIsActive(vm)) {
9621 9622
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
9623
        goto cleanup;
9624 9625 9626
    }

    /* Check the path is one of the domain's network interfaces. */
9627 9628
    for (i = 0 ; i < vm->def->nnets ; i++) {
        if (vm->def->nets[i]->ifname &&
9629 9630 9631 9632
            STREQ (vm->def->nets[i]->ifname, path)) {
            ret = 0;
            break;
        }
9633 9634
    }

9635
    if (ret == 0)
9636
        ret = linuxDomainInterfaceStats(path, stats);
9637
    else
9638 9639
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("invalid path, '%s' is not a known interface"), path);
9640

9641
cleanup:
9642 9643
    if (vm)
        virDomainObjUnlock(vm);
9644 9645
    return ret;
}
9646
#else
9647 9648 9649 9650
static int
qemudDomainInterfaceStats (virDomainPtr dom,
                           const char *path ATTRIBUTE_UNUSED,
                           struct _virDomainInterfaceStats *stats ATTRIBUTE_UNUSED)
9651 9652
    qemuReportError(VIR_ERR_NO_SUPPORT,
                    "%s", __FUNCTION__);
9653 9654
    return -1;
}
9655
#endif
9656

9657 9658 9659 9660 9661 9662 9663 9664 9665 9666 9667 9668 9669 9670 9671 9672
static int
qemudDomainMemoryStats (virDomainPtr dom,
                        struct _virDomainMemoryStat *stats,
                        unsigned int nr_stats)
{
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    unsigned int ret = -1;

    qemuDriverLock(driver);
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
    qemuDriverUnlock(driver);

    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
9673 9674
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
9675 9676 9677
        goto cleanup;
    }

9678 9679 9680
    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

9681 9682 9683
    if (virDomainObjIsActive(vm)) {
        qemuDomainObjPrivatePtr priv = vm->privateData;
        qemuDomainObjEnterMonitor(vm);
9684
        ret = qemuMonitorGetMemoryStats(priv->mon, stats, nr_stats);
9685 9686
        qemuDomainObjExitMonitor(vm);
    } else {
9687 9688
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
9689 9690
    }

9691 9692 9693
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;

9694 9695 9696 9697 9698 9699
cleanup:
    if (vm)
        virDomainObjUnlock(vm);
    return ret;
}

9700 9701 9702 9703 9704 9705 9706
static int
qemudDomainBlockPeek (virDomainPtr dom,
                      const char *path,
                      unsigned long long offset, size_t size,
                      void *buffer,
                      unsigned int flags ATTRIBUTE_UNUSED)
{
9707 9708 9709
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int fd = -1, ret = -1, i;
9710

9711
    qemuDriverLock(driver);
9712
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
9713 9714
    qemuDriverUnlock(driver);

9715
    if (!vm) {
9716 9717
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
9718 9719
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
9720
        goto cleanup;
9721 9722 9723
    }

    if (!path || path[0] == '\0') {
9724 9725
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("NULL or empty path"));
9726
        goto cleanup;
9727 9728 9729
    }

    /* Check the path belongs to this domain. */
9730 9731
    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (vm->def->disks[i]->src != NULL &&
9732 9733 9734 9735
            STREQ (vm->def->disks[i]->src, path)) {
            ret = 0;
            break;
        }
9736 9737
    }

9738 9739 9740 9741 9742
    if (ret == 0) {
        ret = -1;
        /* The path is correct, now try to open it and get its size. */
        fd = open (path, O_RDONLY);
        if (fd == -1) {
9743 9744
            virReportSystemError(errno,
                                 _("%s: failed to open"), path);
9745 9746
            goto cleanup;
        }
9747

9748 9749 9750 9751 9752 9753
        /* Seek and read. */
        /* NB. Because we configure with AC_SYS_LARGEFILE, off_t should
         * be 64 bits on all platforms.
         */
        if (lseek (fd, offset, SEEK_SET) == (off_t) -1 ||
            saferead (fd, buffer, size) == (ssize_t) -1) {
9754 9755
            virReportSystemError(errno,
                                 _("%s: failed to seek or read"), path);
9756 9757 9758 9759 9760
            goto cleanup;
        }

        ret = 0;
    } else {
9761 9762
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("invalid path"));
9763 9764
    }

9765 9766 9767
cleanup:
    if (fd >= 0)
        close (fd);
9768 9769
    if (vm)
        virDomainObjUnlock(vm);
9770 9771 9772
    return ret;
}

R
Richard W.M. Jones 已提交
9773 9774 9775 9776 9777 9778
static int
qemudDomainMemoryPeek (virDomainPtr dom,
                       unsigned long long offset, size_t size,
                       void *buffer,
                       unsigned int flags)
{
9779 9780
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
9781
    char *tmp = NULL;
R
Richard W.M. Jones 已提交
9782 9783
    int fd = -1, ret = -1;

9784
    qemuDriverLock(driver);
9785
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
9786
    qemuDriverUnlock(driver);
R
Richard W.M. Jones 已提交
9787 9788

    if (!vm) {
9789 9790
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
9791 9792
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
9793 9794 9795
        goto cleanup;
    }

9796
    if (flags != VIR_MEMORY_VIRTUAL && flags != VIR_MEMORY_PHYSICAL) {
9797 9798
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("flags parameter must be VIR_MEMORY_VIRTUAL or VIR_MEMORY_PHYSICAL"));
9799
        goto cleanup;
R
Richard W.M. Jones 已提交
9800 9801
    }

9802 9803 9804
    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
9805
    if (!virDomainObjIsActive(vm)) {
9806 9807
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
9808
        goto endjob;
R
Richard W.M. Jones 已提交
9809 9810
    }

9811
    if (virAsprintf(&tmp, "%s/qemu.mem.XXXXXX", driver->cacheDir) < 0) {
9812
        virReportOOMError();
9813
        goto endjob;
9814 9815
    }

R
Richard W.M. Jones 已提交
9816 9817
    /* Create a temporary filename. */
    if ((fd = mkstemp (tmp)) == -1) {
9818 9819
        virReportSystemError(errno,
                             _("mkstemp(\"%s\") failed"), tmp);
9820
        goto endjob;
R
Richard W.M. Jones 已提交
9821 9822
    }

9823
    qemuDomainObjPrivatePtr priv = vm->privateData;
9824
    qemuDomainObjEnterMonitor(vm);
9825
    if (flags == VIR_MEMORY_VIRTUAL) {
9826 9827
        if (qemuMonitorSaveVirtualMemory(priv->mon, offset, size, tmp) < 0) {
            qemuDomainObjExitMonitor(vm);
9828
            goto endjob;
9829
        }
9830
    } else {
9831 9832
        if (qemuMonitorSavePhysicalMemory(priv->mon, offset, size, tmp) < 0) {
            qemuDomainObjExitMonitor(vm);
9833
            goto endjob;
9834
        }
R
Richard W.M. Jones 已提交
9835
    }
9836
    qemuDomainObjExitMonitor(vm);
R
Richard W.M. Jones 已提交
9837 9838 9839

    /* Read the memory file into buffer. */
    if (saferead (fd, buffer, size) == (ssize_t) -1) {
9840 9841 9842
        virReportSystemError(errno,
                             _("failed to read temporary file "
                               "created with template %s"), tmp);
9843
        goto endjob;
R
Richard W.M. Jones 已提交
9844 9845 9846
    }

    ret = 0;
9847

9848
endjob:
9849 9850
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
9851

9852
cleanup:
9853
    VIR_FREE(tmp);
R
Richard W.M. Jones 已提交
9854 9855
    if (fd >= 0) close (fd);
    unlink (tmp);
9856 9857
    if (vm)
        virDomainObjUnlock(vm);
R
Richard W.M. Jones 已提交
9858 9859 9860
    return ret;
}

9861

9862 9863 9864 9865 9866 9867 9868 9869 9870 9871
static int qemuDomainGetBlockInfo(virDomainPtr dom,
                                  const char *path,
                                  virDomainBlockInfoPtr info,
                                  unsigned int flags) {
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
    int fd = -1;
    off_t end;
    virStorageFileMetadata meta;
9872
    virDomainDiskDefPtr disk = NULL;
9873 9874
    struct stat sb;
    int i;
9875
    int format;
9876 9877 9878 9879 9880 9881 9882 9883 9884 9885 9886 9887 9888 9889 9890 9891 9892 9893 9894 9895 9896 9897 9898 9899

    virCheckFlags(0, -1);

    qemuDriverLock(driver);
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
    qemuDriverUnlock(driver);
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
        goto cleanup;
    }

    if (!path || path[0] == '\0') {
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("NULL or empty path"));
        goto cleanup;
    }

    /* Check the path belongs to this domain. */
    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (vm->def->disks[i]->src != NULL &&
            STREQ (vm->def->disks[i]->src, path)) {
9900
            disk = vm->def->disks[i];
9901 9902 9903 9904
            break;
        }
    }

9905
    if (!disk) {
9906 9907 9908 9909 9910 9911 9912 9913 9914 9915 9916 9917 9918 9919
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("invalid path %s not assigned to domain"), path);
        goto cleanup;
    }

    /* The path is correct, now try to open it and get its size. */
    fd = open (path, O_RDONLY);
    if (fd == -1) {
        virReportSystemError(errno,
                             _("failed to open path '%s'"), path);
        goto cleanup;
    }

    /* Probe for magic formats */
9920 9921 9922 9923 9924 9925 9926 9927
    if (disk->driverType) {
        if ((format = virStorageFileFormatTypeFromString(disk->driverType)) < 0) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("unknown disk format %s for %s"),
                            disk->driverType, disk->src);
            goto cleanup;
        }
    } else {
9928 9929 9930 9931 9932 9933 9934
        if (driver->allowDiskFormatProbing) {
            if ((format = virStorageFileProbeFormat(disk->src)) < 0)
                goto cleanup;
        } else {
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("no disk format for %s and probing is disabled"),
                            disk->src);
9935
            goto cleanup;
9936
        }
9937 9938 9939 9940 9941
    }

    if (virStorageFileGetMetadataFromFD(path, fd,
                                        format,
                                        &meta) < 0)
9942 9943 9944 9945 9946 9947 9948 9949 9950 9951
        goto cleanup;

    /* Get info for normal formats */
    if (fstat(fd, &sb) < 0) {
        virReportSystemError(errno,
                             _("cannot stat file '%s'"), path);
        goto cleanup;
    }

    if (S_ISREG(sb.st_mode)) {
9952
#ifndef WIN32
9953 9954 9955 9956 9957 9958 9959 9960 9961 9962 9963 9964 9965 9966 9967 9968 9969 9970 9971 9972 9973 9974 9975 9976 9977 9978 9979 9980
        info->physical = (unsigned long long)sb.st_blocks *
            (unsigned long long)DEV_BSIZE;
#else
        info->physical = sb.st_size;
#endif
        /* Regular files may be sparse, so logical size (capacity) is not same
         * as actual physical above
         */
        info->capacity = sb.st_size;
    } else {
        /* NB. Because we configure with AC_SYS_LARGEFILE, off_t should
         * be 64 bits on all platforms.
         */
        end = lseek (fd, 0, SEEK_END);
        if (end == (off_t)-1) {
            virReportSystemError(errno,
                                 _("failed to seek to end of %s"), path);
            goto cleanup;
        }
        info->physical = end;
        info->capacity = end;
    }

    /* If the file we probed has a capacity set, then override
     * what we calculated from file/block extents */
    if (meta.capacity)
        info->capacity = meta.capacity;

9981
    /* Set default value .. */
9982 9983
    info->allocation = info->physical;

9984 9985 9986 9987 9988
    /* ..but if guest is running & not using raw
       disk format and on a block device, then query
       highest allocated extent from QEMU */
    if (virDomainObjIsActive(vm) &&
        disk->type == VIR_DOMAIN_DISK_TYPE_BLOCK &&
9989
        format != VIR_STORAGE_FILE_RAW &&
9990 9991 9992 9993 9994 9995 9996 9997 9998 9999 10000 10001 10002 10003 10004 10005
        S_ISBLK(sb.st_mode)) {
        qemuDomainObjPrivatePtr priv = vm->privateData;
        if (qemuDomainObjBeginJob(vm) < 0)
            goto cleanup;

        qemuDomainObjEnterMonitor(vm);
        ret = qemuMonitorGetBlockExtent(priv->mon,
                                        disk->info.alias,
                                        &info->allocation);
        qemuDomainObjExitMonitor(vm);

        if (qemuDomainObjEndJob(vm) == 0)
            vm = NULL;
    } else {
        ret = 0;
    }
10006 10007 10008 10009 10010 10011 10012 10013 10014 10015

cleanup:
    if (fd != -1)
        close(fd);
    if (vm)
        virDomainObjUnlock(vm);
    return ret;
}


10016
static int
10017 10018 10019 10020
qemuDomainEventRegister(virConnectPtr conn,
                        virConnectDomainEventCallback callback,
                        void *opaque,
                        virFreeCallback freecb)
10021
{
10022 10023 10024
    struct qemud_driver *driver = conn->privateData;
    int ret;

10025
    qemuDriverLock(driver);
10026 10027
    ret = virDomainEventCallbackListAdd(conn, driver->domainEventCallbacks,
                                        callback, opaque, freecb);
10028
    qemuDriverUnlock(driver);
10029

10030
    return ret;
10031 10032
}

10033

10034
static int
10035 10036
qemuDomainEventDeregister(virConnectPtr conn,
                          virConnectDomainEventCallback callback)
10037
{
10038 10039 10040
    struct qemud_driver *driver = conn->privateData;
    int ret;

10041
    qemuDriverLock(driver);
10042 10043 10044 10045 10046 10047
    if (driver->domainEventDispatching)
        ret = virDomainEventCallbackListMarkDelete(conn, driver->domainEventCallbacks,
                                                   callback);
    else
        ret = virDomainEventCallbackListRemove(conn, driver->domainEventCallbacks,
                                               callback);
10048
    qemuDriverUnlock(driver);
10049

10050
    return ret;
10051 10052
}

10053 10054 10055 10056 10057 10058 10059 10060 10061 10062 10063 10064 10065 10066 10067 10068 10069 10070 10071 10072 10073 10074 10075 10076 10077 10078 10079 10080 10081 10082 10083 10084 10085 10086 10087 10088 10089 10090 10091 10092 10093 10094 10095

static int
qemuDomainEventRegisterAny(virConnectPtr conn,
                           virDomainPtr dom,
                           int eventID,
                           virConnectDomainEventGenericCallback callback,
                           void *opaque,
                           virFreeCallback freecb)
{
    struct qemud_driver *driver = conn->privateData;
    int ret;

    qemuDriverLock(driver);
    ret = virDomainEventCallbackListAddID(conn,
                                          driver->domainEventCallbacks,
                                          dom, eventID,
                                          callback, opaque, freecb);
    qemuDriverUnlock(driver);

    return ret;
}


static int
qemuDomainEventDeregisterAny(virConnectPtr conn,
                             int callbackID)
{
    struct qemud_driver *driver = conn->privateData;
    int ret;

    qemuDriverLock(driver);
    if (driver->domainEventDispatching)
        ret = virDomainEventCallbackListMarkDeleteID(conn, driver->domainEventCallbacks,
                                                     callbackID);
    else
        ret = virDomainEventCallbackListRemoveID(conn, driver->domainEventCallbacks,
                                                 callbackID);
    qemuDriverUnlock(driver);

    return ret;
}


10096 10097
static void qemuDomainEventDispatchFunc(virConnectPtr conn,
                                        virDomainEventPtr event,
10098
                                        virConnectDomainEventGenericCallback cb,
10099 10100
                                        void *cbopaque,
                                        void *opaque)
10101
{
10102
    struct qemud_driver *driver = opaque;
10103

10104 10105 10106 10107 10108 10109 10110 10111 10112 10113 10114 10115 10116 10117 10118 10119 10120 10121 10122 10123 10124 10125 10126 10127 10128 10129 10130 10131 10132 10133 10134 10135 10136 10137 10138 10139 10140 10141 10142 10143 10144 10145 10146 10147
    /* Drop the lock whle dispatching, for sake of re-entrancy */
    qemuDriverUnlock(driver);
    virDomainEventDispatchDefaultFunc(conn, event, cb, cbopaque, NULL);
    qemuDriverLock(driver);
}

static void qemuDomainEventFlush(int timer ATTRIBUTE_UNUSED, void *opaque)
{
    struct qemud_driver *driver = opaque;
    virDomainEventQueue tempQueue;

    qemuDriverLock(driver);

    driver->domainEventDispatching = 1;

    /* Copy the queue, so we're reentrant safe */
    tempQueue.count = driver->domainEventQueue->count;
    tempQueue.events = driver->domainEventQueue->events;
    driver->domainEventQueue->count = 0;
    driver->domainEventQueue->events = NULL;

    virEventUpdateTimeout(driver->domainEventTimer, -1);
    virDomainEventQueueDispatch(&tempQueue,
                                driver->domainEventCallbacks,
                                qemuDomainEventDispatchFunc,
                                driver);

    /* Purge any deleted callbacks */
    virDomainEventCallbackListPurgeMarked(driver->domainEventCallbacks);

    driver->domainEventDispatching = 0;
    qemuDriverUnlock(driver);
}


/* driver must be locked before calling */
static void qemuDomainEventQueue(struct qemud_driver *driver,
                                 virDomainEventPtr event)
{
    if (virDomainEventQueuePush(driver->domainEventQueue,
                                event) < 0)
        virDomainEventFree(event);
    if (qemu_driver->domainEventQueue->count == 1)
        virEventUpdateTimeout(driver->domainEventTimer, 0);
10148 10149
}

D
Daniel Veillard 已提交
10150 10151
/* Migration support. */

C
Chris Lalancette 已提交
10152 10153 10154 10155 10156 10157 10158 10159 10160 10161 10162 10163 10164 10165 10166 10167 10168 10169 10170
/* Tunnelled migration stream support */
struct qemuStreamMigFile {
    int fd;

    int watch;
    unsigned int cbRemoved;
    unsigned int dispatching;
    virStreamEventCallback cb;
    void *opaque;
    virFreeCallback ff;
};

static int qemuStreamMigRemoveCallback(virStreamPtr stream)
{
    struct qemud_driver *driver = stream->conn->privateData;
    struct qemuStreamMigFile *qemust = stream->privateData;
    int ret = -1;

    if (!qemust) {
10171 10172
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream is not open"));
C
Chris Lalancette 已提交
10173 10174 10175 10176 10177
        return -1;
    }

    qemuDriverLock(driver);
    if (qemust->watch == 0) {
10178 10179
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream does not have a callback registered"));
C
Chris Lalancette 已提交
10180 10181 10182 10183 10184 10185 10186 10187 10188 10189 10190 10191 10192 10193 10194 10195 10196 10197 10198 10199 10200 10201 10202 10203 10204 10205 10206 10207
        goto cleanup;
    }

    virEventRemoveHandle(qemust->watch);
    if (qemust->dispatching)
        qemust->cbRemoved = 1;
    else if (qemust->ff)
        (qemust->ff)(qemust->opaque);

    qemust->watch = 0;
    qemust->ff = NULL;
    qemust->cb = NULL;
    qemust->opaque = NULL;

    ret = 0;

cleanup:
    qemuDriverUnlock(driver);
    return ret;
}

static int qemuStreamMigUpdateCallback(virStreamPtr stream, int events)
{
    struct qemud_driver *driver = stream->conn->privateData;
    struct qemuStreamMigFile *qemust = stream->privateData;
    int ret = -1;

    if (!qemust) {
10208 10209
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream is not open"));
C
Chris Lalancette 已提交
10210 10211 10212 10213 10214
        return -1;
    }

    qemuDriverLock(driver);
    if (qemust->watch == 0) {
10215 10216
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream does not have a callback registered"));
C
Chris Lalancette 已提交
10217 10218 10219 10220 10221 10222 10223 10224 10225 10226 10227 10228 10229 10230 10231 10232 10233 10234 10235 10236 10237 10238 10239 10240 10241 10242 10243 10244 10245 10246 10247 10248 10249 10250 10251 10252 10253 10254 10255 10256 10257 10258 10259 10260 10261 10262 10263 10264 10265 10266 10267 10268 10269 10270 10271 10272 10273
        goto cleanup;
    }

    virEventUpdateHandle(qemust->watch, events);

    ret = 0;

cleanup:
    qemuDriverUnlock(driver);
    return ret;
}

static void qemuStreamMigEvent(int watch ATTRIBUTE_UNUSED,
                               int fd ATTRIBUTE_UNUSED,
                               int events,
                               void *opaque)
{
    virStreamPtr stream = opaque;
    struct qemud_driver *driver = stream->conn->privateData;
    struct qemuStreamMigFile *qemust = stream->privateData;
    virStreamEventCallback cb;
    void *cbopaque;
    virFreeCallback ff;

    qemuDriverLock(driver);
    if (!qemust || !qemust->cb) {
        qemuDriverUnlock(driver);
        return;
    }

    cb = qemust->cb;
    cbopaque = qemust->opaque;
    ff = qemust->ff;
    qemust->dispatching = 1;
    qemuDriverUnlock(driver);

    cb(stream, events, cbopaque);

    qemuDriverLock(driver);
    qemust->dispatching = 0;
    if (qemust->cbRemoved && ff)
        (ff)(cbopaque);
    qemuDriverUnlock(driver);
}

static int
qemuStreamMigAddCallback(virStreamPtr st,
                         int events,
                         virStreamEventCallback cb,
                         void *opaque,
                         virFreeCallback ff)
{
    struct qemud_driver *driver = st->conn->privateData;
    struct qemuStreamMigFile *qemust = st->privateData;
    int ret = -1;

    if (!qemust) {
10274 10275
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream is not open"));
C
Chris Lalancette 已提交
10276 10277 10278 10279 10280
        return -1;
    }

    qemuDriverLock(driver);
    if (qemust->watch != 0) {
10281 10282
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream already has a callback registered"));
C
Chris Lalancette 已提交
10283 10284 10285 10286 10287 10288 10289 10290
        goto cleanup;
    }

    if ((qemust->watch = virEventAddHandle(qemust->fd,
                                           events,
                                           qemuStreamMigEvent,
                                           st,
                                           NULL)) < 0) {
10291 10292
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("cannot register file watch on stream"));
C
Chris Lalancette 已提交
10293 10294 10295 10296 10297 10298 10299 10300 10301 10302 10303 10304 10305 10306 10307 10308 10309 10310 10311 10312 10313 10314 10315 10316 10317 10318 10319 10320 10321 10322 10323 10324
        goto cleanup;
    }

    qemust->cbRemoved = 0;
    qemust->cb = cb;
    qemust->opaque = opaque;
    qemust->ff = ff;
    virStreamRef(st);

    ret = 0;

cleanup:
    qemuDriverUnlock(driver);
    return ret;
}

static void qemuStreamMigFree(struct qemuStreamMigFile *qemust)
{
    if (qemust->fd != -1)
        close(qemust->fd);
    VIR_FREE(qemust);
}

static struct qemuStreamMigFile *qemuStreamMigOpen(virStreamPtr st,
                                                   const char *unixfile)
{
    struct qemuStreamMigFile *qemust = NULL;
    struct sockaddr_un sa_qemu;
    int i = 0;
    int timeout = 3;
    int ret;

10325
    if (VIR_ALLOC(qemust) < 0) {
10326
        virReportOOMError();
C
Chris Lalancette 已提交
10327
        return NULL;
10328
    }
C
Chris Lalancette 已提交
10329 10330 10331 10332 10333 10334 10335 10336 10337 10338 10339 10340 10341 10342 10343 10344 10345 10346 10347 10348 10349 10350 10351 10352 10353 10354 10355 10356 10357 10358 10359 10360 10361 10362 10363 10364 10365 10366 10367 10368 10369 10370 10371 10372 10373 10374 10375 10376 10377 10378 10379 10380 10381 10382 10383 10384 10385 10386 10387 10388 10389

    qemust->fd = socket(AF_UNIX, SOCK_STREAM, 0);
    if (qemust->fd < 0)
        goto cleanup;

    memset(&sa_qemu, 0, sizeof(sa_qemu));
    sa_qemu.sun_family = AF_UNIX;
    if (virStrcpy(sa_qemu.sun_path, unixfile, sizeof(sa_qemu.sun_path)) == NULL)
        goto cleanup;

    do {
        ret = connect(qemust->fd, (struct sockaddr *)&sa_qemu, sizeof(sa_qemu));
        if (ret == 0)
            break;

        if (errno == ENOENT || errno == ECONNREFUSED) {
            /* ENOENT       : Socket may not have shown up yet
             * ECONNREFUSED : Leftover socket hasn't been removed yet */
            continue;
        }

        goto cleanup;
    } while ((++i <= timeout*5) && (usleep(.2 * 1000000) <= 0));

    if ((st->flags & VIR_STREAM_NONBLOCK) && virSetNonBlock(qemust->fd) < 0)
        goto cleanup;

    return qemust;

cleanup:
    qemuStreamMigFree(qemust);
    return NULL;
}

static int
qemuStreamMigClose(virStreamPtr st)
{
    struct qemud_driver *driver = st->conn->privateData;
    struct qemuStreamMigFile *qemust = st->privateData;

    if (!qemust)
        return 0;

    qemuDriverLock(driver);

    qemuStreamMigFree(qemust);

    st->privateData = NULL;

    qemuDriverUnlock(driver);

    return 0;
}

static int qemuStreamMigWrite(virStreamPtr st, const char *bytes, size_t nbytes)
{
    struct qemud_driver *driver = st->conn->privateData;
    struct qemuStreamMigFile *qemust = st->privateData;
    int ret;

    if (!qemust) {
10390 10391
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream is not open"));
C
Chris Lalancette 已提交
10392 10393 10394 10395 10396 10397 10398 10399 10400 10401 10402 10403 10404 10405
        return -1;
    }

    qemuDriverLock(driver);

retry:
    ret = write(qemust->fd, bytes, nbytes);
    if (ret < 0) {
        if (errno == EAGAIN || errno == EWOULDBLOCK) {
            ret = -2;
        } else if (errno == EINTR) {
            goto retry;
        } else {
            ret = -1;
10406
            virReportSystemError(errno, "%s",
C
Chris Lalancette 已提交
10407 10408 10409 10410 10411 10412 10413 10414 10415 10416 10417 10418 10419 10420 10421 10422 10423 10424 10425 10426 10427 10428 10429 10430 10431 10432 10433 10434 10435 10436 10437 10438 10439 10440 10441 10442 10443 10444
                                 _("cannot write to stream"));
        }
    }

    qemuDriverUnlock(driver);
    return ret;
}

static virStreamDriver qemuStreamMigDrv = {
    .streamSend = qemuStreamMigWrite,
    .streamFinish = qemuStreamMigClose,
    .streamAbort = qemuStreamMigClose,
    .streamAddCallback = qemuStreamMigAddCallback,
    .streamUpdateCallback = qemuStreamMigUpdateCallback,
    .streamRemoveCallback = qemuStreamMigRemoveCallback
};

/* Prepare is the first step, and it runs on the destination host.
 *
 * This version starts an empty VM listening on a localhost TCP port, and
 * sets up the corresponding virStream to handle the incoming data.
 */
static int
qemudDomainMigratePrepareTunnel(virConnectPtr dconn,
                                virStreamPtr st,
                                unsigned long flags,
                                const char *dname,
                                unsigned long resource ATTRIBUTE_UNUSED,
                                const char *dom_xml)
{
    struct qemud_driver *driver = dconn->privateData;
    virDomainDefPtr def = NULL;
    virDomainObjPtr vm = NULL;
    char *migrateFrom;
    virDomainEventPtr event = NULL;
    int ret = -1;
    int internalret;
    char *unixfile = NULL;
10445
    unsigned long long qemuCmdFlags;
C
Chris Lalancette 已提交
10446
    struct qemuStreamMigFile *qemust = NULL;
10447 10448 10449 10450 10451 10452 10453 10454
    qemuDomainObjPrivatePtr priv = NULL;
    struct timeval now;

    if (gettimeofday(&now, NULL) < 0) {
        virReportSystemError(errno, "%s",
                             _("cannot get time of day"));
        return -1;
    }
C
Chris Lalancette 已提交
10455 10456 10457

    qemuDriverLock(driver);
    if (!dom_xml) {
10458 10459
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("no domain XML passed"));
C
Chris Lalancette 已提交
10460 10461 10462
        goto cleanup;
    }
    if (!(flags & VIR_MIGRATE_TUNNELLED)) {
10463
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
C
Chris Lalancette 已提交
10464 10465 10466 10467
                         "%s", _("PrepareTunnel called but no TUNNELLED flag set"));
        goto cleanup;
    }
    if (st == NULL) {
10468 10469
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("tunnelled migration requested but NULL stream passed"));
C
Chris Lalancette 已提交
10470 10471 10472 10473
        goto cleanup;
    }

    /* Parse the domain XML. */
10474
    if (!(def = virDomainDefParseString(driver->caps, dom_xml,
C
Chris Lalancette 已提交
10475
                                        VIR_DOMAIN_XML_INACTIVE))) {
10476 10477
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to parse XML"));
C
Chris Lalancette 已提交
10478 10479 10480 10481
        goto cleanup;
    }

    /* Target domain name, maybe renamed. */
10482 10483 10484 10485 10486 10487
    if (dname) {
        VIR_FREE(def->name);
        def->name = strdup(dname);
        if (def->name == NULL)
            goto cleanup;
    }
C
Chris Lalancette 已提交
10488

10489 10490
    if (virDomainObjIsDuplicate(&driver->domains, def, 1) < 0)
        goto cleanup;
C
Chris Lalancette 已提交
10491

10492
    if (!(vm = virDomainAssignDef(driver->caps,
C
Chris Lalancette 已提交
10493
                                  &driver->domains,
10494
                                  def, true))) {
10495
        /* virDomainAssignDef already set the error */
C
Chris Lalancette 已提交
10496 10497 10498
        goto cleanup;
    }
    def = NULL;
10499
    priv = vm->privateData;
C
Chris Lalancette 已提交
10500

10501 10502
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;
10503
    priv->jobActive = QEMU_JOB_MIGRATION_OUT;
10504

C
Chris Lalancette 已提交
10505 10506 10507 10508
    /* Domain starts inactive, even if the domain XML had an id field. */
    vm->def->id = -1;

    if (virAsprintf(&unixfile, "%s/qemu.tunnelmigrate.dest.%s",
10509
                    driver->libDir, vm->def->name) < 0) {
10510
        virReportOOMError();
10511
        goto endjob;
C
Chris Lalancette 已提交
10512 10513 10514 10515 10516
    }
    unlink(unixfile);

    /* check that this qemu version supports the interactive exec */
    if (qemudExtractVersionInfo(vm->def->emulator, NULL, &qemuCmdFlags) < 0) {
10517 10518 10519
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Cannot determine QEMU argv syntax %s"),
                        vm->def->emulator);
10520
        goto endjob;
C
Chris Lalancette 已提交
10521 10522 10523 10524 10525 10526
    }
    if (qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_UNIX)
        internalret = virAsprintf(&migrateFrom, "unix:%s", unixfile);
    else if (qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_EXEC)
        internalret = virAsprintf(&migrateFrom, "exec:nc -U -l %s", unixfile);
    else {
10527 10528
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("Destination qemu is too old to support tunnelled migration"));
10529
        goto endjob;
C
Chris Lalancette 已提交
10530 10531
    }
    if (internalret < 0) {
10532
        virReportOOMError();
10533
        goto endjob;
C
Chris Lalancette 已提交
10534 10535 10536 10537
    }
    /* Start the QEMU daemon, with the same command-line arguments plus
     * -incoming unix:/path/to/file or exec:nc -U /path/to/file
     */
10538 10539
    internalret = qemudStartVMDaemon(dconn, driver, vm, migrateFrom, true,
                                     -1, NULL);
C
Chris Lalancette 已提交
10540 10541 10542 10543 10544 10545 10546 10547 10548
    VIR_FREE(migrateFrom);
    if (internalret < 0) {
        /* Note that we don't set an error here because qemudStartVMDaemon
         * should have already done that.
         */
        if (!vm->persistent) {
            virDomainRemoveInactive(&driver->domains, vm);
            vm = NULL;
        }
10549
        goto endjob;
C
Chris Lalancette 已提交
10550 10551 10552 10553
    }

    qemust = qemuStreamMigOpen(st, unixfile);
    if (qemust == NULL) {
10554
        qemudShutdownVMDaemon(driver, vm, 0);
10555
        if (!vm->persistent) {
10556 10557
            if (qemuDomainObjEndJob(vm) > 0)
                virDomainRemoveInactive(&driver->domains, vm);
10558 10559
            vm = NULL;
        }
10560
        virReportSystemError(errno,
C
Chris Lalancette 已提交
10561 10562
                             _("cannot open unix socket '%s' for tunnelled migration"),
                             unixfile);
10563
        goto endjob;
C
Chris Lalancette 已提交
10564 10565 10566 10567 10568 10569 10570 10571 10572 10573
    }

    st->driver = &qemuStreamMigDrv;
    st->privateData = qemust;

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_MIGRATED);
    ret = 0;

10574
endjob:
10575 10576 10577
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
10578

10579 10580 10581 10582 10583 10584 10585 10586 10587 10588 10589 10590
    /* We set a fake job active which is held across
     * API calls until the finish() call. This prevents
     * any other APIs being invoked while incoming
     * migration is taking place
     */
    if (vm &&
        virDomainObjIsActive(vm)) {
        priv->jobActive = QEMU_JOB_MIGRATION_IN;
        priv->jobInfo.type = VIR_DOMAIN_JOB_UNBOUNDED;
        priv->jobStart = (now.tv_sec * 1000ull) + (now.tv_usec / 1000);
    }

C
Chris Lalancette 已提交
10591 10592
cleanup:
    virDomainDefFree(def);
10593 10594
    if (unixfile)
        unlink(unixfile);
C
Chris Lalancette 已提交
10595 10596 10597 10598 10599 10600 10601 10602 10603
    VIR_FREE(unixfile);
    if (vm)
        virDomainObjUnlock(vm);
    if (event)
        qemuDomainEventQueue(driver, event);
    qemuDriverUnlock(driver);
    return ret;
}

D
Daniel Veillard 已提交
10604 10605 10606 10607
/* Prepare is the first step, and it runs on the destination host.
 *
 * This starts an empty VM listening on a TCP port.
 */
10608
static int ATTRIBUTE_NONNULL (5)
D
Daniel Veillard 已提交
10609 10610 10611 10612 10613
qemudDomainMigratePrepare2 (virConnectPtr dconn,
                            char **cookie ATTRIBUTE_UNUSED,
                            int *cookielen ATTRIBUTE_UNUSED,
                            const char *uri_in,
                            char **uri_out,
C
Chris Lalancette 已提交
10614
                            unsigned long flags,
D
Daniel Veillard 已提交
10615 10616 10617 10618 10619
                            const char *dname,
                            unsigned long resource ATTRIBUTE_UNUSED,
                            const char *dom_xml)
{
    static int port = 0;
10620 10621
    struct qemud_driver *driver = dconn->privateData;
    virDomainDefPtr def = NULL;
D
Daniel Veillard 已提交
10622 10623
    virDomainObjPtr vm = NULL;
    int this_port;
C
Chris Lalancette 已提交
10624
    char *hostname = NULL;
D
Daniel Veillard 已提交
10625 10626
    char migrateFrom [64];
    const char *p;
10627
    virDomainEventPtr event = NULL;
10628
    int ret = -1;
10629
    int internalret;
10630 10631 10632 10633 10634 10635 10636 10637
    qemuDomainObjPrivatePtr priv = NULL;
    struct timeval now;

    if (gettimeofday(&now, NULL) < 0) {
        virReportSystemError(errno, "%s",
                             _("cannot get time of day"));
        return -1;
    }
10638

C
Chris Lalancette 已提交
10639 10640 10641 10642 10643 10644 10645 10646 10647
    virCheckFlags(VIR_MIGRATE_LIVE |
                  VIR_MIGRATE_PEER2PEER |
                  VIR_MIGRATE_TUNNELLED |
                  VIR_MIGRATE_PERSIST_DEST |
                  VIR_MIGRATE_UNDEFINE_SOURCE |
                  VIR_MIGRATE_PAUSED |
                  VIR_MIGRATE_NON_SHARED_DISK |
                  VIR_MIGRATE_NON_SHARED_INC, -1);

10648
    *uri_out = NULL;
D
Daniel Veillard 已提交
10649

10650
    qemuDriverLock(driver);
C
Chris Lalancette 已提交
10651 10652 10653 10654
    if (flags & VIR_MIGRATE_TUNNELLED) {
        /* this is a logical error; we never should have gotten here with
         * VIR_MIGRATE_TUNNELLED set
         */
10655 10656
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("Tunnelled migration requested but invalid RPC method called"));
C
Chris Lalancette 已提交
10657 10658 10659
        goto cleanup;
    }

D
Daniel Veillard 已提交
10660
    if (!dom_xml) {
10661 10662
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("no domain XML passed"));
10663
        goto cleanup;
D
Daniel Veillard 已提交
10664 10665 10666 10667 10668 10669 10670 10671 10672 10673 10674 10675 10676 10677 10678 10679 10680
    }

    /* 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 */
C
Chris Lalancette 已提交
10681
        if ((hostname = virGetHostname(NULL)) == NULL)
10682
            goto cleanup;
D
Daniel Veillard 已提交
10683

C
Chris Lalancette 已提交
10684 10685 10686 10687 10688 10689
        if (STRPREFIX(hostname, "localhost")) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("hostname on destination resolved to localhost, but migration requires an FQDN"));
            goto cleanup;
        }

10690 10691 10692 10693 10694
        /* XXX this really should have been a properly well-formed
         * URI, but we can't add in tcp:// now without breaking
         * compatability with old targets. We at least make the
         * new targets accept both syntaxes though.
         */
D
Daniel Veillard 已提交
10695
        /* Caller frees */
10696 10697
        internalret = virAsprintf(uri_out, "tcp:%s:%d", hostname, this_port);
        if (internalret < 0) {
10698
            virReportOOMError();
10699
            goto cleanup;
D
Daniel Veillard 已提交
10700 10701 10702 10703 10704 10705
        }
    } 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.
         */
10706
        if (!STRPREFIX (uri_in, "tcp:")) {
10707 10708
            qemuReportError (VIR_ERR_INVALID_ARG,
                             "%s", _("only tcp URIs are supported for KVM/QEMU migrations"));
10709
            goto cleanup;
D
Daniel Veillard 已提交
10710 10711 10712 10713
        }

        /* Get the port number. */
        p = strrchr (uri_in, ':');
10714 10715 10716 10717 10718 10719 10720 10721
        if (p == strchr(uri_in, ':')) {
            /* Generate a port */
            this_port = QEMUD_MIGRATION_FIRST_PORT + port++;
            if (port == QEMUD_MIGRATION_NUM_PORTS)
                port = 0;

            /* Caller frees */
            if (virAsprintf(uri_out, "%s:%d", uri_in, this_port) < 0) {
10722
                virReportOOMError();
10723 10724 10725 10726 10727 10728 10729
                goto cleanup;
            }

        } else {
            p++; /* definitely has a ':' in it, see above */
            this_port = virParseNumber (&p);
            if (this_port == -1 || p-uri_in != strlen (uri_in)) {
10730 10731
                qemuReportError(VIR_ERR_INVALID_ARG,
                                "%s", _("URI ended with incorrect ':port'"));
10732 10733
                goto cleanup;
            }
D
Daniel Veillard 已提交
10734 10735 10736
        }
    }

10737
    if (*uri_out)
10738 10739
        VIR_DEBUG("Generated uri_out=%s", *uri_out);

D
Daniel Veillard 已提交
10740
    /* Parse the domain XML. */
10741
    if (!(def = virDomainDefParseString(driver->caps, dom_xml,
10742
                                        VIR_DOMAIN_XML_INACTIVE))) {
10743 10744
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to parse XML"));
10745
        goto cleanup;
D
Daniel Veillard 已提交
10746 10747 10748
    }

    /* Target domain name, maybe renamed. */
10749 10750 10751 10752 10753 10754
    if (dname) {
        VIR_FREE(def->name);
        def->name = strdup(dname);
        if (def->name == NULL)
            goto cleanup;
    }
D
Daniel Veillard 已提交
10755

10756 10757
    if (virDomainObjIsDuplicate(&driver->domains, def, 1) < 0)
        goto cleanup;
D
Daniel Veillard 已提交
10758

10759
    if (!(vm = virDomainAssignDef(driver->caps,
D
Daniel Veillard 已提交
10760
                                  &driver->domains,
10761
                                  def, true))) {
10762
        /* virDomainAssignDef already set the error */
10763
        goto cleanup;
D
Daniel Veillard 已提交
10764
    }
10765
    def = NULL;
10766
    priv = vm->privateData;
D
Daniel Veillard 已提交
10767

10768 10769
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;
10770
    priv->jobActive = QEMU_JOB_MIGRATION_OUT;
10771

D
Daniel Veillard 已提交
10772 10773 10774 10775 10776 10777 10778
    /* Domain starts inactive, even if the domain XML had an id field. */
    vm->def->id = -1;

    /* Start the QEMU daemon, with the same command-line arguments plus
     * -incoming tcp:0.0.0.0:port
     */
    snprintf (migrateFrom, sizeof (migrateFrom), "tcp:0.0.0.0:%d", this_port);
10779 10780
    if (qemudStartVMDaemon (dconn, driver, vm, migrateFrom, true,
                            -1, NULL) < 0) {
10781 10782 10783
        /* Note that we don't set an error here because qemudStartVMDaemon
         * should have already done that.
         */
10784
        if (!vm->persistent) {
10785 10786
            if (qemuDomainObjEndJob(vm) > 0)
                virDomainRemoveInactive(&driver->domains, vm);
10787 10788
            vm = NULL;
        }
10789
        goto endjob;
D
Daniel Veillard 已提交
10790
    }
10791 10792 10793 10794

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_MIGRATED);
10795
    ret = 0;
D
Daniel Veillard 已提交
10796

10797
endjob:
10798 10799 10800
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
10801

10802 10803 10804 10805 10806 10807 10808 10809 10810 10811 10812 10813
    /* We set a fake job active which is held across
     * API calls until the finish() call. This prevents
     * any other APIs being invoked while incoming
     * migration is taking place
     */
    if (vm &&
        virDomainObjIsActive(vm)) {
        priv->jobActive = QEMU_JOB_MIGRATION_IN;
        priv->jobInfo.type = VIR_DOMAIN_JOB_UNBOUNDED;
        priv->jobStart = (now.tv_sec * 1000ull) + (now.tv_usec / 1000);
    }

10814
cleanup:
C
Chris Lalancette 已提交
10815
    VIR_FREE(hostname);
10816
    virDomainDefFree(def);
C
Chris Lalancette 已提交
10817
    if (ret != 0)
10818
        VIR_FREE(*uri_out);
10819 10820
    if (vm)
        virDomainObjUnlock(vm);
10821 10822
    if (event)
        qemuDomainEventQueue(driver, event);
10823
    qemuDriverUnlock(driver);
10824
    return ret;
C
Chris Lalancette 已提交
10825 10826 10827

}

10828 10829 10830 10831

/* Perform migration using QEMU's native TCP migrate support,
 * not encrypted obviously
 */
10832
static int doNativeMigrate(struct qemud_driver *driver,
10833 10834
                           virDomainObjPtr vm,
                           const char *uri,
10835
                           unsigned int flags,
10836 10837 10838 10839
                           const char *dname ATTRIBUTE_UNUSED,
                           unsigned long resource)
{
    int ret = -1;
10840
    xmlURIPtr uribits = NULL;
10841
    qemuDomainObjPrivatePtr priv = vm->privateData;
10842
    unsigned int background_flags = QEMU_MONITOR_MIGRATE_BACKGROUND;
10843

10844 10845 10846 10847 10848
    /* Issue the migrate command. */
    if (STRPREFIX(uri, "tcp:") && !STRPREFIX(uri, "tcp://")) {
        /* HACK: source host generates bogus URIs, so fix them up */
        char *tmpuri;
        if (virAsprintf(&tmpuri, "tcp://%s", uri + strlen("tcp:")) < 0) {
10849
            virReportOOMError();
10850 10851 10852 10853 10854 10855 10856 10857
            goto cleanup;
        }
        uribits = xmlParseURI(tmpuri);
        VIR_FREE(tmpuri);
    } else {
        uribits = xmlParseURI(uri);
    }
    if (!uribits) {
10858 10859
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("cannot parse URI %s"), uri);
10860 10861 10862
        goto cleanup;
    }

10863
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
10864
    if (resource > 0 &&
10865
        qemuMonitorSetMigrationSpeed(priv->mon, resource) < 0) {
10866
        qemuDomainObjExitMonitorWithDriver(driver, vm);
10867
        goto cleanup;
10868
    }
10869

10870 10871 10872 10873 10874 10875 10876 10877
    if (flags & VIR_MIGRATE_NON_SHARED_DISK)
        background_flags |= QEMU_MONITOR_MIGRATE_NON_SHARED_DISK;

    if (flags & VIR_MIGRATE_NON_SHARED_INC)
        background_flags |= QEMU_MONITOR_MIGRATE_NON_SHARED_INC;

    if (qemuMonitorMigrateToHost(priv->mon, background_flags, uribits->server,
                                 uribits->port) < 0) {
10878
        qemuDomainObjExitMonitorWithDriver(driver, vm);
10879 10880
        goto cleanup;
    }
10881
    qemuDomainObjExitMonitorWithDriver(driver, vm);
10882

10883
    if (qemuDomainWaitForMigrationComplete(driver, vm) < 0)
10884 10885 10886 10887 10888 10889 10890 10891 10892 10893
        goto cleanup;

    ret = 0;

cleanup:
    xmlFreeURI(uribits);
    return ret;
}


10894 10895
#define TUNNEL_SEND_BUF_SIZE 65536

10896
static int doTunnelSendAll(virStreamPtr st,
10897 10898
                           int sock)
{
10899 10900 10901 10902 10903 10904 10905 10906
    char *buffer;
    int nbytes = TUNNEL_SEND_BUF_SIZE;

    if (VIR_ALLOC_N(buffer, TUNNEL_SEND_BUF_SIZE) < 0) {
        virReportOOMError();
        virStreamAbort(st);
        return -1;
    }
10907 10908 10909 10910 10911

    /* XXX should honour the 'resource' parameter here */
    for (;;) {
        nbytes = saferead(sock, buffer, nbytes);
        if (nbytes < 0) {
10912
            virReportSystemError(errno, "%s",
10913
                                 _("tunnelled migration failed to read from qemu"));
10914 10915
            virStreamAbort(st);
            VIR_FREE(buffer);
10916 10917 10918 10919 10920 10921 10922
            return -1;
        }
        else if (nbytes == 0)
            /* EOF; get out of here */
            break;

        if (virStreamSend(st, buffer, nbytes) < 0) {
10923 10924
            qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
                            _("Failed to write migration data to remote libvirtd"));
10925
            VIR_FREE(buffer);
10926 10927 10928 10929
            return -1;
        }
    }

10930 10931
    VIR_FREE(buffer);

10932 10933 10934 10935 10936 10937 10938
    if (virStreamFinish(st) < 0)
        /* virStreamFinish set the error for us */
        return -1;

    return 0;
}

C
Chris Lalancette 已提交
10939
static int doTunnelMigrate(virDomainPtr dom,
10940
                           struct qemud_driver *driver,
10941
                           virConnectPtr dconn,
C
Chris Lalancette 已提交
10942
                           virDomainObjPtr vm,
10943
                           const char *dom_xml,
C
Chris Lalancette 已提交
10944 10945 10946 10947 10948
                           const char *uri,
                           unsigned long flags,
                           const char *dname,
                           unsigned long resource)
{
10949
    qemuDomainObjPrivatePtr priv = vm->privateData;
10950 10951
    int client_sock = -1;
    int qemu_sock = -1;
C
Chris Lalancette 已提交
10952 10953
    struct sockaddr_un sa_qemu, sa_client;
    socklen_t addrlen;
10954
    virDomainPtr ddomain = NULL;
C
Chris Lalancette 已提交
10955
    int retval = -1;
10956
    virStreamPtr st = NULL;
C
Chris Lalancette 已提交
10957 10958
    char *unixfile = NULL;
    int internalret;
10959
    unsigned long long qemuCmdFlags;
C
Chris Lalancette 已提交
10960 10961
    int status;
    unsigned long long transferred, remaining, total;
10962
    unsigned int background_flags = QEMU_MONITOR_MIGRATE_BACKGROUND;
C
Chris Lalancette 已提交
10963

10964 10965 10966 10967 10968 10969 10970 10971
    /*
     * The order of operations is important here to avoid touching
     * the source VM until we are very sure we can successfully
     * start the migration operation.
     *
     *   1. setup local support infrastructure (eg sockets)
     *   2. setup destination fully
     *   3. start migration on source
C
Chris Lalancette 已提交
10972 10973
     */

10974

10975
    /* Stage 1. setup local support infrastructure */
C
Chris Lalancette 已提交
10976 10977

    if (virAsprintf(&unixfile, "%s/qemu.tunnelmigrate.src.%s",
10978
                    driver->libDir, vm->def->name) < 0) {
10979
        virReportOOMError();
10980
        goto cleanup;
C
Chris Lalancette 已提交
10981 10982 10983 10984
    }

    qemu_sock = socket(AF_UNIX, SOCK_STREAM, 0);
    if (qemu_sock < 0) {
10985
        virReportSystemError(errno, "%s",
C
Chris Lalancette 已提交
10986
                             _("cannot open tunnelled migration socket"));
10987
        goto cleanup;
C
Chris Lalancette 已提交
10988 10989 10990 10991 10992
    }
    memset(&sa_qemu, 0, sizeof(sa_qemu));
    sa_qemu.sun_family = AF_UNIX;
    if (virStrcpy(sa_qemu.sun_path, unixfile,
                  sizeof(sa_qemu.sun_path)) == NULL) {
10993 10994 10995
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Unix socket '%s' too big for destination"),
                        unixfile);
10996
        goto cleanup;
C
Chris Lalancette 已提交
10997 10998 10999
    }
    unlink(unixfile);
    if (bind(qemu_sock, (struct sockaddr *)&sa_qemu, sizeof(sa_qemu)) < 0) {
11000
        virReportSystemError(errno,
C
Chris Lalancette 已提交
11001 11002
                             _("Cannot bind to unix socket '%s' for tunnelled migration"),
                             unixfile);
11003
        goto cleanup;
C
Chris Lalancette 已提交
11004 11005
    }
    if (listen(qemu_sock, 1) < 0) {
11006
        virReportSystemError(errno,
C
Chris Lalancette 已提交
11007 11008
                             _("Cannot listen on unix socket '%s' for tunnelled migration"),
                             unixfile);
11009
        goto cleanup;
C
Chris Lalancette 已提交
11010 11011
    }

11012 11013 11014 11015 11016 11017 11018
    if (chown(unixfile, qemu_driver->user, qemu_driver->group) < 0) {
        virReportSystemError(errno,
                             _("Cannot change unix socket '%s' owner"),
                             unixfile);
        goto cleanup;
    }

C
Chris Lalancette 已提交
11019 11020
    /* check that this qemu version supports the unix migration */
    if (qemudExtractVersionInfo(vm->def->emulator, NULL, &qemuCmdFlags) < 0) {
11021 11022 11023
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Cannot extract Qemu version from '%s'"),
                        vm->def->emulator);
11024 11025 11026 11027 11028
        goto cleanup;
    }

    if (!(qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_UNIX) &&
        !(qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_EXEC)) {
11029 11030
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("Source qemu is too old to support tunnelled migration"));
11031
        goto cleanup;
C
Chris Lalancette 已提交
11032
    }
11033 11034 11035 11036 11037 11038 11039 11040 11041 11042 11043 11044


    /* Stage 2. setup destination fully
     *
     * Once stage 2 has completed successfully, we *must* call finish
     * to cleanup the target whether we succeed or fail
     */
    st = virStreamNew(dconn, 0);
    if (st == NULL)
        /* virStreamNew only fails on OOM, and it reports the error itself */
        goto cleanup;

11045
    qemuDomainObjEnterRemoteWithDriver(driver, vm);
11046 11047 11048
    internalret = dconn->driver->domainMigratePrepareTunnel(dconn, st,
                                                            flags, dname,
                                                            resource, dom_xml);
11049
    qemuDomainObjExitRemoteWithDriver(driver, vm);
11050 11051 11052 11053 11054

    if (internalret < 0)
        /* domainMigratePrepareTunnel sets the error for us */
        goto cleanup;

11055 11056 11057 11058 11059 11060 11061 11062 11063
    /* the domain may have shutdown or crashed while we had the locks dropped
     * in qemuDomainObjEnterRemoteWithDriver, so check again
     */
    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("guest unexpectedly quit"));
        goto cleanup;
    }

11064
    /*   3. start migration on source */
11065
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
11066 11067 11068 11069 11070 11071 11072 11073
    if (flags & VIR_MIGRATE_NON_SHARED_DISK)
        background_flags |= QEMU_MONITOR_MIGRATE_NON_SHARED_DISK;
    if (flags & VIR_MIGRATE_NON_SHARED_INC)
        background_flags |= QEMU_MONITOR_MIGRATE_NON_SHARED_INC;
    if (qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_UNIX){
        internalret = qemuMonitorMigrateToUnix(priv->mon, background_flags,
                                               unixfile);
    }
C
Chris Lalancette 已提交
11074 11075
    else if (qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_EXEC) {
        const char *args[] = { "nc", "-U", unixfile, NULL };
11076
        internalret = qemuMonitorMigrateToCommand(priv->mon, QEMU_MONITOR_MIGRATE_BACKGROUND, args);
11077 11078
    } else {
        internalret = -1;
C
Chris Lalancette 已提交
11079
    }
11080
    qemuDomainObjExitMonitorWithDriver(driver, vm);
C
Chris Lalancette 已提交
11081
    if (internalret < 0) {
11082 11083
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("tunnelled migration monitor command failed"));
11084
        goto finish;
C
Chris Lalancette 已提交
11085 11086
    }

11087 11088 11089 11090 11091 11092
    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("guest unexpectedly quit"));
        goto cleanup;
    }

11093 11094 11095
    /* From this point onwards we *must* call cancel to abort the
     * migration on source if anything goes wrong */

C
Chris Lalancette 已提交
11096 11097 11098
    /* it is also possible that the migrate didn't fail initially, but
     * rather failed later on.  Check the output of "info migrate"
     */
11099
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
11100 11101
    if (qemuMonitorGetMigrationStatus(priv->mon,
                                      &status,
C
Chris Lalancette 已提交
11102 11103 11104
                                      &transferred,
                                      &remaining,
                                      &total) < 0) {
11105
        qemuDomainObjExitMonitorWithDriver(driver, vm);
11106
        goto cancel;
C
Chris Lalancette 已提交
11107
    }
11108
    qemuDomainObjExitMonitorWithDriver(driver, vm);
C
Chris Lalancette 已提交
11109 11110

    if (status == QEMU_MONITOR_MIGRATION_STATUS_ERROR) {
11111 11112
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s",_("migrate failed"));
11113
        goto cancel;
C
Chris Lalancette 已提交
11114 11115 11116 11117 11118 11119
    }

    addrlen = sizeof(sa_client);
    while ((client_sock = accept(qemu_sock, (struct sockaddr *)&sa_client, &addrlen)) < 0) {
        if (errno == EAGAIN || errno == EINTR)
            continue;
11120
        virReportSystemError(errno, "%s",
C
Chris Lalancette 已提交
11121
                             _("tunnelled migration failed to accept from qemu"));
11122
        goto cancel;
C
Chris Lalancette 已提交
11123 11124
    }

11125
    retval = doTunnelSendAll(st, client_sock);
11126

11127
cancel:
11128
    if (retval != 0 && virDomainObjIsActive(vm)) {
11129
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
11130
        qemuMonitorMigrateCancel(priv->mon);
11131
        qemuDomainObjExitMonitorWithDriver(driver, vm);
11132
    }
C
Chris Lalancette 已提交
11133

11134
finish:
C
Chris Lalancette 已提交
11135
    dname = dname ? dname : dom->name;
11136
    qemuDomainObjEnterRemoteWithDriver(driver, vm);
C
Chris Lalancette 已提交
11137 11138
    ddomain = dconn->driver->domainMigrateFinish2
        (dconn, dname, NULL, 0, uri, flags, retval);
11139
    qemuDomainObjExitRemoteWithDriver(driver, vm);
11140 11141 11142 11143 11144 11145 11146

cleanup:
    if (client_sock != -1)
        close(client_sock);
    if (qemu_sock != -1)
        close(qemu_sock);

C
Chris Lalancette 已提交
11147 11148 11149
    if (ddomain)
        virUnrefDomain(ddomain);

11150 11151 11152 11153
    if (unixfile) {
        unlink(unixfile);
        VIR_FREE(unixfile);
    }
C
Chris Lalancette 已提交
11154

11155 11156 11157
    if (st)
        /* don't call virStreamFree(), because that resets any pending errors */
        virUnrefStream(st);
11158 11159 11160 11161
    return retval;
}


11162 11163 11164 11165
/* This is essentially a simplified re-impl of
 * virDomainMigrateVersion2 from libvirt.c, but running in source
 * libvirtd context, instead of client app context */
static int doNonTunnelMigrate(virDomainPtr dom,
11166
                              struct qemud_driver *driver,
11167 11168 11169 11170 11171 11172 11173 11174 11175 11176 11177
                              virConnectPtr dconn,
                              virDomainObjPtr vm,
                              const char *dom_xml,
                              const char *uri ATTRIBUTE_UNUSED,
                              unsigned long flags,
                              const char *dname,
                              unsigned long resource)
{
    virDomainPtr ddomain = NULL;
    int retval = -1;
    char *uri_out = NULL;
11178
    int rc;
11179

11180
    qemuDomainObjEnterRemoteWithDriver(driver, vm);
11181 11182
    /* NB we don't pass 'uri' into this, since that's the libvirtd
     * URI in this context - so we let dest pick it */
11183 11184 11185 11186 11187 11188 11189 11190 11191
    rc = dconn->driver->domainMigratePrepare2(dconn,
                                              NULL, /* cookie */
                                              0, /* cookielen */
                                              NULL, /* uri */
                                              &uri_out,
                                              flags, dname,
                                              resource, dom_xml);
    qemuDomainObjExitRemoteWithDriver(driver, vm);
    if (rc < 0)
11192 11193 11194
        /* domainMigratePrepare2 sets the error for us */
        goto cleanup;

11195 11196 11197 11198 11199 11200 11201 11202 11203
    /* the domain may have shutdown or crashed while we had the locks dropped
     * in qemuDomainObjEnterRemoteWithDriver, so check again
     */
    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("guest unexpectedly quit"));
        goto cleanup;
    }

11204
    if (uri_out == NULL) {
11205 11206
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("domainMigratePrepare2 did not set uri"));
11207
        goto cleanup;
11208 11209
    }

11210
    if (doNativeMigrate(driver, vm, uri_out, flags, dname, resource) < 0)
11211 11212 11213 11214 11215 11216
        goto finish;

    retval = 0;

finish:
    dname = dname ? dname : dom->name;
11217
    qemuDomainObjEnterRemoteWithDriver(driver, vm);
11218 11219
    ddomain = dconn->driver->domainMigrateFinish2
        (dconn, dname, NULL, 0, uri_out, flags, retval);
11220
    qemuDomainObjExitRemoteWithDriver(driver, vm);
11221 11222 11223 11224 11225 11226 11227 11228 11229

    if (ddomain)
        virUnrefDomain(ddomain);

cleanup:
    return retval;
}


11230
static int doPeer2PeerMigrate(virDomainPtr dom,
11231
                              struct qemud_driver *driver,
11232 11233 11234 11235 11236 11237 11238 11239 11240 11241 11242 11243 11244 11245 11246 11247
                              virDomainObjPtr vm,
                              const char *uri,
                              unsigned long flags,
                              const char *dname,
                              unsigned long resource)
{
    int ret = -1;
    virConnectPtr dconn = NULL;
    char *dom_xml;

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

    dconn = virConnectOpen(uri);
    if (dconn == NULL) {
11248 11249
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("Failed to connect to remote libvirt URI %s"), uri);
11250 11251 11252
        return -1;
    }
    if (!VIR_DRV_SUPPORTS_FEATURE(dconn->driver, dconn,
11253
                                  VIR_DRV_FEATURE_MIGRATION_P2P)) {
11254 11255
        qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
                        _("Destination libvirt does not support peer-to-peer migration protocol"));
11256 11257 11258
        goto cleanup;
    }

11259 11260 11261
    dom_xml = qemudVMDumpXML(driver, vm,
                             VIR_DOMAIN_XML_SECURE |
                             VIR_DOMAIN_XML_UPDATE_CPU);
11262
    if (!dom_xml) {
11263 11264
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to get domain xml"));
11265 11266 11267
        goto cleanup;
    }

11268
    if (flags & VIR_MIGRATE_TUNNELLED)
11269
        ret = doTunnelMigrate(dom, driver, dconn, vm, dom_xml, uri, flags, dname, resource);
11270
    else
11271
        ret = doNonTunnelMigrate(dom, driver, dconn, vm, dom_xml, uri, flags, dname, resource);
11272 11273 11274

cleanup:
    VIR_FREE(dom_xml);
C
Chris Lalancette 已提交
11275 11276 11277
    /* don't call virConnectClose(), because that resets any pending errors */
    virUnrefConnect(dconn);

11278
    return ret;
D
Daniel Veillard 已提交
11279 11280
}

11281

D
Daniel Veillard 已提交
11282 11283 11284 11285 11286 11287
/* Perform is the second step, and it runs on the source host. */
static int
qemudDomainMigratePerform (virDomainPtr dom,
                           const char *cookie ATTRIBUTE_UNUSED,
                           int cookielen ATTRIBUTE_UNUSED,
                           const char *uri,
11288
                           unsigned long flags,
11289
                           const char *dname,
D
Daniel Veillard 已提交
11290 11291
                           unsigned long resource)
{
11292 11293
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
11294
    virDomainEventPtr event = NULL;
11295
    int ret = -1;
11296
    int resume = 0;
11297
    qemuDomainObjPrivatePtr priv;
D
Daniel Veillard 已提交
11298

C
Chris Lalancette 已提交
11299 11300 11301 11302 11303 11304 11305 11306 11307
    virCheckFlags(VIR_MIGRATE_LIVE |
                  VIR_MIGRATE_PEER2PEER |
                  VIR_MIGRATE_TUNNELLED |
                  VIR_MIGRATE_PERSIST_DEST |
                  VIR_MIGRATE_UNDEFINE_SOURCE |
                  VIR_MIGRATE_PAUSED |
                  VIR_MIGRATE_NON_SHARED_DISK |
                  VIR_MIGRATE_NON_SHARED_INC, -1);

11308
    qemuDriverLock(driver);
11309
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
D
Daniel Veillard 已提交
11310
    if (!vm) {
11311 11312
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
11313 11314
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
11315
        goto cleanup;
D
Daniel Veillard 已提交
11316
    }
11317
    priv = vm->privateData;
D
Daniel Veillard 已提交
11318

11319 11320
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;
11321
    priv->jobActive = QEMU_JOB_MIGRATION_OUT;
11322

D
Daniel P. Berrange 已提交
11323
    if (!virDomainObjIsActive(vm)) {
11324 11325
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
11326
        goto endjob;
D
Daniel Veillard 已提交
11327 11328
    }

11329 11330 11331
    memset(&priv->jobInfo, 0, sizeof(priv->jobInfo));
    priv->jobInfo.type = VIR_DOMAIN_JOB_UNBOUNDED;

11332
    resume = vm->state == VIR_DOMAIN_RUNNING;
11333
    if (!(flags & VIR_MIGRATE_LIVE) && vm->state == VIR_DOMAIN_RUNNING) {
11334
        if (qemuDomainMigrateOffline(driver, vm) < 0)
11335
            goto endjob;
11336 11337
    }

11338
    if ((flags & (VIR_MIGRATE_TUNNELLED | VIR_MIGRATE_PEER2PEER))) {
11339
        if (doPeer2PeerMigrate(dom, driver, vm, uri, flags, dname, resource) < 0)
11340
            /* doPeer2PeerMigrate already set the error, so just get out */
11341
            goto endjob;
11342
    } else {
11343
        if (doNativeMigrate(driver, vm, uri, flags, dname, resource) < 0)
11344
            goto endjob;
11345 11346
    }

D
Daniel Veillard 已提交
11347
    /* Clean up the source domain. */
11348
    qemudShutdownVMDaemon(driver, vm, 1);
11349
    resume = 0;
11350 11351 11352 11353

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_MIGRATED);
C
Chris Lalancette 已提交
11354
    if (!vm->persistent || (flags & VIR_MIGRATE_UNDEFINE_SOURCE)) {
11355
        virDomainDeleteConfig(driver->configDir, driver->autostartDir, vm);
11356 11357
        if (qemuDomainObjEndJob(vm) > 0)
            virDomainRemoveInactive(&driver->domains, vm);
11358 11359
        vm = NULL;
    }
11360
    ret = 0;
D
Daniel Veillard 已提交
11361

11362
endjob:
11363
    if (resume && vm->state == VIR_DOMAIN_PAUSED) {
11364
        /* we got here through some sort of failure; start the domain again */
11365
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
11366
        if (qemuMonitorStartCPUs(priv->mon, dom->conn) < 0) {
11367 11368 11369 11370
            /* 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
             */
11371
            VIR_ERROR(_("Failed to resume guest %s after failure"),
11372
                      vm->def->name);
11373
        }
11374
        qemuDomainObjExitMonitorWithDriver(driver, vm);
11375

11376
        vm->state = VIR_DOMAIN_RUNNING;
11377 11378 11379 11380
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_MIGRATED);
    }
11381 11382 11383
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
11384

11385
cleanup:
11386 11387
    if (vm)
        virDomainObjUnlock(vm);
11388 11389
    if (event)
        qemuDomainEventQueue(driver, event);
11390
    qemuDriverUnlock(driver);
11391
    return ret;
D
Daniel Veillard 已提交
11392 11393 11394 11395 11396 11397 11398 11399 11400
}

/* Finish is the third and final step, and it runs on the destination host. */
static virDomainPtr
qemudDomainMigrateFinish2 (virConnectPtr dconn,
                           const char *dname,
                           const char *cookie ATTRIBUTE_UNUSED,
                           int cookielen ATTRIBUTE_UNUSED,
                           const char *uri ATTRIBUTE_UNUSED,
C
Chris Lalancette 已提交
11401
                           unsigned long flags,
D
Daniel Veillard 已提交
11402 11403
                           int retcode)
{
11404 11405 11406
    struct qemud_driver *driver = dconn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
11407
    virDomainEventPtr event = NULL;
11408
    virErrorPtr orig_err;
C
Chris Lalancette 已提交
11409
    int newVM = 1;
11410
    qemuDomainObjPrivatePtr priv = NULL;
D
Daniel Veillard 已提交
11411

C
Chris Lalancette 已提交
11412 11413 11414 11415 11416 11417 11418 11419 11420
    virCheckFlags(VIR_MIGRATE_LIVE |
                  VIR_MIGRATE_PEER2PEER |
                  VIR_MIGRATE_TUNNELLED |
                  VIR_MIGRATE_PERSIST_DEST |
                  VIR_MIGRATE_UNDEFINE_SOURCE |
                  VIR_MIGRATE_PAUSED |
                  VIR_MIGRATE_NON_SHARED_DISK |
                  VIR_MIGRATE_NON_SHARED_INC, NULL);

11421 11422 11423
    /* Migration failed. Save the current error so nothing squashes it */
    orig_err = virSaveLastError();

11424
    qemuDriverLock(driver);
11425
    vm = virDomainFindByName(&driver->domains, dname);
D
Daniel Veillard 已提交
11426
    if (!vm) {
11427 11428
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching name '%s'"), dname);
11429
        goto cleanup;
D
Daniel Veillard 已提交
11430 11431
    }

11432 11433 11434 11435 11436 11437 11438 11439 11440
    priv = vm->privateData;
    if (priv->jobActive != QEMU_JOB_MIGRATION_IN) {
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("domain '%s' is not processing incoming migration"), dname);
        goto cleanup;
    }
    priv->jobActive = QEMU_JOB_NONE;
    memset(&priv->jobInfo, 0, sizeof(priv->jobInfo));

11441 11442 11443
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

D
Daniel Veillard 已提交
11444 11445 11446 11447
    /* Did the migration go as planned?  If yes, return the domain
     * object, but if no, clean up the empty qemu process.
     */
    if (retcode == 0) {
11448 11449 11450 11451 11452 11453
        if (!virDomainObjIsActive(vm)) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("guest unexpectedly quit"));
            goto cleanup;
        }

C
Chris Lalancette 已提交
11454 11455 11456 11457 11458
        if (flags & VIR_MIGRATE_PERSIST_DEST) {
            if (vm->persistent)
                newVM = 0;
            vm->persistent = 1;

11459
            if (virDomainSaveConfig(driver->configDir, vm->def) < 0) {
C
Chris Lalancette 已提交
11460 11461 11462 11463 11464 11465 11466 11467 11468 11469
                /* 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.
                 */
                vm = NULL;
11470
                goto endjob;
C
Chris Lalancette 已提交
11471 11472 11473 11474 11475 11476 11477 11478 11479
            }

            event = virDomainEventNewFromObj(vm,
                                             VIR_DOMAIN_EVENT_DEFINED,
                                             newVM ?
                                             VIR_DOMAIN_EVENT_DEFINED_ADDED :
                                             VIR_DOMAIN_EVENT_DEFINED_UPDATED);
            if (event)
                qemuDomainEventQueue(driver, event);
11480
            event = NULL;
C
Chris Lalancette 已提交
11481 11482

        }
D
Daniel Veillard 已提交
11483
        dom = virGetDomain (dconn, vm->def->name, vm->def->uuid);
11484

11485 11486 11487 11488 11489 11490 11491 11492
        if (!(flags & VIR_MIGRATE_PAUSED)) {
            /* 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
             */
            qemuDomainObjEnterMonitorWithDriver(driver, vm);
            if (qemuMonitorStartCPUs(priv->mon, dconn) < 0) {
                if (virGetLastError() == NULL)
11493 11494
                    qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                    "%s", _("resume operation failed"));
11495 11496 11497
                qemuDomainObjExitMonitorWithDriver(driver, vm);
                goto endjob;
            }
11498
            qemuDomainObjExitMonitorWithDriver(driver, vm);
11499 11500

            vm->state = VIR_DOMAIN_RUNNING;
11501 11502
        }

11503 11504 11505
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_MIGRATED);
11506 11507 11508 11509 11510 11511
        if (vm->state == VIR_DOMAIN_PAUSED) {
            qemuDomainEventQueue(driver, event);
            event = virDomainEventNewFromObj(vm,
                                             VIR_DOMAIN_EVENT_SUSPENDED,
                                             VIR_DOMAIN_EVENT_SUSPENDED_PAUSED);
        }
11512 11513 11514 11515
        if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0) {
            VIR_WARN("Failed to save status on vm %s", vm->def->name);
            goto endjob;
        }
D
Daniel Veillard 已提交
11516
    } else {
11517
        qemudShutdownVMDaemon(driver, vm, 1);
11518 11519 11520
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         VIR_DOMAIN_EVENT_STOPPED_FAILED);
11521
        if (!vm->persistent) {
11522 11523
            if (qemuDomainObjEndJob(vm) > 0)
                virDomainRemoveInactive(&driver->domains, vm);
11524 11525
            vm = NULL;
        }
D
Daniel Veillard 已提交
11526
    }
11527

11528
endjob:
11529 11530 11531
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
11532

11533
cleanup:
11534 11535 11536 11537
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
11538 11539
    if (vm)
        virDomainObjUnlock(vm);
11540 11541
    if (event)
        qemuDomainEventQueue(driver, event);
11542
    qemuDriverUnlock(driver);
11543
    return dom;
D
Daniel Veillard 已提交
11544 11545
}

11546 11547 11548 11549 11550 11551 11552 11553 11554 11555 11556 11557 11558 11559 11560 11561
static int
qemudNodeDeviceGetPciInfo (virNodeDevicePtr dev,
                           unsigned *domain,
                           unsigned *bus,
                           unsigned *slot,
                           unsigned *function)
{
    virNodeDeviceDefPtr def = NULL;
    virNodeDevCapsDefPtr cap;
    char *xml = NULL;
    int ret = -1;

    xml = virNodeDeviceGetXMLDesc(dev, 0);
    if (!xml)
        goto out;

11562
    def = virNodeDeviceDefParseString(xml, EXISTING_DEVICE);
11563 11564 11565 11566 11567 11568 11569 11570 11571 11572 11573 11574 11575 11576 11577 11578 11579
    if (!def)
        goto out;

    cap = def->caps;
    while (cap) {
        if (cap->type == VIR_NODE_DEV_CAP_PCI_DEV) {
            *domain   = cap->data.pci_dev.domain;
            *bus      = cap->data.pci_dev.bus;
            *slot     = cap->data.pci_dev.slot;
            *function = cap->data.pci_dev.function;
            break;
        }

        cap = cap->next;
    }

    if (!cap) {
11580 11581
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("device %s is not a PCI device"), dev->name);
11582 11583 11584 11585 11586 11587 11588 11589 11590 11591 11592 11593 11594
        goto out;
    }

    ret = 0;
out:
    virNodeDeviceDefFree(def);
    VIR_FREE(xml);
    return ret;
}

static int
qemudNodeDeviceDettach (virNodeDevicePtr dev)
{
11595
    struct qemud_driver *driver = dev->conn->privateData;
11596 11597 11598 11599 11600 11601 11602
    pciDevice *pci;
    unsigned domain, bus, slot, function;
    int ret = -1;

    if (qemudNodeDeviceGetPciInfo(dev, &domain, &bus, &slot, &function) < 0)
        return -1;

11603
    pci = pciGetDevice(domain, bus, slot, function);
11604 11605 11606
    if (!pci)
        return -1;

11607 11608
    qemuDriverLock(driver);
    if (pciDettachDevice(pci, driver->activePciHostdevs) < 0)
11609 11610 11611 11612
        goto out;

    ret = 0;
out:
11613
    qemuDriverUnlock(driver);
11614
    pciFreeDevice(pci);
11615 11616 11617 11618 11619 11620
    return ret;
}

static int
qemudNodeDeviceReAttach (virNodeDevicePtr dev)
{
11621
    struct qemud_driver *driver = dev->conn->privateData;
11622 11623 11624 11625 11626 11627 11628
    pciDevice *pci;
    unsigned domain, bus, slot, function;
    int ret = -1;

    if (qemudNodeDeviceGetPciInfo(dev, &domain, &bus, &slot, &function) < 0)
        return -1;

11629
    pci = pciGetDevice(domain, bus, slot, function);
11630 11631 11632
    if (!pci)
        return -1;

11633 11634
    qemuDriverLock(driver);
    if (pciReAttachDevice(pci, driver->activePciHostdevs) < 0)
11635 11636 11637 11638
        goto out;

    ret = 0;
out:
11639
    qemuDriverUnlock(driver);
11640
    pciFreeDevice(pci);
11641 11642 11643 11644 11645 11646
    return ret;
}

static int
qemudNodeDeviceReset (virNodeDevicePtr dev)
{
11647
    struct qemud_driver *driver = dev->conn->privateData;
11648 11649 11650 11651 11652 11653 11654
    pciDevice *pci;
    unsigned domain, bus, slot, function;
    int ret = -1;

    if (qemudNodeDeviceGetPciInfo(dev, &domain, &bus, &slot, &function) < 0)
        return -1;

11655
    pci = pciGetDevice(domain, bus, slot, function);
11656 11657 11658
    if (!pci)
        return -1;

11659 11660
    qemuDriverLock(driver);

11661
    if (pciResetDevice(pci, driver->activePciHostdevs, NULL) < 0)
11662 11663 11664 11665
        goto out;

    ret = 0;
out:
11666
    qemuDriverUnlock(driver);
11667
    pciFreeDevice(pci);
11668 11669 11670
    return ret;
}

11671 11672 11673 11674 11675 11676 11677 11678 11679 11680 11681
static int
qemuCPUCompare(virConnectPtr conn,
               const char *xmlDesc,
               unsigned int flags ATTRIBUTE_UNUSED)
{
    struct qemud_driver *driver = conn->privateData;
    int ret = VIR_CPU_COMPARE_ERROR;

    qemuDriverLock(driver);

    if (!driver->caps || !driver->caps->host.cpu) {
11682 11683
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        "%s", _("cannot get host CPU capabilities"));
11684 11685
    }
    else
11686
        ret = cpuCompareXML(driver->caps->host.cpu, xmlDesc);
11687 11688 11689 11690 11691 11692

    qemuDriverUnlock(driver);

    return ret;
}

11693

11694 11695 11696 11697 11698 11699 11700 11701 11702 11703 11704 11705 11706
static char *
qemuCPUBaseline(virConnectPtr conn ATTRIBUTE_UNUSED,
                const char **xmlCPUs,
                unsigned int ncpus,
                unsigned int flags ATTRIBUTE_UNUSED)
{
    char *cpu;

    cpu = cpuBaselineXML(xmlCPUs, ncpus, NULL, 0);

    return cpu;
}

11707 11708 11709 11710 11711 11712 11713 11714 11715 11716 11717 11718 11719 11720 11721 11722 11723 11724 11725 11726 11727 11728 11729

static int qemuDomainGetJobInfo(virDomainPtr dom,
                                virDomainJobInfoPtr info) {
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
    qemuDomainObjPrivatePtr priv;

    qemuDriverLock(driver);
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
    qemuDriverUnlock(driver);
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
        goto cleanup;
    }

    priv = vm->privateData;

    if (virDomainObjIsActive(vm)) {
        if (priv->jobActive) {
11730 11731
            struct timeval now;

11732
            memcpy(info, &priv->jobInfo, sizeof(*info));
11733 11734 11735 11736 11737 11738 11739 11740 11741 11742 11743 11744 11745 11746

            /* Refresh elapsed time again just to ensure it
             * is fully updated. This is primarily for benefit
             * of incoming migration which we don't currently
             * monitor actively in the background thread
             */
            if (gettimeofday(&now, NULL) < 0) {
                virReportSystemError(errno, "%s",
                                     _("cannot get time of day"));
                goto cleanup;
            }
            info->timeElapsed =
                ((now.tv_sec * 1000ull) + (now.tv_usec / 1000)) -
                priv->jobStart;
11747 11748 11749 11750 11751 11752 11753 11754 11755 11756 11757 11758 11759 11760 11761 11762 11763 11764 11765
        } else {
            memset(info, 0, sizeof(*info));
            info->type = VIR_DOMAIN_JOB_NONE;
        }
    } else {
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
        goto cleanup;
    }

    ret = 0;

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
    return ret;
}


11766 11767 11768 11769 11770 11771 11772 11773 11774 11775 11776 11777 11778 11779 11780 11781 11782 11783 11784 11785 11786 11787
static int qemuDomainAbortJob(virDomainPtr dom) {
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
    qemuDomainObjPrivatePtr priv;

    qemuDriverLock(driver);
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
    qemuDriverUnlock(driver);
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
        goto cleanup;
    }

    priv = vm->privateData;

    if (virDomainObjIsActive(vm)) {
        if (priv->jobActive) {
            VIR_DEBUG("Requesting cancellation of job on vm %s", vm->def->name);
11788
            priv->jobSignals |= QEMU_JOB_SIGNAL_CANCEL;
11789 11790 11791 11792 11793 11794 11795 11796 11797 11798 11799 11800 11801 11802 11803 11804 11805 11806 11807 11808
        } else {
            qemuReportError(VIR_ERR_OPERATION_INVALID,
                            "%s", _("no job is active on the domain"));
            goto cleanup;
        }
    } else {
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
        goto cleanup;
    }

    ret = 0;

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
    return ret;
}


11809 11810 11811 11812 11813 11814 11815 11816 11817 11818
static int
qemuDomainMigrateSetMaxDowntime(virDomainPtr dom,
                                unsigned long long downtime,
                                unsigned int flags)
{
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    qemuDomainObjPrivatePtr priv;
    int ret = -1;

11819
    virCheckFlags(0, -1);
11820 11821 11822 11823 11824 11825 11826 11827 11828 11829 11830 11831 11832 11833 11834 11835 11836 11837 11838 11839

    qemuDriverLock(driver);
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);

    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
        goto cleanup;
    }

    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
        goto cleanup;
    }

    priv = vm->privateData;

11840
    if (priv->jobActive != QEMU_JOB_MIGRATION_OUT) {
11841 11842 11843 11844 11845 11846 11847 11848 11849 11850 11851 11852 11853 11854 11855 11856 11857
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not being migrated"));
        goto cleanup;
    }

    VIR_DEBUG("Requesting migration downtime change to %llums", downtime);
    priv->jobSignals |= QEMU_JOB_SIGNAL_MIGRATE_DOWNTIME;
    priv->jobSignalsData.migrateDowntime = downtime;
    ret = 0;

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);
    return ret;
}

C
Chris Lalancette 已提交
11858 11859 11860 11861 11862 11863 11864 11865 11866 11867 11868 11869 11870 11871 11872 11873 11874 11875 11876 11877 11878 11879 11880 11881 11882 11883 11884 11885 11886 11887 11888 11889 11890 11891 11892 11893 11894 11895 11896 11897 11898 11899 11900 11901 11902 11903 11904 11905 11906 11907 11908 11909 11910 11911 11912 11913 11914 11915 11916 11917 11918 11919 11920 11921 11922 11923 11924 11925 11926 11927 11928 11929 11930 11931 11932 11933 11934 11935 11936 11937 11938 11939 11940 11941 11942 11943 11944 11945 11946 11947 11948 11949 11950 11951 11952 11953 11954 11955 11956 11957 11958 11959 11960 11961 11962 11963 11964 11965 11966
static char *qemuFindQemuImgBinary(void)
{
    char *ret;

    ret = virFindFileInPath("kvm-img");
    if (ret == NULL)
        ret = virFindFileInPath("qemu-img");
    if (ret == NULL)
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("unable to find kvm-img or qemu-img"));

    return ret;
}

static int qemuDomainSnapshotWriteSnapshotMetadata(virDomainObjPtr vm,
                                                   char *snapshotDir)
{
    int fd = -1;
    char *newxml = NULL;
    int ret = -1;
    char *snapDir = NULL;
    char *snapFile = NULL;
    int err;
    char uuidstr[VIR_UUID_STRING_BUFLEN];

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

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

    if (virAsprintf(&snapFile, "%s/%s.xml", snapDir,
                    vm->current_snapshot->def->name) < 0) {
        virReportOOMError();
        goto cleanup;
    }
    fd = open(snapFile, O_CREAT|O_TRUNC|O_WRONLY, S_IRUSR|S_IWUSR);
    if (fd < 0) {
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("failed to create snapshot file '%s'"), snapFile);
        goto cleanup;
    }
    if (safewrite(fd, newxml, strlen(newxml)) != strlen(newxml)) {
        virReportSystemError(errno, _("Failed to write snapshot data to %s"),
                             snapFile);
        goto cleanup;
    }

    ret = 0;

cleanup:
    VIR_FREE(snapFile);
    VIR_FREE(snapDir);
    VIR_FREE(newxml);
    if (fd != -1)
        close(fd);
    return ret;
}

static int qemuDomainSnapshotSetActive(virDomainObjPtr vm,
                                       char *snapshotDir)
{
    if (vm->current_snapshot) {
        vm->current_snapshot->def->active = 1;

        return qemuDomainSnapshotWriteSnapshotMetadata(vm, snapshotDir);
    }

    return 0;
}

static int qemuDomainSnapshotSetInactive(virDomainObjPtr vm,
                                         char *snapshotDir)
{
    if (vm->current_snapshot) {
        vm->current_snapshot->def->active = 0;

        return qemuDomainSnapshotWriteSnapshotMetadata(vm, snapshotDir);
    }

    return 0;
}


static int qemuDomainSnapshotIsAllowed(virDomainObjPtr vm)
{
    int i;

    /* FIXME: we need to figure out what else here might succeed; in
     * particular, if it's a raw device but on LVM, we could probably make
     * that succeed as well
     */
    for (i = 0; i < vm->def->ndisks; i++) {
        if (vm->def->disks[i]->device == VIR_DOMAIN_DISK_DEVICE_DISK &&
            (!vm->def->disks[i]->driverType ||
             STRNEQ(vm->def->disks[i]->driverType, "qcow2"))) {
            qemuReportError(VIR_ERR_OPERATION_INVALID,
11967 11968
                            _("Disk '%s' does not support snapshotting"),
                            vm->def->disks[i]->src);
C
Chris Lalancette 已提交
11969 11970 11971 11972 11973 11974 11975 11976 11977
            return 0;
        }
    }

    return 1;
}

static virDomainSnapshotPtr qemuDomainSnapshotCreateXML(virDomainPtr domain,
                                                        const char *xmlDesc,
11978
                                                        unsigned int flags)
C
Chris Lalancette 已提交
11979 11980 11981 11982 11983 11984 11985 11986 11987 11988
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm = NULL;
    virDomainSnapshotObjPtr snap = NULL;
    virDomainSnapshotPtr snapshot = NULL;
    char uuidstr[VIR_UUID_STRING_BUFLEN];
    virDomainSnapshotDefPtr def;
    const char *qemuimgarg[] = { NULL, "snapshot", "-c", NULL, NULL, NULL };
    int i;

11989 11990
    virCheckFlags(0, NULL);

C
Chris Lalancette 已提交
11991 11992 11993 11994 11995 11996 11997 11998 11999 12000 12001 12002 12003 12004 12005 12006 12007 12008 12009 12010 12011 12012 12013 12014 12015 12016 12017 12018 12019 12020 12021 12022 12023 12024 12025 12026 12027 12028 12029 12030 12031 12032 12033 12034 12035 12036 12037 12038 12039 12040 12041 12042 12043 12044 12045 12046 12047 12048 12049 12050 12051
    qemuDriverLock(driver);
    virUUIDFormat(domain->uuid, uuidstr);
    vm = virDomainFindByUUID(&driver->domains, domain->uuid);
    if (!vm) {
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
        goto cleanup;
    }

    /* in a perfect world, we would allow qemu to tell us this.  The problem
     * is that qemu only does this check device-by-device; so if you had a
     * domain that booted from a large qcow2 device, but had a secondary raw
     * device attached, you wouldn't find out that you can't snapshot your
     * guest until *after* it had spent the time to snapshot the boot device.
     * This is probably a bug in qemu, but we'll work around it here for now.
     */
    if (!qemuDomainSnapshotIsAllowed(vm))
        goto cleanup;

    if (!(def = virDomainSnapshotDefParseString(xmlDesc, 1)))
        goto cleanup;

    if (!(snap = virDomainSnapshotAssignDef(&vm->snapshots, def)))
        goto cleanup;

    /* actually do the snapshot */
    if (!virDomainObjIsActive(vm)) {
        qemuimgarg[0] = qemuFindQemuImgBinary();
        if (qemuimgarg[0] == NULL)
            /* qemuFindQemuImgBinary set the error */
            goto cleanup;

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

        for (i = 0; i < vm->def->ndisks; i++) {
            /* 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
             */
            if (vm->def->disks[i]->device == VIR_DOMAIN_DISK_DEVICE_DISK) {
                if (!vm->def->disks[i]->driverType ||
                    STRNEQ(vm->def->disks[i]->driverType, "qcow2")) {
                    qemuReportError(VIR_ERR_OPERATION_INVALID,
                                    _("Disk device '%s' does not support snapshotting"),
                                    vm->def->disks[i]->info.alias);
                    goto cleanup;
                }

                qemuimgarg[4] = vm->def->disks[i]->src;

                if (virRun(qemuimgarg, NULL) < 0) {
                    virReportSystemError(errno,
                                         _("Failed to run '%s' to create snapshot '%s' from disk '%s'"),
                                         qemuimgarg[0], snap->def->name,
                                         vm->def->disks[i]->src);
                    goto cleanup;
                }
            }
        }
    }
    else {
12052 12053 12054 12055 12056
        qemuDomainObjPrivatePtr priv;
        int ret;

        if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
            goto cleanup;
C
Chris Lalancette 已提交
12057 12058
        priv = vm->privateData;
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
12059
        ret = qemuMonitorCreateSnapshot(priv->mon, def->name);
C
Chris Lalancette 已提交
12060
        qemuDomainObjExitMonitorWithDriver(driver, vm);
12061
        if (qemuDomainObjEndJob(vm) == 0) {
12062
            vm = NULL;
12063 12064
            goto cleanup;
        }
12065 12066
        if (ret < 0)
            goto cleanup;
C
Chris Lalancette 已提交
12067 12068 12069 12070 12071 12072 12073 12074 12075 12076 12077 12078 12079 12080 12081 12082 12083 12084 12085 12086 12087 12088 12089 12090 12091 12092 12093 12094 12095 12096 12097 12098 12099 12100 12101 12102
    }

    snap->def->state = vm->state;

    /* FIXME: if we fail after this point, there's not a whole lot we can
     * do; we've successfully taken the snapshot, and we are now running
     * on it, so we have to go forward the best we can
     */

    if (vm->current_snapshot) {
        def->parent = strdup(vm->current_snapshot->def->name);
        if (def->parent == NULL) {
            virReportOOMError();
            goto cleanup;
        }
    }

    /* Now we set the new current_snapshot for the domain */
    vm->current_snapshot = snap;

    if (qemuDomainSnapshotWriteSnapshotMetadata(vm, driver->snapshotDir) < 0)
        /* qemuDomainSnapshotWriteSnapshotMetadata set the error */
        goto cleanup;

    snapshot = virGetDomainSnapshot(domain, snap->def->name);

cleanup:
    VIR_FREE(qemuimgarg[0]);
    if (vm)
        virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);
    return snapshot;
}

static int qemuDomainSnapshotListNames(virDomainPtr domain, char **names,
                                       int nameslen,
12103
                                       unsigned int flags)
C
Chris Lalancette 已提交
12104 12105 12106 12107 12108
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm = NULL;
    int n = -1;

12109 12110
    virCheckFlags(0, -1);

C
Chris Lalancette 已提交
12111 12112 12113 12114 12115 12116 12117 12118 12119 12120 12121 12122 12123 12124 12125 12126 12127 12128 12129 12130
    qemuDriverLock(driver);
    vm = virDomainFindByUUID(&driver->domains, domain->uuid);
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(domain->uuid, uuidstr);
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
        goto cleanup;
    }

    n = virDomainSnapshotObjListGetNames(&vm->snapshots, names, nameslen);

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);
    return n;
}

static int qemuDomainSnapshotNum(virDomainPtr domain,
12131
                                 unsigned int flags)
C
Chris Lalancette 已提交
12132 12133 12134 12135 12136
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm = NULL;
    int n = -1;

12137 12138
    virCheckFlags(0, -1);

C
Chris Lalancette 已提交
12139 12140 12141 12142 12143 12144 12145 12146 12147 12148 12149 12150 12151 12152 12153 12154 12155 12156 12157 12158 12159
    qemuDriverLock(driver);
    vm = virDomainFindByUUID(&driver->domains, domain->uuid);
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(domain->uuid, uuidstr);
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
        goto cleanup;
    }

    n = virDomainSnapshotObjListNum(&vm->snapshots);

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);
    return n;
}

static virDomainSnapshotPtr qemuDomainSnapshotLookupByName(virDomainPtr domain,
                                                           const char *name,
12160
                                                           unsigned int flags)
C
Chris Lalancette 已提交
12161 12162 12163 12164 12165 12166
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm;
    virDomainSnapshotObjPtr snap = NULL;
    virDomainSnapshotPtr snapshot = NULL;

12167 12168
    virCheckFlags(0, NULL);

C
Chris Lalancette 已提交
12169 12170 12171 12172 12173 12174 12175 12176 12177 12178 12179 12180 12181 12182 12183 12184 12185 12186 12187 12188 12189 12190 12191 12192 12193 12194 12195
    qemuDriverLock(driver);
    vm = virDomainFindByUUID(&driver->domains, domain->uuid);
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(domain->uuid, uuidstr);
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
        goto cleanup;
    }

    snap = virDomainSnapshotFindByName(&vm->snapshots, name);
    if (!snap) {
        qemuReportError(VIR_ERR_NO_DOMAIN_SNAPSHOT,
                        _("no snapshot with matching name '%s'"), name);
        goto cleanup;
    }

    snapshot = virGetDomainSnapshot(domain, snap->def->name);

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);
    return snapshot;
}

static int qemuDomainHasCurrentSnapshot(virDomainPtr domain,
12196
                                        unsigned int flags)
C
Chris Lalancette 已提交
12197 12198 12199 12200 12201
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;

12202 12203
    virCheckFlags(0, -1);

C
Chris Lalancette 已提交
12204 12205 12206 12207 12208 12209 12210 12211 12212 12213 12214 12215 12216 12217 12218 12219 12220 12221 12222 12223
    qemuDriverLock(driver);
    vm = virDomainFindByUUID(&driver->domains, domain->uuid);
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(domain->uuid, uuidstr);
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
        goto cleanup;
    }

    ret = (vm->current_snapshot != NULL);

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);
    return ret;
}

static virDomainSnapshotPtr qemuDomainSnapshotCurrent(virDomainPtr domain,
12224
                                                      unsigned int flags)
C
Chris Lalancette 已提交
12225 12226 12227 12228 12229
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm;
    virDomainSnapshotPtr snapshot = NULL;

12230 12231
    virCheckFlags(0, NULL);

C
Chris Lalancette 已提交
12232 12233 12234 12235 12236 12237 12238 12239 12240 12241 12242 12243 12244 12245 12246 12247 12248 12249 12250 12251 12252 12253 12254 12255 12256 12257
    qemuDriverLock(driver);
    vm = virDomainFindByUUID(&driver->domains, domain->uuid);
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(domain->uuid, uuidstr);
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
        goto cleanup;
    }

    if (!vm->current_snapshot) {
        qemuReportError(VIR_ERR_NO_DOMAIN_SNAPSHOT, "%s",
                        _("the domain does not have a current snapshot"));
        goto cleanup;
    }

    snapshot = virGetDomainSnapshot(domain, vm->current_snapshot->def->name);

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);
    return snapshot;
}

static char *qemuDomainSnapshotDumpXML(virDomainSnapshotPtr snapshot,
12258
                                       unsigned int flags)
C
Chris Lalancette 已提交
12259 12260 12261 12262 12263 12264 12265
{
    struct qemud_driver *driver = snapshot->domain->conn->privateData;
    virDomainObjPtr vm = NULL;
    char *xml = NULL;
    virDomainSnapshotObjPtr snap = NULL;
    char uuidstr[VIR_UUID_STRING_BUFLEN];

12266 12267
    virCheckFlags(0, NULL);

C
Chris Lalancette 已提交
12268 12269 12270 12271 12272 12273 12274 12275 12276 12277 12278 12279 12280 12281 12282 12283 12284 12285 12286 12287 12288 12289 12290 12291 12292 12293 12294
    qemuDriverLock(driver);
    virUUIDFormat(snapshot->domain->uuid, uuidstr);
    vm = virDomainFindByUUID(&driver->domains, snapshot->domain->uuid);
    if (!vm) {
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
        goto cleanup;
    }

    snap = virDomainSnapshotFindByName(&vm->snapshots, snapshot->name);
    if (!snap) {
        qemuReportError(VIR_ERR_NO_DOMAIN_SNAPSHOT,
                        _("no domain snapshot with matching name '%s'"),
                        snapshot->name);
        goto cleanup;
    }

    xml = virDomainSnapshotDefFormat(uuidstr, snap->def, 0);

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);
    return xml;
}

static int qemuDomainRevertToSnapshot(virDomainSnapshotPtr snapshot,
12295
                                      unsigned int flags)
C
Chris Lalancette 已提交
12296 12297 12298 12299 12300 12301 12302 12303 12304 12305
{
    struct qemud_driver *driver = snapshot->domain->conn->privateData;
    virDomainObjPtr vm = NULL;
    int ret = -1;
    virDomainSnapshotObjPtr snap = NULL;
    char uuidstr[VIR_UUID_STRING_BUFLEN];
    virDomainEventPtr event = NULL;
    qemuDomainObjPrivatePtr priv;
    int rc;

12306 12307
    virCheckFlags(0, -1);

C
Chris Lalancette 已提交
12308 12309 12310 12311 12312 12313 12314 12315 12316 12317 12318 12319 12320 12321 12322 12323 12324 12325 12326 12327 12328 12329 12330 12331 12332 12333 12334 12335 12336 12337 12338
    qemuDriverLock(driver);
    virUUIDFormat(snapshot->domain->uuid, uuidstr);
    vm = virDomainFindByUUID(&driver->domains, snapshot->domain->uuid);
    if (!vm) {
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
        goto cleanup;
    }

    snap = virDomainSnapshotFindByName(&vm->snapshots, snapshot->name);
    if (!snap) {
        qemuReportError(VIR_ERR_NO_DOMAIN_SNAPSHOT,
                        _("no domain snapshot with matching name '%s'"),
                        snapshot->name);
        goto cleanup;
    }

    vm->current_snapshot = snap;

    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

    if (snap->def->state == VIR_DOMAIN_RUNNING
        || snap->def->state == VIR_DOMAIN_PAUSED) {

        if (virDomainObjIsActive(vm)) {
            priv = vm->privateData;
            qemuDomainObjEnterMonitorWithDriver(driver, vm);
            rc = qemuMonitorLoadSnapshot(priv->mon, snap->def->name);
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            if (rc < 0)
12339
                goto endjob;
C
Chris Lalancette 已提交
12340 12341 12342
        }
        else {
            if (qemuDomainSnapshotSetActive(vm, driver->snapshotDir) < 0)
12343
                goto endjob;
C
Chris Lalancette 已提交
12344 12345

            rc = qemudStartVMDaemon(snapshot->domain->conn, driver, vm, NULL,
12346
                                    false, -1, NULL);
C
Chris Lalancette 已提交
12347
            if (qemuDomainSnapshotSetInactive(vm, driver->snapshotDir) < 0)
12348
                goto endjob;
C
Chris Lalancette 已提交
12349
            if (rc < 0)
12350
                goto endjob;
C
Chris Lalancette 已提交
12351 12352 12353 12354 12355 12356
        }

        if (snap->def->state == VIR_DOMAIN_PAUSED) {
            /* qemu unconditionally starts the domain running again after
             * loadvm, so let's pause it to keep consistency
             */
12357
            int state = vm->state;
C
Chris Lalancette 已提交
12358
            priv = vm->privateData;
12359
            vm->state = VIR_DOMAIN_PAUSED;
C
Chris Lalancette 已提交
12360 12361 12362
            qemuDomainObjEnterMonitorWithDriver(driver, vm);
            rc = qemuMonitorStopCPUs(priv->mon);
            qemuDomainObjExitMonitorWithDriver(driver, vm);
12363 12364
            if (rc < 0) {
                vm->state = state;
12365
                goto endjob;
12366
            }
C
Chris Lalancette 已提交
12367 12368 12369 12370 12371 12372 12373 12374 12375 12376 12377 12378 12379 12380 12381 12382 12383 12384
        }

        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STARTED,
                                         VIR_DOMAIN_EVENT_STARTED_FROM_SNAPSHOT);
    }
    else {
        /* qemu is a little funny with running guests and the restoration
         * of snapshots.  If the snapshot was taken online,
         * then after a "loadvm" monitor command, the VM is set running
         * again.  If the snapshot was taken offline, then after a "loadvm"
         * monitor command the VM is left paused.  Unpausing it leads to
         * the memory state *before* the loadvm with the disk *after* the
         * loadvm, which obviously is bound to corrupt something.
         * Therefore we destroy the domain and set it to "off" in this case.
         */

        if (virDomainObjIsActive(vm)) {
12385
            qemudShutdownVMDaemon(driver, vm, 0);
C
Chris Lalancette 已提交
12386 12387 12388
            event = virDomainEventNewFromObj(vm,
                                             VIR_DOMAIN_EVENT_STOPPED,
                                             VIR_DOMAIN_EVENT_STOPPED_FROM_SNAPSHOT);
12389 12390 12391 12392
            if (!vm->persistent) {
                if (qemuDomainObjEndJob(vm) > 0)
                    virDomainRemoveInactive(&driver->domains, vm);
                vm = NULL;
12393
                goto cleanup;
12394
            }
C
Chris Lalancette 已提交
12395 12396 12397
        }

        if (qemuDomainSnapshotSetActive(vm, driver->snapshotDir) < 0)
12398
            goto endjob;
C
Chris Lalancette 已提交
12399 12400 12401 12402 12403 12404
    }

    vm->state = snap->def->state;

    ret = 0;

12405
endjob:
C
Chris Lalancette 已提交
12406 12407 12408
    if (vm && qemuDomainObjEndJob(vm) == 0)
        vm = NULL;

12409
cleanup:
C
Chris Lalancette 已提交
12410 12411 12412 12413 12414 12415 12416 12417 12418 12419 12420 12421 12422 12423 12424 12425 12426 12427 12428 12429 12430 12431 12432 12433 12434 12435 12436 12437 12438 12439 12440 12441 12442 12443 12444 12445 12446 12447 12448 12449 12450 12451 12452 12453 12454 12455 12456 12457 12458 12459 12460 12461 12462 12463 12464 12465 12466 12467 12468 12469 12470 12471 12472 12473 12474 12475 12476 12477 12478 12479 12480 12481 12482 12483 12484 12485 12486 12487 12488 12489 12490 12491 12492 12493 12494 12495 12496 12497 12498 12499 12500 12501 12502 12503 12504 12505 12506 12507 12508 12509 12510 12511 12512 12513 12514 12515 12516 12517 12518 12519 12520 12521 12522 12523 12524 12525 12526 12527 12528 12529 12530 12531 12532 12533 12534 12535 12536 12537 12538 12539 12540 12541 12542 12543
    if (event)
        qemuDomainEventQueue(driver, event);
    if (vm)
        virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);

    return ret;
}

static int qemuDomainSnapshotDiscard(struct qemud_driver *driver,
                                     virDomainObjPtr vm,
                                     virDomainSnapshotObjPtr snap)
{
    const char *qemuimgarg[] = { NULL, "snapshot", "-d", NULL, NULL, NULL };
    char *snapFile = NULL;
    int ret = -1;
    int i;
    qemuDomainObjPrivatePtr priv;
    virDomainSnapshotObjPtr parentsnap;

    if (!virDomainObjIsActive(vm)) {
        qemuimgarg[0] = qemuFindQemuImgBinary();
        if (qemuimgarg[0] == NULL)
            /* qemuFindQemuImgBinary set the error */
            goto cleanup;

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

        for (i = 0; i < vm->def->ndisks; i++) {
            /* FIXME: we also need to handle LVM here */
            if (vm->def->disks[i]->device == VIR_DOMAIN_DISK_DEVICE_DISK) {
                if (!vm->def->disks[i]->driverType ||
                    STRNEQ(vm->def->disks[i]->driverType, "qcow2")) {
                    /* we continue on even in the face of error, since other
                     * disks in this VM may have this snapshot in place
                     */
                    continue;
                }

                qemuimgarg[4] = vm->def->disks[i]->src;

                if (virRun(qemuimgarg, NULL) < 0) {
                    /* we continue on even in the face of error, since other
                     * disks in this VM may have this snapshot in place
                     */
                    continue;
                }
            }
        }
    }
    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 (snap == vm->current_snapshot) {
        if (snap->def->parent) {
            parentsnap = virDomainSnapshotFindByName(&vm->snapshots,
                                                     snap->def->parent);
            if (!parentsnap) {
                qemuReportError(VIR_ERR_NO_DOMAIN_SNAPSHOT,
                                _("no domain snapshot parent with matching name '%s'"),
                                snap->def->parent);
                goto cleanup;
            }

            /* Now we set the new current_snapshot for the domain */
            vm->current_snapshot = parentsnap;
        }
        else
            vm->current_snapshot = NULL;
    }

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

    virDomainSnapshotObjListRemove(&vm->snapshots, snap);

    ret = 0;

cleanup:
    VIR_FREE(snapFile);
    VIR_FREE(qemuimgarg[0]);

    return ret;
}

struct snap_remove {
    struct qemud_driver *driver;
    virDomainObjPtr vm;
    char *parent;
    int err;
};

static void qemuDomainSnapshotDiscardChildren(void *payload,
                                              const char *name ATTRIBUTE_UNUSED,
                                              void *data)
{
    virDomainSnapshotObjPtr snap = payload;
    struct snap_remove *curr = data;
    struct snap_remove this;

    if (snap->def->parent && STREQ(snap->def->parent, curr->parent)) {
        this.driver = curr->driver;
        this.vm = curr->vm;
        this.parent = snap->def->name;
        this.err = 0;
        virHashForEach(curr->vm->snapshots.objs,
                       qemuDomainSnapshotDiscardChildren, &this);

        if (this.err)
            curr->err = this.err;
        else
            this.err = qemuDomainSnapshotDiscard(curr->driver, curr->vm, snap);
    }
}

static int qemuDomainSnapshotDelete(virDomainSnapshotPtr snapshot,
                                    unsigned int flags)
{
    struct qemud_driver *driver = snapshot->domain->conn->privateData;
    virDomainObjPtr vm = NULL;
    int ret = -1;
    virDomainSnapshotObjPtr snap = NULL;
    char uuidstr[VIR_UUID_STRING_BUFLEN];
    struct snap_remove rem;

12544 12545
    virCheckFlags(VIR_DOMAIN_SNAPSHOT_DELETE_CHILDREN, -1);

C
Chris Lalancette 已提交
12546 12547 12548 12549 12550 12551 12552 12553 12554 12555 12556 12557 12558 12559 12560 12561 12562
    qemuDriverLock(driver);
    virUUIDFormat(snapshot->domain->uuid, uuidstr);
    vm = virDomainFindByUUID(&driver->domains, snapshot->domain->uuid);
    if (!vm) {
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
        goto cleanup;
    }

    snap = virDomainSnapshotFindByName(&vm->snapshots, snapshot->name);
    if (!snap) {
        qemuReportError(VIR_ERR_NO_DOMAIN_SNAPSHOT,
                        _("no domain snapshot with matching name '%s'"),
                        snapshot->name);
        goto cleanup;
    }

12563 12564 12565
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

C
Chris Lalancette 已提交
12566 12567 12568 12569 12570 12571 12572 12573
    if (flags & VIR_DOMAIN_SNAPSHOT_DELETE_CHILDREN) {
        rem.driver = driver;
        rem.vm = vm;
        rem.parent = snap->def->name;
        rem.err = 0;
        virHashForEach(vm->snapshots.objs, qemuDomainSnapshotDiscardChildren,
                       &rem);
        if (rem.err < 0)
12574
            goto endjob;
C
Chris Lalancette 已提交
12575 12576 12577 12578
    }

    ret = qemuDomainSnapshotDiscard(driver, vm, snap);

12579 12580 12581 12582
endjob:
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;

C
Chris Lalancette 已提交
12583 12584 12585 12586 12587 12588
cleanup:
    if (vm)
        virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);
    return ret;
}
12589

12590 12591 12592 12593 12594 12595 12596 12597 12598 12599 12600 12601 12602 12603 12604 12605 12606 12607 12608 12609 12610 12611 12612 12613 12614 12615 12616 12617 12618 12619 12620 12621 12622 12623 12624 12625 12626 12627 12628 12629 12630 12631 12632 12633 12634 12635 12636 12637 12638 12639 12640
static int qemuDomainMonitorCommand(virDomainPtr domain, const char *cmd,
                                    char **result, unsigned int flags)
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm = NULL;
    int ret = -1;
    qemuDomainObjPrivatePtr priv;

    virCheckFlags(0, -1);

    qemuDriverLock(driver);
    vm = virDomainFindByUUID(&driver->domains, domain->uuid);
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(domain->uuid, uuidstr);
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
        goto cleanup;
    }

    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
        goto cleanup;
   }

    priv = vm->privateData;

    if (!priv->monitor_warned) {
        VIR_INFO("Qemu monitor command '%s' executed; libvirt results may be unpredictable!",
                 cmd);
        priv->monitor_warned = 1;
    }

    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
    ret = qemuMonitorArbitraryCommand(priv->mon, cmd, result);
    qemuDomainObjExitMonitorWithDriver(driver, vm);
    if (qemuDomainObjEndJob(vm) == 0) {
        vm = NULL;
        goto cleanup;
    }

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);
    return ret;
}

12641 12642 12643 12644 12645
static virDriver qemuDriver = {
    VIR_DRV_QEMU,
    "QEMU",
    qemudOpen, /* open */
    qemudClose, /* close */
D
Daniel Veillard 已提交
12646
    qemudSupportsFeature, /* supports_feature */
12647 12648
    qemudGetType, /* type */
    qemudGetVersion, /* version */
12649
    NULL, /* libvirtVersion (impl. in libvirt.c) */
12650
    virGetHostname, /* getHostname */
12651
    qemudGetMaxVCPUs, /* getMaxVcpus */
12652
    nodeGetInfo, /* nodeGetInfo */
12653 12654 12655
    qemudGetCapabilities, /* getCapabilities */
    qemudListDomains, /* listDomains */
    qemudNumDomains, /* numOfDomains */
12656
    qemudDomainCreate, /* domainCreateXML */
12657 12658 12659 12660 12661
    qemudDomainLookupByID, /* domainLookupByID */
    qemudDomainLookupByUUID, /* domainLookupByUUID */
    qemudDomainLookupByName, /* domainLookupByName */
    qemudDomainSuspend, /* domainSuspend */
    qemudDomainResume, /* domainResume */
12662
    qemudDomainShutdown, /* domainShutdown */
12663 12664 12665
    NULL, /* domainReboot */
    qemudDomainDestroy, /* domainDestroy */
    qemudDomainGetOSType, /* domainGetOSType */
12666
    qemudDomainGetMaxMemory, /* domainGetMaxMemory */
12667
    NULL, /* domainSetMaxMemory */
12668
    qemudDomainSetMemory, /* domainSetMemory */
12669 12670 12671
    qemudDomainGetInfo, /* domainGetInfo */
    qemudDomainSave, /* domainSave */
    qemudDomainRestore, /* domainRestore */
P
Paolo Bonzini 已提交
12672
    qemudDomainCoreDump, /* domainCoreDump */
12673
    qemudDomainSetVcpus, /* domainSetVcpus */
12674 12675
    qemudDomainPinVcpu, /* domainPinVcpu */
    qemudDomainGetVcpus, /* domainGetVcpus */
12676
    qemudDomainGetMaxVcpus, /* domainGetMaxVcpus */
12677 12678
    qemudDomainGetSecurityLabel, /* domainGetSecurityLabel */
    qemudNodeGetSecurityModel, /* nodeGetSecurityModel */
12679
    qemudDomainDumpXML, /* domainDumpXML */
12680
    qemuDomainXMLFromNative, /* domainXmlFromNative */
12681
    qemuDomainXMLToNative, /* domainXMLToNative */
12682 12683
    qemudListDefinedDomains, /* listDefinedDomains */
    qemudNumDefinedDomains, /* numOfDefinedDomains */
12684
    qemudDomainStart, /* domainCreate */
12685
    qemudDomainStartWithFlags, /* domainCreateWithFlags */
12686 12687
    qemudDomainDefine, /* domainDefineXML */
    qemudDomainUndefine, /* domainUndefine */
12688
    qemudDomainAttachDevice, /* domainAttachDevice */
12689
    qemudDomainAttachDeviceFlags, /* domainAttachDeviceFlags */
12690
    qemudDomainDetachDevice, /* domainDetachDevice */
12691
    qemudDomainDetachDeviceFlags, /* domainDetachDeviceFlags */
12692
    qemuDomainUpdateDeviceFlags, /* domainUpdateDeviceFlags */
12693 12694
    qemudDomainGetAutostart, /* domainGetAutostart */
    qemudDomainSetAutostart, /* domainSetAutostart */
12695 12696 12697
    qemuGetSchedulerType, /* domainGetSchedulerType */
    qemuGetSchedulerParameters, /* domainGetSchedulerParameters */
    qemuSetSchedulerParameters, /* domainSetSchedulerParameters */
D
Daniel Veillard 已提交
12698 12699
    NULL, /* domainMigratePrepare (v1) */
    qemudDomainMigratePerform, /* domainMigratePerform */
12700
    NULL, /* domainMigrateFinish */
12701
    qemudDomainBlockStats, /* domainBlockStats */
12702
    qemudDomainInterfaceStats, /* domainInterfaceStats */
12703
    qemudDomainMemoryStats, /* domainMemoryStats */
12704
    qemudDomainBlockPeek, /* domainBlockPeek */
R
Richard W.M. Jones 已提交
12705
    qemudDomainMemoryPeek, /* domainMemoryPeek */
12706
    qemuDomainGetBlockInfo, /* domainGetBlockInfo */
12707 12708
    nodeGetCellsFreeMemory, /* nodeGetCellsFreeMemory */
    nodeGetFreeMemory,  /* getFreeMemory */
12709 12710
    qemuDomainEventRegister, /* domainEventRegister */
    qemuDomainEventDeregister, /* domainEventDeregister */
D
Daniel Veillard 已提交
12711 12712
    qemudDomainMigratePrepare2, /* domainMigratePrepare2 */
    qemudDomainMigrateFinish2, /* domainMigrateFinish2 */
12713 12714 12715
    qemudNodeDeviceDettach, /* nodeDeviceDettach */
    qemudNodeDeviceReAttach, /* nodeDeviceReAttach */
    qemudNodeDeviceReset, /* nodeDeviceReset */
C
Chris Lalancette 已提交
12716
    qemudDomainMigratePrepareTunnel, /* domainMigratePrepareTunnel */
12717 12718 12719 12720
    qemuIsEncrypted, /* isEncrypted */
    qemuIsSecure, /* isSecure */
    qemuDomainIsActive, /* domainIsActive */
    qemuDomainIsPersistent, /* domainIsPersistent */
12721
    qemuCPUCompare, /* cpuCompare */
12722
    qemuCPUBaseline, /* cpuBaseline */
12723
    qemuDomainGetJobInfo, /* domainGetJobInfo */
12724
    qemuDomainAbortJob, /* domainAbortJob */
12725
    qemuDomainMigrateSetMaxDowntime, /* domainMigrateSetMaxDowntime */
12726 12727
    qemuDomainEventRegisterAny, /* domainEventRegisterAny */
    qemuDomainEventDeregisterAny, /* domainEventDeregisterAny */
12728 12729 12730
    qemuDomainManagedSave, /* domainManagedSave */
    qemuDomainHasManagedSaveImage, /* domainHasManagedSaveImage */
    qemuDomainManagedSaveRemove, /* domainManagedSaveRemove */
C
Chris Lalancette 已提交
12731 12732 12733 12734 12735 12736 12737 12738 12739
    qemuDomainSnapshotCreateXML, /* domainSnapshotCreateXML */
    qemuDomainSnapshotDumpXML, /* domainSnapshotDumpXML */
    qemuDomainSnapshotNum, /* domainSnapshotNum */
    qemuDomainSnapshotListNames, /* domainSnapshotListNames */
    qemuDomainSnapshotLookupByName, /* domainSnapshotLookupByName */
    qemuDomainHasCurrentSnapshot, /* domainHasCurrentSnapshot */
    qemuDomainSnapshotCurrent, /* domainSnapshotCurrent */
    qemuDomainRevertToSnapshot, /* domainRevertToSnapshot */
    qemuDomainSnapshotDelete, /* domainSnapshotDelete */
12740
    qemuDomainMonitorCommand, /* qemuDomainMonitorCommand */
12741 12742 12743
};


12744
static virStateDriver qemuStateDriver = {
12745
    .name = "QEMU",
12746 12747 12748 12749
    .initialize = qemudStartup,
    .cleanup = qemudShutdown,
    .reload = qemudReload,
    .active = qemudActive,
12750
};
12751

S
Stefan Berger 已提交
12752
static int
12753
qemudVMFilterRebuild(virConnectPtr conn ATTRIBUTE_UNUSED,
S
Stefan Berger 已提交
12754 12755
                     virHashIterator iter, void *data)
{
12756 12757 12758
    struct qemud_driver *driver = qemu_driver;

    qemuDriverLock(driver);
S
Stefan Berger 已提交
12759
    virHashForEach(qemu_driver->domains.objs, iter, data);
12760 12761
    qemuDriverUnlock(driver);

S
Stefan Berger 已提交
12762 12763 12764
    return 0;
}

12765 12766 12767 12768 12769 12770 12771 12772 12773 12774 12775 12776 12777 12778 12779 12780 12781 12782 12783 12784 12785 12786
static int
qemudVMFiltersInstantiate(virConnectPtr conn,
                          virDomainDefPtr def)
{
    int err = 0;
    int i;

    if (!conn)
        return 1;

    for (i = 0 ; i < def->nnets ; i++) {
        virDomainNetDefPtr net = def->nets[i];
        if ((net->filter) && (net->ifname)) {
           if (virDomainConfNWFilterInstantiate(conn, net)) {
                err = 1;
                break;
            }
        }
    }

    return err;
}
S
Stefan Berger 已提交
12787 12788 12789 12790 12791 12792

static virNWFilterCallbackDriver qemuCallbackDriver = {
    .name = "QEMU",
    .vmFilterRebuild = qemudVMFilterRebuild,
};

12793
int qemuRegister(void) {
12794 12795
    virRegisterDriver(&qemuDriver);
    virRegisterStateDriver(&qemuStateDriver);
S
Stefan Berger 已提交
12796
    virNWFilterRegisterCallbackDriver(&qemuCallbackDriver);
12797 12798
    return 0;
}