qemu_driver.c 345.6 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 38
#include <stdio.h>
#include <strings.h>
#include <stdarg.h>
#include <stdlib.h>
#include <unistd.h>
#include <errno.h>
39
#include <sys/utsname.h>
40 41 42 43
#include <sys/stat.h>
#include <fcntl.h>
#include <signal.h>
#include <paths.h>
44 45
#include <pwd.h>
#include <stdio.h>
46
#include <sys/wait.h>
47
#include <sys/ioctl.h>
48
#include <sys/un.h>
D
Daniel P. Berrange 已提交
49

50
#ifdef __linux__
51 52 53 54
# include <sys/vfs.h>
# ifndef NFS_SUPER_MAGIC
#  define NFS_SUPER_MAGIC 0x6969
# endif /* NFS_SUPER_MAGIC */
55
#endif /* __linux__ */
56

57
#include "virterror_internal.h"
58
#include "logging.h"
59
#include "datatypes.h"
60 61
#include "qemu_driver.h"
#include "qemu_conf.h"
62
#include "qemu_monitor.h"
63
#include "qemu_bridge_filter.h"
J
Jim Meyering 已提交
64
#include "c-ctype.h"
65
#include "event.h"
66
#include "buf.h"
67
#include "util.h"
68
#include "nodeinfo.h"
69
#include "stats_linux.h"
70
#include "capabilities.h"
71
#include "memory.h"
72
#include "uuid.h"
73
#include "domain_conf.h"
74 75
#include "node_device_conf.h"
#include "pci.h"
76
#include "hostusb.h"
77
#include "processinfo.h"
78 79
#include "qemu_security_stacked.h"
#include "qemu_security_dac.h"
80
#include "cgroup.h"
C
Chris Lalancette 已提交
81
#include "libvirt_internal.h"
82
#include "xml.h"
83
#include "cpu/cpu.h"
84
#include "macvtap.h"
S
Stefan Berger 已提交
85
#include "nwfilter/nwfilter_gentech_driver.h"
86
#include "hooks.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
                                  virDomainObjPtr vm);
154

155
static int qemudDomainGetMaxVcpus(virDomainPtr dom);
156

157
static int qemuDetectVcpuPIDs(struct qemud_driver *driver,
158
                              virDomainObjPtr vm);
159

160 161 162
static int qemuUpdateActivePciHostdevs(struct qemud_driver *driver,
                                       virDomainDefPtr def);

J
Jim Meyering 已提交
163
static struct qemud_driver *qemu_driver = NULL;
164

165 166 167 168 169 170 171 172 173 174 175 176 177 178 179

static void *qemuDomainObjPrivateAlloc(void)
{
    qemuDomainObjPrivatePtr priv;

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

    return priv;
}

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

180
    qemuDomainPCIAddressSetFree(priv->pciaddrs);
181 182 183
    virDomainChrDefFree(priv->monConfig);
    VIR_FREE(priv->vcpupids);

184 185 186 187 188 189 190 191 192
    /* 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);
}


193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210
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
211 212
        if (priv->monJSON)
            virBufferAddLit(buf, " json='1'");
213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238
        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) {
239
        virReportOOMError();
240 241 242
        goto error;
    }

243
    if (!(priv->monConfig->info.alias = strdup("monitor"))) {
244
        virReportOOMError();
245 246 247
        goto error;
    }

248
    if (!(monitorpath =
249
          virXPathString("string(./monitor[1]/@path)", ctxt))) {
250 251
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("no monitor path"));
252 253 254
        goto error;
    }

255
    tmp = virXPathString("string(./monitor[1]/@type)", ctxt);
256 257 258 259 260 261
    if (tmp)
        priv->monConfig->type = virDomainChrTypeFromString(tmp);
    else
        priv->monConfig->type = VIR_DOMAIN_CHR_TYPE_PTY;
    VIR_FREE(tmp);

262
    if (virXPathBoolean("count(./monitor[@json = '1']) > 0", ctxt)) {
D
Daniel P. Berrange 已提交
263
        priv->monJSON = 1;
264 265 266
    } else {
        priv->monJSON = 0;
    }
D
Daniel P. Berrange 已提交
267

268 269 270 271 272 273 274 275 276
    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);
277 278 279
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("unsupported monitor type '%s'"),
                        virDomainChrTypeToString(priv->monConfig->type));
280 281 282
        goto error;
    }

283
    n = virXPathNodeSet("./vcpus/vcpu", ctxt, &nodes);
284 285 286 287 288
    if (n < 0)
        goto error;
    if (n) {
        priv->nvcpupids = n;
        if (VIR_REALLOC_N(priv->vcpupids, priv->nvcpupids) < 0) {
289
            virReportOOMError();
290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309
            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:
310 311
    virDomainChrDefFree(priv->monConfig);
    priv->monConfig = NULL;
312 313 314 315 316 317
    VIR_FREE(nodes);
    return -1;
}



318 319 320 321 322 323 324 325 326
/*
 * 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
 */
327 328 329 330

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

331 332 333 334
static int qemuDomainObjBeginJob(virDomainObjPtr obj) ATTRIBUTE_RETURN_CHECK;
static int qemuDomainObjBeginJob(virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
335 336 337 338
    struct timeval now;
    unsigned long long then;

    if (gettimeofday(&now, NULL) < 0) {
339
        virReportSystemError(errno, "%s",
340 341 342 343 344
                             _("cannot get time of day"));
        return -1;
    }
    then = (now.tv_sec * 1000ull) + (now.tv_usec / 1000);
    then += QEMU_JOB_WAIT_TIME;
345 346 347 348

    virDomainObjRef(obj);

    while (priv->jobActive) {
349
        if (virCondWaitUntil(&priv->jobCond, &obj->lock, then) < 0) {
350
            virDomainObjUnref(obj);
351
            if (errno == ETIMEDOUT)
352 353
                qemuReportError(VIR_ERR_OPERATION_TIMEOUT,
                                "%s", _("cannot acquire state change lock"));
354
            else
355
                virReportSystemError(errno,
356
                                     "%s", _("cannot acquire job mutex"));
357 358 359
            return -1;
        }
    }
360 361
    priv->jobActive = QEMU_JOB_UNSPECIFIED;
    priv->jobSignals = 0;
362
    memset(&priv->jobSignalsData, 0, sizeof(priv->jobSignalsData));
363 364
    priv->jobStart = (now.tv_sec * 1000ull) + (now.tv_usec / 1000);
    memset(&priv->jobInfo, 0, sizeof(priv->jobInfo));
365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380

    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;
381 382 383 384
    struct timeval now;
    unsigned long long then;

    if (gettimeofday(&now, NULL) < 0) {
385
        virReportSystemError(errno, "%s",
386 387 388 389 390
                             _("cannot get time of day"));
        return -1;
    }
    then = (now.tv_sec * 1000ull) + (now.tv_usec / 1000);
    then += QEMU_JOB_WAIT_TIME;
391 392 393 394 395

    virDomainObjRef(obj);
    qemuDriverUnlock(driver);

    while (priv->jobActive) {
396
        if (virCondWaitUntil(&priv->jobCond, &obj->lock, then) < 0) {
397
            virDomainObjUnref(obj);
398
            if (errno == ETIMEDOUT)
399 400
                qemuReportError(VIR_ERR_OPERATION_TIMEOUT,
                                "%s", _("cannot acquire state change lock"));
401
            else
402
                virReportSystemError(errno,
403
                                     "%s", _("cannot acquire job mutex"));
M
Matthias Bolte 已提交
404
            qemuDriverLock(driver);
405 406 407
            return -1;
        }
    }
408 409
    priv->jobActive = QEMU_JOB_UNSPECIFIED;
    priv->jobSignals = 0;
410
    memset(&priv->jobSignalsData, 0, sizeof(priv->jobSignalsData));
411 412
    priv->jobStart = (now.tv_sec * 1000ull) + (now.tv_usec / 1000);
    memset(&priv->jobInfo, 0, sizeof(priv->jobInfo));
413 414 415 416 417 418 419 420 421 422 423 424 425

    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
426 427 428
 *
 * Returns remaining refcount on 'obj', maybe 0 to indicated it
 * was deleted
429
 */
430
static int ATTRIBUTE_RETURN_CHECK qemuDomainObjEndJob(virDomainObjPtr obj)
431 432 433
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

434 435
    priv->jobActive = QEMU_JOB_NONE;
    priv->jobSignals = 0;
436
    memset(&priv->jobSignalsData, 0, sizeof(priv->jobSignalsData));
437 438
    priv->jobStart = 0;
    memset(&priv->jobInfo, 0, sizeof(priv->jobInfo));
439 440
    virCondSignal(&priv->jobCond);

441
    return virDomainObjUnref(obj);
442 443 444 445 446 447 448 449 450 451 452
}


/*
 * 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
 */
453 454 455 456 457
static void qemuDomainObjEnterMonitor(virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;

    qemuMonitorLock(priv->mon);
458
    qemuMonitorRef(priv->mon);
459
    virDomainObjUnlock(obj);
460 461 462
}


463 464 465 466
/* obj must NOT be locked before calling, qemud_driver must be unlocked
 *
 * Should be paired with an earlier  qemuDomainObjEnterMonitor() call
 */
467 468 469
static void qemuDomainObjExitMonitor(virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
470 471 472 473 474 475
    int refs;

    refs = qemuMonitorUnref(priv->mon);

    if (refs > 0)
        qemuMonitorUnlock(priv->mon);
476

477
    virDomainObjLock(obj);
478 479 480 481 482

    if (refs == 0) {
        virDomainObjUnref(obj);
        priv->mon = NULL;
    }
483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498
}


/*
 * 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);
499
    qemuMonitorRef(priv->mon);
500 501 502 503 504 505 506 507 508 509 510 511 512
    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;
513 514 515 516 517 518
    int refs;

    refs = qemuMonitorUnref(priv->mon);

    if (refs > 0)
        qemuMonitorUnlock(priv->mon);
519 520 521

    qemuDriverLock(driver);
    virDomainObjLock(obj);
522 523 524 525 526

    if (refs == 0) {
        virDomainObjUnref(obj);
        priv->mon = NULL;
    }
527 528 529
}


530 531 532 533 534 535 536 537 538
static int qemuCgroupControllerActive(struct qemud_driver *driver,
                                      int controller)
{
    if (driver->cgroup == NULL)
        return 0;
    if (driver->cgroupControllers & (1 << controller))
        return 1;
    return 0;
}
539

540
static int
541
qemudLogFD(struct qemud_driver *driver, const char* name)
542 543 544
{
    char logfile[PATH_MAX];
    mode_t logmode;
G
Guido Günther 已提交
545
    int ret, fd = -1;
546

547 548
    if ((ret = snprintf(logfile, sizeof(logfile), "%s/%s.log",
                        driver->logDir, name))
G
Guido Günther 已提交
549
        < 0 || ret >= sizeof(logfile)) {
550
        virReportOOMError();
551 552 553 554
        return -1;
    }

    logmode = O_CREAT | O_WRONLY;
555 556
    /* Only logrotate files in /var/log, so only append if running privileged */
    if (driver->privileged)
557
        logmode |= O_APPEND;
558 559 560
    else
        logmode |= O_TRUNC;

561
    if ((fd = open(logfile, logmode, S_IRUSR | S_IWUSR)) < 0) {
562
        virReportSystemError(errno,
563 564
                             _("failed to create logfile %s"),
                             logfile);
565 566
        return -1;
    }
567
    if (virSetCloseExec(fd) < 0) {
568
        virReportSystemError(errno, "%s",
569
                             _("Unable to set VM logfile close-on-exec flag"));
570 571 572 573 574 575 576
        close(fd);
        return -1;
    }
    return fd;
}


577
static int
578
qemudLogReadFD(const char* logDir, const char* name, off_t pos)
579 580 581 582 583 584 585
{
    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)) {
586 587 588
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("failed to build logfile name %s/%s.log"),
                        logDir, name);
589 590 591 592 593
        return -1;
    }


    if ((fd = open(logfile, logmode)) < 0) {
594
        virReportSystemError(errno,
595 596
                             _("failed to create logfile %s"),
                             logfile);
597 598
        return -1;
    }
599
    if (virSetCloseExec(fd) < 0) {
600
        virReportSystemError(errno, "%s",
601
                             _("Unable to set VM logfile close-on-exec flag"));
602 603 604
        close(fd);
        return -1;
    }
605
    if (pos < 0 || lseek(fd, pos, SEEK_SET) < 0) {
606
      virReportSystemError(pos < 0 ? 0 : errno,
607 608
                             _("Unable to seek to %lld in %s"),
                             (long long) pos, logfile);
609 610 611 612 613 614
        close(fd);
    }
    return fd;
}


615 616 617 618 619 620 621 622 623 624 625 626
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
627
        !virDomainObjIsActive(vm)) {
628 629 630 631 632 633
        int ret;

        virResetLastError();
        ret = qemudStartVMDaemon(data->conn, data->driver, vm, NULL, -1);
        if (ret < 0) {
            virErrorPtr err = virGetLastError();
634
            VIR_ERROR(_("Failed to autostart VM '%s': %s"),
635 636 637 638 639 640 641 642 643 644 645 646 647 648
                      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);
}

649 650
static void
qemudAutostartConfigs(struct qemud_driver *driver) {
651 652 653 654 655
    /* 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
     */
656 657 658
    virConnectPtr conn = virConnectOpen(driver->privileged ?
                                        "qemu:///system" :
                                        "qemu:///session");
659
    /* Ignoring NULL conn which is mostly harmless here */
660
    struct qemuAutostartData data = { driver, conn };
661

662
    qemuDriverLock(driver);
663
    virHashForEach(driver->domains.objs, qemuAutostartDomain, &data);
664
    qemuDriverUnlock(driver);
665

666 667
    if (conn)
        virConnectClose(conn);
668 669
}

670 671 672 673 674 675 676 677 678

/**
 * qemudRemoveDomainStatus
 *
 * remove all state files of a domain from statedir
 *
 * Returns 0 on success
 */
static int
679
qemudRemoveDomainStatus(struct qemud_driver *driver,
680 681
                        virDomainObjPtr vm)
{
682
    char ebuf[1024];
683 684 685
    char *file = NULL;

    if (virAsprintf(&file, "%s/%s.xml", driver->stateDir, vm->def->name) < 0) {
686
        virReportOOMError();
D
Daniel Veillard 已提交
687
        return(-1);
688 689
    }

690 691
    if (unlink(file) < 0 && errno != ENOENT && errno != ENOTDIR)
        VIR_WARN(_("Failed to remove domain XML for %s: %s"),
D
Daniel Veillard 已提交
692 693 694
                 vm->def->name, virStrerror(errno, ebuf, sizeof(ebuf)));
    VIR_FREE(file);

695 696 697
    if (virFileDeletePid(driver->stateDir, vm->def->name) != 0)
        VIR_WARN(_("Failed to remove PID file for %s: %s"),
                 vm->def->name, virStrerror(errno, ebuf, sizeof(ebuf)));
698

D
Daniel Veillard 已提交
699

700
    return 0;
701 702
}

703 704 705 706 707 708 709 710 711 712 713 714 715 716

/*
 * 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;

717
    VIR_DEBUG("Received EOF on %p '%s'", vm, vm->def->name);
718 719 720 721 722 723 724 725
    virDomainObjLock(vm);

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

726
    qemudShutdownVMDaemon(driver, vm);
727 728 729 730 731 732 733 734 735 736 737 738 739
    if (!vm->persistent)
        virDomainRemoveInactive(&driver->domains, vm);
    else
        virDomainObjUnlock(vm);

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


740 741 742
static virDomainDiskDefPtr
findDomainDiskByPath(virDomainObjPtr vm,
                     const char *path)
743 744 745 746 747 748 749
{
    int i;

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

        disk = vm->def->disks[i];
750 751
        if (disk->src != NULL && STREQ(disk->src, path))
            return disk;
752
    }
753 754 755 756

    qemuReportError(VIR_ERR_INTERNAL_ERROR,
                    _("no disk found with path %s"),
                    path);
757 758 759
    return NULL;
}

760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779
static virDomainDiskDefPtr
findDomainDiskByAlias(virDomainObjPtr vm,
                      const char *alias)
{
    int i;

    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;
}

780
static int
781 782 783 784
getVolumeQcowPassphrase(virConnectPtr conn,
                        virDomainDiskDefPtr disk,
                        char **secretRet,
                        size_t *secretLen)
785 786 787 788 789
{
    virSecretPtr secret;
    char *passphrase;
    unsigned char *data;
    size_t size;
790
    int ret = -1;
791
    virStorageEncryptionPtr enc;
792

793 794 795 796 797 798 799
    if (!disk->encryption) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("disk %s does not have any encryption information"),
                        disk->src);
        return -1;
    }
    enc = disk->encryption;
800 801

    if (!conn) {
802 803
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        "%s", _("cannot find secrets without a connection"));
804
        goto cleanup;
805 806 807 808 809
    }

    if (conn->secretDriver == NULL ||
        conn->secretDriver->lookupByUUID == NULL ||
        conn->secretDriver->getValue == NULL) {
810 811
        qemuReportError(VIR_ERR_NO_SUPPORT, "%s",
                        _("secret storage not supported"));
812
        goto cleanup;
813 814 815 816 817 818
    }

    if (enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_QCOW ||
        enc->nsecrets != 1 ||
        enc->secrets[0]->type !=
        VIR_STORAGE_ENCRYPTION_SECRET_TYPE_PASSPHRASE) {
819
        qemuReportError(VIR_ERR_INVALID_DOMAIN,
820
                        _("invalid <encryption> for volume %s"), disk->src);
821
        goto cleanup;
822 823 824 825 826
    }

    secret = conn->secretDriver->lookupByUUID(conn,
                                              enc->secrets[0]->uuid);
    if (secret == NULL)
827
        goto cleanup;
828 829 830 831
    data = conn->secretDriver->getValue(secret, &size,
                                        VIR_SECRET_GET_VALUE_INTERNAL_CALL);
    virUnrefSecret(secret);
    if (data == NULL)
832
        goto cleanup;
833 834 835 836

    if (memchr(data, '\0', size) != NULL) {
        memset(data, 0, size);
        VIR_FREE(data);
837 838
        qemuReportError(VIR_ERR_INVALID_SECRET,
                        _("format='qcow' passphrase for %s must not contain a "
839
                          "'\\0'"), disk->src);
840
        goto cleanup;
841 842 843 844 845
    }

    if (VIR_ALLOC_N(passphrase, size + 1) < 0) {
        memset(data, 0, size);
        VIR_FREE(data);
846
        virReportOOMError();
847
        goto cleanup;
848 849 850 851 852 853 854 855 856 857
    }
    memcpy(passphrase, data, size);
    passphrase[size] = '\0';

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

    *secretRet = passphrase;
    *secretLen = size;

858 859 860
    ret = 0;

cleanup:
861 862
    return ret;
}
863

864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884
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);
885
    return ret;
886 887
}

888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909

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;
}


910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941
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;
}


942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970
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;
}


971 972 973 974 975 976
static int
qemuHandleDomainWatchdog(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                         virDomainObjPtr vm,
                         int action)
{
    struct qemud_driver *driver = qemu_driver;
977 978
    virDomainEventPtr watchdogEvent = NULL;
    virDomainEventPtr lifecycleEvent = NULL;
979 980

    virDomainObjLock(vm);
981 982 983 984 985 986 987 988 989 990 991 992 993 994
    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);
    }
995 996
    virDomainObjUnlock(vm);

997
    if (watchdogEvent || lifecycleEvent) {
998
        qemuDriverLock(driver);
999 1000 1001 1002
        if (watchdogEvent)
            qemuDomainEventQueue(driver, watchdogEvent);
        if (lifecycleEvent)
            qemuDomainEventQueue(driver, lifecycleEvent);
1003 1004 1005 1006 1007 1008 1009
        qemuDriverUnlock(driver);
    }

    return 0;
}


1010 1011 1012 1013 1014 1015 1016
static int
qemuHandleDomainIOError(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                        virDomainObjPtr vm,
                        const char *diskAlias,
                        int action)
{
    struct qemud_driver *driver = qemu_driver;
1017 1018
    virDomainEventPtr ioErrorEvent = NULL;
    virDomainEventPtr lifecycleEvent = NULL;
1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033
    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 = "";
    }

1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047
    ioErrorEvent = virDomainEventIOErrorNewFromObj(vm, srcPath, devAlias, action);

    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);
    }
1048 1049
    virDomainObjUnlock(vm);

1050
    if (ioErrorEvent || lifecycleEvent) {
1051
        qemuDriverLock(driver);
1052 1053 1054 1055
        if (ioErrorEvent)
            qemuDomainEventQueue(driver, ioErrorEvent);
        if (lifecycleEvent)
            qemuDomainEventQueue(driver, lifecycleEvent);
1056 1057 1058 1059 1060 1061 1062
        qemuDriverUnlock(driver);
    }

    return 0;
}


1063 1064 1065 1066 1067 1068 1069 1070 1071 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
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;
}


1155 1156 1157
static qemuMonitorCallbacks monitorCallbacks = {
    .eofNotify = qemuHandleMonitorEOF,
    .diskSecretLookup = findVolumeQcowPassphrase,
1158
    .domainStop = qemuHandleDomainStop,
1159
    .domainReset = qemuHandleDomainReset,
1160
    .domainRTCChange = qemuHandleDomainRTCChange,
1161
    .domainWatchdog = qemuHandleDomainWatchdog,
1162
    .domainIOError = qemuHandleDomainIOError,
1163
    .domainGraphics = qemuHandleDomainGraphics,
1164 1165
};

1166
static int
1167
qemuConnectMonitor(struct qemud_driver *driver, virDomainObjPtr vm)
1168
{
1169
    qemuDomainObjPrivatePtr priv = vm->privateData;
1170
    int ret;
1171

1172 1173 1174 1175
    /* Hold an extra reference because we can't allow 'vm' to be
     * deleted while the monitor is active */
    virDomainObjRef(vm);

1176 1177
    if ((priv->mon = qemuMonitorOpen(vm,
                                     priv->monConfig,
D
Daniel P. Berrange 已提交
1178
                                     priv->monJSON,
1179
                                     &monitorCallbacks)) == NULL) {
1180
        VIR_ERROR(_("Failed to connect monitor for %s"), vm->def->name);
1181
        return -1;
1182
    }
1183

1184 1185 1186 1187 1188 1189 1190 1191 1192 1193
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
    ret = qemuMonitorSetCapabilities(priv->mon);
    qemuDomainObjExitMonitorWithDriver(driver, vm);

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

    return ret;
1194
}
1195 1196 1197 1198

/*
 * Open an existing VM's monitor, re-detect VCPU threads
 * and re-reserve the security labels in use
1199
 */
1200 1201
static void
qemuReconnectDomain(void *payload, const char *name ATTRIBUTE_UNUSED, void *opaque)
1202
{
1203 1204
    virDomainObjPtr obj = payload;
    struct qemud_driver *driver = opaque;
1205
    qemuDomainObjPrivatePtr priv;
1206
    unsigned long long qemuCmdFlags;
1207 1208

    virDomainObjLock(obj);
1209

1210 1211
    VIR_DEBUG("Reconnect monitor to %p '%s'", obj, obj->def->name);

1212 1213
    priv = obj->privateData;

1214
    /* XXX check PID liveliness & EXE path */
1215
    if (qemuConnectMonitor(driver, obj) < 0)
1216
        goto error;
1217

1218 1219 1220 1221
    if (qemuUpdateActivePciHostdevs(driver, obj->def) < 0) {
        goto error;
    }

1222 1223 1224 1225 1226 1227 1228 1229 1230
    /* 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;

1231 1232 1233
    if (!(priv->pciaddrs = qemuDomainPCIAddressSetCreate(obj->def)))
        goto error;

1234
    if (driver->securityDriver &&
1235
        driver->securityDriver->domainReserveSecurityLabel &&
1236
        driver->securityDriver->domainReserveSecurityLabel(obj) < 0)
1237
        goto error;
1238

1239 1240
    if (obj->def->id >= driver->nextvmid)
        driver->nextvmid = obj->def->id + 1;
1241

1242 1243
    virDomainObjUnlock(obj);
    return;
1244

1245
error:
1246 1247 1248
    /* 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 */
1249
    qemudShutdownVMDaemon(driver, obj);
1250 1251 1252 1253
    if (!obj->persistent)
        virDomainRemoveInactive(&driver->domains, obj);
    else
        virDomainObjUnlock(obj);
1254
}
1255

1256
/**
1257
 * qemudReconnectDomains
1258 1259 1260 1261 1262 1263 1264
 *
 * Try to re-open the resources for live VMs that we care
 * about.
 */
static void
qemuReconnectDomains(struct qemud_driver *driver)
{
1265
    virHashForEach(driver->domains.objs, qemuReconnectDomain, driver);
1266 1267
}

1268

1269 1270 1271 1272 1273 1274
static int
qemudSecurityInit(struct qemud_driver *qemud_drv)
{
    int ret;
    virSecurityDriverPtr security_drv;

1275 1276 1277
    qemuSecurityStackedSetDriver(qemud_drv);
    qemuSecurityDACSetDriver(qemud_drv);

1278 1279 1280 1281 1282 1283
    ret = virSecurityDriverStartup(&security_drv,
                                   qemud_drv->securityDriverName);
    if (ret == -1) {
        VIR_ERROR0(_("Failed to start security driver"));
        return -1;
    }
1284 1285 1286

    /* No primary security driver wanted to be enabled: just setup
     * the DAC driver on its own */
1287
    if (ret == -2) {
1288
        qemud_drv->securityDriver = &qemuDACSecurityDriver;
1289
        VIR_INFO0(_("No security driver available"));
1290 1291 1292 1293 1294
    } else {
        qemud_drv->securityPrimaryDriver = security_drv;
        qemud_drv->securitySecondaryDriver = &qemuDACSecurityDriver;
        qemud_drv->securityDriver = &qemuStackedSecurityDriver;
        VIR_INFO("Initialized security driver %s", security_drv->name);
1295 1296
    }

1297
    return 0;
1298
}
1299 1300


1301 1302
static virCapsPtr
qemuCreateCapabilities(virCapsPtr oldcaps,
1303
                       struct qemud_driver *driver)
1304 1305 1306 1307 1308
{
    virCapsPtr caps;

    /* Basic host arch / guest machine capabilities */
    if (!(caps = qemudCapsInit(oldcaps))) {
1309
        virReportOOMError();
1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320
        return NULL;
    }

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


    /* Security driver data */
1321
    if (driver->securityPrimaryDriver) {
1322 1323
        const char *doi, *model;

1324 1325
        doi = virSecurityDriverGetDOI(driver->securityPrimaryDriver);
        model = virSecurityDriverGetModel(driver->securityPrimaryDriver);
1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338

        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:
1339
    virReportOOMError();
1340 1341 1342
    virCapabilitiesFree(caps);
    return NULL;
}
1343

C
Chris Lalancette 已提交
1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 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 1404 1405 1406
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);
        VIR_FREE(fullpath);
        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)));
            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);
            VIR_FREE(xmlStr);
            continue;
        }

1407
        virDomainSnapshotAssignDef(&vm->snapshots, def);
C
Chris Lalancette 已提交
1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429

        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);
}

1430 1431 1432 1433 1434 1435
/**
 * qemudStartup:
 *
 * Initialization function for the QEmu daemon
 */
static int
1436
qemudStartup(int privileged) {
1437
    char *base = NULL;
D
Daniel P. Berrange 已提交
1438
    char driverConf[PATH_MAX];
1439
    int rc;
1440

1441
    if (VIR_ALLOC(qemu_driver) < 0)
1442 1443
        return -1;

1444
    if (virMutexInit(&qemu_driver->lock) < 0) {
1445
        VIR_ERROR("%s", _("cannot initialize mutex"));
1446 1447 1448
        VIR_FREE(qemu_driver);
        return -1;
    }
1449
    qemuDriverLock(qemu_driver);
1450
    qemu_driver->privileged = privileged;
1451

1452 1453 1454
    /* Don't have a dom0 so start from 1 */
    qemu_driver->nextvmid = 1;

1455 1456 1457
    if (virDomainObjListInit(&qemu_driver->domains) < 0)
        goto out_of_memory;

1458
    /* Init callback list */
1459
    if (VIR_ALLOC(qemu_driver->domainEventCallbacks) < 0)
1460
        goto out_of_memory;
1461 1462 1463 1464 1465 1466
    if (!(qemu_driver->domainEventQueue = virDomainEventQueueNew()))
        goto out_of_memory;

    if ((qemu_driver->domainEventTimer =
         virEventAddTimeout(-1, qemuDomainEventFlush, qemu_driver, NULL)) < 0)
        goto error;
1467

1468
    if (privileged) {
1469 1470
        if (virAsprintf(&qemu_driver->logDir,
                        "%s/log/libvirt/qemu", LOCAL_STATE_DIR) == -1)
1471
            goto out_of_memory;
1472

D
Daniel P. Berrange 已提交
1473
        if ((base = strdup (SYSCONF_DIR "/libvirt")) == NULL)
1474
            goto out_of_memory;
1475 1476

        if (virAsprintf(&qemu_driver->stateDir,
1477
                      "%s/run/libvirt/qemu", LOCAL_STATE_DIR) == -1)
1478
            goto out_of_memory;
1479 1480 1481 1482 1483 1484 1485 1486

        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;
1487 1488 1489
        if (virAsprintf(&qemu_driver->saveDir,
                      "%s/lib/libvirt/qemu/save/", LOCAL_STATE_DIR) == -1)
            goto out_of_memory;
C
Chris Lalancette 已提交
1490 1491 1492
        if (virAsprintf(&qemu_driver->snapshotDir,
                        "%s/lib/libvirt/qemu/snapshot", LOCAL_STATE_DIR) == -1)
            goto out_of_memory;
1493
    } else {
1494
        uid_t uid = geteuid();
1495
        char *userdir = virGetUserDirectory(uid);
1496
        if (!userdir)
1497
            goto error;
1498

1499
        if (virAsprintf(&qemu_driver->logDir,
1500 1501
                        "%s/.libvirt/qemu/log", userdir) == -1) {
            VIR_FREE(userdir);
1502
            goto out_of_memory;
1503
        }
1504

1505 1506
        if (virAsprintf(&base, "%s/.libvirt", userdir) == -1) {
            VIR_FREE(userdir);
1507
            goto out_of_memory;
1508 1509
        }
        VIR_FREE(userdir);
1510 1511 1512

        if (virAsprintf(&qemu_driver->stateDir, "%s/qemu/run", base) == -1)
            goto out_of_memory;
1513 1514 1515 1516
        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;
1517 1518
        if (virAsprintf(&qemu_driver->saveDir, "%s/qemu/save", base) == -1)
            goto out_of_memory;
C
Chris Lalancette 已提交
1519 1520
        if (virAsprintf(&qemu_driver->snapshotDir, "%s/qemu/snapshot", base) == -1)
            goto out_of_memory;
1521 1522
    }

L
Laine Stump 已提交
1523
    if (virFileMakePath(qemu_driver->stateDir) != 0) {
1524
        char ebuf[1024];
1525
        VIR_ERROR(_("Failed to create state dir '%s': %s"),
1526
                  qemu_driver->stateDir, virStrerror(errno, ebuf, sizeof ebuf));
1527
        goto error;
1528
    }
L
Laine Stump 已提交
1529
    if (virFileMakePath(qemu_driver->libDir) != 0) {
1530
        char ebuf[1024];
1531
        VIR_ERROR(_("Failed to create lib dir '%s': %s"),
1532 1533 1534
                  qemu_driver->libDir, virStrerror(errno, ebuf, sizeof ebuf));
        goto error;
    }
L
Laine Stump 已提交
1535
    if (virFileMakePath(qemu_driver->cacheDir) != 0) {
1536
        char ebuf[1024];
1537
        VIR_ERROR(_("Failed to create cache dir '%s': %s"),
1538 1539 1540
                  qemu_driver->cacheDir, virStrerror(errno, ebuf, sizeof ebuf));
        goto error;
    }
1541 1542 1543 1544 1545 1546
    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
1547 1548 1549 1550 1551 1552
    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;
    }
1553 1554 1555 1556

    /* Configuration paths are either ~/.libvirt/qemu/... (session) or
     * /etc/libvirt/qemu/... (system).
     */
D
Daniel P. Berrange 已提交
1557
    if (snprintf (driverConf, sizeof(driverConf), "%s/qemu.conf", base) == -1)
1558
        goto out_of_memory;
D
Daniel P. Berrange 已提交
1559
    driverConf[sizeof(driverConf)-1] = '\0';
1560

1561
    if (virAsprintf(&qemu_driver->configDir, "%s/qemu", base) == -1)
1562 1563
        goto out_of_memory;

1564
    if (virAsprintf(&qemu_driver->autostartDir, "%s/qemu/autostart", base) == -1)
1565 1566
        goto out_of_memory;

1567
    VIR_FREE(base);
1568

1569 1570 1571 1572 1573 1574 1575
    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)));
    }

1576 1577 1578 1579
    if (qemudLoadDriverConfig(qemu_driver, driverConf) < 0) {
        goto error;
    }

1580 1581
    if (qemudSecurityInit(qemu_driver) < 0)
        goto error;
D
Daniel P. Berrange 已提交
1582

1583
    if ((qemu_driver->caps = qemuCreateCapabilities(NULL,
1584
                                                    qemu_driver)) == NULL)
1585
        goto error;
1586

1587
    if ((qemu_driver->activePciHostdevs = pciDeviceListNew()) == NULL)
1588 1589
        goto error;

1590 1591
    if (privileged) {
        if (chown(qemu_driver->libDir, qemu_driver->user, qemu_driver->group) < 0) {
1592
            virReportSystemError(errno,
1593 1594 1595 1596 1597
                                 _("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) {
1598
            virReportSystemError(errno,
1599 1600 1601 1602
                                 _("unable to set ownership of '%s' to %d:%d"),
                                 qemu_driver->cacheDir, qemu_driver->user, qemu_driver->group);
            goto error;
        }
1603 1604 1605 1606 1607 1608
        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
1609 1610 1611 1612 1613 1614
        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;
        }
1615 1616
    }

1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630
    /* 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) {
1631
            virReportSystemError(rc,
1632 1633 1634 1635 1636 1637
                                 _("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) {
1638
            virReportSystemError(errno,
1639 1640 1641 1642 1643 1644 1645 1646 1647
                                 _("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;
    }

1648
    /* Get all the running persistent or transient configs first */
1649
    if (virDomainLoadAllConfigs(qemu_driver->caps,
1650 1651 1652 1653 1654 1655 1656 1657 1658
                                &qemu_driver->domains,
                                qemu_driver->stateDir,
                                NULL,
                                1, NULL, NULL) < 0)
        goto error;

    qemuReconnectDomains(qemu_driver);

    /* Then inactive persistent configs */
1659
    if (virDomainLoadAllConfigs(qemu_driver->caps,
1660 1661
                                &qemu_driver->domains,
                                qemu_driver->configDir,
1662
                                qemu_driver->autostartDir,
1663
                                0, NULL, NULL) < 0)
1664
        goto error;
C
Chris Lalancette 已提交
1665 1666 1667 1668 1669


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

1670 1671
    qemuDriverUnlock(qemu_driver);

1672 1673
    qemudAutostartConfigs(qemu_driver);

1674

1675 1676
    return 0;

1677
out_of_memory:
1678
    virReportOOMError();
1679 1680 1681
error:
    if (qemu_driver)
        qemuDriverUnlock(qemu_driver);
1682
    VIR_FREE(base);
1683
    qemudShutdown();
1684 1685 1686
    return -1;
}

1687 1688 1689 1690
static void qemudNotifyLoadDomain(virDomainObjPtr vm, int newVM, void *opaque)
{
    struct qemud_driver *driver = opaque;

1691 1692 1693 1694 1695 1696 1697 1698
    if (newVM) {
        virDomainEventPtr event =
            virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_DEFINED,
                                     VIR_DOMAIN_EVENT_DEFINED_ADDED);
        if (event)
            qemuDomainEventQueue(driver, event);
    }
1699 1700
}

1701 1702 1703 1704 1705 1706 1707 1708
/**
 * qemudReload:
 *
 * Function to restart the QEmu daemon, it will recheck the configuration
 * files and update its state and the networking
 */
static int
qemudReload(void) {
1709 1710 1711
    if (!qemu_driver)
        return 0;

1712
    qemuDriverLock(qemu_driver);
1713
    virDomainLoadAllConfigs(qemu_driver->caps,
1714 1715
                            &qemu_driver->domains,
                            qemu_driver->configDir,
1716
                            qemu_driver->autostartDir,
1717
                            0, qemudNotifyLoadDomain, qemu_driver);
1718
    qemuDriverUnlock(qemu_driver);
1719

1720
    qemudAutostartConfigs(qemu_driver);
1721 1722

    return 0;
1723 1724
}

1725 1726 1727 1728 1729 1730 1731 1732 1733 1734
/**
 * 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) {
1735
    int active = 0;
1736

1737 1738 1739
    if (!qemu_driver)
        return 0;

1740
    /* XXX having to iterate here is not great because it requires many locks */
1741
    qemuDriverLock(qemu_driver);
1742
    active = virDomainObjListNumOfDomains(&qemu_driver->domains, 1);
1743 1744
    qemuDriverUnlock(qemu_driver);
    return active;
1745 1746
}

1747 1748 1749 1750 1751 1752 1753
/**
 * qemudShutdown:
 *
 * Shutdown the QEmu daemon, it will stop all active domains and networks
 */
static int
qemudShutdown(void) {
1754
    int i;
1755

1756
    if (!qemu_driver)
1757
        return -1;
1758

1759
    qemuDriverLock(qemu_driver);
1760
    pciDeviceListFree(qemu_driver->activePciHostdevs);
1761 1762
    virCapabilitiesFree(qemu_driver->caps);

1763
    virDomainObjListDeinit(&qemu_driver->domains);
1764

1765
    VIR_FREE(qemu_driver->securityDriverName);
1766
    VIR_FREE(qemu_driver->logDir);
1767 1768
    VIR_FREE(qemu_driver->configDir);
    VIR_FREE(qemu_driver->autostartDir);
1769
    VIR_FREE(qemu_driver->stateDir);
1770 1771
    VIR_FREE(qemu_driver->libDir);
    VIR_FREE(qemu_driver->cacheDir);
1772
    VIR_FREE(qemu_driver->saveDir);
C
Chris Lalancette 已提交
1773
    VIR_FREE(qemu_driver->snapshotDir);
1774
    VIR_FREE(qemu_driver->vncTLSx509certdir);
J
Jim Meyering 已提交
1775
    VIR_FREE(qemu_driver->vncListen);
1776
    VIR_FREE(qemu_driver->vncPassword);
1777
    VIR_FREE(qemu_driver->vncSASLdir);
1778
    VIR_FREE(qemu_driver->saveImageFormat);
1779 1780
    VIR_FREE(qemu_driver->hugetlbfs_mount);
    VIR_FREE(qemu_driver->hugepage_path);
D
Daniel P. Berrange 已提交
1781

1782 1783 1784 1785 1786 1787
    if (qemu_driver->cgroupDeviceACL) {
        for (i = 0 ; qemu_driver->cgroupDeviceACL[i] != NULL ; i++)
            VIR_FREE(qemu_driver->cgroupDeviceACL[i]);
        VIR_FREE(qemu_driver->cgroupDeviceACL);
    }

1788 1789
    /* Free domain callback list */
    virDomainEventCallbackListFree(qemu_driver->domainEventCallbacks);
1790 1791 1792 1793
    virDomainEventQueueFree(qemu_driver->domainEventQueue);

    if (qemu_driver->domainEventTimer != -1)
        virEventRemoveTimeout(qemu_driver->domainEventTimer);
1794

1795 1796 1797
    if (qemu_driver->brctl)
        brShutdown(qemu_driver->brctl);

1798 1799
    virCgroupFree(&qemu_driver->cgroup);

1800
    qemuDriverUnlock(qemu_driver);
1801
    virMutexDestroy(&qemu_driver->lock);
1802
    VIR_FREE(qemu_driver);
1803 1804

    return 0;
1805 1806
}

1807
typedef int qemuLogHandleOutput(virDomainObjPtr vm,
1808 1809
                                const char *output,
                                int fd);
1810 1811 1812 1813 1814

/*
 * Returns -1 for error, 0 on success
 */
static int
1815
qemudReadLogOutput(virDomainObjPtr vm,
1816 1817
                   int fd,
                   char *buf,
G
Guido Günther 已提交
1818
                   size_t buflen,
1819
                   qemuLogHandleOutput func,
1820 1821 1822
                   const char *what,
                   int timeout)
{
1823
    int retries = (timeout*10);
1824
    int got = 0;
1825 1826 1827
    buf[0] = '\0';

    while (retries) {
1828
        ssize_t func_ret, ret;
1829
        int isdead = 0;
G
Guido Günther 已提交
1830

1831
        func_ret = func(vm, buf, fd);
1832

1833 1834
        if (kill(vm->pid, 0) == -1 && errno == ESRCH)
            isdead = 1;
1835

1836 1837
        /* Any failures should be detected before we read the log, so we
         * always have something useful to report on failure. */
1838 1839
        ret = saferead(fd, buf+got, buflen-got-1);
        if (ret < 0) {
1840
            virReportSystemError(errno,
1841 1842 1843 1844 1845
                                 _("Failure while reading %s log output"),
                                 what);
            return -1;
        }

1846 1847 1848
        got += ret;
        buf[got] = '\0';
        if (got == buflen-1) {
1849
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
1850 1851
                            _("Out of space while reading %s log output: %s"),
                            what, buf);
1852 1853 1854 1855
            return -1;
        }

        if (isdead) {
1856
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
1857 1858
                            _("Process exited while reading %s log output: %s"),
                            what, buf);
1859 1860 1861
            return -1;
        }

1862 1863
        if (func_ret <= 0)
            return func_ret;
1864 1865 1866 1867

        usleep(100*1000);
        retries--;
    }
1868

1869
    qemuReportError(VIR_ERR_INTERNAL_ERROR,
1870 1871
                    _("Timed out while reading %s log output: %s"),
                    what, buf);
1872 1873 1874
    return -1;
}

1875

1876 1877 1878 1879 1880 1881 1882 1883
/*
 * 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
 */
1884
static int
1885
qemudExtractTTYPath(const char *haystack,
1886 1887
                    size_t *offset,
                    char **path)
1888
{
1889
    static const char needle[] = "char device redirected to";
1890
    char *tmp, *dev;
1891

1892
    VIR_FREE(*path);
1893
    /* First look for our magic string */
1894 1895 1896 1897 1898
    if (!(tmp = strstr(haystack + *offset, needle))) {
        return 1;
    }
    tmp += sizeof(needle);
    dev = tmp;
1899

1900 1901 1902 1903 1904
    /*
     * And look for first whitespace character and nul terminate
     * to mark end of the pty path
     */
    while (*tmp) {
1905
        if (c_isspace(*tmp)) {
1906 1907
            *path = strndup(dev, tmp-dev);
            if (*path == NULL) {
1908
                virReportOOMError();
1909 1910
                return -1;
            }
1911

1912
            /* ... now further update offset till we get EOL */
1913
            *offset = tmp - haystack;
1914 1915
            return 0;
        }
1916
        tmp++;
1917 1918 1919 1920 1921
    }

    /*
     * We found a path, but didn't find any whitespace,
     * so it must be still incomplete - we should at
1922 1923
     * least see a \n - indicate that we want to carry
     * on trying again
1924
     */
1925
    return 1;
1926 1927
}

1928
static int
1929
qemudFindCharDevicePTYsMonitor(virDomainObjPtr vm,
1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944
                               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) {                                           \
1945 1946 1947
                if (chr->data.file.path == NULL) {                        \
                    /* neither the log output nor 'info chardev' had a */ \
                    /* pty path for this chardev, report an error */      \
1948 1949 1950
                    qemuReportError(VIR_ERR_INTERNAL_ERROR,               \
                                    _("no assigned pty for device %s"), id); \
                    return -1;                                            \
1951 1952 1953 1954 1955
                } else {                                                  \
                    /* 'info chardev' had no pty path for this chardev, */\
                    /* but the log output had, so we're fine */           \
                    continue;                                             \
                }                                                         \
1956 1957
            }                                                             \
                                                                          \
1958
            VIR_FREE(chr->data.file.path);                                \
1959
            chr->data.file.path = strdup(path);                           \
1960 1961
                                                                          \
            if (chr->data.file.path == NULL) {                            \
1962
                virReportOOMError();                                      \
1963 1964
                return -1;                                                \
            }                                                             \
1965 1966 1967 1968 1969 1970
        }                                                                 \
    }

    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");
1971
#undef LOOKUP_PTYS
1972 1973 1974 1975

    return 0;
}

1976
static int
1977
qemudFindCharDevicePTYs(virDomainObjPtr vm,
1978 1979
                        const char *output,
                        int fd ATTRIBUTE_UNUSED)
1980
{
1981
    size_t offset = 0;
1982
    int ret, i;
1983 1984

    /* The order in which QEMU prints out the PTY paths is
1985 1986
       the order in which it procsses its serial and parallel
       device args. This code must match that ordering.... */
1987

1988
    /* first comes the serial devices */
1989 1990
    for (i = 0 ; i < vm->def->nserials ; i++) {
        virDomainChrDefPtr chr = vm->def->serials[i];
1991
        if (chr->type == VIR_DOMAIN_CHR_TYPE_PTY) {
1992
            if ((ret = qemudExtractTTYPath(output, &offset,
1993
                                           &chr->data.file.path)) != 0)
1994
                return ret;
1995 1996 1997
        }
    }

1998
    /* then the parallel devices */
1999 2000
    for (i = 0 ; i < vm->def->nparallels ; i++) {
        virDomainChrDefPtr chr = vm->def->parallels[i];
2001
        if (chr->type == VIR_DOMAIN_CHR_TYPE_PTY) {
2002
            if ((ret = qemudExtractTTYPath(output, &offset,
2003
                                           &chr->data.file.path)) != 0)
2004
                return ret;
2005 2006 2007
        }
    }

2008 2009 2010 2011
    /* 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) {
2012
            if ((ret = qemudExtractTTYPath(output, &offset,
2013 2014 2015 2016 2017
                                           &chr->data.file.path)) != 0)
                return ret;
        }
    }

2018
    return 0;
2019 2020
}

2021 2022 2023 2024 2025
static void qemudFreePtyPath(void *payload, const char *name ATTRIBUTE_UNUSED)
{
    VIR_FREE(payload);
}

2026
static int
2027
qemudWaitForMonitor(struct qemud_driver* driver,
2028
                    virDomainObjPtr vm, off_t pos)
2029
{
2030
    char buf[4096]; /* Plenty of space to get startup greeting */
2031
    int logfd;
2032
    int ret = -1;
2033

2034
    if ((logfd = qemudLogReadFD(driver->logDir, vm->def->name, pos))
2035
        < 0)
2036
        return -1;
2037

2038
    ret = qemudReadLogOutput(vm, logfd, buf, sizeof(buf),
2039
                             qemudFindCharDevicePTYs,
2040
                             "console", 30);
2041
    if (close(logfd) < 0) {
2042
        char ebuf[4096];
2043
        VIR_WARN(_("Unable to close logfile: %s"),
2044 2045
                 virStrerror(errno, ebuf, sizeof ebuf));
    }
2046

2047
    if (ret < 0)
2048
        return -1;
2049

2050
    VIR_DEBUG("Connect monitor to %p '%s'", vm, vm->def->name);
2051
    if (qemuConnectMonitor(driver, vm) < 0)
2052 2053
        return -1;

2054 2055 2056 2057 2058 2059
    /* 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. */
    virHashTablePtr paths = virHashCreate(0);
    if (paths == NULL) {
2060
        virReportOOMError();
2061 2062 2063
        goto cleanup;
    }

2064
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
2065 2066
    qemuDomainObjPrivatePtr priv = vm->privateData;
    ret = qemuMonitorGetPtyPaths(priv->mon, paths);
2067
    qemuDomainObjExitMonitorWithDriver(driver, vm);
2068 2069 2070

    VIR_DEBUG("qemuMonitorGetPtyPaths returned %i", ret);
    if (ret == 0) {
2071
        ret = qemudFindCharDevicePTYsMonitor(vm, paths);
2072 2073 2074 2075 2076 2077 2078 2079
    }

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

    return ret;
2080 2081
}

2082
static int
2083
qemuDetectVcpuPIDs(struct qemud_driver *driver,
2084 2085 2086
                   virDomainObjPtr vm) {
    pid_t *cpupids = NULL;
    int ncpupids;
2087
    qemuDomainObjPrivatePtr priv = vm->privateData;
2088

2089
    if (vm->def->virtType != VIR_DOMAIN_VIRT_KVM) {
2090 2091
        priv->nvcpupids = 1;
        if (VIR_ALLOC_N(priv->vcpupids, priv->nvcpupids) < 0) {
2092
            virReportOOMError();
2093 2094
            return -1;
        }
2095
        priv->vcpupids[0] = vm->pid;
2096 2097 2098
        return 0;
    }

2099
    /* What follows is now all KVM specific */
2100

2101
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
2102
    if ((ncpupids = qemuMonitorGetCPUInfo(priv->mon, &cpupids)) < 0) {
2103
        qemuDomainObjExitMonitorWithDriver(driver, vm);
2104
        return -1;
2105
    }
2106
    qemuDomainObjExitMonitorWithDriver(driver, vm);
2107

2108 2109 2110
    /* Treat failure to get VCPU<->PID mapping as non-fatal */
    if (ncpupids == 0)
        return 0;
2111

2112
    if (ncpupids != vm->def->vcpus) {
2113 2114 2115
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("got wrong number of vCPU pids from QEMU monitor. got %d, wanted %d"),
                        ncpupids, (int)vm->def->vcpus);
2116 2117 2118
        VIR_FREE(cpupids);
        return -1;
    }
2119

2120 2121
    priv->nvcpupids = ncpupids;
    priv->vcpupids = cpupids;
2122 2123 2124
    return 0;
}

2125 2126 2127
/*
 * To be run between fork/exec of QEMU only
 */
2128
static int
2129 2130
qemudInitCpuAffinity(virDomainObjPtr vm)
{
2131
    int i, hostcpus, maxcpu = QEMUD_CPUMASK_LEN;
2132
    virNodeInfo nodeinfo;
2133 2134
    unsigned char *cpumap;
    int cpumaplen;
2135 2136

    DEBUG0("Setting CPU affinity");
2137

2138
    if (nodeGetInfo(NULL, &nodeinfo) < 0)
2139 2140 2141 2142
        return -1;

    /* setaffinity fails if you set bits for CPUs which
     * aren't present, so we have to limit ourselves */
2143 2144 2145
    hostcpus = VIR_NODEINFO_MAXCPUS(nodeinfo);
    if (maxcpu > hostcpus)
        maxcpu = hostcpus;
2146

2147 2148
    cpumaplen = VIR_CPU_MAPLEN(maxcpu);
    if (VIR_ALLOC_N(cpumap, cpumaplen) < 0) {
2149
        virReportOOMError();
2150 2151 2152
        return -1;
    }

D
Daniel P. Berrange 已提交
2153
    if (vm->def->cpumask) {
2154 2155 2156
        /* 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
2157
            if (vm->def->cpumask[i])
2158
                VIR_USE_CPU(cpumap, i);
D
Daniel P. Berrange 已提交
2159
    } else {
2160 2161 2162 2163
        /* 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
2164
        for (i = 0 ; i < maxcpu ; i++)
2165
            VIR_USE_CPU(cpumap, i);
D
Daniel P. Berrange 已提交
2166
    }
2167

2168 2169 2170 2171 2172 2173 2174 2175
    /* 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;
2176
    }
2177
    VIR_FREE(cpumap);
2178 2179 2180 2181 2182

    return 0;
}


2183
static int
2184 2185 2186 2187
qemuInitPasswords(virConnectPtr conn,
                  struct qemud_driver *driver,
                  virDomainObjPtr vm,
                  unsigned long long qemuCmdFlags) {
2188
    int ret = 0;
2189
    qemuDomainObjPrivatePtr priv = vm->privateData;
2190

2191 2192 2193
    if ((vm->def->ngraphics == 1) &&
        vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
        (vm->def->graphics[0]->data.vnc.passwd || driver->vncPassword)) {
2194

2195
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
2196
        ret = qemuMonitorSetVNCPassword(priv->mon,
2197 2198 2199
                                        vm->def->graphics[0]->data.vnc.passwd ?
                                        vm->def->graphics[0]->data.vnc.passwd :
                                        driver->vncPassword);
2200
        qemuDomainObjExitMonitorWithDriver(driver, vm);
2201 2202
    }

2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225
    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);
2226
            VIR_FREE(secret);
2227 2228 2229 2230 2231 2232 2233
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            if (ret < 0)
                goto cleanup;
        }
    }

cleanup:
2234
    return ret;
2235 2236 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 2262 2263 2264 2265 2266 2267 2268 2269 2270 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
#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
2467
qemuDetectPCIAddresses(virDomainObjPtr vm,
2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484
                       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) {
2485 2486 2487
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for VirtIO disk %s"),
                            vm->def->disks[i]->dst);
2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498
            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) {
2499 2500 2501
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for %s NIC"),
                            vm->def->nets[i]->model);
2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512
            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) {
2513 2514 2515
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for controller %s"),
                            virDomainControllerTypeToString(vm->def->controllers[i]->type));
2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526
            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) {
2527 2528 2529
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for video adapter %s"),
                            virDomainVideoTypeToString(vm->def->videos[i]->type));
2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540
            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) {
2541 2542 2543
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for sound adapter %s"),
                            virDomainSoundModelTypeToString(vm->def->sounds[i]->model));
2544 2545 2546 2547 2548 2549 2550 2551 2552 2553
            return -1;
        }
    }


    if (vm->def->watchdog &&
        qemuGetPCIWatchdogVendorProduct(vm->def->watchdog, &vendor, &product) == 0) {
        if (qemuAssignNextPCIAddress(&(vm->def->watchdog->info),
                                     vendor, product,
                                     addrs,  naddrs) < 0) {
2554 2555 2556
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for watchdog %s"),
                            virDomainWatchdogModelTypeToString(vm->def->watchdog->model));
2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588
            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);

2589
    ret = qemuDetectPCIAddresses(vm, addrs, naddrs);
2590 2591 2592 2593 2594 2595

    VIR_FREE(addrs);

    return ret;
}

2596
static int qemudNextFreeVNCPort(struct qemud_driver *driver ATTRIBUTE_UNUSED) {
2597 2598
    int i;

2599
    for (i = 5900 ; i < 65535 ; i++) {
2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631
        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;
}

2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669

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;
}


2670
static pciDeviceList *
2671
qemuGetPciHostDeviceList(virDomainDefPtr def)
2672 2673
{
    pciDeviceList *list;
2674 2675
    int i;

2676
    if (!(list = pciDeviceListNew()))
2677
        return NULL;
2678 2679 2680 2681 2682 2683 2684 2685 2686 2687

    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;

2688
        dev = pciGetDevice(hostdev->source.subsys.u.pci.domain,
2689 2690 2691
                           hostdev->source.subsys.u.pci.bus,
                           hostdev->source.subsys.u.pci.slot,
                           hostdev->source.subsys.u.pci.function);
2692
        if (!dev) {
2693
            pciDeviceListFree(list);
2694 2695
            return NULL;
        }
2696

2697 2698 2699
        if (pciDeviceListAdd(list, dev) < 0) {
            pciFreeDevice(dev);
            pciDeviceListFree(list);
2700
            return NULL;
2701 2702
        }

2703
        pciDeviceSetManaged(dev, hostdev->managed);
2704 2705
    }

2706 2707 2708 2709
    return list;
}

static int
2710 2711 2712 2713
qemuUpdateActivePciHostdevs(struct qemud_driver *driver,
                            virDomainDefPtr def)
{
    pciDeviceList *pcidevs;
2714
    int i;
2715
    int ret = -1;
2716 2717 2718 2719

    if (!def->nhostdevs)
        return 0;

2720
    if (!(pcidevs = qemuGetPciHostDeviceList(def)))
2721 2722
        return -1;

2723 2724
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
2725 2726 2727
        pciDeviceListSteal(pcidevs, dev);
        if (pciDeviceListAdd(driver->activePciHostdevs, dev) < 0) {
            pciFreeDevice(dev);
2728
            goto cleanup;
2729 2730 2731
        }
    }

2732 2733 2734
    ret = 0;

cleanup:
2735
    pciDeviceListFree(pcidevs);
2736 2737 2738
    return ret;
}

2739

2740
static int
2741 2742
qemuPrepareHostPCIDevices(struct qemud_driver *driver,
                          virDomainDefPtr def)
2743 2744 2745
{
    pciDeviceList *pcidevs;
    int i;
2746
    int ret = -1;
2747

2748
    if (!(pcidevs = qemuGetPciHostDeviceList(def)))
2749 2750
        return -1;

2751
    /* We have to use 3 loops here. *All* devices must
2752 2753
     * be detached before we reset any of them, because
     * in some cases you have to reset the whole PCI,
2754 2755
     * which impacts all devices on it. Also, all devices
     * must be reset before being marked as active.
2756 2757 2758 2759 2760 2761 2762
     */

    /* 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
     */

2763 2764
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
2765
        if (!pciDeviceIsAssignable(dev, !driver->relaxedACS))
2766 2767
            goto cleanup;

2768
        if (pciDeviceGetManaged(dev) &&
2769
            pciDettachDevice(dev) < 0)
2770 2771
            goto cleanup;
    }
2772 2773 2774

    /* Now that all the PCI hostdevs have be dettached, we can safely
     * reset them */
2775 2776
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
2777
        if (pciResetDevice(dev, driver->activePciHostdevs) < 0)
2778 2779
            goto cleanup;
    }
2780

2781
    /* Now mark all the devices as active */
2782 2783
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
2784 2785 2786
        pciDeviceListSteal(pcidevs, dev);
        if (pciDeviceListAdd(driver->activePciHostdevs, dev) < 0) {
            pciFreeDevice(dev);
2787 2788
            goto cleanup;
        }
2789 2790
    }

2791
    ret = 0;
2792

2793
cleanup:
2794
    pciDeviceListFree(pcidevs);
2795
    return ret;
2796 2797
}

2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808

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;
2809
        if (hostdev->source.subsys.type != VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB)
2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847
            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;
}


2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858
static void
qemudReattachManagedDevice(pciDevice *dev)
{
    int retries = 100;

    if (pciDeviceGetManaged(dev)) {
        while (pciWaitForDeviceCleanup(dev, "kvm_assigned_device")
               && retries) {
            usleep(100*1000);
            retries--;
        }
2859
        if (pciReAttachDevice(dev) < 0) {
2860 2861 2862 2863 2864 2865 2866 2867
            virErrorPtr err = virGetLastError();
            VIR_ERROR(_("Failed to re-attach PCI device: %s"),
                      err ? err->message : "");
            virResetError(err);
        }
    }
}

2868
static void
2869
qemuDomainReAttachHostDevices(struct qemud_driver *driver,
2870
                              virDomainDefPtr def)
2871
{
2872
    pciDeviceList *pcidevs;
2873 2874
    int i;

2875 2876
    if (!def->nhostdevs)
        return;
2877

2878
    if (!(pcidevs = qemuGetPciHostDeviceList(def))) {
2879
        virErrorPtr err = virGetLastError();
2880
        VIR_ERROR(_("Failed to allocate pciDeviceList: %s"),
2881 2882 2883
                  err ? err->message : "");
        virResetError(err);
        return;
2884 2885
    }

2886 2887
    /* Again 3 loops; mark all devices as inactive before reset
     * them and reset all the devices before re-attach */
2888

2889 2890
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
2891
        pciDeviceListDel(driver->activePciHostdevs, dev);
2892
    }
2893

2894 2895
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
2896
        if (pciResetDevice(dev, driver->activePciHostdevs) < 0) {
2897
            virErrorPtr err = virGetLastError();
2898
            VIR_ERROR(_("Failed to reset PCI device: %s"),
2899 2900 2901
                      err ? err->message : "");
            virResetError(err);
        }
2902
    }
2903

2904 2905
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
2906
        qemudReattachManagedDevice(dev);
2907
    }
2908

2909
    pciDeviceListFree(pcidevs);
2910 2911
}

2912 2913 2914 2915 2916 2917 2918 2919 2920 2921
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

2922
static int qemuSetupCgroup(struct qemud_driver *driver,
2923 2924 2925 2926
                           virDomainObjPtr vm)
{
    virCgroupPtr cgroup = NULL;
    int rc;
2927
    unsigned int i;
2928 2929 2930 2931
    const char *const *deviceACL =
        driver->cgroupDeviceACL ?
        (const char *const *)driver->cgroupDeviceACL :
        defaultDeviceACL;
2932 2933 2934 2935 2936 2937

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

    rc = virCgroupForDomain(driver->cgroup, vm->def->name, &cgroup, 1);
    if (rc != 0) {
2938
        virReportSystemError(-rc,
2939 2940 2941 2942 2943
                             _("Unable to create cgroup for %s"),
                             vm->def->name);
        goto cleanup;
    }

2944 2945
    if (qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_DEVICES)) {
        rc = virCgroupDenyAllDevices(cgroup);
2946
        if (rc != 0) {
2947 2948 2949 2950 2951
            if (rc == -EPERM) {
                VIR_WARN0("Group devices ACL is not accessible, disabling whitelisting");
                goto done;
            }

2952
            virReportSystemError(-rc,
2953
                                 _("Unable to deny all devices for %s"), vm->def->name);
2954 2955 2956
            goto cleanup;
        }

2957 2958 2959 2960 2961 2962 2963 2964
        for (i = 0; i < vm->def->ndisks ; i++) {
            if (vm->def->disks[i]->type != VIR_DOMAIN_DISK_TYPE_BLOCK ||
                vm->def->disks[i]->src == NULL)
                continue;

            rc = virCgroupAllowDevicePath(cgroup,
                                          vm->def->disks[i]->src);
            if (rc != 0) {
2965
                virReportSystemError(-rc,
2966 2967 2968 2969 2970
                                     _("Unable to allow device %s for %s"),
                                     vm->def->disks[i]->src, vm->def->name);
                goto cleanup;
            }
        }
2971

2972
        rc = virCgroupAllowDeviceMajor(cgroup, 'c', DEVICE_PTY_MAJOR);
2973
        if (rc != 0) {
2974
            virReportSystemError(-rc, "%s",
2975
                                 _("unable to allow /dev/pts/ devices"));
2976 2977 2978
            goto cleanup;
        }

2979 2980 2981
        if (vm->def->nsounds) {
            rc = virCgroupAllowDeviceMajor(cgroup, 'c', DEVICE_SND_MAJOR);
            if (rc != 0) {
2982
                virReportSystemError(-rc, "%s",
2983 2984 2985 2986 2987 2988 2989 2990 2991 2992
                                     _("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) {
2993
                virReportSystemError(-rc,
2994 2995 2996 2997
                                     _("unable to allow device %s"),
                                     deviceACL[i]);
                goto cleanup;
            }
2998 2999 3000 3001
        }
    }

done:
3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013
    virCgroupFree(&cgroup);
    return 0;

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


3014
static int qemuRemoveCgroup(struct qemud_driver *driver,
3015 3016
                            virDomainObjPtr vm,
                            int quiet)
3017 3018 3019 3020 3021 3022 3023 3024 3025
{
    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) {
3026
        if (!quiet)
3027 3028 3029
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("Unable to find cgroup for %s\n"),
                            vm->def->name);
3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049
        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) {
3050
        virReportSystemError(-rc,
3051 3052 3053 3054 3055 3056 3057
                             _("unable to find cgroup for domain %s"),
                             def->name);
        goto cleanup;
    }

    rc = virCgroupAddTask(cgroup, getpid());
    if (rc != 0) {
3058
        virReportSystemError(-rc,
3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075
                             _("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;
3076 3077 3078
};

static int qemudSecurityHook(void *data) {
3079 3080
    struct qemudHookData *h = data;

3081 3082 3083
    /* This must take place before exec(), so that all QEMU
     * memory allocation is on the correct NUMA node
     */
3084
    if (qemuAddToCgroup(h->driver, h->vm->def) < 0)
3085 3086
        return -1;

3087 3088 3089
    /* This must be done after cgroup placement to avoid resetting CPU
     * affinity */
    if (qemudInitCpuAffinity(h->vm) < 0)
3090
        return -1;
3091

3092 3093
    if (h->driver->securityDriver &&
        h->driver->securityDriver->domainSetSecurityProcessLabel &&
3094
        h->driver->securityDriver->domainSetSecurityProcessLabel(h->driver->securityDriver, h->vm) < 0)
3095 3096 3097
        return -1;

    return 0;
3098 3099
}

3100
static int
3101
qemuPrepareMonitorChr(struct qemud_driver *driver,
3102
                      virDomainChrDefPtr monConfig,
3103 3104
                      const char *vm)
{
3105
    monConfig->targetType = VIR_DOMAIN_CHR_TARGET_TYPE_MONITOR;
3106

3107 3108
    monConfig->type = VIR_DOMAIN_CHR_TYPE_UNIX;
    monConfig->data.nix.listen = 1;
3109

D
Daniel P. Berrange 已提交
3110
    if (!(monConfig->info.alias = strdup("monitor"))) {
3111
        virReportOOMError();
D
Daniel P. Berrange 已提交
3112 3113 3114
        return -1;
    }

3115
    if (virAsprintf(&monConfig->data.nix.path, "%s/%s.monitor",
3116
                    driver->libDir, vm) < 0) {
3117
        virReportOOMError();
3118 3119 3120 3121 3122 3123
        return -1;
    }

    return 0;
}

C
Chris Lalancette 已提交
3124 3125 3126 3127 3128
static int qemuDomainSnapshotSetActive(virDomainObjPtr vm,
                                       char *snapshotDir);
static int qemuDomainSnapshotSetInactive(virDomainObjPtr vm,
                                         char *snapshotDir);

3129 3130
static int qemudStartVMDaemon(virConnectPtr conn,
                              struct qemud_driver *driver,
3131
                              virDomainObjPtr vm,
3132 3133
                              const char *migrateFrom,
                              int stdin_fd) {
3134
    const char **argv = NULL, **tmp;
3135
    const char **progenv = NULL;
3136
    int i, ret;
3137
    struct stat sb;
3138 3139
    int *tapfds = NULL;
    int ntapfds = 0;
3140
    unsigned long long qemuCmdFlags;
3141
    fd_set keepfd;
3142
    const char *emulator;
G
Guido Günther 已提交
3143
    pid_t child;
3144
    int pos = -1;
3145
    char ebuf[1024];
3146
    char *pidfile = NULL;
3147
    int logfile = -1;
3148
    qemuDomainObjPrivatePtr priv = vm->privateData;
3149

3150
    struct qemudHookData hookData;
3151 3152 3153 3154
    hookData.conn = conn;
    hookData.vm = vm;
    hookData.driver = driver;

3155
    FD_ZERO(&keepfd);
3156

3157 3158
    DEBUG0("Beginning VM startup process");

D
Daniel P. Berrange 已提交
3159
    if (virDomainObjIsActive(vm)) {
3160 3161
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("VM is already active"));
3162 3163 3164
        return -1;
    }

3165 3166 3167 3168 3169
    /* Must be run before security labelling */
    DEBUG0("Preparing host devices");
    if (qemuPrepareHostDevices(driver, vm->def) < 0)
        goto cleanup;

3170 3171
    /* If you are using a SecurityDriver with dynamic labelling,
       then generate a security label for isolation */
3172
    DEBUG0("Generating domain security label (if required)");
3173
    if (driver->securityDriver &&
3174
        driver->securityDriver->domainGenSecurityLabel &&
3175
        driver->securityDriver->domainGenSecurityLabel(vm) < 0)
3176 3177
        return -1;

3178
    DEBUG0("Generating setting domain security labels (if required)");
3179 3180
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityAllLabel &&
3181
        driver->securityDriver->domainSetSecurityAllLabel(vm) < 0)
3182 3183
        goto cleanup;

3184 3185 3186
    /* Ensure no historical cgroup for this VM is lying around bogus
     * settings */
    DEBUG0("Ensuring no historical cgroup is lying around");
3187
    qemuRemoveCgroup(driver, vm, 1);
3188

3189 3190 3191
    if ((vm->def->ngraphics == 1) &&
        vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
        vm->def->graphics[0]->data.vnc.autoport) {
3192
        DEBUG0("Determining VNC port");
3193
        int port = qemudNextFreeVNCPort(driver);
3194
        if (port < 0) {
3195 3196
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            "%s", _("Unable to find an unused VNC port"));
3197
            goto cleanup;
3198
        }
3199
        vm->def->graphics[0]->data.vnc.port = port;
3200
    }
3201

L
Laine Stump 已提交
3202
    if (virFileMakePath(driver->logDir) != 0) {
3203
        virReportSystemError(errno,
3204 3205
                             _("cannot create log directory %s"),
                             driver->logDir);
3206
        goto cleanup;
3207 3208
    }

3209
    DEBUG0("Creating domain log file");
3210
    if ((logfile = qemudLogFD(driver, vm->def->name)) < 0)
3211
        goto cleanup;
3212

3213 3214
    emulator = vm->def->emulator;

3215 3216 3217 3218
    /* 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
     */
3219
    if (stat(emulator, &sb) < 0) {
3220
        virReportSystemError(errno,
3221 3222
                             _("Cannot find QEMU binary %s"),
                             emulator);
3223
        goto cleanup;
3224 3225
    }

3226
    DEBUG0("Determing emulator version");
3227
    if (qemudExtractVersionInfo(emulator,
3228
                                NULL,
3229
                                &qemuCmdFlags) < 0)
3230
        goto cleanup;
3231

3232
    DEBUG0("Setting up domain cgroup (if required)");
3233
    if (qemuSetupCgroup(driver, vm) < 0)
3234 3235
        goto cleanup;

3236
    if (VIR_ALLOC(priv->monConfig) < 0) {
3237
        virReportOOMError();
3238 3239 3240
        goto cleanup;
    }

3241
    DEBUG0("Preparing monitor state");
3242
    if (qemuPrepareMonitorChr(driver, priv->monConfig, vm->def->name) < 0)
3243
        goto cleanup;
3244

D
Daniel P. Berrange 已提交
3245 3246 3247
#if HAVE_YAJL
    if (qemuCmdFlags & QEMUD_CMD_FLAG_MONITOR_JSON)
        priv->monJSON = 1;
3248
    else
D
Daniel P. Berrange 已提交
3249
#endif
3250
        priv->monJSON = 0;
D
Daniel P. Berrange 已提交
3251

D
Daniel P. Berrange 已提交
3252
    if ((ret = virFileDeletePid(driver->stateDir, vm->def->name)) != 0) {
3253
        virReportSystemError(ret,
D
Daniel P. Berrange 已提交
3254 3255 3256 3257 3258
                             _("Cannot remove stale PID file for %s"),
                             vm->def->name);
        goto cleanup;
    }

3259
    if (!(pidfile = virFilePid(driver->stateDir, vm->def->name))) {
3260
        virReportSystemError(errno,
3261 3262 3263
                             "%s", _("Failed to build pidfile path."));
        goto cleanup;
    }
D
Daniel P. Berrange 已提交
3264

3265
    /*
M
Matthew Booth 已提交
3266
     * Normally PCI addresses are assigned in the virDomainCreate
3267 3268 3269 3270 3271
     * 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
     */
3272
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
3273
        DEBUG0("Assigning domain PCI addresses");
3274
        /* Populate cache with current addresses */
3275 3276 3277 3278 3279 3280 3281
        if (priv->pciaddrs) {
            qemuDomainPCIAddressSetFree(priv->pciaddrs);
            priv->pciaddrs = NULL;
        }
        if (!(priv->pciaddrs = qemuDomainPCIAddressSetCreate(vm->def)))
            goto cleanup;

3282 3283

        /* Assign any remaining addresses */
3284 3285
        if (qemuAssignDevicePCISlots(vm->def, priv->pciaddrs) < 0)
            goto cleanup;
3286 3287 3288 3289

        priv->persistentAddrs = 1;
    } else {
        priv->persistentAddrs = 0;
3290 3291
    }

3292
    DEBUG0("Building emulator command line");
3293
    vm->def->id = driver->nextvmid++;
3294
    if (qemudBuildCommandLine(conn, driver, vm->def, priv->monConfig,
D
Daniel P. Berrange 已提交
3295
                              priv->monJSON, qemuCmdFlags, &argv, &progenv,
C
Chris Lalancette 已提交
3296 3297 3298 3299 3300
                              &tapfds, &ntapfds, migrateFrom,
                              vm->current_snapshot) < 0)
        goto cleanup;

    if (qemuDomainSnapshotSetInactive(vm, driver->snapshotDir) < 0)
3301
        goto cleanup;
3302

3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318
    /* 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;
    }

3319 3320
    tmp = progenv;
    while (*tmp) {
3321
        if (safewrite(logfile, *tmp, strlen(*tmp)) < 0)
3322
            VIR_WARN(_("Unable to write envv to logfile: %s"),
3323
                     virStrerror(errno, ebuf, sizeof ebuf));
3324
        if (safewrite(logfile, " ", 1) < 0)
3325
            VIR_WARN(_("Unable to write envv to logfile: %s"),
3326
                     virStrerror(errno, ebuf, sizeof ebuf));
3327 3328
        tmp++;
    }
3329 3330
    tmp = argv;
    while (*tmp) {
3331
        if (safewrite(logfile, *tmp, strlen(*tmp)) < 0)
3332
            VIR_WARN(_("Unable to write argv to logfile: %s"),
3333
                     virStrerror(errno, ebuf, sizeof ebuf));
3334
        if (safewrite(logfile, " ", 1) < 0)
3335
            VIR_WARN(_("Unable to write argv to logfile: %s"),
3336
                     virStrerror(errno, ebuf, sizeof ebuf));
3337 3338
        tmp++;
    }
3339
    if (safewrite(logfile, "\n", 1) < 0)
3340
        VIR_WARN(_("Unable to write argv to logfile: %s"),
3341
                 virStrerror(errno, ebuf, sizeof ebuf));
3342

3343
    if ((pos = lseek(logfile, 0, SEEK_END)) < 0)
3344
        VIR_WARN(_("Unable to seek to end of logfile: %s"),
3345
                 virStrerror(errno, ebuf, sizeof ebuf));
3346

3347 3348 3349
    for (i = 0 ; i < ntapfds ; i++)
        FD_SET(tapfds[i], &keepfd);

3350
    ret = virExecDaemonize(argv, progenv, &keepfd, &child,
3351
                           stdin_fd, &logfile, &logfile,
3352
                           VIR_EXEC_NONBLOCK | VIR_EXEC_CLEAR_CAPS,
3353 3354 3355
                           qemudSecurityHook, &hookData,
                           pidfile);
    VIR_FREE(pidfile);
G
Guido Günther 已提交
3356 3357 3358

    /* wait for qemu process to to show up */
    if (ret == 0) {
3359
        if (virFileReadPid(driver->stateDir, vm->def->name, &vm->pid)) {
3360 3361
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("Domain %s didn't show up\n"), vm->def->name);
3362
            ret = -1;
G
Guido Günther 已提交
3363
        }
3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374
    } 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;
    }
3375 3376

    vm->state = migrateFrom ? VIR_DOMAIN_PAUSED : VIR_DOMAIN_RUNNING;
3377

3378
    for (i = 0 ; argv[i] ; i++)
3379 3380
        VIR_FREE(argv[i]);
    VIR_FREE(argv);
3381

3382 3383 3384 3385
    for (i = 0 ; progenv[i] ; i++)
        VIR_FREE(progenv[i]);
    VIR_FREE(progenv);

3386 3387 3388
    if (ret == -1) /* The VM failed to start; tear filters before taps */
        virNWFilterTearVMNWFilters(vm);

3389 3390 3391
    if (tapfds) {
        for (i = 0 ; i < ntapfds ; i++) {
            close(tapfds[i]);
3392
        }
3393
        VIR_FREE(tapfds);
3394 3395
    }

3396
    if (ret == -1) /* The VM failed to start */
3397 3398
        goto cleanup;

3399
    DEBUG0("Waiting for monitor to show up");
3400
    if (qemudWaitForMonitor(driver, vm, pos) < 0)
3401 3402
        goto abort;

3403
    DEBUG0("Detecting VCPU PIDs");
3404
    if (qemuDetectVcpuPIDs(driver, vm) < 0)
3405 3406
        goto abort;

3407
    DEBUG0("Setting any required VM passwords");
3408
    if (qemuInitPasswords(conn, driver, vm, qemuCmdFlags) < 0)
3409 3410
        goto abort;

D
Daniel P. Berrange 已提交
3411 3412 3413
    /* 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)) {
3414
        DEBUG0("Determining domain device PCI addresses");
D
Daniel P. Berrange 已提交
3415 3416 3417
        if (qemuInitPCIAddresses(driver, vm) < 0)
            goto abort;
    }
3418

3419
    DEBUG0("Setting initial memory amount");
3420
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
3421
    if (qemuMonitorSetBalloon(priv->mon, vm->def->memory) < 0) {
3422
        qemuDomainObjExitMonitorWithDriver(driver, vm);
3423
        goto abort;
3424
    }
3425 3426

    if (migrateFrom == NULL) {
3427
        DEBUG0("Starting domain CPUs");
3428 3429 3430
        /* Allow the CPUS to start executing */
        if (qemuMonitorStartCPUs(priv->mon, conn) < 0) {
            if (virGetLastError() == NULL)
3431 3432
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                "%s", _("resume operation failed"));
3433 3434 3435 3436
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            goto abort;
        }
    }
3437
    qemuDomainObjExitMonitorWithDriver(driver, vm);
3438

3439

3440
    DEBUG0("Writing domain status to disk");
3441
    if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
3442
        goto abort;
3443

3444 3445 3446
    if (logfile != -1)
        close(logfile);

3447
    return 0;
3448 3449

cleanup:
3450 3451 3452
    /* 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 */
3453
    qemuDomainReAttachHostDevices(driver, vm->def);
3454

3455 3456
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityAllLabel)
3457
        driver->securityDriver->domainRestoreSecurityAllLabel(vm);
3458 3459
    if (driver->securityDriver &&
        driver->securityDriver->domainReleaseSecurityLabel)
3460
        driver->securityDriver->domainReleaseSecurityLabel(vm);
3461
    qemuRemoveCgroup(driver, vm, 1);
3462 3463 3464 3465
    if ((vm->def->ngraphics == 1) &&
        vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
        vm->def->graphics[0]->data.vnc.autoport)
        vm->def->graphics[0]->data.vnc.port = -1;
3466 3467
    if (logfile != -1)
        close(logfile);
3468 3469
    vm->def->id = -1;
    return -1;
3470 3471 3472 3473

abort:
    /* We jump here if we failed to initialize the now running VM
     * killing it off and pretend we never started it */
3474
    qemudShutdownVMDaemon(driver, vm);
3475 3476 3477 3478 3479

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

    return -1;
3480 3481 3482
}


3483
static void qemudShutdownVMDaemon(struct qemud_driver *driver,
3484
                                  virDomainObjPtr vm) {
D
Daniel P. Berrange 已提交
3485
    int ret;
3486
    int retries = 0;
3487
    qemuDomainObjPrivatePtr priv = vm->privateData;
3488
    virErrorPtr orig_err;
3489 3490
    virDomainDefPtr def;
    int i;
D
Daniel P. Berrange 已提交
3491

D
Daniel P. Berrange 已提交
3492
    if (!virDomainObjIsActive(vm))
3493
        return;
3494

3495
    VIR_DEBUG("Shutting down VM '%s'", vm->def->name);
3496

3497 3498 3499 3500
    /* This method is routinely used in clean up paths. Disable error
     * reporting so we don't squash a legit error. */
    orig_err = virSaveLastError();

3501
    virNWFilterTearVMNWFilters(vm);
S
Stefan Berger 已提交
3502

3503
    if (driver->macFilter) {
3504
        def = vm->def;
3505 3506 3507 3508
        for (i = 0 ; i < def->nnets ; i++) {
            virDomainNetDefPtr net = def->nets[i];
            if (net->ifname == NULL)
                continue;
3509
            if ((errno = networkDisallowMacOnPort(driver, net->ifname,
3510
                                                  net->mac))) {
3511
                virReportSystemError(errno,
3512 3513 3514 3515 3516 3517
             _("failed to remove ebtables rule to allow MAC address on  '%s'"),
                                     net->ifname);
            }
        }
    }

G
Guido Günther 已提交
3518 3519
    if (virKillProcess(vm->pid, 0) == 0 &&
        virKillProcess(vm->pid, SIGTERM) < 0)
3520
        virReportSystemError(errno,
3521 3522
                             _("Failed to send SIGTERM to %s (%d)"),
                             vm->def->name, vm->pid);
3523

3524 3525 3526
    if (priv->mon &&
        qemuMonitorClose(priv->mon) == 0) {
        virDomainObjUnref(vm);
3527
        priv->mon = NULL;
3528
    }
3529

3530 3531 3532 3533 3534
    if (priv->monConfig) {
        if (priv->monConfig->type == VIR_DOMAIN_CHR_TYPE_UNIX)
            unlink(priv->monConfig->data.nix.path);
        virDomainChrDefFree(priv->monConfig);
        priv->monConfig = NULL;
3535 3536
    }

G
Guido Günther 已提交
3537 3538
    /* shut it off for sure */
    virKillProcess(vm->pid, SIGKILL);
3539

3540 3541 3542 3543 3544 3545 3546 3547 3548 3549
    /* 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);
    }

3550
    /* Reset Security Labels */
3551
    if (driver->securityDriver &&
3552
        driver->securityDriver->domainRestoreSecurityAllLabel)
3553
        driver->securityDriver->domainRestoreSecurityAllLabel(vm);
3554 3555
    if (driver->securityDriver &&
        driver->securityDriver->domainReleaseSecurityLabel)
3556
        driver->securityDriver->domainReleaseSecurityLabel(vm);
3557

3558 3559 3560 3561 3562 3563 3564
    /* 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
3565
    virDomainDefClearDeviceAliases(vm->def);
3566 3567 3568 3569 3570
    if (!priv->persistentAddrs) {
        virDomainDefClearPCIAddresses(vm->def);
        qemuDomainPCIAddressSetFree(priv->pciaddrs);
        priv->pciaddrs = NULL;
    }
3571

3572
    qemuDomainReAttachHostDevices(driver, vm->def);
3573

S
Stefan Berger 已提交
3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584
#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

3585
retry:
3586
    if ((ret = qemuRemoveCgroup(driver, vm, 0)) < 0) {
3587 3588 3589 3590 3591 3592 3593 3594
        if (ret == -EBUSY && (retries++ < 5)) {
            usleep(200*1000);
            goto retry;
        }
        VIR_WARN("Failed to remove cgroup for %s",
                 vm->def->name);
    }

3595
    qemudRemoveDomainStatus(driver, vm);
D
Daniel P. Berrange 已提交
3596

3597
    vm->pid = -1;
3598
    vm->def->id = -1;
3599
    vm->state = VIR_DOMAIN_SHUTOFF;
3600 3601
    VIR_FREE(priv->vcpupids);
    priv->nvcpupids = 0;
3602 3603

    if (vm->newDef) {
3604
        virDomainDefFree(vm->def);
3605
        vm->def = vm->newDef;
3606
        vm->def->id = -1;
3607 3608
        vm->newDef = NULL;
    }
3609 3610 3611 3612 3613

    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
3614 3615 3616
}


3617
static virDrvOpenStatus qemudOpen(virConnectPtr conn,
3618
                                  virConnectAuthPtr auth ATTRIBUTE_UNUSED,
3619
                                  int flags ATTRIBUTE_UNUSED) {
3620
    if (conn->uri == NULL) {
3621 3622 3623
        if (qemu_driver == NULL)
            return VIR_DRV_OPEN_DECLINED;

3624
        conn->uri = xmlParseURI(qemu_driver->privileged ?
3625 3626
                                "qemu:///system" :
                                "qemu:///session");
3627
        if (!conn->uri) {
3628
            virReportOOMError();
3629 3630
            return VIR_DRV_OPEN_ERROR;
        }
3631 3632 3633 3634 3635 3636 3637 3638 3639 3640
    } 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;

3641
        if (qemu_driver == NULL) {
3642 3643
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("qemu state driver is not active"));
3644 3645 3646
            return VIR_DRV_OPEN_ERROR;
        }

3647
        if (conn->uri->path == NULL) {
3648 3649 3650 3651 3652
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("no QEMU URI path given, try %s"),
                            qemu_driver->privileged
                            ? "qemu:///system"
                            : "qemu:///session");
3653 3654 3655
                return VIR_DRV_OPEN_ERROR;
        }

3656
        if (qemu_driver->privileged) {
3657 3658
            if (STRNEQ (conn->uri->path, "/system") &&
                STRNEQ (conn->uri->path, "/session")) {
3659 3660 3661
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("unexpected QEMU URI path '%s', try qemu:///system"),
                                conn->uri->path);
3662 3663 3664 3665
                return VIR_DRV_OPEN_ERROR;
            }
        } else {
            if (STRNEQ (conn->uri->path, "/session")) {
3666 3667 3668
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("unexpected QEMU URI path '%s', try qemu:///session"),
                                conn->uri->path);
3669 3670 3671
                return VIR_DRV_OPEN_ERROR;
            }
        }
3672 3673 3674 3675 3676 3677 3678
    }
    conn->privateData = qemu_driver;

    return VIR_DRV_OPEN_SUCCESS;
}

static int qemudClose(virConnectPtr conn) {
3679
    struct qemud_driver *driver = conn->privateData;
3680 3681

    /* Get rid of callbacks registered for this conn */
3682
    qemuDriverLock(driver);
3683
    virDomainEventCallbackListRemoveConn(conn, driver->domainEventCallbacks);
3684
    qemuDriverUnlock(driver);
3685 3686 3687 3688 3689 3690

    conn->privateData = NULL;

    return 0;
}

D
Daniel Veillard 已提交
3691 3692 3693 3694 3695
/* Which features are supported by this driver? */
static int
qemudSupportsFeature (virConnectPtr conn ATTRIBUTE_UNUSED, int feature)
{
    switch (feature) {
3696 3697 3698 3699 3700
    case VIR_DRV_FEATURE_MIGRATION_V2:
    case VIR_DRV_FEATURE_MIGRATION_P2P:
        return 1;
    default:
        return 0;
D
Daniel Veillard 已提交
3701 3702 3703
    }
}

3704
static const char *qemudGetType(virConnectPtr conn ATTRIBUTE_UNUSED) {
3705
    return "QEMU";
3706 3707
}

3708

3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721
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;
}


3722 3723 3724 3725
static int kvmGetMaxVCPUs(void) {
    int maxvcpus = 1;

    int r, fd;
3726

3727 3728
    fd = open(KVM_DEVICE, O_RDONLY);
    if (fd < 0) {
3729
        virReportSystemError(errno, _("Unable to open %s"), KVM_DEVICE);
3730
        return -1;
3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741
    }

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

    close(fd);
    return maxvcpus;
}


3742
static int qemudGetMaxVCPUs(virConnectPtr conn ATTRIBUTE_UNUSED, const char *type) {
3743 3744 3745
    if (!type)
        return 16;

3746
    if (STRCASEEQ(type, "qemu"))
3747 3748
        return 16;

3749
    if (STRCASEEQ(type, "kvm"))
3750
        return kvmGetMaxVCPUs();
3751

3752
    if (STRCASEEQ(type, "kqemu"))
3753
        return 1;
3754

3755 3756
    qemuReportError(VIR_ERR_INVALID_ARG,
                    _("unknown type '%s'"), type);
3757 3758 3759
    return -1;
}

3760

3761
static char *qemudGetCapabilities(virConnectPtr conn) {
3762
    struct qemud_driver *driver = conn->privateData;
3763
    virCapsPtr caps = NULL;
3764
    char *xml = NULL;
3765

3766
    qemuDriverLock(driver);
3767

3768
    if ((caps = qemuCreateCapabilities(qemu_driver->caps,
3769
                                       qemu_driver)) == NULL) {
3770 3771 3772
        virCapabilitiesFree(caps);
        goto cleanup;
    }
3773

3774
    virCapabilitiesFree(qemu_driver->caps);
3775 3776 3777
    qemu_driver->caps = caps;

    if ((xml = virCapabilitiesFormatXML(driver->caps)) == NULL)
3778
        virReportOOMError();
3779 3780

cleanup:
3781
    qemuDriverUnlock(driver);
3782

3783
    return xml;
3784 3785 3786
}


3787
static int qemudGetProcessInfo(unsigned long long *cpuTime, int *lastCpu, int pid, int tid) {
D
Daniel P. Berrange 已提交
3788 3789
    char proc[PATH_MAX];
    FILE *pidinfo;
3790
    unsigned long long usertime, systime;
3791 3792
    int cpu;
    int ret;
D
Daniel P. Berrange 已提交
3793

3794 3795 3796 3797 3798 3799
    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
3800 3801 3802 3803
        return -1;
    }

    if (!(pidinfo = fopen(proc, "r"))) {
3804
        /*printf("cannot read pid info");*/
D
Daniel P. Berrange 已提交
3805
        /* VM probably shut down, so fake 0 */
3806 3807 3808 3809
        if (cpuTime)
            *cpuTime = 0;
        if (lastCpu)
            *lastCpu = 0;
D
Daniel P. Berrange 已提交
3810 3811 3812
        return 0;
    }

3813 3814 3815 3816 3817 3818 3819 3820 3821 3822
    /* 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) {
3823
        fclose(pidinfo);
3824 3825
        VIR_WARN0("cannot parse process status data");
        errno = -EINVAL;
D
Daniel P. Berrange 已提交
3826 3827 3828 3829 3830 3831 3832 3833
        return -1;
    }

    /* We got jiffies
     * We want nanoseconds
     * _SC_CLK_TCK is jiffies per second
     * So calulate thus....
     */
3834 3835 3836 3837 3838
    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
3839

3840 3841
    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
3842 3843 3844 3845 3846 3847 3848

    fclose(pidinfo);

    return 0;
}


3849
static virDomainPtr qemudDomainLookupByID(virConnectPtr conn,
3850
                                          int id) {
3851 3852 3853 3854
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;

3855
    qemuDriverLock(driver);
3856
    vm  = virDomainFindByID(&driver->domains, id);
3857
    qemuDriverUnlock(driver);
3858 3859

    if (!vm) {
3860 3861
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching id %d"), id);
3862
        goto cleanup;
3863 3864
    }

3865
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
3866
    if (dom) dom->id = vm->def->id;
3867 3868

cleanup:
3869 3870
    if (vm)
        virDomainObjUnlock(vm);
3871 3872
    return dom;
}
3873

3874
static virDomainPtr qemudDomainLookupByUUID(virConnectPtr conn,
3875
                                            const unsigned char *uuid) {
3876 3877 3878
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
3879

3880
    qemuDriverLock(driver);
3881
    vm = virDomainFindByUUID(&driver->domains, uuid);
3882 3883
    qemuDriverUnlock(driver);

3884
    if (!vm) {
3885 3886
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(uuid, uuidstr);
3887 3888
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
3889
        goto cleanup;
3890 3891
    }

3892
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
3893
    if (dom) dom->id = vm->def->id;
3894 3895

cleanup:
3896 3897
    if (vm)
        virDomainObjUnlock(vm);
3898 3899
    return dom;
}
3900

3901
static virDomainPtr qemudDomainLookupByName(virConnectPtr conn,
3902
                                            const char *name) {
3903 3904 3905
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
3906

3907
    qemuDriverLock(driver);
3908
    vm = virDomainFindByName(&driver->domains, name);
3909 3910
    qemuDriverUnlock(driver);

3911
    if (!vm) {
3912 3913
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching name '%s'"), name);
3914
        goto cleanup;
3915 3916
    }

3917
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
3918
    if (dom) dom->id = vm->def->id;
3919 3920

cleanup:
3921 3922
    if (vm)
        virDomainObjUnlock(vm);
3923 3924 3925
    return dom;
}

3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936

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) {
3937
        qemuReportError(VIR_ERR_NO_DOMAIN, NULL);
3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957
        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) {
3958
        qemuReportError(VIR_ERR_NO_DOMAIN, NULL);
3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 3969
        goto cleanup;
    }
    ret = obj->persistent;

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


3970
static int qemudGetVersion(virConnectPtr conn, unsigned long *version) {
3971 3972 3973
    struct qemud_driver *driver = conn->privateData;
    int ret = -1;

3974
    qemuDriverLock(driver);
3975
    if (qemudExtractVersion(driver) < 0)
3976
        goto cleanup;
3977

3978
    *version = qemu_driver->qemuVersion;
3979 3980 3981
    ret = 0;

cleanup:
3982
    qemuDriverUnlock(driver);
3983
    return ret;
D
Daniel P. Berrange 已提交
3984 3985
}

3986
static int qemudListDomains(virConnectPtr conn, int *ids, int nids) {
3987
    struct qemud_driver *driver = conn->privateData;
3988
    int n;
3989

3990
    qemuDriverLock(driver);
3991
    n = virDomainObjListGetActiveIDs(&driver->domains, ids, nids);
3992
    qemuDriverUnlock(driver);
3993

3994
    return n;
D
Daniel P. Berrange 已提交
3995
}
3996

3997
static int qemudNumDomains(virConnectPtr conn) {
3998
    struct qemud_driver *driver = conn->privateData;
3999
    int n;
4000

4001
    qemuDriverLock(driver);
4002
    n = virDomainObjListNumOfDomains(&driver->domains, 1);
4003
    qemuDriverUnlock(driver);
4004

4005
    return n;
D
Daniel P. Berrange 已提交
4006
}
4007

4008
static virDomainPtr qemudDomainCreate(virConnectPtr conn, const char *xml,
4009
                                      unsigned int flags ATTRIBUTE_UNUSED) {
4010
    struct qemud_driver *driver = conn->privateData;
4011
    virDomainDefPtr def;
4012
    virDomainObjPtr vm = NULL;
4013
    virDomainPtr dom = NULL;
4014
    virDomainEventPtr event = NULL;
D
Daniel P. Berrange 已提交
4015

4016
    qemuDriverLock(driver);
4017
    if (!(def = virDomainDefParseString(driver->caps, xml,
4018
                                        VIR_DOMAIN_XML_INACTIVE)))
4019
        goto cleanup;
4020

4021
    if (virSecurityDriverVerify(def) < 0)
4022 4023
        goto cleanup;

4024 4025
    if (virDomainObjIsDuplicate(&driver->domains, def, 1) < 0)
        goto cleanup;
4026

4027 4028 4029 4030 4031 4032
    if (qemudCanonicalizeMachine(driver, def) < 0)
        goto cleanup;

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

4033
    if (!(vm = virDomainAssignDef(driver->caps,
4034
                                  &driver->domains,
4035
                                  def, false)))
4036 4037 4038
        goto cleanup;

    def = NULL;
D
Daniel P. Berrange 已提交
4039

4040 4041 4042
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup; /* XXXX free the 'vm' we created ? */

4043
    if (qemudStartVMDaemon(conn, driver, vm, NULL, -1) < 0) {
4044 4045 4046
        if (qemuDomainObjEndJob(vm) > 0)
            virDomainRemoveInactive(&driver->domains,
                                    vm);
4047
        vm = NULL;
4048
        goto endjob;
D
Daniel P. Berrange 已提交
4049
    }
4050 4051 4052 4053

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_BOOTED);
D
Daniel P. Berrange 已提交
4054

4055
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
4056
    if (dom) dom->id = vm->def->id;
4057

4058
endjob:
4059 4060 4061
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4062

4063 4064
cleanup:
    virDomainDefFree(def);
4065 4066
    if (vm)
        virDomainObjUnlock(vm);
4067 4068
    if (event)
        qemuDomainEventQueue(driver, event);
4069
    qemuDriverUnlock(driver);
4070
    return dom;
D
Daniel P. Berrange 已提交
4071 4072 4073
}


4074
static int qemudDomainSuspend(virDomainPtr dom) {
4075 4076 4077
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4078
    virDomainEventPtr event = NULL;
4079
    qemuDomainObjPrivatePtr priv;
4080

4081
    qemuDriverLock(driver);
4082
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4083

D
Daniel P. Berrange 已提交
4084
    if (!vm) {
4085 4086
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4087 4088
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4089
        goto cleanup;
D
Daniel P. Berrange 已提交
4090
    }
D
Daniel P. Berrange 已提交
4091
    if (!virDomainObjIsActive(vm)) {
4092 4093
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
4094
        goto cleanup;
D
Daniel P. Berrange 已提交
4095
    }
4096 4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108 4109 4110 4111 4112 4113

    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"));
4114
            goto endjob;
4115
        }
4116 4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 4127 4128 4129 4130 4131
        if (vm->state != VIR_DOMAIN_PAUSED) {
            int rc;

            qemuDomainObjEnterMonitorWithDriver(driver, vm);
            rc = qemuMonitorStopCPUs(priv->mon);
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            if (rc < 0)
                goto endjob;
            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)
            goto endjob;
        ret = 0;
D
Daniel P. Berrange 已提交
4132
    }
4133

4134
endjob:
4135 4136
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4137

4138
cleanup:
4139 4140
    if (vm)
        virDomainObjUnlock(vm);
4141

4142
    if (event)
4143
        qemuDomainEventQueue(driver, event);
4144
    qemuDriverUnlock(driver);
4145
    return ret;
D
Daniel P. Berrange 已提交
4146 4147 4148
}


4149
static int qemudDomainResume(virDomainPtr dom) {
4150 4151 4152
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4153
    virDomainEventPtr event = NULL;
4154

4155
    qemuDriverLock(driver);
4156
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4157

D
Daniel P. Berrange 已提交
4158
    if (!vm) {
4159 4160
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4161 4162
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4163
        goto cleanup;
D
Daniel P. Berrange 已提交
4164
    }
4165 4166 4167 4168

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

D
Daniel P. Berrange 已提交
4169
    if (!virDomainObjIsActive(vm)) {
4170 4171
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
4172
        goto endjob;
D
Daniel P. Berrange 已提交
4173
    }
4174
    if (vm->state == VIR_DOMAIN_PAUSED) {
4175
        qemuDomainObjPrivatePtr priv = vm->privateData;
4176
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
4177
        if (qemuMonitorStartCPUs(priv->mon, dom->conn) < 0) {
4178
            qemuDomainObjExitMonitorWithDriver(driver, vm);
4179
            if (virGetLastError() == NULL)
4180 4181
                qemuReportError(VIR_ERR_OPERATION_FAILED,
                                "%s", _("resume operation failed"));
4182
            goto endjob;
4183
        }
4184
        qemuDomainObjExitMonitorWithDriver(driver, vm);
4185
        vm->state = VIR_DOMAIN_RUNNING;
4186 4187 4188
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_UNPAUSED);
D
Daniel P. Berrange 已提交
4189
    }
4190
    if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
4191
        goto endjob;
4192 4193
    ret = 0;

4194
endjob:
4195 4196
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4197

4198
cleanup:
4199 4200
    if (vm)
        virDomainObjUnlock(vm);
4201
    if (event)
4202
        qemuDomainEventQueue(driver, event);
4203
    qemuDriverUnlock(driver);
4204
    return ret;
D
Daniel P. Berrange 已提交
4205 4206 4207
}


4208
static int qemudDomainShutdown(virDomainPtr dom) {
4209 4210 4211
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4212

4213
    qemuDriverLock(driver);
4214
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4215 4216
    qemuDriverUnlock(driver);

4217
    if (!vm) {
4218 4219
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4220 4221
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4222
        goto cleanup;
4223 4224
    }

4225 4226 4227
    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
4228
    if (!virDomainObjIsActive(vm)) {
4229 4230
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
4231
        goto endjob;
4232 4233
    }

4234
    qemuDomainObjPrivatePtr priv = vm->privateData;
4235 4236 4237
    qemuDomainObjEnterMonitor(vm);
    ret = qemuMonitorSystemPowerdown(priv->mon);
    qemuDomainObjExitMonitor(vm);
4238

4239
endjob:
4240 4241
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4242

4243
cleanup:
4244 4245
    if (vm)
        virDomainObjUnlock(vm);
4246
    return ret;
4247 4248 4249
}


4250
static int qemudDomainDestroy(virDomainPtr dom) {
4251 4252 4253
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4254
    virDomainEventPtr event = NULL;
4255

4256
    qemuDriverLock(driver);
4257
    vm  = virDomainFindByUUID(&driver->domains, dom->uuid);
D
Daniel P. Berrange 已提交
4258
    if (!vm) {
4259 4260
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4261 4262
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4263
        goto cleanup;
D
Daniel P. Berrange 已提交
4264
    }
4265 4266 4267 4268

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

D
Daniel P. Berrange 已提交
4269
    if (!virDomainObjIsActive(vm)) {
4270 4271
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
4272
        goto endjob;
4273
    }
4274

4275
    qemudShutdownVMDaemon(driver, vm);
4276 4277 4278
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_DESTROYED);
4279
    if (!vm->persistent) {
4280 4281 4282
        if (qemuDomainObjEndJob(vm) > 0)
            virDomainRemoveInactive(&driver->domains,
                                    vm);
4283 4284
        vm = NULL;
    }
4285 4286
    ret = 0;

4287
endjob:
4288 4289 4290
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4291

4292
cleanup:
4293 4294
    if (vm)
        virDomainObjUnlock(vm);
4295 4296
    if (event)
        qemuDomainEventQueue(driver, event);
4297
    qemuDriverUnlock(driver);
4298
    return ret;
D
Daniel P. Berrange 已提交
4299 4300 4301
}


4302
static char *qemudDomainGetOSType(virDomainPtr dom) {
4303 4304 4305
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    char *type = NULL;
4306

4307
    qemuDriverLock(driver);
4308
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4309
    qemuDriverUnlock(driver);
4310
    if (!vm) {
4311 4312
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4313 4314
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4315
        goto cleanup;
4316 4317
    }

4318
    if (!(type = strdup(vm->def->os.type)))
4319
        virReportOOMError();
4320 4321

cleanup:
4322 4323
    if (vm)
        virDomainObjUnlock(vm);
4324 4325 4326
    return type;
}

4327 4328
/* Returns max memory in kb, 0 if error */
static unsigned long qemudDomainGetMaxMemory(virDomainPtr dom) {
4329 4330 4331
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    unsigned long ret = 0;
4332

4333
    qemuDriverLock(driver);
4334
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4335 4336
    qemuDriverUnlock(driver);

4337
    if (!vm) {
4338 4339
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4340 4341
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4342
        goto cleanup;
4343 4344
    }

4345 4346 4347
    ret = vm->def->maxmem;

cleanup:
4348 4349
    if (vm)
        virDomainObjUnlock(vm);
4350
    return ret;
4351 4352 4353
}

static int qemudDomainSetMemory(virDomainPtr dom, unsigned long newmem) {
4354
    struct qemud_driver *driver = dom->conn->privateData;
4355
    qemuDomainObjPrivatePtr priv;
4356
    virDomainObjPtr vm;
4357
    int ret = -1, r;
4358

4359
    qemuDriverLock(driver);
4360
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4361
    qemuDriverUnlock(driver);
4362
    if (!vm) {
4363 4364
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4365 4366
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4367
        goto cleanup;
4368 4369
    }

4370 4371 4372 4373 4374 4375
    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
        goto cleanup;
    }

4376
    if (newmem > vm->def->maxmem) {
4377 4378
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("cannot set memory higher than max memory"));
4379
        goto cleanup;
4380 4381
    }

4382 4383 4384
    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

4385 4386 4387 4388 4389 4390
    priv = vm->privateData;
    qemuDomainObjEnterMonitor(vm);
    r = qemuMonitorSetBalloon(priv->mon, newmem);
    qemuDomainObjExitMonitor(vm);
    if (r < 0)
        goto endjob;
4391

4392 4393 4394 4395 4396
    /* 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;
4397
    }
4398

4399
    ret = 0;
4400
endjob:
4401 4402
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4403

4404
cleanup:
4405 4406
    if (vm)
        virDomainObjUnlock(vm);
4407
    return ret;
4408 4409
}

4410
static int qemudDomainGetInfo(virDomainPtr dom,
4411
                              virDomainInfoPtr info) {
4412 4413 4414
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4415 4416
    int err;
    unsigned long balloon;
4417

4418
    qemuDriverLock(driver);
4419
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4420
    qemuDriverUnlock(driver);
D
Daniel P. Berrange 已提交
4421
    if (!vm) {
4422 4423
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4424 4425
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4426
        goto cleanup;
D
Daniel P. Berrange 已提交
4427 4428
    }

4429
    info->state = vm->state;
D
Daniel P. Berrange 已提交
4430

D
Daniel P. Berrange 已提交
4431
    if (!virDomainObjIsActive(vm)) {
4432
        info->cpuTime = 0;
D
Daniel P. Berrange 已提交
4433
    } else {
4434
        if (qemudGetProcessInfo(&(info->cpuTime), NULL, vm->pid, 0) < 0) {
4435
            qemuReportError(VIR_ERR_OPERATION_FAILED, ("cannot read cputime for domain"));
4436
            goto cleanup;
D
Daniel P. Berrange 已提交
4437 4438 4439
        }
    }

4440
    info->maxMem = vm->def->maxmem;
4441

D
Daniel P. Berrange 已提交
4442
    if (virDomainObjIsActive(vm)) {
4443
        qemuDomainObjPrivatePtr priv = vm->privateData;
4444 4445 4446
        if (!priv->jobActive) {
            if (qemuDomainObjBeginJob(vm) < 0)
                goto cleanup;
4447

4448 4449 4450 4451
            qemuDomainObjEnterMonitor(vm);
            err = qemuMonitorGetBalloonInfo(priv->mon, &balloon);
            qemuDomainObjExitMonitor(vm);
            if (err < 0) {
4452 4453
                if (qemuDomainObjEndJob(vm) == 0)
                    vm = NULL;
4454 4455 4456 4457 4458 4459 4460 4461 4462
                goto cleanup;
            }

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

4463 4464 4465 4466
            if (qemuDomainObjEndJob(vm) == 0) {
                vm = NULL;
                goto cleanup;
            }
4467 4468 4469
        } else {
            info->memory = vm->def->memory;
        }
4470 4471 4472 4473
    } else {
        info->memory = vm->def->memory;
    }

4474
    info->nrVirtCpu = vm->def->vcpus;
4475 4476 4477
    ret = 0;

cleanup:
4478 4479
    if (vm)
        virDomainObjUnlock(vm);
4480
    return ret;
D
Daniel P. Berrange 已提交
4481 4482 4483
}


4484 4485 4486 4487 4488 4489 4490 4491 4492 4493 4494 4495 4496 4497 4498 4499 4500 4501 4502 4503 4504 4505 4506 4507 4508 4509 4510 4511 4512
/** qemuDomainMigrateOffline:
 * Pause domain for non-live migration.
 */
static int
qemuDomainMigrateOffline(struct qemud_driver *driver,
                         virDomainObjPtr vm)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int ret;

    qemuDomainObjEnterMonitorWithDriver(driver, vm);
    ret = qemuMonitorStopCPUs(priv->mon);
    qemuDomainObjExitMonitorWithDriver(driver, vm);

    if (ret == 0) {
        virDomainEventPtr event;

        vm->state = VIR_DOMAIN_PAUSED;
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_SUSPENDED,
                                         VIR_DOMAIN_EVENT_SUSPENDED_MIGRATED);
        if (event)
            qemuDomainEventQueue(driver, event);
    }

    return ret;
}


4513 4514 4515 4516 4517 4518 4519 4520 4521 4522 4523 4524 4525 4526
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 */
4527
        struct timespec ts = { .tv_sec = 0, .tv_nsec = 50 * 1000 * 1000ull };
4528 4529 4530
        struct timeval now;
        int rc;

4531 4532
        if (priv->jobSignals & QEMU_JOB_SIGNAL_CANCEL) {
            priv->jobSignals ^= QEMU_JOB_SIGNAL_CANCEL;
4533 4534 4535 4536 4537 4538 4539
            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");
            }
4540 4541 4542 4543 4544
        } 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");
4545 4546 4547 4548 4549 4550 4551 4552 4553 4554 4555
        } 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");
4556 4557
        }

4558 4559 4560 4561 4562 4563 4564 4565 4566 4567 4568 4569 4570 4571 4572 4573 4574 4575 4576 4577 4578 4579 4580 4581 4582 4583 4584 4585 4586 4587 4588 4589 4590 4591 4592 4593 4594 4595 4596 4597 4598 4599 4600 4601 4602 4603 4604 4605 4606 4607 4608 4609 4610 4611 4612 4613 4614 4615 4616 4617 4618 4619 4620 4621 4622 4623 4624 4625 4626 4627 4628 4629
        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;
}


4630
#define QEMUD_SAVE_MAGIC "LibvirtQemudSave"
4631 4632 4633
#define QEMUD_SAVE_VERSION 2

enum qemud_save_formats {
4634 4635 4636
    QEMUD_SAVE_FORMAT_RAW = 0,
    QEMUD_SAVE_FORMAT_GZIP = 1,
    QEMUD_SAVE_FORMAT_BZIP2 = 2,
4637 4638
    /*
     * Deprecated by xz and never used as part of a release
4639
     * QEMUD_SAVE_FORMAT_LZMA
4640 4641
     */
    QEMUD_SAVE_FORMAT_XZ = 3,
4642
    QEMUD_SAVE_FORMAT_LZOP = 4,
4643 4644 4645
    /* Note: add new members only at the end.
       These values are used in the on-disk format.
       Do not change or re-use numbers. */
4646 4647

    QEMUD_SAVE_FORMAT_LAST
4648
};
4649

4650 4651 4652 4653 4654
VIR_ENUM_DECL(qemudSaveCompression)
VIR_ENUM_IMPL(qemudSaveCompression, QEMUD_SAVE_FORMAT_LAST,
              "raw",
              "gzip",
              "bzip2",
4655 4656
              "xz",
              "lzop")
4657

4658 4659 4660 4661 4662
struct qemud_save_header {
    char magic[sizeof(QEMUD_SAVE_MAGIC)-1];
    int version;
    int xml_len;
    int was_running;
4663 4664
    int compressed;
    int unused[15];
4665 4666
};

4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684 4685 4686 4687 4688 4689 4690 4691 4692 4693 4694 4695
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,
                         _("failed to write save header to '%s'"), hdata->path);
        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;
}


4696 4697
static int qemudDomainSaveFlag(virDomainPtr dom, const char *path,
                               int compressed)
4698
{
4699
    struct qemud_driver *driver = dom->conn->privateData;
4700
    virDomainObjPtr vm = NULL;
4701
    char *xml = NULL;
4702
    struct qemud_save_header header;
4703 4704
    struct fileOpHookData hdata;
    int bypassSecurityDriver = 0;
4705
    int ret = -1;
4706
    int rc;
4707
    virDomainEventPtr event = NULL;
4708
    qemuDomainObjPrivatePtr priv;
4709 4710 4711 4712 4713

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

4714
    qemuDriverLock(driver);
4715 4716

    header.compressed = compressed;
4717

4718
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4719

D
Daniel P. Berrange 已提交
4720
    if (!vm) {
4721 4722
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4723 4724
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4725
        goto cleanup;
D
Daniel P. Berrange 已提交
4726
    }
4727
    priv = vm->privateData;
4728

4729 4730 4731
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
4732
    if (!virDomainObjIsActive(vm)) {
4733 4734
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
4735
        goto endjob;
D
Daniel P. Berrange 已提交
4736
    }
4737

4738 4739 4740
    memset(&priv->jobInfo, 0, sizeof(priv->jobInfo));
    priv->jobInfo.type = VIR_DOMAIN_JOB_UNBOUNDED;

4741 4742 4743
    /* Pause */
    if (vm->state == VIR_DOMAIN_RUNNING) {
        header.was_running = 1;
4744
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
4745
        if (qemuMonitorStopCPUs(priv->mon) < 0) {
4746 4747
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            goto endjob;
4748
        }
4749
        qemuDomainObjExitMonitorWithDriver(driver, vm);
4750
        vm->state = VIR_DOMAIN_PAUSED;
4751 4752 4753
    }

    /* Get XML for the domain */
4754
    xml = virDomainDefFormat(vm->def, VIR_DOMAIN_XML_SECURE);
4755
    if (!xml) {
4756 4757
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to get domain xml"));
4758
        goto endjob;
4759 4760 4761
    }
    header.xml_len = strlen(xml) + 1;

4762 4763 4764 4765 4766 4767
    /* Setup hook data needed by virFileOperation hook function */
    hdata.dom = dom;
    hdata.path = path;
    hdata.xml = xml;
    hdata.header = &header;

4768 4769
    /* Write header to file, followed by XML */

4770 4771 4772 4773 4774 4775 4776 4777 4778 4779 4780 4781 4782 4783 4784 4785 4786 4787 4788 4789
    /* First try creating the file as root */
    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;
        }
4790

4791 4792 4793 4794 4795
#ifdef __linux__
        /* On Linux we can also verify the FS-type of the directory. */
        char *dirpath, *p;
        struct statfs st;
        int statfs_ret;
4796

4797 4798 4799 4800 4801 4802 4803 4804 4805 4806 4807 4808 4809 4810 4811 4812 4813 4814 4815 4816 4817 4818 4819 4820 4821 4822 4823 4824 4825 4826 4827 4828 4829 4830 4831 4832 4833 4834 4835 4836 4837 4838 4839 4840 4841 4842 4843 4844 4845 4846 4847 4848 4849 4850 4851 4852 4853 4854 4855 4856 4857 4858 4859 4860 4861 4862 4863
        if ((dirpath = strdup(path)) == NULL) {
            virReportOOMError();
            goto endjob;
        }

        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;
            }

            if (p == dirpath)
                *(p+1) = '\0';
            else
                *p = '\0';

            statfs_ret = statfs(dirpath, &st);

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

        if (statfs_ret == -1) {
            virReportSystemError(errno,
                                 _("Failed to create domain save file '%s'"
                                   " statfs of all elements of path failed."),
                                 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;
        }
        VIR_FREE(dirpath);
#endif

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

        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'"),
                                 path);
            goto endjob;
        }

        /* 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 */

        bypassSecurityDriver = 1;
4864
    }
4865

4866 4867 4868

    if ((!bypassSecurityDriver) &&
        driver->securityDriver &&
4869
        driver->securityDriver->domainSetSavedStateLabel &&
4870
        driver->securityDriver->domainSetSavedStateLabel(vm, path) == -1)
4871 4872
        goto endjob;

4873 4874
    if (header.compressed == QEMUD_SAVE_FORMAT_RAW) {
        const char *args[] = { "cat", NULL };
M
Matthias Bolte 已提交
4875
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
4876
        rc = qemuMonitorMigrateToCommand(priv->mon, 1, args, path);
M
Matthias Bolte 已提交
4877
        qemuDomainObjExitMonitorWithDriver(driver, vm);
4878
    } else {
4879
        const char *prog = qemudSaveCompressionTypeToString(header.compressed);
4880 4881 4882 4883 4884
        const char *args[] = {
            prog,
            "-c",
            NULL
        };
M
Matthias Bolte 已提交
4885
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
4886
        rc = qemuMonitorMigrateToCommand(priv->mon, 1, args, path);
M
Matthias Bolte 已提交
4887
        qemuDomainObjExitMonitorWithDriver(driver, vm);
4888 4889
    }

4890 4891 4892 4893 4894
    if (rc < 0)
        goto endjob;

    rc = qemuDomainWaitForMigrationComplete(driver, vm);

4895
    if (rc < 0)
4896
        goto endjob;
4897

4898 4899
    if ((!bypassSecurityDriver) &&
        driver->securityDriver &&
4900
        driver->securityDriver->domainRestoreSavedStateLabel &&
4901
        driver->securityDriver->domainRestoreSavedStateLabel(vm, path) == -1)
4902 4903 4904 4905
        goto endjob;

    ret = 0;

4906
    /* Shut it down */
4907
    qemudShutdownVMDaemon(driver, vm);
4908 4909 4910
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_SAVED);
4911
    if (!vm->persistent) {
4912 4913 4914
        if (qemuDomainObjEndJob(vm) > 0)
            virDomainRemoveInactive(&driver->domains,
                                    vm);
4915 4916
        vm = NULL;
    }
4917

4918
endjob:
4919 4920 4921 4922 4923 4924 4925 4926 4927 4928
    if (ret != 0 && header.was_running) {
        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;
    }

4929 4930
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
4931
            vm = NULL;
4932

4933 4934 4935 4936
cleanup:
    VIR_FREE(xml);
    if (ret != 0)
        unlink(path);
4937 4938
    if (vm)
        virDomainObjUnlock(vm);
4939 4940
    if (event)
        qemuDomainEventQueue(driver, event);
4941
    qemuDriverUnlock(driver);
4942
    return ret;
D
Daniel P. Berrange 已提交
4943 4944
}

4945 4946 4947 4948 4949 4950 4951 4952 4953 4954 4955 4956 4957 4958 4959 4960 4961 4962 4963 4964 4965 4966 4967 4968 4969 4970 4971 4972 4973 4974 4975 4976 4977 4978 4979 4980 4981 4982 4983 4984 4985 4986
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;

4987
    virCheckFlags(0, -1);
4988 4989 4990 4991 4992 4993 4994 4995 4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009 5010 5011 5012 5013 5014 5015 5016 5017 5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047

    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;

5048
    virCheckFlags(0, -1);
5049 5050 5051 5052 5053 5054 5055 5056 5057 5058 5059 5060 5061 5062 5063 5064 5065 5066 5067 5068 5069 5070 5071 5072 5073 5074 5075 5076 5077 5078 5079 5080 5081

    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;

5082
    virCheckFlags(0, -1);
5083 5084 5085 5086 5087 5088 5089 5090 5091 5092 5093 5094 5095 5096 5097 5098 5099 5100 5101 5102 5103 5104 5105 5106

    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
5107

P
Paolo Bonzini 已提交
5108 5109 5110 5111 5112 5113
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;
5114
    int ret = -1, fd = -1;
5115
    virDomainEventPtr event = NULL;
5116 5117 5118 5119
    const char *args[] = {
        "cat",
        NULL,
    };
5120
    qemuDomainObjPrivatePtr priv;
P
Paolo Bonzini 已提交
5121 5122 5123 5124 5125 5126 5127

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

    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
5128 5129
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
P
Paolo Bonzini 已提交
5130 5131
        goto cleanup;
    }
5132
    priv = vm->privateData;
P
Paolo Bonzini 已提交
5133

5134
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
5135 5136
        goto cleanup;

D
Daniel P. Berrange 已提交
5137
    if (!virDomainObjIsActive(vm)) {
5138 5139
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
5140
        goto endjob;
P
Paolo Bonzini 已提交
5141 5142
    }

5143 5144
    /* Create an empty file with appropriate ownership.  */
    if ((fd = open(path, O_CREAT|O_TRUNC|O_WRONLY, S_IRUSR|S_IWUSR)) < 0) {
5145 5146
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("failed to create '%s'"), path);
5147 5148 5149 5150
        goto endjob;
    }

    if (close(fd) < 0) {
5151
        virReportSystemError(errno,
5152 5153 5154 5155 5156 5157 5158
                             _("unable to save file %s"),
                             path);
        goto endjob;
    }

    if (driver->securityDriver &&
        driver->securityDriver->domainSetSavedStateLabel &&
5159
        driver->securityDriver->domainSetSavedStateLabel(vm, path) == -1)
5160 5161
        goto endjob;

P
Paolo Bonzini 已提交
5162 5163
    /* 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
5164 5165 5166
    resume = (vm->state == VIR_DOMAIN_RUNNING);

    /* Pause domain for non-live dump */
P
Paolo Bonzini 已提交
5167
    if (!(flags & VIR_DUMP_LIVE) && vm->state == VIR_DOMAIN_RUNNING) {
5168
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
5169
        if (qemuMonitorStopCPUs(priv->mon) < 0) {
5170
            qemuDomainObjExitMonitorWithDriver(driver, vm);
5171
            goto endjob;
5172
        }
5173
        qemuDomainObjExitMonitorWithDriver(driver, vm);
P
Paolo Bonzini 已提交
5174 5175 5176
        paused = 1;
    }

5177
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
5178
    ret = qemuMonitorMigrateToCommand(priv->mon, 1, args, path);
5179
    qemuDomainObjExitMonitorWithDriver(driver, vm);
5180 5181 5182 5183 5184 5185 5186 5187 5188 5189

    if (ret < 0)
        goto endjob;

    ret = qemuDomainWaitForMigrationComplete(driver, vm);

    if (ret < 0)
        goto endjob;

    paused = 1;
5190 5191 5192

    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSavedStateLabel &&
5193
        driver->securityDriver->domainRestoreSavedStateLabel(vm, path) == -1)
5194 5195 5196
        goto endjob;

endjob:
5197
    if ((ret == 0) && (flags & VIR_DUMP_CRASH)) {
5198
        qemudShutdownVMDaemon(driver, vm);
5199 5200 5201 5202 5203
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         VIR_DOMAIN_EVENT_STOPPED_CRASHED);
    }

P
Paolo Bonzini 已提交
5204 5205 5206
    /* 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.  */
5207
    else if (resume && paused) {
5208
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
5209
        if (qemuMonitorStartCPUs(priv->mon, dom->conn) < 0) {
5210
            if (virGetLastError() == NULL)
5211 5212
                qemuReportError(VIR_ERR_OPERATION_FAILED,
                                "%s", _("resuming after dump failed"));
P
Paolo Bonzini 已提交
5213
        }
5214
        qemuDomainObjExitMonitorWithDriver(driver, vm);
P
Paolo Bonzini 已提交
5215
    }
5216

5217 5218
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
5219
    else if ((ret == 0) && (flags & VIR_DUMP_CRASH) && !vm->persistent) {
5220 5221 5222 5223
        virDomainRemoveInactive(&driver->domains,
                                vm);
        vm = NULL;
    }
5224 5225

cleanup:
5226 5227
    if (ret != 0)
        unlink(path);
P
Paolo Bonzini 已提交
5228 5229
    if (vm)
        virDomainObjUnlock(vm);
5230 5231
    if (event)
        qemuDomainEventQueue(driver, event);
5232
    qemuDriverUnlock(driver);
P
Paolo Bonzini 已提交
5233 5234 5235 5236
    return ret;
}


5237 5238 5239 5240 5241 5242 5243 5244 5245 5246 5247 5248
static int qemudDomainHotplugVcpus(virDomainObjPtr vm, unsigned int nvcpus)
{
    qemuDomainObjPrivatePtr priv = vm->privateData;
    int i, rc;
    int ret = -1;

    /* 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 */
5249
            qemuDomainObjEnterMonitor(vm);
5250
            rc = qemuMonitorSetCPU(priv->mon, i, 1);
5251
            qemuDomainObjExitMonitor(vm);
5252 5253 5254 5255 5256 5257 5258 5259 5260 5261
            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 */
5262
            qemuDomainObjEnterMonitor(vm);
5263
            rc = qemuMonitorSetCPU(priv->mon, i, 0);
5264
            qemuDomainObjExitMonitor(vm);
5265 5266 5267 5268 5269 5270 5271 5272 5273 5274 5275 5276 5277 5278 5279 5280 5281 5282 5283 5284 5285 5286
            if (rc == 0)
                goto unsupported;
            if (rc < 0)
                goto cleanup;

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

    ret = 0;

cleanup:
    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) {
5287 5288
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
5289 5290
    const char * type;
    int max;
5291
    int ret = -1;
5292

5293
    qemuDriverLock(driver);
5294
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5295 5296
    qemuDriverUnlock(driver);

5297 5298 5299
    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

5300
    if (!vm) {
5301 5302
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
5303 5304
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
5305
        goto endjob;
5306 5307
    }

5308 5309 5310
    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                         "%s", _("domain is not running"));
5311
        goto endjob;
5312 5313
    }

5314 5315 5316 5317 5318 5319 5320 5321 5322 5323 5324 5325 5326 5327 5328 5329 5330 5331 5332 5333 5334 5335 5336 5337 5338
    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;
5339

5340
cleanup:
5341 5342
    if (vm)
        virDomainObjUnlock(vm);
5343
    return ret;
5344 5345
}

5346 5347 5348 5349 5350 5351

static int
qemudDomainPinVcpu(virDomainPtr dom,
                   unsigned int vcpu,
                   unsigned char *cpumap,
                   int maplen) {
5352 5353
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
5354
    int maxcpu, hostcpus;
5355
    virNodeInfo nodeinfo;
5356
    int ret = -1;
5357
    qemuDomainObjPrivatePtr priv;
5358

5359
    qemuDriverLock(driver);
5360
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5361 5362
    qemuDriverUnlock(driver);

5363 5364 5365
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
5366 5367
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
5368 5369 5370
        goto cleanup;
    }

D
Daniel P. Berrange 已提交
5371
    if (!virDomainObjIsActive(vm)) {
5372 5373
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s",_("cannot pin vcpus on an inactive domain"));
5374
        goto cleanup;
5375 5376
    }

5377 5378 5379
    priv = vm->privateData;

    if (vcpu > (priv->nvcpupids-1)) {
5380 5381 5382
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("vcpu number out of range %d > %d"),
                        vcpu, priv->nvcpupids);
5383
        goto cleanup;
5384 5385
    }

5386
    if (nodeGetInfo(dom->conn, &nodeinfo) < 0)
5387
        goto cleanup;
5388

5389
    hostcpus = VIR_NODEINFO_MAXCPUS(nodeinfo);
5390
    maxcpu = maplen * 8;
5391 5392
    if (maxcpu > hostcpus)
        maxcpu = hostcpus;
5393

5394 5395
    if (priv->vcpupids != NULL) {
        if (virProcessInfoSetAffinity(priv->vcpupids[vcpu],
5396
                                      cpumap, maplen, maxcpu) < 0)
5397
            goto cleanup;
5398
    } else {
5399 5400
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        "%s", _("cpu affinity is not supported"));
5401
        goto cleanup;
5402
    }
5403
    ret = 0;
5404

5405
cleanup:
5406 5407
    if (vm)
        virDomainObjUnlock(vm);
5408
    return ret;
5409 5410 5411 5412 5413 5414 5415 5416
}

static int
qemudDomainGetVcpus(virDomainPtr dom,
                    virVcpuInfoPtr info,
                    int maxinfo,
                    unsigned char *cpumaps,
                    int maplen) {
5417 5418
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
5419
    virNodeInfo nodeinfo;
5420
    int i, v, maxcpu, hostcpus;
5421
    int ret = -1;
5422
    qemuDomainObjPrivatePtr priv;
5423

5424
    qemuDriverLock(driver);
5425
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5426 5427
    qemuDriverUnlock(driver);

5428 5429 5430
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
5431 5432
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
5433 5434 5435
        goto cleanup;
    }

D
Daniel P. Berrange 已提交
5436
    if (!virDomainObjIsActive(vm)) {
5437 5438 5439
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s",
                        _("cannot list vcpu pinning for an inactive domain"));
5440
        goto cleanup;
5441 5442
    }

5443 5444
    priv = vm->privateData;

5445
    if (nodeGetInfo(dom->conn, &nodeinfo) < 0)
5446
        goto cleanup;
5447

5448
    hostcpus = VIR_NODEINFO_MAXCPUS(nodeinfo);
5449
    maxcpu = maplen * 8;
5450 5451
    if (maxcpu > hostcpus)
        maxcpu = hostcpus;
5452 5453

    /* Clamp to actual number of vcpus */
5454 5455
    if (maxinfo > priv->nvcpupids)
        maxinfo = priv->nvcpupids;
5456

5457 5458 5459 5460 5461 5462
    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;
5463

5464
                if (priv->vcpupids != NULL &&
5465 5466 5467
                    qemudGetProcessInfo(&(info[i].cpuTime),
                                        &(info[i].cpu),
                                        vm->pid,
5468
                                        priv->vcpupids[i]) < 0) {
5469
                    virReportSystemError(errno, "%s",
5470 5471 5472
                                         _("cannot get vCPU placement & pCPU time"));
                    goto cleanup;
                }
5473
            }
5474 5475
        }

5476 5477
        if (cpumaps != NULL) {
            memset(cpumaps, 0, maplen * maxinfo);
5478
            if (priv->vcpupids != NULL) {
5479 5480 5481
                for (v = 0 ; v < maxinfo ; v++) {
                    unsigned char *cpumap = VIR_GET_CPUMAP(cpumaps, maplen, v);

5482
                    if (virProcessInfoGetAffinity(priv->vcpupids[v],
5483
                                                  cpumap, maplen, maxcpu) < 0)
5484
                        goto cleanup;
5485
                }
5486
            } else {
5487 5488
                qemuReportError(VIR_ERR_NO_SUPPORT,
                                "%s", _("cpu affinity is not available"));
5489
                goto cleanup;
5490 5491 5492
            }
        }
    }
5493
    ret = maxinfo;
5494

5495
cleanup:
5496 5497
    if (vm)
        virDomainObjUnlock(vm);
5498
    return ret;
5499 5500 5501
}


5502
static int qemudDomainGetMaxVcpus(virDomainPtr dom) {
5503 5504
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
5505
    const char *type;
5506
    int ret = -1;
5507

5508
    qemuDriverLock(driver);
5509
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5510 5511
    qemuDriverUnlock(driver);

5512
    if (!vm) {
5513 5514
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
5515 5516
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
5517
        goto cleanup;
5518 5519
    }

5520
    if (!(type = virDomainVirtTypeToString(vm->def->virtType))) {
5521 5522 5523
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("unknown virt type in domain definition '%d'"),
                        vm->def->virtType);
5524
        goto cleanup;
5525 5526
    }

5527
    ret = qemudGetMaxVCPUs(NULL, type);
5528

5529
cleanup:
5530 5531
    if (vm)
        virDomainObjUnlock(vm);
5532 5533 5534
    return ret;
}

5535 5536 5537 5538 5539 5540 5541 5542 5543
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);

5544 5545
    memset(seclabel, 0, sizeof(*seclabel));

5546 5547 5548
    if (!vm) {
        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 5552 5553
        goto cleanup;
    }

5554
    if (!virDomainVirtTypeToString(vm->def->virtType)) {
5555 5556 5557
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("unknown virt type in domain definition '%d'"),
                        vm->def->virtType);
5558 5559 5560 5561 5562 5563 5564 5565 5566 5567 5568 5569 5570 5571 5572 5573 5574
        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
5575
    if (virDomainObjIsActive(vm)) {
5576
        if (driver->securityDriver && driver->securityDriver->domainGetSecurityProcessLabel) {
5577
            if (driver->securityDriver->domainGetSecurityProcessLabel(vm, seclabel) == -1) {
5578 5579
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                "%s", _("Failed to get security label"));
5580 5581 5582 5583 5584 5585 5586 5587 5588 5589
                goto cleanup;
            }
        }
    }

    ret = 0;

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
5590
    qemuDriverUnlock(driver);
5591 5592 5593
    return ret;
}

5594 5595
static int qemudNodeGetSecurityModel(virConnectPtr conn,
                                     virSecurityModelPtr secmodel)
5596 5597 5598
{
    struct qemud_driver *driver = (struct qemud_driver *)conn->privateData;
    char *p;
5599
    int ret = 0;
5600

5601
    qemuDriverLock(driver);
5602
    if (!driver->securityPrimaryDriver) {
5603
        memset(secmodel, 0, sizeof (*secmodel));
5604 5605
        goto cleanup;
    }
5606

5607 5608
    p = driver->caps->host.secModel.model;
    if (strlen(p) >= VIR_SECURITY_MODEL_BUFLEN-1) {
5609 5610 5611
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("security model string exceeds max %d bytes"),
                        VIR_SECURITY_MODEL_BUFLEN-1);
5612 5613
        ret = -1;
        goto cleanup;
5614 5615 5616 5617 5618
    }
    strcpy(secmodel->model, p);

    p = driver->caps->host.secModel.doi;
    if (strlen(p) >= VIR_SECURITY_DOI_BUFLEN-1) {
5619 5620 5621
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("security DOI string exceeds max %d bytes"),
                        VIR_SECURITY_DOI_BUFLEN-1);
5622 5623
        ret = -1;
        goto cleanup;
5624 5625
    }
    strcpy(secmodel->doi, p);
5626 5627 5628 5629

cleanup:
    qemuDriverUnlock(driver);
    return ret;
5630 5631
}

5632 5633 5634 5635 5636 5637 5638 5639 5640 5641 5642 5643 5644 5645 5646 5647 5648 5649 5650 5651 5652 5653 5654 5655 5656 5657 5658 5659 5660 5661 5662 5663 5664 5665 5666 5667 5668 5669 5670 5671 5672 5673 5674 5675 5676 5677 5678 5679 5680 5681 5682 5683 5684 5685 5686 5687 5688 5689 5690 5691 5692 5693 5694 5695 5696 5697 5698 5699 5700 5701 5702 5703 5704 5705 5706 5707 5708 5709 5710 5711 5712 5713 5714 5715 5716 5717 5718 5719 5720 5721 5722 5723 5724 5725 5726 5727 5728 5729 5730 5731 5732 5733 5734 5735 5736 5737 5738 5739 5740 5741 5742 5743 5744 5745 5746 5747 5748 5749 5750 5751 5752 5753 5754 5755 5756 5757 5758 5759 5760 5761 5762 5763
/* 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);
}

5764
/* TODO: check seclabel restore */
5765
static int qemudDomainRestore(virConnectPtr conn,
5766 5767 5768
                              const char *path) {
    struct qemud_driver *driver = conn->privateData;
    virDomainDefPtr def = NULL;
5769
    virDomainObjPtr vm = NULL;
5770
    int fd = -1;
5771
    pid_t read_pid = -1;
5772 5773
    int ret = -1;
    char *xml = NULL;
5774
    struct qemud_save_header header;
5775
    virDomainEventPtr event = NULL;
5776 5777 5778
    int intermediatefd = -1;
    pid_t intermediate_pid = -1;
    int childstat;
5779

5780
    qemuDriverLock(driver);
5781 5782
    /* Verify the header and read the XML */
    if ((fd = open(path, O_RDONLY)) < 0) {
5783 5784 5785 5786 5787 5788 5789 5790 5791 5792 5793 5794 5795 5796
        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;
        }
5797 5798 5799
    }

    if (saferead(fd, &header, sizeof(header)) != sizeof(header)) {
5800 5801
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to read qemu header"));
5802
        goto cleanup;
5803 5804 5805
    }

    if (memcmp(header.magic, QEMUD_SAVE_MAGIC, sizeof(header.magic)) != 0) {
5806 5807
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("image magic is incorrect"));
5808
        goto cleanup;
5809 5810 5811
    }

    if (header.version > QEMUD_SAVE_VERSION) {
5812 5813 5814
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("image version is not supported (%d > %d)"),
                        header.version, QEMUD_SAVE_VERSION);
5815
        goto cleanup;
5816 5817
    }

5818 5819 5820
    if (header.xml_len <= 0) {
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("invalid XML length: %d"), header.xml_len);
5821
        goto cleanup;
5822 5823
    }

5824 5825 5826 5827 5828 5829
    if (VIR_ALLOC_N(xml, header.xml_len) < 0) {
        virReportOOMError();
        goto cleanup;
    }

    if (saferead(fd, xml, header.xml_len) != header.xml_len) {
5830 5831
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to read XML"));
5832
        goto cleanup;
5833 5834 5835
    }

    /* Create a domain from this XML */
5836
    if (!(def = virDomainDefParseString(driver->caps, xml,
5837
                                        VIR_DOMAIN_XML_INACTIVE))) {
5838 5839
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to parse XML"));
5840
        goto cleanup;
5841 5842
    }

5843 5844
    if (virDomainObjIsDuplicate(&driver->domains, def, 1) < 0)
        goto cleanup;
5845

5846
    if (!(vm = virDomainAssignDef(driver->caps,
5847
                                  &driver->domains,
5848
                                  def, true))) {
5849 5850
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to assign new VM"));
5851
        goto cleanup;
5852
    }
5853
    def = NULL;
5854

5855 5856 5857
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

5858 5859
    if (header.version == 2) {
        const char *intermediate_argv[3] = { NULL, "-dc", NULL };
5860 5861
        const char *prog = qemudSaveCompressionTypeToString(header.compressed);
        if (prog == NULL) {
5862 5863 5864
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("Invalid compressed save format %d"),
                            header.compressed);
5865
            goto endjob;
5866
        }
5867

5868
        if (header.compressed != QEMUD_SAVE_FORMAT_RAW) {
5869
            intermediate_argv[0] = prog;
5870 5871
            intermediatefd = fd;
            fd = -1;
5872
            if (virExec(intermediate_argv, NULL, NULL,
5873
                        &intermediate_pid, intermediatefd, &fd, NULL, 0) < 0) {
5874 5875 5876
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("Failed to start decompression binary %s"),
                                intermediate_argv[0]);
5877
                goto endjob;
5878 5879 5880
            }
        }
    }
5881
    /* Set the migration source and start it up. */
5882
    ret = qemudStartVMDaemon(conn, driver, vm, "stdio", fd);
5883 5884 5885 5886 5887 5888 5889
    if (intermediate_pid != -1) {
        /* Wait for intermediate process to exit */
        while (waitpid(intermediate_pid, &childstat, 0) == -1 &&
               errno == EINTR);
    }
    if (intermediatefd != -1)
        close(intermediatefd);
5890
    close(fd);
5891
    fd = -1;
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
    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;
            }
        }
    }
5921
    if (ret < 0) {
5922
        if (!vm->persistent) {
5923 5924 5925
            if (qemuDomainObjEndJob(vm) > 0)
                virDomainRemoveInactive(&driver->domains,
                                        vm);
5926 5927
            vm = NULL;
        }
5928
        goto endjob;
5929 5930
    }

5931 5932 5933
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_RESTORED);
5934

5935 5936
    /* If it was running before, resume it now. */
    if (header.was_running) {
5937
        qemuDomainObjPrivatePtr priv = vm->privateData;
5938
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
5939
        if (qemuMonitorStartCPUs(priv->mon, conn) < 0) {
5940
            if (virGetLastError() == NULL)
5941 5942
                qemuReportError(VIR_ERR_OPERATION_FAILED,
                                "%s", _("failed to resume domain"));
5943 5944
            qemuDomainObjExitMonitorWithDriver(driver,vm);
            goto endjob;
5945
        }
5946
        qemuDomainObjExitMonitorWithDriver(driver, vm);
5947
        vm->state = VIR_DOMAIN_RUNNING;
5948
        virDomainSaveStatus(driver->caps, driver->stateDir, vm);
5949
    }
5950
    ret = 0;
5951

5952
endjob:
5953 5954 5955
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
5956

5957 5958 5959 5960 5961
cleanup:
    virDomainDefFree(def);
    VIR_FREE(xml);
    if (fd != -1)
        close(fd);
5962 5963 5964 5965 5966 5967 5968
    if (read_pid != 0) {
        /* reap the process that read the file */
        while ((waitpid(read_pid, NULL, 0) == -1)
               && (errno == EINTR)) {
            /* empty */
        }
    }
5969 5970
    if (vm)
        virDomainObjUnlock(vm);
5971 5972
    if (event)
        qemuDomainEventQueue(driver, event);
5973
    qemuDriverUnlock(driver);
5974
    return ret;
D
Daniel P. Berrange 已提交
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 6013 6014 6015
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;
}


6016
static char *qemudDomainDumpXML(virDomainPtr dom,
6017
                                int flags) {
6018 6019 6020
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    char *ret = NULL;
6021 6022
    unsigned long balloon;
    int err;
6023

6024
    qemuDriverLock(driver);
6025
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
6026

D
Daniel P. Berrange 已提交
6027
    if (!vm) {
6028 6029
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
6030 6031
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
6032
        goto cleanup;
D
Daniel P. Berrange 已提交
6033 6034
    }

6035
    /* Refresh current memory based on balloon info */
D
Daniel P. Berrange 已提交
6036
    if (virDomainObjIsActive(vm)) {
6037
        qemuDomainObjPrivatePtr priv = vm->privateData;
6038 6039 6040
        /* Don't delay if someone's using the monitor, just use
         * existing most recent data instead */
        if (!priv->jobActive) {
6041
            if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
6042 6043
                goto cleanup;

6044
            qemuDomainObjEnterMonitorWithDriver(driver, vm);
6045
            err = qemuMonitorGetBalloonInfo(priv->mon, &balloon);
6046
            qemuDomainObjExitMonitorWithDriver(driver, vm);
6047 6048 6049 6050
            if (qemuDomainObjEndJob(vm) == 0) {
                vm = NULL;
                goto cleanup;
            }
6051 6052 6053 6054 6055 6056
            if (err < 0)
                goto cleanup;
            if (err > 0)
                vm->def->memory = balloon;
            /* err == 0 indicates no balloon support, so ignore it */
        }
6057
    }
6058

6059
    ret = qemudVMDumpXML(driver, vm, flags);
6060 6061

cleanup:
6062 6063
    if (vm)
        virDomainObjUnlock(vm);
6064
    qemuDriverUnlock(driver);
6065
    return ret;
D
Daniel P. Berrange 已提交
6066 6067 6068
}


6069 6070 6071 6072
static char *qemuDomainXMLFromNative(virConnectPtr conn,
                                     const char *format,
                                     const char *config,
                                     unsigned int flags ATTRIBUTE_UNUSED) {
6073
    struct qemud_driver *driver = conn->privateData;
6074 6075 6076 6077
    virDomainDefPtr def = NULL;
    char *xml = NULL;

    if (STRNEQ(format, QEMU_CONFIG_FORMAT_ARGV)) {
6078 6079
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("unsupported config type %s"), format);
6080 6081 6082
        goto cleanup;
    }

6083
    qemuDriverLock(driver);
6084
    def = qemuParseCommandLineString(driver->caps, config);
6085
    qemuDriverUnlock(driver);
6086 6087 6088
    if (!def)
        goto cleanup;

6089
    xml = virDomainDefFormat(def, VIR_DOMAIN_XML_INACTIVE);
6090 6091 6092 6093 6094 6095

cleanup:
    virDomainDefFree(def);
    return xml;
}

6096 6097 6098 6099 6100 6101
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;
6102
    virDomainChrDef monConfig;
6103
    const char *emulator;
6104
    unsigned long long qemuCmdFlags;
6105 6106 6107 6108 6109 6110 6111 6112
    struct stat sb;
    const char **retargv = NULL;
    const char **retenv = NULL;
    const char **tmp;
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    char *ret = NULL;
    int i;

6113 6114
    qemuDriverLock(driver);

6115
    if (STRNEQ(format, QEMU_CONFIG_FORMAT_ARGV)) {
6116 6117
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("unsupported config type %s"), format);
6118 6119 6120
        goto cleanup;
    }

6121
    def = virDomainDefParseString(driver->caps, xmlData, 0);
6122 6123 6124 6125 6126 6127 6128 6129 6130 6131 6132 6133 6134 6135 6136 6137 6138 6139 6140 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
    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) {
6166
        virReportSystemError(errno,
6167 6168 6169 6170 6171 6172 6173 6174
                             _("Cannot find QEMU binary %s"),
                             emulator);
        goto cleanup;
    }

    if (qemudExtractVersionInfo(emulator,
                                NULL,
                                &qemuCmdFlags) < 0) {
6175 6176 6177
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Cannot determine QEMU argv syntax %s"),
                        emulator);
6178 6179 6180
        goto cleanup;
    }

6181
    if (qemuPrepareMonitorChr(driver, &monConfig, def->name) < 0)
6182
        goto cleanup;
6183 6184

    if (qemudBuildCommandLine(conn, driver, def,
D
Daniel P. Berrange 已提交
6185
                              &monConfig, 0, qemuCmdFlags,
6186 6187
                              &retargv, &retenv,
                              NULL, NULL, /* Don't want it to create TAP devices */
C
Chris Lalancette 已提交
6188
                              NULL, NULL) < 0) {
6189 6190 6191 6192 6193 6194 6195 6196 6197 6198 6199 6200 6201 6202 6203 6204
        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++;
    }

6205 6206
    if (virBufferError(&buf)) {
        virBufferFreeAndReset(&buf);
6207
        virReportOOMError();
6208
        goto cleanup;
6209
    }
6210 6211 6212 6213

    ret = virBufferContentAndReset(&buf);

cleanup:
6214
    qemuDriverUnlock(driver);
6215 6216 6217 6218 6219 6220 6221 6222 6223 6224 6225 6226 6227
    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;
}


6228
static int qemudListDefinedDomains(virConnectPtr conn,
6229
                            char **const names, int nnames) {
6230
    struct qemud_driver *driver = conn->privateData;
6231
    int n;
6232

6233
    qemuDriverLock(driver);
6234
    n = virDomainObjListGetInactiveNames(&driver->domains, names, nnames);
6235
    qemuDriverUnlock(driver);
6236
    return n;
D
Daniel P. Berrange 已提交
6237 6238
}

6239
static int qemudNumDefinedDomains(virConnectPtr conn) {
6240
    struct qemud_driver *driver = conn->privateData;
6241
    int n;
6242

6243
    qemuDriverLock(driver);
6244
    n = virDomainObjListNumOfDomains(&driver->domains, 0);
6245
    qemuDriverUnlock(driver);
6246

6247
    return n;
D
Daniel P. Berrange 已提交
6248 6249 6250
}


6251
static int qemudDomainStart(virDomainPtr dom) {
6252 6253 6254
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
6255
    virDomainEventPtr event = NULL;
6256
    char *managed_save = NULL;
6257

6258
    qemuDriverLock(driver);
6259
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
6260

6261
    if (!vm) {
6262 6263
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
6264 6265
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
6266
        goto cleanup;
6267 6268
    }

6269 6270 6271 6272
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

    if (virDomainObjIsActive(vm)) {
6273 6274
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is already running"));
6275 6276 6277
        goto endjob;
    }

6278 6279 6280 6281 6282 6283 6284 6285 6286 6287 6288 6289 6290 6291 6292 6293 6294 6295 6296 6297 6298 6299 6300 6301 6302 6303 6304 6305 6306
    /*
     * 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
         */
        if (qemuDomainObjEndJob(vm) == 0)
            vm = NULL;
        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);
    }

6307
    ret = qemudStartVMDaemon(dom->conn, driver, vm, NULL, -1);
6308
    if (ret != -1)
6309 6310 6311
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STARTED,
                                         VIR_DOMAIN_EVENT_STARTED_BOOTED);
6312

6313
endjob:
6314 6315
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
6316

6317
cleanup:
6318
    VIR_FREE(managed_save);
6319 6320
    if (vm)
        virDomainObjUnlock(vm);
6321
    if (event)
6322
        qemuDomainEventQueue(driver, event);
6323
    qemuDriverUnlock(driver);
6324
    return ret;
D
Daniel P. Berrange 已提交
6325 6326
}

6327 6328 6329 6330 6331 6332 6333 6334 6335 6336 6337 6338 6339 6340 6341
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;

6342
        if (STRNEQ(def->os.machine, machine->name))
6343 6344 6345
            continue;

        if (!(*canonical = strdup(machine->canonical))) {
6346
            virReportOOMError();
6347 6348 6349 6350 6351 6352 6353 6354 6355
            return -1;
        }

        break;
    }

    return 0;
}

6356 6357 6358 6359 6360 6361 6362
static int
qemudCanonicalizeMachineDirect(virDomainDefPtr def, char **canonical)
{
    virCapsGuestMachinePtr *machines = NULL;
    int i, nmachines = 0;

    if (qemudProbeMachineTypes(def->emulator, &machines, &nmachines) < 0) {
6363
        virReportOOMError();
6364 6365 6366 6367 6368 6369 6370
        return -1;
    }

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

6371
        if (STRNEQ(def->os.machine, machines[i]->name))
6372 6373 6374 6375 6376 6377 6378 6379 6380 6381 6382 6383
            continue;

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

    virCapabilitiesFreeMachines(machines, nmachines);

    return 0;
}

6384 6385
int
qemudCanonicalizeMachine(struct qemud_driver *driver, virDomainDefPtr def)
6386 6387 6388 6389 6390 6391
{
    char *canonical = NULL;
    int i;

    for (i = 0; i < driver->caps->nguests; i++) {
        virCapsGuestPtr guest = driver->caps->guests[i];
6392
        virCapsGuestDomainInfoPtr info;
6393 6394 6395
        int j;

        for (j = 0; j < guest->arch.ndomains; j++) {
6396
            info = &guest->arch.domains[j]->info;
6397

6398 6399 6400 6401 6402 6403 6404 6405 6406
            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;
6407 6408
        }

6409 6410 6411 6412
        info = &guest->arch.defaultInfo;

        if (info->emulator && STREQ(info->emulator, def->emulator)) {
            if (qemudCanonicalizeMachineFromInfo(def, info, &canonical) < 0)
6413 6414 6415 6416
                return -1;
            goto out;
        }
    }
6417 6418 6419 6420

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

6421 6422 6423 6424 6425 6426 6427
out:
    if (canonical) {
        VIR_FREE(def->os.machine);
        def->os.machine = canonical;
    }
    return 0;
}
D
Daniel P. Berrange 已提交
6428

6429
static virDomainPtr qemudDomainDefine(virConnectPtr conn, const char *xml) {
6430
    struct qemud_driver *driver = conn->privateData;
6431
    virDomainDefPtr def;
6432
    virDomainObjPtr vm = NULL;
6433
    virDomainPtr dom = NULL;
6434
    virDomainEventPtr event = NULL;
6435
    int dupVM;
6436

6437
    qemuDriverLock(driver);
6438
    if (!(def = virDomainDefParseString(driver->caps, xml,
6439
                                        VIR_DOMAIN_XML_INACTIVE)))
6440
        goto cleanup;
6441

6442
    if (virSecurityDriverVerify(def) < 0)
6443 6444
        goto cleanup;

6445 6446
    if ((dupVM = virDomainObjIsDuplicate(&driver->domains, def, 0)) < 0)
        goto cleanup;
6447

6448
    if (qemudCanonicalizeMachine(driver, def) < 0)
6449 6450
        goto cleanup;

6451 6452 6453
    if (qemuAssignPCIAddresses(def) < 0)
        goto cleanup;

6454
    if (!(vm = virDomainAssignDef(driver->caps,
6455
                                  &driver->domains,
6456
                                  def, false))) {
6457
        goto cleanup;
6458
    }
6459
    def = NULL;
6460
    vm->persistent = 1;
6461

6462
    if (virDomainSaveConfig(driver->configDir,
6463
                            vm->newDef ? vm->newDef : vm->def) < 0) {
6464 6465
        virDomainRemoveInactive(&driver->domains,
                                vm);
6466
        vm = NULL;
6467
        goto cleanup;
6468 6469
    }

6470 6471
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_DEFINED,
6472
                                     !dupVM ?
6473 6474
                                     VIR_DOMAIN_EVENT_DEFINED_ADDED :
                                     VIR_DOMAIN_EVENT_DEFINED_UPDATED);
6475

6476
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
6477
    if (dom) dom->id = vm->def->id;
6478 6479

cleanup:
6480
    virDomainDefFree(def);
6481 6482
    if (vm)
        virDomainObjUnlock(vm);
6483 6484
    if (event)
        qemuDomainEventQueue(driver, event);
6485
    qemuDriverUnlock(driver);
6486
    return dom;
D
Daniel P. Berrange 已提交
6487 6488
}

6489
static int qemudDomainUndefine(virDomainPtr dom) {
6490 6491
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
6492
    virDomainEventPtr event = NULL;
6493
    int ret = -1;
D
Daniel P. Berrange 已提交
6494

6495
    qemuDriverLock(driver);
6496
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
6497

D
Daniel P. Berrange 已提交
6498
    if (!vm) {
6499 6500
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
6501 6502
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
6503
        goto cleanup;
D
Daniel P. Berrange 已提交
6504 6505
    }

D
Daniel P. Berrange 已提交
6506
    if (virDomainObjIsActive(vm)) {
6507 6508
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot delete active domain"));
6509
        goto cleanup;
D
Daniel P. Berrange 已提交
6510 6511
    }

6512
    if (!vm->persistent) {
6513 6514
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("cannot undefine transient domain"));
6515
        goto cleanup;
6516 6517
    }

6518
    if (virDomainDeleteConfig(driver->configDir, driver->autostartDir, vm) < 0)
6519
        goto cleanup;
D
Daniel P. Berrange 已提交
6520

6521 6522 6523
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_UNDEFINED,
                                     VIR_DOMAIN_EVENT_UNDEFINED_REMOVED);
6524

6525 6526
    virDomainRemoveInactive(&driver->domains,
                            vm);
6527
    vm = NULL;
6528
    ret = 0;
D
Daniel P. Berrange 已提交
6529

6530
cleanup:
6531 6532
    if (vm)
        virDomainObjUnlock(vm);
6533 6534
    if (event)
        qemuDomainEventQueue(driver, event);
6535
    qemuDriverUnlock(driver);
6536
    return ret;
D
Daniel P. Berrange 已提交
6537 6538
}

6539

6540
static int qemudDomainChangeEjectableMedia(struct qemud_driver *driver,
6541
                                           virDomainObjPtr vm,
6542 6543
                                           virDomainDiskDefPtr disk,
                                           unsigned long long qemuCmdFlags)
6544
{
6545
    virDomainDiskDefPtr origdisk = NULL;
6546
    int i;
6547
    int ret;
6548
    char *driveAlias = NULL;
6549

6550
    origdisk = NULL;
6551
    for (i = 0 ; i < vm->def->ndisks ; i++) {
6552 6553
        if (vm->def->disks[i]->bus == disk->bus &&
            STREQ(vm->def->disks[i]->dst, disk->dst)) {
6554
            origdisk = vm->def->disks[i];
6555
            break;
6556
        }
6557 6558 6559
    }

    if (!origdisk) {
6560 6561 6562 6563
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("No device with bus '%s' and target '%s'"),
                        virDomainDiskBusTypeToString(disk->bus),
                        disk->dst);
6564 6565 6566
        return -1;
    }

6567
    if (!origdisk->info.alias) {
6568 6569
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("missing disk device alias name for %s"), origdisk->dst);
6570 6571
        return -1;
    }
6572

6573 6574
    if (origdisk->device != VIR_DOMAIN_DISK_DEVICE_FLOPPY &&
        origdisk->device != VIR_DOMAIN_DISK_DEVICE_CDROM) {
6575 6576 6577
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Removable media not supported for %s device"),
                        virDomainDiskDeviceTypeToString(disk->device));
6578
        return -1;
6579
    }
6580

6581 6582
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityImageLabel &&
6583
        driver->securityDriver->domainSetSecurityImageLabel(vm, disk) < 0)
6584 6585
        return -1;

6586 6587 6588
    if (!(driveAlias = qemuDeviceDriveHostAlias(origdisk, qemuCmdFlags)))
        goto error;

6589
    qemuDomainObjPrivatePtr priv = vm->privateData;
6590
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
6591
    if (disk->src) {
6592
        const char *format = NULL;
6593 6594 6595
        if (disk->type != VIR_DOMAIN_DISK_TYPE_DIR) {
            if (disk->driverType)
                format = disk->driverType;
6596 6597 6598
            else if (origdisk->driverType)
                format = origdisk->driverType;
        }
6599
        ret = qemuMonitorChangeMedia(priv->mon,
6600
                                     driveAlias,
6601
                                     disk->src, format);
6602
    } else {
6603
        ret = qemuMonitorEjectMedia(priv->mon, driveAlias);
6604
    }
6605
    qemuDomainObjExitMonitorWithDriver(driver, vm);
6606

6607 6608 6609 6610 6611
    if (ret < 0)
        goto error;

    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
6612
        driver->securityDriver->domainRestoreSecurityImageLabel(vm, origdisk) < 0)
6613 6614 6615
        VIR_WARN("Unable to restore security label on ejected image %s", origdisk->src);

    VIR_FREE(origdisk->src);
6616 6617 6618 6619
    origdisk->src = disk->src;
    disk->src = NULL;
    origdisk->type = disk->type;

6620 6621
    VIR_FREE(driveAlias);

6622
    virDomainDiskDefFree(disk);
6623

6624
    return ret;
6625 6626

error:
6627
    VIR_FREE(driveAlias);
6628 6629
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
6630
        driver->securityDriver->domainRestoreSecurityImageLabel(vm, disk) < 0)
6631
        VIR_WARN("Unable to restore security label on new media %s", disk->src);
6632
    return -1;
6633 6634
}

6635

6636
static int qemudDomainAttachPciDiskDevice(struct qemud_driver *driver,
6637
                                          virDomainObjPtr vm,
6638 6639
                                          virDomainDiskDefPtr disk,
                                          int qemuCmdFlags)
6640
{
6641
    int i, ret;
6642
    const char* type = virDomainDiskBusTypeToString(disk->bus);
6643
    qemuDomainObjPrivatePtr priv = vm->privateData;
6644 6645
    char *devstr = NULL;
    char *drivestr = NULL;
6646 6647

    for (i = 0 ; i < vm->def->ndisks ; i++) {
6648
        if (STREQ(vm->def->disks[i]->dst, disk->dst)) {
6649 6650
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("target %s already exists"), disk->dst);
6651 6652 6653 6654
            return -1;
        }
    }

6655 6656
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityImageLabel &&
6657
        driver->securityDriver->domainSetSecurityImageLabel(vm, disk) < 0)
6658 6659
        return -1;

6660
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
6661 6662
        if (qemuDomainPCIAddressEnsureAddr(priv->pciaddrs, &disk->info) < 0)
            goto error;
6663 6664
        if (qemuAssignDeviceDiskAlias(disk, qemuCmdFlags) < 0)
            goto error;
6665

6666 6667 6668
        if (!(drivestr = qemuBuildDriveStr(disk, 0, qemuCmdFlags)))
            goto error;

6669
        if (!(devstr = qemuBuildDriveDevStr(disk)))
6670 6671 6672
            goto error;
    }

6673
    if (VIR_REALLOC_N(vm->def->disks, vm->def->ndisks+1) < 0) {
6674
        virReportOOMError();
6675
        goto error;
6676 6677
    }

6678
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
6679 6680 6681 6682 6683 6684 6685 6686 6687 6688 6689 6690 6691 6692 6693 6694
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        ret = qemuMonitorAddDrive(priv->mon, drivestr);
        if (ret == 0)
            qemuMonitorAddDevice(priv->mon, devstr);
            /* XXX remove the drive upon fail */
    } 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));
        }
    }
6695
    qemuDomainObjExitMonitorWithDriver(driver, vm);
6696

6697 6698
    if (ret < 0)
        goto error;
6699

6700
    virDomainDiskInsertPreAlloced(vm->def, disk);
6701

6702 6703 6704
    VIR_FREE(devstr);
    VIR_FREE(drivestr);

6705 6706 6707
    return 0;

error:
6708 6709 6710
    VIR_FREE(devstr);
    VIR_FREE(drivestr);

6711 6712 6713 6714 6715
    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);

6716 6717
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
6718
        driver->securityDriver->domainRestoreSecurityImageLabel(vm, disk) < 0)
6719
        VIR_WARN("Unable to restore security label on %s", disk->src);
6720 6721

    return -1;
6722
}
6723

6724

6725
static int qemudDomainAttachPciControllerDevice(struct qemud_driver *driver,
6726
                                                virDomainObjPtr vm,
6727 6728
                                                virDomainControllerDefPtr controller,
                                                int qemuCmdFlags)
6729
{
6730 6731
    int i;
    int ret = -1;
6732
    const char* type = virDomainControllerTypeToString(controller->type);
6733
    char *devstr = NULL;
6734 6735 6736
    qemuDomainObjPrivatePtr priv = vm->privateData;

    for (i = 0 ; i < vm->def->ncontrollers ; i++) {
6737 6738
        if ((vm->def->controllers[i]->type == controller->type) &&
            (vm->def->controllers[i]->idx == controller->idx)) {
6739 6740 6741
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("target %s:%d already exists"),
                            type, controller->idx);
6742 6743 6744 6745
            return -1;
        }
    }

6746 6747 6748 6749 6750 6751
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuDomainPCIAddressEnsureAddr(priv->pciaddrs, &controller->info) < 0)
            goto cleanup;
        if (qemuAssignDeviceControllerAlias(controller) < 0)
            goto cleanup;
    }
6752

6753
    if (!(devstr = qemuBuildControllerDevStr(controller))) {
6754
        virReportOOMError();
6755 6756 6757
        goto cleanup;
    }

6758
    if (VIR_REALLOC_N(vm->def->controllers, vm->def->ncontrollers+1) < 0) {
6759
        virReportOOMError();
6760
        goto cleanup;
6761 6762 6763
    }

    qemuDomainObjEnterMonitorWithDriver(driver, vm);
6764 6765 6766 6767 6768 6769 6770
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        ret = qemuMonitorAddDevice(priv->mon, devstr);
    } else {
        ret = qemuMonitorAttachPCIDiskController(priv->mon,
                                                 type,
                                                 &controller->info.addr.pci);
    }
6771 6772 6773
    qemuDomainObjExitMonitorWithDriver(driver, vm);

    if (ret == 0) {
6774 6775
        controller->info.type = VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI;
        virDomainControllerInsertPreAlloced(vm->def, controller);
6776 6777
    }

6778
cleanup:
6779 6780 6781 6782 6783 6784
    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");

6785
    VIR_FREE(devstr);
6786 6787 6788
    return ret;
}

6789

6790
static virDomainControllerDefPtr
6791
qemuDomainFindOrCreateSCSIDiskController(struct qemud_driver *driver,
6792
                                         virDomainObjPtr vm,
6793 6794
                                         int controller,
                                         int qemuCmdFlags)
6795 6796 6797 6798 6799 6800 6801 6802 6803 6804 6805 6806 6807
{
    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
6808
    /* No SCSI controller present, for backward compatibility we
6809 6810
     * now hotplug a controller */
    if (VIR_ALLOC(cont) < 0) {
6811
        virReportOOMError();
6812 6813 6814 6815 6816 6817
        return NULL;
    }
    cont->type = VIR_DOMAIN_CONTROLLER_TYPE_SCSI;
    cont->idx = 0;

    VIR_INFO0("No SCSI controller present, hotplugging one");
6818
    if (qemudDomainAttachPciControllerDevice(driver,
6819
                                             vm, cont, qemuCmdFlags) < 0) {
6820 6821 6822 6823 6824 6825
        VIR_FREE(cont);
        return NULL;
    }
    return cont;
}

6826

6827
static int qemudDomainAttachSCSIDisk(struct qemud_driver *driver,
6828
                                     virDomainObjPtr vm,
6829
                                     virDomainDiskDefPtr disk,
6830 6831 6832 6833
                                     int qemuCmdFlags)
{
    int i;
    qemuDomainObjPrivatePtr priv = vm->privateData;
6834
    virDomainControllerDefPtr cont = NULL;
6835
    char *drivestr = NULL;
6836
    char *devstr = NULL;
6837 6838 6839
    int ret = -1;

    for (i = 0 ; i < vm->def->ndisks ; i++) {
6840
        if (STREQ(vm->def->disks[i]->dst, disk->dst)) {
6841 6842
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("target %s already exists"), disk->dst);
6843
            return -1;
6844 6845 6846
        }
    }

6847 6848 6849

    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityImageLabel &&
6850
        driver->securityDriver->domainSetSecurityImageLabel(vm, disk) < 0)
6851 6852
        return -1;

6853
    /* We should have an address already, so make sure */
6854
    if (disk->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_DRIVE) {
6855 6856 6857
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("unexpected disk address type %s"),
                        virDomainDeviceAddressTypeToString(disk->info.type));
6858
        goto error;
6859 6860
    }

6861 6862 6863
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuAssignDeviceDiskAlias(disk, qemuCmdFlags) < 0)
            goto error;
6864
        if (!(devstr = qemuBuildDriveDevStr(disk)))
6865 6866 6867 6868 6869 6870
            goto error;
    }

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

6871
    for (i = 0 ; i <= disk->info.addr.drive.controller ; i++) {
6872
        cont = qemuDomainFindOrCreateSCSIDiskController(driver, vm, i, qemuCmdFlags);
6873
        if (!cont)
6874
            goto error;
6875 6876
    }

6877 6878 6879 6880 6881
    /* 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);

6882
    if (cont->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI) {
6883 6884
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("SCSI controller %d was missing its PCI address"), cont->idx);
6885
        goto error;
6886 6887 6888
    }

    if (VIR_REALLOC_N(vm->def->disks, vm->def->ndisks+1) < 0) {
6889
        virReportOOMError();
6890
        goto error;
6891 6892 6893
    }

    qemuDomainObjEnterMonitorWithDriver(driver, vm);
6894 6895 6896 6897 6898 6899 6900 6901 6902 6903 6904 6905 6906 6907 6908 6909 6910 6911 6912 6913
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        ret = qemuMonitorAddDrive(priv->mon,
                                  drivestr);
        if (ret == 0)
            ret = qemuMonitorAddDevice(priv->mon,
                                       devstr);
            /* XXX should call 'drive_del' on error but this does not exist yet */
    } 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));
        }
    }
6914 6915
    qemuDomainObjExitMonitorWithDriver(driver, vm);

6916 6917
    if (ret < 0)
        goto error;
6918

6919
    virDomainDiskInsertPreAlloced(vm->def, disk);
6920 6921

    VIR_FREE(devstr);
6922
    VIR_FREE(drivestr);
6923 6924 6925 6926

    return 0;

error:
6927
    VIR_FREE(devstr);
6928
    VIR_FREE(drivestr);
6929

6930 6931
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
6932
        driver->securityDriver->domainRestoreSecurityImageLabel(vm, disk) < 0)
6933
        VIR_WARN("Unable to restore security label on %s", disk->src);
6934 6935

    return -1;
6936 6937
}

6938

6939
static int qemudDomainAttachUsbMassstorageDevice(struct qemud_driver *driver,
6940
                                                 virDomainObjPtr vm,
6941 6942
                                                 virDomainDiskDefPtr disk,
                                                 int qemuCmdFlags)
6943
{
6944
    qemuDomainObjPrivatePtr priv = vm->privateData;
6945
    int i, ret;
6946 6947
    char *drivestr = NULL;
    char *devstr = NULL;
6948

6949
    for (i = 0 ; i < vm->def->ndisks ; i++) {
6950
        if (STREQ(vm->def->disks[i]->dst, disk->dst)) {
6951 6952
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("target %s already exists"), disk->dst);
6953 6954 6955 6956
            return -1;
        }
    }

6957 6958
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityImageLabel &&
6959
        driver->securityDriver->domainSetSecurityImageLabel(vm, disk) < 0)
6960 6961
        return -1;

6962
    if (!disk->src) {
6963 6964
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("disk source path is missing"));
6965
        goto error;
6966 6967
    }

6968
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
6969 6970
        if (qemuAssignDeviceDiskAlias(disk, qemuCmdFlags) < 0)
            goto error;
6971 6972
        if (!(drivestr = qemuBuildDriveStr(disk, 0, qemuCmdFlags)))
            goto error;
6973
        if (!(devstr = qemuBuildDriveDevStr(disk)))
6974 6975 6976
            goto error;
    }

6977
    if (VIR_REALLOC_N(vm->def->disks, vm->def->ndisks+1) < 0) {
6978
        virReportOOMError();
6979
        goto error;
6980 6981
    }

6982
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
6983 6984 6985 6986 6987 6988 6989 6990 6991 6992
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        ret = qemuMonitorAddDrive(priv->mon,
                                  drivestr);
        if (ret == 0)
            ret = qemuMonitorAddDevice(priv->mon,
                                       devstr);
            /* XXX should call 'drive_del' on error but this does not exist yet */
    } else {
        ret = qemuMonitorAddUSBDisk(priv->mon, disk->src);
    }
6993
    qemuDomainObjExitMonitorWithDriver(driver, vm);
6994

6995 6996 6997
    if (ret < 0)
        goto error;

6998
    virDomainDiskInsertPreAlloced(vm->def, disk);
6999

7000 7001 7002
    VIR_FREE(devstr);
    VIR_FREE(drivestr);

7003 7004 7005
    return 0;

error:
7006 7007 7008
    VIR_FREE(devstr);
    VIR_FREE(drivestr);

7009 7010
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
7011
        driver->securityDriver->domainRestoreSecurityImageLabel(vm, disk) < 0)
7012
        VIR_WARN("Unable to restore security label on %s", disk->src);
7013 7014

    return -1;
7015 7016
}

7017

7018
/* XXX conn required for network -> bridge resolution */
M
Mark McLoughlin 已提交
7019
static int qemudDomainAttachNetDevice(virConnectPtr conn,
7020
                                      struct qemud_driver *driver,
M
Mark McLoughlin 已提交
7021
                                      virDomainObjPtr vm,
7022
                                      virDomainNetDefPtr net,
7023
                                      unsigned long long qemuCmdFlags)
M
Mark McLoughlin 已提交
7024
{
7025
    qemuDomainObjPrivatePtr priv = vm->privateData;
7026
    char *tapfd_name = NULL;
7027
    int tapfd = -1;
7028
    char *nicstr = NULL;
7029
    char *netstr = NULL;
7030
    int ret = -1;
7031
    virDomainDevicePCIAddress guestAddr;
7032
    int vlan;
M
Mark McLoughlin 已提交
7033 7034

    if (!(qemuCmdFlags & QEMUD_CMD_FLAG_HOST_NET_ADD)) {
7035 7036
        qemuReportError(VIR_ERR_NO_SUPPORT, "%s",
                        _("installed qemu version does not support host_net_add"));
M
Mark McLoughlin 已提交
7037 7038 7039 7040 7041
        return -1;
    }

    if (net->type == VIR_DOMAIN_NET_TYPE_BRIDGE ||
        net->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
7042
        if (priv->monConfig->type != VIR_DOMAIN_CHR_TYPE_UNIX) {
7043 7044 7045 7046
            qemuReportError(VIR_ERR_NO_SUPPORT,
                            _("network device type '%s' cannot be attached: "
                              "qemu is not using a unix socket monitor"),
                            virDomainNetTypeToString(net->type));
7047 7048 7049 7050 7051
            return -1;
        }

        if ((tapfd = qemudNetworkIfaceConnect(conn, driver, net, qemuCmdFlags)) < 0)
            return -1;
7052 7053 7054 7055 7056 7057 7058 7059 7060
    } 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
7061
        if ((tapfd = qemudPhysIfaceConnect(conn, driver, net,
7062
                                           net->data.direct.linkdev,
7063 7064
                                           net->data.direct.mode,
                                           qemuCmdFlags)) < 0)
7065
            return -1;
M
Mark McLoughlin 已提交
7066 7067
    }

7068 7069
    if (VIR_REALLOC_N(vm->def->nets, vm->def->nnets+1) < 0)
        goto no_memory;
M
Mark McLoughlin 已提交
7070

7071 7072 7073 7074 7075
    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
7076

7077 7078 7079 7080
    if ((qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) &&
        qemuDomainPCIAddressEnsureAddr(priv->pciaddrs, &net->info) < 0)
        goto cleanup;

7081
    vlan = qemuDomainNetVLAN(net);
M
Mark McLoughlin 已提交
7082

7083
    if (vlan < 0) {
7084 7085
        qemuReportError(VIR_ERR_NO_SUPPORT, "%s",
                        _("Unable to attach network devices without vlan"));
7086 7087 7088
        goto cleanup;
    }

7089
    if (tapfd != -1) {
7090
        if (virAsprintf(&tapfd_name, "fd-%s", net->info.alias) < 0)
7091 7092
            goto no_memory;

7093
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
7094
        if (qemuMonitorSendFileHandle(priv->mon, tapfd_name, tapfd) < 0) {
7095
            qemuDomainObjExitMonitorWithDriver(driver, vm);
7096
            goto cleanup;
7097
        }
7098
        qemuDomainObjExitMonitorWithDriver(driver, vm);
7099 7100
    }

7101
    /* FIXME - need to support vhost-net here (5th arg) */
7102
    if (!(netstr = qemuBuildHostNetStr(net, ' ',
7103
                                       vlan, tapfd_name, 0)))
7104
        goto try_tapfd_close;
M
Mark McLoughlin 已提交
7105

7106
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7107
    if (qemuMonitorAddHostNetwork(priv->mon, netstr) < 0) {
7108
        qemuDomainObjExitMonitorWithDriver(driver, vm);
7109
        goto try_tapfd_close;
7110
    }
7111
    qemuDomainObjExitMonitorWithDriver(driver, vm);
M
Mark McLoughlin 已提交
7112

7113 7114 7115
    if (tapfd != -1)
        close(tapfd);
    tapfd = -1;
M
Mark McLoughlin 已提交
7116

7117 7118 7119 7120 7121 7122 7123
    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
7124

7125
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7126 7127 7128 7129 7130 7131 7132 7133 7134 7135 7136 7137 7138
    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));
7139
    }
7140
    qemuDomainObjExitMonitorWithDriver(driver, vm);
7141

7142
    ret = 0;
M
Mark McLoughlin 已提交
7143 7144 7145

    vm->def->nets[vm->def->nnets++] = net;

7146
cleanup:
7147 7148 7149 7150 7151 7152
    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");

7153 7154
    if (ret != 0)
        virNWFilterTearNWFilter(net);
S
Stefan Berger 已提交
7155

7156 7157 7158 7159 7160
    VIR_FREE(nicstr);
    VIR_FREE(netstr);
    VIR_FREE(tapfd_name);
    if (tapfd != -1)
        close(tapfd);
7161

7162
    return ret;
7163

7164
try_remove:
7165
    if (vlan < 0) {
7166
        VIR_WARN0(_("Unable to remove network backend"));
7167 7168 7169 7170
    } else {
        char *hostnet_name;
        if (virAsprintf(&hostnet_name, "host%s", net->info.alias) < 0)
            goto no_memory;
7171
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
7172
        if (qemuMonitorRemoveHostNetwork(priv->mon, vlan, hostnet_name) < 0)
7173
            VIR_WARN(_("Failed to remove network backend for vlan %d, net %s"),
7174
                     vlan, hostnet_name);
7175
        qemuDomainObjExitMonitorWithDriver(driver, vm);
7176
        VIR_FREE(hostnet_name);
7177
    }
7178
    goto cleanup;
7179

7180
try_tapfd_close:
7181
    if (tapfd_name) {
7182
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
7183
        if (qemuMonitorCloseFileHandle(priv->mon, tapfd_name) < 0)
7184
            VIR_WARN(_("Failed to close tapfd with '%s'"), tapfd_name);
7185
        qemuDomainObjExitMonitorWithDriver(driver, vm);
7186
    }
7187

7188 7189
    goto cleanup;

7190
no_memory:
7191
    virReportOOMError();
7192
    goto cleanup;
M
Mark McLoughlin 已提交
7193 7194
}

7195

7196
static int qemudDomainAttachHostPciDevice(struct qemud_driver *driver,
7197
                                          virDomainObjPtr vm,
7198 7199
                                          virDomainHostdevDefPtr hostdev,
                                          int qemuCmdFlags)
7200
{
7201
    qemuDomainObjPrivatePtr priv = vm->privateData;
7202
    pciDevice *pci;
7203
    int ret;
7204
    virDomainDevicePCIAddress guestAddr;
7205
    char *devstr = NULL;
7206 7207

    if (VIR_REALLOC_N(vm->def->hostdevs, vm->def->nhostdevs+1) < 0) {
7208
        virReportOOMError();
7209 7210 7211
        return -1;
    }

7212
    pci = pciGetDevice(hostdev->source.subsys.u.pci.domain,
7213 7214 7215
                       hostdev->source.subsys.u.pci.bus,
                       hostdev->source.subsys.u.pci.slot,
                       hostdev->source.subsys.u.pci.function);
7216
    if (!pci)
7217
        return -1;
7218

7219 7220 7221 7222
    if (!pciDeviceIsAssignable(pci, !driver->relaxedACS) ||
        (hostdev->managed && pciDettachDevice(pci) < 0) ||
        pciResetDevice(pci, driver->activePciHostdevs) < 0) {
        pciFreeDevice(pci);
7223 7224 7225
        return -1;
    }

7226 7227
    if (pciDeviceListAdd(driver->activePciHostdevs, pci) < 0) {
        pciFreeDevice(pci);
7228
        return -1;
7229 7230
    }

7231
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
7232 7233
        if (qemuAssignDeviceHostdevAlias(vm->def, hostdev, -1) < 0)
            goto error;
7234 7235 7236 7237 7238 7239
        if (qemuDomainPCIAddressEnsureAddr(priv->pciaddrs, &hostdev->info) < 0)
            goto error;

        if (!(devstr = qemuBuildPCIHostdevDevStr(hostdev)))
            goto error;
    }
7240

7241
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7242 7243 7244 7245 7246 7247
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)
        ret = qemuMonitorAddDevice(priv->mon, devstr);
    else
        ret = qemuMonitorAddPCIHostDevice(priv->mon,
                                          &hostdev->source.subsys.u.pci,
                                          &guestAddr);
7248
    qemuDomainObjExitMonitorWithDriver(driver, vm);
7249
    if (ret < 0)
7250
        goto error;
7251
    hostdev->info.type = VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI;
7252
    memcpy(&hostdev->info.addr.pci, &guestAddr, sizeof(guestAddr));
7253 7254 7255

    vm->def->hostdevs[vm->def->nhostdevs++] = hostdev;

7256 7257
    VIR_FREE(devstr);

7258
    return 0;
7259 7260

error:
7261 7262 7263 7264 7265
    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");

7266
    VIR_FREE(devstr);
7267
    pciDeviceListDel(driver->activePciHostdevs, pci);
7268 7269

    return -1;
7270 7271
}

7272

7273
static int qemudDomainAttachHostUsbDevice(struct qemud_driver *driver,
M
Mark McLoughlin 已提交
7274
                                          virDomainObjPtr vm,
7275 7276
                                          virDomainHostdevDefPtr hostdev,
                                          int qemuCmdFlags)
7277 7278
{
    int ret;
7279
    qemuDomainObjPrivatePtr priv = vm->privateData;
7280 7281
    char *devstr = NULL;

7282 7283 7284 7285 7286 7287
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuAssignDeviceHostdevAlias(vm->def, hostdev, -1) < 0)
            goto error;
        if (!(devstr = qemuBuildUSBHostdevDevStr(hostdev)))
            goto error;
    }
7288

7289
    if (VIR_REALLOC_N(vm->def->hostdevs, vm->def->nhostdevs+1) < 0) {
7290
        virReportOOMError();
7291
        goto error;
7292
    }
7293

7294
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7295 7296 7297
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)
        ret = qemuMonitorAddDevice(priv->mon, devstr);
    else
7298
        ret = qemuMonitorAddUSBDeviceExact(priv->mon,
7299 7300
                                           hostdev->source.subsys.u.usb.bus,
                                           hostdev->source.subsys.u.usb.device);
7301
    qemuDomainObjExitMonitorWithDriver(driver, vm);
7302 7303 7304 7305
    if (ret < 0)
        goto error;

    vm->def->hostdevs[vm->def->nhostdevs++] = hostdev;
7306

7307
    VIR_FREE(devstr);
7308

7309 7310 7311 7312 7313
    return 0;

error:
    VIR_FREE(devstr);
    return -1;
7314 7315
}

7316

7317
static int qemudDomainAttachHostDevice(struct qemud_driver *driver,
M
Mark McLoughlin 已提交
7318
                                       virDomainObjPtr vm,
7319 7320
                                       virDomainHostdevDefPtr hostdev,
                                       int qemuCmdFlags)
M
Mark McLoughlin 已提交
7321 7322
{
    if (hostdev->mode != VIR_DOMAIN_HOSTDEV_MODE_SUBSYS) {
7323 7324 7325
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        _("hostdev mode '%s' not supported"),
                        virDomainHostdevModeTypeToString(hostdev->mode));
M
Mark McLoughlin 已提交
7326 7327 7328
        return -1;
    }

7329 7330 7331 7332 7333 7334 7335 7336 7337 7338 7339 7340 7341 7342 7343 7344 7345
    /* 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);
    }


7346
    if (driver->securityDriver &&
7347
        driver->securityDriver->domainSetSecurityHostdevLabel &&
7348
        driver->securityDriver->domainSetSecurityHostdevLabel(vm, hostdev) < 0)
7349
        return -1;
M
Mark McLoughlin 已提交
7350 7351

    switch (hostdev->source.subsys.type) {
7352
    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI:
7353
        if (qemudDomainAttachHostPciDevice(driver, vm,
7354
                                           hostdev, qemuCmdFlags) < 0)
7355 7356 7357
            goto error;
        break;

M
Mark McLoughlin 已提交
7358
    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB:
7359
        if (qemudDomainAttachHostUsbDevice(driver, vm,
7360
                                           hostdev, qemuCmdFlags) < 0)
7361 7362 7363
            goto error;
        break;

M
Mark McLoughlin 已提交
7364
    default:
7365 7366 7367
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        _("hostdev subsys type '%s' not supported"),
                        virDomainHostdevSubsysTypeToString(hostdev->source.subsys.type));
7368
        goto error;
M
Mark McLoughlin 已提交
7369
    }
7370 7371 7372 7373 7374 7375

    return 0;

error:
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityHostdevLabel &&
7376
        driver->securityDriver->domainRestoreSecurityHostdevLabel(vm, hostdev) < 0)
7377 7378 7379
        VIR_WARN0("Unable to restore host device labelling on hotplug fail");

    return -1;
M
Mark McLoughlin 已提交
7380 7381
}

7382

7383
static int qemudDomainAttachDevice(virDomainPtr dom,
7384 7385
                                   const char *xml)
{
7386 7387 7388
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    virDomainDeviceDefPtr dev = NULL;
7389
    unsigned long long qemuCmdFlags;
7390
    virCgroupPtr cgroup = NULL;
7391
    int ret = -1;
7392

7393
    qemuDriverLock(driver);
7394
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
7395
    if (!vm) {
7396 7397
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
7398 7399
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
7400
        goto cleanup;
7401 7402
    }

7403 7404 7405
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
7406
    if (!virDomainObjIsActive(vm)) {
7407 7408
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot attach device on inactive domain"));
7409
        goto endjob;
7410 7411
    }

7412
    dev = virDomainDeviceDefParse(driver->caps, vm->def, xml,
7413
                                  VIR_DOMAIN_XML_INACTIVE);
7414
    if (dev == NULL)
7415
        goto endjob;
7416

7417 7418 7419
    if (qemudExtractVersionInfo(vm->def->emulator,
                                NULL,
                                &qemuCmdFlags) < 0)
7420
        goto endjob;
7421

7422
    if (dev->type == VIR_DOMAIN_DEVICE_DISK) {
7423
        if (qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_DEVICES)) {
7424
            if (virCgroupForDomain(driver->cgroup, vm->def->name, &cgroup, 0) !=0 ) {
7425 7426 7427
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("Unable to find cgroup for %s\n"),
                                vm->def->name);
7428
                goto endjob;
7429 7430 7431 7432 7433
            }
            if (dev->data.disk->src != NULL &&
                dev->data.disk->type == VIR_DOMAIN_DISK_TYPE_BLOCK &&
                virCgroupAllowDevicePath(cgroup,
                                         dev->data.disk->src) < 0) {
7434 7435 7436
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("unable to allow device %s"),
                                dev->data.disk->src);
7437
                goto endjob;
7438 7439 7440
            }
        }

7441
        switch (dev->data.disk->device) {
7442 7443
        case VIR_DOMAIN_DISK_DEVICE_CDROM:
        case VIR_DOMAIN_DISK_DEVICE_FLOPPY:
7444 7445 7446
            ret = qemudDomainChangeEjectableMedia(driver, vm,
                                                  dev->data.disk,
                                                  qemuCmdFlags);
7447 7448
            if (ret == 0)
                dev->data.disk = NULL;
7449
            break;
7450

7451 7452
        case VIR_DOMAIN_DISK_DEVICE_DISK:
            if (dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_USB) {
7453
                ret = qemudDomainAttachUsbMassstorageDevice(driver, vm,
7454
                                                            dev->data.disk, qemuCmdFlags);
7455 7456
                if (ret == 0)
                    dev->data.disk = NULL;
7457
            } else if (dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_VIRTIO) {
7458
                ret = qemudDomainAttachPciDiskDevice(driver, vm,
7459
                                                     dev->data.disk, qemuCmdFlags);
7460 7461
                if (ret == 0)
                    dev->data.disk = NULL;
7462
            } else if (dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_SCSI) {
7463
                ret = qemudDomainAttachSCSIDisk(driver, vm,
7464
                                                dev->data.disk, qemuCmdFlags);
7465 7466
                if (ret == 0)
                    dev->data.disk = NULL;
7467
            } else {
7468 7469 7470
                qemuReportError(VIR_ERR_NO_SUPPORT,
                                _("disk bus '%s' cannot be hotplugged."),
                                virDomainDiskBusTypeToString(dev->data.disk->bus));
7471
                /* fallthrough */
7472 7473
            }
            break;
7474

7475
        default:
7476 7477 7478
            qemuReportError(VIR_ERR_NO_SUPPORT,
                            _("disk device type '%s' cannot be hotplugged"),
                            virDomainDiskDeviceTypeToString(dev->data.disk->device));
7479 7480
            /* Fallthrough */
        }
7481
        if (ret != 0 && cgroup) {
7482 7483
            virCgroupDenyDevicePath(cgroup,
                                    dev->data.disk->src);
7484
        }
7485 7486
    } else if (dev->type == VIR_DOMAIN_DEVICE_CONTROLLER) {
        if (dev->data.controller->type == VIR_DOMAIN_CONTROLLER_TYPE_SCSI) {
7487
            ret = qemudDomainAttachPciControllerDevice(driver, vm,
7488
                                                       dev->data.controller, qemuCmdFlags);
7489 7490
            if (ret == 0)
                dev->data.controller = NULL;
7491
        } else {
7492 7493 7494
            qemuReportError(VIR_ERR_NO_SUPPORT,
                            _("disk controller bus '%s' cannot be hotplugged."),
                            virDomainControllerTypeToString(dev->data.controller->type));
7495 7496
            /* fallthrough */
        }
M
Mark McLoughlin 已提交
7497
    } else if (dev->type == VIR_DOMAIN_DEVICE_NET) {
7498 7499
        ret = qemudDomainAttachNetDevice(dom->conn, driver, vm,
                                         dev->data.net, qemuCmdFlags);
7500 7501
        if (ret == 0)
            dev->data.net = NULL;
M
Mark McLoughlin 已提交
7502
    } else if (dev->type == VIR_DOMAIN_DEVICE_HOSTDEV) {
7503
        ret = qemudDomainAttachHostDevice(driver, vm,
7504
                                          dev->data.hostdev, qemuCmdFlags);
7505 7506
        if (ret == 0)
            dev->data.hostdev = NULL;
7507
    } else {
7508 7509 7510
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        _("device type '%s' cannot be attached"),
                        virDomainDeviceTypeToString(dev->type));
7511
        goto endjob;
7512 7513
    }

7514
    if (!ret && virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
7515 7516
        ret = -1;

7517
endjob:
7518 7519
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
7520

7521
cleanup:
7522 7523 7524
    if (cgroup)
        virCgroupFree(&cgroup);

7525
    virDomainDeviceDefFree(dev);
7526 7527
    if (vm)
        virDomainObjUnlock(vm);
7528
    qemuDriverUnlock(driver);
7529 7530 7531
    return ret;
}

7532 7533 7534 7535
static int qemudDomainAttachDeviceFlags(virDomainPtr dom,
                                        const char *xml,
                                        unsigned int flags) {
    if (flags & VIR_DOMAIN_DEVICE_MODIFY_CONFIG) {
7536 7537
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot modify the persistent configuration of a domain"));
7538 7539 7540 7541 7542 7543
        return -1;
    }

    return qemudDomainAttachDevice(dom, xml);
}

7544

7545 7546 7547 7548 7549 7550 7551 7552 7553 7554 7555 7556 7557 7558 7559 7560 7561 7562 7563 7564 7565 7566 7567 7568 7569 7570 7571 7572 7573 7574 7575 7576 7577 7578 7579 7580 7581 7582 7583 7584 7585 7586 7587 7588 7589 7590 7591 7592 7593 7594 7595 7596 7597 7598 7599 7600 7601 7602 7603 7604 7605 7606 7607 7608 7609 7610 7611 7612 7613 7614 7615 7616 7617 7618 7619 7620 7621
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;
}


7622 7623 7624 7625 7626 7627 7628 7629 7630 7631 7632
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;

7633 7634 7635 7636
    virCheckFlags(VIR_DOMAIN_DEVICE_MODIFY_CURRENT |
                  VIR_DOMAIN_DEVICE_MODIFY_LIVE |
                  VIR_DOMAIN_DEVICE_MODIFY_CONFIG, -1);

7637 7638 7639 7640 7641 7642 7643 7644 7645 7646 7647 7648 7649 7650 7651 7652 7653 7654 7655 7656 7657 7658 7659 7660 7661 7662 7663 7664 7665 7666 7667 7668 7669 7670 7671 7672 7673 7674 7675 7676 7677 7678 7679 7680 7681 7682 7683 7684 7685 7686 7687 7688 7689 7690 7691 7692 7693 7694
    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;
            }
            if (dev->data.disk->src != NULL &&
                dev->data.disk->type == VIR_DOMAIN_DISK_TYPE_BLOCK &&
                virCgroupAllowDevicePath(cgroup,
                                         dev->data.disk->src) < 0) {
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("unable to allow device %s"),
                                dev->data.disk->src);
                goto endjob;
            }
        }

        switch (dev->data.disk->device) {
        case VIR_DOMAIN_DISK_DEVICE_CDROM:
        case VIR_DOMAIN_DISK_DEVICE_FLOPPY:
7695 7696 7697
            ret = qemudDomainChangeEjectableMedia(driver, vm,
                                                  dev->data.disk,
                                                  qemuCmdFlags);
7698 7699 7700 7701 7702 7703 7704 7705 7706 7707 7708 7709 7710 7711 7712 7713 7714 7715
            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) {
            virCgroupDenyDevicePath(cgroup,
                                    dev->data.disk->src);
        }
        break;

7716 7717 7718 7719
    case VIR_DOMAIN_DEVICE_GRAPHICS:
        ret = qemuDomainChangeGraphics(driver, vm, dev->data.graphics);
        break;

7720 7721 7722 7723 7724 7725 7726 7727 7728 7729 7730 7731 7732 7733 7734 7735 7736 7737 7738 7739 7740 7741 7742 7743 7744 7745
    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;
}


7746
static int qemudDomainDetachPciDiskDevice(struct qemud_driver *driver,
7747
                                          virDomainObjPtr vm,
7748 7749
                                          virDomainDeviceDefPtr dev,
                                          unsigned long long qemuCmdFlags)
7750 7751 7752
{
    int i, ret = -1;
    virDomainDiskDefPtr detach = NULL;
7753
    qemuDomainObjPrivatePtr priv = vm->privateData;
7754 7755 7756 7757 7758 7759 7760 7761 7762

    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (STREQ(vm->def->disks[i]->dst, dev->data.disk->dst)) {
            detach = vm->def->disks[i];
            break;
        }
    }

    if (!detach) {
7763 7764
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("disk %s not found"), dev->data.disk->dst);
7765
        goto cleanup;
7766 7767
    }

7768 7769
    if (!virDomainDeviceAddressIsValid(&detach->info,
                                       VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI)) {
7770 7771
        qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
                        _("device cannot be detached without a PCI address"));
7772
        goto cleanup;
7773 7774
    }

7775
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7776 7777 7778 7779 7780 7781 7782 7783 7784 7785 7786
    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;
        }
7787
    }
7788
    qemuDomainObjExitMonitorWithDriver(driver, vm);
7789

7790 7791 7792 7793 7794 7795 7796 7797 7798 7799
    if (vm->def->ndisks > 1) {
        memmove(vm->def->disks + i,
                vm->def->disks + i + 1,
                sizeof(*vm->def->disks) *
                (vm->def->ndisks - (i + 1)));
        vm->def->ndisks--;
        if (VIR_REALLOC_N(vm->def->disks, vm->def->ndisks) < 0) {
            /* ignore, harmless */
        }
    } else {
7800
        VIR_FREE(vm->def->disks);
7801
        vm->def->ndisks = 0;
7802
    }
7803
    virDomainDiskDefFree(detach);
7804

7805 7806
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
7807
        driver->securityDriver->domainRestoreSecurityImageLabel(vm, dev->data.disk) < 0)
7808 7809
        VIR_WARN("Unable to restore security label on %s", dev->data.disk->src);

7810
    ret = 0;
7811 7812

cleanup:
7813 7814 7815
    return ret;
}

7816
static int qemudDomainDetachPciControllerDevice(struct qemud_driver *driver,
7817
                                                virDomainObjPtr vm,
7818 7819
                                                virDomainDeviceDefPtr dev,
                                                unsigned long long qemuCmdFlags)
7820 7821 7822 7823 7824 7825 7826 7827 7828 7829 7830 7831 7832 7833
{
    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) {
7834 7835 7836 7837
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("disk controller %s:%d not found"),
                        virDomainControllerTypeToString(dev->data.controller->type),
                        dev->data.controller->idx);
7838 7839 7840 7841 7842
        goto cleanup;
    }

    if (!virDomainDeviceAddressIsValid(&detach->info,
                                       VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI)) {
7843 7844
        qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
                        _("device cannot be detached without a PCI address"));
7845 7846 7847
        goto cleanup;
    }

7848 7849 7850 7851 7852
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuAssignDeviceControllerAlias(detach) < 0)
            goto cleanup;
    }

7853
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7854 7855 7856 7857 7858 7859 7860 7861 7862 7863 7864
    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;
        }
7865 7866 7867 7868 7869 7870 7871 7872 7873 7874 7875 7876 7877 7878 7879 7880
    }
    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;
    }
7881 7882 7883 7884 7885

    if (qemuDomainPCIAddressReleaseAddr(priv->pciaddrs, &detach->info) < 0) {
        VIR_WARN0("Unable to release PCI address on controller");
    }

7886 7887 7888 7889 7890 7891 7892 7893
    virDomainControllerDefFree(detach);

    ret = 0;

cleanup:
    return ret;
}

7894
static int
7895
qemudDomainDetachNetDevice(struct qemud_driver *driver,
7896
                           virDomainObjPtr vm,
7897 7898
                           virDomainDeviceDefPtr dev,
                           unsigned long long qemuCmdFlags)
7899 7900 7901
{
    int i, ret = -1;
    virDomainNetDefPtr detach = NULL;
7902
    qemuDomainObjPrivatePtr priv = vm->privateData;
7903 7904
    int vlan;
    char *hostnet_name = NULL;
7905 7906 7907 7908 7909 7910 7911 7912 7913 7914 7915

    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) {
7916 7917 7918 7919 7920
        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]);
7921 7922 7923
        goto cleanup;
    }

7924 7925
    if (!virDomainDeviceAddressIsValid(&detach->info,
                                       VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI)) {
7926 7927
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("device cannot be detached without a PCI address"));
7928 7929 7930
        goto cleanup;
    }

7931
    if ((vlan = qemuDomainNetVLAN(detach)) < 0) {
7932 7933
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("unable to determine original VLAN"));
7934 7935 7936 7937
        goto cleanup;
    }

    if (virAsprintf(&hostnet_name, "host%s", detach->info.alias) < 0) {
7938
        virReportOOMError();
7939 7940 7941
        goto cleanup;
    }

7942
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7943 7944 7945 7946 7947 7948 7949 7950 7951 7952 7953
    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;
        }
7954
    }
7955

7956
    if (qemuMonitorRemoveHostNetwork(priv->mon, vlan, hostnet_name) < 0) {
7957
        qemuDomainObjExitMonitorWithDriver(driver, vm);
7958
        goto cleanup;
7959
    }
7960
    qemuDomainObjExitMonitorWithDriver(driver, vm);
7961

7962 7963
    virNWFilterTearNWFilter(detach);

7964 7965
#if WITH_MACVTAP
    if (detach->type == VIR_DOMAIN_NET_TYPE_DIRECT) {
S
Stefan Berger 已提交
7966 7967
        if (detach->ifname)
            delMacvtap(detach->ifname);
7968 7969 7970
    }
#endif

7971
    if ((driver->macFilter) && (detach->ifname != NULL)) {
7972
        if ((errno = networkDisallowMacOnPort(driver,
7973 7974
                                              detach->ifname,
                                              detach->mac))) {
7975
            virReportSystemError(errno,
7976 7977 7978 7979 7980
             _("failed to remove ebtables rule on  '%s'"),
                                 detach->ifname);
        }
    }

7981 7982 7983 7984 7985 7986 7987 7988 7989 7990
    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 {
7991
        VIR_FREE(vm->def->nets);
7992
        vm->def->nnets = 0;
7993
    }
7994
    virDomainNetDefFree(detach);
7995

7996 7997 7998
    ret = 0;

cleanup:
7999
    VIR_FREE(hostnet_name);
8000 8001 8002
    return ret;
}

8003
static int qemudDomainDetachHostPciDevice(struct qemud_driver *driver,
8004
                                          virDomainObjPtr vm,
8005 8006
                                          virDomainDeviceDefPtr dev,
                                          unsigned long long qemuCmdFlags)
8007
{
8008
    virDomainHostdevDefPtr detach = NULL;
8009
    qemuDomainObjPrivatePtr priv = vm->privateData;
8010
    int i, ret;
8011
    pciDevice *pci;
8012 8013

    for (i = 0 ; i < vm->def->nhostdevs ; i++) {
8014 8015 8016 8017
        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;

8018 8019 8020 8021 8022 8023 8024 8025 8026 8027 8028 8029 8030 8031 8032
        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) {
8033 8034 8035 8036 8037 8038
        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);
8039 8040 8041
        return -1;
    }

8042 8043
    if (!virDomainDeviceAddressIsValid(&detach->info,
                                       VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI)) {
8044 8045
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("device cannot be detached without a PCI address"));
8046 8047 8048
        return -1;
    }

8049
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
8050 8051 8052 8053 8054 8055 8056 8057 8058 8059 8060
    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;
        }
8061
    }
8062
    qemuDomainObjExitMonitorWithDriver(driver, vm);
8063 8064 8065

    ret = 0;

8066
    pci = pciGetDevice(detach->source.subsys.u.pci.domain,
8067 8068 8069 8070 8071 8072
                       detach->source.subsys.u.pci.bus,
                       detach->source.subsys.u.pci.slot,
                       detach->source.subsys.u.pci.function);
    if (!pci)
        ret = -1;
    else {
8073
        pciDeviceSetManaged(pci, detach->managed);
8074 8075
        pciDeviceListDel(driver->activePciHostdevs, pci);
        if (pciResetDevice(pci, driver->activePciHostdevs) < 0)
8076
            ret = -1;
8077
        qemudReattachManagedDevice(pci);
8078
        pciFreeDevice(pci);
8079 8080
    }

8081 8082 8083 8084 8085 8086 8087 8088 8089 8090 8091 8092
    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;
8093
    }
8094
    virDomainHostdevDefFree(detach);
8095 8096 8097 8098

    return ret;
}

8099 8100 8101 8102 8103 8104 8105 8106 8107 8108 8109 8110 8111 8112 8113 8114
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;
8115 8116 8117 8118 8119 8120 8121 8122 8123 8124 8125 8126 8127 8128 8129 8130
        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;
            }
8131 8132 8133 8134 8135 8136 8137 8138 8139 8140 8141 8142 8143 8144 8145 8146 8147 8148 8149 8150 8151 8152 8153 8154 8155 8156 8157 8158 8159 8160 8161 8162 8163 8164 8165 8166 8167 8168 8169 8170 8171 8172 8173 8174 8175 8176 8177 8178 8179 8180
        }
    }

    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;
}

8181
static int qemudDomainDetachHostDevice(struct qemud_driver *driver,
8182
                                       virDomainObjPtr vm,
8183 8184
                                       virDomainDeviceDefPtr dev,
                                       unsigned long long qemuCmdFlags)
8185 8186 8187 8188 8189
{
    virDomainHostdevDefPtr hostdev = dev->data.hostdev;
    int ret;

    if (hostdev->mode != VIR_DOMAIN_HOSTDEV_MODE_SUBSYS) {
8190 8191 8192
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        _("hostdev mode '%s' not supported"),
                        virDomainHostdevModeTypeToString(hostdev->mode));
8193 8194 8195 8196 8197
        return -1;
    }

    switch (hostdev->source.subsys.type) {
    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI:
8198
        ret = qemudDomainDetachHostPciDevice(driver, vm, dev, qemuCmdFlags);
8199
        break;
8200 8201 8202
    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB:
        ret = qemudDomainDetachHostUsbDevice(driver, vm, dev, qemuCmdFlags);
        break;
8203
    default:
8204 8205 8206
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        _("hostdev subsys type '%s' not supported"),
                        virDomainHostdevSubsysTypeToString(hostdev->source.subsys.type));
8207 8208 8209
        return -1;
    }

8210
    if (driver->securityDriver &&
8211
        driver->securityDriver->domainRestoreSecurityHostdevLabel &&
8212
        driver->securityDriver->domainRestoreSecurityHostdevLabel(vm, dev->data.hostdev) < 0)
8213
        VIR_WARN0("Failed to restore host device labelling");
8214

8215 8216 8217
    return ret;
}

8218 8219
static int qemudDomainDetachDevice(virDomainPtr dom,
                                   const char *xml) {
8220 8221
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
8222
    unsigned long long qemuCmdFlags;
8223 8224
    virDomainDeviceDefPtr dev = NULL;
    int ret = -1;
8225

8226
    qemuDriverLock(driver);
8227
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
8228
    if (!vm) {
8229 8230
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
8231 8232
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
8233
        goto cleanup;
8234 8235
    }

8236 8237 8238
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
8239
    if (!virDomainObjIsActive(vm)) {
8240 8241
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot detach device on inactive domain"));
8242
        goto endjob;
8243 8244
    }

8245
    dev = virDomainDeviceDefParse(driver->caps, vm->def, xml,
8246
                                  VIR_DOMAIN_XML_INACTIVE);
8247
    if (dev == NULL)
8248
        goto endjob;
8249

8250 8251 8252 8253
    if (qemudExtractVersionInfo(vm->def->emulator,
                                NULL,
                                &qemuCmdFlags) < 0)
        goto endjob;
8254 8255 8256

    if (dev->type == VIR_DOMAIN_DEVICE_DISK &&
        dev->data.disk->device == VIR_DOMAIN_DISK_DEVICE_DISK &&
8257
        dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_VIRTIO) {
8258
        ret = qemudDomainDetachPciDiskDevice(driver, vm, dev, qemuCmdFlags);
8259
    } else if (dev->type == VIR_DOMAIN_DEVICE_NET) {
8260
        ret = qemudDomainDetachNetDevice(driver, vm, dev, qemuCmdFlags);
8261 8262
    } else if (dev->type == VIR_DOMAIN_DEVICE_CONTROLLER) {
        if (dev->data.controller->type == VIR_DOMAIN_CONTROLLER_TYPE_SCSI) {
8263 8264
            ret = qemudDomainDetachPciControllerDevice(driver, vm, dev,
                                                       qemuCmdFlags);
8265
        } else {
8266 8267 8268
            qemuReportError(VIR_ERR_NO_SUPPORT,
                            _("disk controller bus '%s' cannot be hotunplugged."),
                            virDomainControllerTypeToString(dev->data.controller->type));
8269 8270
            /* fallthrough */
        }
8271
    } else if (dev->type == VIR_DOMAIN_DEVICE_HOSTDEV) {
8272
        ret = qemudDomainDetachHostDevice(driver, vm, dev, qemuCmdFlags);
8273
    } else {
8274 8275
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        "%s", _("This type of device cannot be hot unplugged"));
8276
    }
8277

8278
    if (!ret && virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
8279 8280
        ret = -1;

8281
endjob:
8282 8283
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
8284

8285 8286
cleanup:
    virDomainDeviceDefFree(dev);
8287 8288
    if (vm)
        virDomainObjUnlock(vm);
8289
    qemuDriverUnlock(driver);
8290 8291 8292
    return ret;
}

8293 8294 8295 8296
static int qemudDomainDetachDeviceFlags(virDomainPtr dom,
                                        const char *xml,
                                        unsigned int flags) {
    if (flags & VIR_DOMAIN_DEVICE_MODIFY_CONFIG) {
8297 8298
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot modify the persistent configuration of a domain"));
8299 8300 8301 8302 8303 8304
        return -1;
    }

    return qemudDomainDetachDevice(dom, xml);
}

8305
static int qemudDomainGetAutostart(virDomainPtr dom,
8306
                                   int *autostart) {
8307 8308 8309
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
8310

8311
    qemuDriverLock(driver);
8312
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
8313 8314
    qemuDriverUnlock(driver);

8315
    if (!vm) {
8316 8317
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
8318 8319
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
8320
        goto cleanup;
8321 8322 8323
    }

    *autostart = vm->autostart;
8324
    ret = 0;
8325

8326
cleanup:
8327 8328
    if (vm)
        virDomainObjUnlock(vm);
8329
    return ret;
8330 8331
}

8332
static int qemudDomainSetAutostart(virDomainPtr dom,
8333
                                   int autostart) {
8334 8335
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
8336 8337
    char *configFile = NULL, *autostartLink = NULL;
    int ret = -1;
8338

8339
    qemuDriverLock(driver);
8340
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
8341

8342
    if (!vm) {
8343 8344
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
8345 8346
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
8347
        goto cleanup;
8348 8349
    }

8350
    if (!vm->persistent) {
8351 8352
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("cannot set autostart for transient domain"));
8353
        goto cleanup;
8354 8355
    }

8356 8357
    autostart = (autostart != 0);

8358
    if (vm->autostart != autostart) {
8359
        if ((configFile = virDomainConfigFile(driver->configDir, vm->def->name)) == NULL)
8360
            goto cleanup;
8361
        if ((autostartLink = virDomainConfigFile(driver->autostartDir, vm->def->name)) == NULL)
8362
            goto cleanup;
8363

8364 8365
        if (autostart) {
            int err;
8366

8367
            if ((err = virFileMakePath(driver->autostartDir))) {
8368
                virReportSystemError(err,
8369 8370
                                     _("cannot create autostart directory %s"),
                                     driver->autostartDir);
8371 8372
                goto cleanup;
            }
8373

8374
            if (symlink(configFile, autostartLink) < 0) {
8375
                virReportSystemError(errno,
8376 8377
                                     _("Failed to create symlink '%s to '%s'"),
                                     autostartLink, configFile);
8378 8379 8380 8381
                goto cleanup;
            }
        } else {
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
8382
                virReportSystemError(errno,
8383 8384
                                     _("Failed to delete symlink '%s'"),
                                     autostartLink);
8385 8386
                goto cleanup;
            }
8387 8388
        }

8389
        vm->autostart = autostart;
8390
    }
8391
    ret = 0;
8392

8393 8394 8395
cleanup:
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
8396 8397
    if (vm)
        virDomainObjUnlock(vm);
8398
    qemuDriverUnlock(driver);
8399
    return ret;
8400 8401
}

8402 8403 8404 8405 8406

static char *qemuGetSchedulerType(virDomainPtr dom,
                                  int *nparams)
{
    struct qemud_driver *driver = dom->conn->privateData;
8407
    char *ret = NULL;
8408

8409
    qemuDriverLock(driver);
8410
    if (!qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_CPU)) {
8411 8412
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        __FUNCTION__);
8413
        goto cleanup;
8414 8415 8416 8417 8418 8419 8420
    }

    if (nparams)
        *nparams = 1;

    ret = strdup("posix");
    if (!ret)
8421
        virReportOOMError();
8422 8423 8424

cleanup:
    qemuDriverUnlock(driver);
8425 8426 8427 8428 8429 8430 8431 8432 8433 8434 8435 8436 8437
    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;

8438
    qemuDriverLock(driver);
8439
    if (!qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_CPU)) {
8440 8441
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        __FUNCTION__);
8442
        goto cleanup;
8443 8444 8445 8446 8447
    }

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

    if (vm == NULL) {
8448 8449
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("No such domain %s"), dom->uuid);
8450 8451 8452 8453
        goto cleanup;
    }

    if (virCgroupForDomain(driver->cgroup, vm->def->name, &group, 0) != 0) {
8454 8455
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("cannot find cgroup for domain %s"), vm->def->name);
8456 8457 8458 8459 8460 8461 8462 8463 8464
        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) {
8465 8466
                qemuReportError(VIR_ERR_INVALID_ARG, "%s",
                                _("invalid type for cpu_shares tunable, expected a 'ullong'"));
8467 8468 8469 8470 8471
                goto cleanup;
            }

            rc = virCgroupSetCpuShares(group, params[i].value.ul);
            if (rc != 0) {
8472
                virReportSystemError(-rc, "%s",
8473 8474 8475 8476
                                     _("unable to set cpu shares tunable"));
                goto cleanup;
            }
        } else {
8477 8478
            qemuReportError(VIR_ERR_INVALID_ARG,
                            _("Invalid parameter `%s'"), param->field);
8479 8480 8481 8482 8483 8484 8485 8486 8487
            goto cleanup;
        }
    }
    ret = 0;

cleanup:
    virCgroupFree(&group);
    if (vm)
        virDomainObjUnlock(vm);
8488
    qemuDriverUnlock(driver);
8489 8490 8491 8492 8493 8494 8495 8496 8497 8498 8499 8500 8501 8502
    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;

8503
    qemuDriverLock(driver);
8504
    if (!qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_CPU)) {
8505 8506
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        __FUNCTION__);
8507
        goto cleanup;
8508 8509 8510
    }

    if ((*nparams) != 1) {
8511 8512
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("Invalid parameter count"));
8513
        goto cleanup;
8514 8515 8516 8517 8518
    }

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

    if (vm == NULL) {
8519 8520
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("No such domain %s"), dom->uuid);
8521 8522 8523 8524
        goto cleanup;
    }

    if (virCgroupForDomain(driver->cgroup, vm->def->name, &group, 0) != 0) {
8525 8526
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("cannot find cgroup for domain %s"), vm->def->name);
8527 8528 8529 8530 8531
        goto cleanup;
    }

    rc = virCgroupGetCpuShares(group, &val);
    if (rc != 0) {
8532
        virReportSystemError(-rc, "%s",
8533 8534 8535 8536 8537
                             _("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
8538
    if (virStrcpyStatic(params[0].field, "cpu_shares") == NULL) {
8539 8540
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("Field cpu_shares too long for destination"));
C
Chris Lalancette 已提交
8541 8542
        goto cleanup;
    }
8543 8544 8545 8546 8547 8548 8549

    ret = 0;

cleanup:
    virCgroupFree(&group);
    if (vm)
        virDomainObjUnlock(vm);
8550
    qemuDriverUnlock(driver);
8551 8552 8553 8554
    return ret;
}


8555 8556 8557 8558 8559 8560 8561 8562 8563
/* 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)
{
8564
    struct qemud_driver *driver = dom->conn->privateData;
8565
    int i, ret = -1;
8566
    virDomainObjPtr vm;
8567
    virDomainDiskDefPtr disk = NULL;
8568

8569
    qemuDriverLock(driver);
8570
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
8571
    qemuDriverUnlock(driver);
8572
    if (!vm) {
8573 8574
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
8575 8576
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
8577
        goto cleanup;
8578
    }
8579 8580 8581 8582

    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
8583
    if (!virDomainObjIsActive (vm)) {
8584 8585
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
8586
        goto endjob;
8587 8588
    }

8589 8590 8591 8592 8593 8594 8595 8596
    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (STREQ(path, vm->def->disks[i]->dst)) {
            disk = vm->def->disks[i];
            break;
        }
    }

    if (!disk) {
8597 8598
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("invalid path: %s"), path);
8599
        goto endjob;
8600 8601
    }

8602
    if (!disk->info.alias) {
8603 8604
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("missing disk device alias name for %s"), disk->dst);
8605
        goto endjob;
8606
    }
8607

8608
    qemuDomainObjPrivatePtr priv = vm->privateData;
8609 8610
    qemuDomainObjEnterMonitor(vm);
    ret = qemuMonitorGetBlockStatsInfo(priv->mon,
8611
                                       disk->info.alias,
8612 8613 8614 8615 8616 8617
                                       &stats->rd_req,
                                       &stats->rd_bytes,
                                       &stats->wr_req,
                                       &stats->wr_bytes,
                                       &stats->errs);
    qemuDomainObjExitMonitor(vm);
8618

8619
endjob:
8620 8621
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
8622

8623
cleanup:
8624 8625
    if (vm)
        virDomainObjUnlock(vm);
8626
    return ret;
8627 8628
}

8629
#ifdef __linux__
8630 8631 8632 8633 8634
static int
qemudDomainInterfaceStats (virDomainPtr dom,
                           const char *path,
                           struct _virDomainInterfaceStats *stats)
{
8635 8636
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
8637
    int i;
8638
    int ret = -1;
8639

8640
    qemuDriverLock(driver);
8641
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
8642 8643
    qemuDriverUnlock(driver);

8644
    if (!vm) {
8645 8646
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
8647 8648
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
8649
        goto cleanup;
8650 8651
    }

D
Daniel P. Berrange 已提交
8652
    if (!virDomainObjIsActive(vm)) {
8653 8654
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
8655
        goto cleanup;
8656 8657 8658
    }

    /* Check the path is one of the domain's network interfaces. */
8659 8660
    for (i = 0 ; i < vm->def->nnets ; i++) {
        if (vm->def->nets[i]->ifname &&
8661 8662 8663 8664
            STREQ (vm->def->nets[i]->ifname, path)) {
            ret = 0;
            break;
        }
8665 8666
    }

8667
    if (ret == 0)
8668
        ret = linuxDomainInterfaceStats(path, stats);
8669
    else
8670 8671
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("invalid path, '%s' is not a known interface"), path);
8672

8673
cleanup:
8674 8675
    if (vm)
        virDomainObjUnlock(vm);
8676 8677
    return ret;
}
8678
#else
8679 8680 8681 8682
static int
qemudDomainInterfaceStats (virDomainPtr dom,
                           const char *path ATTRIBUTE_UNUSED,
                           struct _virDomainInterfaceStats *stats ATTRIBUTE_UNUSED)
8683 8684
    qemuReportError(VIR_ERR_NO_SUPPORT,
                    "%s", __FUNCTION__);
8685 8686
    return -1;
}
8687
#endif
8688

8689 8690 8691 8692 8693 8694 8695 8696 8697 8698 8699 8700 8701 8702 8703 8704
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);
8705 8706
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
8707 8708 8709
        goto cleanup;
    }

8710 8711 8712
    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

8713 8714 8715
    if (virDomainObjIsActive(vm)) {
        qemuDomainObjPrivatePtr priv = vm->privateData;
        qemuDomainObjEnterMonitor(vm);
8716
        ret = qemuMonitorGetMemoryStats(priv->mon, stats, nr_stats);
8717 8718
        qemuDomainObjExitMonitor(vm);
    } else {
8719 8720
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
8721 8722
    }

8723 8724 8725
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;

8726 8727 8728 8729 8730 8731
cleanup:
    if (vm)
        virDomainObjUnlock(vm);
    return ret;
}

8732 8733 8734 8735 8736 8737 8738
static int
qemudDomainBlockPeek (virDomainPtr dom,
                      const char *path,
                      unsigned long long offset, size_t size,
                      void *buffer,
                      unsigned int flags ATTRIBUTE_UNUSED)
{
8739 8740 8741
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int fd = -1, ret = -1, i;
8742

8743
    qemuDriverLock(driver);
8744
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
8745 8746
    qemuDriverUnlock(driver);

8747
    if (!vm) {
8748 8749
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
8750 8751
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
8752
        goto cleanup;
8753 8754 8755
    }

    if (!path || path[0] == '\0') {
8756 8757
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("NULL or empty path"));
8758
        goto cleanup;
8759 8760 8761
    }

    /* Check the path belongs to this domain. */
8762 8763
    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (vm->def->disks[i]->src != NULL &&
8764 8765 8766 8767
            STREQ (vm->def->disks[i]->src, path)) {
            ret = 0;
            break;
        }
8768 8769
    }

8770 8771 8772 8773 8774
    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) {
8775 8776
            virReportSystemError(errno,
                                 _("%s: failed to open"), path);
8777 8778
            goto cleanup;
        }
8779

8780 8781 8782 8783 8784 8785
        /* 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) {
8786 8787
            virReportSystemError(errno,
                                 _("%s: failed to seek or read"), path);
8788 8789 8790 8791 8792
            goto cleanup;
        }

        ret = 0;
    } else {
8793 8794
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("invalid path"));
8795 8796
    }

8797 8798 8799
cleanup:
    if (fd >= 0)
        close (fd);
8800 8801
    if (vm)
        virDomainObjUnlock(vm);
8802 8803 8804
    return ret;
}

R
Richard W.M. Jones 已提交
8805 8806 8807 8808 8809 8810
static int
qemudDomainMemoryPeek (virDomainPtr dom,
                       unsigned long long offset, size_t size,
                       void *buffer,
                       unsigned int flags)
{
8811 8812
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
8813
    char *tmp = NULL;
R
Richard W.M. Jones 已提交
8814 8815
    int fd = -1, ret = -1;

8816
    qemuDriverLock(driver);
8817
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
8818
    qemuDriverUnlock(driver);
R
Richard W.M. Jones 已提交
8819 8820

    if (!vm) {
8821 8822
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
8823 8824
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
8825 8826 8827
        goto cleanup;
    }

8828
    if (flags != VIR_MEMORY_VIRTUAL && flags != VIR_MEMORY_PHYSICAL) {
8829 8830
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("flags parameter must be VIR_MEMORY_VIRTUAL or VIR_MEMORY_PHYSICAL"));
8831
        goto cleanup;
R
Richard W.M. Jones 已提交
8832 8833
    }

8834 8835 8836
    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
8837
    if (!virDomainObjIsActive(vm)) {
8838 8839
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
8840
        goto endjob;
R
Richard W.M. Jones 已提交
8841 8842
    }

8843
    if (virAsprintf(&tmp, driver->cacheDir,  "/qemu.mem.XXXXXX") < 0) {
8844
        virReportOOMError();
8845
        goto endjob;
8846 8847
    }

R
Richard W.M. Jones 已提交
8848 8849
    /* Create a temporary filename. */
    if ((fd = mkstemp (tmp)) == -1) {
8850 8851
        virReportSystemError(errno,
                             _("mkstemp(\"%s\") failed"), tmp);
8852
        goto endjob;
R
Richard W.M. Jones 已提交
8853 8854
    }

8855
    qemuDomainObjPrivatePtr priv = vm->privateData;
8856
    qemuDomainObjEnterMonitor(vm);
8857
    if (flags == VIR_MEMORY_VIRTUAL) {
8858 8859
        if (qemuMonitorSaveVirtualMemory(priv->mon, offset, size, tmp) < 0) {
            qemuDomainObjExitMonitor(vm);
8860
            goto endjob;
8861
        }
8862
    } else {
8863 8864
        if (qemuMonitorSavePhysicalMemory(priv->mon, offset, size, tmp) < 0) {
            qemuDomainObjExitMonitor(vm);
8865
            goto endjob;
8866
        }
R
Richard W.M. Jones 已提交
8867
    }
8868
    qemuDomainObjExitMonitor(vm);
R
Richard W.M. Jones 已提交
8869 8870 8871

    /* Read the memory file into buffer. */
    if (saferead (fd, buffer, size) == (ssize_t) -1) {
8872 8873 8874
        virReportSystemError(errno,
                             _("failed to read temporary file "
                               "created with template %s"), tmp);
8875
        goto endjob;
R
Richard W.M. Jones 已提交
8876 8877 8878
    }

    ret = 0;
8879

8880
endjob:
8881 8882
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
8883

8884
cleanup:
8885
    VIR_FREE(tmp);
R
Richard W.M. Jones 已提交
8886 8887
    if (fd >= 0) close (fd);
    unlink (tmp);
8888 8889
    if (vm)
        virDomainObjUnlock(vm);
R
Richard W.M. Jones 已提交
8890 8891 8892
    return ret;
}

8893

8894
static int
8895 8896 8897 8898
qemuDomainEventRegister(virConnectPtr conn,
                        virConnectDomainEventCallback callback,
                        void *opaque,
                        virFreeCallback freecb)
8899
{
8900 8901 8902
    struct qemud_driver *driver = conn->privateData;
    int ret;

8903
    qemuDriverLock(driver);
8904 8905
    ret = virDomainEventCallbackListAdd(conn, driver->domainEventCallbacks,
                                        callback, opaque, freecb);
8906
    qemuDriverUnlock(driver);
8907

8908
    return ret;
8909 8910
}

8911

8912
static int
8913 8914
qemuDomainEventDeregister(virConnectPtr conn,
                          virConnectDomainEventCallback callback)
8915
{
8916 8917 8918
    struct qemud_driver *driver = conn->privateData;
    int ret;

8919
    qemuDriverLock(driver);
8920 8921 8922 8923 8924 8925
    if (driver->domainEventDispatching)
        ret = virDomainEventCallbackListMarkDelete(conn, driver->domainEventCallbacks,
                                                   callback);
    else
        ret = virDomainEventCallbackListRemove(conn, driver->domainEventCallbacks,
                                               callback);
8926
    qemuDriverUnlock(driver);
8927

8928
    return ret;
8929 8930
}

8931 8932 8933 8934 8935 8936 8937 8938 8939 8940 8941 8942 8943 8944 8945 8946 8947 8948 8949 8950 8951 8952 8953 8954 8955 8956 8957 8958 8959 8960 8961 8962 8963 8964 8965 8966 8967 8968 8969 8970 8971 8972 8973

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;
}


8974 8975
static void qemuDomainEventDispatchFunc(virConnectPtr conn,
                                        virDomainEventPtr event,
8976
                                        virConnectDomainEventGenericCallback cb,
8977 8978
                                        void *cbopaque,
                                        void *opaque)
8979
{
8980
    struct qemud_driver *driver = opaque;
8981

8982 8983 8984 8985 8986 8987 8988 8989 8990 8991 8992 8993 8994 8995 8996 8997 8998 8999 9000 9001 9002 9003 9004 9005 9006 9007 9008 9009 9010 9011 9012 9013 9014 9015 9016 9017 9018 9019 9020 9021 9022 9023 9024 9025
    /* 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);
9026 9027
}

D
Daniel Veillard 已提交
9028 9029
/* Migration support. */

C
Chris Lalancette 已提交
9030 9031 9032 9033 9034 9035 9036 9037 9038 9039 9040 9041 9042 9043 9044 9045 9046 9047 9048
/* 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) {
9049 9050
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream is not open"));
C
Chris Lalancette 已提交
9051 9052 9053 9054 9055
        return -1;
    }

    qemuDriverLock(driver);
    if (qemust->watch == 0) {
9056 9057
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream does not have a callback registered"));
C
Chris Lalancette 已提交
9058 9059 9060 9061 9062 9063 9064 9065 9066 9067 9068 9069 9070 9071 9072 9073 9074 9075 9076 9077 9078 9079 9080 9081 9082 9083 9084 9085
        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) {
9086 9087
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream is not open"));
C
Chris Lalancette 已提交
9088 9089 9090 9091 9092
        return -1;
    }

    qemuDriverLock(driver);
    if (qemust->watch == 0) {
9093 9094
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream does not have a callback registered"));
C
Chris Lalancette 已提交
9095 9096 9097 9098 9099 9100 9101 9102 9103 9104 9105 9106 9107 9108 9109 9110 9111 9112 9113 9114 9115 9116 9117 9118 9119 9120 9121 9122 9123 9124 9125 9126 9127 9128 9129 9130 9131 9132 9133 9134 9135 9136 9137 9138 9139 9140 9141 9142 9143 9144 9145 9146 9147 9148 9149 9150 9151
        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) {
9152 9153
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream is not open"));
C
Chris Lalancette 已提交
9154 9155 9156 9157 9158
        return -1;
    }

    qemuDriverLock(driver);
    if (qemust->watch != 0) {
9159 9160
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream already has a callback registered"));
C
Chris Lalancette 已提交
9161 9162 9163 9164 9165 9166 9167 9168
        goto cleanup;
    }

    if ((qemust->watch = virEventAddHandle(qemust->fd,
                                           events,
                                           qemuStreamMigEvent,
                                           st,
                                           NULL)) < 0) {
9169 9170
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("cannot register file watch on stream"));
C
Chris Lalancette 已提交
9171 9172 9173 9174 9175 9176 9177 9178 9179 9180 9181 9182 9183 9184 9185 9186 9187 9188 9189 9190 9191 9192 9193 9194 9195 9196 9197 9198 9199 9200 9201 9202
        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;

9203
    if (VIR_ALLOC(qemust) < 0) {
9204
        virReportOOMError();
C
Chris Lalancette 已提交
9205
        return NULL;
9206
    }
C
Chris Lalancette 已提交
9207 9208 9209 9210 9211 9212 9213 9214 9215 9216 9217 9218 9219 9220 9221 9222 9223 9224 9225 9226 9227 9228 9229 9230 9231 9232 9233 9234 9235 9236 9237 9238 9239 9240 9241 9242 9243 9244 9245 9246 9247 9248 9249 9250 9251 9252 9253 9254 9255 9256 9257 9258 9259 9260 9261 9262 9263 9264 9265 9266 9267

    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) {
9268 9269
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream is not open"));
C
Chris Lalancette 已提交
9270 9271 9272 9273 9274 9275 9276 9277 9278 9279 9280 9281 9282 9283
        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;
9284
            virReportSystemError(errno, "%s",
C
Chris Lalancette 已提交
9285 9286 9287 9288 9289 9290 9291 9292 9293 9294 9295 9296 9297 9298 9299 9300 9301 9302 9303 9304 9305 9306 9307 9308 9309 9310 9311 9312 9313 9314 9315 9316 9317 9318 9319 9320 9321 9322
                                 _("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;
9323
    unsigned long long qemuCmdFlags;
C
Chris Lalancette 已提交
9324 9325 9326 9327
    struct qemuStreamMigFile *qemust = NULL;

    qemuDriverLock(driver);
    if (!dom_xml) {
9328 9329
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("no domain XML passed"));
C
Chris Lalancette 已提交
9330 9331 9332
        goto cleanup;
    }
    if (!(flags & VIR_MIGRATE_TUNNELLED)) {
9333
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
C
Chris Lalancette 已提交
9334 9335 9336 9337
                         "%s", _("PrepareTunnel called but no TUNNELLED flag set"));
        goto cleanup;
    }
    if (st == NULL) {
9338 9339
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("tunnelled migration requested but NULL stream passed"));
C
Chris Lalancette 已提交
9340 9341 9342 9343
        goto cleanup;
    }

    /* Parse the domain XML. */
9344
    if (!(def = virDomainDefParseString(driver->caps, dom_xml,
C
Chris Lalancette 已提交
9345
                                        VIR_DOMAIN_XML_INACTIVE))) {
9346 9347
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to parse XML"));
C
Chris Lalancette 已提交
9348 9349 9350 9351
        goto cleanup;
    }

    /* Target domain name, maybe renamed. */
9352 9353 9354 9355 9356 9357
    if (dname) {
        VIR_FREE(def->name);
        def->name = strdup(dname);
        if (def->name == NULL)
            goto cleanup;
    }
C
Chris Lalancette 已提交
9358

9359 9360
    if (virDomainObjIsDuplicate(&driver->domains, def, 1) < 0)
        goto cleanup;
C
Chris Lalancette 已提交
9361

9362
    if (!(vm = virDomainAssignDef(driver->caps,
C
Chris Lalancette 已提交
9363
                                  &driver->domains,
9364
                                  def, true))) {
9365 9366
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to assign new VM"));
C
Chris Lalancette 已提交
9367 9368 9369 9370
        goto cleanup;
    }
    def = NULL;

9371 9372 9373
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

C
Chris Lalancette 已提交
9374 9375 9376 9377 9378
    /* 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) {
9379
        virReportOOMError();
9380
        goto endjob;
C
Chris Lalancette 已提交
9381 9382 9383 9384 9385
    }
    unlink(unixfile);

    /* check that this qemu version supports the interactive exec */
    if (qemudExtractVersionInfo(vm->def->emulator, NULL, &qemuCmdFlags) < 0) {
9386 9387 9388
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Cannot determine QEMU argv syntax %s"),
                        vm->def->emulator);
9389
        goto endjob;
C
Chris Lalancette 已提交
9390 9391 9392 9393 9394 9395
    }
    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 {
9396 9397
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("Destination qemu is too old to support tunnelled migration"));
9398
        goto endjob;
C
Chris Lalancette 已提交
9399 9400
    }
    if (internalret < 0) {
9401
        virReportOOMError();
9402
        goto endjob;
C
Chris Lalancette 已提交
9403 9404 9405 9406 9407 9408 9409 9410 9411 9412 9413 9414 9415 9416
    }
    /* 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;
        }
9417
        goto endjob;
C
Chris Lalancette 已提交
9418 9419 9420 9421
    }

    qemust = qemuStreamMigOpen(st, unixfile);
    if (qemust == NULL) {
9422
        qemudShutdownVMDaemon(driver, vm);
9423
        if (!vm->persistent) {
9424 9425
            if (qemuDomainObjEndJob(vm) > 0)
                virDomainRemoveInactive(&driver->domains, vm);
9426 9427
            vm = NULL;
        }
9428
        virReportSystemError(errno,
C
Chris Lalancette 已提交
9429 9430
                             _("cannot open unix socket '%s' for tunnelled migration"),
                             unixfile);
9431
        goto endjob;
C
Chris Lalancette 已提交
9432 9433 9434 9435 9436 9437 9438 9439 9440 9441
    }

    st->driver = &qemuStreamMigDrv;
    st->privateData = qemust;

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_MIGRATED);
    ret = 0;

9442
endjob:
9443 9444 9445
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
9446

C
Chris Lalancette 已提交
9447 9448
cleanup:
    virDomainDefFree(def);
9449 9450
    if (unixfile)
        unlink(unixfile);
C
Chris Lalancette 已提交
9451 9452 9453 9454 9455 9456 9457 9458 9459
    VIR_FREE(unixfile);
    if (vm)
        virDomainObjUnlock(vm);
    if (event)
        qemuDomainEventQueue(driver, event);
    qemuDriverUnlock(driver);
    return ret;
}

D
Daniel Veillard 已提交
9460 9461 9462 9463
/* Prepare is the first step, and it runs on the destination host.
 *
 * This starts an empty VM listening on a TCP port.
 */
9464
static int ATTRIBUTE_NONNULL (5)
D
Daniel Veillard 已提交
9465 9466 9467 9468 9469
qemudDomainMigratePrepare2 (virConnectPtr dconn,
                            char **cookie ATTRIBUTE_UNUSED,
                            int *cookielen ATTRIBUTE_UNUSED,
                            const char *uri_in,
                            char **uri_out,
C
Chris Lalancette 已提交
9470
                            unsigned long flags,
D
Daniel Veillard 已提交
9471 9472 9473 9474 9475
                            const char *dname,
                            unsigned long resource ATTRIBUTE_UNUSED,
                            const char *dom_xml)
{
    static int port = 0;
9476 9477
    struct qemud_driver *driver = dconn->privateData;
    virDomainDefPtr def = NULL;
D
Daniel Veillard 已提交
9478 9479
    virDomainObjPtr vm = NULL;
    int this_port;
9480
    char *hostname;
D
Daniel Veillard 已提交
9481 9482
    char migrateFrom [64];
    const char *p;
9483
    virDomainEventPtr event = NULL;
9484
    int ret = -1;
9485
    int internalret;
9486 9487

    *uri_out = NULL;
D
Daniel Veillard 已提交
9488

9489
    qemuDriverLock(driver);
C
Chris Lalancette 已提交
9490 9491 9492 9493
    if (flags & VIR_MIGRATE_TUNNELLED) {
        /* this is a logical error; we never should have gotten here with
         * VIR_MIGRATE_TUNNELLED set
         */
9494 9495
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("Tunnelled migration requested but invalid RPC method called"));
C
Chris Lalancette 已提交
9496 9497 9498
        goto cleanup;
    }

D
Daniel Veillard 已提交
9499
    if (!dom_xml) {
9500 9501
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("no domain XML passed"));
9502
        goto cleanup;
D
Daniel Veillard 已提交
9503 9504 9505 9506 9507 9508 9509 9510 9511 9512 9513 9514 9515 9516 9517 9518 9519
    }

    /* 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 */
9520
        if ((hostname = virGetHostnameLocalhost(0)) == NULL)
9521
            goto cleanup;
D
Daniel Veillard 已提交
9522

9523 9524 9525 9526 9527
        /* 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
9528
        /* Caller frees */
9529 9530 9531
        internalret = virAsprintf(uri_out, "tcp:%s:%d", hostname, this_port);
        VIR_FREE(hostname);
        if (internalret < 0) {
9532
            virReportOOMError();
9533
            goto cleanup;
D
Daniel Veillard 已提交
9534 9535 9536 9537 9538 9539
        }
    } 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.
         */
9540
        if (!STRPREFIX (uri_in, "tcp:")) {
9541 9542
            qemuReportError (VIR_ERR_INVALID_ARG,
                             "%s", _("only tcp URIs are supported for KVM/QEMU migrations"));
9543
            goto cleanup;
D
Daniel Veillard 已提交
9544 9545 9546 9547
        }

        /* Get the port number. */
        p = strrchr (uri_in, ':');
9548 9549 9550 9551 9552 9553 9554 9555
        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) {
9556
                virReportOOMError();
9557 9558 9559 9560 9561 9562 9563
                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)) {
9564 9565
                qemuReportError(VIR_ERR_INVALID_ARG,
                                "%s", _("URI ended with incorrect ':port'"));
9566 9567
                goto cleanup;
            }
D
Daniel Veillard 已提交
9568 9569 9570
        }
    }

9571
    if (*uri_out)
9572 9573
        VIR_DEBUG("Generated uri_out=%s", *uri_out);

D
Daniel Veillard 已提交
9574
    /* Parse the domain XML. */
9575
    if (!(def = virDomainDefParseString(driver->caps, dom_xml,
9576
                                        VIR_DOMAIN_XML_INACTIVE))) {
9577 9578
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to parse XML"));
9579
        goto cleanup;
D
Daniel Veillard 已提交
9580 9581 9582
    }

    /* Target domain name, maybe renamed. */
9583 9584 9585 9586 9587 9588
    if (dname) {
        VIR_FREE(def->name);
        def->name = strdup(dname);
        if (def->name == NULL)
            goto cleanup;
    }
D
Daniel Veillard 已提交
9589

9590 9591
    if (virDomainObjIsDuplicate(&driver->domains, def, 1) < 0)
        goto cleanup;
D
Daniel Veillard 已提交
9592

9593
    if (!(vm = virDomainAssignDef(driver->caps,
D
Daniel Veillard 已提交
9594
                                  &driver->domains,
9595
                                  def, true))) {
9596 9597
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to assign new VM"));
9598
        goto cleanup;
D
Daniel Veillard 已提交
9599
    }
9600
    def = NULL;
D
Daniel Veillard 已提交
9601

9602 9603 9604
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

D
Daniel Veillard 已提交
9605 9606 9607 9608 9609 9610 9611
    /* 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);
9612
    if (qemudStartVMDaemon (dconn, driver, vm, migrateFrom, -1) < 0) {
9613 9614 9615
        /* Note that we don't set an error here because qemudStartVMDaemon
         * should have already done that.
         */
9616
        if (!vm->persistent) {
9617 9618
            if (qemuDomainObjEndJob(vm) > 0)
                virDomainRemoveInactive(&driver->domains, vm);
9619 9620
            vm = NULL;
        }
9621
        goto endjob;
D
Daniel Veillard 已提交
9622
    }
9623 9624 9625 9626

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_MIGRATED);
9627
    ret = 0;
D
Daniel Veillard 已提交
9628

9629
endjob:
9630 9631 9632
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
9633

9634 9635 9636 9637 9638
cleanup:
    virDomainDefFree(def);
    if (ret != 0) {
        VIR_FREE(*uri_out);
    }
9639 9640
    if (vm)
        virDomainObjUnlock(vm);
9641 9642
    if (event)
        qemuDomainEventQueue(driver, event);
9643
    qemuDriverUnlock(driver);
9644
    return ret;
C
Chris Lalancette 已提交
9645 9646 9647

}

9648 9649 9650 9651

/* Perform migration using QEMU's native TCP migrate support,
 * not encrypted obviously
 */
9652
static int doNativeMigrate(struct qemud_driver *driver,
9653 9654 9655 9656 9657 9658 9659
                           virDomainObjPtr vm,
                           const char *uri,
                           unsigned long flags ATTRIBUTE_UNUSED,
                           const char *dname ATTRIBUTE_UNUSED,
                           unsigned long resource)
{
    int ret = -1;
9660
    xmlURIPtr uribits = NULL;
9661
    qemuDomainObjPrivatePtr priv = vm->privateData;
9662 9663 9664 9665 9666 9667

    /* 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) {
9668
            virReportOOMError();
9669 9670 9671 9672 9673 9674 9675 9676
            goto cleanup;
        }
        uribits = xmlParseURI(tmpuri);
        VIR_FREE(tmpuri);
    } else {
        uribits = xmlParseURI(uri);
    }
    if (!uribits) {
9677 9678
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("cannot parse URI %s"), uri);
9679 9680 9681
        goto cleanup;
    }

9682
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
9683
    if (resource > 0 &&
9684
        qemuMonitorSetMigrationSpeed(priv->mon, resource) < 0) {
9685
        qemuDomainObjExitMonitorWithDriver(driver, vm);
9686
        goto cleanup;
9687
    }
9688

9689
    if (qemuMonitorMigrateToHost(priv->mon, 1, uribits->server, uribits->port) < 0) {
9690
        qemuDomainObjExitMonitorWithDriver(driver, vm);
9691 9692
        goto cleanup;
    }
9693
    qemuDomainObjExitMonitorWithDriver(driver, vm);
9694

9695
    if (qemuDomainWaitForMigrationComplete(driver, vm) < 0)
9696 9697 9698 9699 9700 9701 9702 9703 9704 9705
        goto cleanup;

    ret = 0;

cleanup:
    xmlFreeURI(uribits);
    return ret;
}


9706 9707
#define TUNNEL_SEND_BUF_SIZE 65536

9708
static int doTunnelSendAll(virStreamPtr st,
9709 9710
                           int sock)
{
9711 9712 9713 9714 9715 9716 9717 9718
    char *buffer;
    int nbytes = TUNNEL_SEND_BUF_SIZE;

    if (VIR_ALLOC_N(buffer, TUNNEL_SEND_BUF_SIZE) < 0) {
        virReportOOMError();
        virStreamAbort(st);
        return -1;
    }
9719 9720 9721 9722 9723

    /* XXX should honour the 'resource' parameter here */
    for (;;) {
        nbytes = saferead(sock, buffer, nbytes);
        if (nbytes < 0) {
9724
            virReportSystemError(errno, "%s",
9725
                                 _("tunnelled migration failed to read from qemu"));
9726 9727
            virStreamAbort(st);
            VIR_FREE(buffer);
9728 9729 9730 9731 9732 9733 9734
            return -1;
        }
        else if (nbytes == 0)
            /* EOF; get out of here */
            break;

        if (virStreamSend(st, buffer, nbytes) < 0) {
9735 9736
            qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
                            _("Failed to write migration data to remote libvirtd"));
9737
            VIR_FREE(buffer);
9738 9739 9740 9741
            return -1;
        }
    }

9742 9743
    VIR_FREE(buffer);

9744 9745 9746 9747 9748 9749 9750
    if (virStreamFinish(st) < 0)
        /* virStreamFinish set the error for us */
        return -1;

    return 0;
}

C
Chris Lalancette 已提交
9751
static int doTunnelMigrate(virDomainPtr dom,
9752
                           struct qemud_driver *driver,
9753
                           virConnectPtr dconn,
C
Chris Lalancette 已提交
9754
                           virDomainObjPtr vm,
9755
                           const char *dom_xml,
C
Chris Lalancette 已提交
9756 9757 9758 9759 9760
                           const char *uri,
                           unsigned long flags,
                           const char *dname,
                           unsigned long resource)
{
9761
    qemuDomainObjPrivatePtr priv = vm->privateData;
9762 9763
    int client_sock = -1;
    int qemu_sock = -1;
C
Chris Lalancette 已提交
9764 9765
    struct sockaddr_un sa_qemu, sa_client;
    socklen_t addrlen;
9766
    virDomainPtr ddomain = NULL;
C
Chris Lalancette 已提交
9767
    int retval = -1;
9768
    virStreamPtr st = NULL;
C
Chris Lalancette 已提交
9769 9770
    char *unixfile = NULL;
    int internalret;
9771
    unsigned long long qemuCmdFlags;
C
Chris Lalancette 已提交
9772 9773 9774
    int status;
    unsigned long long transferred, remaining, total;

9775 9776 9777 9778 9779 9780 9781 9782
    /*
     * 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
9783 9784
     */

9785

9786
    /* Stage 1. setup local support infrastructure */
C
Chris Lalancette 已提交
9787 9788 9789

    if (virAsprintf(&unixfile, "%s/qemu.tunnelmigrate.src.%s",
                    driver->stateDir, vm->def->name) < 0) {
9790
        virReportOOMError();
9791
        goto cleanup;
C
Chris Lalancette 已提交
9792 9793 9794 9795
    }

    qemu_sock = socket(AF_UNIX, SOCK_STREAM, 0);
    if (qemu_sock < 0) {
9796
        virReportSystemError(errno, "%s",
C
Chris Lalancette 已提交
9797
                             _("cannot open tunnelled migration socket"));
9798
        goto cleanup;
C
Chris Lalancette 已提交
9799 9800 9801 9802 9803
    }
    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) {
9804 9805 9806
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Unix socket '%s' too big for destination"),
                        unixfile);
9807
        goto cleanup;
C
Chris Lalancette 已提交
9808 9809 9810
    }
    unlink(unixfile);
    if (bind(qemu_sock, (struct sockaddr *)&sa_qemu, sizeof(sa_qemu)) < 0) {
9811
        virReportSystemError(errno,
C
Chris Lalancette 已提交
9812 9813
                             _("Cannot bind to unix socket '%s' for tunnelled migration"),
                             unixfile);
9814
        goto cleanup;
C
Chris Lalancette 已提交
9815 9816
    }
    if (listen(qemu_sock, 1) < 0) {
9817
        virReportSystemError(errno,
C
Chris Lalancette 已提交
9818 9819
                             _("Cannot listen on unix socket '%s' for tunnelled migration"),
                             unixfile);
9820
        goto cleanup;
C
Chris Lalancette 已提交
9821 9822 9823 9824
    }

    /* check that this qemu version supports the unix migration */
    if (qemudExtractVersionInfo(vm->def->emulator, NULL, &qemuCmdFlags) < 0) {
9825 9826 9827
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Cannot extract Qemu version from '%s'"),
                        vm->def->emulator);
9828 9829 9830 9831 9832
        goto cleanup;
    }

    if (!(qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_UNIX) &&
        !(qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_EXEC)) {
9833 9834
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("Source qemu is too old to support tunnelled migration"));
9835
        goto cleanup;
C
Chris Lalancette 已提交
9836
    }
9837 9838 9839 9840 9841 9842 9843 9844 9845 9846 9847 9848 9849 9850 9851 9852 9853 9854 9855 9856 9857


    /* 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 */
9858
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
C
Chris Lalancette 已提交
9859
    if (qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_UNIX)
9860
        internalret = qemuMonitorMigrateToUnix(priv->mon, 1, unixfile);
C
Chris Lalancette 已提交
9861 9862
    else if (qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_EXEC) {
        const char *args[] = { "nc", "-U", unixfile, NULL };
9863
        internalret = qemuMonitorMigrateToCommand(priv->mon, 1, args, "/dev/null");
9864 9865
    } else {
        internalret = -1;
C
Chris Lalancette 已提交
9866
    }
9867
    qemuDomainObjExitMonitorWithDriver(driver, vm);
C
Chris Lalancette 已提交
9868
    if (internalret < 0) {
9869 9870
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("tunnelled migration monitor command failed"));
9871
        goto finish;
C
Chris Lalancette 已提交
9872 9873
    }

9874 9875 9876
    /* From this point onwards we *must* call cancel to abort the
     * migration on source if anything goes wrong */

C
Chris Lalancette 已提交
9877 9878 9879
    /* it is also possible that the migrate didn't fail initially, but
     * rather failed later on.  Check the output of "info migrate"
     */
9880
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
9881 9882
    if (qemuMonitorGetMigrationStatus(priv->mon,
                                      &status,
C
Chris Lalancette 已提交
9883 9884 9885
                                      &transferred,
                                      &remaining,
                                      &total) < 0) {
9886
        qemuDomainObjExitMonitorWithDriver(driver, vm);
9887
        goto cancel;
C
Chris Lalancette 已提交
9888
    }
9889
    qemuDomainObjExitMonitorWithDriver(driver, vm);
C
Chris Lalancette 已提交
9890 9891

    if (status == QEMU_MONITOR_MIGRATION_STATUS_ERROR) {
9892 9893
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s",_("migrate failed"));
9894
        goto cancel;
C
Chris Lalancette 已提交
9895 9896 9897 9898 9899 9900
    }

    addrlen = sizeof(sa_client);
    while ((client_sock = accept(qemu_sock, (struct sockaddr *)&sa_client, &addrlen)) < 0) {
        if (errno == EAGAIN || errno == EINTR)
            continue;
9901
        virReportSystemError(errno, "%s",
C
Chris Lalancette 已提交
9902
                             _("tunnelled migration failed to accept from qemu"));
9903
        goto cancel;
C
Chris Lalancette 已提交
9904 9905
    }

9906
    retval = doTunnelSendAll(st, client_sock);
9907

9908
cancel:
9909
    if (retval != 0) {
9910
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
9911
        qemuMonitorMigrateCancel(priv->mon);
9912
        qemuDomainObjExitMonitorWithDriver(driver, vm);
9913
    }
C
Chris Lalancette 已提交
9914

9915
finish:
C
Chris Lalancette 已提交
9916 9917 9918
    dname = dname ? dname : dom->name;
    ddomain = dconn->driver->domainMigrateFinish2
        (dconn, dname, NULL, 0, uri, flags, retval);
9919 9920 9921 9922 9923 9924 9925

cleanup:
    if (client_sock != -1)
        close(client_sock);
    if (qemu_sock != -1)
        close(qemu_sock);

C
Chris Lalancette 已提交
9926 9927 9928
    if (ddomain)
        virUnrefDomain(ddomain);

9929 9930 9931 9932
    if (unixfile) {
        unlink(unixfile);
        VIR_FREE(unixfile);
    }
C
Chris Lalancette 已提交
9933

9934 9935 9936
    if (st)
        /* don't call virStreamFree(), because that resets any pending errors */
        virUnrefStream(st);
9937 9938 9939 9940
    return retval;
}


9941 9942 9943 9944
/* 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,
9945
                              struct qemud_driver *driver,
9946 9947 9948 9949 9950 9951 9952 9953 9954 9955 9956 9957 9958 9959 9960 9961 9962 9963 9964 9965 9966 9967 9968 9969 9970
                              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) {
9971 9972
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("domainMigratePrepare2 did not set uri"));
9973
        goto cleanup;
9974 9975
    }

9976
    if (doNativeMigrate(driver, vm, uri_out, flags, dname, resource) < 0)
9977 9978 9979 9980 9981 9982 9983 9984 9985 9986 9987 9988 9989 9990 9991 9992 9993
        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;
}


9994
static int doPeer2PeerMigrate(virDomainPtr dom,
9995
                              struct qemud_driver *driver,
9996 9997 9998 9999 10000 10001 10002 10003 10004 10005 10006 10007 10008 10009 10010 10011
                              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) {
10012 10013
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("Failed to connect to remote libvirt URI %s"), uri);
10014 10015 10016
        return -1;
    }
    if (!VIR_DRV_SUPPORTS_FEATURE(dconn->driver, dconn,
10017
                                  VIR_DRV_FEATURE_MIGRATION_P2P)) {
10018 10019
        qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
                        _("Destination libvirt does not support peer-to-peer migration protocol"));
10020 10021 10022
        goto cleanup;
    }

10023 10024 10025
    dom_xml = qemudVMDumpXML(driver, vm,
                             VIR_DOMAIN_XML_SECURE |
                             VIR_DOMAIN_XML_UPDATE_CPU);
10026
    if (!dom_xml) {
10027 10028
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to get domain xml"));
10029 10030 10031
        goto cleanup;
    }

10032
    if (flags & VIR_MIGRATE_TUNNELLED)
10033
        ret = doTunnelMigrate(dom, driver, dconn, vm, dom_xml, uri, flags, dname, resource);
10034
    else
10035
        ret = doNonTunnelMigrate(dom, driver, dconn, vm, dom_xml, uri, flags, dname, resource);
10036 10037 10038

cleanup:
    VIR_FREE(dom_xml);
C
Chris Lalancette 已提交
10039 10040 10041
    /* don't call virConnectClose(), because that resets any pending errors */
    virUnrefConnect(dconn);

10042
    return ret;
D
Daniel Veillard 已提交
10043 10044
}

10045

D
Daniel Veillard 已提交
10046 10047 10048 10049 10050 10051
/* 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,
10052
                           unsigned long flags,
10053
                           const char *dname,
D
Daniel Veillard 已提交
10054 10055
                           unsigned long resource)
{
10056 10057
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
10058
    virDomainEventPtr event = NULL;
10059
    int ret = -1;
10060
    int resume = 0;
10061
    qemuDomainObjPrivatePtr priv;
D
Daniel Veillard 已提交
10062

10063
    qemuDriverLock(driver);
10064
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
D
Daniel Veillard 已提交
10065
    if (!vm) {
10066 10067
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
10068 10069
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
10070
        goto cleanup;
D
Daniel Veillard 已提交
10071
    }
10072
    priv = vm->privateData;
D
Daniel Veillard 已提交
10073

10074 10075
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;
10076
    priv->jobActive = QEMU_JOB_MIGRATION;
10077

D
Daniel P. Berrange 已提交
10078
    if (!virDomainObjIsActive(vm)) {
10079 10080
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
10081
        goto endjob;
D
Daniel Veillard 已提交
10082 10083
    }

10084 10085 10086
    memset(&priv->jobInfo, 0, sizeof(priv->jobInfo));
    priv->jobInfo.type = VIR_DOMAIN_JOB_UNBOUNDED;

10087
    resume = vm->state == VIR_DOMAIN_RUNNING;
10088
    if (!(flags & VIR_MIGRATE_LIVE) && vm->state == VIR_DOMAIN_RUNNING) {
10089
        if (qemuDomainMigrateOffline(driver, vm) < 0)
10090
            goto endjob;
10091 10092
    }

10093
    if ((flags & (VIR_MIGRATE_TUNNELLED | VIR_MIGRATE_PEER2PEER))) {
10094
        if (doPeer2PeerMigrate(dom, driver, vm, uri, flags, dname, resource) < 0)
10095
            /* doPeer2PeerMigrate already set the error, so just get out */
10096
            goto endjob;
10097
    } else {
10098
        if (doNativeMigrate(driver, vm, uri, flags, dname, resource) < 0)
10099
            goto endjob;
10100 10101
    }

D
Daniel Veillard 已提交
10102
    /* Clean up the source domain. */
10103
    qemudShutdownVMDaemon(driver, vm);
10104
    resume = 0;
10105 10106 10107 10108

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_MIGRATED);
C
Chris Lalancette 已提交
10109
    if (!vm->persistent || (flags & VIR_MIGRATE_UNDEFINE_SOURCE)) {
10110
        virDomainDeleteConfig(driver->configDir, driver->autostartDir, vm);
10111 10112
        if (qemuDomainObjEndJob(vm) > 0)
            virDomainRemoveInactive(&driver->domains, vm);
10113 10114
        vm = NULL;
    }
10115
    ret = 0;
D
Daniel Veillard 已提交
10116

10117
endjob:
10118
    if (resume && vm->state == VIR_DOMAIN_PAUSED) {
10119
        /* we got here through some sort of failure; start the domain again */
10120
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
10121
        if (qemuMonitorStartCPUs(priv->mon, dom->conn) < 0) {
10122 10123 10124 10125
            /* 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
             */
10126
            VIR_ERROR(_("Failed to resume guest %s after failure"),
10127
                      vm->def->name);
10128
        }
10129
        qemuDomainObjExitMonitorWithDriver(driver, vm);
10130

10131
        vm->state = VIR_DOMAIN_RUNNING;
10132 10133 10134 10135
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_MIGRATED);
    }
10136 10137 10138
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
10139

10140
cleanup:
10141 10142
    if (vm)
        virDomainObjUnlock(vm);
10143 10144
    if (event)
        qemuDomainEventQueue(driver, event);
10145
    qemuDriverUnlock(driver);
10146
    return ret;
D
Daniel Veillard 已提交
10147 10148 10149 10150 10151 10152 10153 10154 10155
}

/* 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
10156
                           unsigned long flags,
D
Daniel Veillard 已提交
10157 10158
                           int retcode)
{
10159 10160 10161
    struct qemud_driver *driver = dconn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
10162
    virDomainEventPtr event = NULL;
10163
    virErrorPtr orig_err;
C
Chris Lalancette 已提交
10164
    int newVM = 1;
D
Daniel Veillard 已提交
10165

10166 10167 10168
    /* Migration failed. Save the current error so nothing squashes it */
    orig_err = virSaveLastError();

10169
    qemuDriverLock(driver);
10170
    vm = virDomainFindByName(&driver->domains, dname);
D
Daniel Veillard 已提交
10171
    if (!vm) {
10172 10173
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching name '%s'"), dname);
10174
        goto cleanup;
D
Daniel Veillard 已提交
10175 10176
    }

10177 10178 10179
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

D
Daniel Veillard 已提交
10180 10181 10182 10183
    /* Did the migration go as planned?  If yes, return the domain
     * object, but if no, clean up the empty qemu process.
     */
    if (retcode == 0) {
C
Chris Lalancette 已提交
10184 10185 10186 10187 10188
        if (flags & VIR_MIGRATE_PERSIST_DEST) {
            if (vm->persistent)
                newVM = 0;
            vm->persistent = 1;

10189
            if (virDomainSaveConfig(driver->configDir, vm->def) < 0) {
C
Chris Lalancette 已提交
10190 10191 10192 10193 10194 10195 10196 10197 10198 10199
                /* 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;
10200
                goto endjob;
C
Chris Lalancette 已提交
10201 10202 10203 10204 10205 10206 10207 10208 10209
            }

            event = virDomainEventNewFromObj(vm,
                                             VIR_DOMAIN_EVENT_DEFINED,
                                             newVM ?
                                             VIR_DOMAIN_EVENT_DEFINED_ADDED :
                                             VIR_DOMAIN_EVENT_DEFINED_UPDATED);
            if (event)
                qemuDomainEventQueue(driver, event);
10210
            event = NULL;
C
Chris Lalancette 已提交
10211 10212

        }
10213
        qemuDomainObjPrivatePtr priv = vm->privateData;
D
Daniel Veillard 已提交
10214
        dom = virGetDomain (dconn, vm->def->name, vm->def->uuid);
10215

10216 10217 10218 10219 10220 10221 10222 10223
        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)
10224 10225
                    qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                    "%s", _("resume operation failed"));
10226 10227 10228
                qemuDomainObjExitMonitorWithDriver(driver, vm);
                goto endjob;
            }
10229
            qemuDomainObjExitMonitorWithDriver(driver, vm);
10230 10231

            vm->state = VIR_DOMAIN_RUNNING;
10232 10233
        }

10234 10235 10236
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_MIGRATED);
10237 10238 10239 10240 10241 10242
        if (vm->state == VIR_DOMAIN_PAUSED) {
            qemuDomainEventQueue(driver, event);
            event = virDomainEventNewFromObj(vm,
                                             VIR_DOMAIN_EVENT_SUSPENDED,
                                             VIR_DOMAIN_EVENT_SUSPENDED_PAUSED);
        }
10243
        virDomainSaveStatus(driver->caps, driver->stateDir, vm);
D
Daniel Veillard 已提交
10244
    } else {
10245
        qemudShutdownVMDaemon(driver, vm);
10246 10247 10248
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         VIR_DOMAIN_EVENT_STOPPED_FAILED);
10249
        if (!vm->persistent) {
10250 10251
            if (qemuDomainObjEndJob(vm) > 0)
                virDomainRemoveInactive(&driver->domains, vm);
10252 10253
            vm = NULL;
        }
D
Daniel Veillard 已提交
10254
    }
10255

10256
endjob:
10257 10258 10259
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
10260

10261
cleanup:
10262 10263 10264 10265
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
10266 10267
    if (vm)
        virDomainObjUnlock(vm);
10268 10269
    if (event)
        qemuDomainEventQueue(driver, event);
10270
    qemuDriverUnlock(driver);
10271
    return dom;
D
Daniel Veillard 已提交
10272 10273
}

10274 10275 10276 10277 10278 10279 10280 10281 10282 10283 10284 10285 10286 10287 10288 10289
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;

10290
    def = virNodeDeviceDefParseString(xml, EXISTING_DEVICE);
10291 10292 10293 10294 10295 10296 10297 10298 10299 10300 10301 10302 10303 10304 10305 10306 10307
    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) {
10308 10309
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("device %s is not a PCI device"), dev->name);
10310 10311 10312 10313 10314 10315 10316 10317 10318 10319 10320 10321 10322 10323 10324 10325 10326 10327 10328 10329
        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;

10330
    pci = pciGetDevice(domain, bus, slot, function);
10331 10332 10333
    if (!pci)
        return -1;

10334
    if (pciDettachDevice(pci) < 0)
10335 10336 10337 10338
        goto out;

    ret = 0;
out:
10339
    pciFreeDevice(pci);
10340 10341 10342 10343 10344 10345 10346 10347 10348 10349 10350 10351 10352
    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;

10353
    pci = pciGetDevice(domain, bus, slot, function);
10354 10355 10356
    if (!pci)
        return -1;

10357
    if (pciReAttachDevice(pci) < 0)
10358 10359 10360 10361
        goto out;

    ret = 0;
out:
10362
    pciFreeDevice(pci);
10363 10364 10365 10366 10367 10368
    return ret;
}

static int
qemudNodeDeviceReset (virNodeDevicePtr dev)
{
10369
    struct qemud_driver *driver = dev->conn->privateData;
10370 10371 10372 10373 10374 10375 10376
    pciDevice *pci;
    unsigned domain, bus, slot, function;
    int ret = -1;

    if (qemudNodeDeviceGetPciInfo(dev, &domain, &bus, &slot, &function) < 0)
        return -1;

10377
    pci = pciGetDevice(domain, bus, slot, function);
10378 10379 10380
    if (!pci)
        return -1;

10381 10382
    qemuDriverLock(driver);

10383
    if (pciResetDevice(pci, driver->activePciHostdevs) < 0)
10384 10385 10386 10387
        goto out;

    ret = 0;
out:
10388
    qemuDriverUnlock(driver);
10389
    pciFreeDevice(pci);
10390 10391 10392
    return ret;
}

10393 10394 10395 10396 10397 10398 10399 10400 10401 10402 10403
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) {
10404 10405
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        "%s", _("cannot get host CPU capabilities"));
10406 10407
    }
    else
10408
        ret = cpuCompareXML(driver->caps->host.cpu, xmlDesc);
10409 10410 10411 10412 10413 10414

    qemuDriverUnlock(driver);

    return ret;
}

10415

10416 10417 10418 10419 10420 10421 10422 10423 10424 10425 10426 10427 10428
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;
}

10429 10430 10431 10432 10433 10434 10435 10436 10437 10438 10439 10440 10441 10442 10443 10444 10445 10446 10447 10448 10449 10450 10451 10452 10453 10454 10455 10456 10457 10458 10459 10460 10461 10462 10463 10464 10465 10466 10467 10468 10469 10470 10471

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;
}


10472 10473 10474 10475 10476 10477 10478 10479 10480 10481 10482 10483 10484 10485 10486 10487 10488 10489 10490 10491 10492 10493
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);
10494
            priv->jobSignals |= QEMU_JOB_SIGNAL_CANCEL;
10495 10496 10497 10498 10499 10500 10501 10502 10503 10504 10505 10506 10507 10508 10509 10510 10511 10512 10513 10514
        } 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;
}


10515 10516 10517 10518 10519 10520 10521 10522 10523 10524
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;

10525
    virCheckFlags(0, -1);
10526 10527 10528 10529 10530 10531 10532 10533 10534 10535 10536 10537 10538 10539 10540 10541 10542 10543 10544 10545 10546 10547 10548 10549 10550 10551 10552 10553 10554 10555 10556 10557 10558 10559 10560 10561 10562 10563

    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
10564 10565 10566 10567 10568 10569 10570 10571 10572 10573 10574 10575 10576 10577 10578 10579 10580 10581 10582 10583 10584 10585 10586 10587 10588 10589 10590 10591 10592 10593 10594 10595 10596 10597 10598 10599 10600 10601 10602 10603 10604 10605 10606 10607 10608 10609 10610 10611 10612 10613 10614 10615 10616 10617 10618 10619 10620 10621 10622 10623 10624 10625 10626 10627 10628 10629 10630 10631 10632 10633 10634 10635 10636 10637 10638 10639 10640 10641 10642 10643 10644 10645 10646 10647 10648 10649 10650 10651 10652 10653 10654 10655 10656 10657 10658 10659 10660 10661 10662 10663 10664 10665 10666 10667 10668 10669 10670 10671 10672 10673 10674 10675 10676 10677 10678 10679 10680 10681 10682 10683
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,
                            _("Disk device '%s' does not support snapshotting"),
                            vm->def->disks[i]->info.alias);
            return 0;
        }
    }

    return 1;
}

static virDomainSnapshotPtr qemuDomainSnapshotCreateXML(virDomainPtr domain,
                                                        const char *xmlDesc,
10684
                                                        unsigned int flags)
C
Chris Lalancette 已提交
10685 10686 10687 10688 10689 10690 10691 10692 10693 10694 10695
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm = NULL;
    virDomainSnapshotObjPtr snap = NULL;
    virDomainSnapshotPtr snapshot = NULL;
    char uuidstr[VIR_UUID_STRING_BUFLEN];
    virDomainSnapshotDefPtr def;
    qemuDomainObjPrivatePtr priv;
    const char *qemuimgarg[] = { NULL, "snapshot", "-c", NULL, NULL, NULL };
    int i;

10696 10697
    virCheckFlags(0, NULL);

C
Chris Lalancette 已提交
10698 10699 10700 10701 10702 10703 10704 10705 10706 10707 10708 10709 10710 10711 10712 10713 10714 10715 10716 10717 10718 10719 10720 10721 10722 10723 10724 10725 10726 10727 10728 10729 10730 10731 10732 10733 10734 10735 10736 10737 10738 10739 10740 10741 10742 10743 10744 10745 10746 10747 10748 10749 10750 10751 10752 10753 10754 10755 10756 10757 10758 10759 10760 10761 10762 10763 10764 10765 10766 10767 10768 10769 10770 10771 10772 10773 10774 10775 10776 10777 10778 10779 10780 10781 10782 10783 10784 10785 10786 10787 10788 10789 10790 10791 10792 10793 10794 10795 10796 10797 10798 10799 10800 10801 10802
    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 {
        priv = vm->privateData;
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
        if (qemuMonitorCreateSnapshot(priv->mon, def->name) < 0) {
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            goto cleanup;
        }
        qemuDomainObjExitMonitorWithDriver(driver, vm);

    }

    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,
10803
                                       unsigned int flags)
C
Chris Lalancette 已提交
10804 10805 10806 10807 10808
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm = NULL;
    int n = -1;

10809 10810
    virCheckFlags(0, -1);

C
Chris Lalancette 已提交
10811 10812 10813 10814 10815 10816 10817 10818 10819 10820 10821 10822 10823 10824 10825 10826 10827 10828 10829 10830
    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,
10831
                                 unsigned int flags)
C
Chris Lalancette 已提交
10832 10833 10834 10835 10836
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm = NULL;
    int n = -1;

10837 10838
    virCheckFlags(0, -1);

C
Chris Lalancette 已提交
10839 10840 10841 10842 10843 10844 10845 10846 10847 10848 10849 10850 10851 10852 10853 10854 10855 10856 10857 10858 10859
    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,
10860
                                                           unsigned int flags)
C
Chris Lalancette 已提交
10861 10862 10863 10864 10865 10866
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm;
    virDomainSnapshotObjPtr snap = NULL;
    virDomainSnapshotPtr snapshot = NULL;

10867 10868
    virCheckFlags(0, NULL);

C
Chris Lalancette 已提交
10869 10870 10871 10872 10873 10874 10875 10876 10877 10878 10879 10880 10881 10882 10883 10884 10885 10886 10887 10888 10889 10890 10891 10892 10893 10894 10895
    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,
10896
                                        unsigned int flags)
C
Chris Lalancette 已提交
10897 10898 10899 10900 10901
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;

10902 10903
    virCheckFlags(0, -1);

C
Chris Lalancette 已提交
10904 10905 10906 10907 10908 10909 10910 10911 10912 10913 10914 10915 10916 10917 10918 10919 10920 10921 10922 10923
    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,
10924
                                                      unsigned int flags)
C
Chris Lalancette 已提交
10925 10926 10927 10928 10929
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm;
    virDomainSnapshotPtr snapshot = NULL;

10930 10931
    virCheckFlags(0, NULL);

C
Chris Lalancette 已提交
10932 10933 10934 10935 10936 10937 10938 10939 10940 10941 10942 10943 10944 10945 10946 10947 10948 10949 10950 10951 10952 10953 10954 10955 10956 10957
    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,
10958
                                       unsigned int flags)
C
Chris Lalancette 已提交
10959 10960 10961 10962 10963 10964 10965
{
    struct qemud_driver *driver = snapshot->domain->conn->privateData;
    virDomainObjPtr vm = NULL;
    char *xml = NULL;
    virDomainSnapshotObjPtr snap = NULL;
    char uuidstr[VIR_UUID_STRING_BUFLEN];

10966 10967
    virCheckFlags(0, NULL);

C
Chris Lalancette 已提交
10968 10969 10970 10971 10972 10973 10974 10975 10976 10977 10978 10979 10980 10981 10982 10983 10984 10985 10986 10987 10988 10989 10990 10991 10992 10993 10994
    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,
10995
                                      unsigned int flags)
C
Chris Lalancette 已提交
10996 10997 10998 10999 11000 11001 11002 11003 11004 11005
{
    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;

11006 11007
    virCheckFlags(0, -1);

C
Chris Lalancette 已提交
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 11047 11048 11049 11050 11051 11052 11053 11054 11055 11056 11057 11058 11059 11060 11061 11062 11063 11064 11065 11066 11067 11068 11069 11070 11071 11072 11073 11074 11075 11076 11077 11078 11079 11080 11081 11082 11083 11084 11085 11086 11087 11088 11089 11090 11091 11092 11093 11094 11095 11096 11097 11098 11099 11100 11101 11102 11103 11104 11105 11106 11107 11108 11109 11110 11111 11112 11113 11114 11115 11116 11117 11118 11119 11120 11121 11122 11123 11124 11125 11126 11127 11128 11129 11130 11131 11132 11133 11134 11135 11136 11137 11138 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
    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)
                goto cleanup;
        }
        else {
            if (qemuDomainSnapshotSetActive(vm, driver->snapshotDir) < 0)
                goto cleanup;

            rc = qemudStartVMDaemon(snapshot->domain->conn, driver, vm, NULL,
                                    -1);
            if (qemuDomainSnapshotSetInactive(vm, driver->snapshotDir) < 0)
                goto cleanup;
            if (rc < 0)
                goto cleanup;
        }

        if (snap->def->state == VIR_DOMAIN_PAUSED) {
            /* qemu unconditionally starts the domain running again after
             * loadvm, so let's pause it to keep consistency
             */
            priv = vm->privateData;
            qemuDomainObjEnterMonitorWithDriver(driver, vm);
            rc = qemuMonitorStopCPUs(priv->mon);
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            if (rc < 0)
                goto cleanup;
        }

        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)) {
            qemudShutdownVMDaemon(driver, vm);
            event = virDomainEventNewFromObj(vm,
                                             VIR_DOMAIN_EVENT_STOPPED,
                                             VIR_DOMAIN_EVENT_STOPPED_FROM_SNAPSHOT);
        }

        if (qemuDomainSnapshotSetActive(vm, driver->snapshotDir) < 0)
            goto cleanup;
    }

    vm->state = snap->def->state;

    ret = 0;

cleanup:
    if (vm && qemuDomainObjEndJob(vm) == 0)
        vm = NULL;

    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;

11233 11234
    virCheckFlags(VIR_DOMAIN_SNAPSHOT_DELETE_CHILDREN, -1);

C
Chris Lalancette 已提交
11235 11236 11237 11238 11239 11240 11241 11242 11243 11244 11245 11246 11247 11248 11249 11250 11251 11252 11253 11254 11255 11256 11257 11258 11259 11260 11261 11262 11263 11264 11265 11266 11267 11268 11269 11270
    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;
    }

    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)
            goto cleanup;
    }

    ret = qemuDomainSnapshotDiscard(driver, vm, snap);

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);
    return ret;
}
11271

11272 11273 11274 11275 11276
static virDriver qemuDriver = {
    VIR_DRV_QEMU,
    "QEMU",
    qemudOpen, /* open */
    qemudClose, /* close */
D
Daniel Veillard 已提交
11277
    qemudSupportsFeature, /* supports_feature */
11278 11279
    qemudGetType, /* type */
    qemudGetVersion, /* version */
11280
    NULL, /* libvirtVersion (impl. in libvirt.c) */
11281
    virGetHostname, /* getHostname */
11282
    qemudGetMaxVCPUs, /* getMaxVcpus */
11283
    nodeGetInfo, /* nodeGetInfo */
11284 11285 11286
    qemudGetCapabilities, /* getCapabilities */
    qemudListDomains, /* listDomains */
    qemudNumDomains, /* numOfDomains */
11287
    qemudDomainCreate, /* domainCreateXML */
11288 11289 11290 11291 11292
    qemudDomainLookupByID, /* domainLookupByID */
    qemudDomainLookupByUUID, /* domainLookupByUUID */
    qemudDomainLookupByName, /* domainLookupByName */
    qemudDomainSuspend, /* domainSuspend */
    qemudDomainResume, /* domainResume */
11293
    qemudDomainShutdown, /* domainShutdown */
11294 11295 11296
    NULL, /* domainReboot */
    qemudDomainDestroy, /* domainDestroy */
    qemudDomainGetOSType, /* domainGetOSType */
11297
    qemudDomainGetMaxMemory, /* domainGetMaxMemory */
11298
    NULL, /* domainSetMaxMemory */
11299
    qemudDomainSetMemory, /* domainSetMemory */
11300 11301 11302
    qemudDomainGetInfo, /* domainGetInfo */
    qemudDomainSave, /* domainSave */
    qemudDomainRestore, /* domainRestore */
P
Paolo Bonzini 已提交
11303
    qemudDomainCoreDump, /* domainCoreDump */
11304
    qemudDomainSetVcpus, /* domainSetVcpus */
11305 11306
    qemudDomainPinVcpu, /* domainPinVcpu */
    qemudDomainGetVcpus, /* domainGetVcpus */
11307
    qemudDomainGetMaxVcpus, /* domainGetMaxVcpus */
11308 11309
    qemudDomainGetSecurityLabel, /* domainGetSecurityLabel */
    qemudNodeGetSecurityModel, /* nodeGetSecurityModel */
11310
    qemudDomainDumpXML, /* domainDumpXML */
11311
    qemuDomainXMLFromNative, /* domainXmlFromNative */
11312
    qemuDomainXMLToNative, /* domainXMLToNative */
11313 11314
    qemudListDefinedDomains, /* listDefinedDomains */
    qemudNumDefinedDomains, /* numOfDefinedDomains */
11315 11316 11317
    qemudDomainStart, /* domainCreate */
    qemudDomainDefine, /* domainDefineXML */
    qemudDomainUndefine, /* domainUndefine */
11318
    qemudDomainAttachDevice, /* domainAttachDevice */
11319
    qemudDomainAttachDeviceFlags, /* domainAttachDeviceFlags */
11320
    qemudDomainDetachDevice, /* domainDetachDevice */
11321
    qemudDomainDetachDeviceFlags, /* domainDetachDeviceFlags */
11322
    qemuDomainUpdateDeviceFlags, /* domainUpdateDeviceFlags */
11323 11324
    qemudDomainGetAutostart, /* domainGetAutostart */
    qemudDomainSetAutostart, /* domainSetAutostart */
11325 11326 11327
    qemuGetSchedulerType, /* domainGetSchedulerType */
    qemuGetSchedulerParameters, /* domainGetSchedulerParameters */
    qemuSetSchedulerParameters, /* domainSetSchedulerParameters */
D
Daniel Veillard 已提交
11328 11329
    NULL, /* domainMigratePrepare (v1) */
    qemudDomainMigratePerform, /* domainMigratePerform */
11330
    NULL, /* domainMigrateFinish */
11331
    qemudDomainBlockStats, /* domainBlockStats */
11332
    qemudDomainInterfaceStats, /* domainInterfaceStats */
11333
    qemudDomainMemoryStats, /* domainMemoryStats */
11334
    qemudDomainBlockPeek, /* domainBlockPeek */
R
Richard W.M. Jones 已提交
11335
    qemudDomainMemoryPeek, /* domainMemoryPeek */
11336 11337
    nodeGetCellsFreeMemory, /* nodeGetCellsFreeMemory */
    nodeGetFreeMemory,  /* getFreeMemory */
11338 11339
    qemuDomainEventRegister, /* domainEventRegister */
    qemuDomainEventDeregister, /* domainEventDeregister */
D
Daniel Veillard 已提交
11340 11341
    qemudDomainMigratePrepare2, /* domainMigratePrepare2 */
    qemudDomainMigrateFinish2, /* domainMigrateFinish2 */
11342 11343 11344
    qemudNodeDeviceDettach, /* nodeDeviceDettach */
    qemudNodeDeviceReAttach, /* nodeDeviceReAttach */
    qemudNodeDeviceReset, /* nodeDeviceReset */
C
Chris Lalancette 已提交
11345
    qemudDomainMigratePrepareTunnel, /* domainMigratePrepareTunnel */
11346 11347 11348 11349
    qemuIsEncrypted, /* isEncrypted */
    qemuIsSecure, /* isSecure */
    qemuDomainIsActive, /* domainIsActive */
    qemuDomainIsPersistent, /* domainIsPersistent */
11350
    qemuCPUCompare, /* cpuCompare */
11351
    qemuCPUBaseline, /* cpuBaseline */
11352
    qemuDomainGetJobInfo, /* domainGetJobInfo */
11353
    qemuDomainAbortJob, /* domainAbortJob */
11354
    qemuDomainMigrateSetMaxDowntime, /* domainMigrateSetMaxDowntime */
11355 11356
    qemuDomainEventRegisterAny, /* domainEventRegisterAny */
    qemuDomainEventDeregisterAny, /* domainEventDeregisterAny */
11357 11358 11359
    qemuDomainManagedSave, /* domainManagedSave */
    qemuDomainHasManagedSaveImage, /* domainHasManagedSaveImage */
    qemuDomainManagedSaveRemove, /* domainManagedSaveRemove */
C
Chris Lalancette 已提交
11360 11361 11362 11363 11364 11365 11366 11367 11368
    qemuDomainSnapshotCreateXML, /* domainSnapshotCreateXML */
    qemuDomainSnapshotDumpXML, /* domainSnapshotDumpXML */
    qemuDomainSnapshotNum, /* domainSnapshotNum */
    qemuDomainSnapshotListNames, /* domainSnapshotListNames */
    qemuDomainSnapshotLookupByName, /* domainSnapshotLookupByName */
    qemuDomainHasCurrentSnapshot, /* domainHasCurrentSnapshot */
    qemuDomainSnapshotCurrent, /* domainSnapshotCurrent */
    qemuDomainRevertToSnapshot, /* domainRevertToSnapshot */
    qemuDomainSnapshotDelete, /* domainSnapshotDelete */
11369 11370 11371
};


11372
static virStateDriver qemuStateDriver = {
11373
    .name = "QEMU",
11374 11375 11376 11377
    .initialize = qemudStartup,
    .cleanup = qemudShutdown,
    .reload = qemudReload,
    .active = qemudActive,
11378
};
11379

S
Stefan Berger 已提交
11380 11381 11382 11383 11384 11385 11386 11387 11388 11389 11390 11391 11392 11393 11394
static int
qemudVMFilterRebuild(virConnectPtr conn,
                     virHashIterator iter, void *data)
{
    (void)conn;
    virHashForEach(qemu_driver->domains.objs, iter, data);
    return 0;
}


static virNWFilterCallbackDriver qemuCallbackDriver = {
    .name = "QEMU",
    .vmFilterRebuild = qemudVMFilterRebuild,
};

11395
int qemuRegister(void) {
11396 11397
    virRegisterDriver(&qemuDriver);
    virRegisterStateDriver(&qemuStateDriver);
S
Stefan Berger 已提交
11398
    virNWFilterRegisterCallbackDriver(&qemuCallbackDriver);
11399 11400
    return 0;
}