qemu_driver.c 364.4 KB
Newer Older
D
Daniel P. Berrange 已提交
1 2 3
/*
 * driver.c: core driver methods for managing qemu guests
 *
4
 * Copyright (C) 2006, 2007, 2008, 2009, 2010 Red Hat, Inc.
D
Daniel P. Berrange 已提交
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
 * Copyright (C) 2006 Daniel P. Berrange
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with this library; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307  USA
 *
 * Author: Daniel P. Berrange <berrange@redhat.com>
 */

24
#include <config.h>
25

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

49
#ifdef __linux__
50 51 52 53
# include <sys/vfs.h>
# ifndef NFS_SUPER_MAGIC
#  define NFS_SUPER_MAGIC 0x6969
# endif /* NFS_SUPER_MAGIC */
54
#endif /* __linux__ */
55

56
#include "virterror_internal.h"
57
#include "logging.h"
58
#include "datatypes.h"
59 60
#include "qemu_driver.h"
#include "qemu_conf.h"
61
#include "qemu_monitor.h"
62
#include "qemu_bridge_filter.h"
J
Jim Meyering 已提交
63
#include "c-ctype.h"
64
#include "event.h"
65
#include "buf.h"
66
#include "util.h"
67
#include "nodeinfo.h"
68
#include "stats_linux.h"
69
#include "capabilities.h"
70
#include "memory.h"
71
#include "uuid.h"
72
#include "domain_conf.h"
73 74
#include "node_device_conf.h"
#include "pci.h"
75
#include "hostusb.h"
76
#include "processinfo.h"
77 78
#include "qemu_security_stacked.h"
#include "qemu_security_dac.h"
79
#include "cgroup.h"
C
Chris Lalancette 已提交
80
#include "libvirt_internal.h"
81
#include "xml.h"
82
#include "cpu/cpu.h"
83
#include "macvtap.h"
S
Stefan Berger 已提交
84
#include "nwfilter/nwfilter_gentech_driver.h"
85
#include "hooks.h"
86
#include "storage_file.h"
87

88

89 90
#define VIR_FROM_THIS VIR_FROM_QEMU

91 92 93 94 95 96 97 98 99 100 101 102
/* 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,
    QEMU_JOB_MIGRATION,
};

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 */
103 104 105 106 107
    QEMU_JOB_SIGNAL_MIGRATE_DOWNTIME = 1 << 2, /* Request migration downtime change */
};

struct qemuDomainJobSignalsData {
    unsigned long long migrateDowntime; /* Data for QEMU_JOB_SIGNAL_MIGRATE_DOWNTIME */
108 109
};

110 111 112
typedef struct _qemuDomainObjPrivate qemuDomainObjPrivate;
typedef qemuDomainObjPrivate *qemuDomainObjPrivatePtr;
struct _qemuDomainObjPrivate {
113
    virCond jobCond; /* Use in conjunction with main virDomainObjPtr lock */
114 115
    enum qemuDomainJob jobActive;   /* Currently running job */
    unsigned int jobSignals;        /* Signals for running job */
116
    struct qemuDomainJobSignalsData jobSignalsData; /* Signal specific data */
117 118
    virDomainJobInfo jobInfo;
    unsigned long long jobStart;
119

120
    qemuMonitorPtr mon;
121
    virDomainChrDefPtr monConfig;
D
Daniel P. Berrange 已提交
122
    int monJSON;
123 124 125

    int nvcpupids;
    int *vcpupids;
126 127

    qemuDomainPCIAddressSetPtr pciaddrs;
128
    int persistentAddrs;
129 130
};

131 132
static int qemudShutdown(void);

133 134
static void qemuDriverLock(struct qemud_driver *driver)
{
135
    virMutexLock(&driver->lock);
136 137 138
}
static void qemuDriverUnlock(struct qemud_driver *driver)
{
139
    virMutexUnlock(&driver->lock);
140 141
}

142 143 144
static void qemuDomainEventFlush(int timer, void *opaque);
static void qemuDomainEventQueue(struct qemud_driver *driver,
                                 virDomainEventPtr event);
145

146 147
static int qemudStartVMDaemon(virConnectPtr conn,
                              struct qemud_driver *driver,
148
                              virDomainObjPtr vm,
149 150
                              const char *migrateFrom,
                              int stdin_fd);
151

152
static void qemudShutdownVMDaemon(struct qemud_driver *driver,
153 154
                                  virDomainObjPtr vm,
                                  int migrated);
155

156
static int qemudDomainGetMaxVcpus(virDomainPtr dom);
157

158
static int qemuDetectVcpuPIDs(struct qemud_driver *driver,
159
                              virDomainObjPtr vm);
160

161 162 163
static int qemuUpdateActivePciHostdevs(struct qemud_driver *driver,
                                       virDomainDefPtr def);

J
Jim Meyering 已提交
164
static struct qemud_driver *qemu_driver = NULL;
165

166 167 168 169 170 171 172 173 174 175 176 177 178 179 180

static void *qemuDomainObjPrivateAlloc(void)
{
    qemuDomainObjPrivatePtr priv;

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

    return priv;
}

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

181
    qemuDomainPCIAddressSetFree(priv->pciaddrs);
182 183 184
    virDomainChrDefFree(priv->monConfig);
    VIR_FREE(priv->vcpupids);

185 186 187 188 189 190 191 192 193
    /* This should never be non-NULL if we get here, but just in case... */
    if (priv->mon) {
        VIR_ERROR0("Unexpected QEMU monitor still active during domain deletion");
        qemuMonitorClose(priv->mon);
    }
    VIR_FREE(priv);
}


194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211
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
212 213
        if (priv->monJSON)
            virBufferAddLit(buf, " json='1'");
214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239
        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) {
240
        virReportOOMError();
241 242 243
        goto error;
    }

244
    if (!(priv->monConfig->info.alias = strdup("monitor"))) {
245
        virReportOOMError();
246 247 248
        goto error;
    }

249
    if (!(monitorpath =
250
          virXPathString("string(./monitor[1]/@path)", ctxt))) {
251 252
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("no monitor path"));
253 254 255
        goto error;
    }

256
    tmp = virXPathString("string(./monitor[1]/@type)", ctxt);
257 258 259 260 261 262
    if (tmp)
        priv->monConfig->type = virDomainChrTypeFromString(tmp);
    else
        priv->monConfig->type = VIR_DOMAIN_CHR_TYPE_PTY;
    VIR_FREE(tmp);

263
    if (virXPathBoolean("count(./monitor[@json = '1']) > 0", ctxt)) {
D
Daniel P. Berrange 已提交
264
        priv->monJSON = 1;
265 266 267
    } else {
        priv->monJSON = 0;
    }
D
Daniel P. Berrange 已提交
268

269 270 271 272 273 274 275 276 277
    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);
278 279 280
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("unsupported monitor type '%s'"),
                        virDomainChrTypeToString(priv->monConfig->type));
281 282 283
        goto error;
    }

284
    n = virXPathNodeSet("./vcpus/vcpu", ctxt, &nodes);
285 286 287 288 289
    if (n < 0)
        goto error;
    if (n) {
        priv->nvcpupids = n;
        if (VIR_REALLOC_N(priv->vcpupids, priv->nvcpupids) < 0) {
290
            virReportOOMError();
291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310
            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:
311 312
    virDomainChrDefFree(priv->monConfig);
    priv->monConfig = NULL;
313 314 315 316 317 318
    VIR_FREE(nodes);
    return -1;
}



319 320 321 322 323 324 325 326 327
/*
 * 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
 */
328 329 330 331

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

332 333 334 335
static int qemuDomainObjBeginJob(virDomainObjPtr obj) ATTRIBUTE_RETURN_CHECK;
static int qemuDomainObjBeginJob(virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
336 337 338 339
    struct timeval now;
    unsigned long long then;

    if (gettimeofday(&now, NULL) < 0) {
340
        virReportSystemError(errno, "%s",
341 342 343 344 345
                             _("cannot get time of day"));
        return -1;
    }
    then = (now.tv_sec * 1000ull) + (now.tv_usec / 1000);
    then += QEMU_JOB_WAIT_TIME;
346 347 348 349

    virDomainObjRef(obj);

    while (priv->jobActive) {
350
        if (virCondWaitUntil(&priv->jobCond, &obj->lock, then) < 0) {
351
            virDomainObjUnref(obj);
352
            if (errno == ETIMEDOUT)
353 354
                qemuReportError(VIR_ERR_OPERATION_TIMEOUT,
                                "%s", _("cannot acquire state change lock"));
355
            else
356
                virReportSystemError(errno,
357
                                     "%s", _("cannot acquire job mutex"));
358 359 360
            return -1;
        }
    }
361 362
    priv->jobActive = QEMU_JOB_UNSPECIFIED;
    priv->jobSignals = 0;
363
    memset(&priv->jobSignalsData, 0, sizeof(priv->jobSignalsData));
364 365
    priv->jobStart = (now.tv_sec * 1000ull) + (now.tv_usec / 1000);
    memset(&priv->jobInfo, 0, sizeof(priv->jobInfo));
366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381

    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;
382 383 384 385
    struct timeval now;
    unsigned long long then;

    if (gettimeofday(&now, NULL) < 0) {
386
        virReportSystemError(errno, "%s",
387 388 389 390 391
                             _("cannot get time of day"));
        return -1;
    }
    then = (now.tv_sec * 1000ull) + (now.tv_usec / 1000);
    then += QEMU_JOB_WAIT_TIME;
392 393 394 395 396

    virDomainObjRef(obj);
    qemuDriverUnlock(driver);

    while (priv->jobActive) {
397
        if (virCondWaitUntil(&priv->jobCond, &obj->lock, then) < 0) {
398
            virDomainObjUnref(obj);
399
            if (errno == ETIMEDOUT)
400 401
                qemuReportError(VIR_ERR_OPERATION_TIMEOUT,
                                "%s", _("cannot acquire state change lock"));
402
            else
403
                virReportSystemError(errno,
404
                                     "%s", _("cannot acquire job mutex"));
M
Matthias Bolte 已提交
405
            qemuDriverLock(driver);
406 407 408
            return -1;
        }
    }
409 410
    priv->jobActive = QEMU_JOB_UNSPECIFIED;
    priv->jobSignals = 0;
411
    memset(&priv->jobSignalsData, 0, sizeof(priv->jobSignalsData));
412 413
    priv->jobStart = (now.tv_sec * 1000ull) + (now.tv_usec / 1000);
    memset(&priv->jobInfo, 0, sizeof(priv->jobInfo));
414 415 416 417 418 419 420 421 422 423 424 425 426

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

    return 0;
}

/*
 * obj must be locked before calling, qemud_driver does not matter
 *
 * To be called after completing the work associated with the
 * earlier  qemuDomainBeginJob() call
427 428 429
 *
 * Returns remaining refcount on 'obj', maybe 0 to indicated it
 * was deleted
430
 */
431
static int ATTRIBUTE_RETURN_CHECK qemuDomainObjEndJob(virDomainObjPtr obj)
432 433 434
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

435 436
    priv->jobActive = QEMU_JOB_NONE;
    priv->jobSignals = 0;
437
    memset(&priv->jobSignalsData, 0, sizeof(priv->jobSignalsData));
438 439
    priv->jobStart = 0;
    memset(&priv->jobInfo, 0, sizeof(priv->jobInfo));
440 441
    virCondSignal(&priv->jobCond);

442
    return virDomainObjUnref(obj);
443 444 445 446 447 448 449 450 451 452 453
}


/*
 * obj must be locked before calling, qemud_driver must be unlocked
 *
 * To be called immediately before any QEMU monitor API call
 * Must have alrady called qemuDomainObjBeginJob().
 *
 * To be followed with qemuDomainObjExitMonitor() once complete
 */
454 455 456 457 458
static void qemuDomainObjEnterMonitor(virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    qemuMonitorLock(priv->mon);
459
    qemuMonitorRef(priv->mon);
460
    virDomainObjUnlock(obj);
461 462 463
}


464 465 466 467
/* obj must NOT be locked before calling, qemud_driver must be unlocked
 *
 * Should be paired with an earlier  qemuDomainObjEnterMonitor() call
 */
468 469 470
static void qemuDomainObjExitMonitor(virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
471 472 473 474 475 476
    int refs;

    refs = qemuMonitorUnref(priv->mon);

    if (refs > 0)
        qemuMonitorUnlock(priv->mon);
477

478
    virDomainObjLock(obj);
479 480 481 482 483

    if (refs == 0) {
        virDomainObjUnref(obj);
        priv->mon = NULL;
    }
484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499
}


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

    qemuMonitorLock(priv->mon);
500
    qemuMonitorRef(priv->mon);
501 502 503 504 505 506 507 508 509 510 511 512 513
    virDomainObjUnlock(obj);
    qemuDriverUnlock(driver);
}


/* obj must NOT be locked before calling, qemud_driver must be unlocked,
 * and will be locked after returning
 *
 * Should be paired with an earlier  qemuDomainObjEnterMonitor() call
 */
static void qemuDomainObjExitMonitorWithDriver(struct qemud_driver *driver, virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
514 515 516 517 518 519
    int refs;

    refs = qemuMonitorUnref(priv->mon);

    if (refs > 0)
        qemuMonitorUnlock(priv->mon);
520 521 522

    qemuDriverLock(driver);
    virDomainObjLock(obj);
523 524 525 526 527

    if (refs == 0) {
        virDomainObjUnref(obj);
        priv->mon = NULL;
    }
528 529 530
}


531 532 533 534 535 536 537 538 539
static int qemuCgroupControllerActive(struct qemud_driver *driver,
                                      int controller)
{
    if (driver->cgroup == NULL)
        return 0;
    if (driver->cgroupControllers & (1 << controller))
        return 1;
    return 0;
}
540

541
static int
542
qemudLogFD(struct qemud_driver *driver, const char* name)
543 544 545
{
    char logfile[PATH_MAX];
    mode_t logmode;
G
Guido Günther 已提交
546
    int ret, fd = -1;
547

548 549
    if ((ret = snprintf(logfile, sizeof(logfile), "%s/%s.log",
                        driver->logDir, name))
G
Guido Günther 已提交
550
        < 0 || ret >= sizeof(logfile)) {
551
        virReportOOMError();
552 553 554 555
        return -1;
    }

    logmode = O_CREAT | O_WRONLY;
556 557
    /* Only logrotate files in /var/log, so only append if running privileged */
    if (driver->privileged)
558
        logmode |= O_APPEND;
559 560 561
    else
        logmode |= O_TRUNC;

562
    if ((fd = open(logfile, logmode, S_IRUSR | S_IWUSR)) < 0) {
563
        virReportSystemError(errno,
564 565
                             _("failed to create logfile %s"),
                             logfile);
566 567
        return -1;
    }
568
    if (virSetCloseExec(fd) < 0) {
569
        virReportSystemError(errno, "%s",
570
                             _("Unable to set VM logfile close-on-exec flag"));
571 572 573 574 575 576 577
        close(fd);
        return -1;
    }
    return fd;
}


578
static int
579
qemudLogReadFD(const char* logDir, const char* name, off_t pos)
580 581 582 583 584 585 586
{
    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)) {
587 588 589
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("failed to build logfile name %s/%s.log"),
                        logDir, name);
590 591 592 593 594
        return -1;
    }


    if ((fd = open(logfile, logmode)) < 0) {
595
        virReportSystemError(errno,
596 597
                             _("failed to create logfile %s"),
                             logfile);
598 599
        return -1;
    }
600
    if (virSetCloseExec(fd) < 0) {
601
        virReportSystemError(errno, "%s",
602
                             _("Unable to set VM logfile close-on-exec flag"));
603 604 605
        close(fd);
        return -1;
    }
606
    if (pos < 0 || lseek(fd, pos, SEEK_SET) < 0) {
607
      virReportSystemError(pos < 0 ? 0 : errno,
608 609
                             _("Unable to seek to %lld in %s"),
                             (long long) pos, logfile);
610 611 612 613 614 615
        close(fd);
    }
    return fd;
}


616 617 618 619 620 621 622 623 624 625 626 627
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;

    virDomainObjLock(vm);
    if (vm->autostart &&
D
Daniel P. Berrange 已提交
628
        !virDomainObjIsActive(vm)) {
629 630 631 632 633 634
        int ret;

        virResetLastError();
        ret = qemudStartVMDaemon(data->conn, data->driver, vm, NULL, -1);
        if (ret < 0) {
            virErrorPtr err = virGetLastError();
635
            VIR_ERROR(_("Failed to autostart VM '%s': %s"),
636 637 638 639 640 641 642 643 644 645 646 647 648 649
                      vm->def->name,
                      err ? err->message : "");
        } else {
            virDomainEventPtr event =
                virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STARTED,
                                         VIR_DOMAIN_EVENT_STARTED_BOOTED);
            if (event)
                qemuDomainEventQueue(data->driver, event);
        }
    }
    virDomainObjUnlock(vm);
}

650 651
static void
qemudAutostartConfigs(struct qemud_driver *driver) {
652 653 654 655 656
    /* 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
     */
657 658 659
    virConnectPtr conn = virConnectOpen(driver->privileged ?
                                        "qemu:///system" :
                                        "qemu:///session");
660
    /* Ignoring NULL conn which is mostly harmless here */
661
    struct qemuAutostartData data = { driver, conn };
662

663
    qemuDriverLock(driver);
664
    virHashForEach(driver->domains.objs, qemuAutostartDomain, &data);
665
    qemuDriverUnlock(driver);
666

667 668
    if (conn)
        virConnectClose(conn);
669 670
}

671 672 673 674 675 676 677 678 679

/**
 * qemudRemoveDomainStatus
 *
 * remove all state files of a domain from statedir
 *
 * Returns 0 on success
 */
static int
680
qemudRemoveDomainStatus(struct qemud_driver *driver,
681 682
                        virDomainObjPtr vm)
{
683
    char ebuf[1024];
684 685 686
    char *file = NULL;

    if (virAsprintf(&file, "%s/%s.xml", driver->stateDir, vm->def->name) < 0) {
687
        virReportOOMError();
D
Daniel Veillard 已提交
688
        return(-1);
689 690
    }

691
    if (unlink(file) < 0 && errno != ENOENT && errno != ENOTDIR)
692
        VIR_WARN("Failed to remove domain XML for %s: %s",
D
Daniel Veillard 已提交
693 694 695
                 vm->def->name, virStrerror(errno, ebuf, sizeof(ebuf)));
    VIR_FREE(file);

696
    if (virFileDeletePid(driver->stateDir, vm->def->name) != 0)
697
        VIR_WARN("Failed to remove PID file for %s: %s",
698
                 vm->def->name, virStrerror(errno, ebuf, sizeof(ebuf)));
699

D
Daniel Veillard 已提交
700

701
    return 0;
702 703
}

704 705 706 707 708 709 710 711 712 713 714 715 716 717

/*
 * 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;

718
    VIR_DEBUG("Received EOF on %p '%s'", vm, vm->def->name);
719 720 721 722 723 724 725 726
    virDomainObjLock(vm);

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

727
    qemudShutdownVMDaemon(driver, vm, 0);
728 729 730 731 732 733 734 735 736 737 738 739 740
    if (!vm->persistent)
        virDomainRemoveInactive(&driver->domains, vm);
    else
        virDomainObjUnlock(vm);

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


741 742 743
static virDomainDiskDefPtr
findDomainDiskByPath(virDomainObjPtr vm,
                     const char *path)
744 745 746 747 748 749 750
{
    int i;

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

        disk = vm->def->disks[i];
751 752
        if (disk->src != NULL && STREQ(disk->src, path))
            return disk;
753
    }
754 755 756 757

    qemuReportError(VIR_ERR_INTERNAL_ERROR,
                    _("no disk found with path %s"),
                    path);
758 759 760
    return NULL;
}

761 762 763 764 765 766
static virDomainDiskDefPtr
findDomainDiskByAlias(virDomainObjPtr vm,
                      const char *alias)
{
    int i;

767 768 769
    if (STRPREFIX(alias, QEMU_DRIVE_HOST_PREFIX))
        alias += strlen(QEMU_DRIVE_HOST_PREFIX);

770 771 772 773 774 775 776 777 778 779 780 781 782 783
    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;
}

784
static int
785 786 787 788
getVolumeQcowPassphrase(virConnectPtr conn,
                        virDomainDiskDefPtr disk,
                        char **secretRet,
                        size_t *secretLen)
789 790 791 792 793
{
    virSecretPtr secret;
    char *passphrase;
    unsigned char *data;
    size_t size;
794
    int ret = -1;
795
    virStorageEncryptionPtr enc;
796

797 798 799 800 801 802 803
    if (!disk->encryption) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("disk %s does not have any encryption information"),
                        disk->src);
        return -1;
    }
    enc = disk->encryption;
804 805

    if (!conn) {
806 807
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        "%s", _("cannot find secrets without a connection"));
808
        goto cleanup;
809 810 811 812 813
    }

    if (conn->secretDriver == NULL ||
        conn->secretDriver->lookupByUUID == NULL ||
        conn->secretDriver->getValue == NULL) {
814 815
        qemuReportError(VIR_ERR_NO_SUPPORT, "%s",
                        _("secret storage not supported"));
816
        goto cleanup;
817 818 819 820 821 822
    }

    if (enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_QCOW ||
        enc->nsecrets != 1 ||
        enc->secrets[0]->type !=
        VIR_STORAGE_ENCRYPTION_SECRET_TYPE_PASSPHRASE) {
823
        qemuReportError(VIR_ERR_INVALID_DOMAIN,
824
                        _("invalid <encryption> for volume %s"), disk->src);
825
        goto cleanup;
826 827 828 829 830
    }

    secret = conn->secretDriver->lookupByUUID(conn,
                                              enc->secrets[0]->uuid);
    if (secret == NULL)
831
        goto cleanup;
832 833 834 835
    data = conn->secretDriver->getValue(secret, &size,
                                        VIR_SECRET_GET_VALUE_INTERNAL_CALL);
    virUnrefSecret(secret);
    if (data == NULL)
836
        goto cleanup;
837 838 839 840

    if (memchr(data, '\0', size) != NULL) {
        memset(data, 0, size);
        VIR_FREE(data);
841 842
        qemuReportError(VIR_ERR_INVALID_SECRET,
                        _("format='qcow' passphrase for %s must not contain a "
843
                          "'\\0'"), disk->src);
844
        goto cleanup;
845 846 847 848 849
    }

    if (VIR_ALLOC_N(passphrase, size + 1) < 0) {
        memset(data, 0, size);
        VIR_FREE(data);
850
        virReportOOMError();
851
        goto cleanup;
852 853 854 855 856 857 858 859 860 861
    }
    memcpy(passphrase, data, size);
    passphrase[size] = '\0';

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

    *secretRet = passphrase;
    *secretLen = size;

862 863 864
    ret = 0;

cleanup:
865 866
    return ret;
}
867

868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888
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);
889
    return ret;
890 891
}

892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913

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;
}


914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945
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;
}


946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974
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;
}


975 976 977 978 979 980
static int
qemuHandleDomainWatchdog(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                         virDomainObjPtr vm,
                         int action)
{
    struct qemud_driver *driver = qemu_driver;
981 982
    virDomainEventPtr watchdogEvent = NULL;
    virDomainEventPtr lifecycleEvent = NULL;
983 984

    virDomainObjLock(vm);
985 986 987 988 989 990 991 992 993 994 995 996 997 998
    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);
    }
999 1000
    virDomainObjUnlock(vm);

1001
    if (watchdogEvent || lifecycleEvent) {
1002
        qemuDriverLock(driver);
1003 1004 1005 1006
        if (watchdogEvent)
            qemuDomainEventQueue(driver, watchdogEvent);
        if (lifecycleEvent)
            qemuDomainEventQueue(driver, lifecycleEvent);
1007 1008 1009 1010 1011 1012 1013
        qemuDriverUnlock(driver);
    }

    return 0;
}


1014 1015 1016 1017
static int
qemuHandleDomainIOError(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                        virDomainObjPtr vm,
                        const char *diskAlias,
1018 1019
                        int action,
                        const char *reason)
1020 1021
{
    struct qemud_driver *driver = qemu_driver;
1022
    virDomainEventPtr ioErrorEvent = NULL;
1023
    virDomainEventPtr ioErrorEvent2 = NULL;
1024
    virDomainEventPtr lifecycleEvent = NULL;
1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039
    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 = "";
    }

1040
    ioErrorEvent = virDomainEventIOErrorNewFromObj(vm, srcPath, devAlias, action);
1041
    ioErrorEvent2 = virDomainEventIOErrorReasonNewFromObj(vm, srcPath, devAlias, action, reason);
1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054

    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);
    }
1055 1056
    virDomainObjUnlock(vm);

1057
    if (ioErrorEvent || ioErrorEvent2 || lifecycleEvent) {
1058
        qemuDriverLock(driver);
1059 1060
        if (ioErrorEvent)
            qemuDomainEventQueue(driver, ioErrorEvent);
1061 1062
        if (ioErrorEvent2)
            qemuDomainEventQueue(driver, ioErrorEvent2);
1063 1064
        if (lifecycleEvent)
            qemuDomainEventQueue(driver, lifecycleEvent);
1065 1066 1067 1068 1069 1070 1071
        qemuDriverUnlock(driver);
    }

    return 0;
}


1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163
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;
}


1164 1165 1166
static qemuMonitorCallbacks monitorCallbacks = {
    .eofNotify = qemuHandleMonitorEOF,
    .diskSecretLookup = findVolumeQcowPassphrase,
1167
    .domainStop = qemuHandleDomainStop,
1168
    .domainReset = qemuHandleDomainReset,
1169
    .domainRTCChange = qemuHandleDomainRTCChange,
1170
    .domainWatchdog = qemuHandleDomainWatchdog,
1171
    .domainIOError = qemuHandleDomainIOError,
1172
    .domainGraphics = qemuHandleDomainGraphics,
1173 1174
};

1175
static int
1176
qemuConnectMonitor(struct qemud_driver *driver, virDomainObjPtr vm)
1177
{
1178
    qemuDomainObjPrivatePtr priv = vm->privateData;
1179
    int ret;
1180

1181 1182 1183 1184
    /* Hold an extra reference because we can't allow 'vm' to be
     * deleted while the monitor is active */
    virDomainObjRef(vm);

1185 1186
    if ((priv->mon = qemuMonitorOpen(vm,
                                     priv->monConfig,
D
Daniel P. Berrange 已提交
1187
                                     priv->monJSON,
1188
                                     &monitorCallbacks)) == NULL) {
1189
        VIR_ERROR(_("Failed to connect monitor for %s"), vm->def->name);
1190
        return -1;
1191
    }
1192

1193 1194 1195 1196 1197 1198 1199 1200 1201 1202
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
    ret = qemuMonitorSetCapabilities(priv->mon);
    qemuDomainObjExitMonitorWithDriver(driver, vm);

    if (ret < 0) {
        qemuMonitorClose(priv->mon);
        priv->mon = NULL;
    }

    return ret;
1203
}
1204 1205 1206 1207

/*
 * Open an existing VM's monitor, re-detect VCPU threads
 * and re-reserve the security labels in use
1208
 */
1209 1210
static void
qemuReconnectDomain(void *payload, const char *name ATTRIBUTE_UNUSED, void *opaque)
1211
{
1212 1213
    virDomainObjPtr obj = payload;
    struct qemud_driver *driver = opaque;
1214
    qemuDomainObjPrivatePtr priv;
1215
    unsigned long long qemuCmdFlags;
1216 1217

    virDomainObjLock(obj);
1218

1219 1220
    VIR_DEBUG("Reconnect monitor to %p '%s'", obj, obj->def->name);

1221 1222
    priv = obj->privateData;

1223
    /* XXX check PID liveliness & EXE path */
1224
    if (qemuConnectMonitor(driver, obj) < 0)
1225
        goto error;
1226

1227 1228 1229 1230
    if (qemuUpdateActivePciHostdevs(driver, obj->def) < 0) {
        goto error;
    }

1231 1232 1233 1234 1235 1236 1237 1238 1239
    /* 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 &&
        (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE))
        priv->persistentAddrs = 1;

1240 1241 1242
    if (!(priv->pciaddrs = qemuDomainPCIAddressSetCreate(obj->def)))
        goto error;

1243
    if (driver->securityDriver &&
1244
        driver->securityDriver->domainReserveSecurityLabel &&
1245
        driver->securityDriver->domainReserveSecurityLabel(obj) < 0)
1246
        goto error;
1247

1248 1249
    if (obj->def->id >= driver->nextvmid)
        driver->nextvmid = obj->def->id + 1;
1250

1251 1252
    virDomainObjUnlock(obj);
    return;
1253

1254
error:
1255 1256 1257
    /* 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 */
1258
    qemudShutdownVMDaemon(driver, obj, 0);
1259 1260 1261 1262
    if (!obj->persistent)
        virDomainRemoveInactive(&driver->domains, obj);
    else
        virDomainObjUnlock(obj);
1263
}
1264

1265
/**
1266
 * qemudReconnectDomains
1267 1268 1269 1270 1271 1272 1273
 *
 * Try to re-open the resources for live VMs that we care
 * about.
 */
static void
qemuReconnectDomains(struct qemud_driver *driver)
{
1274
    virHashForEach(driver->domains.objs, qemuReconnectDomain, driver);
1275 1276
}

1277

1278 1279 1280 1281 1282 1283
static int
qemudSecurityInit(struct qemud_driver *qemud_drv)
{
    int ret;
    virSecurityDriverPtr security_drv;

1284 1285 1286
    qemuSecurityStackedSetDriver(qemud_drv);
    qemuSecurityDACSetDriver(qemud_drv);

1287 1288 1289 1290 1291 1292
    ret = virSecurityDriverStartup(&security_drv,
                                   qemud_drv->securityDriverName);
    if (ret == -1) {
        VIR_ERROR0(_("Failed to start security driver"));
        return -1;
    }
1293 1294 1295

    /* No primary security driver wanted to be enabled: just setup
     * the DAC driver on its own */
1296
    if (ret == -2) {
1297
        qemud_drv->securityDriver = &qemuDACSecurityDriver;
1298
        VIR_INFO0(_("No security driver available"));
1299 1300 1301 1302 1303
    } else {
        qemud_drv->securityPrimaryDriver = security_drv;
        qemud_drv->securitySecondaryDriver = &qemuDACSecurityDriver;
        qemud_drv->securityDriver = &qemuStackedSecurityDriver;
        VIR_INFO("Initialized security driver %s", security_drv->name);
1304 1305
    }

1306
    return 0;
1307
}
1308 1309


1310 1311
static virCapsPtr
qemuCreateCapabilities(virCapsPtr oldcaps,
1312
                       struct qemud_driver *driver)
1313 1314 1315 1316 1317
{
    virCapsPtr caps;

    /* Basic host arch / guest machine capabilities */
    if (!(caps = qemudCapsInit(oldcaps))) {
1318
        virReportOOMError();
1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329
        return NULL;
    }

    /* Domain XML parser hooks */
    caps->privateDataAllocFunc = qemuDomainObjPrivateAlloc;
    caps->privateDataFreeFunc = qemuDomainObjPrivateFree;
    caps->privateDataXMLFormat = qemuDomainObjPrivateXMLFormat;
    caps->privateDataXMLParse = qemuDomainObjPrivateXMLParse;


    /* Security driver data */
1330
    if (driver->securityPrimaryDriver) {
1331 1332
        const char *doi, *model;

1333 1334
        doi = virSecurityDriverGetDOI(driver->securityPrimaryDriver);
        model = virSecurityDriverGetModel(driver->securityPrimaryDriver);
1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347

        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:
1348
    virReportOOMError();
1349 1350 1351
    virCapabilitiesFree(caps);
    return NULL;
}
1352

C
Chris Lalancette 已提交
1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403
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) {
        VIR_ERROR("Failed to allocate memory for snapshot directory for domain %s",
                   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)
            VIR_ERROR("Failed to open snapshot directory %s for domain %s: %s",
                      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) {
            VIR_ERROR0("Failed to allocate memory for path");
            continue;
        }

        ret = virFileReadAll(fullpath, 1024*1024*1, &xmlStr);
        if (ret < 0) {
            /* Nothing we can do here, skip this one */
            VIR_ERROR("Failed to read snapshot file %s: %s", fullpath,
                      virStrerror(errno, ebuf, sizeof(ebuf)));
1404
            VIR_FREE(fullpath);
C
Chris Lalancette 已提交
1405 1406 1407 1408 1409 1410 1411
            continue;
        }

        def = virDomainSnapshotDefParseString(xmlStr, 0);
        if (def == NULL) {
            /* Nothing we can do here, skip this one */
            VIR_ERROR("Failed to parse snapshot XML from file '%s'", fullpath);
1412
            VIR_FREE(fullpath);
C
Chris Lalancette 已提交
1413 1414 1415 1416
            VIR_FREE(xmlStr);
            continue;
        }

1417
        virDomainSnapshotAssignDef(&vm->snapshots, def);
C
Chris Lalancette 已提交
1418

1419
        VIR_FREE(fullpath);
C
Chris Lalancette 已提交
1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440
        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);
}

1441 1442 1443 1444 1445 1446
/**
 * qemudStartup:
 *
 * Initialization function for the QEmu daemon
 */
static int
1447
qemudStartup(int privileged) {
1448
    char *base = NULL;
D
Daniel P. Berrange 已提交
1449
    char driverConf[PATH_MAX];
1450
    int rc;
1451

1452
    if (VIR_ALLOC(qemu_driver) < 0)
1453 1454
        return -1;

1455
    if (virMutexInit(&qemu_driver->lock) < 0) {
1456
        VIR_ERROR("%s", _("cannot initialize mutex"));
1457 1458 1459
        VIR_FREE(qemu_driver);
        return -1;
    }
1460
    qemuDriverLock(qemu_driver);
1461
    qemu_driver->privileged = privileged;
1462

1463 1464 1465
    /* Don't have a dom0 so start from 1 */
    qemu_driver->nextvmid = 1;

1466 1467 1468
    if (virDomainObjListInit(&qemu_driver->domains) < 0)
        goto out_of_memory;

1469
    /* Init callback list */
1470
    if (VIR_ALLOC(qemu_driver->domainEventCallbacks) < 0)
1471
        goto out_of_memory;
1472 1473 1474 1475 1476 1477
    if (!(qemu_driver->domainEventQueue = virDomainEventQueueNew()))
        goto out_of_memory;

    if ((qemu_driver->domainEventTimer =
         virEventAddTimeout(-1, qemuDomainEventFlush, qemu_driver, NULL)) < 0)
        goto error;
1478

1479
    if (privileged) {
1480 1481
        if (virAsprintf(&qemu_driver->logDir,
                        "%s/log/libvirt/qemu", LOCAL_STATE_DIR) == -1)
1482
            goto out_of_memory;
1483

D
Daniel P. Berrange 已提交
1484
        if ((base = strdup (SYSCONF_DIR "/libvirt")) == NULL)
1485
            goto out_of_memory;
1486 1487

        if (virAsprintf(&qemu_driver->stateDir,
1488
                      "%s/run/libvirt/qemu", LOCAL_STATE_DIR) == -1)
1489
            goto out_of_memory;
1490 1491 1492 1493 1494 1495 1496 1497

        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;
1498 1499 1500
        if (virAsprintf(&qemu_driver->saveDir,
                      "%s/lib/libvirt/qemu/save/", LOCAL_STATE_DIR) == -1)
            goto out_of_memory;
C
Chris Lalancette 已提交
1501 1502 1503
        if (virAsprintf(&qemu_driver->snapshotDir,
                        "%s/lib/libvirt/qemu/snapshot", LOCAL_STATE_DIR) == -1)
            goto out_of_memory;
1504
    } else {
1505
        uid_t uid = geteuid();
1506
        char *userdir = virGetUserDirectory(uid);
1507
        if (!userdir)
1508
            goto error;
1509

1510
        if (virAsprintf(&qemu_driver->logDir,
1511 1512
                        "%s/.libvirt/qemu/log", userdir) == -1) {
            VIR_FREE(userdir);
1513
            goto out_of_memory;
1514
        }
1515

1516 1517
        if (virAsprintf(&base, "%s/.libvirt", userdir) == -1) {
            VIR_FREE(userdir);
1518
            goto out_of_memory;
1519 1520
        }
        VIR_FREE(userdir);
1521 1522 1523

        if (virAsprintf(&qemu_driver->stateDir, "%s/qemu/run", base) == -1)
            goto out_of_memory;
1524 1525 1526 1527
        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;
1528 1529
        if (virAsprintf(&qemu_driver->saveDir, "%s/qemu/save", base) == -1)
            goto out_of_memory;
C
Chris Lalancette 已提交
1530 1531
        if (virAsprintf(&qemu_driver->snapshotDir, "%s/qemu/snapshot", base) == -1)
            goto out_of_memory;
1532 1533
    }

L
Laine Stump 已提交
1534
    if (virFileMakePath(qemu_driver->stateDir) != 0) {
1535
        char ebuf[1024];
1536
        VIR_ERROR(_("Failed to create state dir '%s': %s"),
1537
                  qemu_driver->stateDir, virStrerror(errno, ebuf, sizeof ebuf));
1538
        goto error;
1539
    }
L
Laine Stump 已提交
1540
    if (virFileMakePath(qemu_driver->libDir) != 0) {
1541
        char ebuf[1024];
1542
        VIR_ERROR(_("Failed to create lib dir '%s': %s"),
1543 1544 1545
                  qemu_driver->libDir, virStrerror(errno, ebuf, sizeof ebuf));
        goto error;
    }
L
Laine Stump 已提交
1546
    if (virFileMakePath(qemu_driver->cacheDir) != 0) {
1547
        char ebuf[1024];
1548
        VIR_ERROR(_("Failed to create cache dir '%s': %s"),
1549 1550 1551
                  qemu_driver->cacheDir, virStrerror(errno, ebuf, sizeof ebuf));
        goto error;
    }
1552 1553 1554 1555 1556 1557
    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
1558 1559 1560 1561 1562 1563
    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;
    }
1564 1565 1566 1567

    /* Configuration paths are either ~/.libvirt/qemu/... (session) or
     * /etc/libvirt/qemu/... (system).
     */
D
Daniel P. Berrange 已提交
1568
    if (snprintf (driverConf, sizeof(driverConf), "%s/qemu.conf", base) == -1)
1569
        goto out_of_memory;
D
Daniel P. Berrange 已提交
1570
    driverConf[sizeof(driverConf)-1] = '\0';
1571

1572
    if (virAsprintf(&qemu_driver->configDir, "%s/qemu", base) == -1)
1573 1574
        goto out_of_memory;

1575
    if (virAsprintf(&qemu_driver->autostartDir, "%s/qemu/autostart", base) == -1)
1576 1577
        goto out_of_memory;

1578
    VIR_FREE(base);
1579

1580 1581 1582 1583 1584 1585 1586
    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)));
    }

1587 1588 1589 1590
    if (qemudLoadDriverConfig(qemu_driver, driverConf) < 0) {
        goto error;
    }

1591 1592
    if (qemudSecurityInit(qemu_driver) < 0)
        goto error;
D
Daniel P. Berrange 已提交
1593

1594
    if ((qemu_driver->caps = qemuCreateCapabilities(NULL,
1595
                                                    qemu_driver)) == NULL)
1596
        goto error;
1597

1598
    if ((qemu_driver->activePciHostdevs = pciDeviceListNew()) == NULL)
1599 1600
        goto error;

1601 1602
    if (privileged) {
        if (chown(qemu_driver->libDir, qemu_driver->user, qemu_driver->group) < 0) {
1603
            virReportSystemError(errno,
1604 1605 1606 1607 1608
                                 _("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) {
1609
            virReportSystemError(errno,
1610 1611 1612 1613
                                 _("unable to set ownership of '%s' to %d:%d"),
                                 qemu_driver->cacheDir, qemu_driver->user, qemu_driver->group);
            goto error;
        }
1614 1615 1616 1617 1618 1619
        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
1620 1621 1622 1623 1624 1625
        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;
        }
1626 1627
    }

1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641
    /* 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) {
1642
            virReportSystemError(rc,
1643 1644 1645 1646 1647 1648
                                 _("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) {
1649
            virReportSystemError(errno,
1650 1651 1652 1653 1654 1655 1656 1657 1658
                                 _("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;
    }

1659
    /* Get all the running persistent or transient configs first */
1660
    if (virDomainLoadAllConfigs(qemu_driver->caps,
1661 1662 1663 1664 1665 1666 1667 1668 1669
                                &qemu_driver->domains,
                                qemu_driver->stateDir,
                                NULL,
                                1, NULL, NULL) < 0)
        goto error;

    qemuReconnectDomains(qemu_driver);

    /* Then inactive persistent configs */
1670
    if (virDomainLoadAllConfigs(qemu_driver->caps,
1671 1672
                                &qemu_driver->domains,
                                qemu_driver->configDir,
1673
                                qemu_driver->autostartDir,
1674
                                0, NULL, NULL) < 0)
1675
        goto error;
C
Chris Lalancette 已提交
1676 1677 1678 1679 1680


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

1681 1682
    qemuDriverUnlock(qemu_driver);

1683 1684
    qemudAutostartConfigs(qemu_driver);

1685

1686 1687
    return 0;

1688
out_of_memory:
1689
    virReportOOMError();
1690 1691 1692
error:
    if (qemu_driver)
        qemuDriverUnlock(qemu_driver);
1693
    VIR_FREE(base);
1694
    qemudShutdown();
1695 1696 1697
    return -1;
}

1698 1699 1700 1701
static void qemudNotifyLoadDomain(virDomainObjPtr vm, int newVM, void *opaque)
{
    struct qemud_driver *driver = opaque;

1702 1703 1704 1705 1706 1707 1708 1709
    if (newVM) {
        virDomainEventPtr event =
            virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_DEFINED,
                                     VIR_DOMAIN_EVENT_DEFINED_ADDED);
        if (event)
            qemuDomainEventQueue(driver, event);
    }
1710 1711
}

1712 1713 1714 1715 1716 1717 1718 1719
/**
 * qemudReload:
 *
 * Function to restart the QEmu daemon, it will recheck the configuration
 * files and update its state and the networking
 */
static int
qemudReload(void) {
1720 1721 1722
    if (!qemu_driver)
        return 0;

1723
    qemuDriverLock(qemu_driver);
1724
    virDomainLoadAllConfigs(qemu_driver->caps,
1725 1726
                            &qemu_driver->domains,
                            qemu_driver->configDir,
1727
                            qemu_driver->autostartDir,
1728
                            0, qemudNotifyLoadDomain, qemu_driver);
1729
    qemuDriverUnlock(qemu_driver);
1730

1731
    qemudAutostartConfigs(qemu_driver);
1732 1733

    return 0;
1734 1735
}

1736 1737 1738 1739 1740 1741 1742 1743 1744 1745
/**
 * 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) {
1746
    int active = 0;
1747

1748 1749 1750
    if (!qemu_driver)
        return 0;

1751
    /* XXX having to iterate here is not great because it requires many locks */
1752
    qemuDriverLock(qemu_driver);
1753
    active = virDomainObjListNumOfDomains(&qemu_driver->domains, 1);
1754 1755
    qemuDriverUnlock(qemu_driver);
    return active;
1756 1757
}

1758 1759 1760 1761 1762 1763 1764
/**
 * qemudShutdown:
 *
 * Shutdown the QEmu daemon, it will stop all active domains and networks
 */
static int
qemudShutdown(void) {
1765
    int i;
1766

1767
    if (!qemu_driver)
1768
        return -1;
1769

1770
    qemuDriverLock(qemu_driver);
1771
    pciDeviceListFree(qemu_driver->activePciHostdevs);
1772 1773
    virCapabilitiesFree(qemu_driver->caps);

1774
    virDomainObjListDeinit(&qemu_driver->domains);
1775

1776
    VIR_FREE(qemu_driver->securityDriverName);
1777
    VIR_FREE(qemu_driver->logDir);
1778 1779
    VIR_FREE(qemu_driver->configDir);
    VIR_FREE(qemu_driver->autostartDir);
1780
    VIR_FREE(qemu_driver->stateDir);
1781 1782
    VIR_FREE(qemu_driver->libDir);
    VIR_FREE(qemu_driver->cacheDir);
1783
    VIR_FREE(qemu_driver->saveDir);
C
Chris Lalancette 已提交
1784
    VIR_FREE(qemu_driver->snapshotDir);
1785
    VIR_FREE(qemu_driver->vncTLSx509certdir);
J
Jim Meyering 已提交
1786
    VIR_FREE(qemu_driver->vncListen);
1787
    VIR_FREE(qemu_driver->vncPassword);
1788
    VIR_FREE(qemu_driver->vncSASLdir);
1789
    VIR_FREE(qemu_driver->saveImageFormat);
1790 1791
    VIR_FREE(qemu_driver->hugetlbfs_mount);
    VIR_FREE(qemu_driver->hugepage_path);
D
Daniel P. Berrange 已提交
1792

1793 1794 1795 1796 1797 1798
    if (qemu_driver->cgroupDeviceACL) {
        for (i = 0 ; qemu_driver->cgroupDeviceACL[i] != NULL ; i++)
            VIR_FREE(qemu_driver->cgroupDeviceACL[i]);
        VIR_FREE(qemu_driver->cgroupDeviceACL);
    }

1799 1800
    /* Free domain callback list */
    virDomainEventCallbackListFree(qemu_driver->domainEventCallbacks);
1801 1802 1803 1804
    virDomainEventQueueFree(qemu_driver->domainEventQueue);

    if (qemu_driver->domainEventTimer != -1)
        virEventRemoveTimeout(qemu_driver->domainEventTimer);
1805

1806 1807 1808
    if (qemu_driver->brctl)
        brShutdown(qemu_driver->brctl);

1809 1810
    virCgroupFree(&qemu_driver->cgroup);

1811
    qemuDriverUnlock(qemu_driver);
1812
    virMutexDestroy(&qemu_driver->lock);
1813
    VIR_FREE(qemu_driver);
1814 1815

    return 0;
1816 1817
}

1818
typedef int qemuLogHandleOutput(virDomainObjPtr vm,
1819 1820
                                const char *output,
                                int fd);
1821 1822 1823 1824 1825

/*
 * Returns -1 for error, 0 on success
 */
static int
1826
qemudReadLogOutput(virDomainObjPtr vm,
1827 1828
                   int fd,
                   char *buf,
G
Guido Günther 已提交
1829
                   size_t buflen,
1830
                   qemuLogHandleOutput func,
1831 1832 1833
                   const char *what,
                   int timeout)
{
1834
    int retries = (timeout*10);
1835
    int got = 0;
1836 1837 1838
    buf[0] = '\0';

    while (retries) {
1839
        ssize_t func_ret, ret;
1840
        int isdead = 0;
G
Guido Günther 已提交
1841

1842
        func_ret = func(vm, buf, fd);
1843

1844 1845
        if (kill(vm->pid, 0) == -1 && errno == ESRCH)
            isdead = 1;
1846

1847 1848
        /* Any failures should be detected before we read the log, so we
         * always have something useful to report on failure. */
1849 1850
        ret = saferead(fd, buf+got, buflen-got-1);
        if (ret < 0) {
1851
            virReportSystemError(errno,
1852 1853 1854 1855 1856
                                 _("Failure while reading %s log output"),
                                 what);
            return -1;
        }

1857 1858 1859
        got += ret;
        buf[got] = '\0';
        if (got == buflen-1) {
1860
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
1861 1862
                            _("Out of space while reading %s log output: %s"),
                            what, buf);
1863 1864 1865 1866
            return -1;
        }

        if (isdead) {
1867
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
1868 1869
                            _("Process exited while reading %s log output: %s"),
                            what, buf);
1870 1871 1872
            return -1;
        }

1873 1874
        if (func_ret <= 0)
            return func_ret;
1875 1876 1877 1878

        usleep(100*1000);
        retries--;
    }
1879

1880
    qemuReportError(VIR_ERR_INTERNAL_ERROR,
1881 1882
                    _("Timed out while reading %s log output: %s"),
                    what, buf);
1883 1884 1885
    return -1;
}

1886

1887 1888 1889 1890 1891 1892 1893 1894
/*
 * 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
 */
1895
static int
1896
qemudExtractTTYPath(const char *haystack,
1897 1898
                    size_t *offset,
                    char **path)
1899
{
1900
    static const char needle[] = "char device redirected to";
1901
    char *tmp, *dev;
1902

1903
    VIR_FREE(*path);
1904
    /* First look for our magic string */
1905 1906 1907 1908 1909
    if (!(tmp = strstr(haystack + *offset, needle))) {
        return 1;
    }
    tmp += sizeof(needle);
    dev = tmp;
1910

1911 1912 1913 1914 1915
    /*
     * And look for first whitespace character and nul terminate
     * to mark end of the pty path
     */
    while (*tmp) {
1916
        if (c_isspace(*tmp)) {
1917 1918
            *path = strndup(dev, tmp-dev);
            if (*path == NULL) {
1919
                virReportOOMError();
1920 1921
                return -1;
            }
1922

1923
            /* ... now further update offset till we get EOL */
1924
            *offset = tmp - haystack;
1925 1926
            return 0;
        }
1927
        tmp++;
1928 1929 1930 1931 1932
    }

    /*
     * We found a path, but didn't find any whitespace,
     * so it must be still incomplete - we should at
1933 1934
     * least see a \n - indicate that we want to carry
     * on trying again
1935
     */
1936
    return 1;
1937 1938
}

1939
static int
1940
qemudFindCharDevicePTYsMonitor(virDomainObjPtr vm,
1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955
                               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) {                                           \
1956 1957 1958
                if (chr->data.file.path == NULL) {                        \
                    /* neither the log output nor 'info chardev' had a */ \
                    /* pty path for this chardev, report an error */      \
1959 1960 1961
                    qemuReportError(VIR_ERR_INTERNAL_ERROR,               \
                                    _("no assigned pty for device %s"), id); \
                    return -1;                                            \
1962 1963 1964 1965 1966
                } else {                                                  \
                    /* 'info chardev' had no pty path for this chardev, */\
                    /* but the log output had, so we're fine */           \
                    continue;                                             \
                }                                                         \
1967 1968
            }                                                             \
                                                                          \
1969
            VIR_FREE(chr->data.file.path);                                \
1970
            chr->data.file.path = strdup(path);                           \
1971 1972
                                                                          \
            if (chr->data.file.path == NULL) {                            \
1973
                virReportOOMError();                                      \
1974 1975
                return -1;                                                \
            }                                                             \
1976 1977 1978 1979 1980 1981
        }                                                                 \
    }

    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");
1982
#undef LOOKUP_PTYS
1983 1984 1985 1986

    return 0;
}

1987
static int
1988
qemudFindCharDevicePTYs(virDomainObjPtr vm,
1989 1990
                        const char *output,
                        int fd ATTRIBUTE_UNUSED)
1991
{
1992
    size_t offset = 0;
1993
    int ret, i;
1994 1995

    /* The order in which QEMU prints out the PTY paths is
1996 1997
       the order in which it procsses its serial and parallel
       device args. This code must match that ordering.... */
1998

1999
    /* first comes the serial devices */
2000 2001
    for (i = 0 ; i < vm->def->nserials ; i++) {
        virDomainChrDefPtr chr = vm->def->serials[i];
2002
        if (chr->type == VIR_DOMAIN_CHR_TYPE_PTY) {
2003
            if ((ret = qemudExtractTTYPath(output, &offset,
2004
                                           &chr->data.file.path)) != 0)
2005
                return ret;
2006 2007 2008
        }
    }

2009
    /* then the parallel devices */
2010 2011
    for (i = 0 ; i < vm->def->nparallels ; i++) {
        virDomainChrDefPtr chr = vm->def->parallels[i];
2012
        if (chr->type == VIR_DOMAIN_CHR_TYPE_PTY) {
2013
            if ((ret = qemudExtractTTYPath(output, &offset,
2014
                                           &chr->data.file.path)) != 0)
2015
                return ret;
2016 2017 2018
        }
    }

2019 2020 2021 2022
    /* 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) {
2023
            if ((ret = qemudExtractTTYPath(output, &offset,
2024 2025 2026 2027 2028
                                           &chr->data.file.path)) != 0)
                return ret;
        }
    }

2029
    return 0;
2030 2031
}

2032 2033 2034 2035 2036
static void qemudFreePtyPath(void *payload, const char *name ATTRIBUTE_UNUSED)
{
    VIR_FREE(payload);
}

2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050
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';
}

2051
static int
2052
qemudWaitForMonitor(struct qemud_driver* driver,
2053
                    virDomainObjPtr vm, off_t pos)
2054
{
2055
    char buf[4096] = ""; /* Plenty of space to get startup greeting */
2056
    int logfd;
2057
    int ret = -1;
2058
    virHashTablePtr paths = NULL;
2059

2060
    if ((logfd = qemudLogReadFD(driver->logDir, vm->def->name, pos)) < 0)
2061
        return -1;
2062

2063 2064 2065 2066
    if (qemudReadLogOutput(vm, logfd, buf, sizeof(buf),
                           qemudFindCharDevicePTYs,
                           "console", 30) < 0)
        goto closelog;
2067

2068
    VIR_DEBUG("Connect monitor to %p '%s'", vm, vm->def->name);
2069 2070 2071
    if (qemuConnectMonitor(driver, vm) < 0) {
        goto cleanup;
    }
2072

2073 2074 2075 2076
    /* 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. */
2077
    paths = virHashCreate(0);
2078
    if (paths == NULL) {
2079
        virReportOOMError();
2080 2081 2082
        goto cleanup;
    }

2083
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
2084 2085
    qemuDomainObjPrivatePtr priv = vm->privateData;
    ret = qemuMonitorGetPtyPaths(priv->mon, paths);
2086
    qemuDomainObjExitMonitorWithDriver(driver, vm);
2087 2088 2089

    VIR_DEBUG("qemuMonitorGetPtyPaths returned %i", ret);
    if (ret == 0) {
2090
        ret = qemudFindCharDevicePTYsMonitor(vm, paths);
2091 2092 2093 2094 2095 2096 2097
    }

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

2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110
    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];
2111
        VIR_WARN("Unable to close logfile: %s",
2112 2113 2114
                 virStrerror(errno, ebuf, sizeof ebuf));
    }

2115
    return ret;
2116 2117
}

2118
static int
2119
qemuDetectVcpuPIDs(struct qemud_driver *driver,
2120 2121 2122
                   virDomainObjPtr vm) {
    pid_t *cpupids = NULL;
    int ncpupids;
2123
    qemuDomainObjPrivatePtr priv = vm->privateData;
2124

2125
    if (vm->def->virtType != VIR_DOMAIN_VIRT_KVM) {
2126 2127
        priv->nvcpupids = 1;
        if (VIR_ALLOC_N(priv->vcpupids, priv->nvcpupids) < 0) {
2128
            virReportOOMError();
2129 2130
            return -1;
        }
2131
        priv->vcpupids[0] = vm->pid;
2132 2133 2134
        return 0;
    }

2135
    /* What follows is now all KVM specific */
2136

2137
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
2138
    if ((ncpupids = qemuMonitorGetCPUInfo(priv->mon, &cpupids)) < 0) {
2139
        qemuDomainObjExitMonitorWithDriver(driver, vm);
2140
        return -1;
2141
    }
2142
    qemuDomainObjExitMonitorWithDriver(driver, vm);
2143

2144 2145 2146
    /* Treat failure to get VCPU<->PID mapping as non-fatal */
    if (ncpupids == 0)
        return 0;
2147

2148
    if (ncpupids != vm->def->vcpus) {
2149 2150 2151
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("got wrong number of vCPU pids from QEMU monitor. got %d, wanted %d"),
                        ncpupids, (int)vm->def->vcpus);
2152 2153 2154
        VIR_FREE(cpupids);
        return -1;
    }
2155

2156 2157
    priv->nvcpupids = ncpupids;
    priv->vcpupids = cpupids;
2158 2159 2160
    return 0;
}

2161 2162 2163
/*
 * To be run between fork/exec of QEMU only
 */
2164
static int
2165 2166
qemudInitCpuAffinity(virDomainObjPtr vm)
{
2167
    int i, hostcpus, maxcpu = QEMUD_CPUMASK_LEN;
2168
    virNodeInfo nodeinfo;
2169 2170
    unsigned char *cpumap;
    int cpumaplen;
2171 2172

    DEBUG0("Setting CPU affinity");
2173

2174
    if (nodeGetInfo(NULL, &nodeinfo) < 0)
2175 2176 2177 2178
        return -1;

    /* setaffinity fails if you set bits for CPUs which
     * aren't present, so we have to limit ourselves */
2179 2180 2181
    hostcpus = VIR_NODEINFO_MAXCPUS(nodeinfo);
    if (maxcpu > hostcpus)
        maxcpu = hostcpus;
2182

2183 2184
    cpumaplen = VIR_CPU_MAPLEN(maxcpu);
    if (VIR_ALLOC_N(cpumap, cpumaplen) < 0) {
2185
        virReportOOMError();
2186 2187 2188
        return -1;
    }

D
Daniel P. Berrange 已提交
2189
    if (vm->def->cpumask) {
2190 2191 2192
        /* 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
2193
            if (vm->def->cpumask[i])
2194
                VIR_USE_CPU(cpumap, i);
D
Daniel P. Berrange 已提交
2195
    } else {
2196 2197 2198 2199
        /* 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
2200
        for (i = 0 ; i < maxcpu ; i++)
2201
            VIR_USE_CPU(cpumap, i);
D
Daniel P. Berrange 已提交
2202
    }
2203

2204 2205 2206 2207 2208 2209 2210 2211
    /* 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;
2212
    }
2213
    VIR_FREE(cpumap);
2214 2215 2216 2217 2218

    return 0;
}


2219
static int
2220 2221 2222 2223
qemuInitPasswords(virConnectPtr conn,
                  struct qemud_driver *driver,
                  virDomainObjPtr vm,
                  unsigned long long qemuCmdFlags) {
2224
    int ret = 0;
2225
    qemuDomainObjPrivatePtr priv = vm->privateData;
2226

2227 2228 2229
    if ((vm->def->ngraphics == 1) &&
        vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
        (vm->def->graphics[0]->data.vnc.passwd || driver->vncPassword)) {
2230

2231
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
2232
        ret = qemuMonitorSetVNCPassword(priv->mon,
2233 2234 2235
                                        vm->def->graphics[0]->data.vnc.passwd ?
                                        vm->def->graphics[0]->data.vnc.passwd :
                                        driver->vncPassword);
2236
        qemuDomainObjExitMonitorWithDriver(driver, vm);
2237 2238
    }

2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261
    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);
2262
            VIR_FREE(secret);
2263 2264 2265 2266 2267 2268 2269
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            if (ret < 0)
                goto cleanup;
        }
    }

cleanup:
2270
    return ret;
2271 2272 2273
}


2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502
#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

#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;
}


/*
 * 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
2503
qemuDetectPCIAddresses(virDomainObjPtr vm,
2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520
                       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) {
2521 2522 2523
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for VirtIO disk %s"),
                            vm->def->disks[i]->dst);
2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534
            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) {
2535 2536 2537
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for %s NIC"),
                            vm->def->nets[i]->model);
2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548
            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) {
2549 2550 2551
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for controller %s"),
                            virDomainControllerTypeToString(vm->def->controllers[i]->type));
2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562
            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) {
2563 2564 2565
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for video adapter %s"),
                            virDomainVideoTypeToString(vm->def->videos[i]->type));
2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576
            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),
                                     vendor, product,
                                     addrs,  naddrs) < 0) {
2577 2578 2579
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for sound adapter %s"),
                            virDomainSoundModelTypeToString(vm->def->sounds[i]->model));
2580 2581 2582 2583 2584 2585 2586 2587 2588 2589
            return -1;
        }
    }


    if (vm->def->watchdog &&
        qemuGetPCIWatchdogVendorProduct(vm->def->watchdog, &vendor, &product) == 0) {
        if (qemuAssignNextPCIAddress(&(vm->def->watchdog->info),
                                     vendor, product,
                                     addrs,  naddrs) < 0) {
2590 2591 2592
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for watchdog %s"),
                            virDomainWatchdogModelTypeToString(vm->def->watchdog->model));
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
            return -1;
        }
    }

    /* XXX console (virtio) */


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

    /* XXX USB controller ? */

    /* XXXX virtio balloon ? */

    /* 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);

2625
    ret = qemuDetectPCIAddresses(vm, addrs, naddrs);
2626 2627 2628 2629 2630 2631

    VIR_FREE(addrs);

    return ret;
}

2632
static int qemudNextFreeVNCPort(struct qemud_driver *driver ATTRIBUTE_UNUSED) {
2633 2634
    int i;

2635
    for (i = 5900 ; i < 65535 ; i++) {
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
        int fd;
        int reuse = 1;
        struct sockaddr_in addr;
        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);
            return i;
        }
        close(fd);

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

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

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;
}


2706
static pciDeviceList *
2707
qemuGetPciHostDeviceList(virDomainDefPtr def)
2708 2709
{
    pciDeviceList *list;
2710 2711
    int i;

2712
    if (!(list = pciDeviceListNew()))
2713
        return NULL;
2714 2715 2716 2717 2718 2719 2720 2721 2722 2723

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

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

2724
        dev = pciGetDevice(hostdev->source.subsys.u.pci.domain,
2725 2726 2727
                           hostdev->source.subsys.u.pci.bus,
                           hostdev->source.subsys.u.pci.slot,
                           hostdev->source.subsys.u.pci.function);
2728
        if (!dev) {
2729
            pciDeviceListFree(list);
2730 2731
            return NULL;
        }
2732

2733 2734 2735
        if (pciDeviceListAdd(list, dev) < 0) {
            pciFreeDevice(dev);
            pciDeviceListFree(list);
2736
            return NULL;
2737 2738
        }

2739
        pciDeviceSetManaged(dev, hostdev->managed);
2740 2741
    }

2742 2743 2744 2745
    return list;
}

static int
2746 2747 2748 2749
qemuUpdateActivePciHostdevs(struct qemud_driver *driver,
                            virDomainDefPtr def)
{
    pciDeviceList *pcidevs;
2750
    int i;
2751
    int ret = -1;
2752 2753 2754 2755

    if (!def->nhostdevs)
        return 0;

2756
    if (!(pcidevs = qemuGetPciHostDeviceList(def)))
2757 2758
        return -1;

2759 2760
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
2761 2762 2763
        pciDeviceListSteal(pcidevs, dev);
        if (pciDeviceListAdd(driver->activePciHostdevs, dev) < 0) {
            pciFreeDevice(dev);
2764
            goto cleanup;
2765 2766 2767
        }
    }

2768 2769 2770
    ret = 0;

cleanup:
2771
    pciDeviceListFree(pcidevs);
2772 2773 2774
    return ret;
}

2775

2776
static int
2777 2778
qemuPrepareHostPCIDevices(struct qemud_driver *driver,
                          virDomainDefPtr def)
2779 2780 2781
{
    pciDeviceList *pcidevs;
    int i;
2782
    int ret = -1;
2783

2784
    if (!(pcidevs = qemuGetPciHostDeviceList(def)))
2785 2786
        return -1;

2787
    /* We have to use 3 loops here. *All* devices must
2788 2789
     * be detached before we reset any of them, because
     * in some cases you have to reset the whole PCI,
2790 2791
     * which impacts all devices on it. Also, all devices
     * must be reset before being marked as active.
2792 2793 2794 2795 2796 2797 2798
     */

    /* 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
     */

2799 2800
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
2801
        if (!pciDeviceIsAssignable(dev, !driver->relaxedACS))
2802 2803
            goto cleanup;

2804
        if (pciDeviceGetManaged(dev) &&
2805
            pciDettachDevice(dev) < 0)
2806 2807
            goto cleanup;
    }
2808 2809 2810

    /* Now that all the PCI hostdevs have be dettached, we can safely
     * reset them */
2811 2812
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
2813
        if (pciResetDevice(dev, driver->activePciHostdevs) < 0)
2814 2815
            goto cleanup;
    }
2816

2817
    /* Now mark all the devices as active */
2818 2819
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
2820 2821 2822
        pciDeviceListSteal(pcidevs, dev);
        if (pciDeviceListAdd(driver->activePciHostdevs, dev) < 0) {
            pciFreeDevice(dev);
2823 2824
            goto cleanup;
        }
2825 2826
    }

2827
    ret = 0;
2828

2829
cleanup:
2830
    pciDeviceListFree(pcidevs);
2831
    return ret;
2832 2833
}

2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844

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;
2845
        if (hostdev->source.subsys.type != VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB)
2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883
            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;
}


2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894
static void
qemudReattachManagedDevice(pciDevice *dev)
{
    int retries = 100;

    if (pciDeviceGetManaged(dev)) {
        while (pciWaitForDeviceCleanup(dev, "kvm_assigned_device")
               && retries) {
            usleep(100*1000);
            retries--;
        }
2895
        if (pciReAttachDevice(dev) < 0) {
2896 2897 2898 2899 2900 2901 2902 2903
            virErrorPtr err = virGetLastError();
            VIR_ERROR(_("Failed to re-attach PCI device: %s"),
                      err ? err->message : "");
            virResetError(err);
        }
    }
}

2904
static void
2905
qemuDomainReAttachHostDevices(struct qemud_driver *driver,
2906
                              virDomainDefPtr def)
2907
{
2908
    pciDeviceList *pcidevs;
2909 2910
    int i;

2911 2912
    if (!def->nhostdevs)
        return;
2913

2914
    if (!(pcidevs = qemuGetPciHostDeviceList(def))) {
2915
        virErrorPtr err = virGetLastError();
2916
        VIR_ERROR(_("Failed to allocate pciDeviceList: %s"),
2917 2918 2919
                  err ? err->message : "");
        virResetError(err);
        return;
2920 2921
    }

2922 2923
    /* Again 3 loops; mark all devices as inactive before reset
     * them and reset all the devices before re-attach */
2924

2925 2926
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
2927
        pciDeviceListDel(driver->activePciHostdevs, dev);
2928
    }
2929

2930 2931
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
2932
        if (pciResetDevice(dev, driver->activePciHostdevs) < 0) {
2933
            virErrorPtr err = virGetLastError();
2934
            VIR_ERROR(_("Failed to reset PCI device: %s"),
2935 2936 2937
                      err ? err->message : "");
            virResetError(err);
        }
2938
    }
2939

2940 2941
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
2942
        qemudReattachManagedDevice(dev);
2943
    }
2944

2945
    pciDeviceListFree(pcidevs);
2946 2947
}

2948 2949 2950 2951 2952 2953 2954 2955 2956 2957
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

2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994
static int qemuSetupDiskCgroup(virCgroupPtr cgroup,
                               virDomainObjPtr vm,
                               virDomainDiskDefPtr disk)
{
    char *path = disk->src;
    int ret = -1;

    while (path != NULL) {
        virStorageFileMetadata meta;
        int rc;

        VIR_DEBUG("Process path %s for disk", path);
        rc = virCgroupAllowDevicePath(cgroup, path);
        if (rc != 0) {
            /* Get this for non-block devices */
            if (rc == -EINVAL) {
                VIR_DEBUG("Ignoring EINVAL for %s", path);
            } else {
                virReportSystemError(-rc,
                                     _("Unable to allow device %s for %s"),
                                     path, vm->def->name);
                if (path != disk->src)
                    VIR_FREE(path);
                goto cleanup;
            }
        }

        rc = virStorageFileGetMetadata(path, &meta);

        if (path != disk->src)
            VIR_FREE(path);
        path = NULL;

        if (rc < 0)
            goto cleanup;

        path = meta.backingStore;
2995
    }
2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040

    ret = 0;

cleanup:
    return ret;
}


static int qemuTeardownDiskCgroup(virCgroupPtr cgroup,
                                  virDomainObjPtr vm,
                                  virDomainDiskDefPtr disk)
{
    char *path = disk->src;
    int ret = -1;

    while (path != NULL) {
        virStorageFileMetadata meta;
        int rc;

        VIR_DEBUG("Process path %s for disk", path);
        rc = virCgroupDenyDevicePath(cgroup, path);
        if (rc != 0) {
            /* Get this for non-block devices */
            if (rc == -EINVAL) {
                VIR_DEBUG("Ignoring EINVAL for %s", path);
            } else {
                virReportSystemError(-rc,
                                     _("Unable to deny device %s for %s"),
                                     path, vm->def->name);
                if (path != disk->src)
                    VIR_FREE(path);
                goto cleanup;
            }
        }

        rc = virStorageFileGetMetadata(path, &meta);

        if (path != disk->src)
            VIR_FREE(path);
        path = NULL;

        if (rc < 0)
            goto cleanup;

        path = meta.backingStore;
3041
    }
3042 3043 3044 3045 3046 3047 3048 3049

    ret = 0;

cleanup:
    return ret;
}


3050
static int qemuSetupCgroup(struct qemud_driver *driver,
3051 3052 3053 3054
                           virDomainObjPtr vm)
{
    virCgroupPtr cgroup = NULL;
    int rc;
3055
    unsigned int i;
3056 3057 3058 3059
    const char *const *deviceACL =
        driver->cgroupDeviceACL ?
        (const char *const *)driver->cgroupDeviceACL :
        defaultDeviceACL;
3060 3061 3062 3063 3064 3065

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

    rc = virCgroupForDomain(driver->cgroup, vm->def->name, &cgroup, 1);
    if (rc != 0) {
3066
        virReportSystemError(-rc,
3067 3068 3069 3070 3071
                             _("Unable to create cgroup for %s"),
                             vm->def->name);
        goto cleanup;
    }

3072 3073
    if (qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_DEVICES)) {
        rc = virCgroupDenyAllDevices(cgroup);
3074
        if (rc != 0) {
3075 3076 3077 3078 3079
            if (rc == -EPERM) {
                VIR_WARN0("Group devices ACL is not accessible, disabling whitelisting");
                goto done;
            }

3080
            virReportSystemError(-rc,
3081
                                 _("Unable to deny all devices for %s"), vm->def->name);
3082 3083 3084
            goto cleanup;
        }

3085
        for (i = 0; i < vm->def->ndisks ; i++) {
3086
            if (qemuSetupDiskCgroup(cgroup, vm, vm->def->disks[i]) < 0)
3087 3088
                goto cleanup;
        }
3089

3090
        rc = virCgroupAllowDeviceMajor(cgroup, 'c', DEVICE_PTY_MAJOR);
3091
        if (rc != 0) {
3092
            virReportSystemError(-rc, "%s",
3093
                                 _("unable to allow /dev/pts/ devices"));
3094 3095 3096
            goto cleanup;
        }

3097 3098 3099
        if (vm->def->nsounds) {
            rc = virCgroupAllowDeviceMajor(cgroup, 'c', DEVICE_SND_MAJOR);
            if (rc != 0) {
3100
                virReportSystemError(-rc, "%s",
3101 3102 3103 3104 3105 3106 3107 3108 3109 3110
                                     _("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) {
3111
                virReportSystemError(-rc,
3112 3113 3114 3115
                                     _("unable to allow device %s"),
                                     deviceACL[i]);
                goto cleanup;
            }
3116 3117 3118 3119
        }
    }

done:
3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131
    virCgroupFree(&cgroup);
    return 0;

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


3132
static int qemuRemoveCgroup(struct qemud_driver *driver,
3133 3134
                            virDomainObjPtr vm,
                            int quiet)
3135 3136 3137 3138 3139 3140 3141 3142 3143
{
    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) {
3144
        if (!quiet)
3145 3146 3147
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("Unable to find cgroup for %s\n"),
                            vm->def->name);
3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167
        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) {
3168
        virReportSystemError(-rc,
3169 3170 3171 3172 3173 3174 3175
                             _("unable to find cgroup for domain %s"),
                             def->name);
        goto cleanup;
    }

    rc = virCgroupAddTask(cgroup, getpid());
    if (rc != 0) {
3176
        virReportSystemError(-rc,
3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193
                             _("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;
3194 3195 3196
};

static int qemudSecurityHook(void *data) {
3197 3198
    struct qemudHookData *h = data;

3199 3200 3201
    /* This must take place before exec(), so that all QEMU
     * memory allocation is on the correct NUMA node
     */
3202
    if (qemuAddToCgroup(h->driver, h->vm->def) < 0)
3203 3204
        return -1;

3205 3206 3207
    /* This must be done after cgroup placement to avoid resetting CPU
     * affinity */
    if (qemudInitCpuAffinity(h->vm) < 0)
3208
        return -1;
3209

3210 3211
    if (h->driver->securityDriver &&
        h->driver->securityDriver->domainSetSecurityProcessLabel &&
3212
        h->driver->securityDriver->domainSetSecurityProcessLabel(h->driver->securityDriver, h->vm) < 0)
3213 3214 3215
        return -1;

    return 0;
3216 3217
}

3218
static int
3219
qemuPrepareMonitorChr(struct qemud_driver *driver,
3220
                      virDomainChrDefPtr monConfig,
3221 3222
                      const char *vm)
{
3223
    monConfig->targetType = VIR_DOMAIN_CHR_TARGET_TYPE_MONITOR;
3224

3225 3226
    monConfig->type = VIR_DOMAIN_CHR_TYPE_UNIX;
    monConfig->data.nix.listen = 1;
3227

D
Daniel P. Berrange 已提交
3228
    if (!(monConfig->info.alias = strdup("monitor"))) {
3229
        virReportOOMError();
D
Daniel P. Berrange 已提交
3230 3231 3232
        return -1;
    }

3233
    if (virAsprintf(&monConfig->data.nix.path, "%s/%s.monitor",
3234
                    driver->libDir, vm) < 0) {
3235
        virReportOOMError();
3236 3237 3238 3239 3240 3241
        return -1;
    }

    return 0;
}

C
Chris Lalancette 已提交
3242 3243 3244 3245 3246
static int qemuDomainSnapshotSetActive(virDomainObjPtr vm,
                                       char *snapshotDir);
static int qemuDomainSnapshotSetInactive(virDomainObjPtr vm,
                                         char *snapshotDir);

3247 3248
static int qemudStartVMDaemon(virConnectPtr conn,
                              struct qemud_driver *driver,
3249
                              virDomainObjPtr vm,
3250 3251
                              const char *migrateFrom,
                              int stdin_fd) {
3252
    const char **argv = NULL, **tmp;
3253
    const char **progenv = NULL;
3254
    int i, ret;
3255
    struct stat sb;
3256 3257
    int *vmfds = NULL;
    int nvmfds = 0;
3258
    unsigned long long qemuCmdFlags;
3259
    fd_set keepfd;
3260
    const char *emulator;
G
Guido Günther 已提交
3261
    pid_t child;
3262
    int pos = -1;
3263
    char ebuf[1024];
3264
    char *pidfile = NULL;
3265
    int logfile = -1;
3266
    qemuDomainObjPrivatePtr priv = vm->privateData;
3267

3268
    struct qemudHookData hookData;
3269 3270 3271 3272
    hookData.conn = conn;
    hookData.vm = vm;
    hookData.driver = driver;

3273
    FD_ZERO(&keepfd);
3274

3275 3276
    DEBUG0("Beginning VM startup process");

D
Daniel P. Berrange 已提交
3277
    if (virDomainObjIsActive(vm)) {
3278 3279
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("VM is already active"));
3280 3281 3282
        return -1;
    }

3283 3284 3285 3286 3287
    /* Must be run before security labelling */
    DEBUG0("Preparing host devices");
    if (qemuPrepareHostDevices(driver, vm->def) < 0)
        goto cleanup;

3288 3289
    /* If you are using a SecurityDriver with dynamic labelling,
       then generate a security label for isolation */
3290
    DEBUG0("Generating domain security label (if required)");
3291
    if (driver->securityDriver &&
3292
        driver->securityDriver->domainGenSecurityLabel &&
3293
        driver->securityDriver->domainGenSecurityLabel(vm) < 0)
3294
        goto cleanup;
3295

3296
    DEBUG0("Generating setting domain security labels (if required)");
3297 3298
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityAllLabel &&
3299
        driver->securityDriver->domainSetSecurityAllLabel(vm) < 0)
3300 3301
        goto cleanup;

3302 3303 3304
    /* Ensure no historical cgroup for this VM is lying around bogus
     * settings */
    DEBUG0("Ensuring no historical cgroup is lying around");
3305
    qemuRemoveCgroup(driver, vm, 1);
3306

3307 3308 3309
    if ((vm->def->ngraphics == 1) &&
        vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
        vm->def->graphics[0]->data.vnc.autoport) {
3310
        DEBUG0("Determining VNC port");
3311
        int port = qemudNextFreeVNCPort(driver);
3312
        if (port < 0) {
3313 3314
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            "%s", _("Unable to find an unused VNC port"));
3315
            goto cleanup;
3316
        }
3317
        vm->def->graphics[0]->data.vnc.port = port;
3318
    }
3319

L
Laine Stump 已提交
3320
    if (virFileMakePath(driver->logDir) != 0) {
3321
        virReportSystemError(errno,
3322 3323
                             _("cannot create log directory %s"),
                             driver->logDir);
3324
        goto cleanup;
3325 3326
    }

3327
    DEBUG0("Creating domain log file");
3328
    if ((logfile = qemudLogFD(driver, vm->def->name)) < 0)
3329
        goto cleanup;
3330

3331 3332
    emulator = vm->def->emulator;

3333 3334 3335 3336
    /* 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
     */
3337
    if (stat(emulator, &sb) < 0) {
3338
        virReportSystemError(errno,
3339 3340
                             _("Cannot find QEMU binary %s"),
                             emulator);
3341
        goto cleanup;
3342 3343
    }

3344
    DEBUG0("Determing emulator version");
3345
    if (qemudExtractVersionInfo(emulator,
3346
                                NULL,
3347
                                &qemuCmdFlags) < 0)
3348
        goto cleanup;
3349

3350
    DEBUG0("Setting up domain cgroup (if required)");
3351
    if (qemuSetupCgroup(driver, vm) < 0)
3352 3353
        goto cleanup;

3354
    if (VIR_ALLOC(priv->monConfig) < 0) {
3355
        virReportOOMError();
3356 3357 3358
        goto cleanup;
    }

3359
    DEBUG0("Preparing monitor state");
3360
    if (qemuPrepareMonitorChr(driver, priv->monConfig, vm->def->name) < 0)
3361
        goto cleanup;
3362

D
Daniel P. Berrange 已提交
3363 3364 3365
#if HAVE_YAJL
    if (qemuCmdFlags & QEMUD_CMD_FLAG_MONITOR_JSON)
        priv->monJSON = 1;
3366
    else
D
Daniel P. Berrange 已提交
3367
#endif
3368
        priv->monJSON = 0;
D
Daniel P. Berrange 已提交
3369

D
Daniel P. Berrange 已提交
3370
    if ((ret = virFileDeletePid(driver->stateDir, vm->def->name)) != 0) {
3371
        virReportSystemError(ret,
D
Daniel P. Berrange 已提交
3372 3373 3374 3375 3376
                             _("Cannot remove stale PID file for %s"),
                             vm->def->name);
        goto cleanup;
    }

3377
    if (!(pidfile = virFilePid(driver->stateDir, vm->def->name))) {
3378
        virReportSystemError(errno,
3379 3380 3381
                             "%s", _("Failed to build pidfile path."));
        goto cleanup;
    }
D
Daniel P. Berrange 已提交
3382

3383
    /*
M
Matthew Booth 已提交
3384
     * Normally PCI addresses are assigned in the virDomainCreate
3385 3386 3387 3388 3389
     * 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
     */
3390
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
3391
        DEBUG0("Assigning domain PCI addresses");
3392
        /* Populate cache with current addresses */
3393 3394 3395 3396 3397 3398 3399
        if (priv->pciaddrs) {
            qemuDomainPCIAddressSetFree(priv->pciaddrs);
            priv->pciaddrs = NULL;
        }
        if (!(priv->pciaddrs = qemuDomainPCIAddressSetCreate(vm->def)))
            goto cleanup;

3400 3401

        /* Assign any remaining addresses */
3402 3403
        if (qemuAssignDevicePCISlots(vm->def, priv->pciaddrs) < 0)
            goto cleanup;
3404 3405 3406 3407

        priv->persistentAddrs = 1;
    } else {
        priv->persistentAddrs = 0;
3408 3409
    }

3410
    DEBUG0("Building emulator command line");
3411
    vm->def->id = driver->nextvmid++;
3412
    if (qemudBuildCommandLine(conn, driver, vm->def, priv->monConfig,
D
Daniel P. Berrange 已提交
3413
                              priv->monJSON, qemuCmdFlags, &argv, &progenv,
3414
                              &vmfds, &nvmfds, migrateFrom,
C
Chris Lalancette 已提交
3415 3416 3417 3418
                              vm->current_snapshot) < 0)
        goto cleanup;

    if (qemuDomainSnapshotSetInactive(vm, driver->snapshotDir) < 0)
3419
        goto cleanup;
3420

3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435 3436
    /* 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;
    }

3437 3438
    tmp = progenv;
    while (*tmp) {
3439
        if (safewrite(logfile, *tmp, strlen(*tmp)) < 0)
3440
            VIR_WARN("Unable to write envv to logfile: %s",
3441
                     virStrerror(errno, ebuf, sizeof ebuf));
3442
        if (safewrite(logfile, " ", 1) < 0)
3443
            VIR_WARN("Unable to write envv to logfile: %s",
3444
                     virStrerror(errno, ebuf, sizeof ebuf));
3445 3446
        tmp++;
    }
3447 3448
    tmp = argv;
    while (*tmp) {
3449
        if (safewrite(logfile, *tmp, strlen(*tmp)) < 0)
3450
            VIR_WARN("Unable to write argv to logfile: %s",
3451
                     virStrerror(errno, ebuf, sizeof ebuf));
3452
        if (safewrite(logfile, " ", 1) < 0)
3453
            VIR_WARN("Unable to write argv to logfile: %s",
3454
                     virStrerror(errno, ebuf, sizeof ebuf));
3455 3456
        tmp++;
    }
3457
    if (safewrite(logfile, "\n", 1) < 0)
3458
        VIR_WARN("Unable to write argv to logfile: %s",
3459
                 virStrerror(errno, ebuf, sizeof ebuf));
3460

3461
    if ((pos = lseek(logfile, 0, SEEK_END)) < 0)
3462
        VIR_WARN("Unable to seek to end of logfile: %s",
3463
                 virStrerror(errno, ebuf, sizeof ebuf));
3464

3465 3466
    for (i = 0 ; i < nvmfds ; i++)
        FD_SET(vmfds[i], &keepfd);
3467

3468
    ret = virExecDaemonize(argv, progenv, &keepfd, &child,
3469
                           stdin_fd, &logfile, &logfile,
3470
                           VIR_EXEC_NONBLOCK | VIR_EXEC_CLEAR_CAPS,
3471 3472 3473
                           qemudSecurityHook, &hookData,
                           pidfile);
    VIR_FREE(pidfile);
G
Guido Günther 已提交
3474 3475 3476

    /* wait for qemu process to to show up */
    if (ret == 0) {
3477
        if (virFileReadPid(driver->stateDir, vm->def->name, &vm->pid)) {
3478 3479
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("Domain %s didn't show up\n"), vm->def->name);
3480
            ret = -1;
G
Guido Günther 已提交
3481
        }
3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492
    } 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;
    }
3493 3494

    vm->state = migrateFrom ? VIR_DOMAIN_PAUSED : VIR_DOMAIN_RUNNING;
3495

3496
    for (i = 0 ; argv[i] ; i++)
3497 3498
        VIR_FREE(argv[i]);
    VIR_FREE(argv);
3499

3500 3501 3502 3503
    for (i = 0 ; progenv[i] ; i++)
        VIR_FREE(progenv[i]);
    VIR_FREE(progenv);

3504 3505 3506
    if (ret == -1) /* The VM failed to start; tear filters before taps */
        virNWFilterTearVMNWFilters(vm);

3507 3508 3509
    if (vmfds) {
        for (i = 0 ; i < nvmfds ; i++) {
            close(vmfds[i]);
3510
        }
3511
        VIR_FREE(vmfds);
3512 3513
    }

3514
    if (ret == -1) /* The VM failed to start */
3515 3516
        goto cleanup;

3517
    DEBUG0("Waiting for monitor to show up");
3518
    if (qemudWaitForMonitor(driver, vm, pos) < 0)
3519 3520
        goto abort;

3521
    DEBUG0("Detecting VCPU PIDs");
3522
    if (qemuDetectVcpuPIDs(driver, vm) < 0)
3523 3524
        goto abort;

3525
    DEBUG0("Setting any required VM passwords");
3526
    if (qemuInitPasswords(conn, driver, vm, qemuCmdFlags) < 0)
3527 3528
        goto abort;

D
Daniel P. Berrange 已提交
3529 3530 3531
    /* 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)) {
3532
        DEBUG0("Determining domain device PCI addresses");
D
Daniel P. Berrange 已提交
3533 3534 3535
        if (qemuInitPCIAddresses(driver, vm) < 0)
            goto abort;
    }
3536

3537
    DEBUG0("Setting initial memory amount");
3538
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
3539
    if (qemuMonitorSetBalloon(priv->mon, vm->def->memory) < 0) {
3540
        qemuDomainObjExitMonitorWithDriver(driver, vm);
3541
        goto abort;
3542
    }
3543 3544

    if (migrateFrom == NULL) {
3545
        DEBUG0("Starting domain CPUs");
3546 3547 3548
        /* Allow the CPUS to start executing */
        if (qemuMonitorStartCPUs(priv->mon, conn) < 0) {
            if (virGetLastError() == NULL)
3549 3550
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                "%s", _("resume operation failed"));
3551 3552 3553 3554
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            goto abort;
        }
    }
3555
    qemuDomainObjExitMonitorWithDriver(driver, vm);
3556

3557

3558
    DEBUG0("Writing domain status to disk");
3559
    if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
3560
        goto abort;
3561

3562 3563 3564
    if (logfile != -1)
        close(logfile);

3565
    return 0;
3566 3567

cleanup:
3568 3569 3570
    /* We jump here if we failed to start the VM for any reason
     * XXX investigate if we can kill this block and safely call
     * qemudShutdownVMDaemon even though no PID is running */
3571
    qemuDomainReAttachHostDevices(driver, vm->def);
3572

3573 3574
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityAllLabel)
3575
        driver->securityDriver->domainRestoreSecurityAllLabel(vm, 0);
3576 3577
    if (driver->securityDriver &&
        driver->securityDriver->domainReleaseSecurityLabel)
3578
        driver->securityDriver->domainReleaseSecurityLabel(vm);
3579
    qemuRemoveCgroup(driver, vm, 1);
3580 3581
    if (logfile != -1)
        close(logfile);
3582 3583
    vm->def->id = -1;
    return -1;
3584 3585 3586 3587

abort:
    /* We jump here if we failed to initialize the now running VM
     * killing it off and pretend we never started it */
3588
    qemudShutdownVMDaemon(driver, vm, 0);
3589 3590 3591 3592 3593

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

    return -1;
3594 3595 3596
}


3597
static void qemudShutdownVMDaemon(struct qemud_driver *driver,
3598 3599
                                  virDomainObjPtr vm,
                                  int migrated) {
D
Daniel P. Berrange 已提交
3600
    int ret;
3601
    int retries = 0;
3602
    qemuDomainObjPrivatePtr priv = vm->privateData;
3603
    virErrorPtr orig_err;
3604 3605
    virDomainDefPtr def;
    int i;
D
Daniel P. Berrange 已提交
3606

D
Daniel P. Berrange 已提交
3607
    if (!virDomainObjIsActive(vm))
3608
        return;
3609

3610
    VIR_DEBUG("Shutting down VM '%s' migrated=%d", vm->def->name, migrated);
3611

3612 3613 3614 3615
    /* This method is routinely used in clean up paths. Disable error
     * reporting so we don't squash a legit error. */
    orig_err = virSaveLastError();

3616
    virNWFilterTearVMNWFilters(vm);
S
Stefan Berger 已提交
3617

3618
    if (driver->macFilter) {
3619
        def = vm->def;
3620 3621 3622 3623
        for (i = 0 ; i < def->nnets ; i++) {
            virDomainNetDefPtr net = def->nets[i];
            if (net->ifname == NULL)
                continue;
3624
            if ((errno = networkDisallowMacOnPort(driver, net->ifname,
3625
                                                  net->mac))) {
3626
                virReportSystemError(errno,
3627 3628 3629 3630 3631 3632
             _("failed to remove ebtables rule to allow MAC address on  '%s'"),
                                     net->ifname);
            }
        }
    }

G
Guido Günther 已提交
3633 3634
    if (virKillProcess(vm->pid, 0) == 0 &&
        virKillProcess(vm->pid, SIGTERM) < 0)
3635
        virReportSystemError(errno,
3636 3637
                             _("Failed to send SIGTERM to %s (%d)"),
                             vm->def->name, vm->pid);
3638

3639 3640 3641
    if (priv->mon &&
        qemuMonitorClose(priv->mon) == 0) {
        virDomainObjUnref(vm);
3642
        priv->mon = NULL;
3643
    }
3644

3645 3646 3647 3648 3649
    if (priv->monConfig) {
        if (priv->monConfig->type == VIR_DOMAIN_CHR_TYPE_UNIX)
            unlink(priv->monConfig->data.nix.path);
        virDomainChrDefFree(priv->monConfig);
        priv->monConfig = NULL;
3650 3651
    }

G
Guido Günther 已提交
3652 3653
    /* shut it off for sure */
    virKillProcess(vm->pid, SIGKILL);
3654

3655 3656 3657 3658 3659 3660 3661 3662 3663 3664
    /* 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);
    }

3665
    /* Reset Security Labels */
3666
    if (driver->securityDriver &&
3667
        driver->securityDriver->domainRestoreSecurityAllLabel)
3668
        driver->securityDriver->domainRestoreSecurityAllLabel(vm, migrated);
3669 3670
    if (driver->securityDriver &&
        driver->securityDriver->domainReleaseSecurityLabel)
3671
        driver->securityDriver->domainReleaseSecurityLabel(vm);
3672

3673 3674 3675 3676 3677 3678 3679
    /* 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
3680
    virDomainDefClearDeviceAliases(vm->def);
3681 3682 3683 3684 3685
    if (!priv->persistentAddrs) {
        virDomainDefClearPCIAddresses(vm->def);
        qemuDomainPCIAddressSetFree(priv->pciaddrs);
        priv->pciaddrs = NULL;
    }
3686

3687
    qemuDomainReAttachHostDevices(driver, vm->def);
3688

S
Stefan Berger 已提交
3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699
#if WITH_MACVTAP
    def = vm->def;
    for (i = 0; i < def->nnets; i++) {
        virDomainNetDefPtr net = def->nets[i];
        if (net->type == VIR_DOMAIN_NET_TYPE_DIRECT) {
            if (net->ifname)
                delMacvtap(net->ifname);
        }
    }
#endif

3700
retry:
3701
    if ((ret = qemuRemoveCgroup(driver, vm, 0)) < 0) {
3702 3703 3704 3705 3706 3707 3708 3709
        if (ret == -EBUSY && (retries++ < 5)) {
            usleep(200*1000);
            goto retry;
        }
        VIR_WARN("Failed to remove cgroup for %s",
                 vm->def->name);
    }

3710
    qemudRemoveDomainStatus(driver, vm);
D
Daniel P. Berrange 已提交
3711

3712
    vm->pid = -1;
3713
    vm->def->id = -1;
3714
    vm->state = VIR_DOMAIN_SHUTOFF;
3715 3716
    VIR_FREE(priv->vcpupids);
    priv->nvcpupids = 0;
3717 3718

    if (vm->newDef) {
3719
        virDomainDefFree(vm->def);
3720
        vm->def = vm->newDef;
3721
        vm->def->id = -1;
3722 3723
        vm->newDef = NULL;
    }
3724 3725 3726 3727 3728

    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
3729 3730 3731
}


3732
static virDrvOpenStatus qemudOpen(virConnectPtr conn,
3733
                                  virConnectAuthPtr auth ATTRIBUTE_UNUSED,
3734
                                  int flags ATTRIBUTE_UNUSED) {
3735
    if (conn->uri == NULL) {
3736 3737 3738
        if (qemu_driver == NULL)
            return VIR_DRV_OPEN_DECLINED;

3739
        conn->uri = xmlParseURI(qemu_driver->privileged ?
3740 3741
                                "qemu:///system" :
                                "qemu:///session");
3742
        if (!conn->uri) {
3743
            virReportOOMError();
3744 3745
            return VIR_DRV_OPEN_ERROR;
        }
3746 3747 3748 3749 3750 3751 3752 3753 3754 3755
    } 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;

3756
        if (qemu_driver == NULL) {
3757 3758
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("qemu state driver is not active"));
3759 3760 3761
            return VIR_DRV_OPEN_ERROR;
        }

3762
        if (conn->uri->path == NULL) {
3763 3764 3765 3766 3767
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("no QEMU URI path given, try %s"),
                            qemu_driver->privileged
                            ? "qemu:///system"
                            : "qemu:///session");
3768 3769 3770
                return VIR_DRV_OPEN_ERROR;
        }

3771
        if (qemu_driver->privileged) {
3772 3773
            if (STRNEQ (conn->uri->path, "/system") &&
                STRNEQ (conn->uri->path, "/session")) {
3774 3775 3776
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("unexpected QEMU URI path '%s', try qemu:///system"),
                                conn->uri->path);
3777 3778 3779 3780
                return VIR_DRV_OPEN_ERROR;
            }
        } else {
            if (STRNEQ (conn->uri->path, "/session")) {
3781 3782 3783
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("unexpected QEMU URI path '%s', try qemu:///session"),
                                conn->uri->path);
3784 3785 3786
                return VIR_DRV_OPEN_ERROR;
            }
        }
3787 3788 3789 3790 3791 3792 3793
    }
    conn->privateData = qemu_driver;

    return VIR_DRV_OPEN_SUCCESS;
}

static int qemudClose(virConnectPtr conn) {
3794
    struct qemud_driver *driver = conn->privateData;
3795 3796

    /* Get rid of callbacks registered for this conn */
3797
    qemuDriverLock(driver);
3798
    virDomainEventCallbackListRemoveConn(conn, driver->domainEventCallbacks);
3799
    qemuDriverUnlock(driver);
3800 3801 3802 3803 3804 3805

    conn->privateData = NULL;

    return 0;
}

D
Daniel Veillard 已提交
3806 3807 3808 3809 3810
/* Which features are supported by this driver? */
static int
qemudSupportsFeature (virConnectPtr conn ATTRIBUTE_UNUSED, int feature)
{
    switch (feature) {
3811 3812 3813 3814 3815
    case VIR_DRV_FEATURE_MIGRATION_V2:
    case VIR_DRV_FEATURE_MIGRATION_P2P:
        return 1;
    default:
        return 0;
D
Daniel Veillard 已提交
3816 3817 3818
    }
}

3819
static const char *qemudGetType(virConnectPtr conn ATTRIBUTE_UNUSED) {
3820
    return "QEMU";
3821 3822
}

3823

3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836
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;
}


3837 3838 3839 3840
static int kvmGetMaxVCPUs(void) {
    int maxvcpus = 1;

    int r, fd;
3841

3842 3843
    fd = open(KVM_DEVICE, O_RDONLY);
    if (fd < 0) {
3844
        virReportSystemError(errno, _("Unable to open %s"), KVM_DEVICE);
3845
        return -1;
3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856
    }

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

    close(fd);
    return maxvcpus;
}


3857
static int qemudGetMaxVCPUs(virConnectPtr conn ATTRIBUTE_UNUSED, const char *type) {
3858 3859 3860
    if (!type)
        return 16;

3861
    if (STRCASEEQ(type, "qemu"))
3862 3863
        return 16;

3864
    if (STRCASEEQ(type, "kvm"))
3865
        return kvmGetMaxVCPUs();
3866

3867
    if (STRCASEEQ(type, "kqemu"))
3868
        return 1;
3869

3870 3871
    qemuReportError(VIR_ERR_INVALID_ARG,
                    _("unknown type '%s'"), type);
3872 3873 3874
    return -1;
}

3875

3876
static char *qemudGetCapabilities(virConnectPtr conn) {
3877
    struct qemud_driver *driver = conn->privateData;
3878
    virCapsPtr caps = NULL;
3879
    char *xml = NULL;
3880

3881
    qemuDriverLock(driver);
3882

3883
    if ((caps = qemuCreateCapabilities(qemu_driver->caps,
3884
                                       qemu_driver)) == NULL) {
3885 3886 3887
        virCapabilitiesFree(caps);
        goto cleanup;
    }
3888

3889
    virCapabilitiesFree(qemu_driver->caps);
3890 3891 3892
    qemu_driver->caps = caps;

    if ((xml = virCapabilitiesFormatXML(driver->caps)) == NULL)
3893
        virReportOOMError();
3894 3895

cleanup:
3896
    qemuDriverUnlock(driver);
3897

3898
    return xml;
3899 3900 3901
}


3902
static int qemudGetProcessInfo(unsigned long long *cpuTime, int *lastCpu, int pid, int tid) {
D
Daniel P. Berrange 已提交
3903 3904
    char proc[PATH_MAX];
    FILE *pidinfo;
3905
    unsigned long long usertime, systime;
3906 3907
    int cpu;
    int ret;
D
Daniel P. Berrange 已提交
3908

3909 3910 3911 3912 3913 3914
    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
3915 3916 3917 3918
        return -1;
    }

    if (!(pidinfo = fopen(proc, "r"))) {
3919
        /*printf("cannot read pid info");*/
D
Daniel P. Berrange 已提交
3920
        /* VM probably shut down, so fake 0 */
3921 3922 3923 3924
        if (cpuTime)
            *cpuTime = 0;
        if (lastCpu)
            *lastCpu = 0;
D
Daniel P. Berrange 已提交
3925 3926 3927
        return 0;
    }

3928 3929 3930 3931 3932 3933 3934 3935 3936 3937
    /* 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) {
3938
        fclose(pidinfo);
3939 3940
        VIR_WARN0("cannot parse process status data");
        errno = -EINVAL;
D
Daniel P. Berrange 已提交
3941 3942 3943 3944 3945 3946 3947 3948
        return -1;
    }

    /* We got jiffies
     * We want nanoseconds
     * _SC_CLK_TCK is jiffies per second
     * So calulate thus....
     */
3949 3950 3951 3952 3953
    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
3954

3955 3956
    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
3957 3958 3959 3960 3961 3962 3963

    fclose(pidinfo);

    return 0;
}


3964
static virDomainPtr qemudDomainLookupByID(virConnectPtr conn,
3965
                                          int id) {
3966 3967 3968 3969
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;

3970
    qemuDriverLock(driver);
3971
    vm  = virDomainFindByID(&driver->domains, id);
3972
    qemuDriverUnlock(driver);
3973 3974

    if (!vm) {
3975 3976
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching id %d"), id);
3977
        goto cleanup;
3978 3979
    }

3980
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
3981
    if (dom) dom->id = vm->def->id;
3982 3983

cleanup:
3984 3985
    if (vm)
        virDomainObjUnlock(vm);
3986 3987
    return dom;
}
3988

3989
static virDomainPtr qemudDomainLookupByUUID(virConnectPtr conn,
3990
                                            const unsigned char *uuid) {
3991 3992 3993
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
3994

3995
    qemuDriverLock(driver);
3996
    vm = virDomainFindByUUID(&driver->domains, uuid);
3997 3998
    qemuDriverUnlock(driver);

3999
    if (!vm) {
4000 4001
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(uuid, uuidstr);
4002 4003
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4004
        goto cleanup;
4005 4006
    }

4007
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
4008
    if (dom) dom->id = vm->def->id;
4009 4010

cleanup:
4011 4012
    if (vm)
        virDomainObjUnlock(vm);
4013 4014
    return dom;
}
4015

4016
static virDomainPtr qemudDomainLookupByName(virConnectPtr conn,
4017
                                            const char *name) {
4018 4019 4020
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
4021

4022
    qemuDriverLock(driver);
4023
    vm = virDomainFindByName(&driver->domains, name);
4024 4025
    qemuDriverUnlock(driver);

4026
    if (!vm) {
4027 4028
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching name '%s'"), name);
4029
        goto cleanup;
4030 4031
    }

4032
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
4033
    if (dom) dom->id = vm->def->id;
4034 4035

cleanup:
4036 4037
    if (vm)
        virDomainObjUnlock(vm);
4038 4039 4040
    return dom;
}

4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051

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) {
4052
        qemuReportError(VIR_ERR_NO_DOMAIN, NULL);
4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064 4065 4066 4067 4068 4069 4070 4071 4072
        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) {
4073
        qemuReportError(VIR_ERR_NO_DOMAIN, NULL);
4074 4075 4076 4077 4078 4079 4080 4081 4082 4083 4084
        goto cleanup;
    }
    ret = obj->persistent;

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


4085
static int qemudGetVersion(virConnectPtr conn, unsigned long *version) {
4086 4087 4088
    struct qemud_driver *driver = conn->privateData;
    int ret = -1;

4089
    qemuDriverLock(driver);
4090
    if (qemudExtractVersion(driver) < 0)
4091
        goto cleanup;
4092

4093
    *version = qemu_driver->qemuVersion;
4094 4095 4096
    ret = 0;

cleanup:
4097
    qemuDriverUnlock(driver);
4098
    return ret;
D
Daniel P. Berrange 已提交
4099 4100
}

4101
static int qemudListDomains(virConnectPtr conn, int *ids, int nids) {
4102
    struct qemud_driver *driver = conn->privateData;
4103
    int n;
4104

4105
    qemuDriverLock(driver);
4106
    n = virDomainObjListGetActiveIDs(&driver->domains, ids, nids);
4107
    qemuDriverUnlock(driver);
4108

4109
    return n;
D
Daniel P. Berrange 已提交
4110
}
4111

4112
static int qemudNumDomains(virConnectPtr conn) {
4113
    struct qemud_driver *driver = conn->privateData;
4114
    int n;
4115

4116
    qemuDriverLock(driver);
4117
    n = virDomainObjListNumOfDomains(&driver->domains, 1);
4118
    qemuDriverUnlock(driver);
4119

4120
    return n;
D
Daniel P. Berrange 已提交
4121
}
4122

4123
static virDomainPtr qemudDomainCreate(virConnectPtr conn, const char *xml,
4124
                                      unsigned int flags ATTRIBUTE_UNUSED) {
4125
    struct qemud_driver *driver = conn->privateData;
4126
    virDomainDefPtr def;
4127
    virDomainObjPtr vm = NULL;
4128
    virDomainPtr dom = NULL;
4129
    virDomainEventPtr event = NULL;
D
Daniel P. Berrange 已提交
4130

4131
    qemuDriverLock(driver);
4132
    if (!(def = virDomainDefParseString(driver->caps, xml,
4133
                                        VIR_DOMAIN_XML_INACTIVE)))
4134
        goto cleanup;
4135

4136
    if (virSecurityDriverVerify(def) < 0)
4137 4138
        goto cleanup;

4139 4140
    if (virDomainObjIsDuplicate(&driver->domains, def, 1) < 0)
        goto cleanup;
4141

4142 4143 4144 4145 4146 4147
    if (qemudCanonicalizeMachine(driver, def) < 0)
        goto cleanup;

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

4148
    if (!(vm = virDomainAssignDef(driver->caps,
4149
                                  &driver->domains,
4150
                                  def, false)))
4151 4152 4153
        goto cleanup;

    def = NULL;
D
Daniel P. Berrange 已提交
4154

4155 4156 4157
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup; /* XXXX free the 'vm' we created ? */

4158
    if (qemudStartVMDaemon(conn, driver, vm, NULL, -1) < 0) {
4159 4160 4161
        if (qemuDomainObjEndJob(vm) > 0)
            virDomainRemoveInactive(&driver->domains,
                                    vm);
4162
        vm = NULL;
4163
        goto cleanup;
D
Daniel P. Berrange 已提交
4164
    }
4165 4166 4167 4168

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_BOOTED);
D
Daniel P. Berrange 已提交
4169

4170
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
4171
    if (dom) dom->id = vm->def->id;
4172

4173 4174 4175
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4176

4177 4178
cleanup:
    virDomainDefFree(def);
4179 4180
    if (vm)
        virDomainObjUnlock(vm);
4181 4182
    if (event)
        qemuDomainEventQueue(driver, event);
4183
    qemuDriverUnlock(driver);
4184
    return dom;
D
Daniel P. Berrange 已提交
4185 4186 4187
}


4188
static int qemudDomainSuspend(virDomainPtr dom) {
4189 4190 4191
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4192
    virDomainEventPtr event = NULL;
4193
    qemuDomainObjPrivatePtr priv;
4194

4195
    qemuDriverLock(driver);
4196
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4197

D
Daniel P. Berrange 已提交
4198
    if (!vm) {
4199 4200
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4201 4202
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4203
        goto cleanup;
D
Daniel P. Berrange 已提交
4204
    }
D
Daniel P. Berrange 已提交
4205
    if (!virDomainObjIsActive(vm)) {
4206 4207
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
4208
        goto cleanup;
D
Daniel P. Berrange 已提交
4209
    }
4210 4211 4212 4213 4214 4215 4216 4217 4218 4219 4220 4221 4222 4223 4224 4225 4226 4227

    priv = vm->privateData;

    if (priv->jobActive == QEMU_JOB_MIGRATION) {
        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"));
4228
            goto endjob;
4229
        }
4230 4231
        if (vm->state != VIR_DOMAIN_PAUSED) {
            int rc;
4232
            int state = vm->state;
4233

4234
            vm->state = VIR_DOMAIN_PAUSED;
4235 4236 4237
            qemuDomainObjEnterMonitorWithDriver(driver, vm);
            rc = qemuMonitorStopCPUs(priv->mon);
            qemuDomainObjExitMonitorWithDriver(driver, vm);
4238 4239
            if (rc < 0) {
                vm->state = state;
4240
                goto endjob;
4241
            }
4242 4243 4244 4245 4246 4247 4248
            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
4249
    }
4250

4251
endjob:
4252 4253
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4254

4255
cleanup:
4256 4257
    if (vm)
        virDomainObjUnlock(vm);
4258

4259
    if (event)
4260
        qemuDomainEventQueue(driver, event);
4261
    qemuDriverUnlock(driver);
4262
    return ret;
D
Daniel P. Berrange 已提交
4263 4264 4265
}


4266
static int qemudDomainResume(virDomainPtr dom) {
4267 4268 4269
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4270
    virDomainEventPtr event = NULL;
4271

4272
    qemuDriverLock(driver);
4273
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4274

D
Daniel P. Berrange 已提交
4275
    if (!vm) {
4276 4277
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4278 4279
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4280
        goto cleanup;
D
Daniel P. Berrange 已提交
4281
    }
4282 4283 4284 4285

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

D
Daniel P. Berrange 已提交
4286
    if (!virDomainObjIsActive(vm)) {
4287 4288
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
4289
        goto endjob;
D
Daniel P. Berrange 已提交
4290
    }
4291
    if (vm->state == VIR_DOMAIN_PAUSED) {
4292
        qemuDomainObjPrivatePtr priv = vm->privateData;
4293
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
4294
        if (qemuMonitorStartCPUs(priv->mon, dom->conn) < 0) {
4295
            qemuDomainObjExitMonitorWithDriver(driver, vm);
4296
            if (virGetLastError() == NULL)
4297 4298
                qemuReportError(VIR_ERR_OPERATION_FAILED,
                                "%s", _("resume operation failed"));
4299
            goto endjob;
4300
        }
4301
        qemuDomainObjExitMonitorWithDriver(driver, vm);
4302
        vm->state = VIR_DOMAIN_RUNNING;
4303 4304 4305
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_UNPAUSED);
D
Daniel P. Berrange 已提交
4306
    }
4307
    if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
4308
        goto endjob;
4309 4310
    ret = 0;

4311
endjob:
4312 4313
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4314

4315
cleanup:
4316 4317
    if (vm)
        virDomainObjUnlock(vm);
4318
    if (event)
4319
        qemuDomainEventQueue(driver, event);
4320
    qemuDriverUnlock(driver);
4321
    return ret;
D
Daniel P. Berrange 已提交
4322 4323 4324
}


4325
static int qemudDomainShutdown(virDomainPtr dom) {
4326 4327 4328
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4329

4330
    qemuDriverLock(driver);
4331
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4332 4333
    qemuDriverUnlock(driver);

4334
    if (!vm) {
4335 4336
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4337 4338
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4339
        goto cleanup;
4340 4341
    }

4342 4343 4344
    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
4345
    if (!virDomainObjIsActive(vm)) {
4346 4347
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
4348
        goto endjob;
4349 4350
    }

4351
    qemuDomainObjPrivatePtr priv = vm->privateData;
4352 4353 4354
    qemuDomainObjEnterMonitor(vm);
    ret = qemuMonitorSystemPowerdown(priv->mon);
    qemuDomainObjExitMonitor(vm);
4355

4356
endjob:
4357 4358
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4359

4360
cleanup:
4361 4362
    if (vm)
        virDomainObjUnlock(vm);
4363
    return ret;
4364 4365 4366
}


4367
static int qemudDomainDestroy(virDomainPtr dom) {
4368 4369 4370
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4371
    virDomainEventPtr event = NULL;
4372

4373
    qemuDriverLock(driver);
4374
    vm  = virDomainFindByUUID(&driver->domains, dom->uuid);
D
Daniel P. Berrange 已提交
4375
    if (!vm) {
4376 4377
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4378 4379
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4380
        goto cleanup;
D
Daniel P. Berrange 已提交
4381
    }
4382 4383 4384 4385

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

D
Daniel P. Berrange 已提交
4386
    if (!virDomainObjIsActive(vm)) {
4387 4388
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
4389
        goto endjob;
4390
    }
4391

4392
    qemudShutdownVMDaemon(driver, vm, 0);
4393 4394 4395
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_DESTROYED);
4396
    if (!vm->persistent) {
4397 4398 4399
        if (qemuDomainObjEndJob(vm) > 0)
            virDomainRemoveInactive(&driver->domains,
                                    vm);
4400 4401
        vm = NULL;
    }
4402 4403
    ret = 0;

4404
endjob:
4405 4406 4407
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4408

4409
cleanup:
4410 4411
    if (vm)
        virDomainObjUnlock(vm);
4412 4413
    if (event)
        qemuDomainEventQueue(driver, event);
4414
    qemuDriverUnlock(driver);
4415
    return ret;
D
Daniel P. Berrange 已提交
4416 4417 4418
}


4419
static char *qemudDomainGetOSType(virDomainPtr dom) {
4420 4421 4422
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    char *type = NULL;
4423

4424
    qemuDriverLock(driver);
4425
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4426
    qemuDriverUnlock(driver);
4427
    if (!vm) {
4428 4429
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4430 4431
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4432
        goto cleanup;
4433 4434
    }

4435
    if (!(type = strdup(vm->def->os.type)))
4436
        virReportOOMError();
4437 4438

cleanup:
4439 4440
    if (vm)
        virDomainObjUnlock(vm);
4441 4442 4443
    return type;
}

4444 4445
/* Returns max memory in kb, 0 if error */
static unsigned long qemudDomainGetMaxMemory(virDomainPtr dom) {
4446 4447 4448
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    unsigned long ret = 0;
4449

4450
    qemuDriverLock(driver);
4451
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4452 4453
    qemuDriverUnlock(driver);

4454
    if (!vm) {
4455 4456
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4457 4458
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4459
        goto cleanup;
4460 4461
    }

4462 4463 4464
    ret = vm->def->maxmem;

cleanup:
4465 4466
    if (vm)
        virDomainObjUnlock(vm);
4467
    return ret;
4468 4469 4470
}

static int qemudDomainSetMemory(virDomainPtr dom, unsigned long newmem) {
4471
    struct qemud_driver *driver = dom->conn->privateData;
4472
    qemuDomainObjPrivatePtr priv;
4473
    virDomainObjPtr vm;
4474
    int ret = -1, r;
4475

4476
    qemuDriverLock(driver);
4477
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4478
    qemuDriverUnlock(driver);
4479
    if (!vm) {
4480 4481
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4482 4483
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4484
        goto cleanup;
4485 4486
    }

4487 4488 4489 4490 4491 4492
    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
        goto cleanup;
    }

4493
    if (newmem > vm->def->maxmem) {
4494 4495
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("cannot set memory higher than max memory"));
4496
        goto cleanup;
4497 4498
    }

4499 4500 4501
    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

4502 4503 4504 4505 4506 4507
    priv = vm->privateData;
    qemuDomainObjEnterMonitor(vm);
    r = qemuMonitorSetBalloon(priv->mon, newmem);
    qemuDomainObjExitMonitor(vm);
    if (r < 0)
        goto endjob;
4508

4509 4510 4511 4512 4513
    /* 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;
4514
    }
4515

4516
    ret = 0;
4517
endjob:
4518 4519
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4520

4521
cleanup:
4522 4523
    if (vm)
        virDomainObjUnlock(vm);
4524
    return ret;
4525 4526
}

4527
static int qemudDomainGetInfo(virDomainPtr dom,
4528
                              virDomainInfoPtr info) {
4529 4530 4531
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4532 4533
    int err;
    unsigned long balloon;
4534

4535
    qemuDriverLock(driver);
4536
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4537
    qemuDriverUnlock(driver);
D
Daniel P. Berrange 已提交
4538
    if (!vm) {
4539 4540
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4541 4542
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4543
        goto cleanup;
D
Daniel P. Berrange 已提交
4544 4545
    }

4546
    info->state = vm->state;
D
Daniel P. Berrange 已提交
4547

D
Daniel P. Berrange 已提交
4548
    if (!virDomainObjIsActive(vm)) {
4549
        info->cpuTime = 0;
D
Daniel P. Berrange 已提交
4550
    } else {
4551
        if (qemudGetProcessInfo(&(info->cpuTime), NULL, vm->pid, 0) < 0) {
4552
            qemuReportError(VIR_ERR_OPERATION_FAILED, ("cannot read cputime for domain"));
4553
            goto cleanup;
D
Daniel P. Berrange 已提交
4554 4555 4556
        }
    }

4557
    info->maxMem = vm->def->maxmem;
4558

D
Daniel P. Berrange 已提交
4559
    if (virDomainObjIsActive(vm)) {
4560
        qemuDomainObjPrivatePtr priv = vm->privateData;
4561 4562 4563
        if (!priv->jobActive) {
            if (qemuDomainObjBeginJob(vm) < 0)
                goto cleanup;
4564

4565 4566 4567 4568
            qemuDomainObjEnterMonitor(vm);
            err = qemuMonitorGetBalloonInfo(priv->mon, &balloon);
            qemuDomainObjExitMonitor(vm);
            if (err < 0) {
4569 4570
                if (qemuDomainObjEndJob(vm) == 0)
                    vm = NULL;
4571 4572 4573 4574 4575 4576 4577 4578 4579
                goto cleanup;
            }

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

4580 4581 4582 4583
            if (qemuDomainObjEndJob(vm) == 0) {
                vm = NULL;
                goto cleanup;
            }
4584 4585 4586
        } else {
            info->memory = vm->def->memory;
        }
4587 4588 4589 4590
    } else {
        info->memory = vm->def->memory;
    }

4591
    info->nrVirtCpu = vm->def->vcpus;
4592 4593 4594
    ret = 0;

cleanup:
4595 4596
    if (vm)
        virDomainObjUnlock(vm);
4597
    return ret;
D
Daniel P. Berrange 已提交
4598 4599 4600
}


4601 4602 4603 4604 4605 4606 4607 4608
/** qemuDomainMigrateOffline:
 * Pause domain for non-live migration.
 */
static int
qemuDomainMigrateOffline(struct qemud_driver *driver,
                         virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
4609
    int state = vm->state;
4610 4611
    int ret;

4612
    vm->state = VIR_DOMAIN_PAUSED;
4613 4614 4615 4616 4617 4618 4619 4620 4621 4622 4623 4624
    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);
4625 4626
    } else
        vm->state = state;
4627 4628 4629 4630 4631

    return ret;
}


4632 4633 4634 4635 4636 4637 4638 4639 4640 4641 4642 4643 4644 4645
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 */
4646
        struct timespec ts = { .tv_sec = 0, .tv_nsec = 50 * 1000 * 1000ull };
4647 4648 4649
        struct timeval now;
        int rc;

4650 4651 4652 4653 4654 4655
        if (!virDomainObjIsActive(vm)) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("guest unexpectedly quit during migration"));
            goto cleanup;
        }

4656 4657
        if (priv->jobSignals & QEMU_JOB_SIGNAL_CANCEL) {
            priv->jobSignals ^= QEMU_JOB_SIGNAL_CANCEL;
4658 4659 4660 4661 4662 4663 4664
            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");
            }
4665 4666 4667 4668 4669
        } 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");
4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680
        } 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");
4681 4682
        }

4683 4684 4685 4686 4687 4688 4689 4690 4691
        /* 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;
        }

4692 4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711 4712 4713 4714 4715 4716 4717 4718 4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731 4732 4733 4734 4735 4736 4737 4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758 4759 4760 4761 4762 4763
        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;
}


4764
#define QEMUD_SAVE_MAGIC "LibvirtQemudSave"
4765 4766 4767
#define QEMUD_SAVE_VERSION 2

enum qemud_save_formats {
4768 4769 4770
    QEMUD_SAVE_FORMAT_RAW = 0,
    QEMUD_SAVE_FORMAT_GZIP = 1,
    QEMUD_SAVE_FORMAT_BZIP2 = 2,
4771 4772
    /*
     * Deprecated by xz and never used as part of a release
4773
     * QEMUD_SAVE_FORMAT_LZMA
4774 4775
     */
    QEMUD_SAVE_FORMAT_XZ = 3,
4776
    QEMUD_SAVE_FORMAT_LZOP = 4,
4777 4778 4779
    /* Note: add new members only at the end.
       These values are used in the on-disk format.
       Do not change or re-use numbers. */
4780 4781

    QEMUD_SAVE_FORMAT_LAST
4782
};
4783

4784 4785 4786 4787 4788
VIR_ENUM_DECL(qemudSaveCompression)
VIR_ENUM_IMPL(qemudSaveCompression, QEMUD_SAVE_FORMAT_LAST,
              "raw",
              "gzip",
              "bzip2",
4789 4790
              "xz",
              "lzop")
4791

4792 4793 4794 4795 4796
struct qemud_save_header {
    char magic[sizeof(QEMUD_SAVE_MAGIC)-1];
    int version;
    int xml_len;
    int was_running;
4797 4798
    int compressed;
    int unused[15];
4799 4800
};

4801 4802 4803 4804 4805 4806 4807 4808 4809 4810 4811 4812 4813 4814
struct fileOpHookData {
    virDomainPtr dom;
    const char *path;
    char *xml;
    struct qemud_save_header *header;
};

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)) {
        ret = errno;
        qemuReportError(VIR_ERR_OPERATION_FAILED,
4815 4816
                        _("failed to write header to domain save file '%s'"),
                        hdata->path);
4817 4818 4819 4820 4821 4822 4823 4824 4825 4826 4827 4828 4829 4830
        goto endjob;
    }

    if (safewrite(fd, hdata->xml, hdata->header->xml_len) != hdata->header->xml_len) {
        ret = errno;
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                         _("failed to write xml to '%s'"), hdata->path);
        goto endjob;
    }
endjob:
    return ret;
}


4831 4832
static int qemudDomainSaveFlag(virDomainPtr dom, const char *path,
                               int compressed)
4833
{
4834
    struct qemud_driver *driver = dom->conn->privateData;
4835
    virDomainObjPtr vm = NULL;
4836
    char *xml = NULL;
4837
    struct qemud_save_header header;
4838 4839
    struct fileOpHookData hdata;
    int bypassSecurityDriver = 0;
4840
    int ret = -1;
4841
    int rc;
4842
    virDomainEventPtr event = NULL;
4843
    qemuDomainObjPrivatePtr priv;
4844 4845
    struct stat sb;
    int is_reg = 0;
4846
    unsigned long long offset;
4847
    virCgroupPtr cgroup = NULL;
4848 4849 4850 4851 4852

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

4853
    qemuDriverLock(driver);
4854 4855

    header.compressed = compressed;
4856

4857
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4858

D
Daniel P. Berrange 已提交
4859
    if (!vm) {
4860 4861
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4862 4863
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4864
        goto cleanup;
D
Daniel P. Berrange 已提交
4865
    }
4866
    priv = vm->privateData;
4867

4868 4869 4870
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
4871
    if (!virDomainObjIsActive(vm)) {
4872 4873
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
4874
        goto endjob;
D
Daniel P. Berrange 已提交
4875
    }
4876

4877 4878 4879
    memset(&priv->jobInfo, 0, sizeof(priv->jobInfo));
    priv->jobInfo.type = VIR_DOMAIN_JOB_UNBOUNDED;

4880 4881 4882
    /* Pause */
    if (vm->state == VIR_DOMAIN_RUNNING) {
        header.was_running = 1;
4883
        vm->state = VIR_DOMAIN_PAUSED;
4884
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
4885
        if (qemuMonitorStopCPUs(priv->mon) < 0) {
4886
            qemuDomainObjExitMonitorWithDriver(driver, vm);
4887
            vm->state = VIR_DOMAIN_RUNNING;
4888
            goto endjob;
4889
        }
4890
        qemuDomainObjExitMonitorWithDriver(driver, vm);
4891 4892 4893 4894 4895 4896

        if (!virDomainObjIsActive(vm)) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("guest unexpectedly quit"));
            goto endjob;
        }
4897 4898 4899
    }

    /* Get XML for the domain */
4900
    xml = virDomainDefFormat(vm->def, VIR_DOMAIN_XML_SECURE);
4901
    if (!xml) {
4902 4903
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to get domain xml"));
4904
        goto endjob;
4905 4906 4907
    }
    header.xml_len = strlen(xml) + 1;

4908 4909 4910 4911 4912 4913 4914 4915 4916 4917 4918 4919 4920 4921
    /* 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);
    }


4922 4923 4924 4925 4926
    /* Setup hook data needed by virFileOperation hook function */
    hdata.dom = dom;
    hdata.path = path;
    hdata.xml = xml;
    hdata.header = &header;
4927 4928 4929 4930 4931 4932 4933 4934 4935 4936 4937 4938 4939 4940 4941 4942 4943 4944 4945
    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
     * it by padding the XML string with NULLs */
    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;
    }
4946

4947 4948
    /* Write header to file, followed by XML */

4949
    /* First try creating the file as root */
4950 4951 4952 4953
    if (!is_reg) {
        int fd = open(path, O_WRONLY | O_TRUNC);
        if (fd < 0) {
            virReportSystemError(errno, _("unable to open %s"), path);
4954 4955
            goto endjob;
        }
4956
        if (qemudDomainSaveFileOpHook(fd, &hdata) != 0) {
4957 4958 4959 4960 4961
            close(fd);
            goto endjob;
        }
        if (close(fd) < 0) {
            virReportSystemError(errno, _("unable to close %s"), path);
4962 4963
            goto endjob;
        }
4964 4965 4966 4967 4968 4969 4970 4971 4972 4973 4974 4975 4976 4977 4978 4979 4980 4981 4982
    } else {
        if ((rc = virFileOperation(path, O_CREAT|O_TRUNC|O_WRONLY,
                                  S_IRUSR|S_IWUSR,
                                  getuid(), getgid(),
                                  qemudDomainSaveFileOpHook, &hdata,
                                  0)) != 0) {
            /* If we failed as root, and the error was permission-denied
               (EACCES), assume it's on a network-connected share where
               root access is restricted (eg, root-squashed NFS). If the
               qemu user (driver->user) is non-root, just set a flag to
               bypass security driver shenanigans, and retry the operation
               after doing setuid to qemu user */

            if ((rc != EACCES) ||
                driver->user == getuid()) {
                virReportSystemError(rc, _("Failed to create domain save file '%s'"),
                                     path);
                goto endjob;
            }
4983

4984 4985 4986 4987 4988
#ifdef __linux__
            /* On Linux we can also verify the FS-type of the directory. */
            char *dirpath, *p;
            struct statfs st;
            int statfs_ret;
4989

4990 4991
            if ((dirpath = strdup(path)) == NULL) {
                virReportOOMError();
4992 4993 4994
                goto endjob;
            }

4995 4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009
            do {
                // Try less and less of the path until we get to a
                // directory we can stat. Even if we don't have 'x'
                // permission on any directory in the path on the NFS
                // server (assuming it's NFS), we will be able to stat the
                // mount point, and that will properly tell us if the
                // fstype is NFS.

                if ((p = strrchr(dirpath, '/')) == NULL) {
                    qemuReportError(VIR_ERR_INVALID_ARG,
                                    _("Invalid relative path '%s' for domain save file"),
                                    path);
                    VIR_FREE(dirpath);
                    goto endjob;
                }
5010

5011 5012 5013 5014
                if (p == dirpath)
                    *(p+1) = '\0';
                else
                    *p = '\0';
5015

5016
                statfs_ret = statfs(dirpath, &st);
5017

5018
            } while ((statfs_ret == -1) && (p != dirpath));
5019

5020 5021
            if (statfs_ret == -1) {
                virReportSystemError(errno,
5022 5023 5024
                                     _("Failed to create domain save file "
                                       "'%s': statfs of all elements of path "
                                       "failed"),
5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 5037
                                     path);
                VIR_FREE(dirpath);
                goto endjob;
            }

            if (st.f_type != NFS_SUPER_MAGIC) {
                virReportSystemError(rc,
                                     _("Failed to create domain save file '%s'"
                                       " (fstype of '%s' is 0x%X"),
                                     path, dirpath, (unsigned int) st.f_type);
                VIR_FREE(dirpath);
                goto endjob;
            }
5038 5039 5040
            VIR_FREE(dirpath);
#endif

5041
            /* Retry creating the file as driver->user */
5042

5043 5044 5045 5046 5047 5048
            if ((rc = virFileOperation(path, O_CREAT|O_TRUNC|O_WRONLY,
                                       S_IRUSR|S_IWUSR|S_IRGRP|S_IWGRP,
                                       driver->user, driver->group,
                                       qemudDomainSaveFileOpHook, &hdata,
                                       VIR_FILE_OP_AS_UID)) != 0) {
                virReportSystemError(rc, _("Error from child process creating '%s'"),
5049
                                 path);
5050 5051
                goto endjob;
            }
5052

5053 5054 5055
            /* 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 */
5056

5057 5058
            bypassSecurityDriver = 1;
        }
5059
    }
5060

5061

5062 5063 5064 5065 5066 5067 5068 5069 5070 5071 5072 5073 5074 5075 5076 5077 5078
    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;
        }
    }

5079 5080
    if ((!bypassSecurityDriver) &&
        driver->securityDriver &&
5081
        driver->securityDriver->domainSetSavedStateLabel &&
5082
        driver->securityDriver->domainSetSavedStateLabel(vm, path) == -1)
5083 5084
        goto endjob;

5085 5086
    if (header.compressed == QEMUD_SAVE_FORMAT_RAW) {
        const char *args[] = { "cat", NULL };
M
Matthias Bolte 已提交
5087
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
5088 5089 5090
        rc = qemuMonitorMigrateToFile(priv->mon,
                                      QEMU_MONITOR_MIGRATE_BACKGROUND,
                                      args, path, offset);
M
Matthias Bolte 已提交
5091
        qemuDomainObjExitMonitorWithDriver(driver, vm);
5092
    } else {
5093
        const char *prog = qemudSaveCompressionTypeToString(header.compressed);
5094 5095 5096 5097 5098
        const char *args[] = {
            prog,
            "-c",
            NULL
        };
M
Matthias Bolte 已提交
5099
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
5100 5101 5102
        rc = qemuMonitorMigrateToFile(priv->mon,
                                      QEMU_MONITOR_MIGRATE_BACKGROUND,
                                      args, path, offset);
M
Matthias Bolte 已提交
5103
        qemuDomainObjExitMonitorWithDriver(driver, vm);
5104 5105
    }

5106 5107 5108 5109 5110
    if (rc < 0)
        goto endjob;

    rc = qemuDomainWaitForMigrationComplete(driver, vm);

5111
    if (rc < 0)
5112
        goto endjob;
5113

5114 5115
    if ((!bypassSecurityDriver) &&
        driver->securityDriver &&
5116
        driver->securityDriver->domainRestoreSavedStateLabel &&
5117
        driver->securityDriver->domainRestoreSavedStateLabel(vm, path) == -1)
5118
        VIR_WARN("failed to restore save state label on %s", path);
5119

5120 5121
    if (cgroup != NULL) {
        rc = virCgroupDenyDevicePath(cgroup, path);
5122 5123 5124
        if (rc != 0)
            VIR_WARN("Unable to deny device %s for %s %d",
                     path, vm->def->name, rc);
5125 5126
    }

5127 5128
    ret = 0;

5129
    /* Shut it down */
5130
    qemudShutdownVMDaemon(driver, vm, 0);
5131 5132 5133
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_SAVED);
5134
    if (!vm->persistent) {
5135 5136 5137
        if (qemuDomainObjEndJob(vm) > 0)
            virDomainRemoveInactive(&driver->domains,
                                    vm);
5138 5139
        vm = NULL;
    }
5140

5141
endjob:
5142
    if (vm) {
5143 5144 5145 5146 5147 5148 5149 5150 5151 5152
        if (ret != 0) {
            if (header.was_running && priv->mon) {
                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;
            }
5153

5154 5155 5156 5157 5158
            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);
5159
            }
5160 5161 5162 5163 5164 5165

            if ((!bypassSecurityDriver) &&
                driver->securityDriver &&
                driver->securityDriver->domainRestoreSavedStateLabel &&
                driver->securityDriver->domainRestoreSavedStateLabel(vm, path) == -1)
                VIR_WARN("failed to restore save state label on %s", path);
5166 5167
        }

5168
        if (qemuDomainObjEndJob(vm) == 0)
5169
            vm = NULL;
5170
    }
5171

5172 5173
cleanup:
    VIR_FREE(xml);
5174
    if (ret != 0 && is_reg)
5175
        unlink(path);
5176 5177
    if (vm)
        virDomainObjUnlock(vm);
5178 5179
    if (event)
        qemuDomainEventQueue(driver, event);
5180
    virCgroupFree(&cgroup);
5181
    qemuDriverUnlock(driver);
5182
    return ret;
D
Daniel P. Berrange 已提交
5183 5184
}

5185 5186 5187 5188 5189 5190 5191 5192 5193 5194 5195 5196 5197 5198 5199 5200 5201 5202 5203 5204 5205 5206 5207 5208 5209 5210 5211 5212 5213 5214 5215 5216 5217 5218 5219 5220 5221 5222 5223 5224 5225 5226
static int qemudDomainSave(virDomainPtr dom, const char *path)
{
    struct qemud_driver *driver = dom->conn->privateData;
    int compressed;

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

    return qemudDomainSaveFlag(dom, path, compressed);
}

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

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

    return(ret);
}

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

5227
    virCheckFlags(0, -1);
5228 5229 5230 5231 5232 5233 5234 5235 5236 5237 5238 5239 5240 5241 5242 5243 5244 5245 5246 5247 5248 5249 5250 5251 5252 5253 5254 5255 5256 5257 5258 5259 5260 5261 5262 5263 5264 5265 5266 5267 5268 5269 5270 5271 5272 5273 5274 5275 5276 5277 5278 5279 5280 5281 5282 5283 5284 5285 5286 5287

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

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

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

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

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

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

    ret = qemudDomainSaveFlag(dom, name, compressed);

cleanup:
    VIR_FREE(name);

    return ret;

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

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

5288
    virCheckFlags(0, -1);
5289 5290 5291 5292 5293 5294 5295 5296 5297 5298 5299 5300 5301 5302 5303 5304 5305 5306 5307 5308 5309 5310 5311 5312 5313 5314 5315 5316 5317 5318 5319 5320 5321

    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;

5322
    virCheckFlags(0, -1);
5323 5324 5325 5326 5327 5328 5329 5330 5331 5332 5333 5334 5335 5336 5337 5338 5339 5340 5341 5342 5343 5344 5345 5346

    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
5347

P
Paolo Bonzini 已提交
5348 5349 5350 5351 5352 5353
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;
5354
    int ret = -1, fd = -1;
5355
    virDomainEventPtr event = NULL;
5356 5357 5358 5359
    const char *args[] = {
        "cat",
        NULL,
    };
5360
    qemuDomainObjPrivatePtr priv;
P
Paolo Bonzini 已提交
5361 5362 5363 5364 5365 5366 5367

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

    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
5368 5369
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
P
Paolo Bonzini 已提交
5370 5371
        goto cleanup;
    }
5372
    priv = vm->privateData;
P
Paolo Bonzini 已提交
5373

5374
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
5375 5376
        goto cleanup;

D
Daniel P. Berrange 已提交
5377
    if (!virDomainObjIsActive(vm)) {
5378 5379
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
5380
        goto endjob;
P
Paolo Bonzini 已提交
5381 5382
    }

5383 5384
    /* Create an empty file with appropriate ownership.  */
    if ((fd = open(path, O_CREAT|O_TRUNC|O_WRONLY, S_IRUSR|S_IWUSR)) < 0) {
5385 5386
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("failed to create '%s'"), path);
5387 5388 5389 5390
        goto endjob;
    }

    if (close(fd) < 0) {
5391
        virReportSystemError(errno,
5392 5393 5394 5395 5396 5397 5398
                             _("unable to save file %s"),
                             path);
        goto endjob;
    }

    if (driver->securityDriver &&
        driver->securityDriver->domainSetSavedStateLabel &&
5399
        driver->securityDriver->domainSetSavedStateLabel(vm, path) == -1)
5400 5401
        goto endjob;

P
Paolo Bonzini 已提交
5402 5403
    /* 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
5404 5405 5406
    resume = (vm->state == VIR_DOMAIN_RUNNING);

    /* Pause domain for non-live dump */
P
Paolo Bonzini 已提交
5407
    if (!(flags & VIR_DUMP_LIVE) && vm->state == VIR_DOMAIN_RUNNING) {
5408
        vm->state = VIR_DOMAIN_PAUSED;
5409
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
5410
        if (qemuMonitorStopCPUs(priv->mon) < 0) {
5411
            qemuDomainObjExitMonitorWithDriver(driver, vm);
5412
            vm->state = VIR_DOMAIN_RUNNING;
5413
            goto endjob;
5414
        }
5415
        qemuDomainObjExitMonitorWithDriver(driver, vm);
P
Paolo Bonzini 已提交
5416
        paused = 1;
5417 5418 5419 5420 5421 5422

        if (!virDomainObjIsActive(vm)) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("guest unexpectedly quit"));
            goto endjob;
        }
P
Paolo Bonzini 已提交
5423 5424
    }

5425
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
5426 5427 5428
    ret = qemuMonitorMigrateToFile(priv->mon,
                                   QEMU_MONITOR_MIGRATE_BACKGROUND,
                                   args, path, 0);
5429
    qemuDomainObjExitMonitorWithDriver(driver, vm);
5430 5431 5432 5433 5434 5435 5436 5437 5438
    if (ret < 0)
        goto endjob;

    ret = qemuDomainWaitForMigrationComplete(driver, vm);

    if (ret < 0)
        goto endjob;

    paused = 1;
5439 5440 5441

    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSavedStateLabel &&
5442
        driver->securityDriver->domainRestoreSavedStateLabel(vm, path) == -1)
5443 5444 5445
        goto endjob;

endjob:
5446
    if ((ret == 0) && (flags & VIR_DUMP_CRASH)) {
5447
        qemudShutdownVMDaemon(driver, vm, 0);
5448 5449 5450 5451 5452
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         VIR_DOMAIN_EVENT_STOPPED_CRASHED);
    }

P
Paolo Bonzini 已提交
5453 5454 5455
    /* 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.  */
5456
    else if (resume && paused && priv->mon) {
5457
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
5458
        if (qemuMonitorStartCPUs(priv->mon, dom->conn) < 0) {
5459
            if (virGetLastError() == NULL)
5460 5461
                qemuReportError(VIR_ERR_OPERATION_FAILED,
                                "%s", _("resuming after dump failed"));
P
Paolo Bonzini 已提交
5462
        }
5463
        qemuDomainObjExitMonitorWithDriver(driver, vm);
5464
        vm->state = VIR_DOMAIN_RUNNING;
P
Paolo Bonzini 已提交
5465
    }
5466

5467 5468
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
5469
    else if ((ret == 0) && (flags & VIR_DUMP_CRASH) && !vm->persistent) {
5470 5471 5472 5473
        virDomainRemoveInactive(&driver->domains,
                                vm);
        vm = NULL;
    }
5474 5475

cleanup:
5476 5477
    if (ret != 0)
        unlink(path);
P
Paolo Bonzini 已提交
5478 5479
    if (vm)
        virDomainObjUnlock(vm);
5480 5481
    if (event)
        qemuDomainEventQueue(driver, event);
5482
    qemuDriverUnlock(driver);
P
Paolo Bonzini 已提交
5483 5484 5485 5486
    return ret;
}


5487 5488 5489 5490 5491 5492
static int qemudDomainHotplugVcpus(virDomainObjPtr vm, unsigned int nvcpus)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int i, rc;
    int ret = -1;

5493 5494
    qemuDomainObjEnterMonitor(vm);

5495 5496 5497 5498 5499 5500 5501 5502 5503 5504 5505 5506 5507 5508 5509 5510 5511 5512 5513 5514 5515 5516 5517 5518 5519 5520 5521 5522 5523 5524
    /* 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:
5525
    qemuDomainObjExitMonitor(vm);
5526 5527 5528 5529 5530 5531 5532 5533 5534 5535
    return ret;

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


static int qemudDomainSetVcpus(virDomainPtr dom, unsigned int nvcpus) {
5536 5537
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
5538 5539
    const char * type;
    int max;
5540
    int ret = -1;
5541

5542
    qemuDriverLock(driver);
5543
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5544 5545
    qemuDriverUnlock(driver);

5546
    if (!vm) {
5547 5548
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
5549 5550
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
5551
        goto cleanup;
5552 5553
    }

5554 5555 5556
    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

5557 5558 5559
    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                         "%s", _("domain is not running"));
5560
        goto endjob;
5561 5562
    }

5563 5564 5565 5566 5567 5568 5569 5570 5571 5572 5573 5574 5575 5576 5577 5578 5579 5580 5581 5582 5583 5584 5585 5586 5587
    if (!(type = virDomainVirtTypeToString(vm->def->virtType))) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("unknown virt type in domain definition '%d'"),
                        vm->def->virtType);
        goto endjob;
    }

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

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

    ret = qemudDomainHotplugVcpus(vm, nvcpus);

endjob:
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
5588

5589
cleanup:
5590 5591
    if (vm)
        virDomainObjUnlock(vm);
5592
    return ret;
5593 5594
}

5595 5596 5597 5598 5599 5600

static int
qemudDomainPinVcpu(virDomainPtr dom,
                   unsigned int vcpu,
                   unsigned char *cpumap,
                   int maplen) {
5601 5602
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
5603
    int maxcpu, hostcpus;
5604
    virNodeInfo nodeinfo;
5605
    int ret = -1;
5606
    qemuDomainObjPrivatePtr priv;
5607

5608
    qemuDriverLock(driver);
5609
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5610 5611
    qemuDriverUnlock(driver);

5612 5613 5614
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
5615 5616
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
5617 5618 5619
        goto cleanup;
    }

D
Daniel P. Berrange 已提交
5620
    if (!virDomainObjIsActive(vm)) {
5621 5622
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s",_("cannot pin vcpus on an inactive domain"));
5623
        goto cleanup;
5624 5625
    }

5626 5627 5628
    priv = vm->privateData;

    if (vcpu > (priv->nvcpupids-1)) {
5629 5630 5631
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("vcpu number out of range %d > %d"),
                        vcpu, priv->nvcpupids);
5632
        goto cleanup;
5633 5634
    }

5635
    if (nodeGetInfo(dom->conn, &nodeinfo) < 0)
5636
        goto cleanup;
5637

5638
    hostcpus = VIR_NODEINFO_MAXCPUS(nodeinfo);
5639
    maxcpu = maplen * 8;
5640 5641
    if (maxcpu > hostcpus)
        maxcpu = hostcpus;
5642

5643 5644
    if (priv->vcpupids != NULL) {
        if (virProcessInfoSetAffinity(priv->vcpupids[vcpu],
5645
                                      cpumap, maplen, maxcpu) < 0)
5646
            goto cleanup;
5647
    } else {
5648 5649
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        "%s", _("cpu affinity is not supported"));
5650
        goto cleanup;
5651
    }
5652
    ret = 0;
5653

5654
cleanup:
5655 5656
    if (vm)
        virDomainObjUnlock(vm);
5657
    return ret;
5658 5659 5660 5661 5662 5663 5664 5665
}

static int
qemudDomainGetVcpus(virDomainPtr dom,
                    virVcpuInfoPtr info,
                    int maxinfo,
                    unsigned char *cpumaps,
                    int maplen) {
5666 5667
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
5668
    virNodeInfo nodeinfo;
5669
    int i, v, maxcpu, hostcpus;
5670
    int ret = -1;
5671
    qemuDomainObjPrivatePtr priv;
5672

5673
    qemuDriverLock(driver);
5674
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5675 5676
    qemuDriverUnlock(driver);

5677 5678 5679
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
5680 5681
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
5682 5683 5684
        goto cleanup;
    }

D
Daniel P. Berrange 已提交
5685
    if (!virDomainObjIsActive(vm)) {
5686 5687 5688
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s",
                        _("cannot list vcpu pinning for an inactive domain"));
5689
        goto cleanup;
5690 5691
    }

5692 5693
    priv = vm->privateData;

5694
    if (nodeGetInfo(dom->conn, &nodeinfo) < 0)
5695
        goto cleanup;
5696

5697
    hostcpus = VIR_NODEINFO_MAXCPUS(nodeinfo);
5698
    maxcpu = maplen * 8;
5699 5700
    if (maxcpu > hostcpus)
        maxcpu = hostcpus;
5701 5702

    /* Clamp to actual number of vcpus */
5703 5704
    if (maxinfo > priv->nvcpupids)
        maxinfo = priv->nvcpupids;
5705

5706 5707 5708 5709 5710 5711
    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;
5712

5713
                if (priv->vcpupids != NULL &&
5714 5715 5716
                    qemudGetProcessInfo(&(info[i].cpuTime),
                                        &(info[i].cpu),
                                        vm->pid,
5717
                                        priv->vcpupids[i]) < 0) {
5718
                    virReportSystemError(errno, "%s",
5719 5720 5721
                                         _("cannot get vCPU placement & pCPU time"));
                    goto cleanup;
                }
5722
            }
5723 5724
        }

5725 5726
        if (cpumaps != NULL) {
            memset(cpumaps, 0, maplen * maxinfo);
5727
            if (priv->vcpupids != NULL) {
5728 5729 5730
                for (v = 0 ; v < maxinfo ; v++) {
                    unsigned char *cpumap = VIR_GET_CPUMAP(cpumaps, maplen, v);

5731
                    if (virProcessInfoGetAffinity(priv->vcpupids[v],
5732
                                                  cpumap, maplen, maxcpu) < 0)
5733
                        goto cleanup;
5734
                }
5735
            } else {
5736 5737
                qemuReportError(VIR_ERR_NO_SUPPORT,
                                "%s", _("cpu affinity is not available"));
5738
                goto cleanup;
5739 5740 5741
            }
        }
    }
5742
    ret = maxinfo;
5743

5744
cleanup:
5745 5746
    if (vm)
        virDomainObjUnlock(vm);
5747
    return ret;
5748 5749 5750
}


5751
static int qemudDomainGetMaxVcpus(virDomainPtr dom) {
5752 5753
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
5754
    const char *type;
5755
    int ret = -1;
5756

5757
    qemuDriverLock(driver);
5758
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5759 5760
    qemuDriverUnlock(driver);

5761
    if (!vm) {
5762 5763
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
5764 5765
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
5766
        goto cleanup;
5767 5768
    }

5769
    if (!(type = virDomainVirtTypeToString(vm->def->virtType))) {
5770 5771 5772
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("unknown virt type in domain definition '%d'"),
                        vm->def->virtType);
5773
        goto cleanup;
5774 5775
    }

5776
    ret = qemudGetMaxVCPUs(NULL, type);
5777

5778
cleanup:
5779 5780
    if (vm)
        virDomainObjUnlock(vm);
5781 5782 5783
    return ret;
}

5784 5785 5786 5787 5788 5789 5790 5791 5792
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);

5793 5794
    memset(seclabel, 0, sizeof(*seclabel));

5795 5796 5797
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
5798 5799
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
5800 5801 5802
        goto cleanup;
    }

5803
    if (!virDomainVirtTypeToString(vm->def->virtType)) {
5804 5805 5806
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("unknown virt type in domain definition '%d'"),
                        vm->def->virtType);
5807 5808 5809 5810 5811 5812 5813 5814 5815 5816 5817 5818 5819 5820 5821 5822 5823
        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
5824
    if (virDomainObjIsActive(vm)) {
5825
        if (driver->securityDriver && driver->securityDriver->domainGetSecurityProcessLabel) {
5826
            if (driver->securityDriver->domainGetSecurityProcessLabel(vm, seclabel) == -1) {
5827 5828
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                "%s", _("Failed to get security label"));
5829 5830 5831 5832 5833 5834 5835 5836 5837 5838
                goto cleanup;
            }
        }
    }

    ret = 0;

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
5839
    qemuDriverUnlock(driver);
5840 5841 5842
    return ret;
}

5843 5844
static int qemudNodeGetSecurityModel(virConnectPtr conn,
                                     virSecurityModelPtr secmodel)
5845 5846 5847
{
    struct qemud_driver *driver = (struct qemud_driver *)conn->privateData;
    char *p;
5848
    int ret = 0;
5849

5850
    qemuDriverLock(driver);
5851
    if (!driver->securityPrimaryDriver) {
5852
        memset(secmodel, 0, sizeof (*secmodel));
5853 5854
        goto cleanup;
    }
5855

5856 5857
    p = driver->caps->host.secModel.model;
    if (strlen(p) >= VIR_SECURITY_MODEL_BUFLEN-1) {
5858 5859 5860
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("security model string exceeds max %d bytes"),
                        VIR_SECURITY_MODEL_BUFLEN-1);
5861 5862
        ret = -1;
        goto cleanup;
5863 5864 5865 5866 5867
    }
    strcpy(secmodel->model, p);

    p = driver->caps->host.secModel.doi;
    if (strlen(p) >= VIR_SECURITY_DOI_BUFLEN-1) {
5868 5869 5870
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("security DOI string exceeds max %d bytes"),
                        VIR_SECURITY_DOI_BUFLEN-1);
5871 5872
        ret = -1;
        goto cleanup;
5873 5874
    }
    strcpy(secmodel->doi, p);
5875 5876 5877 5878

cleanup:
    qemuDriverUnlock(driver);
    return ret;
5879 5880
}

5881 5882 5883 5884 5885 5886 5887 5888 5889 5890 5891 5892 5893 5894 5895 5896 5897 5898 5899 5900 5901 5902 5903 5904 5905 5906 5907 5908 5909 5910 5911 5912 5913 5914 5915 5916 5917 5918 5919 5920 5921 5922 5923 5924 5925 5926 5927 5928 5929 5930 5931 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 5965 5966 5967 5968 5969 5970 5971 5972 5973 5974 5975 5976 5977 5978 5979 5980 5981 5982 5983 5984 5985 5986 5987 5988 5989 5990 5991 5992 5993 5994 5995 5996 5997 5998 5999 6000 6001 6002 6003 6004 6005 6006 6007 6008 6009 6010 6011 6012
/* qemudOpenAsUID() - pipe/fork/setuid/open a file, and return the
   pipe fd to caller, so that it can read from the file. Also return
   the pid of the child process, so the caller can wait for it to exit
   after it's finished reading (to avoid a zombie, if nothing
   else). */

static int qemudOpenAsUID(const char *path, uid_t uid, pid_t *child_pid) {
    int pipefd[2];
    int fd = -1;

    *child_pid = -1;

    if (pipe(pipefd) < 0) {
        virReportSystemError(errno,
                             _("failed to create pipe to read '%s'"),
                             path);
        pipefd[0] = pipefd[1] = -1;
        goto parent_cleanup;
    }

    int forkRet = virFork(child_pid);

    if (*child_pid < 0) {
        virReportSystemError(errno,
                             _("failed to fork child to read '%s'"),
                             path);
        goto parent_cleanup;
    }

    if (*child_pid > 0) {

        /* parent */

        /* parent doesn't need the write side of the pipe */
        close(pipefd[1]);
        pipefd[1] = -1;

        if (forkRet < 0) {
            virReportSystemError(errno,
                                 _("failed in parent after forking child to read '%s'"),
                                 path);
            goto parent_cleanup;
        }
        /* caller gets the read side of the pipe */
        fd = pipefd[0];
        pipefd[0] = -1;
parent_cleanup:
        if (pipefd[0] != -1)
            close(pipefd[0]);
        if (pipefd[1] != -1)
            close(pipefd[1]);
        if ((fd < 0) && (*child_pid > 0)) {
            /* a child process was started and subsequently an error
               occurred in the parent, so we need to wait for it to
               exit, but its status is inconsequential. */
            while ((waitpid(*child_pid, NULL, 0) == -1)
                   && (errno == EINTR)) {
                /* empty */
            }
            *child_pid = -1;
        }
        return fd;
    }

    /* child */

    /* setuid to the qemu user, then open the file, read it,
       and stuff it into the pipe for the parent process to
       read */
    int exit_code;
    char *buf = NULL;
    size_t bufsize = 1024 * 1024;
    int bytesread;

    /* child doesn't need the read side of the pipe */
    close(pipefd[0]);

    if (forkRet < 0) {
        exit_code = errno;
        virReportSystemError(errno,
                             _("failed in child after forking to read '%s'"),
                             path);
        goto child_cleanup;
    }

    if (setuid(uid) != 0) {
        exit_code = errno;
        virReportSystemError(errno,
                             _("cannot setuid(%d) to read '%s'"),
                             uid, path);
        goto child_cleanup;
    }
    if ((fd = open(path, O_RDONLY)) < 0) {
        exit_code = errno;
        virReportSystemError(errno,
                             _("cannot open '%s' as uid %d"),
                             path, uid);
        goto child_cleanup;
    }
    if (VIR_ALLOC_N(buf, bufsize) < 0) {
        exit_code = ENOMEM;
        virReportOOMError();
        goto child_cleanup;
    }

    /* read from fd and write to pipefd[1] until EOF */
    do {
        if ((bytesread = saferead(fd, buf, bufsize)) < 0) {
            exit_code = errno;
            virReportSystemError(errno,
                                 _("child failed reading from '%s'"),
                                 path);
            goto child_cleanup;
        }
        if (safewrite(pipefd[1], buf, bytesread) != bytesread) {
            exit_code = errno;
            virReportSystemError(errno, "%s",
                                 _("child failed writing to pipe"));
            goto child_cleanup;
        }
    } while (bytesread > 0);
    exit_code = 0;

child_cleanup:
    VIR_FREE(buf);
    if (fd != -1)
        close(fd);
    if (pipefd[1] != -1)
        close(pipefd[1]);
    _exit(exit_code);
}

6013
/* TODO: check seclabel restore */
6014
static int qemudDomainRestore(virConnectPtr conn,
6015 6016 6017
                              const char *path) {
    struct qemud_driver *driver = conn->privateData;
    virDomainDefPtr def = NULL;
6018
    virDomainObjPtr vm = NULL;
6019
    int fd = -1;
6020
    pid_t read_pid = -1;
6021 6022
    int ret = -1;
    char *xml = NULL;
6023
    struct qemud_save_header header;
6024
    virDomainEventPtr event = NULL;
6025 6026 6027
    int intermediatefd = -1;
    pid_t intermediate_pid = -1;
    int childstat;
6028

6029
    qemuDriverLock(driver);
6030 6031
    /* Verify the header and read the XML */
    if ((fd = open(path, O_RDONLY)) < 0) {
6032 6033 6034 6035 6036 6037 6038 6039 6040 6041 6042 6043 6044 6045
        if ((driver->user == 0) || (getuid() != 0)) {
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            "%s", _("cannot read domain image"));
            goto cleanup;
        }

        /* 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 */
            goto cleanup;
        }
6046 6047 6048
    }

    if (saferead(fd, &header, sizeof(header)) != sizeof(header)) {
6049 6050
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to read qemu header"));
6051
        goto cleanup;
6052 6053 6054
    }

    if (memcmp(header.magic, QEMUD_SAVE_MAGIC, sizeof(header.magic)) != 0) {
6055 6056
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("image magic is incorrect"));
6057
        goto cleanup;
6058 6059 6060
    }

    if (header.version > QEMUD_SAVE_VERSION) {
6061 6062 6063
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("image version is not supported (%d > %d)"),
                        header.version, QEMUD_SAVE_VERSION);
6064
        goto cleanup;
6065 6066
    }

6067 6068 6069
    if (header.xml_len <= 0) {
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("invalid XML length: %d"), header.xml_len);
6070
        goto cleanup;
6071 6072
    }

6073 6074 6075 6076 6077 6078
    if (VIR_ALLOC_N(xml, header.xml_len) < 0) {
        virReportOOMError();
        goto cleanup;
    }

    if (saferead(fd, xml, header.xml_len) != header.xml_len) {
6079 6080
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to read XML"));
6081
        goto cleanup;
6082 6083 6084
    }

    /* Create a domain from this XML */
6085
    if (!(def = virDomainDefParseString(driver->caps, xml,
6086
                                        VIR_DOMAIN_XML_INACTIVE))) {
6087 6088
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to parse XML"));
6089
        goto cleanup;
6090 6091
    }

6092 6093
    if (virDomainObjIsDuplicate(&driver->domains, def, 1) < 0)
        goto cleanup;
6094

6095
    if (!(vm = virDomainAssignDef(driver->caps,
6096
                                  &driver->domains,
6097
                                  def, true))) {
6098 6099
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to assign new VM"));
6100
        goto cleanup;
6101
    }
6102
    def = NULL;
6103

6104 6105 6106
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

6107 6108
    if (header.version == 2) {
        const char *intermediate_argv[3] = { NULL, "-dc", NULL };
6109 6110
        const char *prog = qemudSaveCompressionTypeToString(header.compressed);
        if (prog == NULL) {
6111 6112 6113
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("Invalid compressed save format %d"),
                            header.compressed);
6114
            goto endjob;
6115
        }
6116

6117
        if (header.compressed != QEMUD_SAVE_FORMAT_RAW) {
6118
            intermediate_argv[0] = prog;
6119 6120
            intermediatefd = fd;
            fd = -1;
6121
            if (virExec(intermediate_argv, NULL, NULL,
6122
                        &intermediate_pid, intermediatefd, &fd, NULL, 0) < 0) {
6123 6124 6125
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("Failed to start decompression binary %s"),
                                intermediate_argv[0]);
6126
                goto endjob;
6127 6128 6129
            }
        }
    }
6130
    /* Set the migration source and start it up. */
6131
    ret = qemudStartVMDaemon(conn, driver, vm, "stdio", fd);
6132 6133 6134 6135 6136 6137 6138
    if (intermediate_pid != -1) {
        /* Wait for intermediate process to exit */
        while (waitpid(intermediate_pid, &childstat, 0) == -1 &&
               errno == EINTR);
    }
    if (intermediatefd != -1)
        close(intermediatefd);
6139
    close(fd);
6140
    fd = -1;
6141 6142 6143 6144 6145 6146 6147 6148 6149 6150 6151 6152 6153 6154 6155 6156 6157 6158 6159 6160 6161 6162 6163 6164 6165 6166 6167 6168 6169
    if (read_pid != -1) {
        int wait_ret;
        int status;
        /* reap the process that read the file */
        while (((wait_ret = waitpid(read_pid, &status, 0)) == -1)
               && (errno == EINTR)) {
            /* empty */
        }
        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;
            }
        }
    }
6170
    if (ret < 0) {
6171
        if (!vm->persistent) {
6172 6173 6174
            if (qemuDomainObjEndJob(vm) > 0)
                virDomainRemoveInactive(&driver->domains,
                                        vm);
6175 6176
            vm = NULL;
        }
6177
        goto endjob;
6178 6179
    }

6180 6181 6182
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_RESTORED);
6183

6184 6185
    /* If it was running before, resume it now. */
    if (header.was_running) {
6186
        qemuDomainObjPrivatePtr priv = vm->privateData;
6187
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
6188
        if (qemuMonitorStartCPUs(priv->mon, conn) < 0) {
6189
            if (virGetLastError() == NULL)
6190 6191
                qemuReportError(VIR_ERR_OPERATION_FAILED,
                                "%s", _("failed to resume domain"));
6192 6193
            qemuDomainObjExitMonitorWithDriver(driver,vm);
            goto endjob;
6194
        }
6195
        qemuDomainObjExitMonitorWithDriver(driver, vm);
6196
        vm->state = VIR_DOMAIN_RUNNING;
6197 6198 6199 6200
        if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0) {
            VIR_WARN("Failed to save status on vm %s", vm->def->name);
            goto endjob;
        }
6201
    }
6202
    ret = 0;
6203

6204
endjob:
6205 6206 6207
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
6208

6209 6210 6211 6212 6213
cleanup:
    virDomainDefFree(def);
    VIR_FREE(xml);
    if (fd != -1)
        close(fd);
6214 6215 6216 6217 6218 6219 6220
    if (read_pid != 0) {
        /* reap the process that read the file */
        while ((waitpid(read_pid, NULL, 0) == -1)
               && (errno == EINTR)) {
            /* empty */
        }
    }
6221 6222
    if (vm)
        virDomainObjUnlock(vm);
6223 6224
    if (event)
        qemuDomainEventQueue(driver, event);
6225
    qemuDriverUnlock(driver);
6226
    return ret;
D
Daniel P. Berrange 已提交
6227 6228 6229
}


6230 6231 6232 6233 6234 6235 6236 6237 6238 6239 6240 6241 6242 6243 6244 6245 6246 6247 6248 6249 6250 6251 6252 6253 6254 6255 6256 6257 6258 6259 6260 6261 6262 6263 6264 6265 6266 6267
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;
}


6268
static char *qemudDomainDumpXML(virDomainPtr dom,
6269
                                int flags) {
6270 6271 6272
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    char *ret = NULL;
6273 6274
    unsigned long balloon;
    int err;
6275

6276
    qemuDriverLock(driver);
6277
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
6278

D
Daniel P. Berrange 已提交
6279
    if (!vm) {
6280 6281
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
6282 6283
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
6284
        goto cleanup;
D
Daniel P. Berrange 已提交
6285 6286
    }

6287
    /* Refresh current memory based on balloon info */
D
Daniel P. Berrange 已提交
6288
    if (virDomainObjIsActive(vm)) {
6289
        qemuDomainObjPrivatePtr priv = vm->privateData;
6290 6291 6292
        /* Don't delay if someone's using the monitor, just use
         * existing most recent data instead */
        if (!priv->jobActive) {
6293
            if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
6294 6295
                goto cleanup;

6296
            qemuDomainObjEnterMonitorWithDriver(driver, vm);
6297
            err = qemuMonitorGetBalloonInfo(priv->mon, &balloon);
6298
            qemuDomainObjExitMonitorWithDriver(driver, vm);
6299 6300 6301 6302
            if (qemuDomainObjEndJob(vm) == 0) {
                vm = NULL;
                goto cleanup;
            }
6303 6304 6305 6306 6307 6308
            if (err < 0)
                goto cleanup;
            if (err > 0)
                vm->def->memory = balloon;
            /* err == 0 indicates no balloon support, so ignore it */
        }
6309
    }
6310

6311
    ret = qemudVMDumpXML(driver, vm, flags);
6312 6313

cleanup:
6314 6315
    if (vm)
        virDomainObjUnlock(vm);
6316
    qemuDriverUnlock(driver);
6317
    return ret;
D
Daniel P. Berrange 已提交
6318 6319 6320
}


6321 6322 6323 6324
static char *qemuDomainXMLFromNative(virConnectPtr conn,
                                     const char *format,
                                     const char *config,
                                     unsigned int flags ATTRIBUTE_UNUSED) {
6325
    struct qemud_driver *driver = conn->privateData;
6326 6327 6328 6329
    virDomainDefPtr def = NULL;
    char *xml = NULL;

    if (STRNEQ(format, QEMU_CONFIG_FORMAT_ARGV)) {
6330 6331
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("unsupported config type %s"), format);
6332 6333 6334
        goto cleanup;
    }

6335
    qemuDriverLock(driver);
6336
    def = qemuParseCommandLineString(driver->caps, config);
6337
    qemuDriverUnlock(driver);
6338 6339 6340
    if (!def)
        goto cleanup;

6341
    xml = virDomainDefFormat(def, VIR_DOMAIN_XML_INACTIVE);
6342 6343 6344 6345 6346 6347

cleanup:
    virDomainDefFree(def);
    return xml;
}

6348 6349 6350 6351 6352 6353
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;
6354
    virDomainChrDef monConfig;
6355
    const char *emulator;
6356
    unsigned long long qemuCmdFlags;
6357 6358 6359 6360 6361 6362 6363 6364
    struct stat sb;
    const char **retargv = NULL;
    const char **retenv = NULL;
    const char **tmp;
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    char *ret = NULL;
    int i;

6365 6366
    qemuDriverLock(driver);

6367
    if (STRNEQ(format, QEMU_CONFIG_FORMAT_ARGV)) {
6368 6369
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("unsupported config type %s"), format);
6370 6371 6372
        goto cleanup;
    }

6373
    def = virDomainDefParseString(driver->caps, xmlData, 0);
6374 6375 6376 6377 6378 6379 6380 6381 6382 6383 6384 6385 6386 6387 6388 6389 6390 6391 6392 6393 6394 6395 6396 6397 6398 6399 6400 6401 6402 6403 6404 6405 6406 6407 6408 6409 6410 6411 6412 6413 6414 6415 6416 6417
    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) {
6418
        virReportSystemError(errno,
6419 6420 6421 6422 6423 6424 6425 6426
                             _("Cannot find QEMU binary %s"),
                             emulator);
        goto cleanup;
    }

    if (qemudExtractVersionInfo(emulator,
                                NULL,
                                &qemuCmdFlags) < 0) {
6427 6428 6429
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Cannot determine QEMU argv syntax %s"),
                        emulator);
6430 6431 6432
        goto cleanup;
    }

6433
    if (qemuPrepareMonitorChr(driver, &monConfig, def->name) < 0)
6434
        goto cleanup;
6435 6436

    if (qemudBuildCommandLine(conn, driver, def,
D
Daniel P. Berrange 已提交
6437
                              &monConfig, 0, qemuCmdFlags,
6438 6439
                              &retargv, &retenv,
                              NULL, NULL, /* Don't want it to create TAP devices */
C
Chris Lalancette 已提交
6440
                              NULL, NULL) < 0) {
6441 6442 6443 6444 6445 6446 6447 6448 6449 6450 6451 6452 6453 6454 6455 6456
        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++;
    }

6457 6458
    if (virBufferError(&buf)) {
        virBufferFreeAndReset(&buf);
6459
        virReportOOMError();
6460
        goto cleanup;
6461
    }
6462 6463 6464 6465

    ret = virBufferContentAndReset(&buf);

cleanup:
6466
    qemuDriverUnlock(driver);
6467 6468 6469 6470 6471 6472 6473 6474 6475 6476 6477 6478 6479
    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;
}


6480
static int qemudListDefinedDomains(virConnectPtr conn,
6481
                            char **const names, int nnames) {
6482
    struct qemud_driver *driver = conn->privateData;
6483
    int n;
6484

6485
    qemuDriverLock(driver);
6486
    n = virDomainObjListGetInactiveNames(&driver->domains, names, nnames);
6487
    qemuDriverUnlock(driver);
6488
    return n;
D
Daniel P. Berrange 已提交
6489 6490
}

6491
static int qemudNumDefinedDomains(virConnectPtr conn) {
6492
    struct qemud_driver *driver = conn->privateData;
6493
    int n;
6494

6495
    qemuDriverLock(driver);
6496
    n = virDomainObjListNumOfDomains(&driver->domains, 0);
6497
    qemuDriverUnlock(driver);
6498

6499
    return n;
D
Daniel P. Berrange 已提交
6500 6501 6502
}


6503
static int qemudDomainStart(virDomainPtr dom) {
6504 6505 6506
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
6507
    virDomainEventPtr event = NULL;
6508
    char *managed_save = NULL;
6509

6510
    qemuDriverLock(driver);
6511
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
6512

6513
    if (!vm) {
6514 6515
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
6516 6517
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
6518
        goto cleanup;
6519 6520
    }

6521 6522 6523 6524
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

    if (virDomainObjIsActive(vm)) {
6525 6526
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is already running"));
6527 6528 6529
        goto endjob;
    }

6530 6531 6532 6533 6534 6535 6536 6537 6538 6539
    /*
     * 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))) {
        /*
         * We should still have a reference left to vm but
         * one should check for 0 anyway
         */
6540
        if (qemuDomainObjEndJob(vm) == 0) {
6541
            vm = NULL;
6542 6543 6544
            goto cleanup;
        }

6545 6546 6547 6548 6549 6550 6551 6552 6553 6554 6555 6556 6557 6558 6559 6560 6561
        virDomainObjUnlock(vm);
        qemuDriverUnlock(driver);
        ret = qemudDomainRestore(dom->conn, managed_save);

        if (unlink(managed_save) < 0) {
            VIR_WARN("Failed to remove the managed state %s", managed_save);
        }

        if (ret == 0) {
            /* qemudDomainRestore should have sent the Started/Restore event */
            VIR_FREE(managed_save);
            return(ret);
        }
        qemuDriverLock(driver);
        virDomainObjLock(vm);
    }

6562
    ret = qemudStartVMDaemon(dom->conn, driver, vm, NULL, -1);
6563
    if (ret != -1)
6564 6565 6566
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STARTED,
                                         VIR_DOMAIN_EVENT_STARTED_BOOTED);
6567

6568
endjob:
6569 6570
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
6571

6572
cleanup:
6573
    VIR_FREE(managed_save);
6574 6575
    if (vm)
        virDomainObjUnlock(vm);
6576
    if (event)
6577
        qemuDomainEventQueue(driver, event);
6578
    qemuDriverUnlock(driver);
6579
    return ret;
D
Daniel P. Berrange 已提交
6580 6581
}

6582 6583 6584 6585 6586 6587 6588 6589 6590 6591 6592 6593 6594 6595 6596
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;

6597
        if (STRNEQ(def->os.machine, machine->name))
6598 6599 6600
            continue;

        if (!(*canonical = strdup(machine->canonical))) {
6601
            virReportOOMError();
6602 6603 6604 6605 6606 6607 6608 6609 6610
            return -1;
        }

        break;
    }

    return 0;
}

6611 6612 6613 6614 6615 6616 6617
static int
qemudCanonicalizeMachineDirect(virDomainDefPtr def, char **canonical)
{
    virCapsGuestMachinePtr *machines = NULL;
    int i, nmachines = 0;

    if (qemudProbeMachineTypes(def->emulator, &machines, &nmachines) < 0) {
6618
        virReportOOMError();
6619 6620 6621 6622 6623 6624 6625
        return -1;
    }

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

6626
        if (STRNEQ(def->os.machine, machines[i]->name))
6627 6628 6629 6630 6631 6632 6633 6634 6635 6636 6637 6638
            continue;

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

    virCapabilitiesFreeMachines(machines, nmachines);

    return 0;
}

6639 6640
int
qemudCanonicalizeMachine(struct qemud_driver *driver, virDomainDefPtr def)
6641 6642 6643 6644 6645 6646
{
    char *canonical = NULL;
    int i;

    for (i = 0; i < driver->caps->nguests; i++) {
        virCapsGuestPtr guest = driver->caps->guests[i];
6647
        virCapsGuestDomainInfoPtr info;
6648 6649 6650
        int j;

        for (j = 0; j < guest->arch.ndomains; j++) {
6651
            info = &guest->arch.domains[j]->info;
6652

6653 6654 6655 6656 6657 6658 6659 6660 6661
            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;
6662 6663
        }

6664 6665 6666 6667
        info = &guest->arch.defaultInfo;

        if (info->emulator && STREQ(info->emulator, def->emulator)) {
            if (qemudCanonicalizeMachineFromInfo(def, info, &canonical) < 0)
6668 6669 6670 6671
                return -1;
            goto out;
        }
    }
6672 6673 6674 6675

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

6676 6677 6678 6679 6680 6681 6682
out:
    if (canonical) {
        VIR_FREE(def->os.machine);
        def->os.machine = canonical;
    }
    return 0;
}
D
Daniel P. Berrange 已提交
6683

6684
static virDomainPtr qemudDomainDefine(virConnectPtr conn, const char *xml) {
6685
    struct qemud_driver *driver = conn->privateData;
6686
    virDomainDefPtr def;
6687
    virDomainObjPtr vm = NULL;
6688
    virDomainPtr dom = NULL;
6689
    virDomainEventPtr event = NULL;
6690
    int dupVM;
6691

6692
    qemuDriverLock(driver);
6693
    if (!(def = virDomainDefParseString(driver->caps, xml,
6694
                                        VIR_DOMAIN_XML_INACTIVE)))
6695
        goto cleanup;
6696

6697
    if (virSecurityDriverVerify(def) < 0)
6698 6699
        goto cleanup;

6700 6701
    if ((dupVM = virDomainObjIsDuplicate(&driver->domains, def, 0)) < 0)
        goto cleanup;
6702

6703
    if (qemudCanonicalizeMachine(driver, def) < 0)
6704 6705
        goto cleanup;

6706 6707 6708
    if (qemuAssignPCIAddresses(def) < 0)
        goto cleanup;

6709
    if (!(vm = virDomainAssignDef(driver->caps,
6710
                                  &driver->domains,
6711
                                  def, false))) {
6712
        goto cleanup;
6713
    }
6714
    def = NULL;
6715
    vm->persistent = 1;
6716

6717
    if (virDomainSaveConfig(driver->configDir,
6718
                            vm->newDef ? vm->newDef : vm->def) < 0) {
6719 6720
        virDomainRemoveInactive(&driver->domains,
                                vm);
6721
        vm = NULL;
6722
        goto cleanup;
6723 6724
    }

6725 6726
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_DEFINED,
6727
                                     !dupVM ?
6728 6729
                                     VIR_DOMAIN_EVENT_DEFINED_ADDED :
                                     VIR_DOMAIN_EVENT_DEFINED_UPDATED);
6730

6731
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
6732
    if (dom) dom->id = vm->def->id;
6733 6734

cleanup:
6735
    virDomainDefFree(def);
6736 6737
    if (vm)
        virDomainObjUnlock(vm);
6738 6739
    if (event)
        qemuDomainEventQueue(driver, event);
6740
    qemuDriverUnlock(driver);
6741
    return dom;
D
Daniel P. Berrange 已提交
6742 6743
}

6744
static int qemudDomainUndefine(virDomainPtr dom) {
6745 6746
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
6747
    virDomainEventPtr event = NULL;
6748
    int ret = -1;
D
Daniel P. Berrange 已提交
6749

6750
    qemuDriverLock(driver);
6751
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
6752

D
Daniel P. Berrange 已提交
6753
    if (!vm) {
6754 6755
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
6756 6757
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
6758
        goto cleanup;
D
Daniel P. Berrange 已提交
6759 6760
    }

D
Daniel P. Berrange 已提交
6761
    if (virDomainObjIsActive(vm)) {
6762 6763
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot delete active domain"));
6764
        goto cleanup;
D
Daniel P. Berrange 已提交
6765 6766
    }

6767
    if (!vm->persistent) {
6768 6769
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("cannot undefine transient domain"));
6770
        goto cleanup;
6771 6772
    }

6773
    if (virDomainDeleteConfig(driver->configDir, driver->autostartDir, vm) < 0)
6774
        goto cleanup;
D
Daniel P. Berrange 已提交
6775

6776 6777 6778
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_UNDEFINED,
                                     VIR_DOMAIN_EVENT_UNDEFINED_REMOVED);
6779

6780 6781
    virDomainRemoveInactive(&driver->domains,
                            vm);
6782
    vm = NULL;
6783
    ret = 0;
D
Daniel P. Berrange 已提交
6784

6785
cleanup:
6786 6787
    if (vm)
        virDomainObjUnlock(vm);
6788 6789
    if (event)
        qemuDomainEventQueue(driver, event);
6790
    qemuDriverUnlock(driver);
6791
    return ret;
D
Daniel P. Berrange 已提交
6792 6793
}

6794

6795
static int qemudDomainChangeEjectableMedia(struct qemud_driver *driver,
6796
                                           virDomainObjPtr vm,
6797 6798
                                           virDomainDiskDefPtr disk,
                                           unsigned long long qemuCmdFlags)
6799
{
6800
    virDomainDiskDefPtr origdisk = NULL;
6801
    int i;
6802
    int ret;
6803
    char *driveAlias = NULL;
6804

6805
    origdisk = NULL;
6806
    for (i = 0 ; i < vm->def->ndisks ; i++) {
6807 6808
        if (vm->def->disks[i]->bus == disk->bus &&
            STREQ(vm->def->disks[i]->dst, disk->dst)) {
6809
            origdisk = vm->def->disks[i];
6810
            break;
6811
        }
6812 6813 6814
    }

    if (!origdisk) {
6815 6816 6817 6818
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("No device with bus '%s' and target '%s'"),
                        virDomainDiskBusTypeToString(disk->bus),
                        disk->dst);
6819 6820 6821
        return -1;
    }

6822
    if (!origdisk->info.alias) {
6823 6824
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("missing disk device alias name for %s"), origdisk->dst);
6825 6826
        return -1;
    }
6827

6828 6829
    if (origdisk->device != VIR_DOMAIN_DISK_DEVICE_FLOPPY &&
        origdisk->device != VIR_DOMAIN_DISK_DEVICE_CDROM) {
6830 6831 6832
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Removable media not supported for %s device"),
                        virDomainDiskDeviceTypeToString(disk->device));
6833
        return -1;
6834
    }
6835

6836 6837
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityImageLabel &&
6838
        driver->securityDriver->domainSetSecurityImageLabel(vm, disk) < 0)
6839 6840
        return -1;

6841 6842 6843
    if (!(driveAlias = qemuDeviceDriveHostAlias(origdisk, qemuCmdFlags)))
        goto error;

6844
    qemuDomainObjPrivatePtr priv = vm->privateData;
6845
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
6846
    if (disk->src) {
6847
        const char *format = NULL;
6848 6849 6850
        if (disk->type != VIR_DOMAIN_DISK_TYPE_DIR) {
            if (disk->driverType)
                format = disk->driverType;
6851 6852 6853
            else if (origdisk->driverType)
                format = origdisk->driverType;
        }
6854
        ret = qemuMonitorChangeMedia(priv->mon,
6855
                                     driveAlias,
6856
                                     disk->src, format);
6857
    } else {
6858
        ret = qemuMonitorEjectMedia(priv->mon, driveAlias);
6859
    }
6860
    qemuDomainObjExitMonitorWithDriver(driver, vm);
6861

6862 6863 6864 6865 6866
    if (ret < 0)
        goto error;

    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
6867
        driver->securityDriver->domainRestoreSecurityImageLabel(vm, origdisk) < 0)
6868 6869 6870
        VIR_WARN("Unable to restore security label on ejected image %s", origdisk->src);

    VIR_FREE(origdisk->src);
6871 6872 6873 6874
    origdisk->src = disk->src;
    disk->src = NULL;
    origdisk->type = disk->type;

6875 6876
    VIR_FREE(driveAlias);

6877
    virDomainDiskDefFree(disk);
6878

6879
    return ret;
6880 6881

error:
6882
    VIR_FREE(driveAlias);
6883 6884
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
6885
        driver->securityDriver->domainRestoreSecurityImageLabel(vm, disk) < 0)
6886
        VIR_WARN("Unable to restore security label on new media %s", disk->src);
6887
    return -1;
6888 6889
}

6890

6891
static int qemudDomainAttachPciDiskDevice(struct qemud_driver *driver,
6892
                                          virDomainObjPtr vm,
6893 6894
                                          virDomainDiskDefPtr disk,
                                          int qemuCmdFlags)
6895
{
6896
    int i, ret;
6897
    const char* type = virDomainDiskBusTypeToString(disk->bus);
6898
    qemuDomainObjPrivatePtr priv = vm->privateData;
6899 6900
    char *devstr = NULL;
    char *drivestr = NULL;
6901 6902

    for (i = 0 ; i < vm->def->ndisks ; i++) {
6903
        if (STREQ(vm->def->disks[i]->dst, disk->dst)) {
6904 6905
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("target %s already exists"), disk->dst);
6906 6907 6908 6909
            return -1;
        }
    }

6910 6911
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityImageLabel &&
6912
        driver->securityDriver->domainSetSecurityImageLabel(vm, disk) < 0)
6913 6914
        return -1;

6915
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
6916 6917
        if (qemuDomainPCIAddressEnsureAddr(priv->pciaddrs, &disk->info) < 0)
            goto error;
6918 6919
        if (qemuAssignDeviceDiskAlias(disk, qemuCmdFlags) < 0)
            goto error;
6920

6921 6922 6923
        if (!(drivestr = qemuBuildDriveStr(disk, 0, qemuCmdFlags)))
            goto error;

6924
        if (!(devstr = qemuBuildDriveDevStr(disk)))
6925 6926 6927
            goto error;
    }

6928
    if (VIR_REALLOC_N(vm->def->disks, vm->def->ndisks+1) < 0) {
6929
        virReportOOMError();
6930
        goto error;
6931 6932
    }

6933
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
6934 6935
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        ret = qemuMonitorAddDrive(priv->mon, drivestr);
6936 6937 6938
        if (ret == 0) {
            ret = qemuMonitorAddDevice(priv->mon, devstr);
            if (ret < 0) {
6939
                VIR_WARN("qemuMonitorAddDevice failed on %s (%s)",
6940 6941 6942 6943 6944
                         drivestr, devstr);
                /* XXX should call 'drive_del' on error but this does not
                   exist yet */
            }
        }
6945 6946 6947 6948 6949 6950 6951 6952 6953 6954 6955
    } 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));
        }
    }
6956
    qemuDomainObjExitMonitorWithDriver(driver, vm);
6957

6958 6959
    if (ret < 0)
        goto error;
6960

6961
    virDomainDiskInsertPreAlloced(vm->def, disk);
6962

6963 6964 6965
    VIR_FREE(devstr);
    VIR_FREE(drivestr);

6966 6967 6968
    return 0;

error:
6969 6970 6971
    VIR_FREE(devstr);
    VIR_FREE(drivestr);

6972 6973 6974 6975 6976
    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);

6977 6978
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
6979
        driver->securityDriver->domainRestoreSecurityImageLabel(vm, disk) < 0)
6980
        VIR_WARN("Unable to restore security label on %s", disk->src);
6981 6982

    return -1;
6983
}
6984

6985

6986
static int qemudDomainAttachPciControllerDevice(struct qemud_driver *driver,
6987
                                                virDomainObjPtr vm,
6988 6989
                                                virDomainControllerDefPtr controller,
                                                int qemuCmdFlags)
6990
{
6991 6992
    int i;
    int ret = -1;
6993
    const char* type = virDomainControllerTypeToString(controller->type);
6994
    char *devstr = NULL;
6995 6996 6997
    qemuDomainObjPrivatePtr priv = vm->privateData;

    for (i = 0 ; i < vm->def->ncontrollers ; i++) {
6998 6999
        if ((vm->def->controllers[i]->type == controller->type) &&
            (vm->def->controllers[i]->idx == controller->idx)) {
7000 7001 7002
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("target %s:%d already exists"),
                            type, controller->idx);
7003 7004 7005 7006
            return -1;
        }
    }

7007 7008 7009 7010 7011 7012
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuDomainPCIAddressEnsureAddr(priv->pciaddrs, &controller->info) < 0)
            goto cleanup;
        if (qemuAssignDeviceControllerAlias(controller) < 0)
            goto cleanup;
    }
7013

7014
    if (!(devstr = qemuBuildControllerDevStr(controller))) {
7015
        virReportOOMError();
7016 7017 7018
        goto cleanup;
    }

7019
    if (VIR_REALLOC_N(vm->def->controllers, vm->def->ncontrollers+1) < 0) {
7020
        virReportOOMError();
7021
        goto cleanup;
7022 7023 7024
    }

    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7025 7026 7027 7028 7029 7030 7031
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        ret = qemuMonitorAddDevice(priv->mon, devstr);
    } else {
        ret = qemuMonitorAttachPCIDiskController(priv->mon,
                                                 type,
                                                 &controller->info.addr.pci);
    }
7032 7033 7034
    qemuDomainObjExitMonitorWithDriver(driver, vm);

    if (ret == 0) {
7035 7036
        controller->info.type = VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI;
        virDomainControllerInsertPreAlloced(vm->def, controller);
7037 7038
    }

7039
cleanup:
7040 7041 7042 7043 7044 7045
    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");

7046
    VIR_FREE(devstr);
7047 7048 7049
    return ret;
}

7050

7051
static virDomainControllerDefPtr
7052
qemuDomainFindOrCreateSCSIDiskController(struct qemud_driver *driver,
7053
                                         virDomainObjPtr vm,
7054 7055
                                         int controller,
                                         int qemuCmdFlags)
7056 7057 7058 7059 7060 7061 7062 7063 7064 7065 7066 7067 7068
{
    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
7069
    /* No SCSI controller present, for backward compatibility we
7070 7071
     * now hotplug a controller */
    if (VIR_ALLOC(cont) < 0) {
7072
        virReportOOMError();
7073 7074 7075 7076 7077 7078
        return NULL;
    }
    cont->type = VIR_DOMAIN_CONTROLLER_TYPE_SCSI;
    cont->idx = 0;

    VIR_INFO0("No SCSI controller present, hotplugging one");
7079
    if (qemudDomainAttachPciControllerDevice(driver,
7080
                                             vm, cont, qemuCmdFlags) < 0) {
7081 7082 7083
        VIR_FREE(cont);
        return NULL;
    }
7084 7085 7086 7087 7088 7089 7090 7091 7092

    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;
    }

7093 7094 7095
    return cont;
}

7096

7097
static int qemudDomainAttachSCSIDisk(struct qemud_driver *driver,
7098
                                     virDomainObjPtr vm,
7099
                                     virDomainDiskDefPtr disk,
7100 7101 7102 7103
                                     int qemuCmdFlags)
{
    int i;
    qemuDomainObjPrivatePtr priv = vm->privateData;
7104
    virDomainControllerDefPtr cont = NULL;
7105
    char *drivestr = NULL;
7106
    char *devstr = NULL;
7107 7108 7109
    int ret = -1;

    for (i = 0 ; i < vm->def->ndisks ; i++) {
7110
        if (STREQ(vm->def->disks[i]->dst, disk->dst)) {
7111 7112
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("target %s already exists"), disk->dst);
7113
            return -1;
7114 7115 7116
        }
    }

7117 7118 7119

    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityImageLabel &&
7120
        driver->securityDriver->domainSetSecurityImageLabel(vm, disk) < 0)
7121 7122
        return -1;

7123
    /* We should have an address already, so make sure */
7124
    if (disk->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_DRIVE) {
7125 7126 7127
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("unexpected disk address type %s"),
                        virDomainDeviceAddressTypeToString(disk->info.type));
7128
        goto error;
7129 7130
    }

7131 7132 7133
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuAssignDeviceDiskAlias(disk, qemuCmdFlags) < 0)
            goto error;
7134
        if (!(devstr = qemuBuildDriveDevStr(disk)))
7135 7136 7137 7138 7139 7140
            goto error;
    }

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

7141
    for (i = 0 ; i <= disk->info.addr.drive.controller ; i++) {
7142
        cont = qemuDomainFindOrCreateSCSIDiskController(driver, vm, i, qemuCmdFlags);
7143
        if (!cont)
7144
            goto error;
7145 7146
    }

7147 7148 7149 7150 7151
    /* 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);

7152
    if (cont->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI) {
7153 7154
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("SCSI controller %d was missing its PCI address"), cont->idx);
7155
        goto error;
7156 7157 7158
    }

    if (VIR_REALLOC_N(vm->def->disks, vm->def->ndisks+1) < 0) {
7159
        virReportOOMError();
7160
        goto error;
7161 7162 7163
    }

    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7164
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
7165 7166 7167 7168
        ret = qemuMonitorAddDrive(priv->mon, drivestr);
        if (ret == 0) {
            ret = qemuMonitorAddDevice(priv->mon, devstr);
            if (ret < 0) {
7169
                VIR_WARN("qemuMonitorAddDevice failed on %s (%s)",
7170 7171 7172 7173 7174
                         drivestr, devstr);
                /* XXX should call 'drive_del' on error but this does not
                   exist yet */
            }
        }
7175 7176 7177 7178 7179 7180 7181 7182 7183 7184 7185 7186 7187
    } 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));
        }
    }
7188 7189
    qemuDomainObjExitMonitorWithDriver(driver, vm);

7190 7191
    if (ret < 0)
        goto error;
7192

7193
    virDomainDiskInsertPreAlloced(vm->def, disk);
7194 7195

    VIR_FREE(devstr);
7196
    VIR_FREE(drivestr);
7197 7198 7199 7200

    return 0;

error:
7201
    VIR_FREE(devstr);
7202
    VIR_FREE(drivestr);
7203

7204 7205
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
7206
        driver->securityDriver->domainRestoreSecurityImageLabel(vm, disk) < 0)
7207
        VIR_WARN("Unable to restore security label on %s", disk->src);
7208 7209

    return -1;
7210 7211
}

7212

7213
static int qemudDomainAttachUsbMassstorageDevice(struct qemud_driver *driver,
7214
                                                 virDomainObjPtr vm,
7215 7216
                                                 virDomainDiskDefPtr disk,
                                                 int qemuCmdFlags)
7217
{
7218
    qemuDomainObjPrivatePtr priv = vm->privateData;
7219
    int i, ret;
7220 7221
    char *drivestr = NULL;
    char *devstr = NULL;
7222

7223
    for (i = 0 ; i < vm->def->ndisks ; i++) {
7224
        if (STREQ(vm->def->disks[i]->dst, disk->dst)) {
7225 7226
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("target %s already exists"), disk->dst);
7227 7228 7229 7230
            return -1;
        }
    }

7231 7232
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityImageLabel &&
7233
        driver->securityDriver->domainSetSecurityImageLabel(vm, disk) < 0)
7234 7235
        return -1;

7236
    if (!disk->src) {
7237 7238
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("disk source path is missing"));
7239
        goto error;
7240 7241
    }

7242
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
7243 7244
        if (qemuAssignDeviceDiskAlias(disk, qemuCmdFlags) < 0)
            goto error;
7245 7246
        if (!(drivestr = qemuBuildDriveStr(disk, 0, qemuCmdFlags)))
            goto error;
7247
        if (!(devstr = qemuBuildDriveDevStr(disk)))
7248 7249 7250
            goto error;
    }

7251
    if (VIR_REALLOC_N(vm->def->disks, vm->def->ndisks+1) < 0) {
7252
        virReportOOMError();
7253
        goto error;
7254 7255
    }

7256
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7257
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
7258 7259 7260 7261
        ret = qemuMonitorAddDrive(priv->mon, drivestr);
        if (ret == 0) {
            ret = qemuMonitorAddDevice(priv->mon, devstr);
            if (ret < 0) {
7262
                VIR_WARN("qemuMonitorAddDevice failed on %s (%s)",
7263 7264 7265 7266 7267
                         drivestr, devstr);
                /* XXX should call 'drive_del' on error but this does not
                   exist yet */
            }
        }
7268 7269 7270
    } else {
        ret = qemuMonitorAddUSBDisk(priv->mon, disk->src);
    }
7271
    qemuDomainObjExitMonitorWithDriver(driver, vm);
7272

7273 7274 7275
    if (ret < 0)
        goto error;

7276
    virDomainDiskInsertPreAlloced(vm->def, disk);
7277

7278 7279 7280
    VIR_FREE(devstr);
    VIR_FREE(drivestr);

7281 7282 7283
    return 0;

error:
7284 7285 7286
    VIR_FREE(devstr);
    VIR_FREE(drivestr);

7287 7288
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
7289
        driver->securityDriver->domainRestoreSecurityImageLabel(vm, disk) < 0)
7290
        VIR_WARN("Unable to restore security label on %s", disk->src);
7291 7292

    return -1;
7293 7294
}

7295

7296
/* XXX conn required for network -> bridge resolution */
M
Mark McLoughlin 已提交
7297
static int qemudDomainAttachNetDevice(virConnectPtr conn,
7298
                                      struct qemud_driver *driver,
M
Mark McLoughlin 已提交
7299
                                      virDomainObjPtr vm,
7300
                                      virDomainNetDefPtr net,
7301
                                      unsigned long long qemuCmdFlags)
M
Mark McLoughlin 已提交
7302
{
7303
    qemuDomainObjPrivatePtr priv = vm->privateData;
7304
    char *tapfd_name = NULL;
7305
    int tapfd = -1;
7306
    char *nicstr = NULL;
7307
    char *netstr = NULL;
7308
    int ret = -1;
7309
    virDomainDevicePCIAddress guestAddr;
7310
    int vlan;
M
Mark McLoughlin 已提交
7311 7312

    if (!(qemuCmdFlags & QEMUD_CMD_FLAG_HOST_NET_ADD)) {
7313 7314
        qemuReportError(VIR_ERR_NO_SUPPORT, "%s",
                        _("installed qemu version does not support host_net_add"));
M
Mark McLoughlin 已提交
7315 7316 7317 7318 7319
        return -1;
    }

    if (net->type == VIR_DOMAIN_NET_TYPE_BRIDGE ||
        net->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
7320
        if (priv->monConfig->type != VIR_DOMAIN_CHR_TYPE_UNIX) {
7321 7322 7323 7324
            qemuReportError(VIR_ERR_NO_SUPPORT,
                            _("network device type '%s' cannot be attached: "
                              "qemu is not using a unix socket monitor"),
                            virDomainNetTypeToString(net->type));
7325 7326 7327 7328 7329
            return -1;
        }

        if ((tapfd = qemudNetworkIfaceConnect(conn, driver, net, qemuCmdFlags)) < 0)
            return -1;
7330 7331 7332 7333 7334 7335 7336 7337 7338
    } else if (net->type == VIR_DOMAIN_NET_TYPE_DIRECT) {
        if (priv->monConfig->type != VIR_DOMAIN_CHR_TYPE_UNIX) {
            qemuReportError(VIR_ERR_NO_SUPPORT,
                            _("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
7339
        if ((tapfd = qemudPhysIfaceConnect(conn, driver, net,
7340
                                           net->data.direct.linkdev,
7341 7342
                                           net->data.direct.mode,
                                           qemuCmdFlags)) < 0)
7343
            return -1;
M
Mark McLoughlin 已提交
7344 7345
    }

7346 7347
    if (VIR_REALLOC_N(vm->def->nets, vm->def->nnets+1) < 0)
        goto no_memory;
M
Mark McLoughlin 已提交
7348

7349 7350 7351 7352 7353
    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
7354

7355 7356 7357 7358
    if ((qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) &&
        qemuDomainPCIAddressEnsureAddr(priv->pciaddrs, &net->info) < 0)
        goto cleanup;

7359 7360 7361 7362 7363
    if ((qemuCmdFlags & QEMUD_CMD_FLAG_NETDEV) &&
        (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)) {
        vlan = -1;
    } else {
        vlan = qemuDomainNetVLAN(net);
M
Mark McLoughlin 已提交
7364

7365 7366 7367 7368 7369
        if (vlan < 0) {
            qemuReportError(VIR_ERR_NO_SUPPORT, "%s",
                            _("Unable to attach network devices without vlan"));
            goto cleanup;
        }
7370 7371
    }

7372
    if (tapfd != -1) {
7373
        if (virAsprintf(&tapfd_name, "fd-%s", net->info.alias) < 0)
7374 7375
            goto no_memory;

7376
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
7377
        if (qemuMonitorSendFileHandle(priv->mon, tapfd_name, tapfd) < 0) {
7378
            qemuDomainObjExitMonitorWithDriver(driver, vm);
7379
            goto cleanup;
7380
        }
7381
        qemuDomainObjExitMonitorWithDriver(driver, vm);
7382 7383 7384 7385 7386 7387

        if (!virDomainObjIsActive(vm)) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("guest unexpectedly quit"));
            goto cleanup;
        }
7388 7389
    }

7390
    /* FIXME - need to support vhost-net here (5th arg) */
7391 7392 7393 7394 7395 7396 7397 7398 7399 7400
    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
7401

7402
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7403 7404 7405 7406 7407 7408 7409 7410 7411 7412 7413
    if ((qemuCmdFlags & QEMUD_CMD_FLAG_NETDEV) &&
        (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)) {
        if (qemuMonitorAddNetdev(priv->mon, netstr) < 0) {
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            goto try_tapfd_close;
        }
    } else {
        if (qemuMonitorAddHostNetwork(priv->mon, netstr) < 0) {
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            goto try_tapfd_close;
        }
7414
    }
7415
    qemuDomainObjExitMonitorWithDriver(driver, vm);
M
Mark McLoughlin 已提交
7416

7417 7418 7419
    if (tapfd != -1)
        close(tapfd);
    tapfd = -1;
M
Mark McLoughlin 已提交
7420

7421 7422 7423 7424 7425 7426
    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("guest unexpectedly quit"));
        goto cleanup;
    }

7427 7428 7429 7430 7431 7432 7433
    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
7434

7435
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7436 7437 7438 7439 7440 7441 7442 7443 7444 7445 7446 7447 7448
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuMonitorAddDevice(priv->mon, nicstr) < 0) {
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            goto try_remove;
        }
    } else {
        if (qemuMonitorAddPCINetwork(priv->mon, nicstr,
                                     &guestAddr) < 0) {
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            goto try_remove;
        }
        net->info.type = VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI;
        memcpy(&net->info.addr.pci, &guestAddr, sizeof(guestAddr));
7449
    }
7450
    qemuDomainObjExitMonitorWithDriver(driver, vm);
7451

7452
    ret = 0;
M
Mark McLoughlin 已提交
7453 7454 7455

    vm->def->nets[vm->def->nnets++] = net;

7456
cleanup:
7457 7458 7459 7460 7461 7462
    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");

7463 7464
    if (ret != 0)
        virNWFilterTearNWFilter(net);
S
Stefan Berger 已提交
7465

7466 7467 7468 7469 7470
    VIR_FREE(nicstr);
    VIR_FREE(netstr);
    VIR_FREE(tapfd_name);
    if (tapfd != -1)
        close(tapfd);
7471

7472
    return ret;
7473

7474
try_remove:
7475 7476 7477
    if (!priv->mon)
        goto cleanup;

7478
    if (vlan < 0) {
7479 7480 7481 7482 7483 7484 7485
        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)
7486
                VIR_WARN("Failed to remove network backend for netdev %s",
7487 7488 7489 7490
                         netdev_name);
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            VIR_FREE(netdev_name);
        } else {
7491
            VIR_WARN0("Unable to remove network backend");
7492
        }
7493 7494 7495 7496
    } else {
        char *hostnet_name;
        if (virAsprintf(&hostnet_name, "host%s", net->info.alias) < 0)
            goto no_memory;
7497
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
7498
        if (qemuMonitorRemoveHostNetwork(priv->mon, vlan, hostnet_name) < 0)
7499
            VIR_WARN("Failed to remove network backend for vlan %d, net %s",
7500
                     vlan, hostnet_name);
7501
        qemuDomainObjExitMonitorWithDriver(driver, vm);
7502
        VIR_FREE(hostnet_name);
7503
    }
7504
    goto cleanup;
7505

7506
try_tapfd_close:
7507 7508 7509
    if (!priv->mon)
        goto cleanup;

7510
    if (tapfd_name) {
7511
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
7512
        if (qemuMonitorCloseFileHandle(priv->mon, tapfd_name) < 0)
7513
            VIR_WARN("Failed to close tapfd with '%s'", tapfd_name);
7514
        qemuDomainObjExitMonitorWithDriver(driver, vm);
7515
    }
7516

7517 7518
    goto cleanup;

7519
no_memory:
7520
    virReportOOMError();
7521
    goto cleanup;
M
Mark McLoughlin 已提交
7522 7523
}

7524

7525
static int qemudDomainAttachHostPciDevice(struct qemud_driver *driver,
7526
                                          virDomainObjPtr vm,
7527 7528
                                          virDomainHostdevDefPtr hostdev,
                                          int qemuCmdFlags)
7529
{
7530
    qemuDomainObjPrivatePtr priv = vm->privateData;
7531
    pciDevice *pci;
7532
    int ret;
7533
    virDomainDevicePCIAddress guestAddr;
7534
    char *devstr = NULL;
7535 7536

    if (VIR_REALLOC_N(vm->def->hostdevs, vm->def->nhostdevs+1) < 0) {
7537
        virReportOOMError();
7538 7539 7540
        return -1;
    }

7541
    pci = pciGetDevice(hostdev->source.subsys.u.pci.domain,
7542 7543 7544
                       hostdev->source.subsys.u.pci.bus,
                       hostdev->source.subsys.u.pci.slot,
                       hostdev->source.subsys.u.pci.function);
7545
    if (!pci)
7546
        return -1;
7547

7548 7549 7550 7551
    if (!pciDeviceIsAssignable(pci, !driver->relaxedACS) ||
        (hostdev->managed && pciDettachDevice(pci) < 0) ||
        pciResetDevice(pci, driver->activePciHostdevs) < 0) {
        pciFreeDevice(pci);
7552 7553 7554
        return -1;
    }

7555 7556
    if (pciDeviceListAdd(driver->activePciHostdevs, pci) < 0) {
        pciFreeDevice(pci);
7557
        return -1;
7558 7559
    }

7560
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
7561 7562
        if (qemuAssignDeviceHostdevAlias(vm->def, hostdev, -1) < 0)
            goto error;
7563 7564 7565 7566 7567 7568
        if (qemuDomainPCIAddressEnsureAddr(priv->pciaddrs, &hostdev->info) < 0)
            goto error;

        if (!(devstr = qemuBuildPCIHostdevDevStr(hostdev)))
            goto error;
    }
7569

7570
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7571 7572 7573 7574 7575 7576
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)
        ret = qemuMonitorAddDevice(priv->mon, devstr);
    else
        ret = qemuMonitorAddPCIHostDevice(priv->mon,
                                          &hostdev->source.subsys.u.pci,
                                          &guestAddr);
7577
    qemuDomainObjExitMonitorWithDriver(driver, vm);
7578
    if (ret < 0)
7579
        goto error;
7580
    hostdev->info.type = VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI;
7581
    memcpy(&hostdev->info.addr.pci, &guestAddr, sizeof(guestAddr));
7582 7583 7584

    vm->def->hostdevs[vm->def->nhostdevs++] = hostdev;

7585 7586
    VIR_FREE(devstr);

7587
    return 0;
7588 7589

error:
7590 7591 7592 7593 7594
    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");

7595
    VIR_FREE(devstr);
7596
    pciDeviceListDel(driver->activePciHostdevs, pci);
7597 7598

    return -1;
7599 7600
}

7601

7602
static int qemudDomainAttachHostUsbDevice(struct qemud_driver *driver,
M
Mark McLoughlin 已提交
7603
                                          virDomainObjPtr vm,
7604 7605
                                          virDomainHostdevDefPtr hostdev,
                                          int qemuCmdFlags)
7606 7607
{
    int ret;
7608
    qemuDomainObjPrivatePtr priv = vm->privateData;
7609 7610
    char *devstr = NULL;

7611 7612 7613 7614 7615 7616
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuAssignDeviceHostdevAlias(vm->def, hostdev, -1) < 0)
            goto error;
        if (!(devstr = qemuBuildUSBHostdevDevStr(hostdev)))
            goto error;
    }
7617

7618
    if (VIR_REALLOC_N(vm->def->hostdevs, vm->def->nhostdevs+1) < 0) {
7619
        virReportOOMError();
7620
        goto error;
7621
    }
7622

7623
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7624 7625 7626
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)
        ret = qemuMonitorAddDevice(priv->mon, devstr);
    else
7627
        ret = qemuMonitorAddUSBDeviceExact(priv->mon,
7628 7629
                                           hostdev->source.subsys.u.usb.bus,
                                           hostdev->source.subsys.u.usb.device);
7630
    qemuDomainObjExitMonitorWithDriver(driver, vm);
7631 7632 7633 7634
    if (ret < 0)
        goto error;

    vm->def->hostdevs[vm->def->nhostdevs++] = hostdev;
7635

7636
    VIR_FREE(devstr);
7637

7638 7639 7640 7641 7642
    return 0;

error:
    VIR_FREE(devstr);
    return -1;
7643 7644
}

7645

7646
static int qemudDomainAttachHostDevice(struct qemud_driver *driver,
M
Mark McLoughlin 已提交
7647
                                       virDomainObjPtr vm,
7648 7649
                                       virDomainHostdevDefPtr hostdev,
                                       int qemuCmdFlags)
M
Mark McLoughlin 已提交
7650 7651
{
    if (hostdev->mode != VIR_DOMAIN_HOSTDEV_MODE_SUBSYS) {
7652 7653 7654
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        _("hostdev mode '%s' not supported"),
                        virDomainHostdevModeTypeToString(hostdev->mode));
M
Mark McLoughlin 已提交
7655 7656 7657
        return -1;
    }

7658 7659 7660 7661 7662 7663 7664 7665 7666 7667 7668 7669 7670 7671 7672 7673 7674
    /* 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);
    }


7675
    if (driver->securityDriver &&
7676
        driver->securityDriver->domainSetSecurityHostdevLabel &&
7677
        driver->securityDriver->domainSetSecurityHostdevLabel(vm, hostdev) < 0)
7678
        return -1;
M
Mark McLoughlin 已提交
7679 7680

    switch (hostdev->source.subsys.type) {
7681
    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI:
7682
        if (qemudDomainAttachHostPciDevice(driver, vm,
7683
                                           hostdev, qemuCmdFlags) < 0)
7684 7685 7686
            goto error;
        break;

M
Mark McLoughlin 已提交
7687
    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB:
7688
        if (qemudDomainAttachHostUsbDevice(driver, vm,
7689
                                           hostdev, qemuCmdFlags) < 0)
7690 7691 7692
            goto error;
        break;

M
Mark McLoughlin 已提交
7693
    default:
7694 7695 7696
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        _("hostdev subsys type '%s' not supported"),
                        virDomainHostdevSubsysTypeToString(hostdev->source.subsys.type));
7697
        goto error;
M
Mark McLoughlin 已提交
7698
    }
7699 7700 7701 7702 7703 7704

    return 0;

error:
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityHostdevLabel &&
7705
        driver->securityDriver->domainRestoreSecurityHostdevLabel(vm, hostdev) < 0)
7706 7707 7708
        VIR_WARN0("Unable to restore host device labelling on hotplug fail");

    return -1;
M
Mark McLoughlin 已提交
7709 7710
}

7711

7712
static int qemudDomainAttachDevice(virDomainPtr dom,
7713 7714
                                   const char *xml)
{
7715 7716 7717
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    virDomainDeviceDefPtr dev = NULL;
7718
    unsigned long long qemuCmdFlags;
7719
    virCgroupPtr cgroup = NULL;
7720
    int ret = -1;
7721

7722
    qemuDriverLock(driver);
7723
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
7724
    if (!vm) {
7725 7726
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
7727 7728
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
7729
        goto cleanup;
7730 7731
    }

7732 7733 7734
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
7735
    if (!virDomainObjIsActive(vm)) {
7736 7737
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot attach device on inactive domain"));
7738
        goto endjob;
7739 7740
    }

7741
    dev = virDomainDeviceDefParse(driver->caps, vm->def, xml,
7742
                                  VIR_DOMAIN_XML_INACTIVE);
7743
    if (dev == NULL)
7744
        goto endjob;
7745

7746 7747 7748
    if (qemudExtractVersionInfo(vm->def->emulator,
                                NULL,
                                &qemuCmdFlags) < 0)
7749
        goto endjob;
7750

7751
    if (dev->type == VIR_DOMAIN_DEVICE_DISK) {
7752
        if (qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_DEVICES)) {
7753
            if (virCgroupForDomain(driver->cgroup, vm->def->name, &cgroup, 0) !=0 ) {
7754 7755 7756
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("Unable to find cgroup for %s\n"),
                                vm->def->name);
7757
                goto endjob;
7758
            }
7759
            if (qemuSetupDiskCgroup(cgroup, vm, dev->data.disk) < 0)
7760
                goto endjob;
7761 7762
        }

7763
        switch (dev->data.disk->device) {
7764 7765
        case VIR_DOMAIN_DISK_DEVICE_CDROM:
        case VIR_DOMAIN_DISK_DEVICE_FLOPPY:
7766 7767 7768
            ret = qemudDomainChangeEjectableMedia(driver, vm,
                                                  dev->data.disk,
                                                  qemuCmdFlags);
7769 7770
            if (ret == 0)
                dev->data.disk = NULL;
7771
            break;
7772

7773 7774
        case VIR_DOMAIN_DISK_DEVICE_DISK:
            if (dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_USB) {
7775
                ret = qemudDomainAttachUsbMassstorageDevice(driver, vm,
7776
                                                            dev->data.disk, qemuCmdFlags);
7777 7778
                if (ret == 0)
                    dev->data.disk = NULL;
7779
            } else if (dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_VIRTIO) {
7780
                ret = qemudDomainAttachPciDiskDevice(driver, vm,
7781
                                                     dev->data.disk, qemuCmdFlags);
7782 7783
                if (ret == 0)
                    dev->data.disk = NULL;
7784
            } else if (dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_SCSI) {
7785
                ret = qemudDomainAttachSCSIDisk(driver, vm,
7786
                                                dev->data.disk, qemuCmdFlags);
7787 7788
                if (ret == 0)
                    dev->data.disk = NULL;
7789
            } else {
7790 7791 7792
                qemuReportError(VIR_ERR_NO_SUPPORT,
                                _("disk bus '%s' cannot be hotplugged."),
                                virDomainDiskBusTypeToString(dev->data.disk->bus));
7793
                /* fallthrough */
7794 7795
            }
            break;
7796

7797
        default:
7798 7799 7800
            qemuReportError(VIR_ERR_NO_SUPPORT,
                            _("disk device type '%s' cannot be hotplugged"),
                            virDomainDiskDeviceTypeToString(dev->data.disk->device));
7801 7802
            /* Fallthrough */
        }
7803
        if (ret != 0 && cgroup) {
7804 7805 7806
            if (qemuTeardownDiskCgroup(cgroup, vm, dev->data.disk) < 0)
                VIR_WARN("Failed to teardown cgroup for disk path %s",
                         NULLSTR(dev->data.disk->src));
7807
        }
7808 7809
    } else if (dev->type == VIR_DOMAIN_DEVICE_CONTROLLER) {
        if (dev->data.controller->type == VIR_DOMAIN_CONTROLLER_TYPE_SCSI) {
7810
            ret = qemudDomainAttachPciControllerDevice(driver, vm,
7811
                                                       dev->data.controller, qemuCmdFlags);
7812 7813
            if (ret == 0)
                dev->data.controller = NULL;
7814
        } else {
7815 7816 7817
            qemuReportError(VIR_ERR_NO_SUPPORT,
                            _("disk controller bus '%s' cannot be hotplugged."),
                            virDomainControllerTypeToString(dev->data.controller->type));
7818 7819
            /* fallthrough */
        }
M
Mark McLoughlin 已提交
7820
    } else if (dev->type == VIR_DOMAIN_DEVICE_NET) {
7821 7822
        ret = qemudDomainAttachNetDevice(dom->conn, driver, vm,
                                         dev->data.net, qemuCmdFlags);
7823 7824
        if (ret == 0)
            dev->data.net = NULL;
M
Mark McLoughlin 已提交
7825
    } else if (dev->type == VIR_DOMAIN_DEVICE_HOSTDEV) {
7826
        ret = qemudDomainAttachHostDevice(driver, vm,
7827
                                          dev->data.hostdev, qemuCmdFlags);
7828 7829
        if (ret == 0)
            dev->data.hostdev = NULL;
7830
    } else {
7831 7832 7833
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        _("device type '%s' cannot be attached"),
                        virDomainDeviceTypeToString(dev->type));
7834
        goto endjob;
7835 7836
    }

7837
    if (!ret && virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
7838 7839
        ret = -1;

7840
endjob:
7841 7842
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
7843

7844
cleanup:
7845 7846 7847
    if (cgroup)
        virCgroupFree(&cgroup);

7848
    virDomainDeviceDefFree(dev);
7849 7850
    if (vm)
        virDomainObjUnlock(vm);
7851
    qemuDriverUnlock(driver);
7852 7853 7854
    return ret;
}

7855 7856 7857 7858
static int qemudDomainAttachDeviceFlags(virDomainPtr dom,
                                        const char *xml,
                                        unsigned int flags) {
    if (flags & VIR_DOMAIN_DEVICE_MODIFY_CONFIG) {
7859 7860
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot modify the persistent configuration of a domain"));
7861 7862 7863 7864 7865 7866
        return -1;
    }

    return qemudDomainAttachDevice(dom, xml);
}

7867

7868 7869 7870 7871 7872 7873 7874 7875 7876 7877 7878 7879 7880 7881 7882 7883 7884 7885 7886 7887 7888 7889 7890 7891 7892 7893 7894 7895 7896 7897 7898 7899 7900 7901 7902 7903 7904 7905 7906 7907 7908 7909 7910 7911 7912 7913 7914 7915 7916 7917 7918 7919 7920 7921 7922 7923 7924 7925 7926 7927 7928 7929 7930 7931 7932 7933 7934 7935 7936 7937 7938 7939 7940 7941 7942 7943 7944
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;
}


7945 7946 7947 7948 7949 7950 7951 7952 7953 7954 7955
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;

7956 7957 7958 7959
    virCheckFlags(VIR_DOMAIN_DEVICE_MODIFY_CURRENT |
                  VIR_DOMAIN_DEVICE_MODIFY_LIVE |
                  VIR_DOMAIN_DEVICE_MODIFY_CONFIG, -1);

7960 7961 7962 7963 7964 7965 7966 7967 7968 7969 7970 7971 7972 7973 7974 7975 7976 7977 7978 7979 7980 7981 7982 7983 7984 7985 7986 7987 7988 7989 7990 7991 7992 7993 7994 7995 7996 7997 7998 7999 8000 8001 8002 8003
    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;
            }
8004
            if (qemuSetupDiskCgroup(cgroup, vm, dev->data.disk) < 0)
8005 8006 8007 8008 8009 8010
                goto endjob;
        }

        switch (dev->data.disk->device) {
        case VIR_DOMAIN_DISK_DEVICE_CDROM:
        case VIR_DOMAIN_DISK_DEVICE_FLOPPY:
8011 8012 8013
            ret = qemudDomainChangeEjectableMedia(driver, vm,
                                                  dev->data.disk,
                                                  qemuCmdFlags);
8014 8015 8016 8017 8018 8019 8020 8021 8022 8023 8024 8025 8026
            if (ret == 0)
                dev->data.disk = NULL;
            break;


        default:
            qemuReportError(VIR_ERR_NO_SUPPORT,
                            _("disk bus '%s' cannot be updated."),
                            virDomainDiskBusTypeToString(dev->data.disk->bus));
            break;
        }

        if (ret != 0 && cgroup) {
8027 8028 8029
            if (qemuTeardownDiskCgroup(cgroup, vm, dev->data.disk) < 0)
                VIR_WARN("Failed to teardown cgroup for disk path %s",
                         NULLSTR(dev->data.disk->src));
8030 8031 8032
        }
        break;

8033 8034 8035 8036
    case VIR_DOMAIN_DEVICE_GRAPHICS:
        ret = qemuDomainChangeGraphics(driver, vm, dev->data.graphics);
        break;

8037 8038 8039 8040 8041 8042 8043 8044 8045 8046 8047 8048 8049 8050 8051 8052 8053 8054 8055 8056 8057 8058 8059 8060 8061 8062
    default:
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        _("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
8063
static inline int qemudFindDisk(virDomainDefPtr def, const char *dst)
W
Wolfgang Mauerer 已提交
8064 8065 8066 8067 8068 8069 8070 8071 8072 8073 8074 8075
{
    int i;

    for (i = 0 ; i < def->ndisks ; i++) {
        if (STREQ(def->disks[i]->dst, dst)) {
            return i;
        }
    }

    return -1;
}

E
Eric Blake 已提交
8076
static inline void qemudShrinkDisks(virDomainDefPtr def, size_t i)
W
Wolfgang Mauerer 已提交
8077 8078 8079 8080 8081 8082 8083 8084 8085 8086 8087 8088 8089 8090 8091 8092
{
    if (def->ndisks > 1) {
        memmove(def->disks + i,
                def->disks + i + 1,
                sizeof(*def->disks) *
                (def->ndisks - (i + 1)));
        def->ndisks--;
        if (VIR_REALLOC_N(def->disks, def->ndisks) < 0) {
            /* ignore, harmless */
        }
    } else {
        VIR_FREE(def->disks);
        def->ndisks = 0;
    }
}

8093
static int qemudDomainDetachPciDiskDevice(struct qemud_driver *driver,
8094
                                          virDomainObjPtr vm,
8095 8096
                                          virDomainDeviceDefPtr dev,
                                          unsigned long long qemuCmdFlags)
8097 8098 8099
{
    int i, ret = -1;
    virDomainDiskDefPtr detach = NULL;
8100
    qemuDomainObjPrivatePtr priv = vm->privateData;
8101
    virCgroupPtr cgroup = NULL;
8102

W
Wolfgang Mauerer 已提交
8103
    i = qemudFindDisk(vm->def, dev->data.disk->dst);
8104

W
Wolfgang Mauerer 已提交
8105
    if (i < 0) {
8106 8107
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("disk %s not found"), dev->data.disk->dst);
8108
        goto cleanup;
8109 8110
    }

W
Wolfgang Mauerer 已提交
8111 8112
    detach = vm->def->disks[i];

8113 8114 8115 8116 8117 8118 8119 8120 8121
    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;
        }
    }

8122 8123
    if (!virDomainDeviceAddressIsValid(&detach->info,
                                       VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI)) {
8124 8125
        qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
                        _("device cannot be detached without a PCI address"));
8126
        goto cleanup;
8127 8128
    }

8129
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
8130 8131 8132 8133 8134 8135 8136 8137 8138 8139 8140
    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;
        }
8141
    }
8142
    qemuDomainObjExitMonitorWithDriver(driver, vm);
8143

W
Wolfgang Mauerer 已提交
8144 8145
    qemudShrinkDisks(vm->def, i);

8146 8147
    virDomainDiskDefFree(detach);

8148 8149 8150 8151 8152
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
        driver->securityDriver->domainRestoreSecurityImageLabel(vm, dev->data.disk) < 0)
        VIR_WARN("Unable to restore security label on %s", dev->data.disk->src);

8153 8154 8155 8156 8157 8158
    if (cgroup != NULL) {
        if (qemuTeardownDiskCgroup(cgroup, vm, dev->data.disk) < 0)
            VIR_WARN("Failed to teardown cgroup for disk path %s",
                     NULLSTR(dev->data.disk->src));
    }

8159 8160 8161 8162 8163 8164 8165 8166 8167 8168 8169 8170 8171 8172
    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;
8173
    virCgroupPtr cgroup = NULL;
8174 8175 8176 8177 8178 8179 8180 8181 8182 8183

    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)) {
8184
        qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
8185 8186 8187 8188 8189 8190
                        _("Underlying qemu does not support SCSI disk removal"));
        goto cleanup;
    }

    detach = vm->def->disks[i];

8191 8192 8193 8194 8195 8196 8197 8198 8199
    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;
        }
    }

8200 8201 8202 8203 8204 8205 8206 8207 8208
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
    if (qemuMonitorDelDevice(priv->mon, detach->info.alias) < 0) {
        qemuDomainObjExitMonitor(vm);
        goto cleanup;
    }
    qemuDomainObjExitMonitorWithDriver(driver, vm);

    qemudShrinkDisks(vm->def, i);

8209
    virDomainDiskDefFree(detach);
8210

8211 8212
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
8213
        driver->securityDriver->domainRestoreSecurityImageLabel(vm, dev->data.disk) < 0)
8214 8215
        VIR_WARN("Unable to restore security label on %s", dev->data.disk->src);

8216 8217 8218 8219 8220 8221
    if (cgroup != NULL) {
        if (qemuTeardownDiskCgroup(cgroup, vm, dev->data.disk) < 0)
            VIR_WARN("Failed to teardown cgroup for disk path %s",
                     NULLSTR(dev->data.disk->src));
    }

8222
    ret = 0;
8223 8224

cleanup:
8225 8226 8227
    return ret;
}

8228
static int qemudDomainDetachPciControllerDevice(struct qemud_driver *driver,
8229
                                                virDomainObjPtr vm,
8230 8231
                                                virDomainDeviceDefPtr dev,
                                                unsigned long long qemuCmdFlags)
8232 8233 8234 8235 8236 8237 8238 8239 8240 8241 8242 8243 8244 8245
{
    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) {
8246 8247 8248 8249
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("disk controller %s:%d not found"),
                        virDomainControllerTypeToString(dev->data.controller->type),
                        dev->data.controller->idx);
8250 8251 8252 8253 8254
        goto cleanup;
    }

    if (!virDomainDeviceAddressIsValid(&detach->info,
                                       VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI)) {
8255 8256
        qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
                        _("device cannot be detached without a PCI address"));
8257 8258 8259
        goto cleanup;
    }

8260 8261 8262 8263 8264
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuAssignDeviceControllerAlias(detach) < 0)
            goto cleanup;
    }

8265
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
8266 8267 8268 8269 8270 8271 8272 8273 8274 8275 8276
    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;
        }
8277 8278 8279 8280 8281 8282 8283 8284 8285 8286 8287 8288 8289 8290 8291 8292
    }
    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;
    }
8293 8294 8295 8296 8297

    if (qemuDomainPCIAddressReleaseAddr(priv->pciaddrs, &detach->info) < 0) {
        VIR_WARN0("Unable to release PCI address on controller");
    }

8298 8299 8300 8301 8302 8303 8304 8305
    virDomainControllerDefFree(detach);

    ret = 0;

cleanup:
    return ret;
}

8306
static int
8307
qemudDomainDetachNetDevice(struct qemud_driver *driver,
8308
                           virDomainObjPtr vm,
8309 8310
                           virDomainDeviceDefPtr dev,
                           unsigned long long qemuCmdFlags)
8311 8312 8313
{
    int i, ret = -1;
    virDomainNetDefPtr detach = NULL;
8314
    qemuDomainObjPrivatePtr priv = vm->privateData;
8315 8316
    int vlan;
    char *hostnet_name = NULL;
8317 8318 8319 8320 8321 8322 8323 8324 8325 8326 8327

    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) {
8328 8329 8330 8331 8332
        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]);
8333 8334 8335
        goto cleanup;
    }

8336 8337
    if (!virDomainDeviceAddressIsValid(&detach->info,
                                       VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI)) {
8338 8339
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("device cannot be detached without a PCI address"));
8340 8341 8342
        goto cleanup;
    }

8343
    if ((vlan = qemuDomainNetVLAN(detach)) < 0) {
8344 8345
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("unable to determine original VLAN"));
8346 8347 8348 8349
        goto cleanup;
    }

    if (virAsprintf(&hostnet_name, "host%s", detach->info.alias) < 0) {
8350
        virReportOOMError();
8351 8352 8353
        goto cleanup;
    }

8354
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
8355 8356 8357 8358 8359 8360 8361 8362 8363 8364 8365
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuMonitorDelDevice(priv->mon, detach->info.alias) < 0) {
            qemuDomainObjExitMonitor(vm);
            goto cleanup;
        }
    } else {
        if (qemuMonitorRemovePCIDevice(priv->mon,
                                       &detach->info.addr.pci) < 0) {
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            goto cleanup;
        }
8366
    }
8367

8368 8369 8370 8371 8372 8373 8374 8375 8376 8377 8378
    if ((qemuCmdFlags & QEMUD_CMD_FLAG_NETDEV) &&
        (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)) {
        if (qemuMonitorRemoveNetdev(priv->mon, hostnet_name) < 0) {
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            goto cleanup;
        }
    } else {
        if (qemuMonitorRemoveHostNetwork(priv->mon, vlan, hostnet_name) < 0) {
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            goto cleanup;
        }
8379
    }
8380
    qemuDomainObjExitMonitorWithDriver(driver, vm);
8381

8382 8383
    virNWFilterTearNWFilter(detach);

8384 8385
#if WITH_MACVTAP
    if (detach->type == VIR_DOMAIN_NET_TYPE_DIRECT) {
S
Stefan Berger 已提交
8386 8387
        if (detach->ifname)
            delMacvtap(detach->ifname);
8388 8389 8390
    }
#endif

8391
    if ((driver->macFilter) && (detach->ifname != NULL)) {
8392
        if ((errno = networkDisallowMacOnPort(driver,
8393 8394
                                              detach->ifname,
                                              detach->mac))) {
8395
            virReportSystemError(errno,
8396 8397 8398 8399 8400
             _("failed to remove ebtables rule on  '%s'"),
                                 detach->ifname);
        }
    }

8401 8402 8403 8404 8405 8406 8407 8408 8409 8410
    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 {
8411
        VIR_FREE(vm->def->nets);
8412
        vm->def->nnets = 0;
8413
    }
8414
    virDomainNetDefFree(detach);
8415

8416 8417 8418
    ret = 0;

cleanup:
8419
    VIR_FREE(hostnet_name);
8420 8421 8422
    return ret;
}

8423
static int qemudDomainDetachHostPciDevice(struct qemud_driver *driver,
8424
                                          virDomainObjPtr vm,
8425 8426
                                          virDomainDeviceDefPtr dev,
                                          unsigned long long qemuCmdFlags)
8427
{
8428
    virDomainHostdevDefPtr detach = NULL;
8429
    qemuDomainObjPrivatePtr priv = vm->privateData;
8430
    int i, ret;
8431
    pciDevice *pci;
8432 8433

    for (i = 0 ; i < vm->def->nhostdevs ; i++) {
8434 8435 8436 8437
        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;

8438 8439 8440 8441 8442 8443 8444 8445 8446 8447 8448 8449 8450 8451 8452
        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) {
8453 8454 8455 8456 8457 8458
        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);
8459 8460 8461
        return -1;
    }

8462 8463
    if (!virDomainDeviceAddressIsValid(&detach->info,
                                       VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI)) {
8464 8465
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("device cannot be detached without a PCI address"));
8466 8467 8468
        return -1;
    }

8469
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
8470 8471 8472 8473 8474 8475 8476 8477 8478 8479 8480
    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;
        }
8481
    }
8482
    qemuDomainObjExitMonitorWithDriver(driver, vm);
8483 8484 8485

    ret = 0;

8486
    pci = pciGetDevice(detach->source.subsys.u.pci.domain,
8487 8488 8489 8490 8491 8492
                       detach->source.subsys.u.pci.bus,
                       detach->source.subsys.u.pci.slot,
                       detach->source.subsys.u.pci.function);
    if (!pci)
        ret = -1;
    else {
8493
        pciDeviceSetManaged(pci, detach->managed);
8494 8495
        pciDeviceListDel(driver->activePciHostdevs, pci);
        if (pciResetDevice(pci, driver->activePciHostdevs) < 0)
8496
            ret = -1;
8497
        qemudReattachManagedDevice(pci);
8498
        pciFreeDevice(pci);
8499 8500
    }

8501 8502 8503 8504 8505 8506 8507 8508 8509 8510 8511 8512
    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;
8513
    }
8514
    virDomainHostdevDefFree(detach);
8515 8516 8517 8518

    return ret;
}

8519 8520 8521 8522 8523 8524 8525 8526 8527 8528 8529 8530 8531 8532 8533 8534
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;
8535 8536 8537 8538 8539 8540 8541 8542 8543 8544 8545 8546 8547 8548 8549 8550
        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;
            }
8551 8552 8553 8554 8555 8556 8557 8558 8559 8560 8561 8562 8563 8564 8565 8566 8567 8568 8569 8570 8571 8572 8573 8574 8575 8576 8577 8578 8579 8580 8581 8582 8583 8584 8585 8586 8587 8588 8589 8590 8591 8592 8593 8594 8595 8596 8597 8598 8599 8600
        }
    }

    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;
}

8601
static int qemudDomainDetachHostDevice(struct qemud_driver *driver,
8602
                                       virDomainObjPtr vm,
8603 8604
                                       virDomainDeviceDefPtr dev,
                                       unsigned long long qemuCmdFlags)
8605 8606 8607 8608 8609
{
    virDomainHostdevDefPtr hostdev = dev->data.hostdev;
    int ret;

    if (hostdev->mode != VIR_DOMAIN_HOSTDEV_MODE_SUBSYS) {
8610 8611 8612
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        _("hostdev mode '%s' not supported"),
                        virDomainHostdevModeTypeToString(hostdev->mode));
8613 8614 8615 8616 8617
        return -1;
    }

    switch (hostdev->source.subsys.type) {
    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI:
8618
        ret = qemudDomainDetachHostPciDevice(driver, vm, dev, qemuCmdFlags);
8619
        break;
8620 8621 8622
    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB:
        ret = qemudDomainDetachHostUsbDevice(driver, vm, dev, qemuCmdFlags);
        break;
8623
    default:
8624 8625 8626
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        _("hostdev subsys type '%s' not supported"),
                        virDomainHostdevSubsysTypeToString(hostdev->source.subsys.type));
8627 8628 8629
        return -1;
    }

8630
    if (driver->securityDriver &&
8631
        driver->securityDriver->domainRestoreSecurityHostdevLabel &&
8632
        driver->securityDriver->domainRestoreSecurityHostdevLabel(vm, dev->data.hostdev) < 0)
8633
        VIR_WARN0("Failed to restore host device labelling");
8634

8635 8636 8637
    return ret;
}

8638 8639
static int qemudDomainDetachDevice(virDomainPtr dom,
                                   const char *xml) {
8640 8641
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
8642
    unsigned long long qemuCmdFlags;
8643 8644
    virDomainDeviceDefPtr dev = NULL;
    int ret = -1;
8645

8646
    qemuDriverLock(driver);
8647
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
8648
    if (!vm) {
8649 8650
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
8651 8652
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
8653
        goto cleanup;
8654 8655
    }

8656 8657 8658
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
8659
    if (!virDomainObjIsActive(vm)) {
8660 8661
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot detach device on inactive domain"));
8662
        goto endjob;
8663 8664
    }

8665
    dev = virDomainDeviceDefParse(driver->caps, vm->def, xml,
8666
                                  VIR_DOMAIN_XML_INACTIVE);
8667
    if (dev == NULL)
8668
        goto endjob;
8669

8670 8671 8672 8673
    if (qemudExtractVersionInfo(vm->def->emulator,
                                NULL,
                                &qemuCmdFlags) < 0)
        goto endjob;
8674 8675

    if (dev->type == VIR_DOMAIN_DEVICE_DISK &&
8676 8677 8678 8679 8680 8681 8682 8683 8684 8685 8686 8687
        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 {
            qemuReportError(VIR_ERR_NO_SUPPORT, "%s",
                            _("This type of disk cannot be hot unplugged"));
        }
8688
    } else if (dev->type == VIR_DOMAIN_DEVICE_NET) {
8689
        ret = qemudDomainDetachNetDevice(driver, vm, dev, qemuCmdFlags);
8690 8691
    } else if (dev->type == VIR_DOMAIN_DEVICE_CONTROLLER) {
        if (dev->data.controller->type == VIR_DOMAIN_CONTROLLER_TYPE_SCSI) {
8692 8693
            ret = qemudDomainDetachPciControllerDevice(driver, vm, dev,
                                                       qemuCmdFlags);
8694
        } else {
8695 8696 8697
            qemuReportError(VIR_ERR_NO_SUPPORT,
                            _("disk controller bus '%s' cannot be hotunplugged."),
                            virDomainControllerTypeToString(dev->data.controller->type));
8698 8699
            /* fallthrough */
        }
8700
    } else if (dev->type == VIR_DOMAIN_DEVICE_HOSTDEV) {
8701
        ret = qemudDomainDetachHostDevice(driver, vm, dev, qemuCmdFlags);
8702
    } else {
8703 8704
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        "%s", _("This type of device cannot be hot unplugged"));
8705
    }
8706

8707
    if (!ret && virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
8708 8709
        ret = -1;

8710
endjob:
8711 8712
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
8713

8714 8715
cleanup:
    virDomainDeviceDefFree(dev);
8716 8717
    if (vm)
        virDomainObjUnlock(vm);
8718
    qemuDriverUnlock(driver);
8719 8720 8721
    return ret;
}

8722 8723 8724 8725
static int qemudDomainDetachDeviceFlags(virDomainPtr dom,
                                        const char *xml,
                                        unsigned int flags) {
    if (flags & VIR_DOMAIN_DEVICE_MODIFY_CONFIG) {
8726 8727
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot modify the persistent configuration of a domain"));
8728 8729 8730 8731 8732 8733
        return -1;
    }

    return qemudDomainDetachDevice(dom, xml);
}

8734
static int qemudDomainGetAutostart(virDomainPtr dom,
8735
                                   int *autostart) {
8736 8737 8738
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
8739

8740
    qemuDriverLock(driver);
8741
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
8742 8743
    qemuDriverUnlock(driver);

8744
    if (!vm) {
8745 8746
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
8747 8748
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
8749
        goto cleanup;
8750 8751 8752
    }

    *autostart = vm->autostart;
8753
    ret = 0;
8754

8755
cleanup:
8756 8757
    if (vm)
        virDomainObjUnlock(vm);
8758
    return ret;
8759 8760
}

8761
static int qemudDomainSetAutostart(virDomainPtr dom,
8762
                                   int autostart) {
8763 8764
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
8765 8766
    char *configFile = NULL, *autostartLink = NULL;
    int ret = -1;
8767

8768
    qemuDriverLock(driver);
8769
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
8770

8771
    if (!vm) {
8772 8773
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
8774 8775
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
8776
        goto cleanup;
8777 8778
    }

8779
    if (!vm->persistent) {
8780 8781
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("cannot set autostart for transient domain"));
8782
        goto cleanup;
8783 8784
    }

8785 8786
    autostart = (autostart != 0);

8787
    if (vm->autostart != autostart) {
8788
        if ((configFile = virDomainConfigFile(driver->configDir, vm->def->name)) == NULL)
8789
            goto cleanup;
8790
        if ((autostartLink = virDomainConfigFile(driver->autostartDir, vm->def->name)) == NULL)
8791
            goto cleanup;
8792

8793 8794
        if (autostart) {
            int err;
8795

8796
            if ((err = virFileMakePath(driver->autostartDir))) {
8797
                virReportSystemError(err,
8798 8799
                                     _("cannot create autostart directory %s"),
                                     driver->autostartDir);
8800 8801
                goto cleanup;
            }
8802

8803
            if (symlink(configFile, autostartLink) < 0) {
8804
                virReportSystemError(errno,
8805 8806
                                     _("Failed to create symlink '%s to '%s'"),
                                     autostartLink, configFile);
8807 8808 8809 8810
                goto cleanup;
            }
        } else {
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
8811
                virReportSystemError(errno,
8812 8813
                                     _("Failed to delete symlink '%s'"),
                                     autostartLink);
8814 8815
                goto cleanup;
            }
8816 8817
        }

8818
        vm->autostart = autostart;
8819
    }
8820
    ret = 0;
8821

8822 8823 8824
cleanup:
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
8825 8826
    if (vm)
        virDomainObjUnlock(vm);
8827
    qemuDriverUnlock(driver);
8828
    return ret;
8829 8830
}

8831 8832 8833 8834 8835

static char *qemuGetSchedulerType(virDomainPtr dom,
                                  int *nparams)
{
    struct qemud_driver *driver = dom->conn->privateData;
8836
    char *ret = NULL;
8837

8838
    qemuDriverLock(driver);
8839
    if (!qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_CPU)) {
8840 8841
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        __FUNCTION__);
8842
        goto cleanup;
8843 8844 8845 8846 8847 8848 8849
    }

    if (nparams)
        *nparams = 1;

    ret = strdup("posix");
    if (!ret)
8850
        virReportOOMError();
8851 8852 8853

cleanup:
    qemuDriverUnlock(driver);
8854 8855 8856 8857 8858 8859 8860 8861 8862 8863 8864 8865 8866
    return ret;
}

static int qemuSetSchedulerParameters(virDomainPtr dom,
                                      virSchedParameterPtr params,
                                      int nparams)
{
    struct qemud_driver *driver = dom->conn->privateData;
    int i;
    virCgroupPtr group = NULL;
    virDomainObjPtr vm = NULL;
    int ret = -1;

8867
    qemuDriverLock(driver);
8868
    if (!qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_CPU)) {
8869 8870
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        __FUNCTION__);
8871
        goto cleanup;
8872 8873 8874 8875 8876
    }

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

    if (vm == NULL) {
8877 8878
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("No such domain %s"), dom->uuid);
8879 8880 8881 8882
        goto cleanup;
    }

    if (virCgroupForDomain(driver->cgroup, vm->def->name, &group, 0) != 0) {
8883 8884
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("cannot find cgroup for domain %s"), vm->def->name);
8885 8886 8887 8888 8889 8890 8891 8892 8893
        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) {
8894 8895
                qemuReportError(VIR_ERR_INVALID_ARG, "%s",
                                _("invalid type for cpu_shares tunable, expected a 'ullong'"));
8896 8897 8898 8899 8900
                goto cleanup;
            }

            rc = virCgroupSetCpuShares(group, params[i].value.ul);
            if (rc != 0) {
8901
                virReportSystemError(-rc, "%s",
8902 8903 8904 8905
                                     _("unable to set cpu shares tunable"));
                goto cleanup;
            }
        } else {
8906 8907
            qemuReportError(VIR_ERR_INVALID_ARG,
                            _("Invalid parameter `%s'"), param->field);
8908 8909 8910 8911 8912 8913 8914 8915 8916
            goto cleanup;
        }
    }
    ret = 0;

cleanup:
    virCgroupFree(&group);
    if (vm)
        virDomainObjUnlock(vm);
8917
    qemuDriverUnlock(driver);
8918 8919 8920 8921 8922 8923 8924 8925 8926 8927 8928 8929 8930 8931
    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;

8932
    qemuDriverLock(driver);
8933
    if (!qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_CPU)) {
8934 8935
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        __FUNCTION__);
8936
        goto cleanup;
8937 8938 8939
    }

    if ((*nparams) != 1) {
8940 8941
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("Invalid parameter count"));
8942
        goto cleanup;
8943 8944 8945 8946 8947
    }

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

    if (vm == NULL) {
8948 8949
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("No such domain %s"), dom->uuid);
8950 8951 8952 8953
        goto cleanup;
    }

    if (virCgroupForDomain(driver->cgroup, vm->def->name, &group, 0) != 0) {
8954 8955
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("cannot find cgroup for domain %s"), vm->def->name);
8956 8957 8958 8959 8960
        goto cleanup;
    }

    rc = virCgroupGetCpuShares(group, &val);
    if (rc != 0) {
8961
        virReportSystemError(-rc, "%s",
8962 8963 8964 8965 8966
                             _("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
8967
    if (virStrcpyStatic(params[0].field, "cpu_shares") == NULL) {
8968 8969
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("Field cpu_shares too long for destination"));
C
Chris Lalancette 已提交
8970 8971
        goto cleanup;
    }
8972 8973 8974 8975 8976 8977 8978

    ret = 0;

cleanup:
    virCgroupFree(&group);
    if (vm)
        virDomainObjUnlock(vm);
8979
    qemuDriverUnlock(driver);
8980 8981 8982 8983
    return ret;
}


8984 8985 8986 8987 8988 8989 8990 8991 8992
/* 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)
{
8993
    struct qemud_driver *driver = dom->conn->privateData;
8994
    int i, ret = -1;
8995
    virDomainObjPtr vm;
8996
    virDomainDiskDefPtr disk = NULL;
8997

8998
    qemuDriverLock(driver);
8999
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
9000
    qemuDriverUnlock(driver);
9001
    if (!vm) {
9002 9003
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
9004 9005
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
9006
        goto cleanup;
9007
    }
9008 9009 9010 9011

    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
9012
    if (!virDomainObjIsActive (vm)) {
9013 9014
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
9015
        goto endjob;
9016 9017
    }

9018 9019 9020 9021 9022 9023 9024 9025
    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (STREQ(path, vm->def->disks[i]->dst)) {
            disk = vm->def->disks[i];
            break;
        }
    }

    if (!disk) {
9026 9027
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("invalid path: %s"), path);
9028
        goto endjob;
9029 9030
    }

9031
    if (!disk->info.alias) {
9032 9033
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("missing disk device alias name for %s"), disk->dst);
9034
        goto endjob;
9035
    }
9036

9037
    qemuDomainObjPrivatePtr priv = vm->privateData;
9038 9039
    qemuDomainObjEnterMonitor(vm);
    ret = qemuMonitorGetBlockStatsInfo(priv->mon,
9040
                                       disk->info.alias,
9041 9042 9043 9044 9045 9046
                                       &stats->rd_req,
                                       &stats->rd_bytes,
                                       &stats->wr_req,
                                       &stats->wr_bytes,
                                       &stats->errs);
    qemuDomainObjExitMonitor(vm);
9047

9048
endjob:
9049 9050
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
9051

9052
cleanup:
9053 9054
    if (vm)
        virDomainObjUnlock(vm);
9055
    return ret;
9056 9057
}

9058
#ifdef __linux__
9059 9060 9061 9062 9063
static int
qemudDomainInterfaceStats (virDomainPtr dom,
                           const char *path,
                           struct _virDomainInterfaceStats *stats)
{
9064 9065
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
9066
    int i;
9067
    int ret = -1;
9068

9069
    qemuDriverLock(driver);
9070
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
9071 9072
    qemuDriverUnlock(driver);

9073
    if (!vm) {
9074 9075
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
9076 9077
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
9078
        goto cleanup;
9079 9080
    }

D
Daniel P. Berrange 已提交
9081
    if (!virDomainObjIsActive(vm)) {
9082 9083
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
9084
        goto cleanup;
9085 9086 9087
    }

    /* Check the path is one of the domain's network interfaces. */
9088 9089
    for (i = 0 ; i < vm->def->nnets ; i++) {
        if (vm->def->nets[i]->ifname &&
9090 9091 9092 9093
            STREQ (vm->def->nets[i]->ifname, path)) {
            ret = 0;
            break;
        }
9094 9095
    }

9096
    if (ret == 0)
9097
        ret = linuxDomainInterfaceStats(path, stats);
9098
    else
9099 9100
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("invalid path, '%s' is not a known interface"), path);
9101

9102
cleanup:
9103 9104
    if (vm)
        virDomainObjUnlock(vm);
9105 9106
    return ret;
}
9107
#else
9108 9109 9110 9111
static int
qemudDomainInterfaceStats (virDomainPtr dom,
                           const char *path ATTRIBUTE_UNUSED,
                           struct _virDomainInterfaceStats *stats ATTRIBUTE_UNUSED)
9112 9113
    qemuReportError(VIR_ERR_NO_SUPPORT,
                    "%s", __FUNCTION__);
9114 9115
    return -1;
}
9116
#endif
9117

9118 9119 9120 9121 9122 9123 9124 9125 9126 9127 9128 9129 9130 9131 9132 9133
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);
9134 9135
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
9136 9137 9138
        goto cleanup;
    }

9139 9140 9141
    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

9142 9143 9144
    if (virDomainObjIsActive(vm)) {
        qemuDomainObjPrivatePtr priv = vm->privateData;
        qemuDomainObjEnterMonitor(vm);
9145
        ret = qemuMonitorGetMemoryStats(priv->mon, stats, nr_stats);
9146 9147
        qemuDomainObjExitMonitor(vm);
    } else {
9148 9149
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
9150 9151
    }

9152 9153 9154
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;

9155 9156 9157 9158 9159 9160
cleanup:
    if (vm)
        virDomainObjUnlock(vm);
    return ret;
}

9161 9162 9163 9164 9165 9166 9167
static int
qemudDomainBlockPeek (virDomainPtr dom,
                      const char *path,
                      unsigned long long offset, size_t size,
                      void *buffer,
                      unsigned int flags ATTRIBUTE_UNUSED)
{
9168 9169 9170
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int fd = -1, ret = -1, i;
9171

9172
    qemuDriverLock(driver);
9173
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
9174 9175
    qemuDriverUnlock(driver);

9176
    if (!vm) {
9177 9178
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
9179 9180
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
9181
        goto cleanup;
9182 9183 9184
    }

    if (!path || path[0] == '\0') {
9185 9186
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("NULL or empty path"));
9187
        goto cleanup;
9188 9189 9190
    }

    /* Check the path belongs to this domain. */
9191 9192
    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (vm->def->disks[i]->src != NULL &&
9193 9194 9195 9196
            STREQ (vm->def->disks[i]->src, path)) {
            ret = 0;
            break;
        }
9197 9198
    }

9199 9200 9201 9202 9203
    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) {
9204 9205
            virReportSystemError(errno,
                                 _("%s: failed to open"), path);
9206 9207
            goto cleanup;
        }
9208

9209 9210 9211 9212 9213 9214
        /* 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) {
9215 9216
            virReportSystemError(errno,
                                 _("%s: failed to seek or read"), path);
9217 9218 9219 9220 9221
            goto cleanup;
        }

        ret = 0;
    } else {
9222 9223
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("invalid path"));
9224 9225
    }

9226 9227 9228
cleanup:
    if (fd >= 0)
        close (fd);
9229 9230
    if (vm)
        virDomainObjUnlock(vm);
9231 9232 9233
    return ret;
}

R
Richard W.M. Jones 已提交
9234 9235 9236 9237 9238 9239
static int
qemudDomainMemoryPeek (virDomainPtr dom,
                       unsigned long long offset, size_t size,
                       void *buffer,
                       unsigned int flags)
{
9240 9241
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
9242
    char *tmp = NULL;
R
Richard W.M. Jones 已提交
9243 9244
    int fd = -1, ret = -1;

9245
    qemuDriverLock(driver);
9246
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
9247
    qemuDriverUnlock(driver);
R
Richard W.M. Jones 已提交
9248 9249

    if (!vm) {
9250 9251
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
9252 9253
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
9254 9255 9256
        goto cleanup;
    }

9257
    if (flags != VIR_MEMORY_VIRTUAL && flags != VIR_MEMORY_PHYSICAL) {
9258 9259
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("flags parameter must be VIR_MEMORY_VIRTUAL or VIR_MEMORY_PHYSICAL"));
9260
        goto cleanup;
R
Richard W.M. Jones 已提交
9261 9262
    }

9263 9264 9265
    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
9266
    if (!virDomainObjIsActive(vm)) {
9267 9268
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
9269
        goto endjob;
R
Richard W.M. Jones 已提交
9270 9271
    }

9272
    if (virAsprintf(&tmp, "%s/qemu.mem.XXXXXX", driver->cacheDir) < 0) {
9273
        virReportOOMError();
9274
        goto endjob;
9275 9276
    }

R
Richard W.M. Jones 已提交
9277 9278
    /* Create a temporary filename. */
    if ((fd = mkstemp (tmp)) == -1) {
9279 9280
        virReportSystemError(errno,
                             _("mkstemp(\"%s\") failed"), tmp);
9281
        goto endjob;
R
Richard W.M. Jones 已提交
9282 9283
    }

9284
    qemuDomainObjPrivatePtr priv = vm->privateData;
9285
    qemuDomainObjEnterMonitor(vm);
9286
    if (flags == VIR_MEMORY_VIRTUAL) {
9287 9288
        if (qemuMonitorSaveVirtualMemory(priv->mon, offset, size, tmp) < 0) {
            qemuDomainObjExitMonitor(vm);
9289
            goto endjob;
9290
        }
9291
    } else {
9292 9293
        if (qemuMonitorSavePhysicalMemory(priv->mon, offset, size, tmp) < 0) {
            qemuDomainObjExitMonitor(vm);
9294
            goto endjob;
9295
        }
R
Richard W.M. Jones 已提交
9296
    }
9297
    qemuDomainObjExitMonitor(vm);
R
Richard W.M. Jones 已提交
9298 9299 9300

    /* Read the memory file into buffer. */
    if (saferead (fd, buffer, size) == (ssize_t) -1) {
9301 9302 9303
        virReportSystemError(errno,
                             _("failed to read temporary file "
                               "created with template %s"), tmp);
9304
        goto endjob;
R
Richard W.M. Jones 已提交
9305 9306 9307
    }

    ret = 0;
9308

9309
endjob:
9310 9311
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
9312

9313
cleanup:
9314
    VIR_FREE(tmp);
R
Richard W.M. Jones 已提交
9315 9316
    if (fd >= 0) close (fd);
    unlink (tmp);
9317 9318
    if (vm)
        virDomainObjUnlock(vm);
R
Richard W.M. Jones 已提交
9319 9320 9321
    return ret;
}

9322

9323 9324 9325 9326 9327 9328 9329 9330 9331 9332 9333 9334 9335 9336 9337 9338 9339 9340 9341 9342 9343 9344 9345 9346 9347 9348 9349 9350 9351 9352 9353 9354 9355 9356 9357 9358 9359 9360 9361 9362 9363 9364 9365 9366 9367 9368 9369 9370 9371 9372 9373 9374 9375 9376 9377 9378 9379 9380 9381 9382 9383 9384 9385 9386 9387 9388 9389 9390 9391
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;
    struct stat sb;
    int i;

    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)) {
            ret = 0;
            break;
        }
    }

    if (ret != 0) {
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("invalid path %s not assigned to domain"), path);
        goto cleanup;
    }

    ret = -1;

    /* 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 */
    if (virStorageFileGetMetadataFromFD(path, fd, &meta) < 0)
        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)) {
9392
#ifndef WIN32
9393 9394 9395 9396 9397 9398 9399 9400 9401 9402 9403 9404 9405 9406 9407 9408 9409 9410 9411 9412 9413 9414 9415 9416 9417 9418 9419 9420 9421 9422 9423 9424 9425 9426 9427 9428 9429 9430 9431 9432 9433 9434 9435
        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;

    /* XXX allocation will need to be pulled from QEMU for
     * the qcow inside LVM case */
    info->allocation = info->physical;

    ret = 0;

cleanup:
    if (fd != -1)
        close(fd);
    if (vm)
        virDomainObjUnlock(vm);
    return ret;
}


9436
static int
9437 9438 9439 9440
qemuDomainEventRegister(virConnectPtr conn,
                        virConnectDomainEventCallback callback,
                        void *opaque,
                        virFreeCallback freecb)
9441
{
9442 9443 9444
    struct qemud_driver *driver = conn->privateData;
    int ret;

9445
    qemuDriverLock(driver);
9446 9447
    ret = virDomainEventCallbackListAdd(conn, driver->domainEventCallbacks,
                                        callback, opaque, freecb);
9448
    qemuDriverUnlock(driver);
9449

9450
    return ret;
9451 9452
}

9453

9454
static int
9455 9456
qemuDomainEventDeregister(virConnectPtr conn,
                          virConnectDomainEventCallback callback)
9457
{
9458 9459 9460
    struct qemud_driver *driver = conn->privateData;
    int ret;

9461
    qemuDriverLock(driver);
9462 9463 9464 9465 9466 9467
    if (driver->domainEventDispatching)
        ret = virDomainEventCallbackListMarkDelete(conn, driver->domainEventCallbacks,
                                                   callback);
    else
        ret = virDomainEventCallbackListRemove(conn, driver->domainEventCallbacks,
                                               callback);
9468
    qemuDriverUnlock(driver);
9469

9470
    return ret;
9471 9472
}

9473 9474 9475 9476 9477 9478 9479 9480 9481 9482 9483 9484 9485 9486 9487 9488 9489 9490 9491 9492 9493 9494 9495 9496 9497 9498 9499 9500 9501 9502 9503 9504 9505 9506 9507 9508 9509 9510 9511 9512 9513 9514 9515

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;
}


9516 9517
static void qemuDomainEventDispatchFunc(virConnectPtr conn,
                                        virDomainEventPtr event,
9518
                                        virConnectDomainEventGenericCallback cb,
9519 9520
                                        void *cbopaque,
                                        void *opaque)
9521
{
9522
    struct qemud_driver *driver = opaque;
9523

9524 9525 9526 9527 9528 9529 9530 9531 9532 9533 9534 9535 9536 9537 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
    /* 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);
9568 9569
}

D
Daniel Veillard 已提交
9570 9571
/* Migration support. */

C
Chris Lalancette 已提交
9572 9573 9574 9575 9576 9577 9578 9579 9580 9581 9582 9583 9584 9585 9586 9587 9588 9589 9590
/* 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) {
9591 9592
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream is not open"));
C
Chris Lalancette 已提交
9593 9594 9595 9596 9597
        return -1;
    }

    qemuDriverLock(driver);
    if (qemust->watch == 0) {
9598 9599
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream does not have a callback registered"));
C
Chris Lalancette 已提交
9600 9601 9602 9603 9604 9605 9606 9607 9608 9609 9610 9611 9612 9613 9614 9615 9616 9617 9618 9619 9620 9621 9622 9623 9624 9625 9626 9627
        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) {
9628 9629
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream is not open"));
C
Chris Lalancette 已提交
9630 9631 9632 9633 9634
        return -1;
    }

    qemuDriverLock(driver);
    if (qemust->watch == 0) {
9635 9636
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream does not have a callback registered"));
C
Chris Lalancette 已提交
9637 9638 9639 9640 9641 9642 9643 9644 9645 9646 9647 9648 9649 9650 9651 9652 9653 9654 9655 9656 9657 9658 9659 9660 9661 9662 9663 9664 9665 9666 9667 9668 9669 9670 9671 9672 9673 9674 9675 9676 9677 9678 9679 9680 9681 9682 9683 9684 9685 9686 9687 9688 9689 9690 9691 9692 9693
        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) {
9694 9695
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream is not open"));
C
Chris Lalancette 已提交
9696 9697 9698 9699 9700
        return -1;
    }

    qemuDriverLock(driver);
    if (qemust->watch != 0) {
9701 9702
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream already has a callback registered"));
C
Chris Lalancette 已提交
9703 9704 9705 9706 9707 9708 9709 9710
        goto cleanup;
    }

    if ((qemust->watch = virEventAddHandle(qemust->fd,
                                           events,
                                           qemuStreamMigEvent,
                                           st,
                                           NULL)) < 0) {
9711 9712
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("cannot register file watch on stream"));
C
Chris Lalancette 已提交
9713 9714 9715 9716 9717 9718 9719 9720 9721 9722 9723 9724 9725 9726 9727 9728 9729 9730 9731 9732 9733 9734 9735 9736 9737 9738 9739 9740 9741 9742 9743 9744
        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;

9745
    if (VIR_ALLOC(qemust) < 0) {
9746
        virReportOOMError();
C
Chris Lalancette 已提交
9747
        return NULL;
9748
    }
C
Chris Lalancette 已提交
9749 9750 9751 9752 9753 9754 9755 9756 9757 9758 9759 9760 9761 9762 9763 9764 9765 9766 9767 9768 9769 9770 9771 9772 9773 9774 9775 9776 9777 9778 9779 9780 9781 9782 9783 9784 9785 9786 9787 9788 9789 9790 9791 9792 9793 9794 9795 9796 9797 9798 9799 9800 9801 9802 9803 9804 9805 9806 9807 9808 9809

    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) {
9810 9811
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream is not open"));
C
Chris Lalancette 已提交
9812 9813 9814 9815 9816 9817 9818 9819 9820 9821 9822 9823 9824 9825
        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;
9826
            virReportSystemError(errno, "%s",
C
Chris Lalancette 已提交
9827 9828 9829 9830 9831 9832 9833 9834 9835 9836 9837 9838 9839 9840 9841 9842 9843 9844 9845 9846 9847 9848 9849 9850 9851 9852 9853 9854 9855 9856 9857 9858 9859 9860 9861 9862 9863 9864
                                 _("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;
9865
    unsigned long long qemuCmdFlags;
C
Chris Lalancette 已提交
9866 9867 9868 9869
    struct qemuStreamMigFile *qemust = NULL;

    qemuDriverLock(driver);
    if (!dom_xml) {
9870 9871
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("no domain XML passed"));
C
Chris Lalancette 已提交
9872 9873 9874
        goto cleanup;
    }
    if (!(flags & VIR_MIGRATE_TUNNELLED)) {
9875
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
C
Chris Lalancette 已提交
9876 9877 9878 9879
                         "%s", _("PrepareTunnel called but no TUNNELLED flag set"));
        goto cleanup;
    }
    if (st == NULL) {
9880 9881
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("tunnelled migration requested but NULL stream passed"));
C
Chris Lalancette 已提交
9882 9883 9884 9885
        goto cleanup;
    }

    /* Parse the domain XML. */
9886
    if (!(def = virDomainDefParseString(driver->caps, dom_xml,
C
Chris Lalancette 已提交
9887
                                        VIR_DOMAIN_XML_INACTIVE))) {
9888 9889
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to parse XML"));
C
Chris Lalancette 已提交
9890 9891 9892 9893
        goto cleanup;
    }

    /* Target domain name, maybe renamed. */
9894 9895 9896 9897 9898 9899
    if (dname) {
        VIR_FREE(def->name);
        def->name = strdup(dname);
        if (def->name == NULL)
            goto cleanup;
    }
C
Chris Lalancette 已提交
9900

9901 9902
    if (virDomainObjIsDuplicate(&driver->domains, def, 1) < 0)
        goto cleanup;
C
Chris Lalancette 已提交
9903

9904
    if (!(vm = virDomainAssignDef(driver->caps,
C
Chris Lalancette 已提交
9905
                                  &driver->domains,
9906
                                  def, true))) {
9907 9908
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to assign new VM"));
C
Chris Lalancette 已提交
9909 9910 9911 9912
        goto cleanup;
    }
    def = NULL;

9913 9914 9915
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

C
Chris Lalancette 已提交
9916 9917 9918 9919 9920
    /* Domain starts inactive, even if the domain XML had an id field. */
    vm->def->id = -1;

    if (virAsprintf(&unixfile, "%s/qemu.tunnelmigrate.dest.%s",
                    driver->stateDir, vm->def->name) < 0) {
9921
        virReportOOMError();
9922
        goto endjob;
C
Chris Lalancette 已提交
9923 9924 9925 9926 9927
    }
    unlink(unixfile);

    /* check that this qemu version supports the interactive exec */
    if (qemudExtractVersionInfo(vm->def->emulator, NULL, &qemuCmdFlags) < 0) {
9928 9929 9930
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Cannot determine QEMU argv syntax %s"),
                        vm->def->emulator);
9931
        goto endjob;
C
Chris Lalancette 已提交
9932 9933 9934 9935 9936 9937
    }
    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 {
9938 9939
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("Destination qemu is too old to support tunnelled migration"));
9940
        goto endjob;
C
Chris Lalancette 已提交
9941 9942
    }
    if (internalret < 0) {
9943
        virReportOOMError();
9944
        goto endjob;
C
Chris Lalancette 已提交
9945 9946 9947 9948 9949 9950 9951 9952 9953 9954 9955 9956 9957 9958
    }
    /* Start the QEMU daemon, with the same command-line arguments plus
     * -incoming unix:/path/to/file or exec:nc -U /path/to/file
     */
    internalret = qemudStartVMDaemon(dconn, driver, vm, migrateFrom, -1);
    VIR_FREE(migrateFrom);
    if (internalret < 0) {
        /* Note that we don't set an error here because qemudStartVMDaemon
         * should have already done that.
         */
        if (!vm->persistent) {
            virDomainRemoveInactive(&driver->domains, vm);
            vm = NULL;
        }
9959
        goto endjob;
C
Chris Lalancette 已提交
9960 9961 9962 9963
    }

    qemust = qemuStreamMigOpen(st, unixfile);
    if (qemust == NULL) {
9964
        qemudShutdownVMDaemon(driver, vm, 0);
9965
        if (!vm->persistent) {
9966 9967
            if (qemuDomainObjEndJob(vm) > 0)
                virDomainRemoveInactive(&driver->domains, vm);
9968 9969
            vm = NULL;
        }
9970
        virReportSystemError(errno,
C
Chris Lalancette 已提交
9971 9972
                             _("cannot open unix socket '%s' for tunnelled migration"),
                             unixfile);
9973
        goto endjob;
C
Chris Lalancette 已提交
9974 9975 9976 9977 9978 9979 9980 9981 9982 9983
    }

    st->driver = &qemuStreamMigDrv;
    st->privateData = qemust;

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_MIGRATED);
    ret = 0;

9984
endjob:
9985 9986 9987
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
9988

C
Chris Lalancette 已提交
9989 9990
cleanup:
    virDomainDefFree(def);
9991 9992
    if (unixfile)
        unlink(unixfile);
C
Chris Lalancette 已提交
9993 9994 9995 9996 9997 9998 9999 10000 10001
    VIR_FREE(unixfile);
    if (vm)
        virDomainObjUnlock(vm);
    if (event)
        qemuDomainEventQueue(driver, event);
    qemuDriverUnlock(driver);
    return ret;
}

D
Daniel Veillard 已提交
10002 10003 10004 10005
/* Prepare is the first step, and it runs on the destination host.
 *
 * This starts an empty VM listening on a TCP port.
 */
10006
static int ATTRIBUTE_NONNULL (5)
D
Daniel Veillard 已提交
10007 10008 10009 10010 10011
qemudDomainMigratePrepare2 (virConnectPtr dconn,
                            char **cookie ATTRIBUTE_UNUSED,
                            int *cookielen ATTRIBUTE_UNUSED,
                            const char *uri_in,
                            char **uri_out,
C
Chris Lalancette 已提交
10012
                            unsigned long flags,
D
Daniel Veillard 已提交
10013 10014 10015 10016 10017
                            const char *dname,
                            unsigned long resource ATTRIBUTE_UNUSED,
                            const char *dom_xml)
{
    static int port = 0;
10018 10019
    struct qemud_driver *driver = dconn->privateData;
    virDomainDefPtr def = NULL;
D
Daniel Veillard 已提交
10020 10021
    virDomainObjPtr vm = NULL;
    int this_port;
10022
    char *hostname;
D
Daniel Veillard 已提交
10023 10024
    char migrateFrom [64];
    const char *p;
10025
    virDomainEventPtr event = NULL;
10026
    int ret = -1;
10027
    int internalret;
10028 10029

    *uri_out = NULL;
D
Daniel Veillard 已提交
10030

10031
    qemuDriverLock(driver);
C
Chris Lalancette 已提交
10032 10033 10034 10035
    if (flags & VIR_MIGRATE_TUNNELLED) {
        /* this is a logical error; we never should have gotten here with
         * VIR_MIGRATE_TUNNELLED set
         */
10036 10037
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("Tunnelled migration requested but invalid RPC method called"));
C
Chris Lalancette 已提交
10038 10039 10040
        goto cleanup;
    }

D
Daniel Veillard 已提交
10041
    if (!dom_xml) {
10042 10043
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("no domain XML passed"));
10044
        goto cleanup;
D
Daniel Veillard 已提交
10045 10046 10047 10048 10049 10050 10051 10052 10053 10054 10055 10056 10057 10058 10059 10060 10061
    }

    /* 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 */
10062
        if ((hostname = virGetHostnameLocalhost(0)) == NULL)
10063
            goto cleanup;
D
Daniel Veillard 已提交
10064

10065 10066 10067 10068 10069
        /* 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
10070
        /* Caller frees */
10071 10072 10073
        internalret = virAsprintf(uri_out, "tcp:%s:%d", hostname, this_port);
        VIR_FREE(hostname);
        if (internalret < 0) {
10074
            virReportOOMError();
10075
            goto cleanup;
D
Daniel Veillard 已提交
10076 10077 10078 10079 10080 10081
        }
    } 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.
         */
10082
        if (!STRPREFIX (uri_in, "tcp:")) {
10083 10084
            qemuReportError (VIR_ERR_INVALID_ARG,
                             "%s", _("only tcp URIs are supported for KVM/QEMU migrations"));
10085
            goto cleanup;
D
Daniel Veillard 已提交
10086 10087 10088 10089
        }

        /* Get the port number. */
        p = strrchr (uri_in, ':');
10090 10091 10092 10093 10094 10095 10096 10097
        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) {
10098
                virReportOOMError();
10099 10100 10101 10102 10103 10104 10105
                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)) {
10106 10107
                qemuReportError(VIR_ERR_INVALID_ARG,
                                "%s", _("URI ended with incorrect ':port'"));
10108 10109
                goto cleanup;
            }
D
Daniel Veillard 已提交
10110 10111 10112
        }
    }

10113
    if (*uri_out)
10114 10115
        VIR_DEBUG("Generated uri_out=%s", *uri_out);

D
Daniel Veillard 已提交
10116
    /* Parse the domain XML. */
10117
    if (!(def = virDomainDefParseString(driver->caps, dom_xml,
10118
                                        VIR_DOMAIN_XML_INACTIVE))) {
10119 10120
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to parse XML"));
10121
        goto cleanup;
D
Daniel Veillard 已提交
10122 10123 10124
    }

    /* Target domain name, maybe renamed. */
10125 10126 10127 10128 10129 10130
    if (dname) {
        VIR_FREE(def->name);
        def->name = strdup(dname);
        if (def->name == NULL)
            goto cleanup;
    }
D
Daniel Veillard 已提交
10131

10132 10133
    if (virDomainObjIsDuplicate(&driver->domains, def, 1) < 0)
        goto cleanup;
D
Daniel Veillard 已提交
10134

10135
    if (!(vm = virDomainAssignDef(driver->caps,
D
Daniel Veillard 已提交
10136
                                  &driver->domains,
10137
                                  def, true))) {
10138 10139
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to assign new VM"));
10140
        goto cleanup;
D
Daniel Veillard 已提交
10141
    }
10142
    def = NULL;
D
Daniel Veillard 已提交
10143

10144 10145 10146
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

D
Daniel Veillard 已提交
10147 10148 10149 10150 10151 10152 10153
    /* 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);
10154
    if (qemudStartVMDaemon (dconn, driver, vm, migrateFrom, -1) < 0) {
10155 10156 10157
        /* Note that we don't set an error here because qemudStartVMDaemon
         * should have already done that.
         */
10158
        if (!vm->persistent) {
10159 10160
            if (qemuDomainObjEndJob(vm) > 0)
                virDomainRemoveInactive(&driver->domains, vm);
10161 10162
            vm = NULL;
        }
10163
        goto endjob;
D
Daniel Veillard 已提交
10164
    }
10165 10166 10167 10168

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_MIGRATED);
10169
    ret = 0;
D
Daniel Veillard 已提交
10170

10171
endjob:
10172 10173 10174
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
10175

10176 10177 10178 10179 10180
cleanup:
    virDomainDefFree(def);
    if (ret != 0) {
        VIR_FREE(*uri_out);
    }
10181 10182
    if (vm)
        virDomainObjUnlock(vm);
10183 10184
    if (event)
        qemuDomainEventQueue(driver, event);
10185
    qemuDriverUnlock(driver);
10186
    return ret;
C
Chris Lalancette 已提交
10187 10188 10189

}

10190 10191 10192 10193

/* Perform migration using QEMU's native TCP migrate support,
 * not encrypted obviously
 */
10194
static int doNativeMigrate(struct qemud_driver *driver,
10195 10196
                           virDomainObjPtr vm,
                           const char *uri,
10197
                           unsigned int flags,
10198 10199 10200 10201
                           const char *dname ATTRIBUTE_UNUSED,
                           unsigned long resource)
{
    int ret = -1;
10202
    xmlURIPtr uribits = NULL;
10203
    qemuDomainObjPrivatePtr priv = vm->privateData;
10204 10205 10206 10207
    unsigned int background_flags = 0;

    virCheckFlags(VIR_MIGRATE_NON_SHARED_DISK | VIR_MIGRATE_NON_SHARED_INC,
                  -1);
10208 10209 10210 10211 10212 10213

    /* 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) {
10214
            virReportOOMError();
10215 10216 10217 10218 10219 10220 10221 10222
            goto cleanup;
        }
        uribits = xmlParseURI(tmpuri);
        VIR_FREE(tmpuri);
    } else {
        uribits = xmlParseURI(uri);
    }
    if (!uribits) {
10223 10224
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("cannot parse URI %s"), uri);
10225 10226 10227
        goto cleanup;
    }

10228
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
10229
    if (resource > 0 &&
10230
        qemuMonitorSetMigrationSpeed(priv->mon, resource) < 0) {
10231
        qemuDomainObjExitMonitorWithDriver(driver, vm);
10232
        goto cleanup;
10233
    }
10234

10235 10236 10237 10238 10239 10240 10241 10242
    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) {
10243
        qemuDomainObjExitMonitorWithDriver(driver, vm);
10244 10245
        goto cleanup;
    }
10246
    qemuDomainObjExitMonitorWithDriver(driver, vm);
10247

10248
    if (qemuDomainWaitForMigrationComplete(driver, vm) < 0)
10249 10250 10251 10252 10253 10254 10255 10256 10257 10258
        goto cleanup;

    ret = 0;

cleanup:
    xmlFreeURI(uribits);
    return ret;
}


10259 10260
#define TUNNEL_SEND_BUF_SIZE 65536

10261
static int doTunnelSendAll(virStreamPtr st,
10262 10263
                           int sock)
{
10264 10265 10266 10267 10268 10269 10270 10271
    char *buffer;
    int nbytes = TUNNEL_SEND_BUF_SIZE;

    if (VIR_ALLOC_N(buffer, TUNNEL_SEND_BUF_SIZE) < 0) {
        virReportOOMError();
        virStreamAbort(st);
        return -1;
    }
10272 10273 10274 10275 10276

    /* XXX should honour the 'resource' parameter here */
    for (;;) {
        nbytes = saferead(sock, buffer, nbytes);
        if (nbytes < 0) {
10277
            virReportSystemError(errno, "%s",
10278
                                 _("tunnelled migration failed to read from qemu"));
10279 10280
            virStreamAbort(st);
            VIR_FREE(buffer);
10281 10282 10283 10284 10285 10286 10287
            return -1;
        }
        else if (nbytes == 0)
            /* EOF; get out of here */
            break;

        if (virStreamSend(st, buffer, nbytes) < 0) {
10288 10289
            qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
                            _("Failed to write migration data to remote libvirtd"));
10290
            VIR_FREE(buffer);
10291 10292 10293 10294
            return -1;
        }
    }

10295 10296
    VIR_FREE(buffer);

10297 10298 10299 10300 10301 10302 10303
    if (virStreamFinish(st) < 0)
        /* virStreamFinish set the error for us */
        return -1;

    return 0;
}

C
Chris Lalancette 已提交
10304
static int doTunnelMigrate(virDomainPtr dom,
10305
                           struct qemud_driver *driver,
10306
                           virConnectPtr dconn,
C
Chris Lalancette 已提交
10307
                           virDomainObjPtr vm,
10308
                           const char *dom_xml,
C
Chris Lalancette 已提交
10309 10310 10311 10312 10313
                           const char *uri,
                           unsigned long flags,
                           const char *dname,
                           unsigned long resource)
{
10314
    qemuDomainObjPrivatePtr priv = vm->privateData;
10315 10316
    int client_sock = -1;
    int qemu_sock = -1;
C
Chris Lalancette 已提交
10317 10318
    struct sockaddr_un sa_qemu, sa_client;
    socklen_t addrlen;
10319
    virDomainPtr ddomain = NULL;
C
Chris Lalancette 已提交
10320
    int retval = -1;
10321
    virStreamPtr st = NULL;
C
Chris Lalancette 已提交
10322 10323
    char *unixfile = NULL;
    int internalret;
10324
    unsigned long long qemuCmdFlags;
C
Chris Lalancette 已提交
10325 10326
    int status;
    unsigned long long transferred, remaining, total;
10327
    unsigned int background_flags = QEMU_MONITOR_MIGRATE_BACKGROUND;
C
Chris Lalancette 已提交
10328

10329 10330 10331 10332 10333 10334 10335 10336
    /*
     * 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
10337 10338
     */

10339

10340
    /* Stage 1. setup local support infrastructure */
C
Chris Lalancette 已提交
10341 10342 10343

    if (virAsprintf(&unixfile, "%s/qemu.tunnelmigrate.src.%s",
                    driver->stateDir, vm->def->name) < 0) {
10344
        virReportOOMError();
10345
        goto cleanup;
C
Chris Lalancette 已提交
10346 10347 10348 10349
    }

    qemu_sock = socket(AF_UNIX, SOCK_STREAM, 0);
    if (qemu_sock < 0) {
10350
        virReportSystemError(errno, "%s",
C
Chris Lalancette 已提交
10351
                             _("cannot open tunnelled migration socket"));
10352
        goto cleanup;
C
Chris Lalancette 已提交
10353 10354 10355 10356 10357
    }
    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) {
10358 10359 10360
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Unix socket '%s' too big for destination"),
                        unixfile);
10361
        goto cleanup;
C
Chris Lalancette 已提交
10362 10363 10364
    }
    unlink(unixfile);
    if (bind(qemu_sock, (struct sockaddr *)&sa_qemu, sizeof(sa_qemu)) < 0) {
10365
        virReportSystemError(errno,
C
Chris Lalancette 已提交
10366 10367
                             _("Cannot bind to unix socket '%s' for tunnelled migration"),
                             unixfile);
10368
        goto cleanup;
C
Chris Lalancette 已提交
10369 10370
    }
    if (listen(qemu_sock, 1) < 0) {
10371
        virReportSystemError(errno,
C
Chris Lalancette 已提交
10372 10373
                             _("Cannot listen on unix socket '%s' for tunnelled migration"),
                             unixfile);
10374
        goto cleanup;
C
Chris Lalancette 已提交
10375 10376 10377 10378
    }

    /* check that this qemu version supports the unix migration */
    if (qemudExtractVersionInfo(vm->def->emulator, NULL, &qemuCmdFlags) < 0) {
10379 10380 10381
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Cannot extract Qemu version from '%s'"),
                        vm->def->emulator);
10382 10383 10384 10385 10386
        goto cleanup;
    }

    if (!(qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_UNIX) &&
        !(qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_EXEC)) {
10387 10388
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("Source qemu is too old to support tunnelled migration"));
10389
        goto cleanup;
C
Chris Lalancette 已提交
10390
    }
10391 10392 10393 10394 10395 10396 10397 10398 10399 10400 10401 10402 10403 10404 10405 10406 10407 10408 10409 10410 10411


    /* 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;

    internalret = dconn->driver->domainMigratePrepareTunnel(dconn, st,
                                                            flags, dname,
                                                            resource, dom_xml);

    if (internalret < 0)
        /* domainMigratePrepareTunnel sets the error for us */
        goto cleanup;

    /*   3. start migration on source */
10412
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
10413 10414 10415 10416 10417 10418 10419 10420
    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
10421 10422
    else if (qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_EXEC) {
        const char *args[] = { "nc", "-U", unixfile, NULL };
10423
        internalret = qemuMonitorMigrateToCommand(priv->mon, QEMU_MONITOR_MIGRATE_BACKGROUND, args);
10424 10425
    } else {
        internalret = -1;
C
Chris Lalancette 已提交
10426
    }
10427
    qemuDomainObjExitMonitorWithDriver(driver, vm);
C
Chris Lalancette 已提交
10428
    if (internalret < 0) {
10429 10430
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("tunnelled migration monitor command failed"));
10431
        goto finish;
C
Chris Lalancette 已提交
10432 10433
    }

10434 10435 10436 10437 10438 10439
    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("guest unexpectedly quit"));
        goto cleanup;
    }

10440 10441 10442
    /* From this point onwards we *must* call cancel to abort the
     * migration on source if anything goes wrong */

C
Chris Lalancette 已提交
10443 10444 10445
    /* it is also possible that the migrate didn't fail initially, but
     * rather failed later on.  Check the output of "info migrate"
     */
10446
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
10447 10448
    if (qemuMonitorGetMigrationStatus(priv->mon,
                                      &status,
C
Chris Lalancette 已提交
10449 10450 10451
                                      &transferred,
                                      &remaining,
                                      &total) < 0) {
10452
        qemuDomainObjExitMonitorWithDriver(driver, vm);
10453
        goto cancel;
C
Chris Lalancette 已提交
10454
    }
10455
    qemuDomainObjExitMonitorWithDriver(driver, vm);
C
Chris Lalancette 已提交
10456 10457

    if (status == QEMU_MONITOR_MIGRATION_STATUS_ERROR) {
10458 10459
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s",_("migrate failed"));
10460
        goto cancel;
C
Chris Lalancette 已提交
10461 10462 10463 10464 10465 10466
    }

    addrlen = sizeof(sa_client);
    while ((client_sock = accept(qemu_sock, (struct sockaddr *)&sa_client, &addrlen)) < 0) {
        if (errno == EAGAIN || errno == EINTR)
            continue;
10467
        virReportSystemError(errno, "%s",
C
Chris Lalancette 已提交
10468
                             _("tunnelled migration failed to accept from qemu"));
10469
        goto cancel;
C
Chris Lalancette 已提交
10470 10471
    }

10472
    retval = doTunnelSendAll(st, client_sock);
10473

10474
cancel:
10475
    if (retval != 0 && priv->mon) {
10476
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
10477
        qemuMonitorMigrateCancel(priv->mon);
10478
        qemuDomainObjExitMonitorWithDriver(driver, vm);
10479
    }
C
Chris Lalancette 已提交
10480

10481
finish:
C
Chris Lalancette 已提交
10482 10483 10484
    dname = dname ? dname : dom->name;
    ddomain = dconn->driver->domainMigrateFinish2
        (dconn, dname, NULL, 0, uri, flags, retval);
10485 10486 10487 10488 10489 10490 10491

cleanup:
    if (client_sock != -1)
        close(client_sock);
    if (qemu_sock != -1)
        close(qemu_sock);

C
Chris Lalancette 已提交
10492 10493 10494
    if (ddomain)
        virUnrefDomain(ddomain);

10495 10496 10497 10498
    if (unixfile) {
        unlink(unixfile);
        VIR_FREE(unixfile);
    }
C
Chris Lalancette 已提交
10499

10500 10501 10502
    if (st)
        /* don't call virStreamFree(), because that resets any pending errors */
        virUnrefStream(st);
10503 10504 10505 10506
    return retval;
}


10507 10508 10509 10510
/* 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,
10511
                              struct qemud_driver *driver,
10512 10513 10514 10515 10516 10517 10518 10519 10520 10521 10522 10523 10524 10525 10526 10527 10528 10529 10530 10531 10532 10533 10534 10535 10536
                              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;

    /* NB we don't pass 'uri' into this, since that's the libvirtd
     * URI in this context - so we let dest pick it */
    if (dconn->driver->domainMigratePrepare2(dconn,
                                             NULL, /* cookie */
                                             0, /* cookielen */
                                             NULL, /* uri */
                                             &uri_out,
                                             flags, dname,
                                             resource, dom_xml) < 0)
        /* domainMigratePrepare2 sets the error for us */
        goto cleanup;

    if (uri_out == NULL) {
10537 10538
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("domainMigratePrepare2 did not set uri"));
10539
        goto cleanup;
10540 10541
    }

10542
    if (doNativeMigrate(driver, vm, uri_out, flags, dname, resource) < 0)
10543 10544 10545 10546 10547 10548 10549 10550 10551 10552 10553 10554 10555 10556 10557 10558 10559
        goto finish;

    retval = 0;

finish:
    dname = dname ? dname : dom->name;
    ddomain = dconn->driver->domainMigrateFinish2
        (dconn, dname, NULL, 0, uri_out, flags, retval);

    if (ddomain)
        virUnrefDomain(ddomain);

cleanup:
    return retval;
}


10560
static int doPeer2PeerMigrate(virDomainPtr dom,
10561
                              struct qemud_driver *driver,
10562 10563 10564 10565 10566 10567 10568 10569 10570 10571 10572 10573 10574 10575 10576 10577
                              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) {
10578 10579
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("Failed to connect to remote libvirt URI %s"), uri);
10580 10581 10582
        return -1;
    }
    if (!VIR_DRV_SUPPORTS_FEATURE(dconn->driver, dconn,
10583
                                  VIR_DRV_FEATURE_MIGRATION_P2P)) {
10584 10585
        qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
                        _("Destination libvirt does not support peer-to-peer migration protocol"));
10586 10587 10588
        goto cleanup;
    }

10589 10590 10591
    dom_xml = qemudVMDumpXML(driver, vm,
                             VIR_DOMAIN_XML_SECURE |
                             VIR_DOMAIN_XML_UPDATE_CPU);
10592
    if (!dom_xml) {
10593 10594
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to get domain xml"));
10595 10596 10597
        goto cleanup;
    }

10598
    if (flags & VIR_MIGRATE_TUNNELLED)
10599
        ret = doTunnelMigrate(dom, driver, dconn, vm, dom_xml, uri, flags, dname, resource);
10600
    else
10601
        ret = doNonTunnelMigrate(dom, driver, dconn, vm, dom_xml, uri, flags, dname, resource);
10602 10603 10604

cleanup:
    VIR_FREE(dom_xml);
C
Chris Lalancette 已提交
10605 10606 10607
    /* don't call virConnectClose(), because that resets any pending errors */
    virUnrefConnect(dconn);

10608
    return ret;
D
Daniel Veillard 已提交
10609 10610
}

10611

D
Daniel Veillard 已提交
10612 10613 10614 10615 10616 10617
/* 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,
10618
                           unsigned long flags,
10619
                           const char *dname,
D
Daniel Veillard 已提交
10620 10621
                           unsigned long resource)
{
10622 10623
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
10624
    virDomainEventPtr event = NULL;
10625
    int ret = -1;
10626
    int resume = 0;
10627
    qemuDomainObjPrivatePtr priv;
D
Daniel Veillard 已提交
10628

10629
    qemuDriverLock(driver);
10630
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
D
Daniel Veillard 已提交
10631
    if (!vm) {
10632 10633
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
10634 10635
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
10636
        goto cleanup;
D
Daniel Veillard 已提交
10637
    }
10638
    priv = vm->privateData;
D
Daniel Veillard 已提交
10639

10640 10641
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;
10642
    priv->jobActive = QEMU_JOB_MIGRATION;
10643

D
Daniel P. Berrange 已提交
10644
    if (!virDomainObjIsActive(vm)) {
10645 10646
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
10647
        goto endjob;
D
Daniel Veillard 已提交
10648 10649
    }

10650 10651 10652
    memset(&priv->jobInfo, 0, sizeof(priv->jobInfo));
    priv->jobInfo.type = VIR_DOMAIN_JOB_UNBOUNDED;

10653
    resume = vm->state == VIR_DOMAIN_RUNNING;
10654
    if (!(flags & VIR_MIGRATE_LIVE) && vm->state == VIR_DOMAIN_RUNNING) {
10655
        if (qemuDomainMigrateOffline(driver, vm) < 0)
10656
            goto endjob;
10657 10658
    }

10659
    if ((flags & (VIR_MIGRATE_TUNNELLED | VIR_MIGRATE_PEER2PEER))) {
10660
        if (doPeer2PeerMigrate(dom, driver, vm, uri, flags, dname, resource) < 0)
10661
            /* doPeer2PeerMigrate already set the error, so just get out */
10662
            goto endjob;
10663
    } else {
10664
        if (doNativeMigrate(driver, vm, uri, flags, dname, resource) < 0)
10665
            goto endjob;
10666 10667
    }

D
Daniel Veillard 已提交
10668
    /* Clean up the source domain. */
10669
    qemudShutdownVMDaemon(driver, vm, 1);
10670
    resume = 0;
10671 10672 10673 10674

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_MIGRATED);
C
Chris Lalancette 已提交
10675
    if (!vm->persistent || (flags & VIR_MIGRATE_UNDEFINE_SOURCE)) {
10676
        virDomainDeleteConfig(driver->configDir, driver->autostartDir, vm);
10677 10678
        if (qemuDomainObjEndJob(vm) > 0)
            virDomainRemoveInactive(&driver->domains, vm);
10679 10680
        vm = NULL;
    }
10681
    ret = 0;
D
Daniel Veillard 已提交
10682

10683
endjob:
10684
    if (resume && vm->state == VIR_DOMAIN_PAUSED) {
10685
        /* we got here through some sort of failure; start the domain again */
10686
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
10687
        if (qemuMonitorStartCPUs(priv->mon, dom->conn) < 0) {
10688 10689 10690 10691
            /* 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
             */
10692
            VIR_ERROR(_("Failed to resume guest %s after failure"),
10693
                      vm->def->name);
10694
        }
10695
        qemuDomainObjExitMonitorWithDriver(driver, vm);
10696

10697
        vm->state = VIR_DOMAIN_RUNNING;
10698 10699 10700 10701
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_MIGRATED);
    }
10702 10703 10704
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
10705

10706
cleanup:
10707 10708
    if (vm)
        virDomainObjUnlock(vm);
10709 10710
    if (event)
        qemuDomainEventQueue(driver, event);
10711
    qemuDriverUnlock(driver);
10712
    return ret;
D
Daniel Veillard 已提交
10713 10714 10715 10716 10717 10718 10719 10720 10721
}

/* 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
10722
                           unsigned long flags,
D
Daniel Veillard 已提交
10723 10724
                           int retcode)
{
10725 10726 10727
    struct qemud_driver *driver = dconn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
10728
    virDomainEventPtr event = NULL;
10729
    virErrorPtr orig_err;
C
Chris Lalancette 已提交
10730
    int newVM = 1;
D
Daniel Veillard 已提交
10731

10732 10733 10734
    /* Migration failed. Save the current error so nothing squashes it */
    orig_err = virSaveLastError();

10735
    qemuDriverLock(driver);
10736
    vm = virDomainFindByName(&driver->domains, dname);
D
Daniel Veillard 已提交
10737
    if (!vm) {
10738 10739
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching name '%s'"), dname);
10740
        goto cleanup;
D
Daniel Veillard 已提交
10741 10742
    }

10743 10744 10745
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

D
Daniel Veillard 已提交
10746 10747 10748 10749
    /* Did the migration go as planned?  If yes, return the domain
     * object, but if no, clean up the empty qemu process.
     */
    if (retcode == 0) {
10750 10751 10752 10753 10754 10755
        if (!virDomainObjIsActive(vm)) {
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("guest unexpectedly quit"));
            goto cleanup;
        }

C
Chris Lalancette 已提交
10756 10757 10758 10759 10760
        if (flags & VIR_MIGRATE_PERSIST_DEST) {
            if (vm->persistent)
                newVM = 0;
            vm->persistent = 1;

10761
            if (virDomainSaveConfig(driver->configDir, vm->def) < 0) {
C
Chris Lalancette 已提交
10762 10763 10764 10765 10766 10767 10768 10769 10770 10771
                /* 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;
10772
                goto endjob;
C
Chris Lalancette 已提交
10773 10774 10775 10776 10777 10778 10779 10780 10781
            }

            event = virDomainEventNewFromObj(vm,
                                             VIR_DOMAIN_EVENT_DEFINED,
                                             newVM ?
                                             VIR_DOMAIN_EVENT_DEFINED_ADDED :
                                             VIR_DOMAIN_EVENT_DEFINED_UPDATED);
            if (event)
                qemuDomainEventQueue(driver, event);
10782
            event = NULL;
C
Chris Lalancette 已提交
10783 10784

        }
10785
        qemuDomainObjPrivatePtr priv = vm->privateData;
D
Daniel Veillard 已提交
10786
        dom = virGetDomain (dconn, vm->def->name, vm->def->uuid);
10787

10788 10789 10790 10791 10792 10793 10794 10795
        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)
10796 10797
                    qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                    "%s", _("resume operation failed"));
10798 10799 10800
                qemuDomainObjExitMonitorWithDriver(driver, vm);
                goto endjob;
            }
10801
            qemuDomainObjExitMonitorWithDriver(driver, vm);
10802 10803

            vm->state = VIR_DOMAIN_RUNNING;
10804 10805
        }

10806 10807 10808
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_MIGRATED);
10809 10810 10811 10812 10813 10814
        if (vm->state == VIR_DOMAIN_PAUSED) {
            qemuDomainEventQueue(driver, event);
            event = virDomainEventNewFromObj(vm,
                                             VIR_DOMAIN_EVENT_SUSPENDED,
                                             VIR_DOMAIN_EVENT_SUSPENDED_PAUSED);
        }
10815 10816 10817 10818
        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
10819
    } else {
10820
        qemudShutdownVMDaemon(driver, vm, 0);
10821 10822 10823
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         VIR_DOMAIN_EVENT_STOPPED_FAILED);
10824
        if (!vm->persistent) {
10825 10826
            if (qemuDomainObjEndJob(vm) > 0)
                virDomainRemoveInactive(&driver->domains, vm);
10827 10828
            vm = NULL;
        }
D
Daniel Veillard 已提交
10829
    }
10830

10831
endjob:
10832 10833 10834
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
10835

10836
cleanup:
10837 10838 10839 10840
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
10841 10842
    if (vm)
        virDomainObjUnlock(vm);
10843 10844
    if (event)
        qemuDomainEventQueue(driver, event);
10845
    qemuDriverUnlock(driver);
10846
    return dom;
D
Daniel Veillard 已提交
10847 10848
}

10849 10850 10851 10852 10853 10854 10855 10856 10857 10858 10859 10860 10861 10862 10863 10864
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;

10865
    def = virNodeDeviceDefParseString(xml, EXISTING_DEVICE);
10866 10867 10868 10869 10870 10871 10872 10873 10874 10875 10876 10877 10878 10879 10880 10881 10882
    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) {
10883 10884
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("device %s is not a PCI device"), dev->name);
10885 10886 10887 10888 10889 10890 10891 10892 10893 10894 10895 10896 10897 10898 10899 10900 10901 10902 10903 10904
        goto out;
    }

    ret = 0;
out:
    virNodeDeviceDefFree(def);
    VIR_FREE(xml);
    return ret;
}

static int
qemudNodeDeviceDettach (virNodeDevicePtr dev)
{
    pciDevice *pci;
    unsigned domain, bus, slot, function;
    int ret = -1;

    if (qemudNodeDeviceGetPciInfo(dev, &domain, &bus, &slot, &function) < 0)
        return -1;

10905
    pci = pciGetDevice(domain, bus, slot, function);
10906 10907 10908
    if (!pci)
        return -1;

10909
    if (pciDettachDevice(pci) < 0)
10910 10911 10912 10913
        goto out;

    ret = 0;
out:
10914
    pciFreeDevice(pci);
10915 10916 10917 10918 10919 10920 10921 10922 10923 10924 10925 10926 10927
    return ret;
}

static int
qemudNodeDeviceReAttach (virNodeDevicePtr dev)
{
    pciDevice *pci;
    unsigned domain, bus, slot, function;
    int ret = -1;

    if (qemudNodeDeviceGetPciInfo(dev, &domain, &bus, &slot, &function) < 0)
        return -1;

10928
    pci = pciGetDevice(domain, bus, slot, function);
10929 10930 10931
    if (!pci)
        return -1;

10932
    if (pciReAttachDevice(pci) < 0)
10933 10934 10935 10936
        goto out;

    ret = 0;
out:
10937
    pciFreeDevice(pci);
10938 10939 10940 10941 10942 10943
    return ret;
}

static int
qemudNodeDeviceReset (virNodeDevicePtr dev)
{
10944
    struct qemud_driver *driver = dev->conn->privateData;
10945 10946 10947 10948 10949 10950 10951
    pciDevice *pci;
    unsigned domain, bus, slot, function;
    int ret = -1;

    if (qemudNodeDeviceGetPciInfo(dev, &domain, &bus, &slot, &function) < 0)
        return -1;

10952
    pci = pciGetDevice(domain, bus, slot, function);
10953 10954 10955
    if (!pci)
        return -1;

10956 10957
    qemuDriverLock(driver);

10958
    if (pciResetDevice(pci, driver->activePciHostdevs) < 0)
10959 10960 10961 10962
        goto out;

    ret = 0;
out:
10963
    qemuDriverUnlock(driver);
10964
    pciFreeDevice(pci);
10965 10966 10967
    return ret;
}

10968 10969 10970 10971 10972 10973 10974 10975 10976 10977 10978
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) {
10979 10980
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        "%s", _("cannot get host CPU capabilities"));
10981 10982
    }
    else
10983
        ret = cpuCompareXML(driver->caps->host.cpu, xmlDesc);
10984 10985 10986 10987 10988 10989

    qemuDriverUnlock(driver);

    return ret;
}

10990

10991 10992 10993 10994 10995 10996 10997 10998 10999 11000 11001 11002 11003
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;
}

11004 11005 11006 11007 11008 11009 11010 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

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) {
            memcpy(info, &priv->jobInfo, sizeof(*info));
        } 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;
}


11047 11048 11049 11050 11051 11052 11053 11054 11055 11056 11057 11058 11059 11060 11061 11062 11063 11064 11065 11066 11067 11068
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);
11069
            priv->jobSignals |= QEMU_JOB_SIGNAL_CANCEL;
11070 11071 11072 11073 11074 11075 11076 11077 11078 11079 11080 11081 11082 11083 11084 11085 11086 11087 11088 11089
        } 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;
}


11090 11091 11092 11093 11094 11095 11096 11097 11098 11099
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;

11100
    virCheckFlags(0, -1);
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 11127 11128 11129 11130 11131 11132 11133 11134 11135 11136 11137 11138

    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;

    if (priv->jobActive != QEMU_JOB_MIGRATION) {
        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
11139 11140 11141 11142 11143 11144 11145 11146 11147 11148 11149 11150 11151 11152 11153 11154 11155 11156 11157 11158 11159 11160 11161 11162 11163 11164 11165 11166 11167 11168 11169 11170 11171 11172 11173 11174 11175 11176 11177 11178 11179 11180 11181 11182 11183 11184 11185 11186 11187 11188 11189 11190 11191 11192 11193 11194 11195 11196 11197 11198 11199 11200 11201 11202 11203 11204 11205 11206 11207 11208 11209 11210 11211 11212 11213 11214 11215 11216 11217 11218 11219 11220 11221 11222 11223 11224 11225 11226 11227 11228 11229 11230 11231 11232 11233 11234 11235 11236 11237 11238 11239 11240 11241 11242 11243 11244 11245 11246 11247
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,
11248 11249
                            _("Disk '%s' does not support snapshotting"),
                            vm->def->disks[i]->src);
C
Chris Lalancette 已提交
11250 11251 11252 11253 11254 11255 11256 11257 11258
            return 0;
        }
    }

    return 1;
}

static virDomainSnapshotPtr qemuDomainSnapshotCreateXML(virDomainPtr domain,
                                                        const char *xmlDesc,
11259
                                                        unsigned int flags)
C
Chris Lalancette 已提交
11260 11261 11262 11263 11264 11265 11266 11267 11268 11269
{
    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;

11270 11271
    virCheckFlags(0, NULL);

C
Chris Lalancette 已提交
11272 11273 11274 11275 11276 11277 11278 11279 11280 11281 11282 11283 11284 11285 11286 11287 11288 11289 11290 11291 11292 11293 11294 11295 11296 11297 11298 11299 11300 11301 11302 11303 11304 11305 11306 11307 11308 11309 11310 11311 11312 11313 11314 11315 11316 11317 11318 11319 11320 11321 11322 11323 11324 11325 11326 11327 11328 11329 11330 11331 11332
    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 {
11333 11334 11335 11336 11337
        qemuDomainObjPrivatePtr priv;
        int ret;

        if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
            goto cleanup;
C
Chris Lalancette 已提交
11338 11339
        priv = vm->privateData;
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
11340
        ret = qemuMonitorCreateSnapshot(priv->mon, def->name);
C
Chris Lalancette 已提交
11341
        qemuDomainObjExitMonitorWithDriver(driver, vm);
11342
        if (qemuDomainObjEndJob(vm) == 0) {
11343
            vm = NULL;
11344 11345
            goto cleanup;
        }
11346 11347
        if (ret < 0)
            goto cleanup;
C
Chris Lalancette 已提交
11348 11349 11350 11351 11352 11353 11354 11355 11356 11357 11358 11359 11360 11361 11362 11363 11364 11365 11366 11367 11368 11369 11370 11371 11372 11373 11374 11375 11376 11377 11378 11379 11380 11381 11382 11383
    }

    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,
11384
                                       unsigned int flags)
C
Chris Lalancette 已提交
11385 11386 11387 11388 11389
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm = NULL;
    int n = -1;

11390 11391
    virCheckFlags(0, -1);

C
Chris Lalancette 已提交
11392 11393 11394 11395 11396 11397 11398 11399 11400 11401 11402 11403 11404 11405 11406 11407 11408 11409 11410 11411
    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,
11412
                                 unsigned int flags)
C
Chris Lalancette 已提交
11413 11414 11415 11416 11417
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm = NULL;
    int n = -1;

11418 11419
    virCheckFlags(0, -1);

C
Chris Lalancette 已提交
11420 11421 11422 11423 11424 11425 11426 11427 11428 11429 11430 11431 11432 11433 11434 11435 11436 11437 11438 11439 11440
    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,
11441
                                                           unsigned int flags)
C
Chris Lalancette 已提交
11442 11443 11444 11445 11446 11447
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm;
    virDomainSnapshotObjPtr snap = NULL;
    virDomainSnapshotPtr snapshot = NULL;

11448 11449
    virCheckFlags(0, NULL);

C
Chris Lalancette 已提交
11450 11451 11452 11453 11454 11455 11456 11457 11458 11459 11460 11461 11462 11463 11464 11465 11466 11467 11468 11469 11470 11471 11472 11473 11474 11475 11476
    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,
11477
                                        unsigned int flags)
C
Chris Lalancette 已提交
11478 11479 11480 11481 11482
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;

11483 11484
    virCheckFlags(0, -1);

C
Chris Lalancette 已提交
11485 11486 11487 11488 11489 11490 11491 11492 11493 11494 11495 11496 11497 11498 11499 11500 11501 11502 11503 11504
    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,
11505
                                                      unsigned int flags)
C
Chris Lalancette 已提交
11506 11507 11508 11509 11510
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm;
    virDomainSnapshotPtr snapshot = NULL;

11511 11512
    virCheckFlags(0, NULL);

C
Chris Lalancette 已提交
11513 11514 11515 11516 11517 11518 11519 11520 11521 11522 11523 11524 11525 11526 11527 11528 11529 11530 11531 11532 11533 11534 11535 11536 11537 11538
    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,
11539
                                       unsigned int flags)
C
Chris Lalancette 已提交
11540 11541 11542 11543 11544 11545 11546
{
    struct qemud_driver *driver = snapshot->domain->conn->privateData;
    virDomainObjPtr vm = NULL;
    char *xml = NULL;
    virDomainSnapshotObjPtr snap = NULL;
    char uuidstr[VIR_UUID_STRING_BUFLEN];

11547 11548
    virCheckFlags(0, NULL);

C
Chris Lalancette 已提交
11549 11550 11551 11552 11553 11554 11555 11556 11557 11558 11559 11560 11561 11562 11563 11564 11565 11566 11567 11568 11569 11570 11571 11572 11573 11574 11575
    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,
11576
                                      unsigned int flags)
C
Chris Lalancette 已提交
11577 11578 11579 11580 11581 11582 11583 11584 11585 11586
{
    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;

11587 11588
    virCheckFlags(0, -1);

C
Chris Lalancette 已提交
11589 11590 11591 11592 11593 11594 11595 11596 11597 11598 11599 11600 11601 11602 11603 11604 11605 11606 11607 11608 11609 11610 11611 11612 11613 11614 11615 11616 11617 11618 11619
    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)
11620
                goto endjob;
C
Chris Lalancette 已提交
11621 11622 11623
        }
        else {
            if (qemuDomainSnapshotSetActive(vm, driver->snapshotDir) < 0)
11624
                goto endjob;
C
Chris Lalancette 已提交
11625 11626 11627 11628

            rc = qemudStartVMDaemon(snapshot->domain->conn, driver, vm, NULL,
                                    -1);
            if (qemuDomainSnapshotSetInactive(vm, driver->snapshotDir) < 0)
11629
                goto endjob;
C
Chris Lalancette 已提交
11630
            if (rc < 0)
11631
                goto endjob;
C
Chris Lalancette 已提交
11632 11633 11634 11635 11636 11637
        }

        if (snap->def->state == VIR_DOMAIN_PAUSED) {
            /* qemu unconditionally starts the domain running again after
             * loadvm, so let's pause it to keep consistency
             */
11638
            int state = vm->state;
C
Chris Lalancette 已提交
11639
            priv = vm->privateData;
11640
            vm->state = VIR_DOMAIN_PAUSED;
C
Chris Lalancette 已提交
11641 11642 11643
            qemuDomainObjEnterMonitorWithDriver(driver, vm);
            rc = qemuMonitorStopCPUs(priv->mon);
            qemuDomainObjExitMonitorWithDriver(driver, vm);
11644 11645
            if (rc < 0) {
                vm->state = state;
11646
                goto endjob;
11647
            }
C
Chris Lalancette 已提交
11648 11649 11650 11651 11652 11653 11654 11655 11656 11657 11658 11659 11660 11661 11662 11663 11664 11665
        }

        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)) {
11666
            qemudShutdownVMDaemon(driver, vm, 0);
C
Chris Lalancette 已提交
11667 11668 11669
            event = virDomainEventNewFromObj(vm,
                                             VIR_DOMAIN_EVENT_STOPPED,
                                             VIR_DOMAIN_EVENT_STOPPED_FROM_SNAPSHOT);
11670 11671 11672 11673
            if (!vm->persistent) {
                if (qemuDomainObjEndJob(vm) > 0)
                    virDomainRemoveInactive(&driver->domains, vm);
                vm = NULL;
11674
                goto cleanup;
11675
            }
C
Chris Lalancette 已提交
11676 11677 11678
        }

        if (qemuDomainSnapshotSetActive(vm, driver->snapshotDir) < 0)
11679
            goto endjob;
C
Chris Lalancette 已提交
11680 11681 11682 11683 11684 11685
    }

    vm->state = snap->def->state;

    ret = 0;

11686
endjob:
C
Chris Lalancette 已提交
11687 11688 11689
    if (vm && qemuDomainObjEndJob(vm) == 0)
        vm = NULL;

11690
cleanup:
C
Chris Lalancette 已提交
11691 11692 11693 11694 11695 11696 11697 11698 11699 11700 11701 11702 11703 11704 11705 11706 11707 11708 11709 11710 11711 11712 11713 11714 11715 11716 11717 11718 11719 11720 11721 11722 11723 11724 11725 11726 11727 11728 11729 11730 11731 11732 11733 11734 11735 11736 11737 11738 11739 11740 11741 11742 11743 11744 11745 11746 11747 11748 11749 11750 11751 11752 11753 11754 11755 11756 11757 11758 11759 11760 11761 11762 11763 11764 11765 11766 11767 11768 11769 11770 11771 11772 11773 11774 11775 11776 11777 11778 11779 11780 11781 11782 11783 11784 11785 11786 11787 11788 11789 11790 11791 11792 11793 11794 11795 11796 11797 11798 11799 11800 11801 11802 11803 11804 11805 11806 11807 11808 11809 11810 11811 11812 11813 11814 11815 11816 11817 11818 11819 11820 11821 11822 11823 11824
    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;

11825 11826
    virCheckFlags(VIR_DOMAIN_SNAPSHOT_DELETE_CHILDREN, -1);

C
Chris Lalancette 已提交
11827 11828 11829 11830 11831 11832 11833 11834 11835 11836 11837 11838 11839 11840 11841 11842 11843
    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;
    }

11844 11845 11846
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

C
Chris Lalancette 已提交
11847 11848 11849 11850 11851 11852 11853 11854
    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)
11855
            goto endjob;
C
Chris Lalancette 已提交
11856 11857 11858 11859
    }

    ret = qemuDomainSnapshotDiscard(driver, vm, snap);

11860 11861 11862 11863
endjob:
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;

C
Chris Lalancette 已提交
11864 11865 11866 11867 11868 11869
cleanup:
    if (vm)
        virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);
    return ret;
}
11870

11871 11872 11873 11874 11875
static virDriver qemuDriver = {
    VIR_DRV_QEMU,
    "QEMU",
    qemudOpen, /* open */
    qemudClose, /* close */
D
Daniel Veillard 已提交
11876
    qemudSupportsFeature, /* supports_feature */
11877 11878
    qemudGetType, /* type */
    qemudGetVersion, /* version */
11879
    NULL, /* libvirtVersion (impl. in libvirt.c) */
11880
    virGetHostname, /* getHostname */
11881
    qemudGetMaxVCPUs, /* getMaxVcpus */
11882
    nodeGetInfo, /* nodeGetInfo */
11883 11884 11885
    qemudGetCapabilities, /* getCapabilities */
    qemudListDomains, /* listDomains */
    qemudNumDomains, /* numOfDomains */
11886
    qemudDomainCreate, /* domainCreateXML */
11887 11888 11889 11890 11891
    qemudDomainLookupByID, /* domainLookupByID */
    qemudDomainLookupByUUID, /* domainLookupByUUID */
    qemudDomainLookupByName, /* domainLookupByName */
    qemudDomainSuspend, /* domainSuspend */
    qemudDomainResume, /* domainResume */
11892
    qemudDomainShutdown, /* domainShutdown */
11893 11894 11895
    NULL, /* domainReboot */
    qemudDomainDestroy, /* domainDestroy */
    qemudDomainGetOSType, /* domainGetOSType */
11896
    qemudDomainGetMaxMemory, /* domainGetMaxMemory */
11897
    NULL, /* domainSetMaxMemory */
11898
    qemudDomainSetMemory, /* domainSetMemory */
11899 11900 11901
    qemudDomainGetInfo, /* domainGetInfo */
    qemudDomainSave, /* domainSave */
    qemudDomainRestore, /* domainRestore */
P
Paolo Bonzini 已提交
11902
    qemudDomainCoreDump, /* domainCoreDump */
11903
    qemudDomainSetVcpus, /* domainSetVcpus */
11904 11905
    qemudDomainPinVcpu, /* domainPinVcpu */
    qemudDomainGetVcpus, /* domainGetVcpus */
11906
    qemudDomainGetMaxVcpus, /* domainGetMaxVcpus */
11907 11908
    qemudDomainGetSecurityLabel, /* domainGetSecurityLabel */
    qemudNodeGetSecurityModel, /* nodeGetSecurityModel */
11909
    qemudDomainDumpXML, /* domainDumpXML */
11910
    qemuDomainXMLFromNative, /* domainXmlFromNative */
11911
    qemuDomainXMLToNative, /* domainXMLToNative */
11912 11913
    qemudListDefinedDomains, /* listDefinedDomains */
    qemudNumDefinedDomains, /* numOfDefinedDomains */
11914 11915 11916
    qemudDomainStart, /* domainCreate */
    qemudDomainDefine, /* domainDefineXML */
    qemudDomainUndefine, /* domainUndefine */
11917
    qemudDomainAttachDevice, /* domainAttachDevice */
11918
    qemudDomainAttachDeviceFlags, /* domainAttachDeviceFlags */
11919
    qemudDomainDetachDevice, /* domainDetachDevice */
11920
    qemudDomainDetachDeviceFlags, /* domainDetachDeviceFlags */
11921
    qemuDomainUpdateDeviceFlags, /* domainUpdateDeviceFlags */
11922 11923
    qemudDomainGetAutostart, /* domainGetAutostart */
    qemudDomainSetAutostart, /* domainSetAutostart */
11924 11925 11926
    qemuGetSchedulerType, /* domainGetSchedulerType */
    qemuGetSchedulerParameters, /* domainGetSchedulerParameters */
    qemuSetSchedulerParameters, /* domainSetSchedulerParameters */
D
Daniel Veillard 已提交
11927 11928
    NULL, /* domainMigratePrepare (v1) */
    qemudDomainMigratePerform, /* domainMigratePerform */
11929
    NULL, /* domainMigrateFinish */
11930
    qemudDomainBlockStats, /* domainBlockStats */
11931
    qemudDomainInterfaceStats, /* domainInterfaceStats */
11932
    qemudDomainMemoryStats, /* domainMemoryStats */
11933
    qemudDomainBlockPeek, /* domainBlockPeek */
R
Richard W.M. Jones 已提交
11934
    qemudDomainMemoryPeek, /* domainMemoryPeek */
11935
    qemuDomainGetBlockInfo, /* domainGetBlockInfo */
11936 11937
    nodeGetCellsFreeMemory, /* nodeGetCellsFreeMemory */
    nodeGetFreeMemory,  /* getFreeMemory */
11938 11939
    qemuDomainEventRegister, /* domainEventRegister */
    qemuDomainEventDeregister, /* domainEventDeregister */
D
Daniel Veillard 已提交
11940 11941
    qemudDomainMigratePrepare2, /* domainMigratePrepare2 */
    qemudDomainMigrateFinish2, /* domainMigrateFinish2 */
11942 11943 11944
    qemudNodeDeviceDettach, /* nodeDeviceDettach */
    qemudNodeDeviceReAttach, /* nodeDeviceReAttach */
    qemudNodeDeviceReset, /* nodeDeviceReset */
C
Chris Lalancette 已提交
11945
    qemudDomainMigratePrepareTunnel, /* domainMigratePrepareTunnel */
11946 11947 11948 11949
    qemuIsEncrypted, /* isEncrypted */
    qemuIsSecure, /* isSecure */
    qemuDomainIsActive, /* domainIsActive */
    qemuDomainIsPersistent, /* domainIsPersistent */
11950
    qemuCPUCompare, /* cpuCompare */
11951
    qemuCPUBaseline, /* cpuBaseline */
11952
    qemuDomainGetJobInfo, /* domainGetJobInfo */
11953
    qemuDomainAbortJob, /* domainAbortJob */
11954
    qemuDomainMigrateSetMaxDowntime, /* domainMigrateSetMaxDowntime */
11955 11956
    qemuDomainEventRegisterAny, /* domainEventRegisterAny */
    qemuDomainEventDeregisterAny, /* domainEventDeregisterAny */
11957 11958 11959
    qemuDomainManagedSave, /* domainManagedSave */
    qemuDomainHasManagedSaveImage, /* domainHasManagedSaveImage */
    qemuDomainManagedSaveRemove, /* domainManagedSaveRemove */
C
Chris Lalancette 已提交
11960 11961 11962 11963 11964 11965 11966 11967 11968
    qemuDomainSnapshotCreateXML, /* domainSnapshotCreateXML */
    qemuDomainSnapshotDumpXML, /* domainSnapshotDumpXML */
    qemuDomainSnapshotNum, /* domainSnapshotNum */
    qemuDomainSnapshotListNames, /* domainSnapshotListNames */
    qemuDomainSnapshotLookupByName, /* domainSnapshotLookupByName */
    qemuDomainHasCurrentSnapshot, /* domainHasCurrentSnapshot */
    qemuDomainSnapshotCurrent, /* domainSnapshotCurrent */
    qemuDomainRevertToSnapshot, /* domainRevertToSnapshot */
    qemuDomainSnapshotDelete, /* domainSnapshotDelete */
11969 11970 11971
};


11972
static virStateDriver qemuStateDriver = {
11973
    .name = "QEMU",
11974 11975 11976 11977
    .initialize = qemudStartup,
    .cleanup = qemudShutdown,
    .reload = qemudReload,
    .active = qemudActive,
11978
};
11979

S
Stefan Berger 已提交
11980
static int
11981
qemudVMFilterRebuild(virConnectPtr conn ATTRIBUTE_UNUSED,
S
Stefan Berger 已提交
11982 11983
                     virHashIterator iter, void *data)
{
11984 11985 11986
    struct qemud_driver *driver = qemu_driver;

    qemuDriverLock(driver);
S
Stefan Berger 已提交
11987
    virHashForEach(qemu_driver->domains.objs, iter, data);
11988 11989
    qemuDriverUnlock(driver);

S
Stefan Berger 已提交
11990 11991 11992 11993 11994 11995 11996 11997 11998
    return 0;
}


static virNWFilterCallbackDriver qemuCallbackDriver = {
    .name = "QEMU",
    .vmFilterRebuild = qemudVMFilterRebuild,
};

11999
int qemuRegister(void) {
12000 12001
    virRegisterDriver(&qemuDriver);
    virRegisterStateDriver(&qemuStateDriver);
S
Stefan Berger 已提交
12002
    virNWFilterRegisterCallbackDriver(&qemuCallbackDriver);
12003 12004
    return 0;
}