qemu_driver.c 416.0 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
#include <pwd.h>
44
#include <grp.h>
45
#include <stdio.h>
46
#include <sys/wait.h>
47
#include <sys/ioctl.h>
48
#include <sys/un.h>
D
Daniel P. Berrange 已提交
49

50 51
#include <libxml/xpathInternals.h>

52
#include "virterror_internal.h"
53
#include "logging.h"
54
#include "datatypes.h"
55 56
#include "qemu_driver.h"
#include "qemu_conf.h"
57
#include "qemu_monitor.h"
58
#include "qemu_bridge_filter.h"
J
Jim Meyering 已提交
59
#include "c-ctype.h"
60
#include "event.h"
61
#include "buf.h"
62
#include "util.h"
63
#include "nodeinfo.h"
64
#include "stats_linux.h"
65
#include "capabilities.h"
66
#include "memory.h"
67
#include "uuid.h"
68
#include "domain_conf.h"
69 70
#include "node_device_conf.h"
#include "pci.h"
71
#include "hostusb.h"
72
#include "processinfo.h"
73 74
#include "qemu_security_stacked.h"
#include "qemu_security_dac.h"
75
#include "cgroup.h"
C
Chris Lalancette 已提交
76
#include "libvirt_internal.h"
77
#include "xml.h"
78
#include "cpu/cpu.h"
79
#include "macvtap.h"
80
#include "sysinfo.h"
81
#include "domain_nwfilter.h"
82
#include "hooks.h"
83
#include "storage_file.h"
84
#include "virtaudit.h"
85

86

87 88
#define VIR_FROM_THIS VIR_FROM_QEMU

89 90 91
#define QEMU_VNC_PORT_MIN  5900
#define QEMU_VNC_PORT_MAX  65535

92 93
#define QEMU_NB_MEM_PARAM  3

94 95
#define QEMU_NAMESPACE_HREF "http://libvirt.org/schemas/domain/qemu/1.0"

96 97 98 99 100 101
/* 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,
102 103
    QEMU_JOB_MIGRATION_OUT,
    QEMU_JOB_MIGRATION_IN,
104 105 106 107 108
};

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 */
109 110 111 112 113
    QEMU_JOB_SIGNAL_MIGRATE_DOWNTIME = 1 << 2, /* Request migration downtime change */
};

struct qemuDomainJobSignalsData {
    unsigned long long migrateDowntime; /* Data for QEMU_JOB_SIGNAL_MIGRATE_DOWNTIME */
114 115
};

116 117 118
typedef struct _qemuDomainObjPrivate qemuDomainObjPrivate;
typedef qemuDomainObjPrivate *qemuDomainObjPrivatePtr;
struct _qemuDomainObjPrivate {
119
    virCond jobCond; /* Use in conjunction with main virDomainObjPtr lock */
120 121
    enum qemuDomainJob jobActive;   /* Currently running job */
    unsigned int jobSignals;        /* Signals for running job */
122
    struct qemuDomainJobSignalsData jobSignalsData; /* Signal specific data */
123 124
    virDomainJobInfo jobInfo;
    unsigned long long jobStart;
125

126
    qemuMonitorPtr mon;
127
    virDomainChrDefPtr monConfig;
D
Daniel P. Berrange 已提交
128
    int monJSON;
129
    int monitor_warned;
130 131 132

    int nvcpupids;
    int *vcpupids;
133 134

    qemuDomainPCIAddressSetPtr pciaddrs;
135
    int persistentAddrs;
136 137
};

138 139
static int qemudShutdown(void);

140 141
static void qemuDriverLock(struct qemud_driver *driver)
{
142
    virMutexLock(&driver->lock);
143 144 145
}
static void qemuDriverUnlock(struct qemud_driver *driver)
{
146
    virMutexUnlock(&driver->lock);
147 148
}

149 150 151
static void qemuDomainEventFlush(int timer, void *opaque);
static void qemuDomainEventQueue(struct qemud_driver *driver,
                                 virDomainEventPtr event);
152

J
Jiri Denemark 已提交
153 154
static int qemudDomainObjStart(virConnectPtr conn,
                               struct qemud_driver *driver,
155 156
                               virDomainObjPtr vm,
                               bool start_paused);
J
Jiri Denemark 已提交
157

158 159
static int qemudStartVMDaemon(virConnectPtr conn,
                              struct qemud_driver *driver,
160
                              virDomainObjPtr vm,
161
                              const char *migrateFrom,
162
                              bool start_paused,
163 164
                              int stdin_fd,
                              const char *stdin_path);
165

166
static void qemudShutdownVMDaemon(struct qemud_driver *driver,
167 168
                                  virDomainObjPtr vm,
                                  int migrated);
169

170 171 172
static void qemuDomainStartAudit(virDomainObjPtr vm, const char *reason, bool success);
static void qemuDomainStopAudit(virDomainObjPtr vm, const char *reason);

173
static int qemudDomainGetMaxVcpus(virDomainPtr dom);
174

175
static int qemuDetectVcpuPIDs(struct qemud_driver *driver,
176
                              virDomainObjPtr vm);
177

178 179 180
static int qemuUpdateActivePciHostdevs(struct qemud_driver *driver,
                                       virDomainDefPtr def);

181 182 183
static int qemudVMFiltersInstantiate(virConnectPtr conn,
                                     virDomainDefPtr def);

J
Jim Meyering 已提交
184
static struct qemud_driver *qemu_driver = NULL;
185

186 187 188 189 190 191 192 193 194 195 196 197 198 199 200

static void *qemuDomainObjPrivateAlloc(void)
{
    qemuDomainObjPrivatePtr priv;

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

    return priv;
}

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

201
    qemuDomainPCIAddressSetFree(priv->pciaddrs);
202 203 204
    virDomainChrDefFree(priv->monConfig);
    VIR_FREE(priv->vcpupids);

205 206
    /* This should never be non-NULL if we get here, but just in case... */
    if (priv->mon) {
207
        VIR_ERROR0(_("Unexpected QEMU monitor still active during domain deletion"));
208 209 210 211 212 213
        qemuMonitorClose(priv->mon);
    }
    VIR_FREE(priv);
}


214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231
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
232 233
        if (priv->monJSON)
            virBufferAddLit(buf, " json='1'");
234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259
        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) {
260
        virReportOOMError();
261 262 263
        goto error;
    }

264
    if (!(priv->monConfig->info.alias = strdup("monitor"))) {
265
        virReportOOMError();
266 267 268
        goto error;
    }

269
    if (!(monitorpath =
270
          virXPathString("string(./monitor[1]/@path)", ctxt))) {
271 272
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("no monitor path"));
273 274 275
        goto error;
    }

276
    tmp = virXPathString("string(./monitor[1]/@type)", ctxt);
277 278 279 280 281 282
    if (tmp)
        priv->monConfig->type = virDomainChrTypeFromString(tmp);
    else
        priv->monConfig->type = VIR_DOMAIN_CHR_TYPE_PTY;
    VIR_FREE(tmp);

283
    if (virXPathBoolean("count(./monitor[@json = '1']) > 0", ctxt)) {
D
Daniel P. Berrange 已提交
284
        priv->monJSON = 1;
285 286 287
    } else {
        priv->monJSON = 0;
    }
D
Daniel P. Berrange 已提交
288

289 290 291 292 293 294 295 296 297
    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);
298 299 300
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("unsupported monitor type '%s'"),
                        virDomainChrTypeToString(priv->monConfig->type));
301 302 303
        goto error;
    }

304
    n = virXPathNodeSet("./vcpus/vcpu", ctxt, &nodes);
305 306 307 308 309
    if (n < 0)
        goto error;
    if (n) {
        priv->nvcpupids = n;
        if (VIR_REALLOC_N(priv->vcpupids, priv->nvcpupids) < 0) {
310
            virReportOOMError();
311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330
            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:
331 332
    virDomainChrDefFree(priv->monConfig);
    priv->monConfig = NULL;
333 334 335 336 337 338
    VIR_FREE(nodes);
    return -1;
}



339 340 341 342 343 344 345 346 347
/*
 * 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
 */
348 349 350 351

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

352 353 354 355
static int qemuDomainObjBeginJob(virDomainObjPtr obj) ATTRIBUTE_RETURN_CHECK;
static int qemuDomainObjBeginJob(virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
356 357 358 359
    struct timeval now;
    unsigned long long then;

    if (gettimeofday(&now, NULL) < 0) {
360
        virReportSystemError(errno, "%s",
361 362 363 364 365
                             _("cannot get time of day"));
        return -1;
    }
    then = (now.tv_sec * 1000ull) + (now.tv_usec / 1000);
    then += QEMU_JOB_WAIT_TIME;
366 367 368 369

    virDomainObjRef(obj);

    while (priv->jobActive) {
370
        if (virCondWaitUntil(&priv->jobCond, &obj->lock, then) < 0) {
371
            virDomainObjUnref(obj);
372
            if (errno == ETIMEDOUT)
373 374
                qemuReportError(VIR_ERR_OPERATION_TIMEOUT,
                                "%s", _("cannot acquire state change lock"));
375
            else
376
                virReportSystemError(errno,
377
                                     "%s", _("cannot acquire job mutex"));
378 379 380
            return -1;
        }
    }
381 382
    priv->jobActive = QEMU_JOB_UNSPECIFIED;
    priv->jobSignals = 0;
383
    memset(&priv->jobSignalsData, 0, sizeof(priv->jobSignalsData));
384 385
    priv->jobStart = (now.tv_sec * 1000ull) + (now.tv_usec / 1000);
    memset(&priv->jobInfo, 0, sizeof(priv->jobInfo));
386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401

    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;
402 403 404 405
    struct timeval now;
    unsigned long long then;

    if (gettimeofday(&now, NULL) < 0) {
406
        virReportSystemError(errno, "%s",
407 408 409 410 411
                             _("cannot get time of day"));
        return -1;
    }
    then = (now.tv_sec * 1000ull) + (now.tv_usec / 1000);
    then += QEMU_JOB_WAIT_TIME;
412 413 414 415 416

    virDomainObjRef(obj);
    qemuDriverUnlock(driver);

    while (priv->jobActive) {
417
        if (virCondWaitUntil(&priv->jobCond, &obj->lock, then) < 0) {
418
            virDomainObjUnref(obj);
419
            if (errno == ETIMEDOUT)
420 421
                qemuReportError(VIR_ERR_OPERATION_TIMEOUT,
                                "%s", _("cannot acquire state change lock"));
422
            else
423
                virReportSystemError(errno,
424
                                     "%s", _("cannot acquire job mutex"));
M
Matthias Bolte 已提交
425
            qemuDriverLock(driver);
426 427 428
            return -1;
        }
    }
429 430
    priv->jobActive = QEMU_JOB_UNSPECIFIED;
    priv->jobSignals = 0;
431
    memset(&priv->jobSignalsData, 0, sizeof(priv->jobSignalsData));
432 433
    priv->jobStart = (now.tv_sec * 1000ull) + (now.tv_usec / 1000);
    memset(&priv->jobInfo, 0, sizeof(priv->jobInfo));
434 435 436 437 438 439 440 441 442 443 444 445

    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
446
 * earlier qemuDomainBeginJob() call
447 448 449
 *
 * Returns remaining refcount on 'obj', maybe 0 to indicated it
 * was deleted
450
 */
451
static int ATTRIBUTE_RETURN_CHECK qemuDomainObjEndJob(virDomainObjPtr obj)
452 453 454
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

455 456
    priv->jobActive = QEMU_JOB_NONE;
    priv->jobSignals = 0;
457
    memset(&priv->jobSignalsData, 0, sizeof(priv->jobSignalsData));
458 459
    priv->jobStart = 0;
    memset(&priv->jobInfo, 0, sizeof(priv->jobInfo));
460 461
    virCondSignal(&priv->jobCond);

462
    return virDomainObjUnref(obj);
463 464 465 466 467 468 469
}


/*
 * obj must be locked before calling, qemud_driver must be unlocked
 *
 * To be called immediately before any QEMU monitor API call
470 471
 * Must have already called qemuDomainObjBeginJob(), and checked
 * that the VM is still active.
472 473 474
 *
 * To be followed with qemuDomainObjExitMonitor() once complete
 */
475 476 477 478 479
static void qemuDomainObjEnterMonitor(virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    qemuMonitorLock(priv->mon);
480
    qemuMonitorRef(priv->mon);
481
    virDomainObjUnlock(obj);
482 483 484
}


485 486
/* obj must NOT be locked before calling, qemud_driver must be unlocked
 *
487
 * Should be paired with an earlier qemuDomainObjEnterMonitor() call
488
 */
489 490 491
static void qemuDomainObjExitMonitor(virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
492 493 494 495 496 497
    int refs;

    refs = qemuMonitorUnref(priv->mon);

    if (refs > 0)
        qemuMonitorUnlock(priv->mon);
498

499
    virDomainObjLock(obj);
500 501 502 503 504

    if (refs == 0) {
        virDomainObjUnref(obj);
        priv->mon = NULL;
    }
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
512
 * Must have already called qemuDomainObjBeginJob().
513 514 515 516 517 518 519 520
 *
 * To be followed with qemuDomainObjExitMonitorWithDriver() once complete
 */
static void qemuDomainObjEnterMonitorWithDriver(struct qemud_driver *driver, virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    qemuMonitorLock(priv->mon);
521
    qemuMonitorRef(priv->mon);
522 523 524 525 526 527 528 529
    virDomainObjUnlock(obj);
    qemuDriverUnlock(driver);
}


/* obj must NOT be locked before calling, qemud_driver must be unlocked,
 * and will be locked after returning
 *
530
 * Should be paired with an earlier qemuDomainObjEnterMonitorWithDriver() call
531 532 533 534
 */
static void qemuDomainObjExitMonitorWithDriver(struct qemud_driver *driver, virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
535 536 537 538 539 540
    int refs;

    refs = qemuMonitorUnref(priv->mon);

    if (refs > 0)
        qemuMonitorUnlock(priv->mon);
541 542 543

    qemuDriverLock(driver);
    virDomainObjLock(obj);
544 545 546 547 548

    if (refs == 0) {
        virDomainObjUnref(obj);
        priv->mon = NULL;
    }
549 550
}

551 552 553 554 555 556 557 558 559 560 561 562 563 564 565
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);
}
566

567
void qemuDomainDefNamespaceFree(void *nsdata)
568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586
{
    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);
}

587 588 589 590
int qemuDomainDefNamespaceParse(xmlDocPtr xml,
                                xmlNodePtr root,
                                xmlXPathContextPtr ctxt,
                                void **data)
591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699
{
    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;
}

700 701
int qemuDomainDefNamespaceFormatXML(virBufferPtr buf,
                                    void *nsdata)
702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723
{
    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;
}

724
const char *qemuDomainDefNamespaceHref(void)
725 726 727 728
{
    return "xmlns:qemu='" QEMU_NAMESPACE_HREF "'";
}

729 730 731 732 733
static int qemuCgroupControllerActive(struct qemud_driver *driver,
                                      int controller)
{
    if (driver->cgroup == NULL)
        return 0;
L
Lai Jiangshan 已提交
734 735
    if (!virCgroupMounted(driver->cgroup, controller))
        return 0;
736 737 738 739
    if (driver->cgroupControllers & (1 << controller))
        return 1;
    return 0;
}
740

741
static int
742
qemudLogFD(struct qemud_driver *driver, const char* name)
743 744 745
{
    char logfile[PATH_MAX];
    mode_t logmode;
G
Guido Günther 已提交
746
    int ret, fd = -1;
747

748 749
    if ((ret = snprintf(logfile, sizeof(logfile), "%s/%s.log",
                        driver->logDir, name))
G
Guido Günther 已提交
750
        < 0 || ret >= sizeof(logfile)) {
751
        virReportOOMError();
752 753 754 755
        return -1;
    }

    logmode = O_CREAT | O_WRONLY;
756 757
    /* Only logrotate files in /var/log, so only append if running privileged */
    if (driver->privileged)
758
        logmode |= O_APPEND;
759 760 761
    else
        logmode |= O_TRUNC;

762
    if ((fd = open(logfile, logmode, S_IRUSR | S_IWUSR)) < 0) {
763
        virReportSystemError(errno,
764 765
                             _("failed to create logfile %s"),
                             logfile);
766 767
        return -1;
    }
768
    if (virSetCloseExec(fd) < 0) {
769
        virReportSystemError(errno, "%s",
770
                             _("Unable to set VM logfile close-on-exec flag"));
771 772 773 774 775 776 777
        close(fd);
        return -1;
    }
    return fd;
}


778
static int
779
qemudLogReadFD(const char* logDir, const char* name, off_t pos)
780 781 782 783 784 785 786
{
    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)) {
787 788 789
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("failed to build logfile name %s/%s.log"),
                        logDir, name);
790 791 792 793 794
        return -1;
    }


    if ((fd = open(logfile, logmode)) < 0) {
795
        virReportSystemError(errno,
796 797
                             _("failed to create logfile %s"),
                             logfile);
798 799
        return -1;
    }
800
    if (virSetCloseExec(fd) < 0) {
801
        virReportSystemError(errno, "%s",
802
                             _("Unable to set VM logfile close-on-exec flag"));
803 804 805
        close(fd);
        return -1;
    }
806
    if (pos < 0 || lseek(fd, pos, SEEK_SET) < 0) {
807
      virReportSystemError(pos < 0 ? 0 : errno,
808 809
                             _("Unable to seek to %lld in %s"),
                             (long long) pos, logfile);
810 811 812 813 814 815
        close(fd);
    }
    return fd;
}


816 817 818 819 820 821 822 823 824
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;
825
    virErrorPtr err;
826 827

    virDomainObjLock(vm);
828 829 830 831 832 833 834 835 836
    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) &&
837
            qemudDomainObjStart(data->conn, data->driver, vm, false) < 0) {
838
            err = virGetLastError();
839
            VIR_ERROR(_("Failed to autostart VM '%s': %s"),
840
                      vm->def->name,
841
                      err ? err->message : _("unknown error"));
842
        }
843 844 845

        if (qemuDomainObjEndJob(vm) == 0)
            vm = NULL;
846
    }
847 848 849

    if (vm)
        virDomainObjUnlock(vm);
850 851
}

852 853
static void
qemudAutostartConfigs(struct qemud_driver *driver) {
854 855 856 857 858
    /* 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
     */
859 860 861
    virConnectPtr conn = virConnectOpen(driver->privileged ?
                                        "qemu:///system" :
                                        "qemu:///session");
862
    /* Ignoring NULL conn which is mostly harmless here */
863
    struct qemuAutostartData data = { driver, conn };
864

865
    qemuDriverLock(driver);
866
    virHashForEach(driver->domains.objs, qemuAutostartDomain, &data);
867
    qemuDriverUnlock(driver);
868

869 870
    if (conn)
        virConnectClose(conn);
871 872
}

873 874 875 876 877 878 879 880 881

/**
 * qemudRemoveDomainStatus
 *
 * remove all state files of a domain from statedir
 *
 * Returns 0 on success
 */
static int
882
qemudRemoveDomainStatus(struct qemud_driver *driver,
883 884
                        virDomainObjPtr vm)
{
885
    char ebuf[1024];
886 887 888
    char *file = NULL;

    if (virAsprintf(&file, "%s/%s.xml", driver->stateDir, vm->def->name) < 0) {
889
        virReportOOMError();
D
Daniel Veillard 已提交
890
        return(-1);
891 892
    }

893
    if (unlink(file) < 0 && errno != ENOENT && errno != ENOTDIR)
894
        VIR_WARN("Failed to remove domain XML for %s: %s",
D
Daniel Veillard 已提交
895 896 897
                 vm->def->name, virStrerror(errno, ebuf, sizeof(ebuf)));
    VIR_FREE(file);

898
    if (virFileDeletePid(driver->stateDir, vm->def->name) != 0)
899
        VIR_WARN("Failed to remove PID file for %s: %s",
900
                 vm->def->name, virStrerror(errno, ebuf, sizeof(ebuf)));
901

D
Daniel Veillard 已提交
902

903
    return 0;
904 905
}

906 907 908 909 910 911 912 913 914 915 916 917 918 919

/*
 * 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;

920
    VIR_DEBUG("Received EOF on %p '%s'", vm, vm->def->name);
921

922 923 924 925 926 927 928 929
    virDomainObjLock(vm);

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

930
    qemudShutdownVMDaemon(driver, vm, 0);
931 932
    qemuDomainStopAudit(vm, hasError ? "failed" : "shutdown");

933 934 935 936 937 938 939 940 941 942 943 944 945
    if (!vm->persistent)
        virDomainRemoveInactive(&driver->domains, vm);
    else
        virDomainObjUnlock(vm);

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


946 947 948
static virDomainDiskDefPtr
findDomainDiskByPath(virDomainObjPtr vm,
                     const char *path)
949 950 951 952 953 954 955
{
    int i;

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

        disk = vm->def->disks[i];
956 957
        if (disk->src != NULL && STREQ(disk->src, path))
            return disk;
958
    }
959 960 961 962

    qemuReportError(VIR_ERR_INTERNAL_ERROR,
                    _("no disk found with path %s"),
                    path);
963 964 965
    return NULL;
}

966 967 968 969 970 971
static virDomainDiskDefPtr
findDomainDiskByAlias(virDomainObjPtr vm,
                      const char *alias)
{
    int i;

972 973 974
    if (STRPREFIX(alias, QEMU_DRIVE_HOST_PREFIX))
        alias += strlen(QEMU_DRIVE_HOST_PREFIX);

975 976 977 978 979 980 981 982 983 984 985 986 987 988
    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;
}

989
static int
990 991 992 993
getVolumeQcowPassphrase(virConnectPtr conn,
                        virDomainDiskDefPtr disk,
                        char **secretRet,
                        size_t *secretLen)
994 995 996 997 998
{
    virSecretPtr secret;
    char *passphrase;
    unsigned char *data;
    size_t size;
999
    int ret = -1;
1000
    virStorageEncryptionPtr enc;
1001

1002 1003 1004 1005 1006 1007 1008
    if (!disk->encryption) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("disk %s does not have any encryption information"),
                        disk->src);
        return -1;
    }
    enc = disk->encryption;
1009 1010

    if (!conn) {
1011
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
1012
                        "%s", _("cannot find secrets without a connection"));
1013
        goto cleanup;
1014 1015 1016 1017 1018
    }

    if (conn->secretDriver == NULL ||
        conn->secretDriver->lookupByUUID == NULL ||
        conn->secretDriver->getValue == NULL) {
1019 1020
        qemuReportError(VIR_ERR_NO_SUPPORT, "%s",
                        _("secret storage not supported"));
1021
        goto cleanup;
1022 1023 1024 1025 1026 1027
    }

    if (enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_QCOW ||
        enc->nsecrets != 1 ||
        enc->secrets[0]->type !=
        VIR_STORAGE_ENCRYPTION_SECRET_TYPE_PASSPHRASE) {
1028
        qemuReportError(VIR_ERR_INVALID_DOMAIN,
1029
                        _("invalid <encryption> for volume %s"), disk->src);
1030
        goto cleanup;
1031 1032 1033 1034 1035
    }

    secret = conn->secretDriver->lookupByUUID(conn,
                                              enc->secrets[0]->uuid);
    if (secret == NULL)
1036
        goto cleanup;
1037 1038 1039 1040
    data = conn->secretDriver->getValue(secret, &size,
                                        VIR_SECRET_GET_VALUE_INTERNAL_CALL);
    virUnrefSecret(secret);
    if (data == NULL)
1041
        goto cleanup;
1042 1043 1044 1045

    if (memchr(data, '\0', size) != NULL) {
        memset(data, 0, size);
        VIR_FREE(data);
1046 1047
        qemuReportError(VIR_ERR_INVALID_SECRET,
                        _("format='qcow' passphrase for %s must not contain a "
1048
                          "'\\0'"), disk->src);
1049
        goto cleanup;
1050 1051 1052 1053 1054
    }

    if (VIR_ALLOC_N(passphrase, size + 1) < 0) {
        memset(data, 0, size);
        VIR_FREE(data);
1055
        virReportOOMError();
1056
        goto cleanup;
1057 1058 1059 1060 1061 1062 1063 1064 1065 1066
    }
    memcpy(passphrase, data, size);
    passphrase[size] = '\0';

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

    *secretRet = passphrase;
    *secretLen = size;

1067 1068 1069
    ret = 0;

cleanup:
1070 1071
    return ret;
}
1072

1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093
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);
1094
    return ret;
1095 1096
}

1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118

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;
}


1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150
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;
}


1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179
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;
}


1180 1181 1182 1183 1184 1185
static int
qemuHandleDomainWatchdog(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                         virDomainObjPtr vm,
                         int action)
{
    struct qemud_driver *driver = qemu_driver;
1186 1187
    virDomainEventPtr watchdogEvent = NULL;
    virDomainEventPtr lifecycleEvent = NULL;
1188 1189

    virDomainObjLock(vm);
1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203
    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);
    }
1204 1205
    virDomainObjUnlock(vm);

1206
    if (watchdogEvent || lifecycleEvent) {
1207
        qemuDriverLock(driver);
1208 1209 1210 1211
        if (watchdogEvent)
            qemuDomainEventQueue(driver, watchdogEvent);
        if (lifecycleEvent)
            qemuDomainEventQueue(driver, lifecycleEvent);
1212 1213 1214 1215 1216 1217 1218
        qemuDriverUnlock(driver);
    }

    return 0;
}


1219 1220 1221 1222
static int
qemuHandleDomainIOError(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                        virDomainObjPtr vm,
                        const char *diskAlias,
1223 1224
                        int action,
                        const char *reason)
1225 1226
{
    struct qemud_driver *driver = qemu_driver;
1227
    virDomainEventPtr ioErrorEvent = NULL;
1228
    virDomainEventPtr ioErrorEvent2 = NULL;
1229
    virDomainEventPtr lifecycleEvent = NULL;
1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244
    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 = "";
    }

1245
    ioErrorEvent = virDomainEventIOErrorNewFromObj(vm, srcPath, devAlias, action);
1246
    ioErrorEvent2 = virDomainEventIOErrorReasonNewFromObj(vm, srcPath, devAlias, action, reason);
1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259

    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);
    }
1260 1261
    virDomainObjUnlock(vm);

1262
    if (ioErrorEvent || ioErrorEvent2 || lifecycleEvent) {
1263
        qemuDriverLock(driver);
1264 1265
        if (ioErrorEvent)
            qemuDomainEventQueue(driver, ioErrorEvent);
1266 1267
        if (ioErrorEvent2)
            qemuDomainEventQueue(driver, ioErrorEvent2);
1268 1269
        if (lifecycleEvent)
            qemuDomainEventQueue(driver, lifecycleEvent);
1270 1271 1272 1273 1274 1275 1276
        qemuDriverUnlock(driver);
    }

    return 0;
}


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 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368
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;
}


1369 1370 1371 1372 1373 1374 1375 1376 1377
static void qemuHandleMonitorDestroy(qemuMonitorPtr mon,
                                     virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    if (priv->mon == mon)
        priv->mon = NULL;
    virDomainObjUnref(vm);
}

1378
static qemuMonitorCallbacks monitorCallbacks = {
1379
    .destroy = qemuHandleMonitorDestroy,
1380 1381
    .eofNotify = qemuHandleMonitorEOF,
    .diskSecretLookup = findVolumeQcowPassphrase,
1382
    .domainStop = qemuHandleDomainStop,
1383
    .domainReset = qemuHandleDomainReset,
1384
    .domainRTCChange = qemuHandleDomainRTCChange,
1385
    .domainWatchdog = qemuHandleDomainWatchdog,
1386
    .domainIOError = qemuHandleDomainIOError,
1387
    .domainGraphics = qemuHandleDomainGraphics,
1388 1389
};

1390
static int
1391
qemuConnectMonitor(struct qemud_driver *driver, virDomainObjPtr vm)
1392
{
1393
    qemuDomainObjPrivatePtr priv = vm->privateData;
1394
    int ret = -1;
1395

1396 1397 1398 1399 1400 1401
    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);
1402 1403 1404
        goto error;
    }

1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415
    /* 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);
1416

1417 1418 1419 1420 1421 1422
    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);
1423
        goto error;
1424
    }
1425

1426 1427 1428 1429 1430 1431
    if (priv->mon == NULL) {
        VIR_INFO("Failed to connect monitor for %s", vm->def->name);
        goto error;
    }


1432 1433 1434 1435
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
    ret = qemuMonitorSetCapabilities(priv->mon);
    qemuDomainObjExitMonitorWithDriver(driver, vm);

1436
error:
1437
    if (ret < 0)
1438 1439 1440
        qemuMonitorClose(priv->mon);

    return ret;
1441
}
1442

1443 1444 1445 1446
struct virReconnectDomainData {
    virConnectPtr conn;
    struct qemud_driver *driver;
};
1447 1448 1449
/*
 * Open an existing VM's monitor, re-detect VCPU threads
 * and re-reserve the security labels in use
1450
 */
1451 1452
static void
qemuReconnectDomain(void *payload, const char *name ATTRIBUTE_UNUSED, void *opaque)
1453
{
1454
    virDomainObjPtr obj = payload;
1455 1456
    struct virReconnectDomainData *data = opaque;
    struct qemud_driver *driver = data->driver;
1457
    qemuDomainObjPrivatePtr priv;
1458
    unsigned long long qemuCmdFlags;
1459
    virConnectPtr conn = data->conn;
1460 1461

    virDomainObjLock(obj);
1462

1463 1464
    VIR_DEBUG("Reconnect monitor to %p '%s'", obj, obj->def->name);

1465 1466
    priv = obj->privateData;

1467
    /* XXX check PID liveliness & EXE path */
1468
    if (qemuConnectMonitor(driver, obj) < 0)
1469
        goto error;
1470

1471 1472 1473 1474
    if (qemuUpdateActivePciHostdevs(driver, obj->def) < 0) {
        goto error;
    }

1475 1476 1477 1478 1479 1480
    /* 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 &&
1481
        (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)) {
1482 1483
        priv->persistentAddrs = 1;

1484 1485 1486 1487
        if (!(priv->pciaddrs = qemuDomainPCIAddressSetCreate(obj->def)) ||
            qemuAssignDevicePCISlots(obj->def, priv->pciaddrs) < 0)
            goto error;
    }
1488

1489
    if (driver->securityDriver &&
1490
        driver->securityDriver->domainReserveSecurityLabel &&
1491 1492
        driver->securityDriver->domainReserveSecurityLabel(driver->securityDriver,
                                                           obj) < 0)
1493
        goto error;
1494

1495 1496 1497
    if (qemudVMFiltersInstantiate(conn, obj->def))
        goto error;

1498 1499
    if (obj->def->id >= driver->nextvmid)
        driver->nextvmid = obj->def->id + 1;
1500

1501 1502
    virDomainObjUnlock(obj);
    return;
1503

1504
error:
1505 1506 1507
    /* 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 */
1508
    qemudShutdownVMDaemon(driver, obj, 0);
1509 1510 1511 1512
    if (!obj->persistent)
        virDomainRemoveInactive(&driver->domains, obj);
    else
        virDomainObjUnlock(obj);
1513
}
1514

1515
/**
1516
 * qemudReconnectDomains
1517 1518 1519 1520 1521
 *
 * Try to re-open the resources for live VMs that we care
 * about.
 */
static void
1522
qemuReconnectDomains(virConnectPtr conn, struct qemud_driver *driver)
1523
{
1524 1525
    struct virReconnectDomainData data = {conn, driver};
    virHashForEach(driver->domains.objs, qemuReconnectDomain, &data);
1526 1527
}

1528

1529 1530 1531 1532 1533 1534
static int
qemudSecurityInit(struct qemud_driver *qemud_drv)
{
    int ret;
    virSecurityDriverPtr security_drv;

1535 1536 1537
    qemuSecurityStackedSetDriver(qemud_drv);
    qemuSecurityDACSetDriver(qemud_drv);

1538
    ret = virSecurityDriverStartup(&security_drv,
1539 1540
                                   qemud_drv->securityDriverName,
                                   qemud_drv->allowDiskFormatProbing);
1541 1542 1543 1544
    if (ret == -1) {
        VIR_ERROR0(_("Failed to start security driver"));
        return -1;
    }
1545 1546 1547

    /* No primary security driver wanted to be enabled: just setup
     * the DAC driver on its own */
1548
    if (ret == -2) {
1549
        qemud_drv->securityDriver = &qemuDACSecurityDriver;
1550
        VIR_INFO0(_("No security driver available"));
1551 1552 1553 1554 1555
    } else {
        qemud_drv->securityPrimaryDriver = security_drv;
        qemud_drv->securitySecondaryDriver = &qemuDACSecurityDriver;
        qemud_drv->securityDriver = &qemuStackedSecurityDriver;
        VIR_INFO("Initialized security driver %s", security_drv->name);
1556 1557
    }

1558
    return 0;
1559
}
1560 1561


1562 1563
static virCapsPtr
qemuCreateCapabilities(virCapsPtr oldcaps,
1564
                       struct qemud_driver *driver)
1565 1566 1567 1568 1569
{
    virCapsPtr caps;

    /* Basic host arch / guest machine capabilities */
    if (!(caps = qemudCapsInit(oldcaps))) {
1570
        virReportOOMError();
1571 1572 1573
        return NULL;
    }

1574 1575 1576 1577 1578 1579 1580 1581
    if (driver->allowDiskFormatProbing) {
        caps->defaultDiskDriverName = NULL;
        caps->defaultDiskDriverType = NULL;
    } else {
        caps->defaultDiskDriverName = "qemu";
        caps->defaultDiskDriverType = "raw";
    }

1582 1583 1584 1585 1586 1587
    /* Domain XML parser hooks */
    caps->privateDataAllocFunc = qemuDomainObjPrivateAlloc;
    caps->privateDataFreeFunc = qemuDomainObjPrivateFree;
    caps->privateDataXMLFormat = qemuDomainObjPrivateXMLFormat;
    caps->privateDataXMLParse = qemuDomainObjPrivateXMLParse;

1588 1589 1590 1591 1592
    if (virGetHostUUID(caps->host.host_uuid)) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                         "%s", _("cannot get the host uuid"));
        goto err_exit;
    }
1593

1594 1595 1596 1597 1598 1599
    /* Domain Namespace XML parser hooks */
    caps->ns.parse = qemuDomainDefNamespaceParse;
    caps->ns.free = qemuDomainDefNamespaceFree;
    caps->ns.format = qemuDomainDefNamespaceFormatXML;
    caps->ns.href = qemuDomainDefNamespaceHref;

1600
    /* Security driver data */
1601
    if (driver->securityPrimaryDriver) {
1602 1603
        const char *doi, *model;

1604 1605
        doi = virSecurityDriverGetDOI(driver->securityPrimaryDriver);
        model = virSecurityDriverGetModel(driver->securityPrimaryDriver);
1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618

        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:
1619
    virReportOOMError();
1620
err_exit:
1621 1622 1623
    virCapabilitiesFree(caps);
    return NULL;
}
1624

C
Chris Lalancette 已提交
1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641
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) {
1642
        VIR_ERROR(_("Failed to allocate memory for snapshot directory for domain %s"),
C
Chris Lalancette 已提交
1643 1644 1645 1646 1647 1648 1649 1650 1651
                   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)
1652
            VIR_ERROR(_("Failed to open snapshot directory %s for domain %s: %s"),
C
Chris Lalancette 已提交
1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666
                      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) {
1667
            VIR_ERROR0(_("Failed to allocate memory for path"));
C
Chris Lalancette 已提交
1668 1669 1670 1671 1672 1673
            continue;
        }

        ret = virFileReadAll(fullpath, 1024*1024*1, &xmlStr);
        if (ret < 0) {
            /* Nothing we can do here, skip this one */
1674
            VIR_ERROR(_("Failed to read snapshot file %s: %s"), fullpath,
C
Chris Lalancette 已提交
1675
                      virStrerror(errno, ebuf, sizeof(ebuf)));
1676
            VIR_FREE(fullpath);
C
Chris Lalancette 已提交
1677 1678 1679 1680 1681 1682
            continue;
        }

        def = virDomainSnapshotDefParseString(xmlStr, 0);
        if (def == NULL) {
            /* Nothing we can do here, skip this one */
1683
            VIR_ERROR(_("Failed to parse snapshot XML from file '%s'"), fullpath);
1684
            VIR_FREE(fullpath);
C
Chris Lalancette 已提交
1685 1686 1687 1688
            VIR_FREE(xmlStr);
            continue;
        }

1689
        virDomainSnapshotAssignDef(&vm->snapshots, def);
C
Chris Lalancette 已提交
1690

1691
        VIR_FREE(fullpath);
C
Chris Lalancette 已提交
1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712
        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);
}

1713 1714 1715 1716 1717 1718
/**
 * qemudStartup:
 *
 * Initialization function for the QEmu daemon
 */
static int
1719
qemudStartup(int privileged) {
1720
    char *base = NULL;
D
Daniel P. Berrange 已提交
1721
    char driverConf[PATH_MAX];
1722
    int rc;
1723
    virConnectPtr conn = NULL;
1724

1725
    if (VIR_ALLOC(qemu_driver) < 0)
1726 1727
        return -1;

1728
    if (virMutexInit(&qemu_driver->lock) < 0) {
1729
        VIR_ERROR0(_("cannot initialize mutex"));
1730 1731 1732
        VIR_FREE(qemu_driver);
        return -1;
    }
1733
    qemuDriverLock(qemu_driver);
1734
    qemu_driver->privileged = privileged;
1735

1736 1737 1738
    /* Don't have a dom0 so start from 1 */
    qemu_driver->nextvmid = 1;

1739 1740 1741
    if (virDomainObjListInit(&qemu_driver->domains) < 0)
        goto out_of_memory;

1742
    /* Init callback list */
1743
    if (VIR_ALLOC(qemu_driver->domainEventCallbacks) < 0)
1744
        goto out_of_memory;
1745 1746 1747 1748 1749 1750
    if (!(qemu_driver->domainEventQueue = virDomainEventQueueNew()))
        goto out_of_memory;

    if ((qemu_driver->domainEventTimer =
         virEventAddTimeout(-1, qemuDomainEventFlush, qemu_driver, NULL)) < 0)
        goto error;
1751

1752 1753 1754 1755 1756
    /* Allocate bitmap for vnc port reservation */
    if ((qemu_driver->reservedVNCPorts =
         virBitmapAlloc(QEMU_VNC_PORT_MAX - QEMU_VNC_PORT_MIN)) == NULL)
        goto out_of_memory;

1757 1758 1759 1760
    /* read the host sysinfo */
    if (privileged)
        qemu_driver->hostsysinfo = virSysinfoRead();

1761
    if (privileged) {
1762 1763
        if (virAsprintf(&qemu_driver->logDir,
                        "%s/log/libvirt/qemu", LOCAL_STATE_DIR) == -1)
1764
            goto out_of_memory;
1765

D
Daniel P. Berrange 已提交
1766
        if ((base = strdup (SYSCONF_DIR "/libvirt")) == NULL)
1767
            goto out_of_memory;
1768 1769

        if (virAsprintf(&qemu_driver->stateDir,
1770
                      "%s/run/libvirt/qemu", LOCAL_STATE_DIR) == -1)
1771
            goto out_of_memory;
1772 1773 1774 1775 1776 1777 1778 1779

        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;
1780
        if (virAsprintf(&qemu_driver->saveDir,
1781
                      "%s/lib/libvirt/qemu/save", LOCAL_STATE_DIR) == -1)
1782
            goto out_of_memory;
C
Chris Lalancette 已提交
1783 1784 1785
        if (virAsprintf(&qemu_driver->snapshotDir,
                        "%s/lib/libvirt/qemu/snapshot", LOCAL_STATE_DIR) == -1)
            goto out_of_memory;
1786
    } else {
1787
        uid_t uid = geteuid();
1788
        char *userdir = virGetUserDirectory(uid);
1789
        if (!userdir)
1790
            goto error;
1791

1792
        if (virAsprintf(&qemu_driver->logDir,
1793 1794
                        "%s/.libvirt/qemu/log", userdir) == -1) {
            VIR_FREE(userdir);
1795
            goto out_of_memory;
1796
        }
1797

1798 1799
        if (virAsprintf(&base, "%s/.libvirt", userdir) == -1) {
            VIR_FREE(userdir);
1800
            goto out_of_memory;
1801 1802
        }
        VIR_FREE(userdir);
1803 1804 1805

        if (virAsprintf(&qemu_driver->stateDir, "%s/qemu/run", base) == -1)
            goto out_of_memory;
1806 1807 1808 1809
        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;
1810 1811
        if (virAsprintf(&qemu_driver->saveDir, "%s/qemu/save", base) == -1)
            goto out_of_memory;
C
Chris Lalancette 已提交
1812 1813
        if (virAsprintf(&qemu_driver->snapshotDir, "%s/qemu/snapshot", base) == -1)
            goto out_of_memory;
1814 1815
    }

L
Laine Stump 已提交
1816
    if (virFileMakePath(qemu_driver->stateDir) != 0) {
1817
        char ebuf[1024];
1818
        VIR_ERROR(_("Failed to create state dir '%s': %s"),
1819
                  qemu_driver->stateDir, virStrerror(errno, ebuf, sizeof ebuf));
1820
        goto error;
1821
    }
L
Laine Stump 已提交
1822
    if (virFileMakePath(qemu_driver->libDir) != 0) {
1823
        char ebuf[1024];
1824
        VIR_ERROR(_("Failed to create lib dir '%s': %s"),
1825 1826 1827
                  qemu_driver->libDir, virStrerror(errno, ebuf, sizeof ebuf));
        goto error;
    }
L
Laine Stump 已提交
1828
    if (virFileMakePath(qemu_driver->cacheDir) != 0) {
1829
        char ebuf[1024];
1830
        VIR_ERROR(_("Failed to create cache dir '%s': %s"),
1831 1832 1833
                  qemu_driver->cacheDir, virStrerror(errno, ebuf, sizeof ebuf));
        goto error;
    }
1834 1835 1836 1837 1838 1839
    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
1840 1841 1842 1843 1844 1845
    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;
    }
1846 1847 1848 1849

    /* Configuration paths are either ~/.libvirt/qemu/... (session) or
     * /etc/libvirt/qemu/... (system).
     */
D
Daniel P. Berrange 已提交
1850
    if (snprintf (driverConf, sizeof(driverConf), "%s/qemu.conf", base) == -1)
1851
        goto out_of_memory;
D
Daniel P. Berrange 已提交
1852
    driverConf[sizeof(driverConf)-1] = '\0';
1853

1854
    if (virAsprintf(&qemu_driver->configDir, "%s/qemu", base) == -1)
1855 1856
        goto out_of_memory;

1857
    if (virAsprintf(&qemu_driver->autostartDir, "%s/qemu/autostart", base) == -1)
1858 1859
        goto out_of_memory;

1860
    VIR_FREE(base);
1861

1862 1863 1864 1865 1866 1867 1868
    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)));
    }

1869 1870 1871 1872
    if (qemudLoadDriverConfig(qemu_driver, driverConf) < 0) {
        goto error;
    }

1873 1874
    if (qemudSecurityInit(qemu_driver) < 0)
        goto error;
D
Daniel P. Berrange 已提交
1875

1876
    if ((qemu_driver->caps = qemuCreateCapabilities(NULL,
1877
                                                    qemu_driver)) == NULL)
1878
        goto error;
1879

1880
    if ((qemu_driver->activePciHostdevs = pciDeviceListNew()) == NULL)
1881 1882
        goto error;

1883 1884
    if (privileged) {
        if (chown(qemu_driver->libDir, qemu_driver->user, qemu_driver->group) < 0) {
1885
            virReportSystemError(errno,
1886 1887 1888 1889 1890
                                 _("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) {
1891
            virReportSystemError(errno,
1892 1893 1894 1895
                                 _("unable to set ownership of '%s' to %d:%d"),
                                 qemu_driver->cacheDir, qemu_driver->user, qemu_driver->group);
            goto error;
        }
1896 1897 1898 1899 1900 1901
        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
1902 1903 1904 1905 1906 1907
        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;
        }
1908 1909
    }

1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923
    /* 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) {
1924
            virReportSystemError(rc,
1925 1926 1927 1928 1929 1930
                                 _("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) {
1931
            virReportSystemError(errno,
1932 1933 1934 1935 1936 1937 1938 1939 1940
                                 _("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;
    }

1941
    /* Get all the running persistent or transient configs first */
1942
    if (virDomainLoadAllConfigs(qemu_driver->caps,
1943 1944 1945 1946 1947 1948
                                &qemu_driver->domains,
                                qemu_driver->stateDir,
                                NULL,
                                1, NULL, NULL) < 0)
        goto error;

1949 1950 1951 1952 1953
    conn = virConnectOpen(qemu_driver->privileged ?
                          "qemu:///system" :
                          "qemu:///session");

    qemuReconnectDomains(conn, qemu_driver);
1954 1955

    /* Then inactive persistent configs */
1956
    if (virDomainLoadAllConfigs(qemu_driver->caps,
1957 1958
                                &qemu_driver->domains,
                                qemu_driver->configDir,
1959
                                qemu_driver->autostartDir,
1960
                                0, NULL, NULL) < 0)
1961
        goto error;
C
Chris Lalancette 已提交
1962 1963 1964 1965 1966


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

1967 1968
    qemuDriverUnlock(qemu_driver);

1969 1970
    qemudAutostartConfigs(qemu_driver);

1971 1972
    if (conn)
        virConnectClose(conn);
1973

1974 1975
    return 0;

1976
out_of_memory:
1977
    virReportOOMError();
1978 1979 1980
error:
    if (qemu_driver)
        qemuDriverUnlock(qemu_driver);
1981 1982
    if (conn)
        virConnectClose(conn);
1983
    VIR_FREE(base);
1984
    qemudShutdown();
1985 1986 1987
    return -1;
}

1988 1989 1990 1991
static void qemudNotifyLoadDomain(virDomainObjPtr vm, int newVM, void *opaque)
{
    struct qemud_driver *driver = opaque;

1992 1993 1994 1995 1996 1997 1998 1999
    if (newVM) {
        virDomainEventPtr event =
            virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_DEFINED,
                                     VIR_DOMAIN_EVENT_DEFINED_ADDED);
        if (event)
            qemuDomainEventQueue(driver, event);
    }
2000 2001
}

2002 2003 2004 2005 2006 2007 2008 2009
/**
 * qemudReload:
 *
 * Function to restart the QEmu daemon, it will recheck the configuration
 * files and update its state and the networking
 */
static int
qemudReload(void) {
2010 2011 2012
    if (!qemu_driver)
        return 0;

2013
    qemuDriverLock(qemu_driver);
2014
    virDomainLoadAllConfigs(qemu_driver->caps,
2015 2016
                            &qemu_driver->domains,
                            qemu_driver->configDir,
2017
                            qemu_driver->autostartDir,
2018
                            0, qemudNotifyLoadDomain, qemu_driver);
2019
    qemuDriverUnlock(qemu_driver);
2020

2021
    qemudAutostartConfigs(qemu_driver);
2022 2023

    return 0;
2024 2025
}

2026 2027 2028 2029 2030 2031 2032 2033 2034 2035
/**
 * 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) {
2036
    int active = 0;
2037

2038 2039 2040
    if (!qemu_driver)
        return 0;

2041
    /* XXX having to iterate here is not great because it requires many locks */
2042
    qemuDriverLock(qemu_driver);
2043
    active = virDomainObjListNumOfDomains(&qemu_driver->domains, 1);
2044 2045
    qemuDriverUnlock(qemu_driver);
    return active;
2046 2047
}

2048 2049 2050 2051 2052 2053 2054
/**
 * qemudShutdown:
 *
 * Shutdown the QEmu daemon, it will stop all active domains and networks
 */
static int
qemudShutdown(void) {
2055
    int i;
2056

2057
    if (!qemu_driver)
2058
        return -1;
2059

2060
    qemuDriverLock(qemu_driver);
2061
    pciDeviceListFree(qemu_driver->activePciHostdevs);
2062 2063
    virCapabilitiesFree(qemu_driver->caps);

2064
    virDomainObjListDeinit(&qemu_driver->domains);
2065
    virBitmapFree(qemu_driver->reservedVNCPorts);
2066

2067 2068
    virSysinfoDefFree(qemu_driver->hostsysinfo);

2069
    VIR_FREE(qemu_driver->securityDriverName);
2070
    VIR_FREE(qemu_driver->logDir);
2071 2072
    VIR_FREE(qemu_driver->configDir);
    VIR_FREE(qemu_driver->autostartDir);
2073
    VIR_FREE(qemu_driver->stateDir);
2074 2075
    VIR_FREE(qemu_driver->libDir);
    VIR_FREE(qemu_driver->cacheDir);
2076
    VIR_FREE(qemu_driver->saveDir);
C
Chris Lalancette 已提交
2077
    VIR_FREE(qemu_driver->snapshotDir);
2078
    VIR_FREE(qemu_driver->vncTLSx509certdir);
J
Jim Meyering 已提交
2079
    VIR_FREE(qemu_driver->vncListen);
2080
    VIR_FREE(qemu_driver->vncPassword);
2081
    VIR_FREE(qemu_driver->vncSASLdir);
2082
    VIR_FREE(qemu_driver->saveImageFormat);
2083 2084
    VIR_FREE(qemu_driver->hugetlbfs_mount);
    VIR_FREE(qemu_driver->hugepage_path);
D
Daniel P. Berrange 已提交
2085

2086 2087 2088 2089 2090 2091
    if (qemu_driver->cgroupDeviceACL) {
        for (i = 0 ; qemu_driver->cgroupDeviceACL[i] != NULL ; i++)
            VIR_FREE(qemu_driver->cgroupDeviceACL[i]);
        VIR_FREE(qemu_driver->cgroupDeviceACL);
    }

2092 2093
    /* Free domain callback list */
    virDomainEventCallbackListFree(qemu_driver->domainEventCallbacks);
2094 2095 2096 2097
    virDomainEventQueueFree(qemu_driver->domainEventQueue);

    if (qemu_driver->domainEventTimer != -1)
        virEventRemoveTimeout(qemu_driver->domainEventTimer);
2098

2099 2100 2101
    if (qemu_driver->brctl)
        brShutdown(qemu_driver->brctl);

2102 2103
    virCgroupFree(&qemu_driver->cgroup);

2104
    qemuDriverUnlock(qemu_driver);
2105
    virMutexDestroy(&qemu_driver->lock);
2106
    VIR_FREE(qemu_driver);
2107 2108

    return 0;
2109 2110
}

2111
typedef int qemuLogHandleOutput(virDomainObjPtr vm,
2112 2113
                                const char *output,
                                int fd);
2114 2115 2116 2117 2118

/*
 * Returns -1 for error, 0 on success
 */
static int
2119
qemudReadLogOutput(virDomainObjPtr vm,
2120 2121
                   int fd,
                   char *buf,
G
Guido Günther 已提交
2122
                   size_t buflen,
2123
                   qemuLogHandleOutput func,
2124 2125 2126
                   const char *what,
                   int timeout)
{
2127
    int retries = (timeout*10);
2128
    int got = 0;
2129 2130 2131
    buf[0] = '\0';

    while (retries) {
2132
        ssize_t func_ret, ret;
2133
        int isdead = 0;
G
Guido Günther 已提交
2134

2135
        func_ret = func(vm, buf, fd);
2136

2137 2138
        if (kill(vm->pid, 0) == -1 && errno == ESRCH)
            isdead = 1;
2139

2140 2141
        /* Any failures should be detected before we read the log, so we
         * always have something useful to report on failure. */
2142 2143
        ret = saferead(fd, buf+got, buflen-got-1);
        if (ret < 0) {
2144
            virReportSystemError(errno,
2145 2146 2147 2148 2149
                                 _("Failure while reading %s log output"),
                                 what);
            return -1;
        }

2150 2151 2152
        got += ret;
        buf[got] = '\0';
        if (got == buflen-1) {
2153
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
2154 2155
                            _("Out of space while reading %s log output: %s"),
                            what, buf);
2156 2157 2158 2159
            return -1;
        }

        if (isdead) {
2160
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
2161 2162
                            _("Process exited while reading %s log output: %s"),
                            what, buf);
2163 2164 2165
            return -1;
        }

2166 2167
        if (func_ret <= 0)
            return func_ret;
2168 2169 2170 2171

        usleep(100*1000);
        retries--;
    }
2172

2173
    qemuReportError(VIR_ERR_INTERNAL_ERROR,
2174 2175
                    _("Timed out while reading %s log output: %s"),
                    what, buf);
2176 2177 2178
    return -1;
}

2179

2180 2181 2182 2183 2184 2185 2186 2187
/*
 * 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
 */
2188
static int
2189
qemudExtractTTYPath(const char *haystack,
2190 2191
                    size_t *offset,
                    char **path)
2192
{
2193
    static const char needle[] = "char device redirected to";
2194
    char *tmp, *dev;
2195

2196
    VIR_FREE(*path);
2197
    /* First look for our magic string */
2198 2199 2200 2201 2202
    if (!(tmp = strstr(haystack + *offset, needle))) {
        return 1;
    }
    tmp += sizeof(needle);
    dev = tmp;
2203

2204 2205 2206 2207 2208
    /*
     * And look for first whitespace character and nul terminate
     * to mark end of the pty path
     */
    while (*tmp) {
2209
        if (c_isspace(*tmp)) {
2210 2211
            *path = strndup(dev, tmp-dev);
            if (*path == NULL) {
2212
                virReportOOMError();
2213 2214
                return -1;
            }
2215

2216
            /* ... now further update offset till we get EOL */
2217
            *offset = tmp - haystack;
2218 2219
            return 0;
        }
2220
        tmp++;
2221 2222 2223 2224 2225
    }

    /*
     * We found a path, but didn't find any whitespace,
     * so it must be still incomplete - we should at
2226 2227
     * least see a \n - indicate that we want to carry
     * on trying again
2228
     */
2229
    return 1;
2230 2231
}

2232
static int
2233
qemudFindCharDevicePTYsMonitor(virDomainObjPtr vm,
2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248
                               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) {                                           \
2249 2250 2251
                if (chr->data.file.path == NULL) {                        \
                    /* neither the log output nor 'info chardev' had a */ \
                    /* pty path for this chardev, report an error */      \
2252 2253 2254
                    qemuReportError(VIR_ERR_INTERNAL_ERROR,               \
                                    _("no assigned pty for device %s"), id); \
                    return -1;                                            \
2255 2256 2257 2258 2259
                } else {                                                  \
                    /* 'info chardev' had no pty path for this chardev, */\
                    /* but the log output had, so we're fine */           \
                    continue;                                             \
                }                                                         \
2260 2261
            }                                                             \
                                                                          \
2262
            VIR_FREE(chr->data.file.path);                                \
2263
            chr->data.file.path = strdup(path);                           \
2264 2265
                                                                          \
            if (chr->data.file.path == NULL) {                            \
2266
                virReportOOMError();                                      \
2267 2268
                return -1;                                                \
            }                                                             \
2269 2270 2271 2272 2273 2274
        }                                                                 \
    }

    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
2275 2276
    if (vm->def->console)
        LOOKUP_PTYS(&vm->def->console, 1,  "console");
2277
#undef LOOKUP_PTYS
2278 2279 2280 2281

    return 0;
}

2282
static int
2283
qemudFindCharDevicePTYs(virDomainObjPtr vm,
2284 2285
                        const char *output,
                        int fd ATTRIBUTE_UNUSED)
2286
{
2287
    size_t offset = 0;
2288
    int ret, i;
2289 2290

    /* The order in which QEMU prints out the PTY paths is
2291 2292
       the order in which it procsses its serial and parallel
       device args. This code must match that ordering.... */
2293

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

2304
    /* then the parallel devices */
2305 2306
    for (i = 0 ; i < vm->def->nparallels ; i++) {
        virDomainChrDefPtr chr = vm->def->parallels[i];
2307
        if (chr->type == VIR_DOMAIN_CHR_TYPE_PTY) {
2308
            if ((ret = qemudExtractTTYPath(output, &offset,
2309
                                           &chr->data.file.path)) != 0)
2310
                return ret;
2311 2312 2313
        }
    }

2314 2315 2316 2317
    /* 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) {
2318
            if ((ret = qemudExtractTTYPath(output, &offset,
2319 2320 2321 2322 2323
                                           &chr->data.file.path)) != 0)
                return ret;
        }
    }

2324
    return 0;
2325 2326
}

2327 2328 2329 2330 2331
static void qemudFreePtyPath(void *payload, const char *name ATTRIBUTE_UNUSED)
{
    VIR_FREE(payload);
}

2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345
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';
}

2346
static int
2347
qemudWaitForMonitor(struct qemud_driver* driver,
2348
                    virDomainObjPtr vm, off_t pos)
2349
{
2350
    char buf[4096] = ""; /* Plenty of space to get startup greeting */
2351
    int logfd;
2352
    int ret = -1;
2353
    virHashTablePtr paths = NULL;
2354

2355
    if ((logfd = qemudLogReadFD(driver->logDir, vm->def->name, pos)) < 0)
2356
        return -1;
2357

2358 2359 2360 2361
    if (qemudReadLogOutput(vm, logfd, buf, sizeof(buf),
                           qemudFindCharDevicePTYs,
                           "console", 30) < 0)
        goto closelog;
2362

2363
    VIR_DEBUG("Connect monitor to %p '%s'", vm, vm->def->name);
2364 2365 2366
    if (qemuConnectMonitor(driver, vm) < 0) {
        goto cleanup;
    }
2367

2368 2369 2370 2371
    /* 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. */
2372
    paths = virHashCreate(0);
2373
    if (paths == NULL) {
2374
        virReportOOMError();
2375 2376 2377
        goto cleanup;
    }

2378
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
2379 2380
    qemuDomainObjPrivatePtr priv = vm->privateData;
    ret = qemuMonitorGetPtyPaths(priv->mon, paths);
2381
    qemuDomainObjExitMonitorWithDriver(driver, vm);
2382 2383 2384

    VIR_DEBUG("qemuMonitorGetPtyPaths returned %i", ret);
    if (ret == 0) {
2385
        ret = qemudFindCharDevicePTYsMonitor(vm, paths);
2386 2387 2388 2389 2390 2391 2392
    }

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

2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405
    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];
2406
        VIR_WARN("Unable to close logfile: %s",
2407 2408 2409
                 virStrerror(errno, ebuf, sizeof ebuf));
    }

2410
    return ret;
2411 2412
}

2413
static int
2414
qemuDetectVcpuPIDs(struct qemud_driver *driver,
2415 2416 2417
                   virDomainObjPtr vm) {
    pid_t *cpupids = NULL;
    int ncpupids;
2418
    qemuDomainObjPrivatePtr priv = vm->privateData;
2419

2420
    if (vm->def->virtType != VIR_DOMAIN_VIRT_KVM) {
2421 2422
        priv->nvcpupids = 1;
        if (VIR_ALLOC_N(priv->vcpupids, priv->nvcpupids) < 0) {
2423
            virReportOOMError();
2424 2425
            return -1;
        }
2426
        priv->vcpupids[0] = vm->pid;
2427 2428 2429
        return 0;
    }

2430
    /* What follows is now all KVM specific */
2431

2432
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
2433
    if ((ncpupids = qemuMonitorGetCPUInfo(priv->mon, &cpupids)) < 0) {
2434
        qemuDomainObjExitMonitorWithDriver(driver, vm);
2435
        return -1;
2436
    }
2437
    qemuDomainObjExitMonitorWithDriver(driver, vm);
2438

2439 2440 2441
    /* Treat failure to get VCPU<->PID mapping as non-fatal */
    if (ncpupids == 0)
        return 0;
2442

2443
    if (ncpupids != vm->def->vcpus) {
2444
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
E
Eric Blake 已提交
2445 2446 2447
                        _("got wrong number of vCPU pids from QEMU monitor. "
                          "got %d, wanted %d"),
                        ncpupids, vm->def->vcpus);
2448 2449 2450
        VIR_FREE(cpupids);
        return -1;
    }
2451

2452 2453
    priv->nvcpupids = ncpupids;
    priv->vcpupids = cpupids;
2454 2455 2456
    return 0;
}

2457 2458 2459
/*
 * To be run between fork/exec of QEMU only
 */
2460
static int
2461 2462
qemudInitCpuAffinity(virDomainObjPtr vm)
{
2463
    int i, hostcpus, maxcpu = QEMUD_CPUMASK_LEN;
2464
    virNodeInfo nodeinfo;
2465 2466
    unsigned char *cpumap;
    int cpumaplen;
2467 2468

    DEBUG0("Setting CPU affinity");
2469

2470
    if (nodeGetInfo(NULL, &nodeinfo) < 0)
2471 2472 2473 2474
        return -1;

    /* setaffinity fails if you set bits for CPUs which
     * aren't present, so we have to limit ourselves */
2475 2476 2477
    hostcpus = VIR_NODEINFO_MAXCPUS(nodeinfo);
    if (maxcpu > hostcpus)
        maxcpu = hostcpus;
2478

2479 2480
    cpumaplen = VIR_CPU_MAPLEN(maxcpu);
    if (VIR_ALLOC_N(cpumap, cpumaplen) < 0) {
2481
        virReportOOMError();
2482 2483 2484
        return -1;
    }

D
Daniel P. Berrange 已提交
2485
    if (vm->def->cpumask) {
2486 2487 2488
        /* 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
2489
            if (vm->def->cpumask[i])
2490
                VIR_USE_CPU(cpumap, i);
D
Daniel P. Berrange 已提交
2491
    } else {
2492 2493 2494 2495
        /* 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
2496
        for (i = 0 ; i < maxcpu ; i++)
2497
            VIR_USE_CPU(cpumap, i);
D
Daniel P. Berrange 已提交
2498
    }
2499

2500 2501 2502 2503 2504 2505 2506 2507
    /* 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;
2508
    }
2509
    VIR_FREE(cpumap);
2510 2511 2512 2513 2514

    return 0;
}


2515
static int
2516 2517 2518 2519
qemuInitPasswords(virConnectPtr conn,
                  struct qemud_driver *driver,
                  virDomainObjPtr vm,
                  unsigned long long qemuCmdFlags) {
2520
    int ret = 0;
2521
    qemuDomainObjPrivatePtr priv = vm->privateData;
2522

2523 2524 2525
    if ((vm->def->ngraphics == 1) &&
        vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
        (vm->def->graphics[0]->data.vnc.passwd || driver->vncPassword)) {
2526

2527
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
2528
        ret = qemuMonitorSetVNCPassword(priv->mon,
2529 2530 2531
                                        vm->def->graphics[0]->data.vnc.passwd ?
                                        vm->def->graphics[0]->data.vnc.passwd :
                                        driver->vncPassword);
2532
        qemuDomainObjExitMonitorWithDriver(driver, vm);
2533 2534
    }

2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557
    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);
2558
            VIR_FREE(secret);
2559 2560 2561 2562 2563 2564 2565
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            if (ret < 0)
                goto cleanup;
        }
    }

cleanup:
2566
    return ret;
2567 2568 2569
}


2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581
#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

2582 2583
#define QEMU_PCI_PRODUCT_BALLOON_VIRTIO 0x1002

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 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791
#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;
}


2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810
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;
}


2811 2812 2813 2814 2815 2816 2817 2818 2819
/*
 * 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
2820
qemuDetectPCIAddresses(virDomainObjPtr vm,
2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837
                       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) {
2838 2839 2840
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for VirtIO disk %s"),
                            vm->def->disks[i]->dst);
2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851
            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) {
2852 2853 2854
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for %s NIC"),
                            vm->def->nets[i]->model);
2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865
            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) {
2866 2867 2868
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for controller %s"),
                            virDomainControllerTypeToString(vm->def->controllers[i]->type));
2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879
            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) {
2880 2881 2882
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for video adapter %s"),
                            virDomainVideoTypeToString(vm->def->videos[i]->type));
2883 2884 2885 2886 2887 2888 2889 2890 2891
            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),
2892
                                    vendor, product,
2893
                                     addrs,  naddrs) < 0) {
2894 2895 2896
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for sound adapter %s"),
                            virDomainSoundModelTypeToString(vm->def->sounds[i]->model));
2897 2898 2899 2900 2901 2902 2903 2904 2905 2906
            return -1;
        }
    }


    if (vm->def->watchdog &&
        qemuGetPCIWatchdogVendorProduct(vm->def->watchdog, &vendor, &product) == 0) {
        if (qemuAssignNextPCIAddress(&(vm->def->watchdog->info),
                                     vendor, product,
                                     addrs,  naddrs) < 0) {
2907 2908 2909
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for watchdog %s"),
                            virDomainWatchdogModelTypeToString(vm->def->watchdog->model));
2910 2911 2912 2913
            return -1;
        }
    }

2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925
    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;
        }
    }

2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951
    /* 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);

2952
    ret = qemuDetectPCIAddresses(vm, addrs, naddrs);
2953 2954 2955 2956 2957 2958

    VIR_FREE(addrs);

    return ret;
}

2959 2960 2961

static int qemudNextFreePort(struct qemud_driver *driver,
                             int startPort) {
2962 2963
    int i;

2964
    for (i = startPort ; i < 65535 ; i++) {
2965 2966 2967
        int fd;
        int reuse = 1;
        struct sockaddr_in addr;
2968 2969 2970 2971 2972 2973 2974 2975 2976
        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;

2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991
        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);
2992 2993 2994 2995 2996 2997
            /* 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);
            }
2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011
            return i;
        }
        close(fd);

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

3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049

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;
}


3050
static pciDeviceList *
3051
qemuGetPciHostDeviceList(virDomainHostdevDefPtr *hostdevs, int nhostdevs)
3052 3053
{
    pciDeviceList *list;
3054 3055
    int i;

3056
    if (!(list = pciDeviceListNew()))
3057
        return NULL;
3058

3059 3060
    for (i = 0 ; i < nhostdevs ; i++) {
        virDomainHostdevDefPtr hostdev = hostdevs[i];
3061 3062 3063 3064 3065 3066 3067
        pciDevice *dev;

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

3068
        dev = pciGetDevice(hostdev->source.subsys.u.pci.domain,
3069 3070 3071
                           hostdev->source.subsys.u.pci.bus,
                           hostdev->source.subsys.u.pci.slot,
                           hostdev->source.subsys.u.pci.function);
3072
        if (!dev) {
3073
            pciDeviceListFree(list);
3074 3075
            return NULL;
        }
3076

3077 3078 3079
        if (pciDeviceListAdd(list, dev) < 0) {
            pciFreeDevice(dev);
            pciDeviceListFree(list);
3080
            return NULL;
3081 3082
        }

3083
        pciDeviceSetManaged(dev, hostdev->managed);
3084 3085
    }

3086 3087 3088 3089
    return list;
}

static int
3090 3091 3092 3093
qemuUpdateActivePciHostdevs(struct qemud_driver *driver,
                            virDomainDefPtr def)
{
    pciDeviceList *pcidevs;
3094
    int i;
3095
    int ret = -1;
3096 3097 3098 3099

    if (!def->nhostdevs)
        return 0;

3100
    if (!(pcidevs = qemuGetPciHostDeviceList(def->hostdevs, def->nhostdevs)))
3101 3102
        return -1;

3103 3104
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
3105 3106 3107
        pciDeviceListSteal(pcidevs, dev);
        if (pciDeviceListAdd(driver->activePciHostdevs, dev) < 0) {
            pciFreeDevice(dev);
3108
            goto cleanup;
3109 3110 3111
        }
    }

3112 3113 3114
    ret = 0;

cleanup:
3115
    pciDeviceListFree(pcidevs);
3116 3117 3118
    return ret;
}

3119

3120
static int
3121 3122 3123
qemuPrepareHostdevPCIDevices(struct qemud_driver *driver,
                             virDomainHostdevDefPtr *hostdevs,
                             int nhostdevs)
3124 3125 3126
{
    pciDeviceList *pcidevs;
    int i;
3127
    int ret = -1;
3128

3129
    if (!(pcidevs = qemuGetPciHostDeviceList(hostdevs, nhostdevs)))
3130 3131
        return -1;

3132
    /* We have to use 3 loops here. *All* devices must
3133 3134
     * be detached before we reset any of them, because
     * in some cases you have to reset the whole PCI,
3135 3136
     * which impacts all devices on it. Also, all devices
     * must be reset before being marked as active.
3137 3138 3139 3140 3141 3142 3143
     */

    /* 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
     */

3144 3145
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
3146
        if (!pciDeviceIsAssignable(dev, !driver->relaxedACS))
3147 3148
            goto cleanup;

3149
        if (pciDeviceGetManaged(dev) &&
3150
            pciDettachDevice(dev, driver->activePciHostdevs) < 0)
3151 3152
            goto cleanup;
    }
3153 3154 3155

    /* Now that all the PCI hostdevs have be dettached, we can safely
     * reset them */
3156 3157
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
3158
        if (pciResetDevice(dev, driver->activePciHostdevs, pcidevs) < 0)
3159 3160
            goto cleanup;
    }
3161

3162
    /* Now mark all the devices as active */
3163 3164
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
3165 3166 3167
        pciDeviceListSteal(pcidevs, dev);
        if (pciDeviceListAdd(driver->activePciHostdevs, dev) < 0) {
            pciFreeDevice(dev);
3168 3169
            goto cleanup;
        }
3170 3171
    }

3172
    ret = 0;
3173

3174
cleanup:
3175
    pciDeviceListFree(pcidevs);
3176
    return ret;
3177 3178
}

3179 3180 3181 3182 3183 3184 3185
static int
qemuPrepareHostPCIDevices(struct qemud_driver *driver,
                          virDomainDefPtr def)
{
    return qemuPrepareHostdevPCIDevices(driver, def->hostdevs, def->nhostdevs);
}

3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196

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;
3197
        if (hostdev->source.subsys.type != VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB)
3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235
            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;
}


3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257
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;
}


3258
static void
3259
qemudReattachPciDevice(pciDevice *dev, struct qemud_driver *driver)
3260 3261 3262
{
    int retries = 100;

3263 3264 3265 3266 3267 3268
    while (pciWaitForDeviceCleanup(dev, "kvm_assigned_device")
           && retries) {
        usleep(100*1000);
        retries--;
    }

3269
    if (pciDeviceGetManaged(dev)) {
3270
        if (pciReAttachDevice(dev, driver->activePciHostdevs) < 0) {
3271 3272
            virErrorPtr err = virGetLastError();
            VIR_ERROR(_("Failed to re-attach PCI device: %s"),
3273
                      err ? err->message : _("unknown error"));
3274 3275 3276 3277 3278
            virResetError(err);
        }
    }
}

3279
static void
3280 3281 3282
qemuDomainReAttachHostdevDevices(struct qemud_driver *driver,
                              virDomainHostdevDefPtr *hostdevs,
                              int nhostdevs)
3283
{
3284
    pciDeviceList *pcidevs;
3285 3286
    int i;

3287
    if (!(pcidevs = qemuGetPciHostDeviceList(hostdevs, nhostdevs))) {
3288
        virErrorPtr err = virGetLastError();
3289
        VIR_ERROR(_("Failed to allocate pciDeviceList: %s"),
3290
                  err ? err->message : _("unknown error"));
3291 3292
        virResetError(err);
        return;
3293 3294
    }

3295 3296
    /* Again 3 loops; mark all devices as inactive before reset
     * them and reset all the devices before re-attach */
3297

3298 3299
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
3300
        pciDeviceListDel(driver->activePciHostdevs, dev);
3301
    }
3302

3303 3304
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
3305
        if (pciResetDevice(dev, driver->activePciHostdevs, pcidevs) < 0) {
3306
            virErrorPtr err = virGetLastError();
3307
            VIR_ERROR(_("Failed to reset PCI device: %s"),
3308
                      err ? err->message : _("unknown error"));
3309 3310
            virResetError(err);
        }
3311
    }
3312

3313 3314
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
3315
        qemudReattachPciDevice(dev, driver);
3316
    }
3317

3318
    pciDeviceListFree(pcidevs);
3319 3320
}

3321 3322 3323 3324 3325 3326 3327 3328 3329 3330
static void
qemuDomainReAttachHostDevices(struct qemud_driver *driver,
                              virDomainDefPtr def)
{
    if (!def->nhostdevs)
        return;

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

3331 3332 3333 3334 3335 3336 3337 3338 3339 3340
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

3341

3342 3343 3344 3345 3346 3347 3348
static int qemuSetupDiskPathAllow(virDomainDiskDefPtr disk ATTRIBUTE_UNUSED,
                                  const char *path,
                                  size_t depth ATTRIBUTE_UNUSED,
                                  void *opaque)
{
    virCgroupPtr cgroup = opaque;
    int rc;
3349

3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363
    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;
3364
        }
3365
    }
3366 3367
    return 0;
}
3368 3369


3370 3371
static int qemuSetupDiskCgroup(struct qemud_driver *driver,
                               virCgroupPtr cgroup,
3372 3373 3374
                               virDomainDiskDefPtr disk)
{
    return virDomainDiskDefForeachPath(disk,
3375
                                       driver->allowDiskFormatProbing,
3376 3377 3378
                                       true,
                                       qemuSetupDiskPathAllow,
                                       cgroup);
3379 3380 3381
}


3382 3383 3384 3385
static int qemuTeardownDiskPathDeny(virDomainDiskDefPtr disk ATTRIBUTE_UNUSED,
                                    const char *path,
                                    size_t depth ATTRIBUTE_UNUSED,
                                    void *opaque)
3386
{
3387 3388
    virCgroupPtr cgroup = opaque;
    int rc;
3389

3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403
    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;
3404
        }
3405
    }
3406 3407
    return 0;
}
3408 3409


3410 3411
static int qemuTeardownDiskCgroup(struct qemud_driver *driver,
                                  virCgroupPtr cgroup,
3412 3413 3414
                                  virDomainDiskDefPtr disk)
{
    return virDomainDiskDefForeachPath(disk,
3415
                                       driver->allowDiskFormatProbing,
3416 3417 3418
                                       true,
                                       qemuTeardownDiskPathDeny,
                                       cgroup);
3419 3420 3421
}


3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445
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;
}


3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464
static int qemuSetupHostUsbDeviceCgroup(usbDevice *dev ATTRIBUTE_UNUSED,
                                        const char *path,
                                        void *opaque)
{
    virCgroupPtr cgroup = opaque;
    int rc;

    VIR_DEBUG("Process path '%s' for USB device", path);
    rc = virCgroupAllowDevicePath(cgroup, path);
    if (rc != 0) {
        virReportSystemError(-rc,
                             _("Unable to allow device %s"),
                             path);
        return -1;
    }

    return 0;
}

3465
static int qemuSetupCgroup(struct qemud_driver *driver,
3466 3467 3468 3469
                           virDomainObjPtr vm)
{
    virCgroupPtr cgroup = NULL;
    int rc;
3470
    unsigned int i;
3471 3472 3473 3474
    const char *const *deviceACL =
        driver->cgroupDeviceACL ?
        (const char *const *)driver->cgroupDeviceACL :
        defaultDeviceACL;
3475 3476 3477 3478 3479 3480

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

    rc = virCgroupForDomain(driver->cgroup, vm->def->name, &cgroup, 1);
    if (rc != 0) {
3481
        virReportSystemError(-rc,
3482 3483 3484 3485 3486
                             _("Unable to create cgroup for %s"),
                             vm->def->name);
        goto cleanup;
    }

3487 3488
    if (qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_DEVICES)) {
        rc = virCgroupDenyAllDevices(cgroup);
3489
        if (rc != 0) {
3490 3491 3492 3493 3494
            if (rc == -EPERM) {
                VIR_WARN0("Group devices ACL is not accessible, disabling whitelisting");
                goto done;
            }

3495
            virReportSystemError(-rc,
3496
                                 _("Unable to deny all devices for %s"), vm->def->name);
3497 3498 3499
            goto cleanup;
        }

3500
        for (i = 0; i < vm->def->ndisks ; i++) {
3501
            if (qemuSetupDiskCgroup(driver, cgroup, vm->def->disks[i]) < 0)
3502 3503
                goto cleanup;
        }
3504

3505
        rc = virCgroupAllowDeviceMajor(cgroup, 'c', DEVICE_PTY_MAJOR);
3506
        if (rc != 0) {
3507
            virReportSystemError(-rc, "%s",
3508
                                 _("unable to allow /dev/pts/ devices"));
3509 3510 3511
            goto cleanup;
        }

3512 3513 3514
        if (vm->def->nsounds) {
            rc = virCgroupAllowDeviceMajor(cgroup, 'c', DEVICE_SND_MAJOR);
            if (rc != 0) {
3515
                virReportSystemError(-rc, "%s",
3516 3517 3518 3519 3520 3521 3522 3523 3524 3525
                                     _("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) {
3526
                virReportSystemError(-rc,
3527 3528 3529 3530
                                     _("unable to allow device %s"),
                                     deviceACL[i]);
                goto cleanup;
            }
3531
        }
3532 3533 3534 3535 3536 3537

        if (virDomainChrDefForeach(vm->def,
                                   true,
                                   qemuSetupChardevCgroup,
                                   cgroup) < 0)
            goto cleanup;
3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554

        for (i = 0; i < vm->def->nhostdevs; i++) {
            virDomainHostdevDefPtr hostdev = vm->def->hostdevs[i];
            usbDevice *usb;

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

            if ((usb = usbGetDevice(hostdev->source.subsys.u.usb.bus,
                                    hostdev->source.subsys.u.usb.device)) == NULL)
                goto cleanup;

            if (usbDeviceFileIterate(usb, qemuSetupHostUsbDeviceCgroup, cgroup) < 0 )
                goto cleanup;
        }
3555 3556
    }

3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590
    if ((rc = qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_MEMORY))) {
        if (vm->def->mem.hard_limit != 0) {
            rc = virCgroupSetMemoryHardLimit(cgroup, vm->def->mem.hard_limit);
            if (rc != 0) {
                virReportSystemError(-rc,
                                     _("Unable to set memory hard limit for domain %s"),
                                     vm->def->name);
                goto cleanup;
            }
        }
        if (vm->def->mem.soft_limit != 0) {
            rc = virCgroupSetMemorySoftLimit(cgroup, vm->def->mem.soft_limit);
            if (rc != 0) {
                virReportSystemError(-rc,
                                     _("Unable to set memory soft limit for domain %s"),
                                     vm->def->name);
                goto cleanup;
            }
        }

        if (vm->def->mem.swap_hard_limit != 0) {
            rc = virCgroupSetSwapHardLimit(cgroup, vm->def->mem.swap_hard_limit);
            if (rc != 0) {
                virReportSystemError(-rc,
                                     _("Unable to set swap hard limit for domain %s"),
                                     vm->def->name);
                goto cleanup;
            }
        }
    } else {
        VIR_WARN("Memory cgroup is disabled in qemu configuration file: %s",
                 vm->def->name);
    }

3591
done:
3592 3593 3594 3595 3596 3597 3598 3599 3600 3601 3602 3603
    virCgroupFree(&cgroup);
    return 0;

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


3604
static int qemuRemoveCgroup(struct qemud_driver *driver,
3605 3606
                            virDomainObjPtr vm,
                            int quiet)
3607 3608 3609 3610 3611 3612 3613 3614 3615
{
    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) {
3616
        if (!quiet)
3617 3618 3619
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("Unable to find cgroup for %s\n"),
                            vm->def->name);
3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639
        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) {
3640
        virReportSystemError(-rc,
3641 3642 3643 3644 3645 3646 3647
                             _("unable to find cgroup for domain %s"),
                             def->name);
        goto cleanup;
    }

    rc = virCgroupAddTask(cgroup, getpid());
    if (rc != 0) {
3648
        virReportSystemError(-rc,
3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665
                             _("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;
3666 3667 3668
};

static int qemudSecurityHook(void *data) {
3669 3670
    struct qemudHookData *h = data;

3671 3672 3673
    /* This must take place before exec(), so that all QEMU
     * memory allocation is on the correct NUMA node
     */
3674
    if (qemuAddToCgroup(h->driver, h->vm->def) < 0)
3675 3676
        return -1;

3677 3678 3679
    /* This must be done after cgroup placement to avoid resetting CPU
     * affinity */
    if (qemudInitCpuAffinity(h->vm) < 0)
3680
        return -1;
3681

3682 3683
    if (h->driver->securityDriver &&
        h->driver->securityDriver->domainSetSecurityProcessLabel &&
3684
        h->driver->securityDriver->domainSetSecurityProcessLabel(h->driver->securityDriver, h->vm) < 0)
3685 3686 3687
        return -1;

    return 0;
3688 3689
}

3690
static int
3691
qemuPrepareMonitorChr(struct qemud_driver *driver,
3692
                      virDomainChrDefPtr monConfig,
3693 3694
                      const char *vm)
{
3695
    monConfig->deviceType = VIR_DOMAIN_CHR_DEVICE_TYPE_MONITOR;
3696

3697 3698
    monConfig->type = VIR_DOMAIN_CHR_TYPE_UNIX;
    monConfig->data.nix.listen = 1;
3699

D
Daniel P. Berrange 已提交
3700
    if (!(monConfig->info.alias = strdup("monitor"))) {
3701
        virReportOOMError();
D
Daniel P. Berrange 已提交
3702 3703 3704
        return -1;
    }

3705
    if (virAsprintf(&monConfig->data.nix.path, "%s/%s.monitor",
3706
                    driver->libDir, vm) < 0) {
3707
        virReportOOMError();
3708 3709 3710 3711 3712 3713
        return -1;
    }

    return 0;
}

C
Chris Lalancette 已提交
3714 3715 3716 3717 3718
static int qemuDomainSnapshotSetActive(virDomainObjPtr vm,
                                       char *snapshotDir);
static int qemuDomainSnapshotSetInactive(virDomainObjPtr vm,
                                         char *snapshotDir);

3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791
static void qemuDomainDiskAudit(virDomainObjPtr vm,
                                virDomainDiskDefPtr oldDef,
                                virDomainDiskDefPtr newDef,
                                const char *reason,
                                bool success)
{
    char uuidstr[VIR_UUID_STRING_BUFLEN];
    char *vmname;
    char *oldsrc = NULL;
    char *newsrc = NULL;

    virUUIDFormat(vm->def->uuid, uuidstr);
    if (!(vmname = virAuditEncode("vm", vm->def->name))) {
        VIR_WARN0("OOM while encoding audit message");
        return;
    }

    if (!(oldsrc = virAuditEncode("old-disk",
                                  oldDef && oldDef->src ?
                                  oldDef->src : "?"))) {
        VIR_WARN0("OOM while encoding audit message");
        goto cleanup;
    }
    if (!(newsrc = virAuditEncode("new-disk",
                                  newDef && newDef->src ?
                                  newDef->src : "?"))) {
        VIR_WARN0("OOM while encoding audit message");
        goto cleanup;
    }

    VIR_AUDIT(VIR_AUDIT_RECORD_RESOURCE, success,
              "resrc=disk reason=%s %s uuid=%s %s %s",
              reason, vmname, uuidstr,
              oldsrc, newsrc);

cleanup:
    VIR_FREE(vmname);
    VIR_FREE(oldsrc);
    VIR_FREE(newsrc);
}


static void qemuDomainNetAudit(virDomainObjPtr vm,
                               virDomainNetDefPtr oldDef,
                               virDomainNetDefPtr newDef,
                               const char *reason,
                               bool success)
{
    char uuidstr[VIR_UUID_STRING_BUFLEN];
    char newMacstr[VIR_MAC_STRING_BUFLEN];
    char oldMacstr[VIR_MAC_STRING_BUFLEN];
    char *vmname;

    virUUIDFormat(vm->def->uuid, uuidstr);
    if (oldDef)
        virFormatMacAddr(oldDef->mac, oldMacstr);
    if (newDef)
        virFormatMacAddr(newDef->mac, newMacstr);
    if (!(vmname = virAuditEncode("vm", vm->def->name))) {
        VIR_WARN0("OOM while encoding audit message");
        return;
    }

    VIR_AUDIT(VIR_AUDIT_RECORD_RESOURCE, success,
              "resrc=net reason=%s %s uuid=%s old-net='%s' new-net='%s'",
              reason, vmname, uuidstr,
              oldDef ? oldMacstr : "?",
              newDef ? newMacstr : "?");

    VIR_FREE(vmname);
}


3792 3793 3794 3795 3796 3797 3798 3799 3800
static void qemuDomainLifecycleAudit(virDomainObjPtr vm,
                                     const char *op,
                                     const char *reason,
                                     bool success)
{
    char uuidstr[VIR_UUID_STRING_BUFLEN];
    char *vmname;

    virUUIDFormat(vm->def->uuid, uuidstr);
3801

3802 3803 3804 3805 3806 3807 3808 3809 3810 3811 3812 3813 3814
    if (!(vmname = virAuditEncode("vm", vm->def->name))) {
        VIR_WARN0("OOM while encoding audit message");
        return;
    }

    VIR_AUDIT(VIR_AUDIT_RECORD_MACHINE_CONTROL, success,
              "op=%s reason=%s %s uuid=%s", op, reason, vmname, uuidstr);

    VIR_FREE(vmname);
}

static void qemuDomainStartAudit(virDomainObjPtr vm, const char *reason, bool success)
{
3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827
    int i;

    for (i = 0 ; i < vm->def->ndisks ; i++) {
        virDomainDiskDefPtr disk = vm->def->disks[i];
        if (disk->src) /* Skips CDROM without media initially inserted */
            qemuDomainDiskAudit(vm, NULL, disk, "start", true);
    }

    for (i = 0 ; i < vm->def->nnets ; i++) {
        virDomainNetDefPtr net = vm->def->nets[i];
        qemuDomainNetAudit(vm, NULL, net, "start", true);
    }

3828 3829 3830 3831 3832 3833 3834 3835
    qemuDomainLifecycleAudit(vm, "start", reason, success);
}

static void qemuDomainStopAudit(virDomainObjPtr vm, const char *reason)
{
    qemuDomainLifecycleAudit(vm, "stop", reason, true);
}

3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856
static void qemuDomainSecurityLabelAudit(virDomainObjPtr vm, bool success)
{
    char uuidstr[VIR_UUID_STRING_BUFLEN];
    char *vmname;

    virUUIDFormat(vm->def->uuid, uuidstr);
    if (!(vmname = virAuditEncode("vm", vm->def->name))) {
        VIR_WARN0("OOM while encoding audit message");
        return;
    }

    VIR_AUDIT(VIR_AUDIT_RECORD_MACHINE_ID, success,
              "%s uuid=%s vm-ctx=%s img-ctx=%s",
              vmname, uuidstr,
              VIR_AUDIT_STR(vm->def->seclabel.label),
              VIR_AUDIT_STR(vm->def->seclabel.imagelabel));

    VIR_FREE(vmname);
}


3857 3858
static int qemudStartVMDaemon(virConnectPtr conn,
                              struct qemud_driver *driver,
3859
                              virDomainObjPtr vm,
3860
                              const char *migrateFrom,
3861
                              bool start_paused,
3862 3863
                              int stdin_fd,
                              const char *stdin_path) {
3864
    const char **argv = NULL, **tmp;
3865
    const char **progenv = NULL;
3866
    int i, ret, runflags;
3867
    struct stat sb;
3868 3869
    int *vmfds = NULL;
    int nvmfds = 0;
3870
    unsigned long long qemuCmdFlags;
3871
    fd_set keepfd;
3872
    const char *emulator;
G
Guido Günther 已提交
3873
    pid_t child;
3874
    int pos = -1;
3875
    char ebuf[1024];
3876
    char *pidfile = NULL;
3877
    int logfile = -1;
3878
    qemuDomainObjPrivatePtr priv = vm->privateData;
3879

3880
    struct qemudHookData hookData;
3881 3882 3883 3884
    hookData.conn = conn;
    hookData.vm = vm;
    hookData.driver = driver;

3885
    FD_ZERO(&keepfd);
3886

3887 3888
    DEBUG0("Beginning VM startup process");

D
Daniel P. Berrange 已提交
3889
    if (virDomainObjIsActive(vm)) {
3890 3891
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("VM is already active"));
3892 3893 3894
        return -1;
    }

3895 3896 3897 3898 3899
    /* Must be run before security labelling */
    DEBUG0("Preparing host devices");
    if (qemuPrepareHostDevices(driver, vm->def) < 0)
        goto cleanup;

3900 3901 3902 3903 3904 3905 3906
    DEBUG0("Preparing chr devices");
    if (virDomainChrDefForeach(vm->def,
                               true,
                               qemuPrepareChardevDevice,
                               NULL) < 0)
        goto cleanup;

3907 3908
    /* If you are using a SecurityDriver with dynamic labelling,
       then generate a security label for isolation */
3909
    DEBUG0("Generating domain security label (if required)");
3910
    if (driver->securityDriver &&
3911 3912 3913 3914 3915 3916 3917
        driver->securityDriver->domainGenSecurityLabel) {
        ret = driver->securityDriver->domainGenSecurityLabel(driver->securityDriver,
                                                             vm);
        qemuDomainSecurityLabelAudit(vm, ret >= 0);
        if (ret < 0)
            goto cleanup;
    }
3918

3919
    DEBUG0("Generating setting domain security labels (if required)");
3920 3921
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityAllLabel &&
3922 3923
        driver->securityDriver->domainSetSecurityAllLabel(driver->securityDriver,
                                                          vm, stdin_path) < 0) {
3924
        goto cleanup;
3925
    }
3926

3927 3928 3929
    /* Ensure no historical cgroup for this VM is lying around bogus
     * settings */
    DEBUG0("Ensuring no historical cgroup is lying around");
3930
    qemuRemoveCgroup(driver, vm, 1);
3931

3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944
    if (vm->def->ngraphics == 1) {
        if (vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
            vm->def->graphics[0]->data.vnc.autoport) {
            int port = qemudNextFreePort(driver, 5900);
            if (port < 0) {
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                "%s", _("Unable to find an unused VNC port"));
                goto cleanup;
            }
            vm->def->graphics[0]->data.vnc.port = port;
        } else if (vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_SPICE &&
                   vm->def->graphics[0]->data.spice.autoport) {
            int port = qemudNextFreePort(driver, 5900);
3945 3946
            int tlsPort = -1;
            if (port < 0) {
3947
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
3948
                                "%s", _("Unable to find an unused SPICE port"));
3949 3950 3951
                goto cleanup;
            }

3952 3953 3954 3955 3956 3957 3958 3959 3960
            if (driver->spiceTLS) {
                tlsPort = qemudNextFreePort(driver, port + 1);
                if (tlsPort < 0) {
                    qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                    "%s", _("Unable to find an unused SPICE TLS port"));
                    goto cleanup;
                }
            }

3961 3962
            vm->def->graphics[0]->data.spice.port = port;
            vm->def->graphics[0]->data.spice.tlsPort = tlsPort;
3963
        }
3964
    }
3965

L
Laine Stump 已提交
3966
    if (virFileMakePath(driver->logDir) != 0) {
3967
        virReportSystemError(errno,
3968 3969
                             _("cannot create log directory %s"),
                             driver->logDir);
3970
        goto cleanup;
3971 3972
    }

3973
    DEBUG0("Creating domain log file");
3974
    if ((logfile = qemudLogFD(driver, vm->def->name)) < 0)
3975
        goto cleanup;
3976

3977 3978
    emulator = vm->def->emulator;

3979 3980 3981 3982
    /* 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
     */
3983
    if (stat(emulator, &sb) < 0) {
3984
        virReportSystemError(errno,
3985 3986
                             _("Cannot find QEMU binary %s"),
                             emulator);
3987
        goto cleanup;
3988 3989
    }

3990
    DEBUG0("Determing emulator version");
3991
    if (qemudExtractVersionInfo(emulator,
3992
                                NULL,
3993
                                &qemuCmdFlags) < 0)
3994
        goto cleanup;
3995

3996
    DEBUG0("Setting up domain cgroup (if required)");
3997
    if (qemuSetupCgroup(driver, vm) < 0)
3998 3999
        goto cleanup;

4000
    if (VIR_ALLOC(priv->monConfig) < 0) {
4001
        virReportOOMError();
4002 4003 4004
        goto cleanup;
    }

4005
    DEBUG0("Preparing monitor state");
4006
    if (qemuPrepareMonitorChr(driver, priv->monConfig, vm->def->name) < 0)
4007
        goto cleanup;
4008

D
Daniel P. Berrange 已提交
4009 4010 4011
#if HAVE_YAJL
    if (qemuCmdFlags & QEMUD_CMD_FLAG_MONITOR_JSON)
        priv->monJSON = 1;
4012
    else
D
Daniel P. Berrange 已提交
4013
#endif
4014
        priv->monJSON = 0;
D
Daniel P. Berrange 已提交
4015

4016 4017
    priv->monitor_warned = 0;

D
Daniel P. Berrange 已提交
4018
    if ((ret = virFileDeletePid(driver->stateDir, vm->def->name)) != 0) {
4019
        virReportSystemError(ret,
D
Daniel P. Berrange 已提交
4020 4021 4022 4023 4024
                             _("Cannot remove stale PID file for %s"),
                             vm->def->name);
        goto cleanup;
    }

4025
    if (!(pidfile = virFilePid(driver->stateDir, vm->def->name))) {
4026
        virReportSystemError(errno,
4027 4028 4029
                             "%s", _("Failed to build pidfile path."));
        goto cleanup;
    }
D
Daniel P. Berrange 已提交
4030

4031
    /*
M
Matthew Booth 已提交
4032
     * Normally PCI addresses are assigned in the virDomainCreate
4033 4034 4035 4036 4037
     * 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
     */
4038
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
4039
        DEBUG0("Assigning domain PCI addresses");
4040
        /* Populate cache with current addresses */
4041 4042 4043 4044 4045 4046 4047
        if (priv->pciaddrs) {
            qemuDomainPCIAddressSetFree(priv->pciaddrs);
            priv->pciaddrs = NULL;
        }
        if (!(priv->pciaddrs = qemuDomainPCIAddressSetCreate(vm->def)))
            goto cleanup;

4048 4049

        /* Assign any remaining addresses */
4050 4051
        if (qemuAssignDevicePCISlots(vm->def, priv->pciaddrs) < 0)
            goto cleanup;
4052 4053 4054 4055

        priv->persistentAddrs = 1;
    } else {
        priv->persistentAddrs = 0;
4056 4057
    }

4058
    DEBUG0("Building emulator command line");
4059
    vm->def->id = driver->nextvmid++;
4060
    if (qemudBuildCommandLine(conn, driver, vm->def, priv->monConfig,
D
Daniel P. Berrange 已提交
4061
                              priv->monJSON, qemuCmdFlags, &argv, &progenv,
4062
                              &vmfds, &nvmfds, migrateFrom,
C
Chris Lalancette 已提交
4063 4064 4065 4066
                              vm->current_snapshot) < 0)
        goto cleanup;

    if (qemuDomainSnapshotSetInactive(vm, driver->snapshotDir) < 0)
4067
        goto cleanup;
4068

4069 4070 4071 4072 4073 4074 4075 4076 4077 4078 4079 4080 4081 4082 4083 4084
    /* 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;
    }

4085 4086
    tmp = progenv;
    while (*tmp) {
4087
        if (safewrite(logfile, *tmp, strlen(*tmp)) < 0)
4088
            VIR_WARN("Unable to write envv to logfile: %s",
4089
                     virStrerror(errno, ebuf, sizeof ebuf));
4090
        if (safewrite(logfile, " ", 1) < 0)
4091
            VIR_WARN("Unable to write envv to logfile: %s",
4092
                     virStrerror(errno, ebuf, sizeof ebuf));
4093 4094
        tmp++;
    }
4095 4096
    tmp = argv;
    while (*tmp) {
4097
        if (safewrite(logfile, *tmp, strlen(*tmp)) < 0)
4098
            VIR_WARN("Unable to write argv to logfile: %s",
4099
                     virStrerror(errno, ebuf, sizeof ebuf));
4100
        if (safewrite(logfile, " ", 1) < 0)
4101
            VIR_WARN("Unable to write argv to logfile: %s",
4102
                     virStrerror(errno, ebuf, sizeof ebuf));
4103 4104
        tmp++;
    }
4105
    if (safewrite(logfile, "\n", 1) < 0)
4106
        VIR_WARN("Unable to write argv to logfile: %s",
4107
                 virStrerror(errno, ebuf, sizeof ebuf));
4108

4109
    if ((pos = lseek(logfile, 0, SEEK_END)) < 0)
4110
        VIR_WARN("Unable to seek to end of logfile: %s",
4111
                 virStrerror(errno, ebuf, sizeof ebuf));
4112

4113 4114
    for (i = 0 ; i < nvmfds ; i++)
        FD_SET(vmfds[i], &keepfd);
4115

4116 4117 4118 4119 4120 4121 4122
    VIR_DEBUG("Clear emulator capabilities: %d",
              driver->clearEmulatorCapabilities);
    runflags = VIR_EXEC_NONBLOCK;
    if (driver->clearEmulatorCapabilities) {
        runflags |= VIR_EXEC_CLEAR_CAPS;
    }

4123
    ret = virExecDaemonize(argv, progenv, &keepfd, &child,
4124
                           stdin_fd, &logfile, &logfile,
4125
                           runflags,
4126 4127 4128
                           qemudSecurityHook, &hookData,
                           pidfile);
    VIR_FREE(pidfile);
G
Guido Günther 已提交
4129 4130 4131

    /* wait for qemu process to to show up */
    if (ret == 0) {
4132
        if (virFileReadPid(driver->stateDir, vm->def->name, &vm->pid)) {
4133 4134
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("Domain %s didn't show up\n"), vm->def->name);
4135
            ret = -1;
G
Guido Günther 已提交
4136
        }
4137 4138 4139 4140 4141 4142 4143 4144 4145 4146 4147
    } 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;
    }
4148

4149 4150 4151
    if (migrateFrom)
        start_paused = true;
    vm->state = start_paused ? VIR_DOMAIN_PAUSED : VIR_DOMAIN_RUNNING;
4152

4153
    for (i = 0 ; argv[i] ; i++)
4154 4155
        VIR_FREE(argv[i]);
    VIR_FREE(argv);
4156

4157 4158 4159 4160
    for (i = 0 ; progenv[i] ; i++)
        VIR_FREE(progenv[i]);
    VIR_FREE(progenv);

4161
    if (ret == -1) /* The VM failed to start; tear filters before taps */
4162
        virDomainConfVMNWFilterTeardown(vm);
4163

4164 4165 4166
    if (vmfds) {
        for (i = 0 ; i < nvmfds ; i++) {
            close(vmfds[i]);
4167
        }
4168
        VIR_FREE(vmfds);
4169 4170
    }

4171
    if (ret == -1) /* The VM failed to start */
4172 4173
        goto cleanup;

4174
    DEBUG0("Waiting for monitor to show up");
4175
    if (qemudWaitForMonitor(driver, vm, pos) < 0)
4176
        goto cleanup;
4177

4178
    DEBUG0("Detecting VCPU PIDs");
4179
    if (qemuDetectVcpuPIDs(driver, vm) < 0)
4180
        goto cleanup;
4181

4182
    DEBUG0("Setting any required VM passwords");
4183
    if (qemuInitPasswords(conn, driver, vm, qemuCmdFlags) < 0)
4184
        goto cleanup;
4185

D
Daniel P. Berrange 已提交
4186 4187 4188
    /* 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)) {
4189
        DEBUG0("Determining domain device PCI addresses");
D
Daniel P. Berrange 已提交
4190
        if (qemuInitPCIAddresses(driver, vm) < 0)
4191
            goto cleanup;
D
Daniel P. Berrange 已提交
4192
    }
4193

4194
    DEBUG0("Setting initial memory amount");
4195
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
4196
    if (qemuMonitorSetBalloon(priv->mon, vm->def->mem.cur_balloon) < 0) {
4197
        qemuDomainObjExitMonitorWithDriver(driver, vm);
4198
        goto cleanup;
4199
    }
4200

4201
    if (!start_paused) {
4202
        DEBUG0("Starting domain CPUs");
4203 4204 4205
        /* Allow the CPUS to start executing */
        if (qemuMonitorStartCPUs(priv->mon, conn) < 0) {
            if (virGetLastError() == NULL)
4206 4207
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                "%s", _("resume operation failed"));
4208
            qemuDomainObjExitMonitorWithDriver(driver, vm);
4209
            goto cleanup;
4210 4211
        }
    }
4212
    qemuDomainObjExitMonitorWithDriver(driver, vm);
4213

4214

4215
    DEBUG0("Writing domain status to disk");
4216
    if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
4217
        goto cleanup;
4218

4219 4220 4221
    if (logfile != -1)
        close(logfile);

4222
    return 0;
4223 4224

cleanup:
4225 4226 4227
    /* 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 */
4228
    qemudShutdownVMDaemon(driver, vm, 0);
4229 4230 4231 4232 4233

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

    return -1;
4234 4235 4236
}


4237
static void qemudShutdownVMDaemon(struct qemud_driver *driver,
4238 4239
                                  virDomainObjPtr vm,
                                  int migrated) {
D
Daniel P. Berrange 已提交
4240
    int ret;
4241
    int retries = 0;
4242
    qemuDomainObjPrivatePtr priv = vm->privateData;
4243
    virErrorPtr orig_err;
4244 4245
    virDomainDefPtr def;
    int i;
D
Daniel P. Berrange 已提交
4246

4247 4248
    VIR_DEBUG("Shutting down VM '%s' pid=%d migrated=%d",
              vm->def->name, vm->pid, migrated);
4249

4250 4251 4252 4253
    /* This method is routinely used in clean up paths. Disable error
     * reporting so we don't squash a legit error. */
    orig_err = virSaveLastError();

4254
    virDomainConfVMNWFilterTeardown(vm);
S
Stefan Berger 已提交
4255

4256
    if (driver->macFilter) {
4257
        def = vm->def;
4258 4259 4260 4261
        for (i = 0 ; i < def->nnets ; i++) {
            virDomainNetDefPtr net = def->nets[i];
            if (net->ifname == NULL)
                continue;
4262
            if ((errno = networkDisallowMacOnPort(driver, net->ifname,
4263
                                                  net->mac))) {
4264
                virReportSystemError(errno,
4265 4266 4267 4268 4269 4270
             _("failed to remove ebtables rule to allow MAC address on  '%s'"),
                                     net->ifname);
            }
        }
    }

4271
    /* This will safely handle a non-running guest with pid=0 or pid=-1*/
G
Guido Günther 已提交
4272 4273
    if (virKillProcess(vm->pid, 0) == 0 &&
        virKillProcess(vm->pid, SIGTERM) < 0)
4274
        virReportSystemError(errno,
4275 4276
                             _("Failed to send SIGTERM to %s (%d)"),
                             vm->def->name, vm->pid);
4277

4278 4279
    if (priv->mon)
        qemuMonitorClose(priv->mon);
4280

4281 4282 4283 4284 4285
    if (priv->monConfig) {
        if (priv->monConfig->type == VIR_DOMAIN_CHR_TYPE_UNIX)
            unlink(priv->monConfig->data.nix.path);
        virDomainChrDefFree(priv->monConfig);
        priv->monConfig = NULL;
4286 4287
    }

G
Guido Günther 已提交
4288 4289
    /* shut it off for sure */
    virKillProcess(vm->pid, SIGKILL);
4290

4291 4292 4293 4294 4295 4296 4297 4298 4299 4300
    /* 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);
    }

4301
    /* Reset Security Labels */
4302
    if (driver->securityDriver &&
4303
        driver->securityDriver->domainRestoreSecurityAllLabel)
4304 4305
        driver->securityDriver->domainRestoreSecurityAllLabel(driver->securityDriver,
                                                              vm, migrated);
4306 4307
    if (driver->securityDriver &&
        driver->securityDriver->domainReleaseSecurityLabel)
4308 4309
        driver->securityDriver->domainReleaseSecurityLabel(driver->securityDriver,
                                                           vm);
4310

4311 4312 4313 4314 4315 4316 4317
    /* 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
4318
    virDomainDefClearDeviceAliases(vm->def);
4319 4320 4321 4322 4323
    if (!priv->persistentAddrs) {
        virDomainDefClearPCIAddresses(vm->def);
        qemuDomainPCIAddressSetFree(priv->pciaddrs);
        priv->pciaddrs = NULL;
    }
4324

4325
    qemuDomainReAttachHostDevices(driver, vm->def);
4326

S
Stefan Berger 已提交
4327 4328 4329 4330
#if WITH_MACVTAP
    def = vm->def;
    for (i = 0; i < def->nnets; i++) {
        virDomainNetDefPtr net = def->nets[i];
4331
        if (net->type == VIR_DOMAIN_NET_TYPE_DIRECT) {
4332
            delMacvtap(net->ifname, net->mac, net->data.direct.linkdev,
4333
                       &net->data.direct.virtPortProfile);
4334 4335
            VIR_FREE(net->ifname);
        }
S
Stefan Berger 已提交
4336 4337 4338
    }
#endif

4339
retry:
4340
    if ((ret = qemuRemoveCgroup(driver, vm, 0)) < 0) {
4341 4342 4343 4344 4345 4346 4347 4348
        if (ret == -EBUSY && (retries++ < 5)) {
            usleep(200*1000);
            goto retry;
        }
        VIR_WARN("Failed to remove cgroup for %s",
                 vm->def->name);
    }

4349
    qemudRemoveDomainStatus(driver, vm);
D
Daniel P. Berrange 已提交
4350

4351 4352 4353 4354 4355 4356
    /* 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 &&
4357
        vm->def->graphics[0]->data.vnc.port >= QEMU_VNC_PORT_MIN) {
4358 4359 4360 4361 4362 4363 4364 4365
        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);
       }
    }

4366
    vm->pid = -1;
4367
    vm->def->id = -1;
4368
    vm->state = VIR_DOMAIN_SHUTOFF;
4369 4370
    VIR_FREE(priv->vcpupids);
    priv->nvcpupids = 0;
4371 4372

    if (vm->newDef) {
4373
        virDomainDefFree(vm->def);
4374
        vm->def = vm->newDef;
4375
        vm->def->id = -1;
4376 4377
        vm->newDef = NULL;
    }
4378 4379 4380 4381 4382

    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
4383 4384 4385
}


4386
static virDrvOpenStatus qemudOpen(virConnectPtr conn,
4387
                                  virConnectAuthPtr auth ATTRIBUTE_UNUSED,
4388
                                  int flags ATTRIBUTE_UNUSED) {
4389
    if (conn->uri == NULL) {
4390 4391 4392
        if (qemu_driver == NULL)
            return VIR_DRV_OPEN_DECLINED;

4393
        conn->uri = xmlParseURI(qemu_driver->privileged ?
4394 4395
                                "qemu:///system" :
                                "qemu:///session");
4396
        if (!conn->uri) {
4397
            virReportOOMError();
4398 4399
            return VIR_DRV_OPEN_ERROR;
        }
4400 4401 4402 4403 4404 4405 4406 4407 4408 4409
    } 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;

4410
        if (qemu_driver == NULL) {
4411 4412
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("qemu state driver is not active"));
4413 4414 4415
            return VIR_DRV_OPEN_ERROR;
        }

4416
        if (conn->uri->path == NULL) {
4417 4418 4419 4420 4421
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("no QEMU URI path given, try %s"),
                            qemu_driver->privileged
                            ? "qemu:///system"
                            : "qemu:///session");
4422 4423 4424
                return VIR_DRV_OPEN_ERROR;
        }

4425
        if (qemu_driver->privileged) {
4426 4427
            if (STRNEQ (conn->uri->path, "/system") &&
                STRNEQ (conn->uri->path, "/session")) {
4428 4429 4430
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("unexpected QEMU URI path '%s', try qemu:///system"),
                                conn->uri->path);
4431 4432 4433 4434
                return VIR_DRV_OPEN_ERROR;
            }
        } else {
            if (STRNEQ (conn->uri->path, "/session")) {
4435 4436 4437
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("unexpected QEMU URI path '%s', try qemu:///session"),
                                conn->uri->path);
4438 4439 4440
                return VIR_DRV_OPEN_ERROR;
            }
        }
4441 4442 4443 4444 4445 4446 4447
    }
    conn->privateData = qemu_driver;

    return VIR_DRV_OPEN_SUCCESS;
}

static int qemudClose(virConnectPtr conn) {
4448
    struct qemud_driver *driver = conn->privateData;
4449 4450

    /* Get rid of callbacks registered for this conn */
4451
    qemuDriverLock(driver);
4452
    virDomainEventCallbackListRemoveConn(conn, driver->domainEventCallbacks);
4453
    qemuDriverUnlock(driver);
4454 4455 4456 4457 4458 4459

    conn->privateData = NULL;

    return 0;
}

D
Daniel Veillard 已提交
4460 4461 4462 4463 4464
/* Which features are supported by this driver? */
static int
qemudSupportsFeature (virConnectPtr conn ATTRIBUTE_UNUSED, int feature)
{
    switch (feature) {
4465 4466 4467 4468 4469
    case VIR_DRV_FEATURE_MIGRATION_V2:
    case VIR_DRV_FEATURE_MIGRATION_P2P:
        return 1;
    default:
        return 0;
D
Daniel Veillard 已提交
4470 4471 4472
    }
}

4473
static const char *qemudGetType(virConnectPtr conn ATTRIBUTE_UNUSED) {
4474
    return "QEMU";
4475 4476
}

4477

4478 4479 4480 4481 4482 4483 4484 4485 4486 4487 4488 4489 4490
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;
}


4491 4492 4493 4494
static int kvmGetMaxVCPUs(void) {
    int maxvcpus = 1;

    int r, fd;
4495

4496 4497
    fd = open(KVM_DEVICE, O_RDONLY);
    if (fd < 0) {
4498
        virReportSystemError(errno, _("Unable to open %s"), KVM_DEVICE);
4499
        return -1;
4500 4501 4502 4503 4504 4505 4506 4507 4508 4509 4510
    }

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

    close(fd);
    return maxvcpus;
}


4511
static int qemudGetMaxVCPUs(virConnectPtr conn ATTRIBUTE_UNUSED, const char *type) {
4512 4513 4514
    if (!type)
        return 16;

4515
    if (STRCASEEQ(type, "qemu"))
4516 4517
        return 16;

4518
    if (STRCASEEQ(type, "kvm"))
4519
        return kvmGetMaxVCPUs();
4520

4521
    if (STRCASEEQ(type, "kqemu"))
4522
        return 1;
4523

4524 4525
    qemuReportError(VIR_ERR_INVALID_ARG,
                    _("unknown type '%s'"), type);
4526 4527 4528
    return -1;
}

4529

4530
static char *qemudGetCapabilities(virConnectPtr conn) {
4531
    struct qemud_driver *driver = conn->privateData;
4532
    virCapsPtr caps = NULL;
4533
    char *xml = NULL;
4534

4535
    qemuDriverLock(driver);
4536

4537
    if ((caps = qemuCreateCapabilities(qemu_driver->caps,
4538
                                       qemu_driver)) == NULL) {
4539 4540 4541
        virCapabilitiesFree(caps);
        goto cleanup;
    }
4542

4543
    virCapabilitiesFree(qemu_driver->caps);
4544 4545 4546
    qemu_driver->caps = caps;

    if ((xml = virCapabilitiesFormatXML(driver->caps)) == NULL)
4547
        virReportOOMError();
4548 4549

cleanup:
4550
    qemuDriverUnlock(driver);
4551

4552
    return xml;
4553 4554 4555
}


4556
static int qemudGetProcessInfo(unsigned long long *cpuTime, int *lastCpu, int pid, int tid) {
D
Daniel P. Berrange 已提交
4557 4558
    char proc[PATH_MAX];
    FILE *pidinfo;
4559
    unsigned long long usertime, systime;
4560 4561
    int cpu;
    int ret;
D
Daniel P. Berrange 已提交
4562

4563 4564 4565 4566 4567 4568
    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
4569 4570 4571 4572 4573
        return -1;
    }

    if (!(pidinfo = fopen(proc, "r"))) {
        /* VM probably shut down, so fake 0 */
4574 4575 4576 4577
        if (cpuTime)
            *cpuTime = 0;
        if (lastCpu)
            *lastCpu = 0;
D
Daniel P. Berrange 已提交
4578 4579 4580
        return 0;
    }

4581 4582 4583 4584 4585 4586 4587 4588 4589 4590
    /* 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) {
4591
        fclose(pidinfo);
4592 4593
        VIR_WARN0("cannot parse process status data");
        errno = -EINVAL;
D
Daniel P. Berrange 已提交
4594 4595 4596 4597 4598 4599 4600 4601
        return -1;
    }

    /* We got jiffies
     * We want nanoseconds
     * _SC_CLK_TCK is jiffies per second
     * So calulate thus....
     */
4602 4603 4604 4605 4606
    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
4607

4608 4609
    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
4610 4611 4612 4613 4614 4615 4616

    fclose(pidinfo);

    return 0;
}


4617
static virDomainPtr qemudDomainLookupByID(virConnectPtr conn,
4618
                                          int id) {
4619 4620 4621 4622
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;

4623
    qemuDriverLock(driver);
4624
    vm  = virDomainFindByID(&driver->domains, id);
4625
    qemuDriverUnlock(driver);
4626 4627

    if (!vm) {
4628 4629
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching id %d"), id);
4630
        goto cleanup;
4631 4632
    }

4633
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
4634
    if (dom) dom->id = vm->def->id;
4635 4636

cleanup:
4637 4638
    if (vm)
        virDomainObjUnlock(vm);
4639 4640
    return dom;
}
4641

4642
static virDomainPtr qemudDomainLookupByUUID(virConnectPtr conn,
4643
                                            const unsigned char *uuid) {
4644 4645 4646
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
4647

4648
    qemuDriverLock(driver);
4649
    vm = virDomainFindByUUID(&driver->domains, uuid);
4650 4651
    qemuDriverUnlock(driver);

4652
    if (!vm) {
4653 4654
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(uuid, uuidstr);
4655 4656
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4657
        goto cleanup;
4658 4659
    }

4660
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
4661
    if (dom) dom->id = vm->def->id;
4662 4663

cleanup:
4664 4665
    if (vm)
        virDomainObjUnlock(vm);
4666 4667
    return dom;
}
4668

4669
static virDomainPtr qemudDomainLookupByName(virConnectPtr conn,
4670
                                            const char *name) {
4671 4672 4673
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
4674

4675
    qemuDriverLock(driver);
4676
    vm = virDomainFindByName(&driver->domains, name);
4677 4678
    qemuDriverUnlock(driver);

4679
    if (!vm) {
4680 4681
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching name '%s'"), name);
4682
        goto cleanup;
4683 4684
    }

4685
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
4686
    if (dom) dom->id = vm->def->id;
4687 4688

cleanup:
4689 4690
    if (vm)
        virDomainObjUnlock(vm);
4691 4692 4693
    return dom;
}

4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704

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) {
4705
        qemuReportError(VIR_ERR_NO_DOMAIN, NULL);
4706 4707 4708 4709 4710 4711 4712 4713 4714 4715 4716 4717 4718 4719 4720 4721 4722 4723 4724 4725
        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) {
4726
        qemuReportError(VIR_ERR_NO_DOMAIN, NULL);
4727 4728 4729 4730 4731 4732 4733 4734 4735 4736 4737
        goto cleanup;
    }
    ret = obj->persistent;

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


4738
static int qemudGetVersion(virConnectPtr conn, unsigned long *version) {
4739 4740 4741
    struct qemud_driver *driver = conn->privateData;
    int ret = -1;

4742
    qemuDriverLock(driver);
4743
    if (qemudExtractVersion(driver) < 0)
4744
        goto cleanup;
4745

4746
    *version = qemu_driver->qemuVersion;
4747 4748 4749
    ret = 0;

cleanup:
4750
    qemuDriverUnlock(driver);
4751
    return ret;
D
Daniel P. Berrange 已提交
4752 4753
}

4754
static int qemudListDomains(virConnectPtr conn, int *ids, int nids) {
4755
    struct qemud_driver *driver = conn->privateData;
4756
    int n;
4757

4758
    qemuDriverLock(driver);
4759
    n = virDomainObjListGetActiveIDs(&driver->domains, ids, nids);
4760
    qemuDriverUnlock(driver);
4761

4762
    return n;
D
Daniel P. Berrange 已提交
4763
}
4764

4765
static int qemudNumDomains(virConnectPtr conn) {
4766
    struct qemud_driver *driver = conn->privateData;
4767
    int n;
4768

4769
    qemuDriverLock(driver);
4770
    n = virDomainObjListNumOfDomains(&driver->domains, 1);
4771
    qemuDriverUnlock(driver);
4772

4773
    return n;
D
Daniel P. Berrange 已提交
4774
}
4775

4776
static virDomainPtr qemudDomainCreate(virConnectPtr conn, const char *xml,
4777
                                      unsigned int flags) {
4778
    struct qemud_driver *driver = conn->privateData;
4779
    virDomainDefPtr def;
4780
    virDomainObjPtr vm = NULL;
4781
    virDomainPtr dom = NULL;
4782
    virDomainEventPtr event = NULL;
D
Daniel P. Berrange 已提交
4783

4784
    virCheckFlags(VIR_DOMAIN_START_PAUSED, NULL);
4785

4786
    qemuDriverLock(driver);
4787
    if (!(def = virDomainDefParseString(driver->caps, xml,
4788
                                        VIR_DOMAIN_XML_INACTIVE)))
4789
        goto cleanup;
4790

4791
    if (virSecurityDriverVerify(def) < 0)
4792 4793
        goto cleanup;

4794 4795
    if (virDomainObjIsDuplicate(&driver->domains, def, 1) < 0)
        goto cleanup;
4796

4797 4798 4799 4800 4801 4802
    if (qemudCanonicalizeMachine(driver, def) < 0)
        goto cleanup;

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

4803
    if (!(vm = virDomainAssignDef(driver->caps,
4804
                                  &driver->domains,
4805
                                  def, false)))
4806 4807 4808
        goto cleanup;

    def = NULL;
D
Daniel P. Berrange 已提交
4809

4810 4811 4812
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup; /* XXXX free the 'vm' we created ? */

4813 4814 4815
    if (qemudStartVMDaemon(conn, driver, vm, NULL,
                           (flags & VIR_DOMAIN_START_PAUSED) != 0,
                           -1, NULL) < 0) {
4816
        qemuDomainStartAudit(vm, "booted", false);
4817 4818 4819
        if (qemuDomainObjEndJob(vm) > 0)
            virDomainRemoveInactive(&driver->domains,
                                    vm);
4820
        vm = NULL;
4821
        goto cleanup;
D
Daniel P. Berrange 已提交
4822
    }
4823 4824 4825 4826

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_BOOTED);
4827
    qemuDomainStartAudit(vm, "booted", true);
D
Daniel P. Berrange 已提交
4828

4829
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
4830
    if (dom) dom->id = vm->def->id;
4831

4832 4833 4834
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4835

4836 4837
cleanup:
    virDomainDefFree(def);
4838 4839
    if (vm)
        virDomainObjUnlock(vm);
4840 4841
    if (event)
        qemuDomainEventQueue(driver, event);
4842
    qemuDriverUnlock(driver);
4843
    return dom;
D
Daniel P. Berrange 已提交
4844 4845 4846
}


4847
static int qemudDomainSuspend(virDomainPtr dom) {
4848 4849 4850
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4851
    virDomainEventPtr event = NULL;
4852
    qemuDomainObjPrivatePtr priv;
4853

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

D
Daniel P. Berrange 已提交
4857
    if (!vm) {
4858 4859
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4860 4861
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4862
        goto cleanup;
D
Daniel P. Berrange 已提交
4863
    }
D
Daniel P. Berrange 已提交
4864
    if (!virDomainObjIsActive(vm)) {
4865 4866
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
4867
        goto cleanup;
D
Daniel P. Berrange 已提交
4868
    }
4869 4870 4871

    priv = vm->privateData;

4872
    if (priv->jobActive == QEMU_JOB_MIGRATION_OUT) {
4873 4874 4875 4876 4877 4878 4879 4880 4881 4882 4883 4884 4885 4886
        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"));
4887
            goto endjob;
4888
        }
4889 4890
        if (vm->state != VIR_DOMAIN_PAUSED) {
            int rc;
4891
            int state = vm->state;
4892

4893
            vm->state = VIR_DOMAIN_PAUSED;
4894 4895 4896
            qemuDomainObjEnterMonitorWithDriver(driver, vm);
            rc = qemuMonitorStopCPUs(priv->mon);
            qemuDomainObjExitMonitorWithDriver(driver, vm);
4897 4898
            if (rc < 0) {
                vm->state = state;
4899
                goto endjob;
4900
            }
4901 4902 4903 4904 4905 4906 4907
            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
4908
    }
4909

4910
endjob:
4911 4912
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4913

4914
cleanup:
4915 4916
    if (vm)
        virDomainObjUnlock(vm);
4917

4918
    if (event)
4919
        qemuDomainEventQueue(driver, event);
4920
    qemuDriverUnlock(driver);
4921
    return ret;
D
Daniel P. Berrange 已提交
4922 4923 4924
}


4925
static int qemudDomainResume(virDomainPtr dom) {
4926 4927 4928
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4929
    virDomainEventPtr event = NULL;
4930

4931
    qemuDriverLock(driver);
4932
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4933

D
Daniel P. Berrange 已提交
4934
    if (!vm) {
4935 4936
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4937 4938
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4939
        goto cleanup;
D
Daniel P. Berrange 已提交
4940
    }
4941 4942 4943 4944

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

D
Daniel P. Berrange 已提交
4945
    if (!virDomainObjIsActive(vm)) {
4946 4947
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
4948
        goto endjob;
D
Daniel P. Berrange 已提交
4949
    }
4950
    if (vm->state == VIR_DOMAIN_PAUSED) {
4951
        qemuDomainObjPrivatePtr priv = vm->privateData;
4952
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
4953
        if (qemuMonitorStartCPUs(priv->mon, dom->conn) < 0) {
4954
            qemuDomainObjExitMonitorWithDriver(driver, vm);
4955
            if (virGetLastError() == NULL)
4956 4957
                qemuReportError(VIR_ERR_OPERATION_FAILED,
                                "%s", _("resume operation failed"));
4958
            goto endjob;
4959
        }
4960
        qemuDomainObjExitMonitorWithDriver(driver, vm);
4961
        vm->state = VIR_DOMAIN_RUNNING;
4962 4963 4964
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_UNPAUSED);
D
Daniel P. Berrange 已提交
4965
    }
4966
    if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
4967
        goto endjob;
4968 4969
    ret = 0;

4970
endjob:
4971 4972
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4973

4974
cleanup:
4975 4976
    if (vm)
        virDomainObjUnlock(vm);
4977
    if (event)
4978
        qemuDomainEventQueue(driver, event);
4979
    qemuDriverUnlock(driver);
4980
    return ret;
D
Daniel P. Berrange 已提交
4981 4982 4983
}


4984
static int qemudDomainShutdown(virDomainPtr dom) {
4985 4986 4987
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4988

4989
    qemuDriverLock(driver);
4990
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4991 4992
    qemuDriverUnlock(driver);

4993
    if (!vm) {
4994 4995
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4996 4997
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4998
        goto cleanup;
4999 5000
    }

5001 5002 5003
    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
5004
    if (!virDomainObjIsActive(vm)) {
5005 5006
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
5007
        goto endjob;
5008 5009
    }

5010
    qemuDomainObjPrivatePtr priv = vm->privateData;
5011 5012 5013
    qemuDomainObjEnterMonitor(vm);
    ret = qemuMonitorSystemPowerdown(priv->mon);
    qemuDomainObjExitMonitor(vm);
5014

5015
endjob:
5016 5017
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
5018

5019
cleanup:
5020 5021
    if (vm)
        virDomainObjUnlock(vm);
5022
    return ret;
5023 5024 5025
}


5026
static int qemudDomainDestroy(virDomainPtr dom) {
5027 5028 5029
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
5030
    virDomainEventPtr event = NULL;
5031

5032
    qemuDriverLock(driver);
5033
    vm  = virDomainFindByUUID(&driver->domains, dom->uuid);
D
Daniel P. Berrange 已提交
5034
    if (!vm) {
5035 5036
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
5037 5038
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
5039
        goto cleanup;
D
Daniel P. Berrange 已提交
5040
    }
5041 5042 5043 5044

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

D
Daniel P. Berrange 已提交
5045
    if (!virDomainObjIsActive(vm)) {
5046 5047
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
5048
        goto endjob;
5049
    }
5050

5051
    qemudShutdownVMDaemon(driver, vm, 0);
5052 5053 5054
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_DESTROYED);
5055 5056
    qemuDomainStopAudit(vm, "destroyed");

5057
    if (!vm->persistent) {
5058 5059 5060
        if (qemuDomainObjEndJob(vm) > 0)
            virDomainRemoveInactive(&driver->domains,
                                    vm);
5061 5062
        vm = NULL;
    }
5063 5064
    ret = 0;

5065
endjob:
5066 5067 5068
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
5069

5070
cleanup:
5071 5072
    if (vm)
        virDomainObjUnlock(vm);
5073 5074
    if (event)
        qemuDomainEventQueue(driver, event);
5075
    qemuDriverUnlock(driver);
5076
    return ret;
D
Daniel P. Berrange 已提交
5077 5078 5079
}


5080
static char *qemudDomainGetOSType(virDomainPtr dom) {
5081 5082 5083
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    char *type = NULL;
5084

5085
    qemuDriverLock(driver);
5086
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5087
    qemuDriverUnlock(driver);
5088
    if (!vm) {
5089 5090
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
5091 5092
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
5093
        goto cleanup;
5094 5095
    }

5096
    if (!(type = strdup(vm->def->os.type)))
5097
        virReportOOMError();
5098 5099

cleanup:
5100 5101
    if (vm)
        virDomainObjUnlock(vm);
5102 5103 5104
    return type;
}

5105 5106
/* Returns max memory in kb, 0 if error */
static unsigned long qemudDomainGetMaxMemory(virDomainPtr dom) {
5107 5108 5109
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    unsigned long ret = 0;
5110

5111
    qemuDriverLock(driver);
5112
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5113 5114
    qemuDriverUnlock(driver);

5115
    if (!vm) {
5116 5117
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
5118 5119
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
5120
        goto cleanup;
5121 5122
    }

5123
    ret = vm->def->mem.max_balloon;
5124 5125

cleanup:
5126 5127
    if (vm)
        virDomainObjUnlock(vm);
5128
    return ret;
5129 5130 5131
}

static int qemudDomainSetMemory(virDomainPtr dom, unsigned long newmem) {
5132
    struct qemud_driver *driver = dom->conn->privateData;
5133
    qemuDomainObjPrivatePtr priv;
5134
    virDomainObjPtr vm;
5135
    int ret = -1, r;
5136

5137
    qemuDriverLock(driver);
5138
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5139
    qemuDriverUnlock(driver);
5140
    if (!vm) {
5141 5142
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
5143 5144
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
5145
        goto cleanup;
5146 5147
    }

5148
    if (newmem > vm->def->mem.max_balloon) {
5149 5150
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("cannot set memory higher than max memory"));
5151
        goto cleanup;
5152 5153
    }

5154 5155 5156
    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

5157 5158 5159 5160 5161 5162
    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
        goto endjob;
    }

5163 5164 5165 5166 5167 5168
    priv = vm->privateData;
    qemuDomainObjEnterMonitor(vm);
    r = qemuMonitorSetBalloon(priv->mon, newmem);
    qemuDomainObjExitMonitor(vm);
    if (r < 0)
        goto endjob;
5169

5170 5171 5172 5173 5174
    /* 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;
5175
    }
5176

5177
    ret = 0;
5178
endjob:
5179 5180
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
5181

5182
cleanup:
5183 5184
    if (vm)
        virDomainObjUnlock(vm);
5185
    return ret;
5186 5187
}

5188
static int qemudDomainGetInfo(virDomainPtr dom,
5189
                              virDomainInfoPtr info) {
5190 5191 5192
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
5193 5194
    int err;
    unsigned long balloon;
5195

5196
    qemuDriverLock(driver);
5197
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5198
    qemuDriverUnlock(driver);
D
Daniel P. Berrange 已提交
5199
    if (!vm) {
5200 5201
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
5202 5203
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
5204
        goto cleanup;
D
Daniel P. Berrange 已提交
5205 5206
    }

5207
    info->state = vm->state;
D
Daniel P. Berrange 已提交
5208

D
Daniel P. Berrange 已提交
5209
    if (!virDomainObjIsActive(vm)) {
5210
        info->cpuTime = 0;
D
Daniel P. Berrange 已提交
5211
    } else {
5212
        if (qemudGetProcessInfo(&(info->cpuTime), NULL, vm->pid, 0) < 0) {
5213
            qemuReportError(VIR_ERR_OPERATION_FAILED, ("cannot read cputime for domain"));
5214
            goto cleanup;
D
Daniel P. Berrange 已提交
5215 5216 5217
        }
    }

5218
    info->maxMem = vm->def->mem.max_balloon;
5219

D
Daniel P. Berrange 已提交
5220
    if (virDomainObjIsActive(vm)) {
5221
        qemuDomainObjPrivatePtr priv = vm->privateData;
5222 5223 5224

        if ((vm->def->memballoon != NULL) &&
            (vm->def->memballoon->model == VIR_DOMAIN_MEMBALLOON_MODEL_NONE)) {
5225
            info->memory = vm->def->mem.max_balloon;
5226
        } else if (!priv->jobActive) {
5227 5228
            if (qemuDomainObjBeginJob(vm) < 0)
                goto cleanup;
5229 5230 5231 5232 5233 5234 5235 5236 5237
            if (!virDomainObjIsActive(vm))
                err = 0;
            else {
                qemuDomainObjEnterMonitor(vm);
                err = qemuMonitorGetBalloonInfo(priv->mon, &balloon);
                qemuDomainObjExitMonitor(vm);
            }
            if (qemuDomainObjEndJob(vm) == 0) {
                vm = NULL;
5238 5239 5240
                goto cleanup;
            }

5241 5242
            if (err < 0)
                goto cleanup;
5243 5244
            if (err == 0)
                /* Balloon not supported, so maxmem is always the allocation */
5245
                info->memory = vm->def->mem.max_balloon;
5246 5247 5248
            else
                info->memory = balloon;
        } else {
5249
            info->memory = vm->def->mem.cur_balloon;
5250
        }
5251
    } else {
5252
        info->memory = vm->def->mem.cur_balloon;
5253 5254
    }

5255
    info->nrVirtCpu = vm->def->vcpus;
5256 5257 5258
    ret = 0;

cleanup:
5259 5260
    if (vm)
        virDomainObjUnlock(vm);
5261
    return ret;
D
Daniel P. Berrange 已提交
5262 5263 5264
}


5265 5266 5267 5268 5269 5270 5271 5272
/** qemuDomainMigrateOffline:
 * Pause domain for non-live migration.
 */
static int
qemuDomainMigrateOffline(struct qemud_driver *driver,
                         virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
5273
    int state = vm->state;
5274 5275
    int ret;

5276
    vm->state = VIR_DOMAIN_PAUSED;
5277 5278 5279 5280 5281 5282 5283 5284 5285 5286 5287 5288
    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);
5289 5290
    } else
        vm->state = state;
5291 5292 5293 5294 5295

    return ret;
}


5296 5297 5298 5299 5300 5301 5302 5303 5304 5305 5306 5307 5308 5309
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 */
5310
        struct timespec ts = { .tv_sec = 0, .tv_nsec = 50 * 1000 * 1000ull };
5311 5312 5313
        struct timeval now;
        int rc;

5314 5315 5316 5317 5318 5319
        if (!virDomainObjIsActive(vm)) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("guest unexpectedly quit during migration"));
            goto cleanup;
        }

5320 5321
        if (priv->jobSignals & QEMU_JOB_SIGNAL_CANCEL) {
            priv->jobSignals ^= QEMU_JOB_SIGNAL_CANCEL;
5322 5323 5324 5325 5326 5327 5328
            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");
            }
5329 5330 5331 5332 5333
        } 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");
5334 5335 5336 5337 5338 5339 5340 5341 5342 5343 5344
        } 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");
5345 5346
        }

5347 5348 5349 5350 5351 5352 5353 5354 5355
        /* 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;
        }

5356 5357 5358 5359 5360 5361 5362 5363 5364 5365 5366 5367 5368 5369 5370 5371 5372 5373 5374 5375 5376 5377 5378 5379 5380 5381 5382 5383 5384 5385 5386 5387 5388 5389 5390 5391 5392 5393 5394 5395 5396 5397 5398 5399 5400 5401 5402 5403 5404 5405 5406 5407 5408 5409 5410 5411 5412 5413 5414 5415 5416 5417 5418 5419 5420 5421 5422 5423 5424 5425 5426 5427
        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;
}


5428
#define QEMUD_SAVE_MAGIC "LibvirtQemudSave"
5429 5430 5431
#define QEMUD_SAVE_VERSION 2

enum qemud_save_formats {
5432 5433 5434
    QEMUD_SAVE_FORMAT_RAW = 0,
    QEMUD_SAVE_FORMAT_GZIP = 1,
    QEMUD_SAVE_FORMAT_BZIP2 = 2,
5435 5436
    /*
     * Deprecated by xz and never used as part of a release
5437
     * QEMUD_SAVE_FORMAT_LZMA
5438 5439
     */
    QEMUD_SAVE_FORMAT_XZ = 3,
5440
    QEMUD_SAVE_FORMAT_LZOP = 4,
5441 5442 5443
    /* Note: add new members only at the end.
       These values are used in the on-disk format.
       Do not change or re-use numbers. */
5444 5445

    QEMUD_SAVE_FORMAT_LAST
5446
};
5447

5448 5449 5450 5451 5452
VIR_ENUM_DECL(qemudSaveCompression)
VIR_ENUM_IMPL(qemudSaveCompression, QEMUD_SAVE_FORMAT_LAST,
              "raw",
              "gzip",
              "bzip2",
5453 5454
              "xz",
              "lzop")
5455

5456 5457 5458 5459 5460
struct qemud_save_header {
    char magic[sizeof(QEMUD_SAVE_MAGIC)-1];
    int version;
    int xml_len;
    int was_running;
5461 5462
    int compressed;
    int unused[15];
5463 5464
};

5465 5466 5467 5468 5469 5470 5471
struct fileOpHookData {
    virDomainPtr dom;
    const char *path;
    char *xml;
    struct qemud_save_header *header;
};

5472
/* return -errno on failure, or 0 on success */
5473 5474 5475 5476 5477
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)) {
5478
        ret = -errno;
5479
        qemuReportError(VIR_ERR_OPERATION_FAILED,
5480 5481
                        _("failed to write header to domain save file '%s'"),
                        hdata->path);
5482 5483 5484 5485
        goto endjob;
    }

    if (safewrite(fd, hdata->xml, hdata->header->xml_len) != hdata->header->xml_len) {
5486
        ret = -errno;
5487 5488 5489 5490 5491 5492 5493 5494
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                         _("failed to write xml to '%s'"), hdata->path);
        goto endjob;
    }
endjob:
    return ret;
}

5495 5496 5497
/* this internal function expects the driver lock to already be held on entry */
static int qemudDomainSaveFlag(struct qemud_driver *driver, virDomainPtr dom,
                               virDomainObjPtr vm, const char *path,
5498
                               int compressed)
5499
{
5500
    char *xml = NULL;
5501
    struct qemud_save_header header;
5502 5503
    struct fileOpHookData hdata;
    int bypassSecurityDriver = 0;
5504
    int ret = -1;
5505
    int rc;
5506
    virDomainEventPtr event = NULL;
5507
    qemuDomainObjPrivatePtr priv;
5508 5509
    struct stat sb;
    int is_reg = 0;
5510
    unsigned long long offset;
5511
    virCgroupPtr cgroup = NULL;
5512 5513 5514 5515 5516

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

5517
    header.compressed = compressed;
5518

5519
    priv = vm->privateData;
5520

5521 5522 5523
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
5524
    if (!virDomainObjIsActive(vm)) {
5525 5526
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
5527
        goto endjob;
D
Daniel P. Berrange 已提交
5528
    }
5529

5530 5531 5532
    memset(&priv->jobInfo, 0, sizeof(priv->jobInfo));
    priv->jobInfo.type = VIR_DOMAIN_JOB_UNBOUNDED;

5533 5534 5535
    /* Pause */
    if (vm->state == VIR_DOMAIN_RUNNING) {
        header.was_running = 1;
5536
        vm->state = VIR_DOMAIN_PAUSED;
5537
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
5538
        if (qemuMonitorStopCPUs(priv->mon) < 0) {
5539
            qemuDomainObjExitMonitorWithDriver(driver, vm);
5540
            vm->state = VIR_DOMAIN_RUNNING;
5541
            goto endjob;
5542
        }
5543
        qemuDomainObjExitMonitorWithDriver(driver, vm);
5544 5545 5546 5547 5548 5549

        if (!virDomainObjIsActive(vm)) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("guest unexpectedly quit"));
            goto endjob;
        }
5550 5551 5552
    }

    /* Get XML for the domain */
5553
    xml = virDomainDefFormat(vm->def, VIR_DOMAIN_XML_SECURE);
5554
    if (!xml) {
5555 5556
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to get domain xml"));
5557
        goto endjob;
5558 5559 5560
    }
    header.xml_len = strlen(xml) + 1;

5561 5562 5563 5564 5565 5566 5567 5568 5569 5570 5571 5572 5573
    /* 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);
    }

5574 5575 5576 5577 5578
    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
5579 5580
     * it by padding the XML string with NULLs.
     */
5581 5582 5583 5584 5585 5586 5587 5588 5589 5590 5591 5592 5593
    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;
    }
5594

5595 5596 5597 5598 5599 5600
    /* Setup hook data needed by virFileOperation hook function */
    hdata.dom = dom;
    hdata.path = path;
    hdata.xml = xml;
    hdata.header = &header;

5601 5602
    /* Write header to file, followed by XML */

5603
    /* First try creating the file as root */
5604 5605 5606 5607
    if (!is_reg) {
        int fd = open(path, O_WRONLY | O_TRUNC);
        if (fd < 0) {
            virReportSystemError(errno, _("unable to open %s"), path);
5608 5609
            goto endjob;
        }
5610
        if (qemudDomainSaveFileOpHook(fd, &hdata) < 0) {
5611 5612 5613 5614 5615
            close(fd);
            goto endjob;
        }
        if (close(fd) < 0) {
            virReportSystemError(errno, _("unable to close %s"), path);
5616 5617
            goto endjob;
        }
5618 5619 5620 5621 5622
    } else {
        if ((rc = virFileOperation(path, O_CREAT|O_TRUNC|O_WRONLY,
                                  S_IRUSR|S_IWUSR,
                                  getuid(), getgid(),
                                  qemudDomainSaveFileOpHook, &hdata,
5623
                                  0)) < 0) {
5624
            /* If we failed as root, and the error was permission-denied
5625 5626
               (EACCES or EPERM), assume it's on a network-connected share
               where root access is restricted (eg, root-squashed NFS). If the
5627 5628 5629 5630
               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 */

5631
            if (((rc != -EACCES) && (rc != -EPERM)) ||
5632
                driver->user == getuid()) {
5633
                virReportSystemError(-rc, _("Failed to create domain save file '%s'"),
5634 5635 5636
                                     path);
                goto endjob;
            }
5637

5638
            /* On Linux we can also verify the FS-type of the directory. */
5639 5640 5641 5642 5643 5644 5645 5646 5647 5648 5649 5650 5651 5652 5653 5654 5655 5656
            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 */
5657
                   virReportSystemError(-rc,
5658 5659 5660 5661
                                        _("Failed to create domain save file '%s'"),
                                        path);
                   goto endjob;
                   break;
5662 5663 5664

            }

5665
            /* Retry creating the file as driver->user */
5666

5667 5668 5669 5670
            if ((rc = virFileOperation(path, O_CREAT|O_TRUNC|O_WRONLY,
                                       S_IRUSR|S_IWUSR|S_IRGRP|S_IWGRP,
                                       driver->user, driver->group,
                                       qemudDomainSaveFileOpHook, &hdata,
5671 5672
                                       VIR_FILE_OP_AS_UID)) < 0) {
                virReportSystemError(-rc, _("Error from child process creating '%s'"),
5673
                                 path);
5674 5675
                goto endjob;
            }
5676

5677 5678 5679
            /* 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 */
5680

5681 5682
            bypassSecurityDriver = 1;
        }
5683
    }
5684

5685

5686 5687 5688 5689 5690 5691 5692 5693 5694 5695 5696 5697 5698 5699 5700 5701 5702
    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;
        }
    }

5703 5704
    if ((!bypassSecurityDriver) &&
        driver->securityDriver &&
5705
        driver->securityDriver->domainSetSavedStateLabel &&
5706 5707
        driver->securityDriver->domainSetSavedStateLabel(driver->securityDriver,
                                                         vm, path) == -1)
5708 5709
        goto endjob;

5710 5711
    if (header.compressed == QEMUD_SAVE_FORMAT_RAW) {
        const char *args[] = { "cat", NULL };
M
Matthias Bolte 已提交
5712
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
5713 5714 5715
        rc = qemuMonitorMigrateToFile(priv->mon,
                                      QEMU_MONITOR_MIGRATE_BACKGROUND,
                                      args, path, offset);
M
Matthias Bolte 已提交
5716
        qemuDomainObjExitMonitorWithDriver(driver, vm);
5717
    } else {
5718
        const char *prog = qemudSaveCompressionTypeToString(header.compressed);
5719 5720 5721 5722 5723
        const char *args[] = {
            prog,
            "-c",
            NULL
        };
M
Matthias Bolte 已提交
5724
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
5725 5726 5727
        rc = qemuMonitorMigrateToFile(priv->mon,
                                      QEMU_MONITOR_MIGRATE_BACKGROUND,
                                      args, path, offset);
M
Matthias Bolte 已提交
5728
        qemuDomainObjExitMonitorWithDriver(driver, vm);
5729 5730
    }

5731 5732 5733 5734 5735
    if (rc < 0)
        goto endjob;

    rc = qemuDomainWaitForMigrationComplete(driver, vm);

5736
    if (rc < 0)
5737
        goto endjob;
5738

5739 5740
    if ((!bypassSecurityDriver) &&
        driver->securityDriver &&
5741
        driver->securityDriver->domainRestoreSavedStateLabel &&
5742 5743
        driver->securityDriver->domainRestoreSavedStateLabel(driver->securityDriver,
                                                             vm, path) == -1)
5744
        VIR_WARN("failed to restore save state label on %s", path);
5745

5746 5747
    if (cgroup != NULL) {
        rc = virCgroupDenyDevicePath(cgroup, path);
5748 5749 5750
        if (rc != 0)
            VIR_WARN("Unable to deny device %s for %s %d",
                     path, vm->def->name, rc);
5751 5752
    }

5753 5754
    ret = 0;

5755
    /* Shut it down */
5756
    qemudShutdownVMDaemon(driver, vm, 0);
5757
    qemuDomainStopAudit(vm, "saved");
5758 5759 5760
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_SAVED);
5761
    if (!vm->persistent) {
5762 5763 5764
        if (qemuDomainObjEndJob(vm) > 0)
            virDomainRemoveInactive(&driver->domains,
                                    vm);
5765 5766
        vm = NULL;
    }
5767

5768
endjob:
5769
    if (vm) {
5770
        if (ret != 0) {
5771
            if (header.was_running && virDomainObjIsActive(vm)) {
5772 5773 5774 5775 5776 5777 5778 5779
                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;
            }
5780

5781 5782 5783 5784 5785
            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);
5786
            }
5787 5788 5789 5790

            if ((!bypassSecurityDriver) &&
                driver->securityDriver &&
                driver->securityDriver->domainRestoreSavedStateLabel &&
5791 5792
                driver->securityDriver->domainRestoreSavedStateLabel(driver->securityDriver,
                                                                     vm, path) == -1)
5793
                VIR_WARN("failed to restore save state label on %s", path);
5794 5795
        }

5796
        if (qemuDomainObjEndJob(vm) == 0)
5797
            vm = NULL;
5798
    }
5799

5800 5801
cleanup:
    VIR_FREE(xml);
5802
    if (ret != 0 && is_reg)
5803
        unlink(path);
5804 5805
    if (event)
        qemuDomainEventQueue(driver, event);
5806
    virCgroupFree(&cgroup);
5807
    return ret;
D
Daniel P. Berrange 已提交
5808 5809
}

5810 5811 5812 5813 5814 5815 5816 5817 5818 5819 5820 5821 5822 5823 5824 5825
/* Returns true if a compression program is available in PATH */
static bool qemudCompressProgramAvailable(enum qemud_save_formats compress)
{
    const char *prog;
    char *c;

    if (compress == QEMUD_SAVE_FORMAT_RAW)
        return true;
    prog = qemudSaveCompressionTypeToString(compress);
    c = virFindFileInPath(prog);
    if (!c)
        return false;
    VIR_FREE(c);
    return true;
}

5826 5827 5828 5829
static int qemudDomainSave(virDomainPtr dom, const char *path)
{
    struct qemud_driver *driver = dom->conn->privateData;
    int compressed;
5830 5831 5832 5833
    int ret = -1;
    virDomainObjPtr vm = NULL;

    qemuDriverLock(driver);
5834 5835 5836 5837 5838 5839 5840 5841 5842 5843 5844

    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;
        }
5845 5846 5847 5848 5849 5850
        if (!qemudCompressProgramAvailable(compressed)) {
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            "%s", _("Compression program for image format "
                                    "in configuration file isn't available"));
            return -1;
        }
5851 5852
    }

5853 5854 5855 5856 5857 5858 5859 5860 5861 5862 5863 5864 5865 5866 5867 5868 5869
    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;
    }

    ret = qemudDomainSaveFlag(driver, dom, vm, path, compressed);

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);

    return ret;
5870 5871 5872 5873 5874 5875 5876 5877 5878 5879 5880 5881 5882 5883 5884 5885 5886 5887 5888 5889 5890 5891 5892
}

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;

5893
    virCheckFlags(0, -1);
5894 5895 5896 5897 5898 5899 5900 5901

    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);
5902
        goto cleanup;
5903 5904 5905 5906
    }

    name = qemuDomainManagedSavePath(driver, vm);
    if (name == NULL)
5907
        goto cleanup;
5908 5909 5910 5911

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

    compressed = QEMUD_SAVE_FORMAT_RAW;
5912
    ret = qemudDomainSaveFlag(driver, dom, vm, name, compressed);
5913 5914 5915 5916 5917

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);
5918 5919 5920
    VIR_FREE(name);

    return ret;
5921 5922 5923 5924 5925 5926 5927 5928 5929 5930
}

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

5931
    virCheckFlags(0, -1);
5932 5933 5934 5935 5936 5937 5938 5939 5940 5941 5942 5943 5944 5945 5946 5947 5948 5949 5950 5951 5952 5953 5954 5955 5956 5957 5958 5959 5960 5961 5962 5963 5964

    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;

5965
    virCheckFlags(0, -1);
5966 5967 5968 5969 5970 5971 5972 5973 5974 5975 5976 5977 5978 5979 5980 5981 5982 5983 5984 5985 5986 5987 5988 5989

    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
5990

P
Paolo Bonzini 已提交
5991 5992 5993 5994 5995 5996
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;
5997
    int ret = -1, fd = -1;
5998
    virDomainEventPtr event = NULL;
5999
    int compress;
6000
    qemuDomainObjPrivatePtr priv;
6001 6002 6003 6004 6005 6006 6007 6008 6009 6010 6011 6012 6013
    /*
     * We reuse "save" flag for "dump" here. Then, we can support the same
     * format in "save" and "dump".
     */
    compress = QEMUD_SAVE_FORMAT_RAW;
    if (driver->dumpImageFormat) {
        compress = qemudSaveCompressionTypeFromString(driver->dumpImageFormat);
        if (compress < 0) {
           qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
                           _("Invalid dump image format specified in "
                             "configuration file"));
           return -1;
        }
6014 6015 6016 6017 6018 6019
        if (!qemudCompressProgramAvailable(compress)) {
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            "%s", _("Compression program for dump image format "
                                    "in configuration file isn't available"));
            return -1;
        }
6020
    }
P
Paolo Bonzini 已提交
6021 6022 6023 6024 6025 6026 6027

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

    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
6028 6029
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
P
Paolo Bonzini 已提交
6030 6031
        goto cleanup;
    }
6032
    priv = vm->privateData;
P
Paolo Bonzini 已提交
6033

6034
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
6035 6036
        goto cleanup;

D
Daniel P. Berrange 已提交
6037
    if (!virDomainObjIsActive(vm)) {
6038 6039
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
6040
        goto endjob;
P
Paolo Bonzini 已提交
6041 6042
    }

6043 6044
    /* Create an empty file with appropriate ownership.  */
    if ((fd = open(path, O_CREAT|O_TRUNC|O_WRONLY, S_IRUSR|S_IWUSR)) < 0) {
6045 6046
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("failed to create '%s'"), path);
6047 6048 6049 6050
        goto endjob;
    }

    if (close(fd) < 0) {
6051
        virReportSystemError(errno,
6052 6053 6054 6055 6056 6057 6058
                             _("unable to save file %s"),
                             path);
        goto endjob;
    }

    if (driver->securityDriver &&
        driver->securityDriver->domainSetSavedStateLabel &&
6059 6060
        driver->securityDriver->domainSetSavedStateLabel(driver->securityDriver,
                                                         vm, path) == -1)
6061 6062
        goto endjob;

P
Paolo Bonzini 已提交
6063 6064
    /* 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
6065 6066 6067
    resume = (vm->state == VIR_DOMAIN_RUNNING);

    /* Pause domain for non-live dump */
P
Paolo Bonzini 已提交
6068
    if (!(flags & VIR_DUMP_LIVE) && vm->state == VIR_DOMAIN_RUNNING) {
6069
        vm->state = VIR_DOMAIN_PAUSED;
6070
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
6071
        if (qemuMonitorStopCPUs(priv->mon) < 0) {
6072
            qemuDomainObjExitMonitorWithDriver(driver, vm);
6073
            vm->state = VIR_DOMAIN_RUNNING;
6074
            goto endjob;
6075
        }
6076
        qemuDomainObjExitMonitorWithDriver(driver, vm);
P
Paolo Bonzini 已提交
6077
        paused = 1;
6078 6079 6080 6081 6082 6083

        if (!virDomainObjIsActive(vm)) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("guest unexpectedly quit"));
            goto endjob;
        }
P
Paolo Bonzini 已提交
6084 6085
    }

6086
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
6087 6088 6089 6090 6091 6092 6093 6094 6095 6096 6097 6098 6099 6100 6101 6102 6103 6104 6105
    if (compress == QEMUD_SAVE_FORMAT_RAW) {
        const char *args[] = {
            "cat",
            NULL,
        };
        ret = qemuMonitorMigrateToFile(priv->mon,
                                       QEMU_MONITOR_MIGRATE_BACKGROUND,
                                       args, path, 0);
    } else {
        const char *prog = qemudSaveCompressionTypeToString(compress);
        const char *args[] = {
            prog,
            "-c",
            NULL,
        };
        ret = qemuMonitorMigrateToFile(priv->mon,
                                       QEMU_MONITOR_MIGRATE_BACKGROUND,
                                       args, path, 0);
    }
6106
    qemuDomainObjExitMonitorWithDriver(driver, vm);
6107 6108 6109 6110 6111 6112 6113 6114 6115
    if (ret < 0)
        goto endjob;

    ret = qemuDomainWaitForMigrationComplete(driver, vm);

    if (ret < 0)
        goto endjob;

    paused = 1;
6116 6117 6118

    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSavedStateLabel &&
6119 6120
        driver->securityDriver->domainRestoreSavedStateLabel(driver->securityDriver,
                                                             vm, path) == -1)
6121 6122 6123
        goto endjob;

endjob:
6124
    if ((ret == 0) && (flags & VIR_DUMP_CRASH)) {
6125
        qemudShutdownVMDaemon(driver, vm, 0);
6126
        qemuDomainStopAudit(vm, "crashed");
6127 6128 6129 6130 6131
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         VIR_DOMAIN_EVENT_STOPPED_CRASHED);
    }

P
Paolo Bonzini 已提交
6132 6133 6134
    /* 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.  */
6135
    else if (resume && paused && virDomainObjIsActive(vm)) {
6136
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
6137
        if (qemuMonitorStartCPUs(priv->mon, dom->conn) < 0) {
6138
            if (virGetLastError() == NULL)
6139 6140
                qemuReportError(VIR_ERR_OPERATION_FAILED,
                                "%s", _("resuming after dump failed"));
P
Paolo Bonzini 已提交
6141
        }
6142
        qemuDomainObjExitMonitorWithDriver(driver, vm);
6143
        vm->state = VIR_DOMAIN_RUNNING;
P
Paolo Bonzini 已提交
6144
    }
6145

6146 6147
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
6148
    else if ((ret == 0) && (flags & VIR_DUMP_CRASH) && !vm->persistent) {
6149 6150 6151 6152
        virDomainRemoveInactive(&driver->domains,
                                vm);
        vm = NULL;
    }
6153 6154

cleanup:
6155 6156
    if (ret != 0)
        unlink(path);
P
Paolo Bonzini 已提交
6157 6158
    if (vm)
        virDomainObjUnlock(vm);
6159 6160
    if (event)
        qemuDomainEventQueue(driver, event);
6161
    qemuDriverUnlock(driver);
P
Paolo Bonzini 已提交
6162 6163 6164 6165
    return ret;
}


6166 6167 6168 6169 6170 6171
static int qemudDomainHotplugVcpus(virDomainObjPtr vm, unsigned int nvcpus)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int i, rc;
    int ret = -1;

6172 6173
    qemuDomainObjEnterMonitor(vm);

6174 6175 6176 6177 6178 6179 6180 6181 6182 6183 6184 6185 6186 6187 6188 6189 6190 6191 6192 6193 6194 6195 6196 6197 6198 6199 6200 6201 6202 6203
    /* 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:
6204
    qemuDomainObjExitMonitor(vm);
6205 6206 6207 6208 6209 6210 6211 6212 6213
    return ret;

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


6214 6215 6216 6217
static int
qemudDomainSetVcpusFlags(virDomainPtr dom, unsigned int nvcpus,
                         unsigned int flags)
{
6218 6219
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
6220
    virDomainDefPtr def;
6221 6222
    const char * type;
    int max;
6223
    int ret = -1;
6224

6225 6226 6227 6228 6229 6230 6231 6232 6233 6234 6235 6236 6237 6238 6239 6240
    virCheckFlags(VIR_DOMAIN_VCPU_LIVE |
                  VIR_DOMAIN_VCPU_CONFIG |
                  VIR_DOMAIN_VCPU_MAXIMUM, -1);

    /* At least one of LIVE or CONFIG must be set.  MAXIMUM cannot be
     * mixed with LIVE.  */
    if ((flags & (VIR_DOMAIN_VCPU_LIVE | VIR_DOMAIN_VCPU_CONFIG)) == 0 ||
        (flags & (VIR_DOMAIN_VCPU_MAXIMUM | VIR_DOMAIN_VCPU_LIVE)) ==
         (VIR_DOMAIN_VCPU_MAXIMUM | VIR_DOMAIN_VCPU_LIVE)) {
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("invalid flag combination: (0x%x)"), flags);
        return -1;
    }
    if (!nvcpus || (unsigned short) nvcpus != nvcpus) {
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("argument out of range: %d"), nvcpus);
6241 6242 6243
        return -1;
    }

6244
    qemuDriverLock(driver);
6245
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
6246 6247
    qemuDriverUnlock(driver);

6248
    if (!vm) {
6249 6250
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
6251 6252
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
6253
        goto cleanup;
6254 6255
    }

6256 6257 6258
    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

6259
    if (!virDomainObjIsActive(vm) && (flags & VIR_DOMAIN_VCPU_LIVE)) {
6260 6261
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                         "%s", _("domain is not running"));
6262
        goto endjob;
6263 6264
    }

6265 6266 6267 6268 6269 6270 6271 6272 6273 6274 6275 6276 6277
    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;
    }

6278 6279 6280 6281 6282
    if ((flags & (VIR_DOMAIN_VCPU_MAXIMUM | VIR_DOMAIN_VCPU_LIVE)) ==
        VIR_DOMAIN_VCPU_LIVE && vm->def->maxvcpus < max) {
        max = vm->def->maxvcpus;
    }

6283 6284 6285 6286 6287 6288 6289
    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;
    }

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
    switch (flags) {
    case VIR_DOMAIN_VCPU_MAXIMUM | VIR_DOMAIN_VCPU_CONFIG:
        def = vm->def;
        if (virDomainObjIsActive(vm)) {
            if (vm->newDef)
                def = vm->newDef;
            else{
                qemuReportError(VIR_ERR_OPERATION_INVALID, "%s",
                                _("no persistent state"));
                goto endjob;
            }
        }
        def->maxvcpus = nvcpus;
        if (nvcpus < vm->newDef->vcpus)
            def->vcpus = nvcpus;
        ret = 0;
        break;

    case VIR_DOMAIN_VCPU_CONFIG:
        def = vm->def;
        if (virDomainObjIsActive(vm)) {
            if (vm->newDef)
                def = vm->newDef;
            else {
                qemuReportError(VIR_ERR_OPERATION_INVALID, "%s",
                                _("no persistent state"));
                goto endjob;
            }
        }
        def->vcpus = nvcpus;
        ret = 0;
        break;

    case VIR_DOMAIN_VCPU_LIVE:
        ret = qemudDomainHotplugVcpus(vm, nvcpus);
        break;

    case VIR_DOMAIN_VCPU_LIVE | VIR_DOMAIN_VCPU_CONFIG:
        ret = qemudDomainHotplugVcpus(vm, nvcpus);
        if (ret == 0 && vm->newDef)
            vm->newDef->vcpus = nvcpus;
        break;
    }
6333 6334 6335 6336

endjob:
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
6337

6338
cleanup:
6339 6340
    if (vm)
        virDomainObjUnlock(vm);
6341
    return ret;
6342 6343
}

6344 6345 6346 6347 6348 6349
static int
qemudDomainSetVcpus(virDomainPtr dom, unsigned int nvcpus)
{
    return qemudDomainSetVcpusFlags(dom, nvcpus, VIR_DOMAIN_VCPU_LIVE);
}

6350 6351 6352 6353 6354 6355

static int
qemudDomainPinVcpu(virDomainPtr dom,
                   unsigned int vcpu,
                   unsigned char *cpumap,
                   int maplen) {
6356 6357
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
6358
    int maxcpu, hostcpus;
6359
    virNodeInfo nodeinfo;
6360
    int ret = -1;
6361
    qemuDomainObjPrivatePtr priv;
6362

6363
    qemuDriverLock(driver);
6364
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
6365 6366
    qemuDriverUnlock(driver);

6367 6368 6369
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
6370 6371
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
6372 6373 6374
        goto cleanup;
    }

D
Daniel P. Berrange 已提交
6375
    if (!virDomainObjIsActive(vm)) {
6376 6377
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s",_("cannot pin vcpus on an inactive domain"));
6378
        goto cleanup;
6379 6380
    }

6381 6382 6383
    priv = vm->privateData;

    if (vcpu > (priv->nvcpupids-1)) {
6384 6385 6386
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("vcpu number out of range %d > %d"),
                        vcpu, priv->nvcpupids);
6387
        goto cleanup;
6388 6389
    }

6390
    if (nodeGetInfo(dom->conn, &nodeinfo) < 0)
6391
        goto cleanup;
6392

6393
    hostcpus = VIR_NODEINFO_MAXCPUS(nodeinfo);
6394
    maxcpu = maplen * 8;
6395 6396
    if (maxcpu > hostcpus)
        maxcpu = hostcpus;
6397

6398 6399
    if (priv->vcpupids != NULL) {
        if (virProcessInfoSetAffinity(priv->vcpupids[vcpu],
6400
                                      cpumap, maplen, maxcpu) < 0)
6401
            goto cleanup;
6402
    } else {
6403 6404
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        "%s", _("cpu affinity is not supported"));
6405
        goto cleanup;
6406
    }
6407
    ret = 0;
6408

6409
cleanup:
6410 6411
    if (vm)
        virDomainObjUnlock(vm);
6412
    return ret;
6413 6414 6415 6416 6417 6418 6419 6420
}

static int
qemudDomainGetVcpus(virDomainPtr dom,
                    virVcpuInfoPtr info,
                    int maxinfo,
                    unsigned char *cpumaps,
                    int maplen) {
6421 6422
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
6423
    virNodeInfo nodeinfo;
6424
    int i, v, maxcpu, hostcpus;
6425
    int ret = -1;
6426
    qemuDomainObjPrivatePtr priv;
6427

6428
    qemuDriverLock(driver);
6429
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
6430 6431
    qemuDriverUnlock(driver);

6432 6433 6434
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
6435 6436
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
6437 6438 6439
        goto cleanup;
    }

D
Daniel P. Berrange 已提交
6440
    if (!virDomainObjIsActive(vm)) {
6441 6442 6443
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s",
                        _("cannot list vcpu pinning for an inactive domain"));
6444
        goto cleanup;
6445 6446
    }

6447 6448
    priv = vm->privateData;

6449
    if (nodeGetInfo(dom->conn, &nodeinfo) < 0)
6450
        goto cleanup;
6451

6452
    hostcpus = VIR_NODEINFO_MAXCPUS(nodeinfo);
6453
    maxcpu = maplen * 8;
6454 6455
    if (maxcpu > hostcpus)
        maxcpu = hostcpus;
6456 6457

    /* Clamp to actual number of vcpus */
6458 6459
    if (maxinfo > priv->nvcpupids)
        maxinfo = priv->nvcpupids;
6460

6461 6462 6463 6464 6465 6466
    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;
6467

6468
                if (priv->vcpupids != NULL &&
6469 6470 6471
                    qemudGetProcessInfo(&(info[i].cpuTime),
                                        &(info[i].cpu),
                                        vm->pid,
6472
                                        priv->vcpupids[i]) < 0) {
6473
                    virReportSystemError(errno, "%s",
6474 6475 6476
                                         _("cannot get vCPU placement & pCPU time"));
                    goto cleanup;
                }
6477
            }
6478 6479
        }

6480 6481
        if (cpumaps != NULL) {
            memset(cpumaps, 0, maplen * maxinfo);
6482
            if (priv->vcpupids != NULL) {
6483 6484 6485
                for (v = 0 ; v < maxinfo ; v++) {
                    unsigned char *cpumap = VIR_GET_CPUMAP(cpumaps, maplen, v);

6486
                    if (virProcessInfoGetAffinity(priv->vcpupids[v],
6487
                                                  cpumap, maplen, maxcpu) < 0)
6488
                        goto cleanup;
6489
                }
6490
            } else {
6491 6492
                qemuReportError(VIR_ERR_NO_SUPPORT,
                                "%s", _("cpu affinity is not available"));
6493
                goto cleanup;
6494 6495 6496
            }
        }
    }
6497
    ret = maxinfo;
6498

6499
cleanup:
6500 6501
    if (vm)
        virDomainObjUnlock(vm);
6502
    return ret;
6503 6504 6505
}


6506 6507 6508
static int
qemudDomainGetVcpusFlags(virDomainPtr dom, unsigned int flags)
{
6509 6510
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
6511
    virDomainDefPtr def;
6512
    int ret = -1;
6513

6514 6515 6516 6517 6518 6519 6520 6521
    virCheckFlags(VIR_DOMAIN_VCPU_LIVE |
                  VIR_DOMAIN_VCPU_CONFIG |
                  VIR_DOMAIN_VCPU_MAXIMUM, -1);

    /* Exactly one of LIVE or CONFIG must be set.  */
    if (!(flags & VIR_DOMAIN_VCPU_LIVE) == !(flags & VIR_DOMAIN_VCPU_CONFIG)) {
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("invalid flag combination: (0x%x)"), flags);
6522 6523 6524
        return -1;
    }

6525
    qemuDriverLock(driver);
6526
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
6527 6528
    qemuDriverUnlock(driver);

6529
    if (!vm) {
6530 6531
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
6532 6533
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
6534
        goto cleanup;
6535 6536
    }

6537 6538 6539 6540 6541 6542 6543 6544 6545
    if (flags & VIR_DOMAIN_VCPU_LIVE) {
        if (!virDomainObjIsActive(vm)) {
            qemuReportError(VIR_ERR_OPERATION_INVALID, "%s",
                            _("domain not active"));
            goto cleanup;
        }
        def = vm->def;
    } else {
        def = vm->newDef ? vm->newDef : vm->def;
6546 6547
    }

6548
    ret = (flags & VIR_DOMAIN_VCPU_MAXIMUM) ? def->maxvcpus : def->vcpus;
6549

6550
cleanup:
6551 6552
    if (vm)
        virDomainObjUnlock(vm);
6553 6554 6555
    return ret;
}

6556 6557 6558 6559 6560 6561 6562
static int
qemudDomainGetMaxVcpus(virDomainPtr dom)
{
    return qemudDomainGetVcpusFlags(dom, (VIR_DOMAIN_VCPU_LIVE |
                                          VIR_DOMAIN_VCPU_MAXIMUM));
}

6563 6564 6565 6566 6567 6568 6569 6570 6571
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);

6572 6573
    memset(seclabel, 0, sizeof(*seclabel));

6574 6575 6576
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
6577 6578
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
6579 6580 6581
        goto cleanup;
    }

6582
    if (!virDomainVirtTypeToString(vm->def->virtType)) {
6583 6584 6585
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("unknown virt type in domain definition '%d'"),
                        vm->def->virtType);
6586 6587 6588 6589 6590 6591 6592 6593 6594 6595 6596 6597 6598 6599 6600 6601 6602
        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
6603
    if (virDomainObjIsActive(vm)) {
6604 6605 6606 6607 6608 6609 6610
        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;
6611 6612 6613 6614 6615 6616 6617 6618
        }
    }

    ret = 0;

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
6619
    qemuDriverUnlock(driver);
6620 6621 6622
    return ret;
}

6623 6624
static int qemudNodeGetSecurityModel(virConnectPtr conn,
                                     virSecurityModelPtr secmodel)
6625 6626 6627
{
    struct qemud_driver *driver = (struct qemud_driver *)conn->privateData;
    char *p;
6628
    int ret = 0;
6629

6630
    qemuDriverLock(driver);
6631
    if (!driver->securityPrimaryDriver) {
6632
        memset(secmodel, 0, sizeof (*secmodel));
6633 6634
        goto cleanup;
    }
6635

6636 6637
    p = driver->caps->host.secModel.model;
    if (strlen(p) >= VIR_SECURITY_MODEL_BUFLEN-1) {
6638 6639 6640
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("security model string exceeds max %d bytes"),
                        VIR_SECURITY_MODEL_BUFLEN-1);
6641 6642
        ret = -1;
        goto cleanup;
6643 6644 6645 6646 6647
    }
    strcpy(secmodel->model, p);

    p = driver->caps->host.secModel.doi;
    if (strlen(p) >= VIR_SECURITY_DOI_BUFLEN-1) {
6648 6649 6650
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("security DOI string exceeds max %d bytes"),
                        VIR_SECURITY_DOI_BUFLEN-1);
6651 6652
        ret = -1;
        goto cleanup;
6653 6654
    }
    strcpy(secmodel->doi, p);
6655 6656 6657 6658

cleanup:
    qemuDriverUnlock(driver);
    return ret;
6659 6660
}

6661 6662 6663 6664 6665 6666 6667 6668 6669 6670 6671 6672 6673 6674 6675 6676 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 6718 6719 6720 6721 6722 6723 6724 6725 6726 6727 6728 6729 6730 6731 6732 6733
/* 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;
6734
    struct passwd pwd, *pwd_result;
6735 6736 6737 6738 6739 6740 6741 6742 6743 6744 6745 6746

    /* 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;
    }

6747 6748 6749 6750 6751 6752 6753 6754 6755 6756 6757 6758 6759 6760 6761 6762 6763 6764 6765 6766
    if (VIR_ALLOC_N(buf, bufsize) < 0) {
        exit_code = ENOMEM;
        virReportOOMError();
        goto child_cleanup;
    }

    exit_code = getpwuid_r(uid, &pwd, buf, bufsize, &pwd_result);
    if (pwd_result == NULL) {
        virReportSystemError(errno,
                             _("cannot getpwuid_r(%d) to read '%s'"),
                             uid, path);
        goto child_cleanup;
    }
    if (initgroups(pwd.pw_name, pwd.pw_gid) != 0) {
        exit_code = errno;
        virReportSystemError(errno,
                             _("cannot initgroups(\"%s\", %d) to read '%s'"),
                             pwd.pw_name, pwd.pw_gid, path);
        goto child_cleanup;
    }
6767 6768 6769 6770 6771 6772 6773 6774 6775 6776 6777 6778 6779 6780 6781 6782 6783 6784 6785 6786 6787 6788 6789 6790 6791 6792 6793 6794 6795 6796 6797 6798 6799 6800 6801 6802 6803 6804 6805 6806 6807 6808
    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;
    }

    /* 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
6809 6810 6811 6812 6813 6814 6815 6816 6817 6818 6819 6820 6821 6822 6823 6824 6825 6826 6827 6828 6829 6830 6831 6832 6833 6834 6835 6836
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;
6837
    pid_t read_pid = -1;
6838
    struct qemud_save_header header;
J
Jiri Denemark 已提交
6839 6840
    char *xml = NULL;
    virDomainDefPtr def = NULL;
6841 6842

    if ((fd = open(path, O_RDONLY)) < 0) {
6843 6844 6845
        if ((driver->user == 0) || (getuid() != 0)) {
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            "%s", _("cannot read domain image"));
J
Jiri Denemark 已提交
6846
            goto error;
6847 6848 6849 6850 6851 6852 6853 6854
        }

        /* 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
6855
            goto error;
6856
        }
6857 6858 6859
    }

    if (saferead(fd, &header, sizeof(header)) != sizeof(header)) {
6860 6861
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to read qemu header"));
J
Jiri Denemark 已提交
6862
        goto error;
6863 6864 6865
    }

    if (memcmp(header.magic, QEMUD_SAVE_MAGIC, sizeof(header.magic)) != 0) {
6866 6867
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("image magic is incorrect"));
J
Jiri Denemark 已提交
6868
        goto error;
6869 6870 6871
    }

    if (header.version > QEMUD_SAVE_VERSION) {
6872 6873 6874
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("image version is not supported (%d > %d)"),
                        header.version, QEMUD_SAVE_VERSION);
J
Jiri Denemark 已提交
6875
        goto error;
6876 6877
    }

6878 6879 6880
    if (header.xml_len <= 0) {
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("invalid XML length: %d"), header.xml_len);
J
Jiri Denemark 已提交
6881
        goto error;
6882 6883
    }

6884 6885
    if (VIR_ALLOC_N(xml, header.xml_len) < 0) {
        virReportOOMError();
J
Jiri Denemark 已提交
6886
        goto error;
6887 6888 6889
    }

    if (saferead(fd, xml, header.xml_len) != header.xml_len) {
6890 6891
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to read XML"));
J
Jiri Denemark 已提交
6892
        goto error;
6893 6894 6895
    }

    /* Create a domain from this XML */
6896
    if (!(def = virDomainDefParseString(driver->caps, xml,
6897
                                        VIR_DOMAIN_XML_INACTIVE))) {
6898 6899
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to parse XML"));
J
Jiri Denemark 已提交
6900
        goto error;
6901 6902
    }

J
Jiri Denemark 已提交
6903
    VIR_FREE(xml);
6904

J
Jiri Denemark 已提交
6905 6906 6907
    *ret_def = def;
    *ret_header = header;
    *ret_read_pid = read_pid;
6908

J
Jiri Denemark 已提交
6909
    return fd;
6910

J
Jiri Denemark 已提交
6911 6912 6913 6914 6915 6916 6917 6918 6919 6920 6921 6922 6923 6924 6925 6926 6927 6928 6929 6930 6931 6932 6933 6934 6935 6936
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) {
6937
        const char *intermediate_argv[3] = { NULL, "-dc", NULL };
J
Jiri Denemark 已提交
6938
        const char *prog = qemudSaveCompressionTypeToString(header->compressed);
6939
        if (prog == NULL) {
6940 6941
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("Invalid compressed save format %d"),
J
Jiri Denemark 已提交
6942 6943
                            header->compressed);
            goto out;
6944
        }
6945

J
Jiri Denemark 已提交
6946
        if (header->compressed != QEMUD_SAVE_FORMAT_RAW) {
6947
            intermediate_argv[0] = prog;
6948 6949
            intermediatefd = fd;
            fd = -1;
6950
            if (virExec(intermediate_argv, NULL, NULL,
6951
                        &intermediate_pid, intermediatefd, &fd, NULL, 0) < 0) {
6952 6953 6954
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("Failed to start decompression binary %s"),
                                intermediate_argv[0]);
J
Jiri Denemark 已提交
6955
                goto out;
6956 6957 6958
            }
        }
    }
J
Jiri Denemark 已提交
6959

6960
    /* Set the migration source and start it up. */
6961
    ret = qemudStartVMDaemon(conn, driver, vm, "stdio", true, fd, path);
J
Jiri Denemark 已提交
6962

6963 6964 6965
    if (intermediate_pid != -1) {
        /* Wait for intermediate process to exit */
        while (waitpid(intermediate_pid, &childstat, 0) == -1 &&
J
Jiri Denemark 已提交
6966 6967 6968
               errno == EINTR) {
            /* empty */
        }
6969 6970 6971
    }
    if (intermediatefd != -1)
        close(intermediatefd);
J
Jiri Denemark 已提交
6972 6973

    wait_ret = qemudDomainSaveImageClose(fd, read_pid, &status);
6974
    fd = -1;
6975 6976 6977 6978 6979 6980 6981 6982 6983 6984 6985 6986 6987 6988 6989 6990 6991 6992 6993 6994 6995
    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
6996

6997 6998
    if (ret < 0) {
        qemuDomainStartAudit(vm, "restored", false);
J
Jiri Denemark 已提交
6999
        goto out;
7000
    }
7001

7002 7003 7004
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_RESTORED);
7005
    qemuDomainStartAudit(vm, "restored", true);
J
Jiri Denemark 已提交
7006 7007 7008
    if (event)
        qemuDomainEventQueue(driver, event);

7009

7010
    /* If it was running before, resume it now. */
J
Jiri Denemark 已提交
7011
    if (header->was_running) {
7012
        qemuDomainObjPrivatePtr priv = vm->privateData;
7013
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
7014
        if (qemuMonitorStartCPUs(priv->mon, conn) < 0) {
7015
            if (virGetLastError() == NULL)
7016 7017
                qemuReportError(VIR_ERR_OPERATION_FAILED,
                                "%s", _("failed to resume domain"));
7018
            qemuDomainObjExitMonitorWithDriver(driver,vm);
J
Jiri Denemark 已提交
7019
            goto out;
7020
        }
7021
        qemuDomainObjExitMonitorWithDriver(driver, vm);
7022
        vm->state = VIR_DOMAIN_RUNNING;
7023 7024
        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
7025
            goto out;
7026
        }
7027
    }
J
Jiri Denemark 已提交
7028

7029
    ret = 0;
7030

J
Jiri Denemark 已提交
7031
out:
7032 7033
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSavedStateLabel &&
7034 7035
        driver->securityDriver->domainRestoreSavedStateLabel(driver->securityDriver,
                                                             vm, path) == -1)
7036 7037
        VIR_WARN("failed to restore save state label on %s", path);

J
Jiri Denemark 已提交
7038 7039 7040 7041 7042 7043 7044 7045 7046 7047 7048 7049 7050 7051 7052 7053 7054 7055 7056 7057 7058 7059 7060 7061 7062
    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))) {
7063
        /* virDomainAssignDef already set the error */
J
Jiri Denemark 已提交
7064 7065 7066 7067 7068 7069 7070 7071 7072 7073 7074
        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)
7075
        vm = NULL;
J
Jiri Denemark 已提交
7076 7077 7078 7079
    else if (ret < 0 && !vm->persistent) {
        virDomainRemoveInactive(&driver->domains, vm);
        vm = NULL;
    }
7080

7081 7082
cleanup:
    virDomainDefFree(def);
J
Jiri Denemark 已提交
7083
    qemudDomainSaveImageClose(fd, read_pid, NULL);
7084 7085 7086
    if (vm)
        virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);
7087
    return ret;
D
Daniel P. Berrange 已提交
7088 7089
}

J
Jiri Denemark 已提交
7090 7091 7092 7093 7094 7095 7096 7097 7098 7099 7100 7101 7102 7103 7104 7105 7106 7107 7108 7109 7110 7111 7112 7113 7114 7115 7116 7117 7118 7119 7120 7121 7122 7123 7124 7125 7126 7127 7128 7129 7130
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
7131

7132 7133 7134 7135 7136 7137 7138 7139 7140 7141 7142 7143 7144 7145 7146 7147 7148 7149 7150 7151 7152 7153 7154 7155 7156 7157 7158 7159 7160 7161 7162 7163 7164 7165 7166 7167 7168 7169
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;
}


7170
static char *qemudDomainDumpXML(virDomainPtr dom,
7171
                                int flags) {
7172 7173 7174
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    char *ret = NULL;
7175 7176
    unsigned long balloon;
    int err;
7177

7178
    qemuDriverLock(driver);
7179
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
7180

D
Daniel P. Berrange 已提交
7181
    if (!vm) {
7182 7183
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
7184 7185
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
7186
        goto cleanup;
D
Daniel P. Berrange 已提交
7187 7188
    }

7189 7190 7191 7192
    /* 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))) {
7193
        qemuDomainObjPrivatePtr priv = vm->privateData;
7194 7195 7196
        /* Don't delay if someone's using the monitor, just use
         * existing most recent data instead */
        if (!priv->jobActive) {
7197
            if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
7198 7199
                goto cleanup;

7200
            qemuDomainObjEnterMonitorWithDriver(driver, vm);
7201
            err = qemuMonitorGetBalloonInfo(priv->mon, &balloon);
7202
            qemuDomainObjExitMonitorWithDriver(driver, vm);
7203 7204 7205 7206
            if (qemuDomainObjEndJob(vm) == 0) {
                vm = NULL;
                goto cleanup;
            }
7207 7208 7209
            if (err < 0)
                goto cleanup;
            if (err > 0)
7210
                vm->def->mem.cur_balloon = balloon;
7211 7212
            /* err == 0 indicates no balloon support, so ignore it */
        }
7213
    }
7214

7215
    ret = qemudVMDumpXML(driver, vm, flags);
7216 7217

cleanup:
7218 7219
    if (vm)
        virDomainObjUnlock(vm);
7220
    qemuDriverUnlock(driver);
7221
    return ret;
D
Daniel P. Berrange 已提交
7222 7223 7224
}


7225 7226 7227 7228
static char *qemuDomainXMLFromNative(virConnectPtr conn,
                                     const char *format,
                                     const char *config,
                                     unsigned int flags ATTRIBUTE_UNUSED) {
7229
    struct qemud_driver *driver = conn->privateData;
7230 7231 7232 7233
    virDomainDefPtr def = NULL;
    char *xml = NULL;

    if (STRNEQ(format, QEMU_CONFIG_FORMAT_ARGV)) {
7234 7235
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("unsupported config type %s"), format);
7236 7237 7238
        goto cleanup;
    }

7239
    qemuDriverLock(driver);
7240
    def = qemuParseCommandLineString(driver->caps, config);
7241
    qemuDriverUnlock(driver);
7242 7243 7244
    if (!def)
        goto cleanup;

7245
    xml = virDomainDefFormat(def, VIR_DOMAIN_XML_INACTIVE);
7246 7247 7248 7249 7250 7251

cleanup:
    virDomainDefFree(def);
    return xml;
}

7252 7253 7254 7255 7256 7257
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;
7258
    virDomainChrDef monConfig;
7259
    const char *emulator;
7260
    unsigned long long qemuCmdFlags;
7261 7262 7263 7264 7265 7266 7267 7268
    struct stat sb;
    const char **retargv = NULL;
    const char **retenv = NULL;
    const char **tmp;
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    char *ret = NULL;
    int i;

7269 7270
    qemuDriverLock(driver);

7271
    if (STRNEQ(format, QEMU_CONFIG_FORMAT_ARGV)) {
7272 7273
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("unsupported config type %s"), format);
7274 7275 7276
        goto cleanup;
    }

7277
    def = virDomainDefParseString(driver->caps, xmlData, 0);
7278 7279 7280 7281 7282 7283 7284 7285 7286 7287 7288 7289 7290 7291 7292 7293 7294 7295 7296 7297 7298 7299 7300 7301 7302 7303 7304 7305 7306 7307 7308 7309 7310 7311 7312 7313 7314 7315 7316 7317 7318 7319 7320 7321
    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) {
7322
        virReportSystemError(errno,
7323 7324 7325 7326 7327 7328 7329 7330
                             _("Cannot find QEMU binary %s"),
                             emulator);
        goto cleanup;
    }

    if (qemudExtractVersionInfo(emulator,
                                NULL,
                                &qemuCmdFlags) < 0) {
7331 7332 7333
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Cannot determine QEMU argv syntax %s"),
                        emulator);
7334 7335 7336
        goto cleanup;
    }

7337
    if (qemuPrepareMonitorChr(driver, &monConfig, def->name) < 0)
7338
        goto cleanup;
7339 7340

    if (qemudBuildCommandLine(conn, driver, def,
D
Daniel P. Berrange 已提交
7341
                              &monConfig, 0, qemuCmdFlags,
7342 7343
                              &retargv, &retenv,
                              NULL, NULL, /* Don't want it to create TAP devices */
C
Chris Lalancette 已提交
7344
                              NULL, NULL) < 0) {
7345 7346 7347 7348 7349 7350 7351 7352 7353 7354 7355 7356 7357 7358 7359 7360
        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++;
    }

7361 7362
    if (virBufferError(&buf)) {
        virBufferFreeAndReset(&buf);
7363
        virReportOOMError();
7364
        goto cleanup;
7365
    }
7366 7367 7368 7369

    ret = virBufferContentAndReset(&buf);

cleanup:
7370
    qemuDriverUnlock(driver);
7371 7372 7373 7374 7375 7376 7377 7378 7379 7380 7381 7382 7383
    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;
}


7384
static int qemudListDefinedDomains(virConnectPtr conn,
7385
                            char **const names, int nnames) {
7386
    struct qemud_driver *driver = conn->privateData;
7387
    int n;
7388

7389
    qemuDriverLock(driver);
7390
    n = virDomainObjListGetInactiveNames(&driver->domains, names, nnames);
7391
    qemuDriverUnlock(driver);
7392
    return n;
D
Daniel P. Berrange 已提交
7393 7394
}

7395
static int qemudNumDefinedDomains(virConnectPtr conn) {
7396
    struct qemud_driver *driver = conn->privateData;
7397
    int n;
7398

7399
    qemuDriverLock(driver);
7400
    n = virDomainObjListNumOfDomains(&driver->domains, 0);
7401
    qemuDriverUnlock(driver);
7402

7403
    return n;
D
Daniel P. Berrange 已提交
7404 7405 7406
}


J
Jiri Denemark 已提交
7407 7408
static int qemudDomainObjStart(virConnectPtr conn,
                               struct qemud_driver *driver,
7409 7410
                               virDomainObjPtr vm,
                               bool start_paused)
J
Jiri Denemark 已提交
7411 7412 7413 7414 7415 7416 7417 7418 7419 7420 7421 7422 7423 7424 7425 7426 7427 7428 7429 7430
{
    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;
    }

7431
    ret = qemudStartVMDaemon(conn, driver, vm, NULL, start_paused, -1, NULL);
7432 7433
    qemuDomainStartAudit(vm, "booted", ret >= 0);
    if (ret >= 0) {
J
Jiri Denemark 已提交
7434 7435 7436 7437 7438 7439 7440 7441 7442 7443 7444 7445 7446
        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;
}

7447 7448 7449
static int
qemudDomainStartWithFlags(virDomainPtr dom, unsigned int flags)
{
7450 7451 7452
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
7453

7454
    virCheckFlags(VIR_DOMAIN_START_PAUSED, -1);
7455

7456
    qemuDriverLock(driver);
7457
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
7458

7459
    if (!vm) {
7460 7461
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
7462 7463
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
7464
        goto cleanup;
7465 7466
    }

7467 7468 7469 7470
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

    if (virDomainObjIsActive(vm)) {
7471 7472
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is already running"));
7473 7474 7475
        goto endjob;
    }

7476 7477
    ret = qemudDomainObjStart(dom->conn, driver, vm,
                              (flags & VIR_DOMAIN_START_PAUSED) != 0);
7478

7479
endjob:
7480 7481
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
7482

7483
cleanup:
7484 7485
    if (vm)
        virDomainObjUnlock(vm);
7486
    qemuDriverUnlock(driver);
7487
    return ret;
D
Daniel P. Berrange 已提交
7488 7489
}

7490 7491 7492 7493 7494 7495
static int
qemudDomainStart(virDomainPtr dom)
{
    return qemudDomainStartWithFlags(dom, 0);
}

7496 7497 7498 7499 7500 7501 7502 7503 7504 7505 7506 7507 7508 7509 7510
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;

7511
        if (STRNEQ(def->os.machine, machine->name))
7512 7513 7514
            continue;

        if (!(*canonical = strdup(machine->canonical))) {
7515
            virReportOOMError();
7516 7517 7518 7519 7520 7521 7522 7523 7524
            return -1;
        }

        break;
    }

    return 0;
}

7525 7526 7527 7528 7529 7530 7531
static int
qemudCanonicalizeMachineDirect(virDomainDefPtr def, char **canonical)
{
    virCapsGuestMachinePtr *machines = NULL;
    int i, nmachines = 0;

    if (qemudProbeMachineTypes(def->emulator, &machines, &nmachines) < 0) {
7532
        virReportOOMError();
7533 7534 7535 7536 7537 7538 7539
        return -1;
    }

    for (i = 0; i < nmachines; i++) {
        if (!machines[i]->canonical)
            continue;

7540
        if (STRNEQ(def->os.machine, machines[i]->name))
7541 7542 7543 7544 7545 7546 7547 7548 7549 7550 7551 7552
            continue;

        *canonical = machines[i]->canonical;
        machines[i]->canonical = NULL;
        break;
    }

    virCapabilitiesFreeMachines(machines, nmachines);

    return 0;
}

7553 7554
int
qemudCanonicalizeMachine(struct qemud_driver *driver, virDomainDefPtr def)
7555 7556 7557 7558 7559 7560
{
    char *canonical = NULL;
    int i;

    for (i = 0; i < driver->caps->nguests; i++) {
        virCapsGuestPtr guest = driver->caps->guests[i];
7561
        virCapsGuestDomainInfoPtr info;
7562 7563 7564
        int j;

        for (j = 0; j < guest->arch.ndomains; j++) {
7565
            info = &guest->arch.domains[j]->info;
7566

7567 7568 7569 7570 7571 7572 7573 7574 7575
            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;
7576 7577
        }

7578 7579 7580 7581
        info = &guest->arch.defaultInfo;

        if (info->emulator && STREQ(info->emulator, def->emulator)) {
            if (qemudCanonicalizeMachineFromInfo(def, info, &canonical) < 0)
7582 7583 7584 7585
                return -1;
            goto out;
        }
    }
7586 7587 7588 7589

    if (qemudCanonicalizeMachineDirect(def, &canonical) < 0)
        return -1;

7590 7591 7592 7593 7594 7595 7596
out:
    if (canonical) {
        VIR_FREE(def->os.machine);
        def->os.machine = canonical;
    }
    return 0;
}
D
Daniel P. Berrange 已提交
7597

7598
static virDomainPtr qemudDomainDefine(virConnectPtr conn, const char *xml) {
7599
    struct qemud_driver *driver = conn->privateData;
7600
    virDomainDefPtr def;
7601
    virDomainObjPtr vm = NULL;
7602
    virDomainPtr dom = NULL;
7603
    virDomainEventPtr event = NULL;
7604
    int dupVM;
7605

7606
    qemuDriverLock(driver);
7607
    if (!(def = virDomainDefParseString(driver->caps, xml,
7608
                                        VIR_DOMAIN_XML_INACTIVE)))
7609
        goto cleanup;
7610

7611
    if (virSecurityDriverVerify(def) < 0)
7612 7613
        goto cleanup;

7614 7615
    if ((dupVM = virDomainObjIsDuplicate(&driver->domains, def, 0)) < 0)
        goto cleanup;
7616

7617
    if (qemudCanonicalizeMachine(driver, def) < 0)
7618 7619
        goto cleanup;

7620 7621 7622
    if (qemuAssignPCIAddresses(def) < 0)
        goto cleanup;

7623
    if (!(vm = virDomainAssignDef(driver->caps,
7624
                                  &driver->domains,
7625
                                  def, false))) {
7626
        goto cleanup;
7627
    }
7628
    def = NULL;
7629
    vm->persistent = 1;
7630

7631
    if (virDomainSaveConfig(driver->configDir,
7632
                            vm->newDef ? vm->newDef : vm->def) < 0) {
7633 7634
        virDomainRemoveInactive(&driver->domains,
                                vm);
7635
        vm = NULL;
7636
        goto cleanup;
7637 7638
    }

7639 7640
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_DEFINED,
7641
                                     !dupVM ?
7642 7643
                                     VIR_DOMAIN_EVENT_DEFINED_ADDED :
                                     VIR_DOMAIN_EVENT_DEFINED_UPDATED);
7644

7645
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
7646
    if (dom) dom->id = vm->def->id;
7647 7648

cleanup:
7649
    virDomainDefFree(def);
7650 7651
    if (vm)
        virDomainObjUnlock(vm);
7652 7653
    if (event)
        qemuDomainEventQueue(driver, event);
7654
    qemuDriverUnlock(driver);
7655
    return dom;
D
Daniel P. Berrange 已提交
7656 7657
}

7658
static int qemudDomainUndefine(virDomainPtr dom) {
7659 7660
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
7661
    virDomainEventPtr event = NULL;
7662
    int ret = -1;
D
Daniel P. Berrange 已提交
7663

7664
    qemuDriverLock(driver);
7665
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
7666

D
Daniel P. Berrange 已提交
7667
    if (!vm) {
7668 7669
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
7670 7671
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
7672
        goto cleanup;
D
Daniel P. Berrange 已提交
7673 7674
    }

D
Daniel P. Berrange 已提交
7675
    if (virDomainObjIsActive(vm)) {
7676 7677
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot delete active domain"));
7678
        goto cleanup;
D
Daniel P. Berrange 已提交
7679 7680
    }

7681
    if (!vm->persistent) {
7682 7683
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("cannot undefine transient domain"));
7684
        goto cleanup;
7685 7686
    }

7687
    if (virDomainDeleteConfig(driver->configDir, driver->autostartDir, vm) < 0)
7688
        goto cleanup;
D
Daniel P. Berrange 已提交
7689

7690 7691 7692
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_UNDEFINED,
                                     VIR_DOMAIN_EVENT_UNDEFINED_REMOVED);
7693

7694 7695
    virDomainRemoveInactive(&driver->domains,
                            vm);
7696
    vm = NULL;
7697
    ret = 0;
D
Daniel P. Berrange 已提交
7698

7699
cleanup:
7700 7701
    if (vm)
        virDomainObjUnlock(vm);
7702 7703
    if (event)
        qemuDomainEventQueue(driver, event);
7704
    qemuDriverUnlock(driver);
7705
    return ret;
D
Daniel P. Berrange 已提交
7706 7707
}

7708

7709
static int qemudDomainChangeEjectableMedia(struct qemud_driver *driver,
7710
                                           virDomainObjPtr vm,
7711 7712
                                           virDomainDiskDefPtr disk,
                                           unsigned long long qemuCmdFlags)
7713
{
7714
    virDomainDiskDefPtr origdisk = NULL;
7715
    int i;
7716
    int ret;
7717
    char *driveAlias = NULL;
7718

7719
    origdisk = NULL;
7720
    for (i = 0 ; i < vm->def->ndisks ; i++) {
7721 7722
        if (vm->def->disks[i]->bus == disk->bus &&
            STREQ(vm->def->disks[i]->dst, disk->dst)) {
7723
            origdisk = vm->def->disks[i];
7724
            break;
7725
        }
7726 7727 7728
    }

    if (!origdisk) {
7729 7730 7731 7732
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("No device with bus '%s' and target '%s'"),
                        virDomainDiskBusTypeToString(disk->bus),
                        disk->dst);
7733 7734 7735
        return -1;
    }

7736
    if (!origdisk->info.alias) {
7737 7738
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("missing disk device alias name for %s"), origdisk->dst);
7739 7740
        return -1;
    }
7741

7742 7743
    if (origdisk->device != VIR_DOMAIN_DISK_DEVICE_FLOPPY &&
        origdisk->device != VIR_DOMAIN_DISK_DEVICE_CDROM) {
7744 7745 7746
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Removable media not supported for %s device"),
                        virDomainDiskDeviceTypeToString(disk->device));
7747
        return -1;
7748
    }
7749

7750 7751
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityImageLabel &&
7752 7753
        driver->securityDriver->domainSetSecurityImageLabel(driver->securityDriver,
                                                            vm, disk) < 0)
7754 7755
        return -1;

7756 7757 7758
    if (!(driveAlias = qemuDeviceDriveHostAlias(origdisk, qemuCmdFlags)))
        goto error;

7759
    qemuDomainObjPrivatePtr priv = vm->privateData;
7760
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7761
    if (disk->src) {
7762
        const char *format = NULL;
7763 7764 7765
        if (disk->type != VIR_DOMAIN_DISK_TYPE_DIR) {
            if (disk->driverType)
                format = disk->driverType;
7766 7767 7768
            else if (origdisk->driverType)
                format = origdisk->driverType;
        }
7769
        ret = qemuMonitorChangeMedia(priv->mon,
7770
                                     driveAlias,
7771
                                     disk->src, format);
7772
    } else {
7773
        ret = qemuMonitorEjectMedia(priv->mon, driveAlias);
7774
    }
7775
    qemuDomainObjExitMonitorWithDriver(driver, vm);
7776

7777 7778
    qemuDomainDiskAudit(vm, origdisk, disk, "update", ret >= 0);

7779 7780 7781 7782 7783
    if (ret < 0)
        goto error;

    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
7784 7785
        driver->securityDriver->domainRestoreSecurityImageLabel(driver->securityDriver,
                                                                vm, origdisk) < 0)
7786 7787 7788
        VIR_WARN("Unable to restore security label on ejected image %s", origdisk->src);

    VIR_FREE(origdisk->src);
7789 7790 7791 7792
    origdisk->src = disk->src;
    disk->src = NULL;
    origdisk->type = disk->type;

7793 7794
    VIR_FREE(driveAlias);

7795
    virDomainDiskDefFree(disk);
7796

7797
    return ret;
7798 7799

error:
7800
    VIR_FREE(driveAlias);
7801 7802
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
7803 7804
        driver->securityDriver->domainRestoreSecurityImageLabel(driver->securityDriver,
                                                                vm, disk) < 0)
7805
        VIR_WARN("Unable to restore security label on new media %s", disk->src);
7806
    return -1;
7807 7808
}

7809

7810
static int qemudDomainAttachPciDiskDevice(struct qemud_driver *driver,
7811
                                          virDomainObjPtr vm,
7812
                                          virDomainDiskDefPtr disk,
7813
                                          unsigned long long qemuCmdFlags)
7814
{
7815
    int i, ret;
7816
    const char* type = virDomainDiskBusTypeToString(disk->bus);
7817
    qemuDomainObjPrivatePtr priv = vm->privateData;
7818 7819
    char *devstr = NULL;
    char *drivestr = NULL;
7820 7821

    for (i = 0 ; i < vm->def->ndisks ; i++) {
7822
        if (STREQ(vm->def->disks[i]->dst, disk->dst)) {
7823 7824
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("target %s already exists"), disk->dst);
7825 7826 7827 7828
            return -1;
        }
    }

7829 7830
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityImageLabel &&
7831 7832
        driver->securityDriver->domainSetSecurityImageLabel(driver->securityDriver,
                                                            vm, disk) < 0)
7833 7834
        return -1;

7835
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
7836 7837
        if (qemuDomainPCIAddressEnsureAddr(priv->pciaddrs, &disk->info) < 0)
            goto error;
7838 7839
        if (qemuAssignDeviceDiskAlias(disk, qemuCmdFlags) < 0)
            goto error;
7840

7841 7842 7843
        if (!(drivestr = qemuBuildDriveStr(disk, 0, qemuCmdFlags)))
            goto error;

7844
        if (!(devstr = qemuBuildDriveDevStr(disk)))
7845 7846 7847
            goto error;
    }

7848
    if (VIR_REALLOC_N(vm->def->disks, vm->def->ndisks+1) < 0) {
7849
        virReportOOMError();
7850
        goto error;
7851 7852
    }

7853
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7854 7855
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        ret = qemuMonitorAddDrive(priv->mon, drivestr);
7856 7857 7858
        if (ret == 0) {
            ret = qemuMonitorAddDevice(priv->mon, devstr);
            if (ret < 0) {
7859
                VIR_WARN("qemuMonitorAddDevice failed on %s (%s)",
7860 7861 7862 7863 7864
                         drivestr, devstr);
                /* XXX should call 'drive_del' on error but this does not
                   exist yet */
            }
        }
7865 7866 7867 7868 7869 7870 7871 7872 7873 7874 7875
    } 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));
        }
    }
7876
    qemuDomainObjExitMonitorWithDriver(driver, vm);
7877

7878 7879
    qemuDomainDiskAudit(vm, NULL, disk, "attach", ret >= 0);

7880 7881
    if (ret < 0)
        goto error;
7882

7883
    virDomainDiskInsertPreAlloced(vm->def, disk);
7884

7885 7886 7887
    VIR_FREE(devstr);
    VIR_FREE(drivestr);

7888 7889 7890
    return 0;

error:
7891 7892 7893
    VIR_FREE(devstr);
    VIR_FREE(drivestr);

7894 7895 7896 7897 7898
    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);

7899 7900
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
7901 7902
        driver->securityDriver->domainRestoreSecurityImageLabel(driver->securityDriver,
                                                                vm, disk) < 0)
7903
        VIR_WARN("Unable to restore security label on %s", disk->src);
7904 7905

    return -1;
7906
}
7907

7908

7909
static int qemudDomainAttachPciControllerDevice(struct qemud_driver *driver,
7910
                                                virDomainObjPtr vm,
7911
                                                virDomainControllerDefPtr controller,
7912
                                                unsigned long long qemuCmdFlags)
7913
{
7914 7915
    int i;
    int ret = -1;
7916
    const char* type = virDomainControllerTypeToString(controller->type);
7917
    char *devstr = NULL;
7918 7919 7920
    qemuDomainObjPrivatePtr priv = vm->privateData;

    for (i = 0 ; i < vm->def->ncontrollers ; i++) {
7921 7922
        if ((vm->def->controllers[i]->type == controller->type) &&
            (vm->def->controllers[i]->idx == controller->idx)) {
7923 7924 7925
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("target %s:%d already exists"),
                            type, controller->idx);
7926 7927 7928 7929
            return -1;
        }
    }

7930 7931 7932 7933 7934 7935
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuDomainPCIAddressEnsureAddr(priv->pciaddrs, &controller->info) < 0)
            goto cleanup;
        if (qemuAssignDeviceControllerAlias(controller) < 0)
            goto cleanup;
    }
7936

7937
    if (!(devstr = qemuBuildControllerDevStr(controller))) {
7938
        virReportOOMError();
7939 7940 7941
        goto cleanup;
    }

7942
    if (VIR_REALLOC_N(vm->def->controllers, vm->def->ncontrollers+1) < 0) {
7943
        virReportOOMError();
7944
        goto cleanup;
7945 7946 7947
    }

    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7948 7949 7950 7951 7952 7953 7954
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        ret = qemuMonitorAddDevice(priv->mon, devstr);
    } else {
        ret = qemuMonitorAttachPCIDiskController(priv->mon,
                                                 type,
                                                 &controller->info.addr.pci);
    }
7955 7956 7957
    qemuDomainObjExitMonitorWithDriver(driver, vm);

    if (ret == 0) {
7958 7959
        controller->info.type = VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI;
        virDomainControllerInsertPreAlloced(vm->def, controller);
7960 7961
    }

7962
cleanup:
7963 7964 7965 7966 7967 7968
    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");

7969
    VIR_FREE(devstr);
7970 7971 7972
    return ret;
}

7973

7974
static virDomainControllerDefPtr
7975
qemuDomainFindOrCreateSCSIDiskController(struct qemud_driver *driver,
7976
                                         virDomainObjPtr vm,
7977
                                         int controller,
7978
                                         unsigned long long qemuCmdFlags)
7979 7980 7981 7982 7983 7984 7985 7986 7987 7988 7989 7990 7991
{
    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
7992
    /* No SCSI controller present, for backward compatibility we
7993 7994
     * now hotplug a controller */
    if (VIR_ALLOC(cont) < 0) {
7995
        virReportOOMError();
7996 7997 7998 7999
        return NULL;
    }
    cont->type = VIR_DOMAIN_CONTROLLER_TYPE_SCSI;
    cont->idx = 0;
8000
    cont->model = -1;
8001 8002

    VIR_INFO0("No SCSI controller present, hotplugging one");
8003
    if (qemudDomainAttachPciControllerDevice(driver,
8004
                                             vm, cont, qemuCmdFlags) < 0) {
8005 8006 8007
        VIR_FREE(cont);
        return NULL;
    }
8008 8009 8010 8011 8012 8013 8014 8015 8016

    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;
    }

8017 8018 8019
    return cont;
}

8020

8021
static int qemudDomainAttachSCSIDisk(struct qemud_driver *driver,
8022
                                     virDomainObjPtr vm,
8023
                                     virDomainDiskDefPtr disk,
8024
                                     unsigned long long qemuCmdFlags)
8025 8026 8027
{
    int i;
    qemuDomainObjPrivatePtr priv = vm->privateData;
8028
    virDomainControllerDefPtr cont = NULL;
8029
    char *drivestr = NULL;
8030
    char *devstr = NULL;
8031 8032 8033
    int ret = -1;

    for (i = 0 ; i < vm->def->ndisks ; i++) {
8034
        if (STREQ(vm->def->disks[i]->dst, disk->dst)) {
8035 8036
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("target %s already exists"), disk->dst);
8037
            return -1;
8038 8039 8040
        }
    }

8041 8042 8043

    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityImageLabel &&
8044 8045
        driver->securityDriver->domainSetSecurityImageLabel(driver->securityDriver,
                                                            vm, disk) < 0)
8046 8047
        return -1;

8048
    /* We should have an address already, so make sure */
8049
    if (disk->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_DRIVE) {
8050 8051 8052
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("unexpected disk address type %s"),
                        virDomainDeviceAddressTypeToString(disk->info.type));
8053
        goto error;
8054 8055
    }

8056 8057 8058
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuAssignDeviceDiskAlias(disk, qemuCmdFlags) < 0)
            goto error;
8059
        if (!(devstr = qemuBuildDriveDevStr(disk)))
8060 8061 8062 8063 8064 8065
            goto error;
    }

    if (!(drivestr = qemuBuildDriveStr(disk, 0, qemuCmdFlags)))
        goto error;

8066
    for (i = 0 ; i <= disk->info.addr.drive.controller ; i++) {
8067
        cont = qemuDomainFindOrCreateSCSIDiskController(driver, vm, i, qemuCmdFlags);
8068
        if (!cont)
8069
            goto error;
8070 8071
    }

8072 8073 8074 8075 8076
    /* 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);

8077
    if (cont->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI) {
8078 8079
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("SCSI controller %d was missing its PCI address"), cont->idx);
8080
        goto error;
8081 8082 8083
    }

    if (VIR_REALLOC_N(vm->def->disks, vm->def->ndisks+1) < 0) {
8084
        virReportOOMError();
8085
        goto error;
8086 8087 8088
    }

    qemuDomainObjEnterMonitorWithDriver(driver, vm);
8089
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
8090 8091 8092 8093
        ret = qemuMonitorAddDrive(priv->mon, drivestr);
        if (ret == 0) {
            ret = qemuMonitorAddDevice(priv->mon, devstr);
            if (ret < 0) {
8094
                VIR_WARN("qemuMonitorAddDevice failed on %s (%s)",
8095 8096 8097 8098 8099
                         drivestr, devstr);
                /* XXX should call 'drive_del' on error but this does not
                   exist yet */
            }
        }
8100 8101 8102 8103 8104 8105 8106 8107 8108 8109 8110 8111 8112
    } 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));
        }
    }
8113 8114
    qemuDomainObjExitMonitorWithDriver(driver, vm);

8115 8116
    qemuDomainDiskAudit(vm, NULL, disk, "attach", ret >= 0);

8117 8118
    if (ret < 0)
        goto error;
8119

8120
    virDomainDiskInsertPreAlloced(vm->def, disk);
8121 8122

    VIR_FREE(devstr);
8123
    VIR_FREE(drivestr);
8124 8125 8126 8127

    return 0;

error:
8128
    VIR_FREE(devstr);
8129
    VIR_FREE(drivestr);
8130

8131 8132
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
8133 8134
        driver->securityDriver->domainRestoreSecurityImageLabel(driver->securityDriver,
                                                                vm, disk) < 0)
8135
        VIR_WARN("Unable to restore security label on %s", disk->src);
8136 8137

    return -1;
8138 8139
}

8140

8141
static int qemudDomainAttachUsbMassstorageDevice(struct qemud_driver *driver,
8142
                                                 virDomainObjPtr vm,
8143
                                                 virDomainDiskDefPtr disk,
8144
                                                 unsigned long long qemuCmdFlags)
8145
{
8146
    qemuDomainObjPrivatePtr priv = vm->privateData;
8147
    int i, ret;
8148 8149
    char *drivestr = NULL;
    char *devstr = NULL;
8150

8151
    for (i = 0 ; i < vm->def->ndisks ; i++) {
8152
        if (STREQ(vm->def->disks[i]->dst, disk->dst)) {
8153 8154
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("target %s already exists"), disk->dst);
8155 8156 8157 8158
            return -1;
        }
    }

8159 8160
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityImageLabel &&
8161 8162
        driver->securityDriver->domainSetSecurityImageLabel(driver->securityDriver,
                                                            vm, disk) < 0)
8163 8164
        return -1;

8165
    if (!disk->src) {
8166 8167
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("disk source path is missing"));
8168
        goto error;
8169 8170
    }

8171
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
8172 8173
        if (qemuAssignDeviceDiskAlias(disk, qemuCmdFlags) < 0)
            goto error;
8174 8175
        if (!(drivestr = qemuBuildDriveStr(disk, 0, qemuCmdFlags)))
            goto error;
8176
        if (!(devstr = qemuBuildDriveDevStr(disk)))
8177 8178 8179
            goto error;
    }

8180
    if (VIR_REALLOC_N(vm->def->disks, vm->def->ndisks+1) < 0) {
8181
        virReportOOMError();
8182
        goto error;
8183 8184
    }

8185
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
8186
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
8187 8188 8189 8190
        ret = qemuMonitorAddDrive(priv->mon, drivestr);
        if (ret == 0) {
            ret = qemuMonitorAddDevice(priv->mon, devstr);
            if (ret < 0) {
8191
                VIR_WARN("qemuMonitorAddDevice failed on %s (%s)",
8192 8193 8194 8195 8196
                         drivestr, devstr);
                /* XXX should call 'drive_del' on error but this does not
                   exist yet */
            }
        }
8197 8198 8199
    } else {
        ret = qemuMonitorAddUSBDisk(priv->mon, disk->src);
    }
8200
    qemuDomainObjExitMonitorWithDriver(driver, vm);
8201

8202 8203
    qemuDomainDiskAudit(vm, NULL, disk, "attach", ret >= 0);

8204 8205 8206
    if (ret < 0)
        goto error;

8207
    virDomainDiskInsertPreAlloced(vm->def, disk);
8208

8209 8210 8211
    VIR_FREE(devstr);
    VIR_FREE(drivestr);

8212 8213 8214
    return 0;

error:
8215 8216 8217
    VIR_FREE(devstr);
    VIR_FREE(drivestr);

8218 8219
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
8220 8221
        driver->securityDriver->domainRestoreSecurityImageLabel(driver->securityDriver,
                                                                vm, disk) < 0)
8222
        VIR_WARN("Unable to restore security label on %s", disk->src);
8223 8224

    return -1;
8225 8226
}

8227

8228
/* XXX conn required for network -> bridge resolution */
M
Mark McLoughlin 已提交
8229
static int qemudDomainAttachNetDevice(virConnectPtr conn,
8230
                                      struct qemud_driver *driver,
M
Mark McLoughlin 已提交
8231
                                      virDomainObjPtr vm,
8232
                                      virDomainNetDefPtr net,
8233
                                      unsigned long long qemuCmdFlags)
M
Mark McLoughlin 已提交
8234
{
8235
    qemuDomainObjPrivatePtr priv = vm->privateData;
8236
    char *tapfd_name = NULL;
8237
    int tapfd = -1;
8238
    char *nicstr = NULL;
8239
    char *netstr = NULL;
8240
    int ret = -1;
8241
    virDomainDevicePCIAddress guestAddr;
8242
    int vlan;
M
Mark McLoughlin 已提交
8243 8244

    if (!(qemuCmdFlags & QEMUD_CMD_FLAG_HOST_NET_ADD)) {
8245
        qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
8246
                        _("installed qemu version does not support host_net_add"));
M
Mark McLoughlin 已提交
8247 8248 8249 8250 8251
        return -1;
    }

    if (net->type == VIR_DOMAIN_NET_TYPE_BRIDGE ||
        net->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
8252
        if (priv->monConfig->type != VIR_DOMAIN_CHR_TYPE_UNIX) {
8253
            qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
8254 8255 8256
                            _("network device type '%s' cannot be attached: "
                              "qemu is not using a unix socket monitor"),
                            virDomainNetTypeToString(net->type));
8257 8258 8259 8260 8261
            return -1;
        }

        if ((tapfd = qemudNetworkIfaceConnect(conn, driver, net, qemuCmdFlags)) < 0)
            return -1;
8262 8263
    } else if (net->type == VIR_DOMAIN_NET_TYPE_DIRECT) {
        if (priv->monConfig->type != VIR_DOMAIN_CHR_TYPE_UNIX) {
8264
            qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
8265 8266 8267 8268 8269 8270
                            _("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
8271
        if ((tapfd = qemudPhysIfaceConnect(conn, driver, net,
8272 8273
                                           qemuCmdFlags,
                                           vm->def->uuid)) < 0)
8274
            return -1;
M
Mark McLoughlin 已提交
8275 8276
    }

8277 8278
    if (VIR_REALLOC_N(vm->def->nets, vm->def->nnets+1) < 0)
        goto no_memory;
M
Mark McLoughlin 已提交
8279

8280 8281 8282 8283 8284
    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
8285

8286 8287 8288 8289
    if ((qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) &&
        qemuDomainPCIAddressEnsureAddr(priv->pciaddrs, &net->info) < 0)
        goto cleanup;

8290 8291 8292 8293 8294
    if ((qemuCmdFlags & QEMUD_CMD_FLAG_NETDEV) &&
        (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)) {
        vlan = -1;
    } else {
        vlan = qemuDomainNetVLAN(net);
M
Mark McLoughlin 已提交
8295

8296
        if (vlan < 0) {
8297
            qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
8298 8299 8300
                            _("Unable to attach network devices without vlan"));
            goto cleanup;
        }
8301 8302
    }

8303
    if (tapfd != -1) {
8304
        if (virAsprintf(&tapfd_name, "fd-%s", net->info.alias) < 0)
8305 8306
            goto no_memory;

8307
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
8308
        if (qemuMonitorSendFileHandle(priv->mon, tapfd_name, tapfd) < 0) {
8309
            qemuDomainObjExitMonitorWithDriver(driver, vm);
8310
            goto cleanup;
8311
        }
8312
        qemuDomainObjExitMonitorWithDriver(driver, vm);
8313 8314 8315 8316 8317 8318

        if (!virDomainObjIsActive(vm)) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("guest unexpectedly quit"));
            goto cleanup;
        }
8319 8320
    }

8321
    /* FIXME - need to support vhost-net here (5th arg) */
8322 8323 8324 8325 8326 8327 8328 8329 8330 8331
    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
8332

8333
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
8334 8335 8336 8337
    if ((qemuCmdFlags & QEMUD_CMD_FLAG_NETDEV) &&
        (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)) {
        if (qemuMonitorAddNetdev(priv->mon, netstr) < 0) {
            qemuDomainObjExitMonitorWithDriver(driver, vm);
8338
            qemuDomainNetAudit(vm, NULL, net, "attach", false);
8339 8340 8341 8342 8343
            goto try_tapfd_close;
        }
    } else {
        if (qemuMonitorAddHostNetwork(priv->mon, netstr) < 0) {
            qemuDomainObjExitMonitorWithDriver(driver, vm);
8344
            qemuDomainNetAudit(vm, NULL, net, "attach", false);
8345 8346
            goto try_tapfd_close;
        }
8347
    }
8348
    qemuDomainObjExitMonitorWithDriver(driver, vm);
M
Mark McLoughlin 已提交
8349

8350 8351 8352
    if (tapfd != -1)
        close(tapfd);
    tapfd = -1;
M
Mark McLoughlin 已提交
8353

8354 8355 8356 8357 8358 8359
    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("guest unexpectedly quit"));
        goto cleanup;
    }

8360 8361 8362 8363 8364 8365 8366
    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
8367

8368
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
8369 8370 8371
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuMonitorAddDevice(priv->mon, nicstr) < 0) {
            qemuDomainObjExitMonitorWithDriver(driver, vm);
8372
            qemuDomainNetAudit(vm, NULL, net, "attach", false);
8373 8374 8375 8376 8377 8378
            goto try_remove;
        }
    } else {
        if (qemuMonitorAddPCINetwork(priv->mon, nicstr,
                                     &guestAddr) < 0) {
            qemuDomainObjExitMonitorWithDriver(driver, vm);
8379
            qemuDomainNetAudit(vm, NULL, net, "attach", false);
8380 8381 8382 8383
            goto try_remove;
        }
        net->info.type = VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI;
        memcpy(&net->info.addr.pci, &guestAddr, sizeof(guestAddr));
8384
    }
8385
    qemuDomainObjExitMonitorWithDriver(driver, vm);
8386

8387 8388
    qemuDomainNetAudit(vm, NULL, net, "attach", true);

8389
    ret = 0;
M
Mark McLoughlin 已提交
8390 8391 8392

    vm->def->nets[vm->def->nnets++] = net;

8393
cleanup:
8394 8395 8396 8397 8398 8399
    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");

8400
    if (ret != 0)
8401
        virDomainConfNWFilterTeardown(net);
S
Stefan Berger 已提交
8402

8403 8404 8405 8406 8407
    VIR_FREE(nicstr);
    VIR_FREE(netstr);
    VIR_FREE(tapfd_name);
    if (tapfd != -1)
        close(tapfd);
8408

8409
    return ret;
8410

8411
try_remove:
8412
    if (!virDomainObjIsActive(vm))
8413 8414
        goto cleanup;

8415
    if (vlan < 0) {
8416 8417 8418 8419 8420 8421 8422
        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)
8423
                VIR_WARN("Failed to remove network backend for netdev %s",
8424 8425 8426 8427
                         netdev_name);
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            VIR_FREE(netdev_name);
        } else {
8428
            VIR_WARN0("Unable to remove network backend");
8429
        }
8430 8431 8432 8433
    } else {
        char *hostnet_name;
        if (virAsprintf(&hostnet_name, "host%s", net->info.alias) < 0)
            goto no_memory;
8434
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
8435
        if (qemuMonitorRemoveHostNetwork(priv->mon, vlan, hostnet_name) < 0)
8436
            VIR_WARN("Failed to remove network backend for vlan %d, net %s",
8437
                     vlan, hostnet_name);
8438
        qemuDomainObjExitMonitorWithDriver(driver, vm);
8439
        VIR_FREE(hostnet_name);
8440
    }
8441
    goto cleanup;
8442

8443
try_tapfd_close:
8444
    if (!virDomainObjIsActive(vm))
8445 8446
        goto cleanup;

8447
    if (tapfd_name) {
8448
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
8449
        if (qemuMonitorCloseFileHandle(priv->mon, tapfd_name) < 0)
8450
            VIR_WARN("Failed to close tapfd with '%s'", tapfd_name);
8451
        qemuDomainObjExitMonitorWithDriver(driver, vm);
8452
    }
8453

8454 8455
    goto cleanup;

8456
no_memory:
8457
    virReportOOMError();
8458
    goto cleanup;
M
Mark McLoughlin 已提交
8459 8460
}

8461

8462
static int qemudDomainAttachHostPciDevice(struct qemud_driver *driver,
8463
                                          virDomainObjPtr vm,
8464
                                          virDomainHostdevDefPtr hostdev,
8465
                                          unsigned long long qemuCmdFlags)
8466
{
8467
    qemuDomainObjPrivatePtr priv = vm->privateData;
8468
    int ret;
8469
    char *devstr = NULL;
8470 8471
    int configfd = -1;
    char *configfd_name = NULL;
8472 8473

    if (VIR_REALLOC_N(vm->def->hostdevs, vm->def->nhostdevs+1) < 0) {
8474
        virReportOOMError();
8475 8476 8477
        return -1;
    }

8478
    if (qemuPrepareHostdevPCIDevices(driver, &hostdev, 1) < 0)
8479
        return -1;
8480

8481
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
8482 8483
        if (qemuAssignDeviceHostdevAlias(vm->def, hostdev, -1) < 0)
            goto error;
8484 8485
        if (qemuDomainPCIAddressEnsureAddr(priv->pciaddrs, &hostdev->info) < 0)
            goto error;
8486 8487 8488 8489 8490 8491 8492 8493 8494 8495 8496 8497 8498 8499 8500 8501 8502 8503 8504 8505 8506 8507 8508 8509
        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;
        }
8510

8511
        if (!(devstr = qemuBuildPCIHostdevDevStr(hostdev, configfd_name)))
8512
            goto error;
8513

8514
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
8515
        ret = qemuMonitorAddDevice(priv->mon, devstr);
8516 8517 8518 8519 8520
        qemuDomainObjExitMonitorWithDriver(driver, vm);
    } else {
        virDomainDevicePCIAddress guestAddr;

        qemuDomainObjEnterMonitorWithDriver(driver, vm);
8521 8522 8523
        ret = qemuMonitorAddPCIHostDevice(priv->mon,
                                          &hostdev->source.subsys.u.pci,
                                          &guestAddr);
8524 8525 8526 8527 8528
        qemuDomainObjExitMonitorWithDriver(driver, vm);

        hostdev->info.type = VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI;
        memcpy(&hostdev->info.addr.pci, &guestAddr, sizeof(guestAddr));
    }
8529
    if (ret < 0)
8530
        goto error;
8531 8532 8533

    vm->def->hostdevs[vm->def->nhostdevs++] = hostdev;

8534
    VIR_FREE(devstr);
8535 8536 8537
    VIR_FREE(configfd_name);
    if (configfd >= 0)
        close(configfd);
8538

8539
    return 0;
8540 8541

error:
8542 8543 8544 8545 8546
    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");

8547
    qemuDomainReAttachHostdevDevices(driver, &hostdev, 1);
8548 8549

    VIR_FREE(devstr);
8550 8551 8552
    VIR_FREE(configfd_name);
    if (configfd >= 0)
        close(configfd);
8553 8554

    return -1;
8555 8556
}

8557

8558
static int qemudDomainAttachHostUsbDevice(struct qemud_driver *driver,
M
Mark McLoughlin 已提交
8559
                                          virDomainObjPtr vm,
8560
                                          virDomainHostdevDefPtr hostdev,
8561
                                          unsigned long long qemuCmdFlags)
8562 8563
{
    int ret;
8564
    qemuDomainObjPrivatePtr priv = vm->privateData;
8565 8566
    char *devstr = NULL;

8567 8568 8569 8570 8571 8572
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuAssignDeviceHostdevAlias(vm->def, hostdev, -1) < 0)
            goto error;
        if (!(devstr = qemuBuildUSBHostdevDevStr(hostdev)))
            goto error;
    }
8573

8574
    if (VIR_REALLOC_N(vm->def->hostdevs, vm->def->nhostdevs+1) < 0) {
8575
        virReportOOMError();
8576
        goto error;
8577
    }
8578

8579 8580 8581 8582 8583 8584 8585 8586 8587 8588 8589 8590 8591 8592 8593 8594 8595 8596 8597
    if (qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_DEVICES)) {
        virCgroupPtr cgroup = NULL;
        usbDevice *usb;

        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 error;
        }

        if ((usb = usbGetDevice(hostdev->source.subsys.u.usb.bus,
                                hostdev->source.subsys.u.usb.device)) == NULL)
            goto error;

        if (usbDeviceFileIterate(usb, qemuSetupHostUsbDeviceCgroup, cgroup) < 0 )
            goto error;
    }

8598
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
8599 8600 8601
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)
        ret = qemuMonitorAddDevice(priv->mon, devstr);
    else
8602
        ret = qemuMonitorAddUSBDeviceExact(priv->mon,
8603 8604
                                           hostdev->source.subsys.u.usb.bus,
                                           hostdev->source.subsys.u.usb.device);
8605
    qemuDomainObjExitMonitorWithDriver(driver, vm);
8606 8607 8608 8609
    if (ret < 0)
        goto error;

    vm->def->hostdevs[vm->def->nhostdevs++] = hostdev;
8610

8611
    VIR_FREE(devstr);
8612

8613 8614 8615 8616 8617
    return 0;

error:
    VIR_FREE(devstr);
    return -1;
8618 8619
}

8620

8621
static int qemudDomainAttachHostDevice(struct qemud_driver *driver,
M
Mark McLoughlin 已提交
8622
                                       virDomainObjPtr vm,
8623
                                       virDomainHostdevDefPtr hostdev,
8624
                                       unsigned long long qemuCmdFlags)
M
Mark McLoughlin 已提交
8625 8626
{
    if (hostdev->mode != VIR_DOMAIN_HOSTDEV_MODE_SUBSYS) {
8627
        qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
8628 8629
                        _("hostdev mode '%s' not supported"),
                        virDomainHostdevModeTypeToString(hostdev->mode));
M
Mark McLoughlin 已提交
8630 8631 8632
        return -1;
    }

8633 8634 8635 8636 8637 8638 8639 8640 8641 8642 8643 8644 8645 8646 8647 8648 8649
    /* 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);
    }


8650
    if (driver->securityDriver &&
8651
        driver->securityDriver->domainSetSecurityHostdevLabel &&
8652 8653
        driver->securityDriver->domainSetSecurityHostdevLabel(driver->securityDriver,
                                                              vm, hostdev) < 0)
8654
        return -1;
M
Mark McLoughlin 已提交
8655 8656

    switch (hostdev->source.subsys.type) {
8657
    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI:
8658
        if (qemudDomainAttachHostPciDevice(driver, vm,
8659
                                           hostdev, qemuCmdFlags) < 0)
8660 8661 8662
            goto error;
        break;

M
Mark McLoughlin 已提交
8663
    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB:
8664
        if (qemudDomainAttachHostUsbDevice(driver, vm,
8665
                                           hostdev, qemuCmdFlags) < 0)
8666 8667 8668
            goto error;
        break;

M
Mark McLoughlin 已提交
8669
    default:
8670
        qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
8671 8672
                        _("hostdev subsys type '%s' not supported"),
                        virDomainHostdevSubsysTypeToString(hostdev->source.subsys.type));
8673
        goto error;
M
Mark McLoughlin 已提交
8674
    }
8675 8676 8677 8678 8679 8680

    return 0;

error:
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityHostdevLabel &&
8681 8682
        driver->securityDriver->domainRestoreSecurityHostdevLabel(driver->securityDriver,
                                                                  vm, hostdev) < 0)
8683 8684 8685
        VIR_WARN0("Unable to restore host device labelling on hotplug fail");

    return -1;
M
Mark McLoughlin 已提交
8686 8687
}

8688

8689
static int qemudDomainAttachDevice(virDomainPtr dom,
8690 8691
                                   const char *xml)
{
8692 8693 8694
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    virDomainDeviceDefPtr dev = NULL;
8695
    unsigned long long qemuCmdFlags;
8696
    virCgroupPtr cgroup = NULL;
8697
    int ret = -1;
8698

8699
    qemuDriverLock(driver);
8700
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
8701
    if (!vm) {
8702 8703
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
8704 8705
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
8706
        goto cleanup;
8707 8708
    }

8709 8710 8711
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
8712
    if (!virDomainObjIsActive(vm)) {
8713 8714
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot attach device on inactive domain"));
8715
        goto endjob;
8716 8717
    }

8718
    dev = virDomainDeviceDefParse(driver->caps, vm->def, xml,
8719
                                  VIR_DOMAIN_XML_INACTIVE);
8720
    if (dev == NULL)
8721
        goto endjob;
8722

8723 8724 8725
    if (qemudExtractVersionInfo(vm->def->emulator,
                                NULL,
                                &qemuCmdFlags) < 0)
8726
        goto endjob;
8727

8728
    if (dev->type == VIR_DOMAIN_DEVICE_DISK) {
8729
        if (qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_DEVICES)) {
8730
            if (virCgroupForDomain(driver->cgroup, vm->def->name, &cgroup, 0) !=0 ) {
8731 8732 8733
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("Unable to find cgroup for %s\n"),
                                vm->def->name);
8734
                goto endjob;
8735
            }
8736
            if (qemuSetupDiskCgroup(driver, cgroup, dev->data.disk) < 0)
8737
                goto endjob;
8738 8739
        }

8740
        switch (dev->data.disk->device) {
8741 8742
        case VIR_DOMAIN_DISK_DEVICE_CDROM:
        case VIR_DOMAIN_DISK_DEVICE_FLOPPY:
8743 8744 8745
            ret = qemudDomainChangeEjectableMedia(driver, vm,
                                                  dev->data.disk,
                                                  qemuCmdFlags);
8746 8747
            if (ret == 0)
                dev->data.disk = NULL;
8748
            break;
8749

8750 8751
        case VIR_DOMAIN_DISK_DEVICE_DISK:
            if (dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_USB) {
8752
                ret = qemudDomainAttachUsbMassstorageDevice(driver, vm,
8753
                                                            dev->data.disk, qemuCmdFlags);
8754 8755
                if (ret == 0)
                    dev->data.disk = NULL;
8756
            } else if (dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_VIRTIO) {
8757
                ret = qemudDomainAttachPciDiskDevice(driver, vm,
8758
                                                     dev->data.disk, qemuCmdFlags);
8759 8760
                if (ret == 0)
                    dev->data.disk = NULL;
8761
            } else if (dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_SCSI) {
8762
                ret = qemudDomainAttachSCSIDisk(driver, vm,
8763
                                                dev->data.disk, qemuCmdFlags);
8764 8765
                if (ret == 0)
                    dev->data.disk = NULL;
8766
            } else {
8767
                qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
8768 8769
                                _("disk bus '%s' cannot be hotplugged."),
                                virDomainDiskBusTypeToString(dev->data.disk->bus));
8770
                /* fallthrough */
8771 8772
            }
            break;
8773

8774
        default:
8775
            qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
8776 8777
                            _("disk device type '%s' cannot be hotplugged"),
                            virDomainDiskDeviceTypeToString(dev->data.disk->device));
8778 8779
            /* Fallthrough */
        }
8780
        if (ret != 0 && cgroup) {
8781
            if (qemuTeardownDiskCgroup(driver, cgroup, dev->data.disk) < 0)
8782 8783
                VIR_WARN("Failed to teardown cgroup for disk path %s",
                         NULLSTR(dev->data.disk->src));
8784
        }
8785 8786
    } else if (dev->type == VIR_DOMAIN_DEVICE_CONTROLLER) {
        if (dev->data.controller->type == VIR_DOMAIN_CONTROLLER_TYPE_SCSI) {
8787
            ret = qemudDomainAttachPciControllerDevice(driver, vm,
8788
                                                       dev->data.controller, qemuCmdFlags);
8789 8790
            if (ret == 0)
                dev->data.controller = NULL;
8791
        } else {
8792
            qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
8793 8794
                            _("disk controller bus '%s' cannot be hotplugged."),
                            virDomainControllerTypeToString(dev->data.controller->type));
8795 8796
            /* fallthrough */
        }
M
Mark McLoughlin 已提交
8797
    } else if (dev->type == VIR_DOMAIN_DEVICE_NET) {
8798 8799
        ret = qemudDomainAttachNetDevice(dom->conn, driver, vm,
                                         dev->data.net, qemuCmdFlags);
8800 8801
        if (ret == 0)
            dev->data.net = NULL;
M
Mark McLoughlin 已提交
8802
    } else if (dev->type == VIR_DOMAIN_DEVICE_HOSTDEV) {
8803
        ret = qemudDomainAttachHostDevice(driver, vm,
8804
                                          dev->data.hostdev, qemuCmdFlags);
8805 8806
        if (ret == 0)
            dev->data.hostdev = NULL;
8807
    } else {
8808
        qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
8809 8810
                        _("device type '%s' cannot be attached"),
                        virDomainDeviceTypeToString(dev->type));
8811
        goto endjob;
8812 8813
    }

8814
    if (!ret && virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
8815 8816
        ret = -1;

8817
endjob:
8818 8819
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
8820

8821
cleanup:
8822 8823 8824
    if (cgroup)
        virCgroupFree(&cgroup);

8825
    virDomainDeviceDefFree(dev);
8826 8827
    if (vm)
        virDomainObjUnlock(vm);
8828
    qemuDriverUnlock(driver);
8829 8830 8831
    return ret;
}

8832 8833 8834 8835
static int qemudDomainAttachDeviceFlags(virDomainPtr dom,
                                        const char *xml,
                                        unsigned int flags) {
    if (flags & VIR_DOMAIN_DEVICE_MODIFY_CONFIG) {
8836 8837
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot modify the persistent configuration of a domain"));
8838 8839 8840 8841 8842 8843
        return -1;
    }

    return qemudDomainAttachDevice(dom, xml);
}

8844

8845 8846 8847 8848 8849 8850 8851 8852 8853 8854 8855 8856 8857 8858 8859 8860 8861 8862 8863 8864 8865 8866 8867 8868 8869 8870 8871 8872 8873 8874 8875 8876 8877 8878 8879 8880 8881 8882 8883 8884 8885 8886 8887 8888 8889 8890 8891 8892 8893 8894 8895 8896 8897 8898 8899 8900 8901 8902 8903 8904 8905 8906 8907 8908 8909 8910 8911 8912 8913 8914 8915 8916 8917 8918 8919 8920 8921
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;
}


8922 8923 8924 8925 8926 8927 8928 8929 8930 8931 8932
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;

8933 8934 8935 8936
    virCheckFlags(VIR_DOMAIN_DEVICE_MODIFY_CURRENT |
                  VIR_DOMAIN_DEVICE_MODIFY_LIVE |
                  VIR_DOMAIN_DEVICE_MODIFY_CONFIG, -1);

8937 8938 8939 8940 8941 8942 8943 8944 8945 8946 8947 8948 8949 8950 8951 8952 8953 8954 8955 8956 8957 8958 8959 8960 8961 8962 8963 8964 8965 8966 8967 8968 8969 8970 8971 8972 8973 8974 8975 8976 8977 8978 8979 8980
    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;
            }
8981
            if (qemuSetupDiskCgroup(driver, cgroup, dev->data.disk) < 0)
8982 8983 8984 8985 8986 8987
                goto endjob;
        }

        switch (dev->data.disk->device) {
        case VIR_DOMAIN_DISK_DEVICE_CDROM:
        case VIR_DOMAIN_DISK_DEVICE_FLOPPY:
8988 8989 8990
            ret = qemudDomainChangeEjectableMedia(driver, vm,
                                                  dev->data.disk,
                                                  qemuCmdFlags);
8991 8992 8993 8994 8995 8996
            if (ret == 0)
                dev->data.disk = NULL;
            break;


        default:
8997
            qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
8998 8999 9000 9001 9002 9003
                            _("disk bus '%s' cannot be updated."),
                            virDomainDiskBusTypeToString(dev->data.disk->bus));
            break;
        }

        if (ret != 0 && cgroup) {
9004
            if (qemuTeardownDiskCgroup(driver, cgroup, dev->data.disk) < 0)
9005 9006
                VIR_WARN("Failed to teardown cgroup for disk path %s",
                         NULLSTR(dev->data.disk->src));
9007 9008 9009
        }
        break;

9010 9011 9012 9013
    case VIR_DOMAIN_DEVICE_GRAPHICS:
        ret = qemuDomainChangeGraphics(driver, vm, dev->data.graphics);
        break;

9014
    default:
9015
        qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
9016 9017 9018 9019 9020 9021 9022 9023 9024 9025 9026 9027 9028 9029 9030 9031 9032 9033 9034 9035 9036 9037 9038 9039
                        _("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
9040
static inline int qemudFindDisk(virDomainDefPtr def, const char *dst)
W
Wolfgang Mauerer 已提交
9041 9042 9043 9044 9045 9046 9047 9048 9049 9050 9051 9052 9053
{
    int i;

    for (i = 0 ; i < def->ndisks ; i++) {
        if (STREQ(def->disks[i]->dst, dst)) {
            return i;
        }
    }

    return -1;
}


9054
static int qemudDomainDetachPciDiskDevice(struct qemud_driver *driver,
9055
                                          virDomainObjPtr vm,
9056 9057
                                          virDomainDeviceDefPtr dev,
                                          unsigned long long qemuCmdFlags)
9058 9059 9060
{
    int i, ret = -1;
    virDomainDiskDefPtr detach = NULL;
9061
    qemuDomainObjPrivatePtr priv = vm->privateData;
9062
    virCgroupPtr cgroup = NULL;
9063

W
Wolfgang Mauerer 已提交
9064
    i = qemudFindDisk(vm->def, dev->data.disk->dst);
9065

W
Wolfgang Mauerer 已提交
9066
    if (i < 0) {
9067 9068
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("disk %s not found"), dev->data.disk->dst);
9069
        goto cleanup;
9070 9071
    }

W
Wolfgang Mauerer 已提交
9072 9073
    detach = vm->def->disks[i];

9074 9075 9076 9077 9078 9079 9080 9081 9082
    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;
        }
    }

9083 9084
    if (!virDomainDeviceAddressIsValid(&detach->info,
                                       VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI)) {
9085 9086
        qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
                        _("device cannot be detached without a PCI address"));
9087
        goto cleanup;
9088 9089
    }

9090
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
9091 9092 9093 9094 9095 9096 9097 9098 9099 9100 9101
    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;
        }
9102
    }
9103
    qemuDomainObjExitMonitorWithDriver(driver, vm);
9104

9105 9106
    qemuDomainDiskAudit(vm, detach, NULL, "detach", ret >= 0);

9107 9108 9109 9110
    if ((qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) &&
        qemuDomainPCIAddressReleaseAddr(priv->pciaddrs, &detach->info) < 0)
        VIR_WARN("Unable to release PCI address on %s", dev->data.disk->src);

9111
    virDomainDiskRemove(vm->def, i);
W
Wolfgang Mauerer 已提交
9112

9113 9114
    virDomainDiskDefFree(detach);

9115 9116
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
9117 9118
        driver->securityDriver->domainRestoreSecurityImageLabel(driver->securityDriver,
                                                                vm, dev->data.disk) < 0)
9119 9120
        VIR_WARN("Unable to restore security label on %s", dev->data.disk->src);

9121
    if (cgroup != NULL) {
9122
        if (qemuTeardownDiskCgroup(driver, cgroup, dev->data.disk) < 0)
9123 9124 9125 9126
            VIR_WARN("Failed to teardown cgroup for disk path %s",
                     NULLSTR(dev->data.disk->src));
    }

9127 9128 9129 9130 9131 9132 9133 9134 9135 9136 9137 9138 9139 9140
    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;
9141
    virCgroupPtr cgroup = NULL;
9142 9143 9144 9145 9146 9147 9148 9149 9150 9151

    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)) {
9152
        qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
9153 9154 9155 9156 9157 9158
                        _("Underlying qemu does not support SCSI disk removal"));
        goto cleanup;
    }

    detach = vm->def->disks[i];

9159 9160 9161 9162 9163 9164 9165 9166 9167
    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;
        }
    }

9168 9169 9170 9171 9172 9173 9174
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
    if (qemuMonitorDelDevice(priv->mon, detach->info.alias) < 0) {
        qemuDomainObjExitMonitor(vm);
        goto cleanup;
    }
    qemuDomainObjExitMonitorWithDriver(driver, vm);

9175 9176
    qemuDomainDiskAudit(vm, detach, NULL, "detach", ret >= 0);

9177
    virDomainDiskRemove(vm->def, i);
9178

9179
    virDomainDiskDefFree(detach);
9180

9181 9182
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
9183 9184
        driver->securityDriver->domainRestoreSecurityImageLabel(driver->securityDriver,
                                                                vm, dev->data.disk) < 0)
9185 9186
        VIR_WARN("Unable to restore security label on %s", dev->data.disk->src);

9187
    if (cgroup != NULL) {
9188
        if (qemuTeardownDiskCgroup(driver, cgroup, dev->data.disk) < 0)
9189 9190 9191 9192
            VIR_WARN("Failed to teardown cgroup for disk path %s",
                     NULLSTR(dev->data.disk->src));
    }

9193
    ret = 0;
9194 9195

cleanup:
9196
    virCgroupFree(&cgroup);
9197 9198 9199
    return ret;
}

9200
static int qemudDomainDetachPciControllerDevice(struct qemud_driver *driver,
9201
                                                virDomainObjPtr vm,
9202 9203
                                                virDomainDeviceDefPtr dev,
                                                unsigned long long qemuCmdFlags)
9204 9205 9206 9207 9208 9209 9210 9211 9212 9213 9214 9215 9216 9217
{
    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) {
9218 9219 9220 9221
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("disk controller %s:%d not found"),
                        virDomainControllerTypeToString(dev->data.controller->type),
                        dev->data.controller->idx);
9222 9223 9224 9225 9226
        goto cleanup;
    }

    if (!virDomainDeviceAddressIsValid(&detach->info,
                                       VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI)) {
9227 9228
        qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
                        _("device cannot be detached without a PCI address"));
9229 9230 9231
        goto cleanup;
    }

9232 9233 9234 9235 9236
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuAssignDeviceControllerAlias(detach) < 0)
            goto cleanup;
    }

9237
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
9238 9239 9240 9241 9242 9243 9244 9245 9246 9247 9248
    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;
        }
9249 9250 9251 9252 9253 9254 9255 9256 9257 9258 9259 9260 9261 9262 9263 9264
    }
    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;
    }
9265

9266 9267
    if ((qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) &&
        qemuDomainPCIAddressReleaseAddr(priv->pciaddrs, &detach->info) < 0)
9268 9269
        VIR_WARN0("Unable to release PCI address on controller");

9270 9271 9272 9273 9274 9275 9276 9277
    virDomainControllerDefFree(detach);

    ret = 0;

cleanup:
    return ret;
}

9278
static int
9279
qemudDomainDetachNetDevice(struct qemud_driver *driver,
9280
                           virDomainObjPtr vm,
9281 9282
                           virDomainDeviceDefPtr dev,
                           unsigned long long qemuCmdFlags)
9283 9284 9285
{
    int i, ret = -1;
    virDomainNetDefPtr detach = NULL;
9286
    qemuDomainObjPrivatePtr priv = vm->privateData;
9287 9288
    int vlan;
    char *hostnet_name = NULL;
9289 9290 9291 9292 9293 9294 9295 9296 9297 9298 9299

    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) {
9300 9301 9302 9303 9304
        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]);
9305 9306 9307
        goto cleanup;
    }

9308 9309
    if (!virDomainDeviceAddressIsValid(&detach->info,
                                       VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI)) {
9310 9311
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("device cannot be detached without a PCI address"));
9312 9313 9314
        goto cleanup;
    }

9315
    if ((vlan = qemuDomainNetVLAN(detach)) < 0) {
9316 9317
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("unable to determine original VLAN"));
9318 9319 9320 9321
        goto cleanup;
    }

    if (virAsprintf(&hostnet_name, "host%s", detach->info.alias) < 0) {
9322
        virReportOOMError();
9323 9324 9325
        goto cleanup;
    }

9326
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
9327 9328 9329
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuMonitorDelDevice(priv->mon, detach->info.alias) < 0) {
            qemuDomainObjExitMonitor(vm);
9330
            qemuDomainNetAudit(vm, detach, NULL, "detach", false);
9331 9332 9333 9334 9335 9336
            goto cleanup;
        }
    } else {
        if (qemuMonitorRemovePCIDevice(priv->mon,
                                       &detach->info.addr.pci) < 0) {
            qemuDomainObjExitMonitorWithDriver(driver, vm);
9337
            qemuDomainNetAudit(vm, detach, NULL, "detach", false);
9338 9339
            goto cleanup;
        }
9340
    }
9341

9342 9343 9344 9345
    if ((qemuCmdFlags & QEMUD_CMD_FLAG_NETDEV) &&
        (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)) {
        if (qemuMonitorRemoveNetdev(priv->mon, hostnet_name) < 0) {
            qemuDomainObjExitMonitorWithDriver(driver, vm);
9346
            qemuDomainNetAudit(vm, detach, NULL, "detach", false);
9347 9348 9349 9350 9351
            goto cleanup;
        }
    } else {
        if (qemuMonitorRemoveHostNetwork(priv->mon, vlan, hostnet_name) < 0) {
            qemuDomainObjExitMonitorWithDriver(driver, vm);
9352
            qemuDomainNetAudit(vm, detach, NULL, "detach", false);
9353 9354
            goto cleanup;
        }
9355
    }
9356
    qemuDomainObjExitMonitorWithDriver(driver, vm);
9357

9358 9359
    qemuDomainNetAudit(vm, detach, NULL, "detach", true);

9360 9361 9362 9363
    if ((qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) &&
        qemuDomainPCIAddressReleaseAddr(priv->pciaddrs, &detach->info) < 0)
        VIR_WARN0("Unable to release PCI address on NIC");

9364
    virDomainConfNWFilterTeardown(detach);
9365

9366
#if WITH_MACVTAP
9367
    if (detach->type == VIR_DOMAIN_NET_TYPE_DIRECT) {
9368
        delMacvtap(detach->ifname, detach->mac, detach->data.direct.linkdev,
9369
                   &detach->data.direct.virtPortProfile);
9370 9371
        VIR_FREE(detach->ifname);
    }
9372 9373
#endif

9374
    if ((driver->macFilter) && (detach->ifname != NULL)) {
9375
        if ((errno = networkDisallowMacOnPort(driver,
9376 9377
                                              detach->ifname,
                                              detach->mac))) {
9378
            virReportSystemError(errno,
9379 9380 9381 9382 9383
             _("failed to remove ebtables rule on  '%s'"),
                                 detach->ifname);
        }
    }

9384 9385 9386 9387 9388 9389 9390 9391 9392 9393
    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 {
9394
        VIR_FREE(vm->def->nets);
9395
        vm->def->nnets = 0;
9396
    }
9397
    virDomainNetDefFree(detach);
9398

9399 9400 9401
    ret = 0;

cleanup:
9402
    VIR_FREE(hostnet_name);
9403 9404 9405
    return ret;
}

9406
static int qemudDomainDetachHostPciDevice(struct qemud_driver *driver,
9407
                                          virDomainObjPtr vm,
9408 9409
                                          virDomainDeviceDefPtr dev,
                                          unsigned long long qemuCmdFlags)
9410
{
9411
    virDomainHostdevDefPtr detach = NULL;
9412
    qemuDomainObjPrivatePtr priv = vm->privateData;
9413
    int i, ret;
9414
    pciDevice *pci;
9415 9416

    for (i = 0 ; i < vm->def->nhostdevs ; i++) {
9417 9418 9419 9420
        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;

9421 9422 9423 9424 9425 9426 9427 9428 9429 9430 9431 9432 9433 9434 9435
        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) {
9436 9437 9438 9439 9440 9441
        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);
9442 9443 9444
        return -1;
    }

9445 9446
    if (!virDomainDeviceAddressIsValid(&detach->info,
                                       VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI)) {
9447 9448
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("device cannot be detached without a PCI address"));
9449 9450 9451
        return -1;
    }

9452
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
9453 9454 9455 9456 9457 9458 9459 9460 9461 9462 9463
    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;
        }
9464
    }
9465
    qemuDomainObjExitMonitorWithDriver(driver, vm);
9466 9467 9468

    ret = 0;

9469
    pci = pciGetDevice(detach->source.subsys.u.pci.domain,
9470 9471 9472 9473 9474 9475
                       detach->source.subsys.u.pci.bus,
                       detach->source.subsys.u.pci.slot,
                       detach->source.subsys.u.pci.function);
    if (!pci)
        ret = -1;
    else {
9476
        pciDeviceSetManaged(pci, detach->managed);
9477
        pciDeviceListDel(driver->activePciHostdevs, pci);
9478
        if (pciResetDevice(pci, driver->activePciHostdevs, NULL) < 0)
9479
            ret = -1;
9480
        qemudReattachPciDevice(pci, driver);
9481
        pciFreeDevice(pci);
9482 9483
    }

9484 9485
    if ((qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) &&
        qemuDomainPCIAddressReleaseAddr(priv->pciaddrs, &detach->info) < 0)
9486
        VIR_WARN0("Unable to release PCI address on host device");
9487

9488 9489 9490 9491 9492 9493 9494 9495 9496 9497 9498 9499
    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;
9500
    }
9501
    virDomainHostdevDefFree(detach);
9502 9503 9504 9505

    return ret;
}

9506 9507 9508 9509 9510 9511 9512 9513 9514 9515 9516 9517 9518 9519 9520 9521
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;
9522 9523 9524 9525 9526 9527 9528 9529 9530 9531 9532 9533 9534 9535 9536 9537
        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;
            }
9538 9539 9540 9541 9542 9543 9544 9545 9546 9547 9548 9549 9550 9551 9552 9553 9554 9555 9556 9557 9558 9559 9560 9561 9562 9563 9564 9565 9566 9567 9568 9569 9570 9571 9572 9573 9574 9575 9576 9577 9578 9579 9580 9581 9582 9583 9584 9585 9586 9587
        }
    }

    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;
}

9588
static int qemudDomainDetachHostDevice(struct qemud_driver *driver,
9589
                                       virDomainObjPtr vm,
9590 9591
                                       virDomainDeviceDefPtr dev,
                                       unsigned long long qemuCmdFlags)
9592 9593 9594 9595 9596
{
    virDomainHostdevDefPtr hostdev = dev->data.hostdev;
    int ret;

    if (hostdev->mode != VIR_DOMAIN_HOSTDEV_MODE_SUBSYS) {
9597
        qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
9598 9599
                        _("hostdev mode '%s' not supported"),
                        virDomainHostdevModeTypeToString(hostdev->mode));
9600 9601 9602 9603 9604
        return -1;
    }

    switch (hostdev->source.subsys.type) {
    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI:
9605
        ret = qemudDomainDetachHostPciDevice(driver, vm, dev, qemuCmdFlags);
9606
        break;
9607 9608 9609
    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB:
        ret = qemudDomainDetachHostUsbDevice(driver, vm, dev, qemuCmdFlags);
        break;
9610
    default:
9611
        qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
9612 9613
                        _("hostdev subsys type '%s' not supported"),
                        virDomainHostdevSubsysTypeToString(hostdev->source.subsys.type));
9614 9615 9616
        return -1;
    }

9617
    if (driver->securityDriver &&
9618
        driver->securityDriver->domainRestoreSecurityHostdevLabel &&
9619 9620
        driver->securityDriver->domainRestoreSecurityHostdevLabel(driver->securityDriver,
                                                                  vm, dev->data.hostdev) < 0)
9621
        VIR_WARN0("Failed to restore host device labelling");
9622

9623 9624 9625
    return ret;
}

9626 9627
static int qemudDomainDetachDevice(virDomainPtr dom,
                                   const char *xml) {
9628 9629
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
9630
    unsigned long long qemuCmdFlags;
9631 9632
    virDomainDeviceDefPtr dev = NULL;
    int ret = -1;
9633

9634
    qemuDriverLock(driver);
9635
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
9636
    if (!vm) {
9637 9638
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
9639 9640
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
9641
        goto cleanup;
9642 9643
    }

9644 9645 9646
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
9647
    if (!virDomainObjIsActive(vm)) {
9648 9649
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot detach device on inactive domain"));
9650
        goto endjob;
9651 9652
    }

9653
    dev = virDomainDeviceDefParse(driver->caps, vm->def, xml,
9654
                                  VIR_DOMAIN_XML_INACTIVE);
9655
    if (dev == NULL)
9656
        goto endjob;
9657

9658 9659 9660 9661
    if (qemudExtractVersionInfo(vm->def->emulator,
                                NULL,
                                &qemuCmdFlags) < 0)
        goto endjob;
9662 9663

    if (dev->type == VIR_DOMAIN_DEVICE_DISK &&
9664 9665 9666 9667 9668 9669 9670 9671 9672
        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 {
9673
            qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
9674 9675
                            _("This type of disk cannot be hot unplugged"));
        }
9676
    } else if (dev->type == VIR_DOMAIN_DEVICE_NET) {
9677
        ret = qemudDomainDetachNetDevice(driver, vm, dev, qemuCmdFlags);
9678 9679
    } else if (dev->type == VIR_DOMAIN_DEVICE_CONTROLLER) {
        if (dev->data.controller->type == VIR_DOMAIN_CONTROLLER_TYPE_SCSI) {
9680 9681
            ret = qemudDomainDetachPciControllerDevice(driver, vm, dev,
                                                       qemuCmdFlags);
9682
        } else {
9683
            qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
9684 9685
                            _("disk controller bus '%s' cannot be hotunplugged."),
                            virDomainControllerTypeToString(dev->data.controller->type));
9686 9687
            /* fallthrough */
        }
9688
    } else if (dev->type == VIR_DOMAIN_DEVICE_HOSTDEV) {
9689
        ret = qemudDomainDetachHostDevice(driver, vm, dev, qemuCmdFlags);
9690
    } else {
9691
        qemuReportError(VIR_ERR_CONFIG_UNSUPPORTED,
9692
                        "%s", _("This type of device cannot be hot unplugged"));
9693
    }
9694

9695
    if (!ret && virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
9696 9697
        ret = -1;

9698
endjob:
9699 9700
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
9701

9702 9703
cleanup:
    virDomainDeviceDefFree(dev);
9704 9705
    if (vm)
        virDomainObjUnlock(vm);
9706
    qemuDriverUnlock(driver);
9707 9708 9709
    return ret;
}

9710 9711 9712 9713
static int qemudDomainDetachDeviceFlags(virDomainPtr dom,
                                        const char *xml,
                                        unsigned int flags) {
    if (flags & VIR_DOMAIN_DEVICE_MODIFY_CONFIG) {
9714 9715
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot modify the persistent configuration of a domain"));
9716 9717 9718 9719 9720 9721
        return -1;
    }

    return qemudDomainDetachDevice(dom, xml);
}

9722
static int qemudDomainGetAutostart(virDomainPtr dom,
9723
                                   int *autostart) {
9724 9725 9726
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
9727

9728
    qemuDriverLock(driver);
9729
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
9730 9731
    qemuDriverUnlock(driver);

9732
    if (!vm) {
9733 9734
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
9735 9736
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
9737
        goto cleanup;
9738 9739 9740
    }

    *autostart = vm->autostart;
9741
    ret = 0;
9742

9743
cleanup:
9744 9745
    if (vm)
        virDomainObjUnlock(vm);
9746
    return ret;
9747 9748
}

9749
static int qemudDomainSetAutostart(virDomainPtr dom,
9750
                                   int autostart) {
9751 9752
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
9753 9754
    char *configFile = NULL, *autostartLink = NULL;
    int ret = -1;
9755

9756
    qemuDriverLock(driver);
9757
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
9758

9759
    if (!vm) {
9760 9761
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
9762 9763
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
9764
        goto cleanup;
9765 9766
    }

9767
    if (!vm->persistent) {
9768 9769
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("cannot set autostart for transient domain"));
9770
        goto cleanup;
9771 9772
    }

9773 9774
    autostart = (autostart != 0);

9775
    if (vm->autostart != autostart) {
9776
        if ((configFile = virDomainConfigFile(driver->configDir, vm->def->name)) == NULL)
9777
            goto cleanup;
9778
        if ((autostartLink = virDomainConfigFile(driver->autostartDir, vm->def->name)) == NULL)
9779
            goto cleanup;
9780

9781 9782
        if (autostart) {
            int err;
9783

9784
            if ((err = virFileMakePath(driver->autostartDir))) {
9785
                virReportSystemError(err,
9786 9787
                                     _("cannot create autostart directory %s"),
                                     driver->autostartDir);
9788 9789
                goto cleanup;
            }
9790

9791
            if (symlink(configFile, autostartLink) < 0) {
9792
                virReportSystemError(errno,
9793 9794
                                     _("Failed to create symlink '%s to '%s'"),
                                     autostartLink, configFile);
9795 9796 9797 9798
                goto cleanup;
            }
        } else {
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
9799
                virReportSystemError(errno,
9800 9801
                                     _("Failed to delete symlink '%s'"),
                                     autostartLink);
9802 9803
                goto cleanup;
            }
9804 9805
        }

9806
        vm->autostart = autostart;
9807
    }
9808
    ret = 0;
9809

9810 9811 9812
cleanup:
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
9813 9814
    if (vm)
        virDomainObjUnlock(vm);
9815
    qemuDriverUnlock(driver);
9816
    return ret;
9817 9818
}

9819 9820 9821 9822 9823

static char *qemuGetSchedulerType(virDomainPtr dom,
                                  int *nparams)
{
    struct qemud_driver *driver = dom->conn->privateData;
9824
    char *ret = NULL;
9825

9826
    qemuDriverLock(driver);
9827
    if (!qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_CPU)) {
9828 9829
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        __FUNCTION__);
9830
        goto cleanup;
9831 9832 9833 9834 9835 9836 9837
    }

    if (nparams)
        *nparams = 1;

    ret = strdup("posix");
    if (!ret)
9838
        virReportOOMError();
9839 9840 9841

cleanup:
    qemuDriverUnlock(driver);
9842 9843 9844
    return ret;
}

9845 9846 9847 9848 9849 9850 9851 9852 9853 9854 9855 9856 9857 9858 9859 9860 9861 9862 9863 9864 9865 9866 9867 9868 9869 9870 9871 9872 9873 9874 9875 9876 9877 9878 9879 9880 9881 9882 9883 9884 9885 9886 9887 9888 9889 9890 9891 9892 9893 9894 9895 9896 9897 9898 9899 9900 9901 9902 9903 9904 9905 9906 9907 9908 9909 9910 9911

static int qemuDomainSetMemoryParameters(virDomainPtr dom,
                                         virMemoryParameterPtr params,
                                         int nparams,
                                         unsigned int flags ATTRIBUTE_UNUSED)
{
    struct qemud_driver *driver = dom->conn->privateData;
    int i;
    virCgroupPtr group = NULL;
    virDomainObjPtr vm = NULL;
    int ret = -1;

    qemuDriverLock(driver);
    if (!qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_MEMORY)) {
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        __FUNCTION__);
        goto cleanup;
    }

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

    if (vm == NULL) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("No such domain %s"), dom->uuid);
        goto cleanup;
    }

    if (virCgroupForDomain(driver->cgroup, vm->def->name, &group, 0) != 0) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("cannot find cgroup for domain %s"), vm->def->name);
        goto cleanup;
    }

    ret = 0;
    for (i = 0; i < nparams; i++) {
        virMemoryParameterPtr param = &params[i];

        if (STREQ(param->field, VIR_DOMAIN_MEMORY_HARD_LIMIT)) {
            int rc;
            if (param->type != VIR_DOMAIN_MEMORY_PARAM_ULLONG) {
                qemuReportError(VIR_ERR_INVALID_ARG, "%s",
                                _("invalid type for memory hard_limit tunable, expected a 'ullong'"));
                ret = -1;
                continue;
            }

            rc = virCgroupSetMemoryHardLimit(group, params[i].value.ul);
            if (rc != 0) {
                virReportSystemError(-rc, "%s",
                                     _("unable to set memory hard_limit tunable"));
                ret = -1;
            }
        } else if (STREQ(param->field, VIR_DOMAIN_MEMORY_SOFT_LIMIT)) {
            int rc;
            if (param->type != VIR_DOMAIN_MEMORY_PARAM_ULLONG) {
                qemuReportError(VIR_ERR_INVALID_ARG, "%s",
                                _("invalid type for memory soft_limit tunable, expected a 'ullong'"));
                ret = -1;
                continue;
            }

            rc = virCgroupSetMemorySoftLimit(group, params[i].value.ul);
            if (rc != 0) {
                virReportSystemError(-rc, "%s",
                                     _("unable to set memory soft_limit tunable"));
                ret = -1;
            }
9912
        } else if (STREQ(param->field, VIR_DOMAIN_MEMORY_SWAP_HARD_LIMIT)) {
9913 9914 9915 9916 9917 9918 9919 9920 9921 9922 9923 9924 9925 9926 9927 9928 9929 9930 9931 9932 9933 9934 9935 9936 9937 9938 9939 9940 9941 9942 9943 9944 9945
            int rc;
            if (param->type != VIR_DOMAIN_MEMORY_PARAM_ULLONG) {
                qemuReportError(VIR_ERR_INVALID_ARG, "%s",
                                _("invalid type for swap_hard_limit tunable, expected a 'ullong'"));
                ret = -1;
                continue;
            }

            rc = virCgroupSetSwapHardLimit(group, params[i].value.ul);
            if (rc != 0) {
                virReportSystemError(-rc, "%s",
                                     _("unable to set swap_hard_limit tunable"));
                ret = -1;
            }
        } else if (STREQ(param->field, VIR_DOMAIN_MEMORY_MIN_GUARANTEE)) {
            qemuReportError(VIR_ERR_INVALID_ARG,
                            _("Memory tunable `%s' not implemented"), param->field);
            ret = -1;
        } else {
            qemuReportError(VIR_ERR_INVALID_ARG,
                            _("Parameter `%s' not supported"), param->field);
            ret = -1;
        }
    }

cleanup:
    virCgroupFree(&group);
    if (vm)
        virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);
    return ret;
}

9946 9947 9948 9949 9950 9951 9952 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 9981 9982 9983 9984 9985 9986 9987 9988 9989 9990 9991 9992 9993 9994 9995 9996 9997 9998 9999 10000 10001 10002 10003 10004 10005
static int qemuDomainGetMemoryParameters(virDomainPtr dom,
                                         virMemoryParameterPtr params,
                                         int *nparams,
                                         unsigned int flags ATTRIBUTE_UNUSED)
{
    struct qemud_driver *driver = dom->conn->privateData;
    int i;
    virCgroupPtr group = NULL;
    virDomainObjPtr vm = NULL;
    unsigned long val;
    int ret = -1;
    int rc;

    qemuDriverLock(driver);

    if (!qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_MEMORY)) {
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        __FUNCTION__);
        goto cleanup;
    }

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

    if (vm == NULL) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("No such domain %s"), dom->uuid);
        goto cleanup;
    }

    if ((*nparams) == 0) {
        /* Current number of memory parameters supported by cgroups */
        *nparams = QEMU_NB_MEM_PARAM;
        ret = 0;
        goto cleanup;
    }

    if ((*nparams) != QEMU_NB_MEM_PARAM) {
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("Invalid parameter count"));
        goto cleanup;
    }

    if (virCgroupForDomain(driver->cgroup, vm->def->name, &group, 0) != 0) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("cannot find cgroup for domain %s"), vm->def->name);
        goto cleanup;
    }

    for (i = 0; i < *nparams; i++) {
        virMemoryParameterPtr param = &params[i];
        val = 0;
        param->value.ul = 0;
        param->type = VIR_DOMAIN_MEMORY_PARAM_ULLONG;

        switch(i) {
        case 0: /* fill memory hard limit here */
            rc = virCgroupGetMemoryHardLimit(group, &val);
            if (rc != 0) {
                virReportSystemError(-rc, "%s",
                                     _("unable to get memory hard limit"));
10006
                goto cleanup;
10007 10008 10009 10010
            }
            if (virStrcpyStatic(param->field, VIR_DOMAIN_MEMORY_HARD_LIMIT) == NULL) {
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                "%s", _("Field memory hard limit too long for destination"));
10011
                goto cleanup;
10012 10013 10014 10015 10016 10017 10018 10019 10020
            }
            param->value.ul = val;
            break;

        case 1: /* fill memory soft limit here */
            rc = virCgroupGetMemorySoftLimit(group, &val);
            if (rc != 0) {
                virReportSystemError(-rc, "%s",
                                     _("unable to get memory soft limit"));
10021
                goto cleanup;
10022 10023 10024 10025
            }
            if (virStrcpyStatic(param->field, VIR_DOMAIN_MEMORY_SOFT_LIMIT) == NULL) {
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                "%s", _("Field memory soft limit too long for destination"));
10026
                goto cleanup;
10027 10028 10029 10030 10031 10032 10033 10034 10035
            }
            param->value.ul = val;
            break;

        case 2: /* fill swap hard limit here */
            rc = virCgroupGetSwapHardLimit(group, &val);
            if (rc != 0) {
                virReportSystemError(-rc, "%s",
                                     _("unable to get swap hard limit"));
10036
                goto cleanup;
10037
            }
10038
            if (virStrcpyStatic(param->field, VIR_DOMAIN_MEMORY_SWAP_HARD_LIMIT) == NULL) {
10039 10040
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                "%s", _("Field swap hard limit too long for destination"));
10041
                goto cleanup;
10042 10043 10044 10045 10046 10047 10048 10049 10050 10051
            }
            param->value.ul = val;
            break;

        default:
            break;
            /* should not hit here */
        }
    }

10052 10053
    ret = 0;

10054 10055 10056 10057 10058 10059 10060 10061 10062
cleanup:
    if (group)
        virCgroupFree(&group);
    if (vm)
        virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);
    return ret;
}

10063 10064 10065 10066 10067 10068 10069 10070 10071 10072
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;

10073
    qemuDriverLock(driver);
10074
    if (!qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_CPU)) {
10075 10076
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        __FUNCTION__);
10077
        goto cleanup;
10078 10079 10080 10081 10082
    }

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

    if (vm == NULL) {
10083 10084
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("No such domain %s"), dom->uuid);
10085 10086 10087 10088
        goto cleanup;
    }

    if (virCgroupForDomain(driver->cgroup, vm->def->name, &group, 0) != 0) {
10089 10090
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("cannot find cgroup for domain %s"), vm->def->name);
10091 10092 10093 10094 10095 10096 10097 10098 10099
        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) {
10100 10101
                qemuReportError(VIR_ERR_INVALID_ARG, "%s",
                                _("invalid type for cpu_shares tunable, expected a 'ullong'"));
10102 10103 10104 10105 10106
                goto cleanup;
            }

            rc = virCgroupSetCpuShares(group, params[i].value.ul);
            if (rc != 0) {
10107
                virReportSystemError(-rc, "%s",
10108 10109 10110 10111
                                     _("unable to set cpu shares tunable"));
                goto cleanup;
            }
        } else {
10112 10113
            qemuReportError(VIR_ERR_INVALID_ARG,
                            _("Invalid parameter `%s'"), param->field);
10114 10115 10116 10117 10118 10119 10120 10121 10122
            goto cleanup;
        }
    }
    ret = 0;

cleanup:
    virCgroupFree(&group);
    if (vm)
        virDomainObjUnlock(vm);
10123
    qemuDriverUnlock(driver);
10124 10125 10126 10127 10128 10129 10130 10131 10132 10133 10134 10135 10136 10137
    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;

10138
    qemuDriverLock(driver);
10139
    if (!qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_CPU)) {
10140 10141
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        __FUNCTION__);
10142
        goto cleanup;
10143 10144 10145
    }

    if ((*nparams) != 1) {
10146 10147
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("Invalid parameter count"));
10148
        goto cleanup;
10149 10150 10151 10152 10153
    }

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

    if (vm == NULL) {
10154 10155
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("No such domain %s"), dom->uuid);
10156 10157 10158 10159
        goto cleanup;
    }

    if (virCgroupForDomain(driver->cgroup, vm->def->name, &group, 0) != 0) {
10160 10161
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("cannot find cgroup for domain %s"), vm->def->name);
10162 10163 10164 10165 10166
        goto cleanup;
    }

    rc = virCgroupGetCpuShares(group, &val);
    if (rc != 0) {
10167
        virReportSystemError(-rc, "%s",
10168 10169 10170 10171 10172
                             _("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
10173
    if (virStrcpyStatic(params[0].field, "cpu_shares") == NULL) {
10174 10175
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("Field cpu_shares too long for destination"));
C
Chris Lalancette 已提交
10176 10177
        goto cleanup;
    }
10178 10179 10180 10181 10182 10183 10184

    ret = 0;

cleanup:
    virCgroupFree(&group);
    if (vm)
        virDomainObjUnlock(vm);
10185
    qemuDriverUnlock(driver);
10186 10187 10188 10189
    return ret;
}


10190 10191 10192 10193 10194 10195 10196 10197 10198
/* 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)
{
10199
    struct qemud_driver *driver = dom->conn->privateData;
10200
    int i, ret = -1;
10201
    virDomainObjPtr vm;
10202
    virDomainDiskDefPtr disk = NULL;
10203

10204
    qemuDriverLock(driver);
10205
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
10206
    qemuDriverUnlock(driver);
10207
    if (!vm) {
10208 10209
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
10210 10211
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
10212
        goto cleanup;
10213
    }
10214 10215 10216 10217

    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
10218
    if (!virDomainObjIsActive (vm)) {
10219 10220
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
10221
        goto endjob;
10222 10223
    }

10224 10225 10226 10227 10228 10229 10230 10231
    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (STREQ(path, vm->def->disks[i]->dst)) {
            disk = vm->def->disks[i];
            break;
        }
    }

    if (!disk) {
10232 10233
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("invalid path: %s"), path);
10234
        goto endjob;
10235 10236
    }

10237
    if (!disk->info.alias) {
10238 10239
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("missing disk device alias name for %s"), disk->dst);
10240
        goto endjob;
10241
    }
10242

10243
    qemuDomainObjPrivatePtr priv = vm->privateData;
10244 10245
    qemuDomainObjEnterMonitor(vm);
    ret = qemuMonitorGetBlockStatsInfo(priv->mon,
10246
                                       disk->info.alias,
10247 10248 10249 10250 10251 10252
                                       &stats->rd_req,
                                       &stats->rd_bytes,
                                       &stats->wr_req,
                                       &stats->wr_bytes,
                                       &stats->errs);
    qemuDomainObjExitMonitor(vm);
10253

10254
endjob:
10255 10256
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
10257

10258
cleanup:
10259 10260
    if (vm)
        virDomainObjUnlock(vm);
10261
    return ret;
10262 10263
}

10264
#ifdef __linux__
10265 10266 10267 10268 10269
static int
qemudDomainInterfaceStats (virDomainPtr dom,
                           const char *path,
                           struct _virDomainInterfaceStats *stats)
{
10270 10271
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
10272
    int i;
10273
    int ret = -1;
10274

10275
    qemuDriverLock(driver);
10276
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
10277 10278
    qemuDriverUnlock(driver);

10279
    if (!vm) {
10280 10281
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
10282 10283
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
10284
        goto cleanup;
10285 10286
    }

D
Daniel P. Berrange 已提交
10287
    if (!virDomainObjIsActive(vm)) {
10288 10289
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
10290
        goto cleanup;
10291 10292 10293
    }

    /* Check the path is one of the domain's network interfaces. */
10294 10295
    for (i = 0 ; i < vm->def->nnets ; i++) {
        if (vm->def->nets[i]->ifname &&
10296 10297 10298 10299
            STREQ (vm->def->nets[i]->ifname, path)) {
            ret = 0;
            break;
        }
10300 10301
    }

10302
    if (ret == 0)
10303
        ret = linuxDomainInterfaceStats(path, stats);
10304
    else
10305 10306
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("invalid path, '%s' is not a known interface"), path);
10307

10308
cleanup:
10309 10310
    if (vm)
        virDomainObjUnlock(vm);
10311 10312
    return ret;
}
10313
#else
10314 10315 10316 10317
static int
qemudDomainInterfaceStats (virDomainPtr dom,
                           const char *path ATTRIBUTE_UNUSED,
                           struct _virDomainInterfaceStats *stats ATTRIBUTE_UNUSED)
10318 10319
    qemuReportError(VIR_ERR_NO_SUPPORT,
                    "%s", __FUNCTION__);
10320 10321
    return -1;
}
10322
#endif
10323

10324 10325 10326 10327 10328 10329 10330 10331 10332 10333 10334 10335 10336 10337 10338 10339
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);
10340 10341
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
10342 10343 10344
        goto cleanup;
    }

10345 10346 10347
    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

10348 10349 10350
    if (virDomainObjIsActive(vm)) {
        qemuDomainObjPrivatePtr priv = vm->privateData;
        qemuDomainObjEnterMonitor(vm);
10351
        ret = qemuMonitorGetMemoryStats(priv->mon, stats, nr_stats);
10352 10353
        qemuDomainObjExitMonitor(vm);
    } else {
10354 10355
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
10356 10357
    }

10358 10359 10360
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;

10361 10362 10363 10364 10365 10366
cleanup:
    if (vm)
        virDomainObjUnlock(vm);
    return ret;
}

10367 10368 10369 10370 10371 10372 10373
static int
qemudDomainBlockPeek (virDomainPtr dom,
                      const char *path,
                      unsigned long long offset, size_t size,
                      void *buffer,
                      unsigned int flags ATTRIBUTE_UNUSED)
{
10374 10375 10376
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int fd = -1, ret = -1, i;
10377

10378
    qemuDriverLock(driver);
10379
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
10380 10381
    qemuDriverUnlock(driver);

10382
    if (!vm) {
10383 10384
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
10385 10386
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
10387
        goto cleanup;
10388 10389 10390
    }

    if (!path || path[0] == '\0') {
10391 10392
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("NULL or empty path"));
10393
        goto cleanup;
10394 10395 10396
    }

    /* Check the path belongs to this domain. */
10397 10398
    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (vm->def->disks[i]->src != NULL &&
10399 10400 10401 10402
            STREQ (vm->def->disks[i]->src, path)) {
            ret = 0;
            break;
        }
10403 10404
    }

10405 10406 10407 10408 10409
    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) {
10410 10411
            virReportSystemError(errno,
                                 _("%s: failed to open"), path);
10412 10413
            goto cleanup;
        }
10414

10415 10416 10417 10418 10419 10420
        /* 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) {
10421 10422
            virReportSystemError(errno,
                                 _("%s: failed to seek or read"), path);
10423 10424 10425 10426 10427
            goto cleanup;
        }

        ret = 0;
    } else {
10428 10429
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("invalid path"));
10430 10431
    }

10432 10433 10434
cleanup:
    if (fd >= 0)
        close (fd);
10435 10436
    if (vm)
        virDomainObjUnlock(vm);
10437 10438 10439
    return ret;
}

R
Richard W.M. Jones 已提交
10440 10441 10442 10443 10444 10445
static int
qemudDomainMemoryPeek (virDomainPtr dom,
                       unsigned long long offset, size_t size,
                       void *buffer,
                       unsigned int flags)
{
10446 10447
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
10448
    char *tmp = NULL;
R
Richard W.M. Jones 已提交
10449 10450
    int fd = -1, ret = -1;

10451
    qemuDriverLock(driver);
10452
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
10453
    qemuDriverUnlock(driver);
R
Richard W.M. Jones 已提交
10454 10455

    if (!vm) {
10456 10457
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
10458 10459
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
10460 10461 10462
        goto cleanup;
    }

10463
    if (flags != VIR_MEMORY_VIRTUAL && flags != VIR_MEMORY_PHYSICAL) {
10464 10465
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("flags parameter must be VIR_MEMORY_VIRTUAL or VIR_MEMORY_PHYSICAL"));
10466
        goto cleanup;
R
Richard W.M. Jones 已提交
10467 10468
    }

10469 10470 10471
    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
10472
    if (!virDomainObjIsActive(vm)) {
10473 10474
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
10475
        goto endjob;
R
Richard W.M. Jones 已提交
10476 10477
    }

10478
    if (virAsprintf(&tmp, "%s/qemu.mem.XXXXXX", driver->cacheDir) < 0) {
10479
        virReportOOMError();
10480
        goto endjob;
10481 10482
    }

R
Richard W.M. Jones 已提交
10483 10484
    /* Create a temporary filename. */
    if ((fd = mkstemp (tmp)) == -1) {
10485 10486
        virReportSystemError(errno,
                             _("mkstemp(\"%s\") failed"), tmp);
10487
        goto endjob;
R
Richard W.M. Jones 已提交
10488 10489
    }

10490
    qemuDomainObjPrivatePtr priv = vm->privateData;
10491
    qemuDomainObjEnterMonitor(vm);
10492
    if (flags == VIR_MEMORY_VIRTUAL) {
10493 10494
        if (qemuMonitorSaveVirtualMemory(priv->mon, offset, size, tmp) < 0) {
            qemuDomainObjExitMonitor(vm);
10495
            goto endjob;
10496
        }
10497
    } else {
10498 10499
        if (qemuMonitorSavePhysicalMemory(priv->mon, offset, size, tmp) < 0) {
            qemuDomainObjExitMonitor(vm);
10500
            goto endjob;
10501
        }
R
Richard W.M. Jones 已提交
10502
    }
10503
    qemuDomainObjExitMonitor(vm);
R
Richard W.M. Jones 已提交
10504 10505 10506

    /* Read the memory file into buffer. */
    if (saferead (fd, buffer, size) == (ssize_t) -1) {
10507 10508 10509
        virReportSystemError(errno,
                             _("failed to read temporary file "
                               "created with template %s"), tmp);
10510
        goto endjob;
R
Richard W.M. Jones 已提交
10511 10512 10513
    }

    ret = 0;
10514

10515
endjob:
10516 10517
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
10518

10519
cleanup:
10520
    VIR_FREE(tmp);
R
Richard W.M. Jones 已提交
10521 10522
    if (fd >= 0) close (fd);
    unlink (tmp);
10523 10524
    if (vm)
        virDomainObjUnlock(vm);
R
Richard W.M. Jones 已提交
10525 10526 10527
    return ret;
}

10528

10529 10530 10531 10532 10533 10534 10535 10536 10537 10538
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;
10539
    virDomainDiskDefPtr disk = NULL;
10540 10541
    struct stat sb;
    int i;
10542
    int format;
10543 10544 10545 10546 10547 10548 10549 10550 10551 10552 10553 10554 10555 10556 10557 10558 10559 10560 10561 10562 10563 10564 10565 10566

    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)) {
10567
            disk = vm->def->disks[i];
10568 10569 10570 10571
            break;
        }
    }

10572
    if (!disk) {
10573 10574 10575 10576 10577 10578 10579 10580 10581 10582 10583 10584 10585 10586
        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 */
10587 10588 10589 10590 10591 10592 10593 10594
    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 {
10595 10596 10597 10598 10599 10600 10601
        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);
10602
            goto cleanup;
10603
        }
10604 10605 10606 10607 10608
    }

    if (virStorageFileGetMetadataFromFD(path, fd,
                                        format,
                                        &meta) < 0)
10609 10610 10611 10612 10613 10614 10615 10616 10617 10618
        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)) {
10619
#ifndef WIN32
10620 10621 10622 10623 10624 10625 10626 10627 10628 10629 10630 10631 10632 10633 10634 10635 10636 10637 10638 10639 10640 10641 10642 10643 10644 10645 10646 10647
        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;

10648
    /* Set default value .. */
10649 10650
    info->allocation = info->physical;

10651 10652 10653
    /* ..but if guest is running & not using raw
       disk format and on a block device, then query
       highest allocated extent from QEMU */
10654
    if (disk->type == VIR_DOMAIN_DISK_TYPE_BLOCK &&
10655
        format != VIR_STORAGE_FILE_RAW &&
10656 10657 10658 10659
        S_ISBLK(sb.st_mode)) {
        qemuDomainObjPrivatePtr priv = vm->privateData;
        if (qemuDomainObjBeginJob(vm) < 0)
            goto cleanup;
10660 10661 10662 10663 10664 10665 10666 10667 10668
        if (!virDomainObjIsActive(vm))
            ret = 0;
        else {
            qemuDomainObjEnterMonitor(vm);
            ret = qemuMonitorGetBlockExtent(priv->mon,
                                            disk->info.alias,
                                            &info->allocation);
            qemuDomainObjExitMonitor(vm);
        }
10669 10670 10671 10672 10673 10674

        if (qemuDomainObjEndJob(vm) == 0)
            vm = NULL;
    } else {
        ret = 0;
    }
10675 10676 10677 10678 10679 10680 10681 10682 10683 10684

cleanup:
    if (fd != -1)
        close(fd);
    if (vm)
        virDomainObjUnlock(vm);
    return ret;
}


10685
static int
10686 10687 10688 10689
qemuDomainEventRegister(virConnectPtr conn,
                        virConnectDomainEventCallback callback,
                        void *opaque,
                        virFreeCallback freecb)
10690
{
10691 10692 10693
    struct qemud_driver *driver = conn->privateData;
    int ret;

10694
    qemuDriverLock(driver);
10695 10696
    ret = virDomainEventCallbackListAdd(conn, driver->domainEventCallbacks,
                                        callback, opaque, freecb);
10697
    qemuDriverUnlock(driver);
10698

10699
    return ret;
10700 10701
}

10702

10703
static int
10704 10705
qemuDomainEventDeregister(virConnectPtr conn,
                          virConnectDomainEventCallback callback)
10706
{
10707 10708 10709
    struct qemud_driver *driver = conn->privateData;
    int ret;

10710
    qemuDriverLock(driver);
10711 10712 10713 10714 10715 10716
    if (driver->domainEventDispatching)
        ret = virDomainEventCallbackListMarkDelete(conn, driver->domainEventCallbacks,
                                                   callback);
    else
        ret = virDomainEventCallbackListRemove(conn, driver->domainEventCallbacks,
                                               callback);
10717
    qemuDriverUnlock(driver);
10718

10719
    return ret;
10720 10721
}

10722 10723 10724 10725 10726 10727 10728 10729 10730 10731 10732 10733 10734 10735 10736 10737 10738 10739 10740 10741 10742 10743 10744 10745 10746 10747 10748 10749 10750 10751 10752 10753 10754 10755 10756 10757 10758 10759 10760 10761 10762 10763 10764

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;
}


10765 10766
static void qemuDomainEventDispatchFunc(virConnectPtr conn,
                                        virDomainEventPtr event,
10767
                                        virConnectDomainEventGenericCallback cb,
10768 10769
                                        void *cbopaque,
                                        void *opaque)
10770
{
10771
    struct qemud_driver *driver = opaque;
10772

10773 10774 10775 10776 10777 10778 10779 10780 10781 10782 10783 10784 10785 10786 10787 10788 10789 10790 10791 10792 10793 10794 10795 10796 10797 10798 10799 10800 10801 10802 10803 10804 10805 10806 10807 10808 10809 10810 10811 10812 10813 10814 10815 10816
    /* 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);
10817 10818
}

D
Daniel Veillard 已提交
10819 10820
/* Migration support. */

10821 10822 10823 10824 10825
static bool ATTRIBUTE_NONNULL(1)
qemuDomainIsMigratable(virDomainDefPtr def)
{
    if (def->nhostdevs > 0) {
        qemuReportError(VIR_ERR_OPERATION_INVALID,
10826
            "%s", _("Domain with assigned host devices cannot be migrated"));
10827 10828 10829 10830 10831 10832 10833
        return false;
    }

    return true;
}


C
Chris Lalancette 已提交
10834 10835 10836 10837 10838 10839 10840 10841 10842 10843 10844 10845 10846 10847 10848 10849 10850 10851 10852
/* 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) {
10853 10854
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream is not open"));
C
Chris Lalancette 已提交
10855 10856 10857 10858 10859
        return -1;
    }

    qemuDriverLock(driver);
    if (qemust->watch == 0) {
10860 10861
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream does not have a callback registered"));
C
Chris Lalancette 已提交
10862 10863 10864 10865 10866 10867 10868 10869 10870 10871 10872 10873 10874 10875 10876 10877 10878 10879 10880 10881 10882 10883 10884 10885 10886 10887 10888 10889
        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) {
10890 10891
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream is not open"));
C
Chris Lalancette 已提交
10892 10893 10894 10895 10896
        return -1;
    }

    qemuDriverLock(driver);
    if (qemust->watch == 0) {
10897 10898
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream does not have a callback registered"));
C
Chris Lalancette 已提交
10899 10900 10901 10902 10903 10904 10905 10906 10907 10908 10909 10910 10911 10912 10913 10914 10915 10916 10917 10918 10919 10920 10921 10922 10923 10924 10925 10926 10927 10928 10929 10930 10931 10932 10933 10934 10935 10936 10937 10938 10939 10940 10941 10942 10943 10944 10945 10946 10947 10948 10949 10950 10951 10952 10953 10954 10955
        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) {
10956 10957
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream is not open"));
C
Chris Lalancette 已提交
10958 10959 10960 10961 10962
        return -1;
    }

    qemuDriverLock(driver);
    if (qemust->watch != 0) {
10963 10964
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream already has a callback registered"));
C
Chris Lalancette 已提交
10965 10966 10967 10968 10969 10970 10971 10972
        goto cleanup;
    }

    if ((qemust->watch = virEventAddHandle(qemust->fd,
                                           events,
                                           qemuStreamMigEvent,
                                           st,
                                           NULL)) < 0) {
10973 10974
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("cannot register file watch on stream"));
C
Chris Lalancette 已提交
10975 10976 10977 10978 10979 10980 10981 10982 10983 10984 10985 10986 10987 10988 10989 10990 10991 10992 10993 10994 10995 10996 10997 10998 10999 11000 11001 11002 11003 11004 11005 11006
        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;

11007
    if (VIR_ALLOC(qemust) < 0) {
11008
        virReportOOMError();
C
Chris Lalancette 已提交
11009
        return NULL;
11010
    }
C
Chris Lalancette 已提交
11011 11012 11013 11014 11015 11016 11017 11018 11019 11020 11021 11022 11023 11024 11025 11026 11027 11028 11029 11030 11031 11032 11033 11034 11035 11036 11037 11038 11039 11040 11041 11042 11043 11044 11045 11046 11047 11048 11049 11050 11051 11052 11053 11054 11055 11056 11057 11058 11059 11060 11061 11062 11063 11064 11065 11066 11067 11068 11069 11070 11071

    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) {
11072 11073
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream is not open"));
C
Chris Lalancette 已提交
11074 11075 11076 11077 11078 11079 11080 11081 11082 11083 11084 11085 11086 11087
        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;
11088
            virReportSystemError(errno, "%s",
C
Chris Lalancette 已提交
11089 11090 11091 11092 11093 11094 11095 11096 11097 11098 11099 11100 11101 11102 11103 11104 11105 11106 11107 11108 11109 11110 11111 11112 11113 11114 11115 11116 11117 11118 11119 11120 11121 11122 11123 11124 11125 11126
                                 _("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;
11127
    unsigned long long qemuCmdFlags;
C
Chris Lalancette 已提交
11128
    struct qemuStreamMigFile *qemust = NULL;
11129 11130 11131 11132 11133 11134 11135 11136
    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
11137 11138 11139

    qemuDriverLock(driver);
    if (!dom_xml) {
11140 11141
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("no domain XML passed"));
C
Chris Lalancette 已提交
11142 11143 11144
        goto cleanup;
    }
    if (!(flags & VIR_MIGRATE_TUNNELLED)) {
11145
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
C
Chris Lalancette 已提交
11146 11147 11148 11149
                         "%s", _("PrepareTunnel called but no TUNNELLED flag set"));
        goto cleanup;
    }
    if (st == NULL) {
11150 11151
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("tunnelled migration requested but NULL stream passed"));
C
Chris Lalancette 已提交
11152 11153 11154 11155
        goto cleanup;
    }

    /* Parse the domain XML. */
11156
    if (!(def = virDomainDefParseString(driver->caps, dom_xml,
C
Chris Lalancette 已提交
11157
                                        VIR_DOMAIN_XML_INACTIVE))) {
11158 11159
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to parse XML"));
C
Chris Lalancette 已提交
11160 11161 11162
        goto cleanup;
    }

11163 11164 11165
    if (!qemuDomainIsMigratable(def))
        goto cleanup;

C
Chris Lalancette 已提交
11166
    /* Target domain name, maybe renamed. */
11167 11168 11169 11170 11171 11172
    if (dname) {
        VIR_FREE(def->name);
        def->name = strdup(dname);
        if (def->name == NULL)
            goto cleanup;
    }
C
Chris Lalancette 已提交
11173

11174 11175
    if (virDomainObjIsDuplicate(&driver->domains, def, 1) < 0)
        goto cleanup;
C
Chris Lalancette 已提交
11176

11177
    if (!(vm = virDomainAssignDef(driver->caps,
C
Chris Lalancette 已提交
11178
                                  &driver->domains,
11179
                                  def, true))) {
11180
        /* virDomainAssignDef already set the error */
C
Chris Lalancette 已提交
11181 11182 11183
        goto cleanup;
    }
    def = NULL;
11184
    priv = vm->privateData;
C
Chris Lalancette 已提交
11185

11186 11187
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;
11188
    priv->jobActive = QEMU_JOB_MIGRATION_OUT;
11189

C
Chris Lalancette 已提交
11190 11191 11192 11193
    /* Domain starts inactive, even if the domain XML had an id field. */
    vm->def->id = -1;

    if (virAsprintf(&unixfile, "%s/qemu.tunnelmigrate.dest.%s",
11194
                    driver->libDir, vm->def->name) < 0) {
11195
        virReportOOMError();
11196
        goto endjob;
C
Chris Lalancette 已提交
11197 11198 11199 11200 11201
    }
    unlink(unixfile);

    /* check that this qemu version supports the interactive exec */
    if (qemudExtractVersionInfo(vm->def->emulator, NULL, &qemuCmdFlags) < 0) {
11202 11203 11204
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Cannot determine QEMU argv syntax %s"),
                        vm->def->emulator);
11205
        goto endjob;
C
Chris Lalancette 已提交
11206 11207 11208 11209 11210 11211
    }
    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 {
11212 11213
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("Destination qemu is too old to support tunnelled migration"));
11214
        goto endjob;
C
Chris Lalancette 已提交
11215 11216
    }
    if (internalret < 0) {
11217
        virReportOOMError();
11218
        goto endjob;
C
Chris Lalancette 已提交
11219 11220 11221 11222
    }
    /* Start the QEMU daemon, with the same command-line arguments plus
     * -incoming unix:/path/to/file or exec:nc -U /path/to/file
     */
11223 11224
    internalret = qemudStartVMDaemon(dconn, driver, vm, migrateFrom, true,
                                     -1, NULL);
C
Chris Lalancette 已提交
11225 11226
    VIR_FREE(migrateFrom);
    if (internalret < 0) {
11227
        qemuDomainStartAudit(vm, "migrated", false);
C
Chris Lalancette 已提交
11228 11229 11230 11231 11232 11233 11234
        /* 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;
        }
11235
        goto endjob;
C
Chris Lalancette 已提交
11236 11237 11238 11239
    }

    qemust = qemuStreamMigOpen(st, unixfile);
    if (qemust == NULL) {
11240
        qemuDomainStartAudit(vm, "migrated", false);
11241
        qemudShutdownVMDaemon(driver, vm, 0);
11242
        if (!vm->persistent) {
11243 11244
            if (qemuDomainObjEndJob(vm) > 0)
                virDomainRemoveInactive(&driver->domains, vm);
11245 11246
            vm = NULL;
        }
11247
        virReportSystemError(errno,
C
Chris Lalancette 已提交
11248 11249
                             _("cannot open unix socket '%s' for tunnelled migration"),
                             unixfile);
11250
        goto endjob;
C
Chris Lalancette 已提交
11251 11252 11253 11254 11255
    }

    st->driver = &qemuStreamMigDrv;
    st->privateData = qemust;

11256
    qemuDomainStartAudit(vm, "migrated", true);
C
Chris Lalancette 已提交
11257 11258 11259 11260 11261
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_MIGRATED);
    ret = 0;

11262
endjob:
11263 11264 11265
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
11266

11267 11268 11269 11270 11271 11272 11273 11274 11275 11276 11277 11278
    /* 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
11279 11280
cleanup:
    virDomainDefFree(def);
11281 11282
    if (unixfile)
        unlink(unixfile);
C
Chris Lalancette 已提交
11283 11284 11285 11286 11287 11288 11289 11290 11291
    VIR_FREE(unixfile);
    if (vm)
        virDomainObjUnlock(vm);
    if (event)
        qemuDomainEventQueue(driver, event);
    qemuDriverUnlock(driver);
    return ret;
}

D
Daniel Veillard 已提交
11292 11293 11294 11295
/* Prepare is the first step, and it runs on the destination host.
 *
 * This starts an empty VM listening on a TCP port.
 */
11296
static int ATTRIBUTE_NONNULL (5)
D
Daniel Veillard 已提交
11297 11298 11299 11300 11301
qemudDomainMigratePrepare2 (virConnectPtr dconn,
                            char **cookie ATTRIBUTE_UNUSED,
                            int *cookielen ATTRIBUTE_UNUSED,
                            const char *uri_in,
                            char **uri_out,
C
Chris Lalancette 已提交
11302
                            unsigned long flags,
D
Daniel Veillard 已提交
11303 11304 11305 11306 11307
                            const char *dname,
                            unsigned long resource ATTRIBUTE_UNUSED,
                            const char *dom_xml)
{
    static int port = 0;
11308 11309
    struct qemud_driver *driver = dconn->privateData;
    virDomainDefPtr def = NULL;
D
Daniel Veillard 已提交
11310 11311
    virDomainObjPtr vm = NULL;
    int this_port;
C
Chris Lalancette 已提交
11312
    char *hostname = NULL;
D
Daniel Veillard 已提交
11313 11314
    char migrateFrom [64];
    const char *p;
11315
    virDomainEventPtr event = NULL;
11316
    int ret = -1;
11317
    int internalret;
11318 11319 11320 11321 11322 11323 11324 11325
    qemuDomainObjPrivatePtr priv = NULL;
    struct timeval now;

    if (gettimeofday(&now, NULL) < 0) {
        virReportSystemError(errno, "%s",
                             _("cannot get time of day"));
        return -1;
    }
11326

C
Chris Lalancette 已提交
11327 11328 11329 11330 11331 11332 11333 11334 11335
    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);

11336
    *uri_out = NULL;
D
Daniel Veillard 已提交
11337

11338
    qemuDriverLock(driver);
C
Chris Lalancette 已提交
11339 11340 11341 11342
    if (flags & VIR_MIGRATE_TUNNELLED) {
        /* this is a logical error; we never should have gotten here with
         * VIR_MIGRATE_TUNNELLED set
         */
11343 11344
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("Tunnelled migration requested but invalid RPC method called"));
C
Chris Lalancette 已提交
11345 11346 11347
        goto cleanup;
    }

D
Daniel Veillard 已提交
11348
    if (!dom_xml) {
11349 11350
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("no domain XML passed"));
11351
        goto cleanup;
D
Daniel Veillard 已提交
11352 11353 11354 11355 11356 11357 11358 11359 11360 11361 11362 11363 11364 11365 11366 11367 11368
    }

    /* 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
11369
        if ((hostname = virGetHostname(NULL)) == NULL)
11370
            goto cleanup;
D
Daniel Veillard 已提交
11371

C
Chris Lalancette 已提交
11372 11373 11374 11375 11376 11377
        if (STRPREFIX(hostname, "localhost")) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("hostname on destination resolved to localhost, but migration requires an FQDN"));
            goto cleanup;
        }

11378 11379 11380 11381 11382
        /* 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
11383
        /* Caller frees */
11384 11385
        internalret = virAsprintf(uri_out, "tcp:%s:%d", hostname, this_port);
        if (internalret < 0) {
11386
            virReportOOMError();
11387
            goto cleanup;
D
Daniel Veillard 已提交
11388 11389 11390 11391 11392 11393
        }
    } 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.
         */
11394
        if (!STRPREFIX (uri_in, "tcp:")) {
11395 11396
            qemuReportError (VIR_ERR_INVALID_ARG,
                             "%s", _("only tcp URIs are supported for KVM/QEMU migrations"));
11397
            goto cleanup;
D
Daniel Veillard 已提交
11398 11399 11400 11401
        }

        /* Get the port number. */
        p = strrchr (uri_in, ':');
11402 11403 11404 11405 11406 11407 11408 11409
        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) {
11410
                virReportOOMError();
11411 11412 11413 11414 11415 11416 11417
                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)) {
11418 11419
                qemuReportError(VIR_ERR_INVALID_ARG,
                                "%s", _("URI ended with incorrect ':port'"));
11420 11421
                goto cleanup;
            }
D
Daniel Veillard 已提交
11422 11423 11424
        }
    }

11425
    if (*uri_out)
11426 11427
        VIR_DEBUG("Generated uri_out=%s", *uri_out);

D
Daniel Veillard 已提交
11428
    /* Parse the domain XML. */
11429
    if (!(def = virDomainDefParseString(driver->caps, dom_xml,
11430
                                        VIR_DOMAIN_XML_INACTIVE))) {
11431 11432
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to parse XML"));
11433
        goto cleanup;
D
Daniel Veillard 已提交
11434 11435
    }

11436 11437 11438
    if (!qemuDomainIsMigratable(def))
        goto cleanup;

D
Daniel Veillard 已提交
11439
    /* Target domain name, maybe renamed. */
11440 11441 11442 11443 11444 11445
    if (dname) {
        VIR_FREE(def->name);
        def->name = strdup(dname);
        if (def->name == NULL)
            goto cleanup;
    }
D
Daniel Veillard 已提交
11446

11447 11448
    if (virDomainObjIsDuplicate(&driver->domains, def, 1) < 0)
        goto cleanup;
D
Daniel Veillard 已提交
11449

11450
    if (!(vm = virDomainAssignDef(driver->caps,
D
Daniel Veillard 已提交
11451
                                  &driver->domains,
11452
                                  def, true))) {
11453
        /* virDomainAssignDef already set the error */
11454
        goto cleanup;
D
Daniel Veillard 已提交
11455
    }
11456
    def = NULL;
11457
    priv = vm->privateData;
D
Daniel Veillard 已提交
11458

11459 11460
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;
11461
    priv->jobActive = QEMU_JOB_MIGRATION_OUT;
11462

D
Daniel Veillard 已提交
11463 11464 11465 11466 11467 11468 11469
    /* 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);
11470 11471
    if (qemudStartVMDaemon (dconn, driver, vm, migrateFrom, true,
                            -1, NULL) < 0) {
11472
        qemuDomainStartAudit(vm, "migrated", false);
11473 11474 11475
        /* Note that we don't set an error here because qemudStartVMDaemon
         * should have already done that.
         */
11476
        if (!vm->persistent) {
11477 11478
            if (qemuDomainObjEndJob(vm) > 0)
                virDomainRemoveInactive(&driver->domains, vm);
11479 11480
            vm = NULL;
        }
11481
        goto endjob;
D
Daniel Veillard 已提交
11482
    }
11483

11484
    qemuDomainStartAudit(vm, "migrated", true);
11485 11486 11487
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_MIGRATED);
11488
    ret = 0;
D
Daniel Veillard 已提交
11489

11490
endjob:
11491 11492 11493
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
11494

11495 11496 11497 11498 11499 11500 11501 11502 11503 11504 11505 11506
    /* 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);
    }

11507
cleanup:
C
Chris Lalancette 已提交
11508
    VIR_FREE(hostname);
11509
    virDomainDefFree(def);
C
Chris Lalancette 已提交
11510
    if (ret != 0)
11511
        VIR_FREE(*uri_out);
11512 11513
    if (vm)
        virDomainObjUnlock(vm);
11514 11515
    if (event)
        qemuDomainEventQueue(driver, event);
11516
    qemuDriverUnlock(driver);
11517
    return ret;
C
Chris Lalancette 已提交
11518 11519 11520

}

11521 11522 11523 11524

/* Perform migration using QEMU's native TCP migrate support,
 * not encrypted obviously
 */
11525
static int doNativeMigrate(struct qemud_driver *driver,
11526 11527
                           virDomainObjPtr vm,
                           const char *uri,
11528
                           unsigned int flags,
11529 11530 11531 11532
                           const char *dname ATTRIBUTE_UNUSED,
                           unsigned long resource)
{
    int ret = -1;
11533
    xmlURIPtr uribits = NULL;
11534
    qemuDomainObjPrivatePtr priv = vm->privateData;
11535
    unsigned int background_flags = QEMU_MONITOR_MIGRATE_BACKGROUND;
11536

11537 11538 11539 11540 11541
    /* 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) {
11542
            virReportOOMError();
11543 11544 11545 11546 11547 11548 11549 11550
            goto cleanup;
        }
        uribits = xmlParseURI(tmpuri);
        VIR_FREE(tmpuri);
    } else {
        uribits = xmlParseURI(uri);
    }
    if (!uribits) {
11551 11552
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("cannot parse URI %s"), uri);
11553 11554 11555
        goto cleanup;
    }

11556
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
11557
    if (resource > 0 &&
11558
        qemuMonitorSetMigrationSpeed(priv->mon, resource) < 0) {
11559
        qemuDomainObjExitMonitorWithDriver(driver, vm);
11560
        goto cleanup;
11561
    }
11562

11563 11564 11565 11566 11567 11568 11569 11570
    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) {
11571
        qemuDomainObjExitMonitorWithDriver(driver, vm);
11572 11573
        goto cleanup;
    }
11574
    qemuDomainObjExitMonitorWithDriver(driver, vm);
11575

11576
    if (qemuDomainWaitForMigrationComplete(driver, vm) < 0)
11577 11578 11579 11580 11581 11582 11583 11584 11585 11586
        goto cleanup;

    ret = 0;

cleanup:
    xmlFreeURI(uribits);
    return ret;
}


11587 11588
#define TUNNEL_SEND_BUF_SIZE 65536

11589
static int doTunnelSendAll(virStreamPtr st,
11590 11591
                           int sock)
{
11592 11593 11594 11595 11596 11597 11598 11599
    char *buffer;
    int nbytes = TUNNEL_SEND_BUF_SIZE;

    if (VIR_ALLOC_N(buffer, TUNNEL_SEND_BUF_SIZE) < 0) {
        virReportOOMError();
        virStreamAbort(st);
        return -1;
    }
11600 11601 11602 11603 11604

    /* XXX should honour the 'resource' parameter here */
    for (;;) {
        nbytes = saferead(sock, buffer, nbytes);
        if (nbytes < 0) {
11605
            virReportSystemError(errno, "%s",
11606
                                 _("tunnelled migration failed to read from qemu"));
11607 11608
            virStreamAbort(st);
            VIR_FREE(buffer);
11609 11610 11611 11612 11613 11614 11615
            return -1;
        }
        else if (nbytes == 0)
            /* EOF; get out of here */
            break;

        if (virStreamSend(st, buffer, nbytes) < 0) {
11616 11617
            qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
                            _("Failed to write migration data to remote libvirtd"));
11618
            VIR_FREE(buffer);
11619 11620 11621 11622
            return -1;
        }
    }

11623 11624
    VIR_FREE(buffer);

11625 11626 11627 11628 11629 11630 11631
    if (virStreamFinish(st) < 0)
        /* virStreamFinish set the error for us */
        return -1;

    return 0;
}

C
Chris Lalancette 已提交
11632
static int doTunnelMigrate(virDomainPtr dom,
11633
                           struct qemud_driver *driver,
11634
                           virConnectPtr dconn,
C
Chris Lalancette 已提交
11635
                           virDomainObjPtr vm,
11636
                           const char *dom_xml,
C
Chris Lalancette 已提交
11637 11638 11639 11640 11641
                           const char *uri,
                           unsigned long flags,
                           const char *dname,
                           unsigned long resource)
{
11642
    qemuDomainObjPrivatePtr priv = vm->privateData;
11643 11644
    int client_sock = -1;
    int qemu_sock = -1;
C
Chris Lalancette 已提交
11645 11646
    struct sockaddr_un sa_qemu, sa_client;
    socklen_t addrlen;
11647
    virDomainPtr ddomain = NULL;
C
Chris Lalancette 已提交
11648
    int retval = -1;
11649
    virStreamPtr st = NULL;
C
Chris Lalancette 已提交
11650 11651
    char *unixfile = NULL;
    int internalret;
11652
    unsigned long long qemuCmdFlags;
C
Chris Lalancette 已提交
11653 11654
    int status;
    unsigned long long transferred, remaining, total;
11655
    unsigned int background_flags = QEMU_MONITOR_MIGRATE_BACKGROUND;
C
Chris Lalancette 已提交
11656

11657 11658 11659 11660 11661 11662 11663 11664
    /*
     * 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
11665 11666
     */

11667

11668
    /* Stage 1. setup local support infrastructure */
C
Chris Lalancette 已提交
11669 11670

    if (virAsprintf(&unixfile, "%s/qemu.tunnelmigrate.src.%s",
11671
                    driver->libDir, vm->def->name) < 0) {
11672
        virReportOOMError();
11673
        goto cleanup;
C
Chris Lalancette 已提交
11674 11675 11676 11677
    }

    qemu_sock = socket(AF_UNIX, SOCK_STREAM, 0);
    if (qemu_sock < 0) {
11678
        virReportSystemError(errno, "%s",
C
Chris Lalancette 已提交
11679
                             _("cannot open tunnelled migration socket"));
11680
        goto cleanup;
C
Chris Lalancette 已提交
11681 11682 11683 11684 11685
    }
    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) {
11686 11687 11688
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Unix socket '%s' too big for destination"),
                        unixfile);
11689
        goto cleanup;
C
Chris Lalancette 已提交
11690 11691 11692
    }
    unlink(unixfile);
    if (bind(qemu_sock, (struct sockaddr *)&sa_qemu, sizeof(sa_qemu)) < 0) {
11693
        virReportSystemError(errno,
C
Chris Lalancette 已提交
11694 11695
                             _("Cannot bind to unix socket '%s' for tunnelled migration"),
                             unixfile);
11696
        goto cleanup;
C
Chris Lalancette 已提交
11697 11698
    }
    if (listen(qemu_sock, 1) < 0) {
11699
        virReportSystemError(errno,
C
Chris Lalancette 已提交
11700 11701
                             _("Cannot listen on unix socket '%s' for tunnelled migration"),
                             unixfile);
11702
        goto cleanup;
C
Chris Lalancette 已提交
11703 11704
    }

11705 11706 11707 11708 11709 11710 11711
    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
11712 11713
    /* check that this qemu version supports the unix migration */
    if (qemudExtractVersionInfo(vm->def->emulator, NULL, &qemuCmdFlags) < 0) {
11714 11715 11716
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Cannot extract Qemu version from '%s'"),
                        vm->def->emulator);
11717 11718 11719 11720 11721
        goto cleanup;
    }

    if (!(qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_UNIX) &&
        !(qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_EXEC)) {
11722 11723
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("Source qemu is too old to support tunnelled migration"));
11724
        goto cleanup;
C
Chris Lalancette 已提交
11725
    }
11726 11727 11728 11729 11730 11731 11732 11733 11734 11735 11736 11737


    /* 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;

11738
    qemuDomainObjEnterRemoteWithDriver(driver, vm);
11739 11740 11741
    internalret = dconn->driver->domainMigratePrepareTunnel(dconn, st,
                                                            flags, dname,
                                                            resource, dom_xml);
11742
    qemuDomainObjExitRemoteWithDriver(driver, vm);
11743 11744 11745 11746 11747

    if (internalret < 0)
        /* domainMigratePrepareTunnel sets the error for us */
        goto cleanup;

11748 11749 11750 11751 11752 11753 11754 11755 11756
    /* 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;
    }

11757
    /*   3. start migration on source */
11758
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
11759 11760 11761 11762 11763 11764 11765 11766
    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
11767 11768
    else if (qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_EXEC) {
        const char *args[] = { "nc", "-U", unixfile, NULL };
11769
        internalret = qemuMonitorMigrateToCommand(priv->mon, QEMU_MONITOR_MIGRATE_BACKGROUND, args);
11770 11771
    } else {
        internalret = -1;
C
Chris Lalancette 已提交
11772
    }
11773
    qemuDomainObjExitMonitorWithDriver(driver, vm);
C
Chris Lalancette 已提交
11774
    if (internalret < 0) {
11775 11776
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("tunnelled migration monitor command failed"));
11777
        goto finish;
C
Chris Lalancette 已提交
11778 11779
    }

11780 11781 11782 11783 11784 11785
    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("guest unexpectedly quit"));
        goto cleanup;
    }

11786 11787 11788
    /* From this point onwards we *must* call cancel to abort the
     * migration on source if anything goes wrong */

C
Chris Lalancette 已提交
11789 11790 11791
    /* it is also possible that the migrate didn't fail initially, but
     * rather failed later on.  Check the output of "info migrate"
     */
11792
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
11793 11794
    if (qemuMonitorGetMigrationStatus(priv->mon,
                                      &status,
C
Chris Lalancette 已提交
11795 11796 11797
                                      &transferred,
                                      &remaining,
                                      &total) < 0) {
11798
        qemuDomainObjExitMonitorWithDriver(driver, vm);
11799
        goto cancel;
C
Chris Lalancette 已提交
11800
    }
11801
    qemuDomainObjExitMonitorWithDriver(driver, vm);
C
Chris Lalancette 已提交
11802 11803

    if (status == QEMU_MONITOR_MIGRATION_STATUS_ERROR) {
11804 11805
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s",_("migrate failed"));
11806
        goto cancel;
C
Chris Lalancette 已提交
11807 11808 11809 11810 11811 11812
    }

    addrlen = sizeof(sa_client);
    while ((client_sock = accept(qemu_sock, (struct sockaddr *)&sa_client, &addrlen)) < 0) {
        if (errno == EAGAIN || errno == EINTR)
            continue;
11813
        virReportSystemError(errno, "%s",
C
Chris Lalancette 已提交
11814
                             _("tunnelled migration failed to accept from qemu"));
11815
        goto cancel;
C
Chris Lalancette 已提交
11816 11817
    }

11818
    retval = doTunnelSendAll(st, client_sock);
11819

11820
cancel:
11821
    if (retval != 0 && virDomainObjIsActive(vm)) {
11822
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
11823
        qemuMonitorMigrateCancel(priv->mon);
11824
        qemuDomainObjExitMonitorWithDriver(driver, vm);
11825
    }
C
Chris Lalancette 已提交
11826

11827
finish:
C
Chris Lalancette 已提交
11828
    dname = dname ? dname : dom->name;
11829
    qemuDomainObjEnterRemoteWithDriver(driver, vm);
C
Chris Lalancette 已提交
11830 11831
    ddomain = dconn->driver->domainMigrateFinish2
        (dconn, dname, NULL, 0, uri, flags, retval);
11832
    qemuDomainObjExitRemoteWithDriver(driver, vm);
11833 11834 11835 11836 11837 11838 11839

cleanup:
    if (client_sock != -1)
        close(client_sock);
    if (qemu_sock != -1)
        close(qemu_sock);

C
Chris Lalancette 已提交
11840 11841 11842
    if (ddomain)
        virUnrefDomain(ddomain);

11843 11844 11845 11846
    if (unixfile) {
        unlink(unixfile);
        VIR_FREE(unixfile);
    }
C
Chris Lalancette 已提交
11847

11848 11849 11850
    if (st)
        /* don't call virStreamFree(), because that resets any pending errors */
        virUnrefStream(st);
11851 11852 11853 11854
    return retval;
}


11855 11856 11857 11858
/* 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,
11859
                              struct qemud_driver *driver,
11860 11861 11862 11863 11864 11865 11866 11867 11868 11869 11870
                              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;
11871
    int rc;
11872

11873
    qemuDomainObjEnterRemoteWithDriver(driver, vm);
11874 11875
    /* NB we don't pass 'uri' into this, since that's the libvirtd
     * URI in this context - so we let dest pick it */
11876 11877 11878 11879 11880 11881 11882 11883 11884
    rc = dconn->driver->domainMigratePrepare2(dconn,
                                              NULL, /* cookie */
                                              0, /* cookielen */
                                              NULL, /* uri */
                                              &uri_out,
                                              flags, dname,
                                              resource, dom_xml);
    qemuDomainObjExitRemoteWithDriver(driver, vm);
    if (rc < 0)
11885 11886 11887
        /* domainMigratePrepare2 sets the error for us */
        goto cleanup;

11888 11889 11890 11891 11892 11893 11894 11895 11896
    /* 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;
    }

11897
    if (uri_out == NULL) {
11898 11899
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("domainMigratePrepare2 did not set uri"));
11900
        goto cleanup;
11901 11902
    }

11903
    if (doNativeMigrate(driver, vm, uri_out, flags, dname, resource) < 0)
11904 11905 11906 11907 11908 11909
        goto finish;

    retval = 0;

finish:
    dname = dname ? dname : dom->name;
11910
    qemuDomainObjEnterRemoteWithDriver(driver, vm);
11911 11912
    ddomain = dconn->driver->domainMigrateFinish2
        (dconn, dname, NULL, 0, uri_out, flags, retval);
11913
    qemuDomainObjExitRemoteWithDriver(driver, vm);
11914 11915 11916 11917 11918 11919 11920 11921 11922

    if (ddomain)
        virUnrefDomain(ddomain);

cleanup:
    return retval;
}


11923
static int doPeer2PeerMigrate(virDomainPtr dom,
11924
                              struct qemud_driver *driver,
11925 11926 11927 11928 11929 11930 11931 11932 11933 11934 11935 11936 11937 11938 11939 11940
                              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) {
11941 11942
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("Failed to connect to remote libvirt URI %s"), uri);
11943 11944 11945
        return -1;
    }
    if (!VIR_DRV_SUPPORTS_FEATURE(dconn->driver, dconn,
11946
                                  VIR_DRV_FEATURE_MIGRATION_P2P)) {
11947 11948
        qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
                        _("Destination libvirt does not support peer-to-peer migration protocol"));
11949 11950 11951
        goto cleanup;
    }

11952 11953 11954
    dom_xml = qemudVMDumpXML(driver, vm,
                             VIR_DOMAIN_XML_SECURE |
                             VIR_DOMAIN_XML_UPDATE_CPU);
11955
    if (!dom_xml) {
11956 11957
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to get domain xml"));
11958 11959 11960
        goto cleanup;
    }

11961
    if (flags & VIR_MIGRATE_TUNNELLED)
11962
        ret = doTunnelMigrate(dom, driver, dconn, vm, dom_xml, uri, flags, dname, resource);
11963
    else
11964
        ret = doNonTunnelMigrate(dom, driver, dconn, vm, dom_xml, uri, flags, dname, resource);
11965 11966 11967

cleanup:
    VIR_FREE(dom_xml);
C
Chris Lalancette 已提交
11968 11969 11970
    /* don't call virConnectClose(), because that resets any pending errors */
    virUnrefConnect(dconn);

11971
    return ret;
D
Daniel Veillard 已提交
11972 11973
}

11974

D
Daniel Veillard 已提交
11975 11976 11977 11978 11979 11980
/* 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,
11981
                           unsigned long flags,
11982
                           const char *dname,
D
Daniel Veillard 已提交
11983 11984
                           unsigned long resource)
{
11985 11986
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
11987
    virDomainEventPtr event = NULL;
11988
    int ret = -1;
11989
    int resume = 0;
11990
    qemuDomainObjPrivatePtr priv;
D
Daniel Veillard 已提交
11991

C
Chris Lalancette 已提交
11992 11993 11994 11995 11996 11997 11998 11999 12000
    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);

12001
    qemuDriverLock(driver);
12002
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
D
Daniel Veillard 已提交
12003
    if (!vm) {
12004 12005
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
12006 12007
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
12008
        goto cleanup;
D
Daniel Veillard 已提交
12009
    }
12010
    priv = vm->privateData;
D
Daniel Veillard 已提交
12011

12012 12013
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;
12014
    priv->jobActive = QEMU_JOB_MIGRATION_OUT;
12015

D
Daniel P. Berrange 已提交
12016
    if (!virDomainObjIsActive(vm)) {
12017 12018
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
12019
        goto endjob;
D
Daniel Veillard 已提交
12020 12021
    }

12022 12023 12024
    memset(&priv->jobInfo, 0, sizeof(priv->jobInfo));
    priv->jobInfo.type = VIR_DOMAIN_JOB_UNBOUNDED;

12025
    resume = vm->state == VIR_DOMAIN_RUNNING;
12026
    if (!(flags & VIR_MIGRATE_LIVE) && vm->state == VIR_DOMAIN_RUNNING) {
12027
        if (qemuDomainMigrateOffline(driver, vm) < 0)
12028
            goto endjob;
12029 12030
    }

12031
    if ((flags & (VIR_MIGRATE_TUNNELLED | VIR_MIGRATE_PEER2PEER))) {
12032
        if (doPeer2PeerMigrate(dom, driver, vm, uri, flags, dname, resource) < 0)
12033
            /* doPeer2PeerMigrate already set the error, so just get out */
12034
            goto endjob;
12035
    } else {
12036
        if (doNativeMigrate(driver, vm, uri, flags, dname, resource) < 0)
12037
            goto endjob;
12038 12039
    }

D
Daniel Veillard 已提交
12040
    /* Clean up the source domain. */
12041
    qemudShutdownVMDaemon(driver, vm, 1);
12042
    qemuDomainStopAudit(vm, "migrated");
12043
    resume = 0;
12044 12045 12046 12047

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_MIGRATED);
C
Chris Lalancette 已提交
12048
    if (!vm->persistent || (flags & VIR_MIGRATE_UNDEFINE_SOURCE)) {
12049
        virDomainDeleteConfig(driver->configDir, driver->autostartDir, vm);
12050 12051
        if (qemuDomainObjEndJob(vm) > 0)
            virDomainRemoveInactive(&driver->domains, vm);
12052 12053
        vm = NULL;
    }
12054
    ret = 0;
D
Daniel Veillard 已提交
12055

12056
endjob:
12057
    if (resume && vm->state == VIR_DOMAIN_PAUSED) {
12058
        /* we got here through some sort of failure; start the domain again */
12059
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
12060
        if (qemuMonitorStartCPUs(priv->mon, dom->conn) < 0) {
12061 12062 12063 12064
            /* 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
             */
12065
            VIR_ERROR(_("Failed to resume guest %s after failure"),
12066
                      vm->def->name);
12067
        }
12068
        qemuDomainObjExitMonitorWithDriver(driver, vm);
12069

12070
        vm->state = VIR_DOMAIN_RUNNING;
12071 12072 12073 12074
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_MIGRATED);
    }
12075 12076 12077
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
12078

12079
cleanup:
12080 12081
    if (vm)
        virDomainObjUnlock(vm);
12082 12083
    if (event)
        qemuDomainEventQueue(driver, event);
12084
    qemuDriverUnlock(driver);
12085
    return ret;
D
Daniel Veillard 已提交
12086 12087 12088 12089 12090 12091 12092 12093 12094
}

/* 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
12095
                           unsigned long flags,
D
Daniel Veillard 已提交
12096 12097
                           int retcode)
{
12098 12099 12100
    struct qemud_driver *driver = dconn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
12101
    virDomainEventPtr event = NULL;
12102
    virErrorPtr orig_err;
C
Chris Lalancette 已提交
12103
    int newVM = 1;
12104
    qemuDomainObjPrivatePtr priv = NULL;
D
Daniel Veillard 已提交
12105

C
Chris Lalancette 已提交
12106 12107 12108 12109 12110 12111 12112 12113 12114
    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);

12115 12116 12117
    /* Migration failed. Save the current error so nothing squashes it */
    orig_err = virSaveLastError();

12118
    qemuDriverLock(driver);
12119
    vm = virDomainFindByName(&driver->domains, dname);
D
Daniel Veillard 已提交
12120
    if (!vm) {
12121 12122
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching name '%s'"), dname);
12123
        goto cleanup;
D
Daniel Veillard 已提交
12124 12125
    }

12126 12127 12128 12129 12130 12131 12132 12133 12134
    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));

12135 12136 12137
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

D
Daniel Veillard 已提交
12138 12139 12140 12141
    /* Did the migration go as planned?  If yes, return the domain
     * object, but if no, clean up the empty qemu process.
     */
    if (retcode == 0) {
12142 12143 12144 12145 12146 12147
        if (!virDomainObjIsActive(vm)) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("guest unexpectedly quit"));
            goto cleanup;
        }

C
Chris Lalancette 已提交
12148 12149 12150 12151 12152
        if (flags & VIR_MIGRATE_PERSIST_DEST) {
            if (vm->persistent)
                newVM = 0;
            vm->persistent = 1;

12153
            if (virDomainSaveConfig(driver->configDir, vm->def) < 0) {
C
Chris Lalancette 已提交
12154 12155 12156 12157 12158 12159 12160 12161 12162 12163
                /* 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;
12164
                goto endjob;
C
Chris Lalancette 已提交
12165 12166 12167 12168 12169 12170 12171 12172 12173
            }

            event = virDomainEventNewFromObj(vm,
                                             VIR_DOMAIN_EVENT_DEFINED,
                                             newVM ?
                                             VIR_DOMAIN_EVENT_DEFINED_ADDED :
                                             VIR_DOMAIN_EVENT_DEFINED_UPDATED);
            if (event)
                qemuDomainEventQueue(driver, event);
12174
            event = NULL;
C
Chris Lalancette 已提交
12175 12176

        }
D
Daniel Veillard 已提交
12177
        dom = virGetDomain (dconn, vm->def->name, vm->def->uuid);
12178

12179 12180 12181 12182 12183 12184 12185 12186
        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)
12187 12188
                    qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                    "%s", _("resume operation failed"));
12189 12190 12191
                qemuDomainObjExitMonitorWithDriver(driver, vm);
                goto endjob;
            }
12192
            qemuDomainObjExitMonitorWithDriver(driver, vm);
12193 12194

            vm->state = VIR_DOMAIN_RUNNING;
12195 12196
        }

12197 12198 12199
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_MIGRATED);
12200 12201 12202 12203 12204 12205
        if (vm->state == VIR_DOMAIN_PAUSED) {
            qemuDomainEventQueue(driver, event);
            event = virDomainEventNewFromObj(vm,
                                             VIR_DOMAIN_EVENT_SUSPENDED,
                                             VIR_DOMAIN_EVENT_SUSPENDED_PAUSED);
        }
12206 12207 12208 12209
        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
12210
    } else {
12211
        qemudShutdownVMDaemon(driver, vm, 1);
12212
        qemuDomainStopAudit(vm, "failed");
12213 12214 12215
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         VIR_DOMAIN_EVENT_STOPPED_FAILED);
12216
        if (!vm->persistent) {
12217 12218
            if (qemuDomainObjEndJob(vm) > 0)
                virDomainRemoveInactive(&driver->domains, vm);
12219 12220
            vm = NULL;
        }
D
Daniel Veillard 已提交
12221
    }
12222

12223
endjob:
12224 12225 12226
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
12227

12228
cleanup:
12229 12230 12231 12232
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
12233 12234
    if (vm)
        virDomainObjUnlock(vm);
12235 12236
    if (event)
        qemuDomainEventQueue(driver, event);
12237
    qemuDriverUnlock(driver);
12238
    return dom;
D
Daniel Veillard 已提交
12239 12240
}

12241 12242 12243 12244 12245 12246 12247 12248 12249 12250 12251 12252 12253 12254 12255 12256
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;

12257
    def = virNodeDeviceDefParseString(xml, EXISTING_DEVICE);
12258 12259 12260 12261 12262 12263 12264 12265 12266 12267 12268 12269 12270 12271 12272 12273 12274
    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) {
12275 12276
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("device %s is not a PCI device"), dev->name);
12277 12278 12279 12280 12281 12282 12283 12284 12285 12286 12287 12288 12289
        goto out;
    }

    ret = 0;
out:
    virNodeDeviceDefFree(def);
    VIR_FREE(xml);
    return ret;
}

static int
qemudNodeDeviceDettach (virNodeDevicePtr dev)
{
12290
    struct qemud_driver *driver = dev->conn->privateData;
12291 12292 12293 12294 12295 12296 12297
    pciDevice *pci;
    unsigned domain, bus, slot, function;
    int ret = -1;

    if (qemudNodeDeviceGetPciInfo(dev, &domain, &bus, &slot, &function) < 0)
        return -1;

12298
    pci = pciGetDevice(domain, bus, slot, function);
12299 12300 12301
    if (!pci)
        return -1;

12302 12303
    qemuDriverLock(driver);
    if (pciDettachDevice(pci, driver->activePciHostdevs) < 0)
12304 12305 12306 12307
        goto out;

    ret = 0;
out:
12308
    qemuDriverUnlock(driver);
12309
    pciFreeDevice(pci);
12310 12311 12312 12313 12314 12315
    return ret;
}

static int
qemudNodeDeviceReAttach (virNodeDevicePtr dev)
{
12316
    struct qemud_driver *driver = dev->conn->privateData;
12317 12318 12319 12320 12321 12322 12323
    pciDevice *pci;
    unsigned domain, bus, slot, function;
    int ret = -1;

    if (qemudNodeDeviceGetPciInfo(dev, &domain, &bus, &slot, &function) < 0)
        return -1;

12324
    pci = pciGetDevice(domain, bus, slot, function);
12325 12326 12327
    if (!pci)
        return -1;

12328 12329
    qemuDriverLock(driver);
    if (pciReAttachDevice(pci, driver->activePciHostdevs) < 0)
12330 12331 12332 12333
        goto out;

    ret = 0;
out:
12334
    qemuDriverUnlock(driver);
12335
    pciFreeDevice(pci);
12336 12337 12338 12339 12340 12341
    return ret;
}

static int
qemudNodeDeviceReset (virNodeDevicePtr dev)
{
12342
    struct qemud_driver *driver = dev->conn->privateData;
12343 12344 12345 12346 12347 12348 12349
    pciDevice *pci;
    unsigned domain, bus, slot, function;
    int ret = -1;

    if (qemudNodeDeviceGetPciInfo(dev, &domain, &bus, &slot, &function) < 0)
        return -1;

12350
    pci = pciGetDevice(domain, bus, slot, function);
12351 12352 12353
    if (!pci)
        return -1;

12354 12355
    qemuDriverLock(driver);

12356
    if (pciResetDevice(pci, driver->activePciHostdevs, NULL) < 0)
12357 12358 12359 12360
        goto out;

    ret = 0;
out:
12361
    qemuDriverUnlock(driver);
12362
    pciFreeDevice(pci);
12363 12364 12365
    return ret;
}

12366 12367 12368 12369 12370 12371 12372 12373 12374 12375 12376
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) {
12377 12378
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        "%s", _("cannot get host CPU capabilities"));
12379 12380
    }
    else
12381
        ret = cpuCompareXML(driver->caps->host.cpu, xmlDesc);
12382 12383 12384 12385 12386 12387

    qemuDriverUnlock(driver);

    return ret;
}

12388

12389 12390 12391 12392 12393 12394 12395 12396 12397 12398 12399 12400 12401
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;
}

12402 12403 12404 12405 12406 12407 12408 12409 12410 12411 12412 12413 12414 12415 12416 12417 12418 12419 12420 12421 12422 12423 12424

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) {
12425 12426
            struct timeval now;

12427
            memcpy(info, &priv->jobInfo, sizeof(*info));
12428 12429 12430 12431 12432 12433 12434 12435 12436 12437 12438 12439 12440 12441

            /* 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;
12442 12443 12444 12445 12446 12447 12448 12449 12450 12451 12452 12453 12454 12455 12456 12457 12458 12459 12460
        } 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;
}


12461 12462 12463 12464 12465 12466 12467 12468 12469 12470 12471 12472 12473 12474 12475 12476 12477 12478 12479 12480 12481 12482
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);
12483
            priv->jobSignals |= QEMU_JOB_SIGNAL_CANCEL;
12484 12485 12486 12487 12488 12489 12490 12491 12492 12493 12494 12495 12496 12497 12498 12499 12500 12501 12502 12503
        } 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;
}


12504 12505 12506 12507 12508 12509 12510 12511 12512 12513
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;

12514
    virCheckFlags(0, -1);
12515 12516 12517 12518 12519 12520 12521 12522 12523 12524 12525 12526 12527 12528 12529 12530 12531 12532 12533 12534

    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;

12535
    if (priv->jobActive != QEMU_JOB_MIGRATION_OUT) {
12536 12537 12538 12539 12540 12541 12542 12543 12544 12545 12546 12547 12548 12549 12550 12551 12552
        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
12553 12554 12555 12556 12557 12558 12559 12560 12561 12562 12563 12564 12565 12566 12567 12568 12569 12570 12571 12572 12573 12574 12575 12576 12577 12578 12579 12580 12581 12582 12583 12584 12585 12586 12587 12588 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 12641 12642 12643 12644 12645 12646 12647 12648 12649 12650 12651 12652 12653 12654 12655 12656 12657 12658 12659 12660 12661
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,
12662 12663
                            _("Disk '%s' does not support snapshotting"),
                            vm->def->disks[i]->src);
C
Chris Lalancette 已提交
12664 12665 12666 12667 12668 12669 12670 12671 12672
            return 0;
        }
    }

    return 1;
}

static virDomainSnapshotPtr qemuDomainSnapshotCreateXML(virDomainPtr domain,
                                                        const char *xmlDesc,
12673
                                                        unsigned int flags)
C
Chris Lalancette 已提交
12674 12675 12676 12677 12678 12679 12680 12681 12682 12683
{
    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;

12684 12685
    virCheckFlags(0, NULL);

C
Chris Lalancette 已提交
12686 12687 12688 12689 12690 12691 12692 12693 12694 12695 12696 12697 12698 12699 12700 12701 12702 12703 12704 12705 12706 12707 12708 12709 12710 12711 12712 12713 12714 12715 12716 12717 12718 12719 12720 12721 12722 12723 12724 12725 12726 12727 12728 12729 12730 12731 12732 12733 12734 12735 12736 12737 12738 12739 12740 12741 12742 12743 12744 12745 12746
    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 {
12747 12748 12749 12750 12751
        qemuDomainObjPrivatePtr priv;
        int ret;

        if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
            goto cleanup;
C
Chris Lalancette 已提交
12752 12753
        priv = vm->privateData;
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
12754
        ret = qemuMonitorCreateSnapshot(priv->mon, def->name);
C
Chris Lalancette 已提交
12755
        qemuDomainObjExitMonitorWithDriver(driver, vm);
12756
        if (qemuDomainObjEndJob(vm) == 0) {
12757
            vm = NULL;
12758 12759
            goto cleanup;
        }
12760 12761
        if (ret < 0)
            goto cleanup;
C
Chris Lalancette 已提交
12762 12763 12764 12765 12766 12767 12768 12769 12770 12771 12772 12773 12774 12775 12776 12777 12778 12779 12780 12781 12782 12783 12784 12785 12786 12787 12788 12789 12790 12791 12792 12793 12794 12795 12796 12797
    }

    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,
12798
                                       unsigned int flags)
C
Chris Lalancette 已提交
12799 12800 12801 12802 12803
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm = NULL;
    int n = -1;

12804 12805
    virCheckFlags(0, -1);

C
Chris Lalancette 已提交
12806 12807 12808 12809 12810 12811 12812 12813 12814 12815 12816 12817 12818 12819 12820 12821 12822 12823 12824 12825
    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,
12826
                                 unsigned int flags)
C
Chris Lalancette 已提交
12827 12828 12829 12830 12831
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm = NULL;
    int n = -1;

12832 12833
    virCheckFlags(0, -1);

C
Chris Lalancette 已提交
12834 12835 12836 12837 12838 12839 12840 12841 12842 12843 12844 12845 12846 12847 12848 12849 12850 12851 12852 12853 12854
    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,
12855
                                                           unsigned int flags)
C
Chris Lalancette 已提交
12856 12857 12858 12859 12860 12861
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm;
    virDomainSnapshotObjPtr snap = NULL;
    virDomainSnapshotPtr snapshot = NULL;

12862 12863
    virCheckFlags(0, NULL);

C
Chris Lalancette 已提交
12864 12865 12866 12867 12868 12869 12870 12871 12872 12873 12874 12875 12876 12877 12878 12879 12880 12881 12882 12883 12884 12885 12886 12887 12888 12889 12890
    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,
12891
                                        unsigned int flags)
C
Chris Lalancette 已提交
12892 12893 12894 12895 12896
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;

12897 12898
    virCheckFlags(0, -1);

C
Chris Lalancette 已提交
12899 12900 12901 12902 12903 12904 12905 12906 12907 12908 12909 12910 12911 12912 12913 12914 12915 12916 12917 12918
    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,
12919
                                                      unsigned int flags)
C
Chris Lalancette 已提交
12920 12921 12922 12923 12924
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm;
    virDomainSnapshotPtr snapshot = NULL;

12925 12926
    virCheckFlags(0, NULL);

C
Chris Lalancette 已提交
12927 12928 12929 12930 12931 12932 12933 12934 12935 12936 12937 12938 12939 12940 12941 12942 12943 12944 12945 12946 12947 12948 12949 12950 12951 12952
    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,
12953
                                       unsigned int flags)
C
Chris Lalancette 已提交
12954 12955 12956 12957 12958 12959 12960
{
    struct qemud_driver *driver = snapshot->domain->conn->privateData;
    virDomainObjPtr vm = NULL;
    char *xml = NULL;
    virDomainSnapshotObjPtr snap = NULL;
    char uuidstr[VIR_UUID_STRING_BUFLEN];

12961 12962
    virCheckFlags(0, NULL);

C
Chris Lalancette 已提交
12963 12964 12965 12966 12967 12968 12969 12970 12971 12972 12973 12974 12975 12976 12977 12978 12979 12980 12981 12982 12983 12984 12985 12986 12987 12988 12989
    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,
12990
                                      unsigned int flags)
C
Chris Lalancette 已提交
12991 12992 12993 12994 12995 12996 12997 12998 12999 13000
{
    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;

13001 13002
    virCheckFlags(0, -1);

C
Chris Lalancette 已提交
13003 13004 13005 13006 13007 13008 13009 13010 13011 13012 13013 13014 13015 13016 13017 13018 13019 13020 13021 13022 13023 13024 13025 13026 13027 13028 13029 13030 13031 13032 13033
    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)
13034
                goto endjob;
C
Chris Lalancette 已提交
13035 13036 13037
        }
        else {
            if (qemuDomainSnapshotSetActive(vm, driver->snapshotDir) < 0)
13038
                goto endjob;
C
Chris Lalancette 已提交
13039 13040

            rc = qemudStartVMDaemon(snapshot->domain->conn, driver, vm, NULL,
13041
                                    false, -1, NULL);
13042
            qemuDomainStartAudit(vm, "from-snapshot", rc >= 0);
C
Chris Lalancette 已提交
13043
            if (qemuDomainSnapshotSetInactive(vm, driver->snapshotDir) < 0)
13044
                goto endjob;
C
Chris Lalancette 已提交
13045
            if (rc < 0)
13046
                goto endjob;
C
Chris Lalancette 已提交
13047 13048 13049 13050 13051 13052
        }

        if (snap->def->state == VIR_DOMAIN_PAUSED) {
            /* qemu unconditionally starts the domain running again after
             * loadvm, so let's pause it to keep consistency
             */
13053
            int state = vm->state;
C
Chris Lalancette 已提交
13054
            priv = vm->privateData;
13055
            vm->state = VIR_DOMAIN_PAUSED;
C
Chris Lalancette 已提交
13056 13057 13058
            qemuDomainObjEnterMonitorWithDriver(driver, vm);
            rc = qemuMonitorStopCPUs(priv->mon);
            qemuDomainObjExitMonitorWithDriver(driver, vm);
13059 13060
            if (rc < 0) {
                vm->state = state;
13061
                goto endjob;
13062
            }
C
Chris Lalancette 已提交
13063 13064 13065 13066 13067 13068 13069 13070 13071 13072 13073 13074 13075 13076 13077 13078 13079 13080
        }

        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)) {
13081
            qemudShutdownVMDaemon(driver, vm, 0);
13082
            qemuDomainStopAudit(vm, "from-snapshot");
C
Chris Lalancette 已提交
13083 13084 13085
            event = virDomainEventNewFromObj(vm,
                                             VIR_DOMAIN_EVENT_STOPPED,
                                             VIR_DOMAIN_EVENT_STOPPED_FROM_SNAPSHOT);
13086 13087 13088 13089
            if (!vm->persistent) {
                if (qemuDomainObjEndJob(vm) > 0)
                    virDomainRemoveInactive(&driver->domains, vm);
                vm = NULL;
13090
                goto cleanup;
13091
            }
C
Chris Lalancette 已提交
13092 13093 13094
        }

        if (qemuDomainSnapshotSetActive(vm, driver->snapshotDir) < 0)
13095
            goto endjob;
C
Chris Lalancette 已提交
13096 13097 13098 13099 13100 13101
    }

    vm->state = snap->def->state;

    ret = 0;

13102
endjob:
C
Chris Lalancette 已提交
13103 13104 13105
    if (vm && qemuDomainObjEndJob(vm) == 0)
        vm = NULL;

13106
cleanup:
C
Chris Lalancette 已提交
13107 13108 13109 13110 13111 13112 13113 13114 13115 13116 13117 13118 13119 13120 13121 13122 13123 13124 13125 13126 13127 13128 13129 13130 13131 13132 13133 13134 13135 13136 13137 13138 13139 13140 13141 13142 13143 13144 13145 13146 13147 13148 13149 13150 13151 13152 13153 13154 13155 13156 13157 13158 13159 13160 13161 13162 13163 13164 13165 13166 13167 13168 13169 13170 13171 13172 13173 13174 13175 13176 13177 13178 13179 13180 13181 13182 13183 13184 13185 13186 13187 13188 13189 13190 13191 13192 13193 13194 13195 13196 13197 13198 13199 13200 13201 13202 13203 13204 13205 13206 13207 13208 13209 13210 13211 13212 13213 13214 13215 13216 13217 13218 13219 13220 13221 13222 13223 13224 13225 13226 13227 13228 13229 13230 13231 13232 13233 13234 13235 13236 13237 13238 13239 13240
    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;

13241 13242
    virCheckFlags(VIR_DOMAIN_SNAPSHOT_DELETE_CHILDREN, -1);

C
Chris Lalancette 已提交
13243 13244 13245 13246 13247 13248 13249 13250 13251 13252 13253 13254 13255 13256 13257 13258 13259
    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;
    }

13260 13261 13262
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

C
Chris Lalancette 已提交
13263 13264 13265 13266 13267 13268 13269 13270
    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)
13271
            goto endjob;
C
Chris Lalancette 已提交
13272 13273 13274 13275
    }

    ret = qemuDomainSnapshotDiscard(driver, vm, snap);

13276 13277 13278 13279
endjob:
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;

C
Chris Lalancette 已提交
13280 13281 13282 13283 13284 13285
cleanup:
    if (vm)
        virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);
    return ret;
}
13286

13287 13288 13289 13290 13291 13292 13293 13294 13295 13296 13297 13298 13299 13300 13301 13302 13303 13304 13305 13306 13307 13308 13309 13310 13311 13312 13313 13314 13315 13316 13317 13318 13319 13320 13321 13322 13323 13324 13325 13326 13327 13328 13329 13330 13331 13332 13333 13334 13335 13336 13337
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;
}

13338 13339 13340 13341 13342
static virDriver qemuDriver = {
    VIR_DRV_QEMU,
    "QEMU",
    qemudOpen, /* open */
    qemudClose, /* close */
D
Daniel Veillard 已提交
13343
    qemudSupportsFeature, /* supports_feature */
13344 13345
    qemudGetType, /* type */
    qemudGetVersion, /* version */
13346
    NULL, /* libvirtVersion (impl. in libvirt.c) */
13347
    virGetHostname, /* getHostname */
13348
    qemudGetMaxVCPUs, /* getMaxVcpus */
13349
    nodeGetInfo, /* nodeGetInfo */
13350 13351 13352
    qemudGetCapabilities, /* getCapabilities */
    qemudListDomains, /* listDomains */
    qemudNumDomains, /* numOfDomains */
13353
    qemudDomainCreate, /* domainCreateXML */
13354 13355 13356 13357 13358
    qemudDomainLookupByID, /* domainLookupByID */
    qemudDomainLookupByUUID, /* domainLookupByUUID */
    qemudDomainLookupByName, /* domainLookupByName */
    qemudDomainSuspend, /* domainSuspend */
    qemudDomainResume, /* domainResume */
13359
    qemudDomainShutdown, /* domainShutdown */
13360 13361 13362
    NULL, /* domainReboot */
    qemudDomainDestroy, /* domainDestroy */
    qemudDomainGetOSType, /* domainGetOSType */
13363
    qemudDomainGetMaxMemory, /* domainGetMaxMemory */
13364
    NULL, /* domainSetMaxMemory */
13365
    qemudDomainSetMemory, /* domainSetMemory */
13366 13367 13368
    qemudDomainGetInfo, /* domainGetInfo */
    qemudDomainSave, /* domainSave */
    qemudDomainRestore, /* domainRestore */
P
Paolo Bonzini 已提交
13369
    qemudDomainCoreDump, /* domainCoreDump */
13370
    qemudDomainSetVcpus, /* domainSetVcpus */
13371 13372
    qemudDomainSetVcpusFlags, /* domainSetVcpusFlags */
    qemudDomainGetVcpusFlags, /* domainGetVcpusFlags */
13373 13374
    qemudDomainPinVcpu, /* domainPinVcpu */
    qemudDomainGetVcpus, /* domainGetVcpus */
13375
    qemudDomainGetMaxVcpus, /* domainGetMaxVcpus */
13376 13377
    qemudDomainGetSecurityLabel, /* domainGetSecurityLabel */
    qemudNodeGetSecurityModel, /* nodeGetSecurityModel */
13378
    qemudDomainDumpXML, /* domainDumpXML */
13379
    qemuDomainXMLFromNative, /* domainXmlFromNative */
13380
    qemuDomainXMLToNative, /* domainXMLToNative */
13381 13382
    qemudListDefinedDomains, /* listDefinedDomains */
    qemudNumDefinedDomains, /* numOfDefinedDomains */
13383
    qemudDomainStart, /* domainCreate */
13384
    qemudDomainStartWithFlags, /* domainCreateWithFlags */
13385 13386
    qemudDomainDefine, /* domainDefineXML */
    qemudDomainUndefine, /* domainUndefine */
13387
    qemudDomainAttachDevice, /* domainAttachDevice */
13388
    qemudDomainAttachDeviceFlags, /* domainAttachDeviceFlags */
13389
    qemudDomainDetachDevice, /* domainDetachDevice */
13390
    qemudDomainDetachDeviceFlags, /* domainDetachDeviceFlags */
13391
    qemuDomainUpdateDeviceFlags, /* domainUpdateDeviceFlags */
13392 13393
    qemudDomainGetAutostart, /* domainGetAutostart */
    qemudDomainSetAutostart, /* domainSetAutostart */
13394 13395 13396
    qemuGetSchedulerType, /* domainGetSchedulerType */
    qemuGetSchedulerParameters, /* domainGetSchedulerParameters */
    qemuSetSchedulerParameters, /* domainSetSchedulerParameters */
D
Daniel Veillard 已提交
13397 13398
    NULL, /* domainMigratePrepare (v1) */
    qemudDomainMigratePerform, /* domainMigratePerform */
13399
    NULL, /* domainMigrateFinish */
13400
    qemudDomainBlockStats, /* domainBlockStats */
13401
    qemudDomainInterfaceStats, /* domainInterfaceStats */
13402
    qemudDomainMemoryStats, /* domainMemoryStats */
13403
    qemudDomainBlockPeek, /* domainBlockPeek */
R
Richard W.M. Jones 已提交
13404
    qemudDomainMemoryPeek, /* domainMemoryPeek */
13405
    qemuDomainGetBlockInfo, /* domainGetBlockInfo */
13406 13407
    nodeGetCellsFreeMemory, /* nodeGetCellsFreeMemory */
    nodeGetFreeMemory,  /* getFreeMemory */
13408 13409
    qemuDomainEventRegister, /* domainEventRegister */
    qemuDomainEventDeregister, /* domainEventDeregister */
D
Daniel Veillard 已提交
13410 13411
    qemudDomainMigratePrepare2, /* domainMigratePrepare2 */
    qemudDomainMigrateFinish2, /* domainMigrateFinish2 */
13412 13413 13414
    qemudNodeDeviceDettach, /* nodeDeviceDettach */
    qemudNodeDeviceReAttach, /* nodeDeviceReAttach */
    qemudNodeDeviceReset, /* nodeDeviceReset */
C
Chris Lalancette 已提交
13415
    qemudDomainMigratePrepareTunnel, /* domainMigratePrepareTunnel */
13416 13417 13418 13419
    qemuIsEncrypted, /* isEncrypted */
    qemuIsSecure, /* isSecure */
    qemuDomainIsActive, /* domainIsActive */
    qemuDomainIsPersistent, /* domainIsPersistent */
13420
    qemuCPUCompare, /* cpuCompare */
13421
    qemuCPUBaseline, /* cpuBaseline */
13422
    qemuDomainGetJobInfo, /* domainGetJobInfo */
13423
    qemuDomainAbortJob, /* domainAbortJob */
13424
    qemuDomainMigrateSetMaxDowntime, /* domainMigrateSetMaxDowntime */
13425 13426
    qemuDomainEventRegisterAny, /* domainEventRegisterAny */
    qemuDomainEventDeregisterAny, /* domainEventDeregisterAny */
13427 13428 13429
    qemuDomainManagedSave, /* domainManagedSave */
    qemuDomainHasManagedSaveImage, /* domainHasManagedSaveImage */
    qemuDomainManagedSaveRemove, /* domainManagedSaveRemove */
C
Chris Lalancette 已提交
13430 13431 13432 13433 13434 13435 13436 13437 13438
    qemuDomainSnapshotCreateXML, /* domainSnapshotCreateXML */
    qemuDomainSnapshotDumpXML, /* domainSnapshotDumpXML */
    qemuDomainSnapshotNum, /* domainSnapshotNum */
    qemuDomainSnapshotListNames, /* domainSnapshotListNames */
    qemuDomainSnapshotLookupByName, /* domainSnapshotLookupByName */
    qemuDomainHasCurrentSnapshot, /* domainHasCurrentSnapshot */
    qemuDomainSnapshotCurrent, /* domainSnapshotCurrent */
    qemuDomainRevertToSnapshot, /* domainRevertToSnapshot */
    qemuDomainSnapshotDelete, /* domainSnapshotDelete */
13439
    qemuDomainMonitorCommand, /* qemuDomainMonitorCommand */
13440
    qemuDomainSetMemoryParameters, /* domainSetMemoryParameters */
13441
    qemuDomainGetMemoryParameters, /* domainGetMemoryParameters */
13442 13443 13444
};


13445
static virStateDriver qemuStateDriver = {
13446
    .name = "QEMU",
13447 13448 13449 13450
    .initialize = qemudStartup,
    .cleanup = qemudShutdown,
    .reload = qemudReload,
    .active = qemudActive,
13451
};
13452

S
Stefan Berger 已提交
13453
static int
13454
qemudVMFilterRebuild(virConnectPtr conn ATTRIBUTE_UNUSED,
S
Stefan Berger 已提交
13455 13456 13457
                     virHashIterator iter, void *data)
{
    virHashForEach(qemu_driver->domains.objs, iter, data);
13458

S
Stefan Berger 已提交
13459 13460 13461
    return 0;
}

13462 13463 13464 13465 13466 13467 13468 13469 13470 13471 13472 13473 13474 13475 13476 13477 13478 13479 13480 13481 13482 13483
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
13484

13485 13486 13487 13488 13489 13490 13491 13492 13493 13494 13495 13496 13497

static void
qemudVMDriverLock(void) {
    qemuDriverLock(qemu_driver);
};


static void
qemudVMDriverUnlock(void) {
    qemuDriverUnlock(qemu_driver);
};


S
Stefan Berger 已提交
13498 13499 13500
static virNWFilterCallbackDriver qemuCallbackDriver = {
    .name = "QEMU",
    .vmFilterRebuild = qemudVMFilterRebuild,
13501 13502
    .vmDriverLock = qemudVMDriverLock,
    .vmDriverUnlock = qemudVMDriverUnlock,
S
Stefan Berger 已提交
13503 13504
};

13505
int qemuRegister(void) {
13506 13507
    virRegisterDriver(&qemuDriver);
    virRegisterStateDriver(&qemuStateDriver);
S
Stefan Berger 已提交
13508
    virNWFilterRegisterCallbackDriver(&qemuCallbackDriver);
13509 13510
    return 0;
}