qemu_driver.c 344.9 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_monitor_text.h"
64
#include "qemu_bridge_filter.h"
J
Jim Meyering 已提交
65
#include "c-ctype.h"
66
#include "event.h"
67
#include "buf.h"
68
#include "util.h"
69
#include "nodeinfo.h"
70
#include "stats_linux.h"
71
#include "capabilities.h"
72
#include "memory.h"
73
#include "uuid.h"
74
#include "domain_conf.h"
75 76
#include "node_device_conf.h"
#include "pci.h"
77
#include "hostusb.h"
78
#include "processinfo.h"
79 80
#include "qemu_security_stacked.h"
#include "qemu_security_dac.h"
81
#include "cgroup.h"
C
Chris Lalancette 已提交
82
#include "libvirt_internal.h"
83
#include "xml.h"
84
#include "cpu/cpu.h"
85
#include "macvtap.h"
S
Stefan Berger 已提交
86
#include "nwfilter/nwfilter_gentech_driver.h"
87
#include "hooks.h"
88

89

90 91
#define VIR_FROM_THIS VIR_FROM_QEMU

92 93 94 95 96 97 98 99 100 101 102 103
/* 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 */
104 105 106 107 108
    QEMU_JOB_SIGNAL_MIGRATE_DOWNTIME = 1 << 2, /* Request migration downtime change */
};

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

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

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

    int nvcpupids;
    int *vcpupids;
127 128

    qemuDomainPCIAddressSetPtr pciaddrs;
129
    int persistentAddrs;
130 131
};

132 133
static int qemudShutdown(void);

134 135
static void qemuDriverLock(struct qemud_driver *driver)
{
136
    virMutexLock(&driver->lock);
137 138 139
}
static void qemuDriverUnlock(struct qemud_driver *driver)
{
140
    virMutexUnlock(&driver->lock);
141 142
}

143 144 145
static void qemuDomainEventFlush(int timer, void *opaque);
static void qemuDomainEventQueue(struct qemud_driver *driver,
                                 virDomainEventPtr event);
146

147 148
static int qemudStartVMDaemon(virConnectPtr conn,
                              struct qemud_driver *driver,
149
                              virDomainObjPtr vm,
150 151
                              const char *migrateFrom,
                              int stdin_fd);
152

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

156
static int qemudDomainGetMaxVcpus(virDomainPtr dom);
157

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

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

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

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

static void *qemuDomainObjPrivateAlloc(void)
{
    qemuDomainObjPrivatePtr priv;

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

    return priv;
}

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

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

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


194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211
static int qemuDomainObjPrivateXMLFormat(virBufferPtr buf, void *data)
{
    qemuDomainObjPrivatePtr priv = data;
    const char *monitorpath;

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

        virBufferEscapeString(buf, "  <monitor path='%s'", monitorpath);
D
Daniel P. Berrange 已提交
212 213
        if (priv->monJSON)
            virBufferAddLit(buf, " json='1'");
214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239
        virBufferVSprintf(buf, " type='%s'/>\n",
                          virDomainChrTypeToString(priv->monConfig->type));
    }


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

    return 0;
}

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

    if (VIR_ALLOC(priv->monConfig) < 0) {
240
        virReportOOMError();
241 242 243
        goto error;
    }

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

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

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

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

269 270 271 272 273 274 275 276 277
    switch (priv->monConfig->type) {
    case VIR_DOMAIN_CHR_TYPE_PTY:
        priv->monConfig->data.file.path = monitorpath;
        break;
    case VIR_DOMAIN_CHR_TYPE_UNIX:
        priv->monConfig->data.nix.path = monitorpath;
        break;
    default:
        VIR_FREE(monitorpath);
278 279 280
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("unsupported monitor type '%s'"),
                        virDomainChrTypeToString(priv->monConfig->type));
281 282 283
        goto error;
    }

284
    n = virXPathNodeSet("./vcpus/vcpu", ctxt, &nodes);
285 286 287 288 289
    if (n < 0)
        goto error;
    if (n) {
        priv->nvcpupids = n;
        if (VIR_REALLOC_N(priv->vcpupids, priv->nvcpupids) < 0) {
290
            virReportOOMError();
291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310
            goto error;
        }

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

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

    return 0;

error:
311 312
    virDomainChrDefFree(priv->monConfig);
    priv->monConfig = NULL;
313 314 315 316 317 318
    VIR_FREE(nodes);
    return -1;
}



319 320 321 322 323 324 325 326 327
/*
 * obj must be locked before calling, qemud_driver must NOT be locked
 *
 * This must be called by anything that will change the VM state
 * in any way, or anything that will use the QEMU monitor.
 *
 * Upon successful return, the object will have its ref count increased,
 * successful calls must be followed by EndJob eventually
 */
328 329 330 331

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

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

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

    virDomainObjRef(obj);

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

    return 0;
}

/*
 * obj must be locked before calling, qemud_driver must be locked
 *
 * This must be called by anything that will change the VM state
 * in any way, or anything that will use the QEMU monitor.
 */
static int qemuDomainObjBeginJobWithDriver(struct qemud_driver *driver,
                                           virDomainObjPtr obj) ATTRIBUTE_RETURN_CHECK;
static int qemuDomainObjBeginJobWithDriver(struct qemud_driver *driver,
                                           virDomainObjPtr obj)
{
    qemuDomainObjPrivatePtr priv = obj->privateData;
382 383 384 385
    struct timeval now;
    unsigned long long then;

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

    virDomainObjRef(obj);
    qemuDriverUnlock(driver);

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

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

    return 0;
}

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

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

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


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

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


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

    refs = qemuMonitorUnref(priv->mon);

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

478
    virDomainObjLock(obj);
479 480 481 482 483

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


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

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


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

    refs = qemuMonitorUnref(priv->mon);

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

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

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


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

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

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

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

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


578
static int
579
qemudLogReadFD(const char* logDir, const char* name, off_t pos)
580 581 582 583 584 585 586
{
    char logfile[PATH_MAX];
    mode_t logmode = O_RDONLY;
    int ret, fd = -1;

    if ((ret = snprintf(logfile, sizeof(logfile), "%s/%s.log", logDir, name))
        < 0 || ret >= sizeof(logfile)) {
587 588 589
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("failed to build logfile name %s/%s.log"),
                        logDir, name);
590 591 592 593 594
        return -1;
    }


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


616 617 618 619 620 621 622 623 624 625 626 627
struct qemuAutostartData {
    struct qemud_driver *driver;
    virConnectPtr conn;
};
static void
qemuAutostartDomain(void *payload, const char *name ATTRIBUTE_UNUSED, void *opaque)
{
    virDomainObjPtr vm = payload;
    struct qemuAutostartData *data = opaque;

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

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

650 651
static void
qemudAutostartConfigs(struct qemud_driver *driver) {
652 653 654 655 656
    /* XXX: Figure out a better way todo this. The domain
     * startup code needs a connection handle in order
     * to lookup the bridge associated with a virtual
     * network
     */
657 658 659
    virConnectPtr conn = virConnectOpen(driver->privileged ?
                                        "qemu:///system" :
                                        "qemu:///session");
660
    /* Ignoring NULL conn which is mostly harmless here */
661
    struct qemuAutostartData data = { driver, conn };
662

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

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

671 672 673 674 675 676 677 678 679

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

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

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

696 697 698
    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)));
699

D
Daniel Veillard 已提交
700

701
    return 0;
702 703
}

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

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

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

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

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

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


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

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

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

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

761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780
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;
}

781
static int
782 783 784 785
getVolumeQcowPassphrase(virConnectPtr conn,
                        virDomainDiskDefPtr disk,
                        char **secretRet,
                        size_t *secretLen)
786 787 788 789 790
{
    virSecretPtr secret;
    char *passphrase;
    unsigned char *data;
    size_t size;
791
    int ret = -1;
792
    virStorageEncryptionPtr enc;
793

794 795 796 797 798 799 800
    if (!disk->encryption) {
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("disk %s does not have any encryption information"),
                        disk->src);
        return -1;
    }
    enc = disk->encryption;
801 802

    if (!conn) {
803 804
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        "%s", _("cannot find secrets without a connection"));
805
        goto cleanup;
806 807 808 809 810
    }

    if (conn->secretDriver == NULL ||
        conn->secretDriver->lookupByUUID == NULL ||
        conn->secretDriver->getValue == NULL) {
811 812
        qemuReportError(VIR_ERR_NO_SUPPORT, "%s",
                        _("secret storage not supported"));
813
        goto cleanup;
814 815 816 817 818 819
    }

    if (enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_QCOW ||
        enc->nsecrets != 1 ||
        enc->secrets[0]->type !=
        VIR_STORAGE_ENCRYPTION_SECRET_TYPE_PASSPHRASE) {
820
        qemuReportError(VIR_ERR_INVALID_DOMAIN,
821
                        _("invalid <encryption> for volume %s"), disk->src);
822
        goto cleanup;
823 824 825 826 827
    }

    secret = conn->secretDriver->lookupByUUID(conn,
                                              enc->secrets[0]->uuid);
    if (secret == NULL)
828
        goto cleanup;
829 830 831 832
    data = conn->secretDriver->getValue(secret, &size,
                                        VIR_SECRET_GET_VALUE_INTERNAL_CALL);
    virUnrefSecret(secret);
    if (data == NULL)
833
        goto cleanup;
834 835 836 837

    if (memchr(data, '\0', size) != NULL) {
        memset(data, 0, size);
        VIR_FREE(data);
838 839
        qemuReportError(VIR_ERR_INVALID_SECRET,
                        _("format='qcow' passphrase for %s must not contain a "
840
                          "'\\0'"), disk->src);
841
        goto cleanup;
842 843 844 845 846
    }

    if (VIR_ALLOC_N(passphrase, size + 1) < 0) {
        memset(data, 0, size);
        VIR_FREE(data);
847
        virReportOOMError();
848
        goto cleanup;
849 850 851 852 853 854 855 856 857 858
    }
    memcpy(passphrase, data, size);
    passphrase[size] = '\0';

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

    *secretRet = passphrase;
    *secretLen = size;

859 860 861
    ret = 0;

cleanup:
862 863
    return ret;
}
864

865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885
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);
886
    return ret;
887 888
}

889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910

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;
}


911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942
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;
}


943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971
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;
}


972 973 974 975 976 977
static int
qemuHandleDomainWatchdog(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                         virDomainObjPtr vm,
                         int action)
{
    struct qemud_driver *driver = qemu_driver;
978 979
    virDomainEventPtr watchdogEvent = NULL;
    virDomainEventPtr lifecycleEvent = NULL;
980 981

    virDomainObjLock(vm);
982 983 984 985 986 987 988 989 990 991 992 993 994 995
    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);
    }
996 997
    virDomainObjUnlock(vm);

998
    if (watchdogEvent || lifecycleEvent) {
999
        qemuDriverLock(driver);
1000 1001 1002 1003
        if (watchdogEvent)
            qemuDomainEventQueue(driver, watchdogEvent);
        if (lifecycleEvent)
            qemuDomainEventQueue(driver, lifecycleEvent);
1004 1005 1006 1007 1008 1009 1010
        qemuDriverUnlock(driver);
    }

    return 0;
}


1011 1012 1013 1014 1015 1016 1017
static int
qemuHandleDomainIOError(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
                        virDomainObjPtr vm,
                        const char *diskAlias,
                        int action)
{
    struct qemud_driver *driver = qemu_driver;
1018 1019
    virDomainEventPtr ioErrorEvent = NULL;
    virDomainEventPtr lifecycleEvent = NULL;
1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034
    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 = "";
    }

1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048
    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);
    }
1049 1050
    virDomainObjUnlock(vm);

1051
    if (ioErrorEvent || lifecycleEvent) {
1052
        qemuDriverLock(driver);
1053 1054 1055 1056
        if (ioErrorEvent)
            qemuDomainEventQueue(driver, ioErrorEvent);
        if (lifecycleEvent)
            qemuDomainEventQueue(driver, lifecycleEvent);
1057 1058 1059 1060 1061 1062 1063
        qemuDriverUnlock(driver);
    }

    return 0;
}


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 1155
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;
}


1156 1157 1158
static qemuMonitorCallbacks monitorCallbacks = {
    .eofNotify = qemuHandleMonitorEOF,
    .diskSecretLookup = findVolumeQcowPassphrase,
1159
    .domainStop = qemuHandleDomainStop,
1160
    .domainReset = qemuHandleDomainReset,
1161
    .domainRTCChange = qemuHandleDomainRTCChange,
1162
    .domainWatchdog = qemuHandleDomainWatchdog,
1163
    .domainIOError = qemuHandleDomainIOError,
1164
    .domainGraphics = qemuHandleDomainGraphics,
1165 1166
};

1167
static int
1168
qemuConnectMonitor(struct qemud_driver *driver, virDomainObjPtr vm)
1169
{
1170
    qemuDomainObjPrivatePtr priv = vm->privateData;
1171
    int ret;
1172

1173 1174 1175 1176
    /* Hold an extra reference because we can't allow 'vm' to be
     * deleted while the monitor is active */
    virDomainObjRef(vm);

1177 1178
    if ((priv->mon = qemuMonitorOpen(vm,
                                     priv->monConfig,
D
Daniel P. Berrange 已提交
1179
                                     priv->monJSON,
1180
                                     &monitorCallbacks)) == NULL) {
1181
        VIR_ERROR(_("Failed to connect monitor for %s"), vm->def->name);
1182
        return -1;
1183
    }
1184

1185 1186 1187 1188 1189 1190 1191 1192 1193 1194
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
    ret = qemuMonitorSetCapabilities(priv->mon);
    qemuDomainObjExitMonitorWithDriver(driver, vm);

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

    return ret;
1195
}
1196 1197 1198 1199

/*
 * Open an existing VM's monitor, re-detect VCPU threads
 * and re-reserve the security labels in use
1200
 */
1201 1202
static void
qemuReconnectDomain(void *payload, const char *name ATTRIBUTE_UNUSED, void *opaque)
1203
{
1204 1205
    virDomainObjPtr obj = payload;
    struct qemud_driver *driver = opaque;
1206
    qemuDomainObjPrivatePtr priv;
1207
    unsigned long long qemuCmdFlags;
1208 1209

    virDomainObjLock(obj);
1210

1211 1212
    VIR_DEBUG("Reconnect monitor to %p '%s'", obj, obj->def->name);

1213 1214
    priv = obj->privateData;

1215
    /* XXX check PID liveliness & EXE path */
1216
    if (qemuConnectMonitor(driver, obj) < 0)
1217
        goto error;
1218

1219 1220 1221 1222
    if (qemuUpdateActivePciHostdevs(driver, obj->def) < 0) {
        goto error;
    }

1223 1224 1225 1226 1227 1228 1229 1230 1231
    /* 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;

1232 1233 1234
    if (!(priv->pciaddrs = qemuDomainPCIAddressSetCreate(obj->def)))
        goto error;

1235
    if (driver->securityDriver &&
1236
        driver->securityDriver->domainReserveSecurityLabel &&
1237
        driver->securityDriver->domainReserveSecurityLabel(obj) < 0)
1238
        goto error;
1239

1240 1241
    if (obj->def->id >= driver->nextvmid)
        driver->nextvmid = obj->def->id + 1;
1242

1243 1244
    virDomainObjUnlock(obj);
    return;
1245

1246
error:
1247 1248 1249
    /* 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 */
1250
    qemudShutdownVMDaemon(driver, obj);
1251 1252 1253 1254
    if (!obj->persistent)
        virDomainRemoveInactive(&driver->domains, obj);
    else
        virDomainObjUnlock(obj);
1255
}
1256

1257
/**
1258
 * qemudReconnectDomains
1259 1260 1261 1262 1263 1264 1265
 *
 * Try to re-open the resources for live VMs that we care
 * about.
 */
static void
qemuReconnectDomains(struct qemud_driver *driver)
{
1266
    virHashForEach(driver->domains.objs, qemuReconnectDomain, driver);
1267 1268
}

1269

1270 1271 1272 1273 1274 1275
static int
qemudSecurityInit(struct qemud_driver *qemud_drv)
{
    int ret;
    virSecurityDriverPtr security_drv;

1276 1277 1278
    qemuSecurityStackedSetDriver(qemud_drv);
    qemuSecurityDACSetDriver(qemud_drv);

1279 1280 1281 1282 1283 1284
    ret = virSecurityDriverStartup(&security_drv,
                                   qemud_drv->securityDriverName);
    if (ret == -1) {
        VIR_ERROR0(_("Failed to start security driver"));
        return -1;
    }
1285 1286 1287

    /* No primary security driver wanted to be enabled: just setup
     * the DAC driver on its own */
1288
    if (ret == -2) {
1289
        qemud_drv->securityDriver = &qemuDACSecurityDriver;
1290
        VIR_INFO0(_("No security driver available"));
1291 1292 1293 1294 1295
    } else {
        qemud_drv->securityPrimaryDriver = security_drv;
        qemud_drv->securitySecondaryDriver = &qemuDACSecurityDriver;
        qemud_drv->securityDriver = &qemuStackedSecurityDriver;
        VIR_INFO("Initialized security driver %s", security_drv->name);
1296 1297
    }

1298
    return 0;
1299
}
1300 1301


1302 1303
static virCapsPtr
qemuCreateCapabilities(virCapsPtr oldcaps,
1304
                       struct qemud_driver *driver)
1305 1306 1307 1308 1309
{
    virCapsPtr caps;

    /* Basic host arch / guest machine capabilities */
    if (!(caps = qemudCapsInit(oldcaps))) {
1310
        virReportOOMError();
1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321
        return NULL;
    }

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


    /* Security driver data */
1322
    if (driver->securityPrimaryDriver) {
1323 1324
        const char *doi, *model;

1325 1326
        doi = virSecurityDriverGetDOI(driver->securityPrimaryDriver);
        model = virSecurityDriverGetModel(driver->securityPrimaryDriver);
1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339

        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:
1340
    virReportOOMError();
1341 1342 1343
    virCapabilitiesFree(caps);
    return NULL;
}
1344

C
Chris Lalancette 已提交
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 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431
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;
    virDomainSnapshotObjPtr snap = NULL;
    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;
        }

        snap = virDomainSnapshotAssignDef(&vm->snapshots, def);

        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);
}

1432 1433 1434 1435 1436 1437
/**
 * qemudStartup:
 *
 * Initialization function for the QEmu daemon
 */
static int
1438
qemudStartup(int privileged) {
1439
    char *base = NULL;
D
Daniel P. Berrange 已提交
1440
    char driverConf[PATH_MAX];
1441
    int rc;
1442

1443
    if (VIR_ALLOC(qemu_driver) < 0)
1444 1445
        return -1;

1446
    if (virMutexInit(&qemu_driver->lock) < 0) {
1447
        VIR_ERROR("%s", _("cannot initialize mutex"));
1448 1449 1450
        VIR_FREE(qemu_driver);
        return -1;
    }
1451
    qemuDriverLock(qemu_driver);
1452
    qemu_driver->privileged = privileged;
1453

1454 1455 1456
    /* Don't have a dom0 so start from 1 */
    qemu_driver->nextvmid = 1;

1457 1458 1459
    if (virDomainObjListInit(&qemu_driver->domains) < 0)
        goto out_of_memory;

1460
    /* Init callback list */
1461
    if (VIR_ALLOC(qemu_driver->domainEventCallbacks) < 0)
1462
        goto out_of_memory;
1463 1464 1465 1466 1467 1468
    if (!(qemu_driver->domainEventQueue = virDomainEventQueueNew()))
        goto out_of_memory;

    if ((qemu_driver->domainEventTimer =
         virEventAddTimeout(-1, qemuDomainEventFlush, qemu_driver, NULL)) < 0)
        goto error;
1469

1470
    if (privileged) {
1471 1472
        if (virAsprintf(&qemu_driver->logDir,
                        "%s/log/libvirt/qemu", LOCAL_STATE_DIR) == -1)
1473
            goto out_of_memory;
1474

D
Daniel P. Berrange 已提交
1475
        if ((base = strdup (SYSCONF_DIR "/libvirt")) == NULL)
1476
            goto out_of_memory;
1477 1478

        if (virAsprintf(&qemu_driver->stateDir,
1479
                      "%s/run/libvirt/qemu", LOCAL_STATE_DIR) == -1)
1480
            goto out_of_memory;
1481 1482 1483 1484 1485 1486 1487 1488

        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;
1489 1490 1491
        if (virAsprintf(&qemu_driver->saveDir,
                      "%s/lib/libvirt/qemu/save/", LOCAL_STATE_DIR) == -1)
            goto out_of_memory;
C
Chris Lalancette 已提交
1492 1493 1494
        if (virAsprintf(&qemu_driver->snapshotDir,
                        "%s/lib/libvirt/qemu/snapshot", LOCAL_STATE_DIR) == -1)
            goto out_of_memory;
1495
    } else {
1496
        uid_t uid = geteuid();
1497
        char *userdir = virGetUserDirectory(uid);
1498
        if (!userdir)
1499
            goto error;
1500

1501
        if (virAsprintf(&qemu_driver->logDir,
1502 1503
                        "%s/.libvirt/qemu/log", userdir) == -1) {
            VIR_FREE(userdir);
1504
            goto out_of_memory;
1505
        }
1506

1507 1508
        if (virAsprintf(&base, "%s/.libvirt", userdir) == -1) {
            VIR_FREE(userdir);
1509
            goto out_of_memory;
1510 1511
        }
        VIR_FREE(userdir);
1512 1513 1514

        if (virAsprintf(&qemu_driver->stateDir, "%s/qemu/run", base) == -1)
            goto out_of_memory;
1515 1516 1517 1518
        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;
1519 1520
        if (virAsprintf(&qemu_driver->saveDir, "%s/qemu/save", base) == -1)
            goto out_of_memory;
C
Chris Lalancette 已提交
1521 1522
        if (virAsprintf(&qemu_driver->snapshotDir, "%s/qemu/snapshot", base) == -1)
            goto out_of_memory;
1523 1524
    }

L
Laine Stump 已提交
1525
    if (virFileMakePath(qemu_driver->stateDir) != 0) {
1526
        char ebuf[1024];
1527
        VIR_ERROR(_("Failed to create state dir '%s': %s"),
1528
                  qemu_driver->stateDir, virStrerror(errno, ebuf, sizeof ebuf));
1529
        goto error;
1530
    }
L
Laine Stump 已提交
1531
    if (virFileMakePath(qemu_driver->libDir) != 0) {
1532
        char ebuf[1024];
1533
        VIR_ERROR(_("Failed to create lib dir '%s': %s"),
1534 1535 1536
                  qemu_driver->libDir, virStrerror(errno, ebuf, sizeof ebuf));
        goto error;
    }
L
Laine Stump 已提交
1537
    if (virFileMakePath(qemu_driver->cacheDir) != 0) {
1538
        char ebuf[1024];
1539
        VIR_ERROR(_("Failed to create cache dir '%s': %s"),
1540 1541 1542
                  qemu_driver->cacheDir, virStrerror(errno, ebuf, sizeof ebuf));
        goto error;
    }
1543 1544 1545 1546 1547 1548
    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
1549 1550 1551 1552 1553 1554
    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;
    }
1555 1556 1557 1558

    /* Configuration paths are either ~/.libvirt/qemu/... (session) or
     * /etc/libvirt/qemu/... (system).
     */
D
Daniel P. Berrange 已提交
1559
    if (snprintf (driverConf, sizeof(driverConf), "%s/qemu.conf", base) == -1)
1560
        goto out_of_memory;
D
Daniel P. Berrange 已提交
1561
    driverConf[sizeof(driverConf)-1] = '\0';
1562

1563
    if (virAsprintf(&qemu_driver->configDir, "%s/qemu", base) == -1)
1564 1565
        goto out_of_memory;

1566
    if (virAsprintf(&qemu_driver->autostartDir, "%s/qemu/autostart", base) == -1)
1567 1568
        goto out_of_memory;

1569
    VIR_FREE(base);
1570

1571 1572 1573 1574 1575 1576 1577
    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)));
    }

1578 1579 1580 1581
    if (qemudLoadDriverConfig(qemu_driver, driverConf) < 0) {
        goto error;
    }

1582 1583
    if (qemudSecurityInit(qemu_driver) < 0)
        goto error;
D
Daniel P. Berrange 已提交
1584

1585
    if ((qemu_driver->caps = qemuCreateCapabilities(NULL,
1586
                                                    qemu_driver)) == NULL)
1587
        goto error;
1588

1589
    if ((qemu_driver->activePciHostdevs = pciDeviceListNew()) == NULL)
1590 1591
        goto error;

1592 1593
    if (privileged) {
        if (chown(qemu_driver->libDir, qemu_driver->user, qemu_driver->group) < 0) {
1594
            virReportSystemError(errno,
1595 1596 1597 1598 1599
                                 _("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) {
1600
            virReportSystemError(errno,
1601 1602 1603 1604
                                 _("unable to set ownership of '%s' to %d:%d"),
                                 qemu_driver->cacheDir, qemu_driver->user, qemu_driver->group);
            goto error;
        }
1605 1606 1607 1608 1609 1610
        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
1611 1612 1613 1614 1615 1616
        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;
        }
1617 1618
    }

1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632
    /* 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) {
1633
            virReportSystemError(rc,
1634 1635 1636 1637 1638 1639
                                 _("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) {
1640
            virReportSystemError(errno,
1641 1642 1643 1644 1645 1646 1647 1648 1649
                                 _("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;
    }

1650
    /* Get all the running persistent or transient configs first */
1651
    if (virDomainLoadAllConfigs(qemu_driver->caps,
1652 1653 1654 1655 1656 1657 1658 1659 1660
                                &qemu_driver->domains,
                                qemu_driver->stateDir,
                                NULL,
                                1, NULL, NULL) < 0)
        goto error;

    qemuReconnectDomains(qemu_driver);

    /* Then inactive persistent configs */
1661
    if (virDomainLoadAllConfigs(qemu_driver->caps,
1662 1663
                                &qemu_driver->domains,
                                qemu_driver->configDir,
1664
                                qemu_driver->autostartDir,
1665
                                0, NULL, NULL) < 0)
1666
        goto error;
C
Chris Lalancette 已提交
1667 1668 1669 1670 1671


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

1672 1673
    qemuDriverUnlock(qemu_driver);

1674 1675
    qemudAutostartConfigs(qemu_driver);

1676

1677 1678
    return 0;

1679
out_of_memory:
1680
    virReportOOMError();
1681 1682 1683
error:
    if (qemu_driver)
        qemuDriverUnlock(qemu_driver);
1684
    VIR_FREE(base);
1685
    qemudShutdown();
1686 1687 1688
    return -1;
}

1689 1690 1691 1692
static void qemudNotifyLoadDomain(virDomainObjPtr vm, int newVM, void *opaque)
{
    struct qemud_driver *driver = opaque;

1693 1694 1695 1696 1697 1698 1699 1700
    if (newVM) {
        virDomainEventPtr event =
            virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_DEFINED,
                                     VIR_DOMAIN_EVENT_DEFINED_ADDED);
        if (event)
            qemuDomainEventQueue(driver, event);
    }
1701 1702
}

1703 1704 1705 1706 1707 1708 1709 1710
/**
 * qemudReload:
 *
 * Function to restart the QEmu daemon, it will recheck the configuration
 * files and update its state and the networking
 */
static int
qemudReload(void) {
1711 1712 1713
    if (!qemu_driver)
        return 0;

1714
    qemuDriverLock(qemu_driver);
1715
    virDomainLoadAllConfigs(qemu_driver->caps,
1716 1717
                            &qemu_driver->domains,
                            qemu_driver->configDir,
1718
                            qemu_driver->autostartDir,
1719
                            0, qemudNotifyLoadDomain, qemu_driver);
1720
    qemuDriverUnlock(qemu_driver);
1721

1722
    qemudAutostartConfigs(qemu_driver);
1723 1724

    return 0;
1725 1726
}

1727 1728 1729 1730 1731 1732 1733 1734 1735 1736
/**
 * 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) {
1737
    int active = 0;
1738

1739 1740 1741
    if (!qemu_driver)
        return 0;

1742
    /* XXX having to iterate here is not great because it requires many locks */
1743
    qemuDriverLock(qemu_driver);
1744
    active = virDomainObjListNumOfDomains(&qemu_driver->domains, 1);
1745 1746
    qemuDriverUnlock(qemu_driver);
    return active;
1747 1748
}

1749 1750 1751 1752 1753 1754 1755
/**
 * qemudShutdown:
 *
 * Shutdown the QEmu daemon, it will stop all active domains and networks
 */
static int
qemudShutdown(void) {
1756
    int i;
1757

1758
    if (!qemu_driver)
1759
        return -1;
1760

1761
    qemuDriverLock(qemu_driver);
1762
    pciDeviceListFree(qemu_driver->activePciHostdevs);
1763 1764
    virCapabilitiesFree(qemu_driver->caps);

1765
    virDomainObjListDeinit(&qemu_driver->domains);
1766

1767
    VIR_FREE(qemu_driver->securityDriverName);
1768
    VIR_FREE(qemu_driver->logDir);
1769 1770
    VIR_FREE(qemu_driver->configDir);
    VIR_FREE(qemu_driver->autostartDir);
1771
    VIR_FREE(qemu_driver->stateDir);
1772 1773
    VIR_FREE(qemu_driver->libDir);
    VIR_FREE(qemu_driver->cacheDir);
1774
    VIR_FREE(qemu_driver->saveDir);
C
Chris Lalancette 已提交
1775
    VIR_FREE(qemu_driver->snapshotDir);
1776
    VIR_FREE(qemu_driver->vncTLSx509certdir);
J
Jim Meyering 已提交
1777
    VIR_FREE(qemu_driver->vncListen);
1778
    VIR_FREE(qemu_driver->vncPassword);
1779
    VIR_FREE(qemu_driver->vncSASLdir);
1780
    VIR_FREE(qemu_driver->saveImageFormat);
1781 1782
    VIR_FREE(qemu_driver->hugetlbfs_mount);
    VIR_FREE(qemu_driver->hugepage_path);
D
Daniel P. Berrange 已提交
1783

1784 1785 1786 1787 1788 1789
    if (qemu_driver->cgroupDeviceACL) {
        for (i = 0 ; qemu_driver->cgroupDeviceACL[i] != NULL ; i++)
            VIR_FREE(qemu_driver->cgroupDeviceACL[i]);
        VIR_FREE(qemu_driver->cgroupDeviceACL);
    }

1790 1791
    /* Free domain callback list */
    virDomainEventCallbackListFree(qemu_driver->domainEventCallbacks);
1792 1793 1794 1795
    virDomainEventQueueFree(qemu_driver->domainEventQueue);

    if (qemu_driver->domainEventTimer != -1)
        virEventRemoveTimeout(qemu_driver->domainEventTimer);
1796

1797 1798 1799
    if (qemu_driver->brctl)
        brShutdown(qemu_driver->brctl);

1800 1801
    virCgroupFree(&qemu_driver->cgroup);

1802
    qemuDriverUnlock(qemu_driver);
1803
    virMutexDestroy(&qemu_driver->lock);
1804
    VIR_FREE(qemu_driver);
1805 1806

    return 0;
1807 1808
}

1809
typedef int qemuLogHandleOutput(virDomainObjPtr vm,
1810 1811
                                const char *output,
                                int fd);
1812 1813 1814 1815 1816

/*
 * Returns -1 for error, 0 on success
 */
static int
1817
qemudReadLogOutput(virDomainObjPtr vm,
1818 1819
                   int fd,
                   char *buf,
G
Guido Günther 已提交
1820
                   size_t buflen,
1821
                   qemuLogHandleOutput func,
1822 1823 1824
                   const char *what,
                   int timeout)
{
1825
    int retries = (timeout*10);
1826
    int got = 0;
1827 1828 1829
    buf[0] = '\0';

    while (retries) {
1830
        ssize_t func_ret, ret;
1831
        int isdead = 0;
G
Guido Günther 已提交
1832

1833
        func_ret = func(vm, buf, fd);
1834

1835 1836
        if (kill(vm->pid, 0) == -1 && errno == ESRCH)
            isdead = 1;
1837

1838 1839
        /* Any failures should be detected before we read the log, so we
         * always have something useful to report on failure. */
1840 1841
        ret = saferead(fd, buf+got, buflen-got-1);
        if (ret < 0) {
1842
            virReportSystemError(errno,
1843 1844 1845 1846 1847
                                 _("Failure while reading %s log output"),
                                 what);
            return -1;
        }

1848 1849 1850
        got += ret;
        buf[got] = '\0';
        if (got == buflen-1) {
1851
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
1852 1853
                            _("Out of space while reading %s log output: %s"),
                            what, buf);
1854 1855 1856 1857
            return -1;
        }

        if (isdead) {
1858
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
1859 1860
                            _("Process exited while reading %s log output: %s"),
                            what, buf);
1861 1862 1863
            return -1;
        }

1864 1865
        if (func_ret <= 0)
            return func_ret;
1866 1867 1868 1869

        usleep(100*1000);
        retries--;
    }
1870

1871
    qemuReportError(VIR_ERR_INTERNAL_ERROR,
1872 1873
                    _("Timed out while reading %s log output: %s"),
                    what, buf);
1874 1875 1876
    return -1;
}

1877

1878 1879 1880 1881 1882 1883 1884 1885
/*
 * 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
 */
1886
static int
1887
qemudExtractTTYPath(const char *haystack,
1888 1889
                    size_t *offset,
                    char **path)
1890
{
1891
    static const char needle[] = "char device redirected to";
1892
    char *tmp, *dev;
1893

1894
    VIR_FREE(*path);
1895
    /* First look for our magic string */
1896 1897 1898 1899 1900
    if (!(tmp = strstr(haystack + *offset, needle))) {
        return 1;
    }
    tmp += sizeof(needle);
    dev = tmp;
1901

1902 1903 1904 1905 1906
    /*
     * And look for first whitespace character and nul terminate
     * to mark end of the pty path
     */
    while (*tmp) {
1907
        if (c_isspace(*tmp)) {
1908 1909
            *path = strndup(dev, tmp-dev);
            if (*path == NULL) {
1910
                virReportOOMError();
1911 1912
                return -1;
            }
1913

1914
            /* ... now further update offset till we get EOL */
1915
            *offset = tmp - haystack;
1916 1917
            return 0;
        }
1918
        tmp++;
1919 1920 1921 1922 1923
    }

    /*
     * We found a path, but didn't find any whitespace,
     * so it must be still incomplete - we should at
1924 1925
     * least see a \n - indicate that we want to carry
     * on trying again
1926
     */
1927
    return 1;
1928 1929
}

1930
static int
1931
qemudFindCharDevicePTYsMonitor(virDomainObjPtr vm,
1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946
                               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) {                                           \
1947 1948 1949
                if (chr->data.file.path == NULL) {                        \
                    /* neither the log output nor 'info chardev' had a */ \
                    /* pty path for this chardev, report an error */      \
1950 1951 1952
                    qemuReportError(VIR_ERR_INTERNAL_ERROR,               \
                                    _("no assigned pty for device %s"), id); \
                    return -1;                                            \
1953 1954 1955 1956 1957
                } else {                                                  \
                    /* 'info chardev' had no pty path for this chardev, */\
                    /* but the log output had, so we're fine */           \
                    continue;                                             \
                }                                                         \
1958 1959
            }                                                             \
                                                                          \
1960
            VIR_FREE(chr->data.file.path);                                \
1961
            chr->data.file.path = strdup(path);                           \
1962 1963
                                                                          \
            if (chr->data.file.path == NULL) {                            \
1964
                virReportOOMError();                                      \
1965 1966
                return -1;                                                \
            }                                                             \
1967 1968 1969 1970 1971 1972
        }                                                                 \
    }

    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");
1973
#undef LOOKUP_PTYS
1974 1975 1976 1977

    return 0;
}

1978
static int
1979
qemudFindCharDevicePTYs(virDomainObjPtr vm,
1980 1981
                        const char *output,
                        int fd ATTRIBUTE_UNUSED)
1982
{
1983
    size_t offset = 0;
1984
    int ret, i;
1985 1986

    /* The order in which QEMU prints out the PTY paths is
1987 1988
       the order in which it procsses its serial and parallel
       device args. This code must match that ordering.... */
1989

1990
    /* first comes the serial devices */
1991 1992
    for (i = 0 ; i < vm->def->nserials ; i++) {
        virDomainChrDefPtr chr = vm->def->serials[i];
1993
        if (chr->type == VIR_DOMAIN_CHR_TYPE_PTY) {
1994
            if ((ret = qemudExtractTTYPath(output, &offset,
1995
                                           &chr->data.file.path)) != 0)
1996
                return ret;
1997 1998 1999
        }
    }

2000
    /* then the parallel devices */
2001 2002
    for (i = 0 ; i < vm->def->nparallels ; i++) {
        virDomainChrDefPtr chr = vm->def->parallels[i];
2003
        if (chr->type == VIR_DOMAIN_CHR_TYPE_PTY) {
2004
            if ((ret = qemudExtractTTYPath(output, &offset,
2005
                                           &chr->data.file.path)) != 0)
2006
                return ret;
2007 2008 2009
        }
    }

2010 2011 2012 2013
    /* 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) {
2014
            if ((ret = qemudExtractTTYPath(output, &offset,
2015 2016 2017 2018 2019
                                           &chr->data.file.path)) != 0)
                return ret;
        }
    }

2020
    return 0;
2021 2022
}

2023 2024 2025 2026 2027
static void qemudFreePtyPath(void *payload, const char *name ATTRIBUTE_UNUSED)
{
    VIR_FREE(payload);
}

2028
static int
2029
qemudWaitForMonitor(struct qemud_driver* driver,
2030
                    virDomainObjPtr vm, off_t pos)
2031
{
2032
    char buf[4096]; /* Plenty of space to get startup greeting */
2033
    int logfd;
2034
    int ret = -1;
2035

2036
    if ((logfd = qemudLogReadFD(driver->logDir, vm->def->name, pos))
2037
        < 0)
2038
        return -1;
2039

2040
    ret = qemudReadLogOutput(vm, logfd, buf, sizeof(buf),
2041
                             qemudFindCharDevicePTYs,
2042
                             "console", 30);
2043
    if (close(logfd) < 0) {
2044
        char ebuf[4096];
2045
        VIR_WARN(_("Unable to close logfile: %s"),
2046 2047
                 virStrerror(errno, ebuf, sizeof ebuf));
    }
2048

2049
    if (ret < 0)
2050
        return -1;
2051

2052
    VIR_DEBUG("Connect monitor to %p '%s'", vm, vm->def->name);
2053
    if (qemuConnectMonitor(driver, vm) < 0)
2054 2055
        return -1;

2056 2057 2058 2059 2060 2061
    /* 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) {
2062
        virReportOOMError();
2063 2064 2065
        goto cleanup;
    }

2066
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
2067 2068
    qemuDomainObjPrivatePtr priv = vm->privateData;
    ret = qemuMonitorGetPtyPaths(priv->mon, paths);
2069
    qemuDomainObjExitMonitorWithDriver(driver, vm);
2070 2071 2072

    VIR_DEBUG("qemuMonitorGetPtyPaths returned %i", ret);
    if (ret == 0) {
2073
        ret = qemudFindCharDevicePTYsMonitor(vm, paths);
2074 2075 2076 2077 2078 2079 2080 2081
    }

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

    return ret;
2082 2083
}

2084
static int
2085
qemuDetectVcpuPIDs(struct qemud_driver *driver,
2086 2087 2088
                   virDomainObjPtr vm) {
    pid_t *cpupids = NULL;
    int ncpupids;
2089
    qemuDomainObjPrivatePtr priv = vm->privateData;
2090

2091
    if (vm->def->virtType != VIR_DOMAIN_VIRT_KVM) {
2092 2093
        priv->nvcpupids = 1;
        if (VIR_ALLOC_N(priv->vcpupids, priv->nvcpupids) < 0) {
2094
            virReportOOMError();
2095 2096
            return -1;
        }
2097
        priv->vcpupids[0] = vm->pid;
2098 2099 2100
        return 0;
    }

2101
    /* What follows is now all KVM specific */
2102

2103
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
2104
    if ((ncpupids = qemuMonitorGetCPUInfo(priv->mon, &cpupids)) < 0) {
2105
        qemuDomainObjExitMonitorWithDriver(driver, vm);
2106
        return -1;
2107
    }
2108
    qemuDomainObjExitMonitorWithDriver(driver, vm);
2109

2110 2111 2112
    /* Treat failure to get VCPU<->PID mapping as non-fatal */
    if (ncpupids == 0)
        return 0;
2113

2114
    if (ncpupids != vm->def->vcpus) {
2115 2116 2117
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("got wrong number of vCPU pids from QEMU monitor. got %d, wanted %d"),
                        ncpupids, (int)vm->def->vcpus);
2118 2119 2120
        VIR_FREE(cpupids);
        return -1;
    }
2121

2122 2123
    priv->nvcpupids = ncpupids;
    priv->vcpupids = cpupids;
2124 2125 2126
    return 0;
}

2127 2128 2129
/*
 * To be run between fork/exec of QEMU only
 */
2130
static int
2131 2132
qemudInitCpuAffinity(virDomainObjPtr vm)
{
2133
    int i, hostcpus, maxcpu = QEMUD_CPUMASK_LEN;
2134
    virNodeInfo nodeinfo;
2135 2136
    unsigned char *cpumap;
    int cpumaplen;
2137 2138

    DEBUG0("Setting CPU affinity");
2139

2140
    if (nodeGetInfo(NULL, &nodeinfo) < 0)
2141 2142 2143 2144
        return -1;

    /* setaffinity fails if you set bits for CPUs which
     * aren't present, so we have to limit ourselves */
2145 2146 2147
    hostcpus = VIR_NODEINFO_MAXCPUS(nodeinfo);
    if (maxcpu > hostcpus)
        maxcpu = hostcpus;
2148

2149 2150
    cpumaplen = VIR_CPU_MAPLEN(maxcpu);
    if (VIR_ALLOC_N(cpumap, cpumaplen) < 0) {
2151
        virReportOOMError();
2152 2153 2154
        return -1;
    }

D
Daniel P. Berrange 已提交
2155
    if (vm->def->cpumask) {
2156 2157 2158
        /* 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
2159
            if (vm->def->cpumask[i])
2160
                VIR_USE_CPU(cpumap, i);
D
Daniel P. Berrange 已提交
2161
    } else {
2162 2163 2164 2165
        /* 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
2166
        for (i = 0 ; i < maxcpu ; i++)
2167
            VIR_USE_CPU(cpumap, i);
D
Daniel P. Berrange 已提交
2168
    }
2169

2170 2171 2172 2173 2174 2175 2176 2177
    /* 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;
2178
    }
2179
    VIR_FREE(cpumap);
2180 2181 2182 2183 2184

    return 0;
}


2185
static int
2186 2187 2188 2189
qemuInitPasswords(virConnectPtr conn,
                  struct qemud_driver *driver,
                  virDomainObjPtr vm,
                  unsigned long long qemuCmdFlags) {
2190
    int ret = 0;
2191
    qemuDomainObjPrivatePtr priv = vm->privateData;
2192

2193 2194 2195
    if ((vm->def->ngraphics == 1) &&
        vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
        (vm->def->graphics[0]->data.vnc.passwd || driver->vncPassword)) {
2196

2197
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
2198
        ret = qemuMonitorSetVNCPassword(priv->mon,
2199 2200 2201
                                        vm->def->graphics[0]->data.vnc.passwd ?
                                        vm->def->graphics[0]->data.vnc.passwd :
                                        driver->vncPassword);
2202
        qemuDomainObjExitMonitorWithDriver(driver, vm);
2203 2204
    }

2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227
    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);
2228
            VIR_FREE(secret);
2229 2230 2231 2232 2233 2234 2235
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            if (ret < 0)
                goto cleanup;
        }
    }

cleanup:
2236
    return ret;
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 2467 2468
#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
2469
qemuDetectPCIAddresses(virDomainObjPtr vm,
2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486
                       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) {
2487 2488 2489
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for VirtIO disk %s"),
                            vm->def->disks[i]->dst);
2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500
            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) {
2501 2502 2503
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for %s NIC"),
                            vm->def->nets[i]->model);
2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514
            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) {
2515 2516 2517
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for controller %s"),
                            virDomainControllerTypeToString(vm->def->controllers[i]->type));
2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528
            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) {
2529 2530 2531
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for video adapter %s"),
                            virDomainVideoTypeToString(vm->def->videos[i]->type));
2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542
            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) {
2543 2544 2545
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for sound adapter %s"),
                            virDomainSoundModelTypeToString(vm->def->sounds[i]->model));
2546 2547 2548 2549 2550 2551 2552 2553 2554 2555
            return -1;
        }
    }


    if (vm->def->watchdog &&
        qemuGetPCIWatchdogVendorProduct(vm->def->watchdog, &vendor, &product) == 0) {
        if (qemuAssignNextPCIAddress(&(vm->def->watchdog->info),
                                     vendor, product,
                                     addrs,  naddrs) < 0) {
2556 2557 2558
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("cannot find PCI address for watchdog %s"),
                            virDomainWatchdogModelTypeToString(vm->def->watchdog->model));
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 2589 2590
            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);

2591
    ret = qemuDetectPCIAddresses(vm, addrs, naddrs);
2592 2593 2594 2595 2596 2597

    VIR_FREE(addrs);

    return ret;
}

2598
static int qemudNextFreeVNCPort(struct qemud_driver *driver ATTRIBUTE_UNUSED) {
2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633
    int i;

    for (i = 5900 ; i < 6000 ; i++) {
        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;
}

2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671

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;
}


2672
static pciDeviceList *
2673
qemuGetPciHostDeviceList(virDomainDefPtr def)
2674 2675
{
    pciDeviceList *list;
2676 2677
    int i;

2678
    if (!(list = pciDeviceListNew()))
2679
        return NULL;
2680 2681 2682 2683 2684 2685 2686 2687 2688 2689

    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;

2690
        dev = pciGetDevice(hostdev->source.subsys.u.pci.domain,
2691 2692 2693
                           hostdev->source.subsys.u.pci.bus,
                           hostdev->source.subsys.u.pci.slot,
                           hostdev->source.subsys.u.pci.function);
2694
        if (!dev) {
2695
            pciDeviceListFree(list);
2696 2697
            return NULL;
        }
2698

2699 2700 2701
        if (pciDeviceListAdd(list, dev) < 0) {
            pciFreeDevice(dev);
            pciDeviceListFree(list);
2702
            return NULL;
2703 2704
        }

2705
        pciDeviceSetManaged(dev, hostdev->managed);
2706 2707
    }

2708 2709 2710 2711
    return list;
}

static int
2712 2713 2714 2715
qemuUpdateActivePciHostdevs(struct qemud_driver *driver,
                            virDomainDefPtr def)
{
    pciDeviceList *pcidevs;
2716
    int i;
2717
    int ret = -1;
2718 2719 2720 2721

    if (!def->nhostdevs)
        return 0;

2722
    if (!(pcidevs = qemuGetPciHostDeviceList(def)))
2723 2724
        return -1;

2725 2726
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
2727 2728 2729
        pciDeviceListSteal(pcidevs, dev);
        if (pciDeviceListAdd(driver->activePciHostdevs, dev) < 0) {
            pciFreeDevice(dev);
2730
            goto cleanup;
2731 2732 2733
        }
    }

2734 2735 2736
    ret = 0;

cleanup:
2737
    pciDeviceListFree(pcidevs);
2738 2739 2740
    return ret;
}

2741

2742
static int
2743 2744
qemuPrepareHostPCIDevices(struct qemud_driver *driver,
                          virDomainDefPtr def)
2745 2746 2747
{
    pciDeviceList *pcidevs;
    int i;
2748
    int ret = -1;
2749

2750
    if (!(pcidevs = qemuGetPciHostDeviceList(def)))
2751 2752
        return -1;

2753
    /* We have to use 3 loops here. *All* devices must
2754 2755
     * be detached before we reset any of them, because
     * in some cases you have to reset the whole PCI,
2756 2757
     * which impacts all devices on it. Also, all devices
     * must be reset before being marked as active.
2758 2759 2760 2761 2762 2763 2764
     */

    /* 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
     */

2765 2766
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
2767
        if (!pciDeviceIsAssignable(dev, !driver->relaxedACS))
2768 2769
            goto cleanup;

2770
        if (pciDeviceGetManaged(dev) &&
2771
            pciDettachDevice(dev) < 0)
2772 2773
            goto cleanup;
    }
2774 2775 2776

    /* Now that all the PCI hostdevs have be dettached, we can safely
     * reset them */
2777 2778
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
2779
        if (pciResetDevice(dev, driver->activePciHostdevs) < 0)
2780 2781
            goto cleanup;
    }
2782

2783
    /* Now mark all the devices as active */
2784 2785
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
2786 2787 2788
        pciDeviceListSteal(pcidevs, dev);
        if (pciDeviceListAdd(driver->activePciHostdevs, dev) < 0) {
            pciFreeDevice(dev);
2789 2790
            goto cleanup;
        }
2791 2792
    }

2793
    ret = 0;
2794

2795
cleanup:
2796
    pciDeviceListFree(pcidevs);
2797
    return ret;
2798 2799
}

2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810

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;
2811
        if (hostdev->source.subsys.type != VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB)
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 2848 2849
            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;
}


2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860
static void
qemudReattachManagedDevice(pciDevice *dev)
{
    int retries = 100;

    if (pciDeviceGetManaged(dev)) {
        while (pciWaitForDeviceCleanup(dev, "kvm_assigned_device")
               && retries) {
            usleep(100*1000);
            retries--;
        }
2861
        if (pciReAttachDevice(dev) < 0) {
2862 2863 2864 2865 2866 2867 2868 2869
            virErrorPtr err = virGetLastError();
            VIR_ERROR(_("Failed to re-attach PCI device: %s"),
                      err ? err->message : "");
            virResetError(err);
        }
    }
}

2870
static void
2871
qemuDomainReAttachHostDevices(struct qemud_driver *driver,
2872
                              virDomainDefPtr def)
2873
{
2874
    pciDeviceList *pcidevs;
2875 2876
    int i;

2877 2878
    if (!def->nhostdevs)
        return;
2879

2880
    if (!(pcidevs = qemuGetPciHostDeviceList(def))) {
2881
        virErrorPtr err = virGetLastError();
2882
        VIR_ERROR(_("Failed to allocate pciDeviceList: %s"),
2883 2884 2885
                  err ? err->message : "");
        virResetError(err);
        return;
2886 2887
    }

2888 2889
    /* Again 3 loops; mark all devices as inactive before reset
     * them and reset all the devices before re-attach */
2890

2891 2892
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
2893
        pciDeviceListDel(driver->activePciHostdevs, dev);
2894
    }
2895

2896 2897
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
2898
        if (pciResetDevice(dev, driver->activePciHostdevs) < 0) {
2899
            virErrorPtr err = virGetLastError();
2900
            VIR_ERROR(_("Failed to reset PCI device: %s"),
2901 2902 2903
                      err ? err->message : "");
            virResetError(err);
        }
2904
    }
2905

2906 2907
    for (i = 0; i < pciDeviceListCount(pcidevs); i++) {
        pciDevice *dev = pciDeviceListGet(pcidevs, i);
2908
        qemudReattachManagedDevice(dev);
2909
    }
2910

2911
    pciDeviceListFree(pcidevs);
2912 2913
}

2914 2915 2916 2917 2918 2919 2920 2921 2922 2923
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

2924
static int qemuSetupCgroup(struct qemud_driver *driver,
2925 2926 2927 2928
                           virDomainObjPtr vm)
{
    virCgroupPtr cgroup = NULL;
    int rc;
2929
    unsigned int i;
2930 2931 2932 2933
    const char *const *deviceACL =
        driver->cgroupDeviceACL ?
        (const char *const *)driver->cgroupDeviceACL :
        defaultDeviceACL;
2934 2935 2936 2937 2938 2939

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

    rc = virCgroupForDomain(driver->cgroup, vm->def->name, &cgroup, 1);
    if (rc != 0) {
2940
        virReportSystemError(-rc,
2941 2942 2943 2944 2945
                             _("Unable to create cgroup for %s"),
                             vm->def->name);
        goto cleanup;
    }

2946 2947
    if (qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_DEVICES)) {
        rc = virCgroupDenyAllDevices(cgroup);
2948
        if (rc != 0) {
2949 2950 2951 2952 2953
            if (rc == -EPERM) {
                VIR_WARN0("Group devices ACL is not accessible, disabling whitelisting");
                goto done;
            }

2954
            virReportSystemError(-rc,
2955
                                 _("Unable to deny all devices for %s"), vm->def->name);
2956 2957 2958
            goto cleanup;
        }

2959 2960 2961 2962 2963 2964 2965 2966
        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) {
2967
                virReportSystemError(-rc,
2968 2969 2970 2971 2972
                                     _("Unable to allow device %s for %s"),
                                     vm->def->disks[i]->src, vm->def->name);
                goto cleanup;
            }
        }
2973

2974
        rc = virCgroupAllowDeviceMajor(cgroup, 'c', DEVICE_PTY_MAJOR);
2975
        if (rc != 0) {
2976
            virReportSystemError(-rc, "%s",
2977
                                 _("unable to allow /dev/pts/ devices"));
2978 2979 2980
            goto cleanup;
        }

2981 2982 2983
        if (vm->def->nsounds) {
            rc = virCgroupAllowDeviceMajor(cgroup, 'c', DEVICE_SND_MAJOR);
            if (rc != 0) {
2984
                virReportSystemError(-rc, "%s",
2985 2986 2987 2988 2989 2990 2991 2992 2993 2994
                                     _("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) {
2995
                virReportSystemError(-rc,
2996 2997 2998 2999
                                     _("unable to allow device %s"),
                                     deviceACL[i]);
                goto cleanup;
            }
3000 3001 3002 3003
        }
    }

done:
3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015
    virCgroupFree(&cgroup);
    return 0;

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


3016
static int qemuRemoveCgroup(struct qemud_driver *driver,
3017 3018
                            virDomainObjPtr vm,
                            int quiet)
3019 3020 3021 3022 3023 3024 3025 3026 3027
{
    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) {
3028
        if (!quiet)
3029 3030 3031
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("Unable to find cgroup for %s\n"),
                            vm->def->name);
3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051
        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) {
3052
        virReportSystemError(-rc,
3053 3054 3055 3056 3057 3058 3059
                             _("unable to find cgroup for domain %s"),
                             def->name);
        goto cleanup;
    }

    rc = virCgroupAddTask(cgroup, getpid());
    if (rc != 0) {
3060
        virReportSystemError(-rc,
3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073
                             _("unable to add domain %s task %d to cgroup"),
                             def->name, getpid());
        goto cleanup;
    }

    ret = 0;

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


S
Stefan Berger 已提交
3074 3075 3076 3077 3078 3079 3080 3081 3082 3083 3084 3085 3086 3087 3088
static void
qemuTearNWFilter(virDomainNetDefPtr net) {
    if ((net->filter) && (net->ifname))
        virNWFilterTeardownFilter(net);
}


static void
qemuTearVMNWFilters(virDomainObjPtr vm) {
    int i;
    for (i = 0; i < vm->def->nnets; i++)
        qemuTearNWFilter(vm->def->nets[i]);
}


3089 3090 3091 3092
struct qemudHookData {
    virConnectPtr conn;
    virDomainObjPtr vm;
    struct qemud_driver *driver;
3093 3094 3095
};

static int qemudSecurityHook(void *data) {
3096 3097
    struct qemudHookData *h = data;

3098 3099 3100 3101 3102 3103
    /* This must take place before exec(), so that all QEMU
     * memory allocation is on the correct NUMA node
     */
    if (qemudInitCpuAffinity(h->vm) < 0)
        return -1;

3104 3105
    if (qemuAddToCgroup(h->driver, h->vm->def) < 0)
        return -1;
3106

3107 3108
    if (h->driver->securityDriver &&
        h->driver->securityDriver->domainSetSecurityProcessLabel &&
3109
        h->driver->securityDriver->domainSetSecurityProcessLabel(h->driver->securityDriver, h->vm) < 0)
3110 3111 3112
        return -1;

    return 0;
3113 3114
}

3115
static int
3116
qemuPrepareMonitorChr(struct qemud_driver *driver,
3117
                      virDomainChrDefPtr monConfig,
3118 3119
                      const char *vm)
{
3120
    monConfig->targetType = VIR_DOMAIN_CHR_TARGET_TYPE_MONITOR;
3121

3122 3123
    monConfig->type = VIR_DOMAIN_CHR_TYPE_UNIX;
    monConfig->data.nix.listen = 1;
3124

D
Daniel P. Berrange 已提交
3125
    if (!(monConfig->info.alias = strdup("monitor"))) {
3126
        virReportOOMError();
D
Daniel P. Berrange 已提交
3127 3128 3129
        return -1;
    }

3130
    if (virAsprintf(&monConfig->data.nix.path, "%s/%s.monitor",
3131
                    driver->libDir, vm) < 0) {
3132
        virReportOOMError();
3133 3134 3135 3136 3137 3138
        return -1;
    }

    return 0;
}

C
Chris Lalancette 已提交
3139 3140 3141 3142 3143
static int qemuDomainSnapshotSetActive(virDomainObjPtr vm,
                                       char *snapshotDir);
static int qemuDomainSnapshotSetInactive(virDomainObjPtr vm,
                                         char *snapshotDir);

3144 3145
static int qemudStartVMDaemon(virConnectPtr conn,
                              struct qemud_driver *driver,
3146
                              virDomainObjPtr vm,
3147 3148
                              const char *migrateFrom,
                              int stdin_fd) {
3149
    const char **argv = NULL, **tmp;
3150
    const char **progenv = NULL;
3151
    int i, ret;
3152
    struct stat sb;
3153 3154
    int *tapfds = NULL;
    int ntapfds = 0;
3155
    unsigned long long qemuCmdFlags;
3156
    fd_set keepfd;
3157
    const char *emulator;
G
Guido Günther 已提交
3158
    pid_t child;
3159
    int pos = -1;
3160
    char ebuf[1024];
3161
    char *pidfile = NULL;
3162
    int logfile;
3163
    qemuDomainObjPrivatePtr priv = vm->privateData;
3164

3165
    struct qemudHookData hookData;
3166 3167 3168 3169
    hookData.conn = conn;
    hookData.vm = vm;
    hookData.driver = driver;

3170
    FD_ZERO(&keepfd);
3171

3172 3173
    DEBUG0("Beginning VM startup process");

D
Daniel P. Berrange 已提交
3174
    if (virDomainObjIsActive(vm)) {
3175 3176
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("VM is already active"));
3177 3178 3179
        return -1;
    }

3180 3181 3182 3183 3184
    /* Must be run before security labelling */
    DEBUG0("Preparing host devices");
    if (qemuPrepareHostDevices(driver, vm->def) < 0)
        goto cleanup;

3185 3186
    /* If you are using a SecurityDriver with dynamic labelling,
       then generate a security label for isolation */
3187
    DEBUG0("Generating domain security label (if required)");
3188
    if (driver->securityDriver &&
3189
        driver->securityDriver->domainGenSecurityLabel &&
3190
        driver->securityDriver->domainGenSecurityLabel(vm) < 0)
3191 3192
        return -1;

3193
    DEBUG0("Generating setting domain security labels (if required)");
3194 3195
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityAllLabel &&
3196
        driver->securityDriver->domainSetSecurityAllLabel(vm) < 0)
3197 3198
        goto cleanup;

3199 3200 3201
    /* Ensure no historical cgroup for this VM is lying around bogus
     * settings */
    DEBUG0("Ensuring no historical cgroup is lying around");
3202
    qemuRemoveCgroup(driver, vm, 1);
3203

3204 3205 3206
    if ((vm->def->ngraphics == 1) &&
        vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
        vm->def->graphics[0]->data.vnc.autoport) {
3207
        DEBUG0("Determining VNC port");
3208
        int port = qemudNextFreeVNCPort(driver);
3209
        if (port < 0) {
3210 3211
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            "%s", _("Unable to find an unused VNC port"));
3212
            goto cleanup;
3213
        }
3214
        vm->def->graphics[0]->data.vnc.port = port;
3215
    }
3216

L
Laine Stump 已提交
3217
    if (virFileMakePath(driver->logDir) != 0) {
3218
        virReportSystemError(errno,
3219 3220
                             _("cannot create log directory %s"),
                             driver->logDir);
3221
        goto cleanup;
3222 3223
    }

3224
    DEBUG0("Creating domain log file");
3225
    if ((logfile = qemudLogFD(driver, vm->def->name)) < 0)
3226
        goto cleanup;
3227

3228 3229
    emulator = vm->def->emulator;

3230 3231 3232 3233
    /* 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
     */
3234
    if (stat(emulator, &sb) < 0) {
3235
        virReportSystemError(errno,
3236 3237
                             _("Cannot find QEMU binary %s"),
                             emulator);
3238
        goto cleanup;
3239 3240
    }

3241
    DEBUG0("Determing emulator version");
3242
    if (qemudExtractVersionInfo(emulator,
3243
                                NULL,
3244
                                &qemuCmdFlags) < 0)
3245
        goto cleanup;
3246

3247
    DEBUG0("Setting up domain cgroup (if required)");
3248
    if (qemuSetupCgroup(driver, vm) < 0)
3249 3250
        goto cleanup;

3251
    if (VIR_ALLOC(priv->monConfig) < 0) {
3252
        virReportOOMError();
3253 3254 3255
        goto cleanup;
    }

3256
    DEBUG0("Preparing monitor state");
3257
    if (qemuPrepareMonitorChr(driver, priv->monConfig, vm->def->name) < 0)
3258
        goto cleanup;
3259

D
Daniel P. Berrange 已提交
3260 3261 3262
#if HAVE_YAJL
    if (qemuCmdFlags & QEMUD_CMD_FLAG_MONITOR_JSON)
        priv->monJSON = 1;
3263
    else
D
Daniel P. Berrange 已提交
3264
#endif
3265
        priv->monJSON = 0;
D
Daniel P. Berrange 已提交
3266

D
Daniel P. Berrange 已提交
3267
    if ((ret = virFileDeletePid(driver->stateDir, vm->def->name)) != 0) {
3268
        virReportSystemError(ret,
D
Daniel P. Berrange 已提交
3269 3270 3271 3272 3273
                             _("Cannot remove stale PID file for %s"),
                             vm->def->name);
        goto cleanup;
    }

3274
    if (!(pidfile = virFilePid(driver->stateDir, vm->def->name))) {
3275
        virReportSystemError(errno,
3276 3277 3278
                             "%s", _("Failed to build pidfile path."));
        goto cleanup;
    }
D
Daniel P. Berrange 已提交
3279

3280
    /*
M
Matthew Booth 已提交
3281
     * Normally PCI addresses are assigned in the virDomainCreate
3282 3283 3284 3285 3286
     * 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
     */
3287
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
3288
        DEBUG0("Assigning domain PCI addresses");
3289
        /* Populate cache with current addresses */
3290 3291 3292 3293 3294 3295 3296
        if (priv->pciaddrs) {
            qemuDomainPCIAddressSetFree(priv->pciaddrs);
            priv->pciaddrs = NULL;
        }
        if (!(priv->pciaddrs = qemuDomainPCIAddressSetCreate(vm->def)))
            goto cleanup;

3297 3298

        /* Assign any remaining addresses */
3299 3300
        if (qemuAssignDevicePCISlots(vm->def, priv->pciaddrs) < 0)
            goto cleanup;
3301 3302 3303 3304

        priv->persistentAddrs = 1;
    } else {
        priv->persistentAddrs = 0;
3305 3306
    }

3307
    DEBUG0("Building emulator command line");
3308
    vm->def->id = driver->nextvmid++;
3309
    if (qemudBuildCommandLine(conn, driver, vm->def, priv->monConfig,
D
Daniel P. Berrange 已提交
3310
                              priv->monJSON, qemuCmdFlags, &argv, &progenv,
C
Chris Lalancette 已提交
3311 3312 3313 3314 3315
                              &tapfds, &ntapfds, migrateFrom,
                              vm->current_snapshot) < 0)
        goto cleanup;

    if (qemuDomainSnapshotSetInactive(vm, driver->snapshotDir) < 0)
3316
        goto cleanup;
3317

3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333
    /* 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;
    }

3334 3335
    tmp = progenv;
    while (*tmp) {
3336
        if (safewrite(logfile, *tmp, strlen(*tmp)) < 0)
3337
            VIR_WARN(_("Unable to write envv to logfile: %s"),
3338
                     virStrerror(errno, ebuf, sizeof ebuf));
3339
        if (safewrite(logfile, " ", 1) < 0)
3340
            VIR_WARN(_("Unable to write envv to logfile: %s"),
3341
                     virStrerror(errno, ebuf, sizeof ebuf));
3342 3343
        tmp++;
    }
3344 3345
    tmp = argv;
    while (*tmp) {
3346
        if (safewrite(logfile, *tmp, strlen(*tmp)) < 0)
3347
            VIR_WARN(_("Unable to write argv to logfile: %s"),
3348
                     virStrerror(errno, ebuf, sizeof ebuf));
3349
        if (safewrite(logfile, " ", 1) < 0)
3350
            VIR_WARN(_("Unable to write argv to logfile: %s"),
3351
                     virStrerror(errno, ebuf, sizeof ebuf));
3352 3353
        tmp++;
    }
3354
    if (safewrite(logfile, "\n", 1) < 0)
3355
        VIR_WARN(_("Unable to write argv to logfile: %s"),
3356
                 virStrerror(errno, ebuf, sizeof ebuf));
3357

3358
    if ((pos = lseek(logfile, 0, SEEK_END)) < 0)
3359
        VIR_WARN(_("Unable to seek to end of logfile: %s"),
3360
                 virStrerror(errno, ebuf, sizeof ebuf));
3361

3362 3363 3364
    for (i = 0 ; i < ntapfds ; i++)
        FD_SET(tapfds[i], &keepfd);

3365
    ret = virExecDaemonize(argv, progenv, &keepfd, &child,
3366
                           stdin_fd, &logfile, &logfile,
3367
                           VIR_EXEC_NONBLOCK | VIR_EXEC_CLEAR_CAPS,
3368 3369 3370
                           qemudSecurityHook, &hookData,
                           pidfile);
    VIR_FREE(pidfile);
G
Guido Günther 已提交
3371 3372 3373

    /* wait for qemu process to to show up */
    if (ret == 0) {
3374
        if (virFileReadPid(driver->stateDir, vm->def->name, &vm->pid)) {
3375 3376
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("Domain %s didn't show up\n"), vm->def->name);
3377
            ret = -1;
G
Guido Günther 已提交
3378
        }
3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389
    } 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;
    }
3390 3391

    vm->state = migrateFrom ? VIR_DOMAIN_PAUSED : VIR_DOMAIN_RUNNING;
3392

3393
    for (i = 0 ; argv[i] ; i++)
3394 3395
        VIR_FREE(argv[i]);
    VIR_FREE(argv);
3396

3397 3398 3399 3400
    for (i = 0 ; progenv[i] ; i++)
        VIR_FREE(progenv[i]);
    VIR_FREE(progenv);

3401 3402 3403
    if (tapfds) {
        for (i = 0 ; i < ntapfds ; i++) {
            close(tapfds[i]);
3404
        }
3405
        VIR_FREE(tapfds);
3406 3407
    }

3408
    if (ret == -1) /* The VM failed to start */
3409 3410
        goto cleanup;

3411
    DEBUG0("Waiting for monitor to show up");
3412
    if (qemudWaitForMonitor(driver, vm, pos) < 0)
3413 3414
        goto abort;

3415
    DEBUG0("Detecting VCPU PIDs");
3416
    if (qemuDetectVcpuPIDs(driver, vm) < 0)
3417 3418
        goto abort;

3419
    DEBUG0("Setting any required VM passwords");
3420
    if (qemuInitPasswords(conn, driver, vm, qemuCmdFlags) < 0)
3421 3422
        goto abort;

D
Daniel P. Berrange 已提交
3423 3424 3425
    /* 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)) {
3426
        DEBUG0("Determining domain device PCI addresses");
D
Daniel P. Berrange 已提交
3427 3428 3429
        if (qemuInitPCIAddresses(driver, vm) < 0)
            goto abort;
    }
3430

3431
    DEBUG0("Setting initial memory amount");
3432
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
3433
    if (qemuMonitorSetBalloon(priv->mon, vm->def->memory) < 0) {
3434
        qemuDomainObjExitMonitorWithDriver(driver, vm);
3435
        goto abort;
3436
    }
3437 3438

    if (migrateFrom == NULL) {
3439
        DEBUG0("Starting domain CPUs");
3440 3441 3442
        /* Allow the CPUS to start executing */
        if (qemuMonitorStartCPUs(priv->mon, conn) < 0) {
            if (virGetLastError() == NULL)
3443 3444
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                "%s", _("resume operation failed"));
3445 3446 3447 3448
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            goto abort;
        }
    }
3449
    qemuDomainObjExitMonitorWithDriver(driver, vm);
3450

3451

3452
    DEBUG0("Writing domain status to disk");
3453
    if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
3454
        goto abort;
3455

3456 3457 3458
    if (logfile != -1)
        close(logfile);

3459
    return 0;
3460 3461

cleanup:
3462 3463 3464
    /* 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 */
S
Stefan Berger 已提交
3465 3466
    qemuTearVMNWFilters(vm);

3467
    qemuDomainReAttachHostDevices(driver, vm->def);
3468

3469 3470
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityAllLabel)
3471
        driver->securityDriver->domainRestoreSecurityAllLabel(vm);
3472 3473
    if (driver->securityDriver &&
        driver->securityDriver->domainReleaseSecurityLabel)
3474
        driver->securityDriver->domainReleaseSecurityLabel(vm);
3475
    qemuRemoveCgroup(driver, vm, 1);
3476 3477 3478 3479
    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;
3480 3481
    if (logfile != -1)
        close(logfile);
3482 3483
    vm->def->id = -1;
    return -1;
3484 3485 3486 3487

abort:
    /* We jump here if we failed to initialize the now running VM
     * killing it off and pretend we never started it */
3488
    qemudShutdownVMDaemon(driver, vm);
3489 3490 3491 3492 3493

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

    return -1;
3494 3495 3496
}


3497
static void qemudShutdownVMDaemon(struct qemud_driver *driver,
3498
                                  virDomainObjPtr vm) {
D
Daniel P. Berrange 已提交
3499
    int ret;
3500
    int retries = 0;
3501
    qemuDomainObjPrivatePtr priv = vm->privateData;
3502
    virErrorPtr orig_err;
3503 3504
    virDomainDefPtr def;
    int i;
D
Daniel P. Berrange 已提交
3505

D
Daniel P. Berrange 已提交
3506
    if (!virDomainObjIsActive(vm))
3507
        return;
3508

3509
    VIR_DEBUG("Shutting down VM '%s'", vm->def->name);
3510

3511 3512 3513 3514
    /* This method is routinely used in clean up paths. Disable error
     * reporting so we don't squash a legit error. */
    orig_err = virSaveLastError();

S
Stefan Berger 已提交
3515 3516
    qemuTearVMNWFilters(vm);

3517
    if (driver->macFilter) {
3518
        def = vm->def;
3519 3520 3521 3522
        for (i = 0 ; i < def->nnets ; i++) {
            virDomainNetDefPtr net = def->nets[i];
            if (net->ifname == NULL)
                continue;
3523
            if ((errno = networkDisallowMacOnPort(driver, net->ifname,
3524
                                                  net->mac))) {
3525
                virReportSystemError(errno,
3526 3527 3528 3529 3530 3531
             _("failed to remove ebtables rule to allow MAC address on  '%s'"),
                                     net->ifname);
            }
        }
    }

G
Guido Günther 已提交
3532 3533
    if (virKillProcess(vm->pid, 0) == 0 &&
        virKillProcess(vm->pid, SIGTERM) < 0)
3534
        virReportSystemError(errno,
3535 3536
                             _("Failed to send SIGTERM to %s (%d)"),
                             vm->def->name, vm->pid);
3537

3538 3539 3540
    if (priv->mon &&
        qemuMonitorClose(priv->mon) == 0) {
        virDomainObjUnref(vm);
3541
        priv->mon = NULL;
3542
    }
3543

3544 3545 3546 3547 3548
    if (priv->monConfig) {
        if (priv->monConfig->type == VIR_DOMAIN_CHR_TYPE_UNIX)
            unlink(priv->monConfig->data.nix.path);
        virDomainChrDefFree(priv->monConfig);
        priv->monConfig = NULL;
3549 3550
    }

G
Guido Günther 已提交
3551 3552
    /* shut it off for sure */
    virKillProcess(vm->pid, SIGKILL);
3553

3554 3555 3556 3557 3558 3559 3560 3561 3562 3563
    /* 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);
    }

3564
    /* Reset Security Labels */
3565
    if (driver->securityDriver &&
3566
        driver->securityDriver->domainRestoreSecurityAllLabel)
3567
        driver->securityDriver->domainRestoreSecurityAllLabel(vm);
3568 3569
    if (driver->securityDriver &&
        driver->securityDriver->domainReleaseSecurityLabel)
3570
        driver->securityDriver->domainReleaseSecurityLabel(vm);
3571

3572 3573 3574 3575 3576 3577 3578
    /* 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
3579
    virDomainDefClearDeviceAliases(vm->def);
3580 3581 3582 3583 3584
    if (!priv->persistentAddrs) {
        virDomainDefClearPCIAddresses(vm->def);
        qemuDomainPCIAddressSetFree(priv->pciaddrs);
        priv->pciaddrs = NULL;
    }
3585

3586
    qemuDomainReAttachHostDevices(driver, vm->def);
3587

S
Stefan Berger 已提交
3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598
#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

3599
retry:
3600
    if ((ret = qemuRemoveCgroup(driver, vm, 0)) < 0) {
3601 3602 3603 3604 3605 3606 3607 3608
        if (ret == -EBUSY && (retries++ < 5)) {
            usleep(200*1000);
            goto retry;
        }
        VIR_WARN("Failed to remove cgroup for %s",
                 vm->def->name);
    }

3609
    qemudRemoveDomainStatus(driver, vm);
D
Daniel P. Berrange 已提交
3610

3611
    vm->pid = -1;
3612
    vm->def->id = -1;
3613
    vm->state = VIR_DOMAIN_SHUTOFF;
3614 3615
    VIR_FREE(priv->vcpupids);
    priv->nvcpupids = 0;
3616 3617

    if (vm->newDef) {
3618
        virDomainDefFree(vm->def);
3619
        vm->def = vm->newDef;
3620
        vm->def->id = -1;
3621 3622
        vm->newDef = NULL;
    }
3623 3624 3625 3626 3627

    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
3628 3629 3630
}


3631
static virDrvOpenStatus qemudOpen(virConnectPtr conn,
3632
                                  virConnectAuthPtr auth ATTRIBUTE_UNUSED,
3633
                                  int flags ATTRIBUTE_UNUSED) {
3634
    if (conn->uri == NULL) {
3635 3636 3637
        if (qemu_driver == NULL)
            return VIR_DRV_OPEN_DECLINED;

3638
        conn->uri = xmlParseURI(qemu_driver->privileged ?
3639 3640
                                "qemu:///system" :
                                "qemu:///session");
3641
        if (!conn->uri) {
3642
            virReportOOMError();
3643 3644
            return VIR_DRV_OPEN_ERROR;
        }
3645 3646 3647 3648 3649 3650 3651 3652 3653 3654
    } 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;

3655
        if (qemu_driver == NULL) {
3656 3657
            qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                            _("qemu state driver is not active"));
3658 3659 3660
            return VIR_DRV_OPEN_ERROR;
        }

3661
        if (conn->uri->path == NULL) {
3662 3663 3664 3665 3666
            qemuReportError(VIR_ERR_INTERNAL_ERROR,
                            _("no QEMU URI path given, try %s"),
                            qemu_driver->privileged
                            ? "qemu:///system"
                            : "qemu:///session");
3667 3668 3669
                return VIR_DRV_OPEN_ERROR;
        }

3670
        if (qemu_driver->privileged) {
3671 3672
            if (STRNEQ (conn->uri->path, "/system") &&
                STRNEQ (conn->uri->path, "/session")) {
3673 3674 3675
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("unexpected QEMU URI path '%s', try qemu:///system"),
                                conn->uri->path);
3676 3677 3678 3679
                return VIR_DRV_OPEN_ERROR;
            }
        } else {
            if (STRNEQ (conn->uri->path, "/session")) {
3680 3681 3682
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("unexpected QEMU URI path '%s', try qemu:///session"),
                                conn->uri->path);
3683 3684 3685
                return VIR_DRV_OPEN_ERROR;
            }
        }
3686 3687 3688 3689 3690 3691 3692
    }
    conn->privateData = qemu_driver;

    return VIR_DRV_OPEN_SUCCESS;
}

static int qemudClose(virConnectPtr conn) {
3693
    struct qemud_driver *driver = conn->privateData;
3694 3695

    /* Get rid of callbacks registered for this conn */
3696
    qemuDriverLock(driver);
3697
    virDomainEventCallbackListRemoveConn(conn, driver->domainEventCallbacks);
3698
    qemuDriverUnlock(driver);
3699 3700 3701 3702 3703 3704

    conn->privateData = NULL;

    return 0;
}

D
Daniel Veillard 已提交
3705 3706 3707 3708 3709
/* Which features are supported by this driver? */
static int
qemudSupportsFeature (virConnectPtr conn ATTRIBUTE_UNUSED, int feature)
{
    switch (feature) {
3710 3711 3712 3713 3714
    case VIR_DRV_FEATURE_MIGRATION_V2:
    case VIR_DRV_FEATURE_MIGRATION_P2P:
        return 1;
    default:
        return 0;
D
Daniel Veillard 已提交
3715 3716 3717
    }
}

3718
static const char *qemudGetType(virConnectPtr conn ATTRIBUTE_UNUSED) {
3719
    return "QEMU";
3720 3721
}

3722

3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735
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;
}


3736 3737 3738 3739
static int kvmGetMaxVCPUs(void) {
    int maxvcpus = 1;

    int r, fd;
3740

3741 3742
    fd = open(KVM_DEVICE, O_RDONLY);
    if (fd < 0) {
3743
        virReportSystemError(errno, _("Unable to open %s"), KVM_DEVICE);
3744
        return -1;
3745 3746 3747 3748 3749 3750 3751 3752 3753 3754 3755
    }

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

    close(fd);
    return maxvcpus;
}


3756
static int qemudGetMaxVCPUs(virConnectPtr conn ATTRIBUTE_UNUSED, const char *type) {
3757 3758 3759
    if (!type)
        return 16;

3760
    if (STRCASEEQ(type, "qemu"))
3761 3762
        return 16;

3763
    if (STRCASEEQ(type, "kvm"))
3764
        return kvmGetMaxVCPUs();
3765

3766
    if (STRCASEEQ(type, "kqemu"))
3767
        return 1;
3768

3769 3770
    qemuReportError(VIR_ERR_INVALID_ARG,
                    _("unknown type '%s'"), type);
3771 3772 3773
    return -1;
}

3774

3775
static char *qemudGetCapabilities(virConnectPtr conn) {
3776
    struct qemud_driver *driver = conn->privateData;
3777
    virCapsPtr caps = NULL;
3778
    char *xml = NULL;
3779

3780
    qemuDriverLock(driver);
3781

3782
    if ((caps = qemuCreateCapabilities(qemu_driver->caps,
3783
                                       qemu_driver)) == NULL) {
3784 3785 3786
        virCapabilitiesFree(caps);
        goto cleanup;
    }
3787

3788
    virCapabilitiesFree(qemu_driver->caps);
3789 3790 3791
    qemu_driver->caps = caps;

    if ((xml = virCapabilitiesFormatXML(driver->caps)) == NULL)
3792
        virReportOOMError();
3793 3794

cleanup:
3795
    qemuDriverUnlock(driver);
3796

3797
    return xml;
3798 3799 3800
}


3801
static int qemudGetProcessInfo(unsigned long long *cpuTime, int *lastCpu, int pid, int tid) {
D
Daniel P. Berrange 已提交
3802 3803
    char proc[PATH_MAX];
    FILE *pidinfo;
3804
    unsigned long long usertime, systime;
3805 3806
    int cpu;
    int ret;
D
Daniel P. Berrange 已提交
3807

3808 3809 3810 3811 3812 3813
    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
3814 3815 3816 3817
        return -1;
    }

    if (!(pidinfo = fopen(proc, "r"))) {
3818
        /*printf("cannot read pid info");*/
D
Daniel P. Berrange 已提交
3819
        /* VM probably shut down, so fake 0 */
3820 3821 3822 3823
        if (cpuTime)
            *cpuTime = 0;
        if (lastCpu)
            *lastCpu = 0;
D
Daniel P. Berrange 已提交
3824 3825 3826
        return 0;
    }

3827 3828 3829 3830 3831 3832 3833 3834 3835 3836
    /* 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) {
3837
        fclose(pidinfo);
3838 3839
        VIR_WARN0("cannot parse process status data");
        errno = -EINVAL;
D
Daniel P. Berrange 已提交
3840 3841 3842 3843 3844 3845 3846 3847
        return -1;
    }

    /* We got jiffies
     * We want nanoseconds
     * _SC_CLK_TCK is jiffies per second
     * So calulate thus....
     */
3848 3849 3850 3851 3852
    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
3853

3854 3855
    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
3856 3857 3858 3859 3860 3861 3862

    fclose(pidinfo);

    return 0;
}


3863
static virDomainPtr qemudDomainLookupByID(virConnectPtr conn,
3864
                                          int id) {
3865 3866 3867 3868
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;

3869
    qemuDriverLock(driver);
3870
    vm  = virDomainFindByID(&driver->domains, id);
3871
    qemuDriverUnlock(driver);
3872 3873

    if (!vm) {
3874 3875
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching id %d"), id);
3876
        goto cleanup;
3877 3878
    }

3879
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
3880
    if (dom) dom->id = vm->def->id;
3881 3882

cleanup:
3883 3884
    if (vm)
        virDomainObjUnlock(vm);
3885 3886
    return dom;
}
3887

3888
static virDomainPtr qemudDomainLookupByUUID(virConnectPtr conn,
3889
                                            const unsigned char *uuid) {
3890 3891 3892
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
3893

3894
    qemuDriverLock(driver);
3895
    vm = virDomainFindByUUID(&driver->domains, uuid);
3896 3897
    qemuDriverUnlock(driver);

3898
    if (!vm) {
3899 3900
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(uuid, uuidstr);
3901 3902
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
3903
        goto cleanup;
3904 3905
    }

3906
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
3907
    if (dom) dom->id = vm->def->id;
3908 3909

cleanup:
3910 3911
    if (vm)
        virDomainObjUnlock(vm);
3912 3913
    return dom;
}
3914

3915
static virDomainPtr qemudDomainLookupByName(virConnectPtr conn,
3916
                                            const char *name) {
3917 3918 3919
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
3920

3921
    qemuDriverLock(driver);
3922
    vm = virDomainFindByName(&driver->domains, name);
3923 3924
    qemuDriverUnlock(driver);

3925
    if (!vm) {
3926 3927
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching name '%s'"), name);
3928
        goto cleanup;
3929 3930
    }

3931
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
3932
    if (dom) dom->id = vm->def->id;
3933 3934

cleanup:
3935 3936
    if (vm)
        virDomainObjUnlock(vm);
3937 3938 3939
    return dom;
}

3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950

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) {
3951
        qemuReportError(VIR_ERR_NO_DOMAIN, NULL);
3952 3953 3954 3955 3956 3957 3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 3969 3970 3971
        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) {
3972
        qemuReportError(VIR_ERR_NO_DOMAIN, NULL);
3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983
        goto cleanup;
    }
    ret = obj->persistent;

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


3984
static int qemudGetVersion(virConnectPtr conn, unsigned long *version) {
3985 3986 3987
    struct qemud_driver *driver = conn->privateData;
    int ret = -1;

3988
    qemuDriverLock(driver);
3989
    if (qemudExtractVersion(driver) < 0)
3990
        goto cleanup;
3991

3992
    *version = qemu_driver->qemuVersion;
3993 3994 3995
    ret = 0;

cleanup:
3996
    qemuDriverUnlock(driver);
3997
    return ret;
D
Daniel P. Berrange 已提交
3998 3999
}

4000
static int qemudListDomains(virConnectPtr conn, int *ids, int nids) {
4001
    struct qemud_driver *driver = conn->privateData;
4002
    int n;
4003

4004
    qemuDriverLock(driver);
4005
    n = virDomainObjListGetActiveIDs(&driver->domains, ids, nids);
4006
    qemuDriverUnlock(driver);
4007

4008
    return n;
D
Daniel P. Berrange 已提交
4009
}
4010

4011
static int qemudNumDomains(virConnectPtr conn) {
4012
    struct qemud_driver *driver = conn->privateData;
4013
    int n;
4014

4015
    qemuDriverLock(driver);
4016
    n = virDomainObjListNumOfDomains(&driver->domains, 1);
4017
    qemuDriverUnlock(driver);
4018

4019
    return n;
D
Daniel P. Berrange 已提交
4020
}
4021

4022
static virDomainPtr qemudDomainCreate(virConnectPtr conn, const char *xml,
4023
                                      unsigned int flags ATTRIBUTE_UNUSED) {
4024
    struct qemud_driver *driver = conn->privateData;
4025
    virDomainDefPtr def;
4026
    virDomainObjPtr vm = NULL;
4027
    virDomainPtr dom = NULL;
4028
    virDomainEventPtr event = NULL;
D
Daniel P. Berrange 已提交
4029

4030
    qemuDriverLock(driver);
4031
    if (!(def = virDomainDefParseString(driver->caps, xml,
4032
                                        VIR_DOMAIN_XML_INACTIVE)))
4033
        goto cleanup;
4034

4035
    if (virSecurityDriverVerify(def) < 0)
4036 4037
        goto cleanup;

4038 4039
    if (virDomainObjIsDuplicate(&driver->domains, def, 1) < 0)
        goto cleanup;
4040

4041 4042 4043 4044 4045 4046
    if (qemudCanonicalizeMachine(driver, def) < 0)
        goto cleanup;

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

4047
    if (!(vm = virDomainAssignDef(driver->caps,
4048
                                  &driver->domains,
4049
                                  def, false)))
4050 4051 4052
        goto cleanup;

    def = NULL;
D
Daniel P. Berrange 已提交
4053

4054 4055 4056
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup; /* XXXX free the 'vm' we created ? */

4057
    if (qemudStartVMDaemon(conn, driver, vm, NULL, -1) < 0) {
4058 4059 4060
        if (qemuDomainObjEndJob(vm) > 0)
            virDomainRemoveInactive(&driver->domains,
                                    vm);
4061
        vm = NULL;
4062
        goto endjob;
D
Daniel P. Berrange 已提交
4063
    }
4064 4065 4066 4067

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_BOOTED);
D
Daniel P. Berrange 已提交
4068

4069
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
4070
    if (dom) dom->id = vm->def->id;
4071

4072
endjob:
4073 4074 4075
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4076

4077 4078
cleanup:
    virDomainDefFree(def);
4079 4080
    if (vm)
        virDomainObjUnlock(vm);
4081 4082
    if (event)
        qemuDomainEventQueue(driver, event);
4083
    qemuDriverUnlock(driver);
4084
    return dom;
D
Daniel P. Berrange 已提交
4085 4086 4087
}


4088
static int qemudDomainSuspend(virDomainPtr dom) {
4089 4090 4091
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4092
    virDomainEventPtr event = NULL;
4093
    qemuDomainObjPrivatePtr priv;
4094

4095
    qemuDriverLock(driver);
4096
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4097

D
Daniel P. Berrange 已提交
4098
    if (!vm) {
4099 4100
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4101 4102
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4103
        goto cleanup;
D
Daniel P. Berrange 已提交
4104
    }
D
Daniel P. Berrange 已提交
4105
    if (!virDomainObjIsActive(vm)) {
4106 4107
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
4108
        goto cleanup;
D
Daniel P. Berrange 已提交
4109
    }
4110 4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 4127

    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"));
4128
            goto endjob;
4129
        }
4130 4131 4132 4133 4134 4135 4136 4137 4138 4139 4140 4141 4142 4143 4144 4145
        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
4146
    }
4147

4148
endjob:
4149 4150
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4151

4152
cleanup:
4153 4154
    if (vm)
        virDomainObjUnlock(vm);
4155

4156
    if (event)
4157
        qemuDomainEventQueue(driver, event);
4158
    qemuDriverUnlock(driver);
4159
    return ret;
D
Daniel P. Berrange 已提交
4160 4161 4162
}


4163
static int qemudDomainResume(virDomainPtr dom) {
4164 4165 4166
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4167
    virDomainEventPtr event = NULL;
4168

4169
    qemuDriverLock(driver);
4170
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4171

D
Daniel P. Berrange 已提交
4172
    if (!vm) {
4173 4174
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4175 4176
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4177
        goto cleanup;
D
Daniel P. Berrange 已提交
4178
    }
4179 4180 4181 4182

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

D
Daniel P. Berrange 已提交
4183
    if (!virDomainObjIsActive(vm)) {
4184 4185
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
4186
        goto endjob;
D
Daniel P. Berrange 已提交
4187
    }
4188
    if (vm->state == VIR_DOMAIN_PAUSED) {
4189
        qemuDomainObjPrivatePtr priv = vm->privateData;
4190
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
4191
        if (qemuMonitorStartCPUs(priv->mon, dom->conn) < 0) {
4192
            qemuDomainObjExitMonitorWithDriver(driver, vm);
4193
            if (virGetLastError() == NULL)
4194 4195
                qemuReportError(VIR_ERR_OPERATION_FAILED,
                                "%s", _("resume operation failed"));
4196
            goto endjob;
4197
        }
4198
        qemuDomainObjExitMonitorWithDriver(driver, vm);
4199
        vm->state = VIR_DOMAIN_RUNNING;
4200 4201 4202
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_UNPAUSED);
D
Daniel P. Berrange 已提交
4203
    }
4204
    if (virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
4205
        goto endjob;
4206 4207
    ret = 0;

4208
endjob:
4209 4210
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4211

4212
cleanup:
4213 4214
    if (vm)
        virDomainObjUnlock(vm);
4215
    if (event)
4216
        qemuDomainEventQueue(driver, event);
4217
    qemuDriverUnlock(driver);
4218
    return ret;
D
Daniel P. Berrange 已提交
4219 4220 4221
}


4222
static int qemudDomainShutdown(virDomainPtr dom) {
4223 4224 4225
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4226

4227
    qemuDriverLock(driver);
4228
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4229 4230
    qemuDriverUnlock(driver);

4231
    if (!vm) {
4232 4233
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4234 4235
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4236
        goto cleanup;
4237 4238
    }

4239 4240 4241
    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
4242
    if (!virDomainObjIsActive(vm)) {
4243 4244
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
4245
        goto endjob;
4246 4247
    }

4248
    qemuDomainObjPrivatePtr priv = vm->privateData;
4249 4250 4251
    qemuDomainObjEnterMonitor(vm);
    ret = qemuMonitorSystemPowerdown(priv->mon);
    qemuDomainObjExitMonitor(vm);
4252

4253
endjob:
4254 4255
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4256

4257
cleanup:
4258 4259
    if (vm)
        virDomainObjUnlock(vm);
4260
    return ret;
4261 4262 4263
}


4264
static int qemudDomainDestroy(virDomainPtr dom) {
4265 4266 4267
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4268
    virDomainEventPtr event = NULL;
4269

4270
    qemuDriverLock(driver);
4271
    vm  = virDomainFindByUUID(&driver->domains, dom->uuid);
D
Daniel P. Berrange 已提交
4272
    if (!vm) {
4273 4274
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4275 4276
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4277
        goto cleanup;
D
Daniel P. Berrange 已提交
4278
    }
4279 4280 4281 4282

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

D
Daniel P. Berrange 已提交
4283
    if (!virDomainObjIsActive(vm)) {
4284 4285
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
4286
        goto endjob;
4287
    }
4288

4289
    qemudShutdownVMDaemon(driver, vm);
4290 4291 4292
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_DESTROYED);
4293
    if (!vm->persistent) {
4294 4295 4296
        if (qemuDomainObjEndJob(vm) > 0)
            virDomainRemoveInactive(&driver->domains,
                                    vm);
4297 4298
        vm = NULL;
    }
4299 4300
    ret = 0;

4301
endjob:
4302 4303 4304
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4305

4306
cleanup:
4307 4308
    if (vm)
        virDomainObjUnlock(vm);
4309 4310
    if (event)
        qemuDomainEventQueue(driver, event);
4311
    qemuDriverUnlock(driver);
4312
    return ret;
D
Daniel P. Berrange 已提交
4313 4314 4315
}


4316
static char *qemudDomainGetOSType(virDomainPtr dom) {
4317 4318 4319
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    char *type = NULL;
4320

4321
    qemuDriverLock(driver);
4322
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4323
    qemuDriverUnlock(driver);
4324
    if (!vm) {
4325 4326
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4327 4328
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4329
        goto cleanup;
4330 4331
    }

4332
    if (!(type = strdup(vm->def->os.type)))
4333
        virReportOOMError();
4334 4335

cleanup:
4336 4337
    if (vm)
        virDomainObjUnlock(vm);
4338 4339 4340
    return type;
}

4341 4342
/* Returns max memory in kb, 0 if error */
static unsigned long qemudDomainGetMaxMemory(virDomainPtr dom) {
4343 4344 4345
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    unsigned long ret = 0;
4346

4347
    qemuDriverLock(driver);
4348
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4349 4350
    qemuDriverUnlock(driver);

4351
    if (!vm) {
4352 4353
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4354 4355
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4356
        goto cleanup;
4357 4358
    }

4359 4360 4361
    ret = vm->def->maxmem;

cleanup:
4362 4363
    if (vm)
        virDomainObjUnlock(vm);
4364
    return ret;
4365 4366 4367
}

static int qemudDomainSetMemory(virDomainPtr dom, unsigned long newmem) {
4368
    struct qemud_driver *driver = dom->conn->privateData;
4369
    qemuDomainObjPrivatePtr priv;
4370
    virDomainObjPtr vm;
4371
    int ret = -1, r;
4372

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

4384 4385 4386 4387 4388 4389
    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
        goto cleanup;
    }

4390
    if (newmem > vm->def->maxmem) {
4391 4392
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("cannot set memory higher than max memory"));
4393
        goto cleanup;
4394 4395
    }

4396 4397 4398
    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

4399 4400 4401 4402 4403 4404
    priv = vm->privateData;
    qemuDomainObjEnterMonitor(vm);
    r = qemuMonitorSetBalloon(priv->mon, newmem);
    qemuDomainObjExitMonitor(vm);
    if (r < 0)
        goto endjob;
4405

4406 4407 4408 4409 4410
    /* 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;
4411
    }
4412

4413
    ret = 0;
4414
endjob:
4415 4416
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
4417

4418
cleanup:
4419 4420
    if (vm)
        virDomainObjUnlock(vm);
4421
    return ret;
4422 4423
}

4424
static int qemudDomainGetInfo(virDomainPtr dom,
4425
                              virDomainInfoPtr info) {
4426 4427 4428
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4429 4430
    int err;
    unsigned long balloon;
4431

4432
    qemuDriverLock(driver);
4433
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4434
    qemuDriverUnlock(driver);
D
Daniel P. Berrange 已提交
4435
    if (!vm) {
4436 4437
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4438 4439
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4440
        goto cleanup;
D
Daniel P. Berrange 已提交
4441 4442
    }

4443
    info->state = vm->state;
D
Daniel P. Berrange 已提交
4444

D
Daniel P. Berrange 已提交
4445
    if (!virDomainObjIsActive(vm)) {
4446
        info->cpuTime = 0;
D
Daniel P. Berrange 已提交
4447
    } else {
4448
        if (qemudGetProcessInfo(&(info->cpuTime), NULL, vm->pid, 0) < 0) {
4449
            qemuReportError(VIR_ERR_OPERATION_FAILED, ("cannot read cputime for domain"));
4450
            goto cleanup;
D
Daniel P. Berrange 已提交
4451 4452 4453
        }
    }

4454
    info->maxMem = vm->def->maxmem;
4455

D
Daniel P. Berrange 已提交
4456
    if (virDomainObjIsActive(vm)) {
4457
        qemuDomainObjPrivatePtr priv = vm->privateData;
4458 4459 4460
        if (!priv->jobActive) {
            if (qemuDomainObjBeginJob(vm) < 0)
                goto cleanup;
4461

4462 4463 4464 4465
            qemuDomainObjEnterMonitor(vm);
            err = qemuMonitorGetBalloonInfo(priv->mon, &balloon);
            qemuDomainObjExitMonitor(vm);
            if (err < 0) {
4466 4467
                if (qemuDomainObjEndJob(vm) == 0)
                    vm = NULL;
4468 4469 4470 4471 4472 4473 4474 4475 4476
                goto cleanup;
            }

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

4477 4478 4479 4480
            if (qemuDomainObjEndJob(vm) == 0) {
                vm = NULL;
                goto cleanup;
            }
4481 4482 4483
        } else {
            info->memory = vm->def->memory;
        }
4484 4485 4486 4487
    } else {
        info->memory = vm->def->memory;
    }

4488
    info->nrVirtCpu = vm->def->vcpus;
4489 4490 4491
    ret = 0;

cleanup:
4492 4493
    if (vm)
        virDomainObjUnlock(vm);
4494
    return ret;
D
Daniel P. Berrange 已提交
4495 4496 4497
}


4498 4499 4500 4501 4502 4503 4504 4505 4506 4507 4508 4509 4510 4511 4512 4513 4514 4515 4516 4517 4518 4519 4520 4521 4522 4523 4524 4525 4526
/** 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;
}


4527 4528 4529 4530 4531 4532 4533 4534 4535 4536 4537 4538 4539 4540 4541 4542 4543 4544
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 */
        struct timespec ts = { .tv_sec = 0, .tv_nsec = 50 * 1000ull };
        struct timeval now;
        int rc;

4545 4546
        if (priv->jobSignals & QEMU_JOB_SIGNAL_CANCEL) {
            priv->jobSignals ^= QEMU_JOB_SIGNAL_CANCEL;
4547 4548 4549 4550 4551 4552 4553
            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");
            }
4554 4555 4556 4557 4558
        } 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");
4559 4560 4561 4562 4563 4564 4565 4566 4567 4568 4569
        } 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");
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 4630 4631 4632 4633 4634 4635 4636 4637 4638 4639 4640 4641 4642 4643
        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;
}


4644
#define QEMUD_SAVE_MAGIC "LibvirtQemudSave"
4645 4646 4647
#define QEMUD_SAVE_VERSION 2

enum qemud_save_formats {
4648 4649 4650
    QEMUD_SAVE_FORMAT_RAW = 0,
    QEMUD_SAVE_FORMAT_GZIP = 1,
    QEMUD_SAVE_FORMAT_BZIP2 = 2,
4651 4652
    /*
     * Deprecated by xz and never used as part of a release
4653
     * QEMUD_SAVE_FORMAT_LZMA
4654 4655
     */
    QEMUD_SAVE_FORMAT_XZ = 3,
4656
    QEMUD_SAVE_FORMAT_LZOP = 4,
4657 4658 4659
    /* Note: add new members only at the end.
       These values are used in the on-disk format.
       Do not change or re-use numbers. */
4660 4661

    QEMUD_SAVE_FORMAT_LAST
4662
};
4663

4664 4665 4666 4667 4668
VIR_ENUM_DECL(qemudSaveCompression)
VIR_ENUM_IMPL(qemudSaveCompression, QEMUD_SAVE_FORMAT_LAST,
              "raw",
              "gzip",
              "bzip2",
4669 4670
              "xz",
              "lzop")
4671

4672 4673 4674 4675 4676
struct qemud_save_header {
    char magic[sizeof(QEMUD_SAVE_MAGIC)-1];
    int version;
    int xml_len;
    int was_running;
4677 4678
    int compressed;
    int unused[15];
4679 4680
};

4681 4682 4683 4684 4685 4686 4687 4688 4689 4690 4691 4692 4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709
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;
}


4710 4711
static int qemudDomainSaveFlag(virDomainPtr dom, const char *path,
                               int compressed)
4712
{
4713
    struct qemud_driver *driver = dom->conn->privateData;
4714
    virDomainObjPtr vm = NULL;
4715
    char *xml = NULL;
4716
    struct qemud_save_header header;
4717 4718
    struct fileOpHookData hdata;
    int bypassSecurityDriver = 0;
4719
    int ret = -1;
4720
    int rc;
4721
    virDomainEventPtr event = NULL;
4722
    qemuDomainObjPrivatePtr priv;
4723 4724 4725 4726 4727

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

4728
    qemuDriverLock(driver);
4729 4730

    header.compressed = compressed;
4731

4732
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4733

D
Daniel P. Berrange 已提交
4734
    if (!vm) {
4735 4736
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4737 4738
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
4739
        goto cleanup;
D
Daniel P. Berrange 已提交
4740
    }
4741
    priv = vm->privateData;
4742

4743 4744 4745
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
4746
    if (!virDomainObjIsActive(vm)) {
4747 4748
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
4749
        goto endjob;
D
Daniel P. Berrange 已提交
4750
    }
4751

4752 4753 4754
    memset(&priv->jobInfo, 0, sizeof(priv->jobInfo));
    priv->jobInfo.type = VIR_DOMAIN_JOB_UNBOUNDED;

4755 4756 4757
    /* Pause */
    if (vm->state == VIR_DOMAIN_RUNNING) {
        header.was_running = 1;
4758
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
4759
        if (qemuMonitorStopCPUs(priv->mon) < 0) {
4760 4761
            qemuDomainObjExitMonitorWithDriver(driver, vm);
            goto endjob;
4762
        }
4763
        qemuDomainObjExitMonitorWithDriver(driver, vm);
4764
        vm->state = VIR_DOMAIN_PAUSED;
4765 4766 4767
    }

    /* Get XML for the domain */
4768
    xml = virDomainDefFormat(vm->def, VIR_DOMAIN_XML_SECURE);
4769
    if (!xml) {
4770 4771
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to get domain xml"));
4772
        goto endjob;
4773 4774 4775
    }
    header.xml_len = strlen(xml) + 1;

4776 4777 4778 4779 4780 4781
    /* Setup hook data needed by virFileOperation hook function */
    hdata.dom = dom;
    hdata.path = path;
    hdata.xml = xml;
    hdata.header = &header;

4782 4783
    /* Write header to file, followed by XML */

4784 4785 4786 4787 4788 4789 4790 4791 4792 4793 4794 4795 4796 4797 4798 4799 4800 4801 4802 4803
    /* 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;
        }
4804

4805 4806 4807 4808 4809
#ifdef __linux__
        /* On Linux we can also verify the FS-type of the directory. */
        char *dirpath, *p;
        struct statfs st;
        int statfs_ret;
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 4864 4865 4866 4867 4868 4869 4870 4871 4872 4873 4874 4875 4876 4877
        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;
4878
    }
4879

4880 4881 4882

    if ((!bypassSecurityDriver) &&
        driver->securityDriver &&
4883
        driver->securityDriver->domainSetSavedStateLabel &&
4884
        driver->securityDriver->domainSetSavedStateLabel(vm, path) == -1)
4885 4886
        goto endjob;

4887 4888
    if (header.compressed == QEMUD_SAVE_FORMAT_RAW) {
        const char *args[] = { "cat", NULL };
M
Matthias Bolte 已提交
4889
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
4890
        rc = qemuMonitorMigrateToCommand(priv->mon, 1, args, path);
M
Matthias Bolte 已提交
4891
        qemuDomainObjExitMonitorWithDriver(driver, vm);
4892
    } else {
4893
        const char *prog = qemudSaveCompressionTypeToString(header.compressed);
4894 4895 4896 4897 4898
        const char *args[] = {
            prog,
            "-c",
            NULL
        };
M
Matthias Bolte 已提交
4899
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
4900
        rc = qemuMonitorMigrateToCommand(priv->mon, 1, args, path);
M
Matthias Bolte 已提交
4901
        qemuDomainObjExitMonitorWithDriver(driver, vm);
4902 4903
    }

4904 4905 4906 4907 4908
    if (rc < 0)
        goto endjob;

    rc = qemuDomainWaitForMigrationComplete(driver, vm);

4909
    if (rc < 0)
4910
        goto endjob;
4911

4912 4913
    if ((!bypassSecurityDriver) &&
        driver->securityDriver &&
4914
        driver->securityDriver->domainRestoreSavedStateLabel &&
4915
        driver->securityDriver->domainRestoreSavedStateLabel(vm, path) == -1)
4916 4917 4918 4919
        goto endjob;

    ret = 0;

4920
    /* Shut it down */
4921
    qemudShutdownVMDaemon(driver, vm);
4922 4923 4924
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_SAVED);
4925
    if (!vm->persistent) {
4926 4927 4928
        if (qemuDomainObjEndJob(vm) > 0)
            virDomainRemoveInactive(&driver->domains,
                                    vm);
4929 4930
        vm = NULL;
    }
4931

4932
endjob:
4933 4934 4935 4936 4937 4938 4939 4940 4941 4942
    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;
    }

4943 4944
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
4945
            vm = NULL;
4946

4947 4948 4949 4950
cleanup:
    VIR_FREE(xml);
    if (ret != 0)
        unlink(path);
4951 4952
    if (vm)
        virDomainObjUnlock(vm);
4953 4954
    if (event)
        qemuDomainEventQueue(driver, event);
4955
    qemuDriverUnlock(driver);
4956
    return ret;
D
Daniel P. Berrange 已提交
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 4987 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 5048 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 5082 5083 5084 5085 5086 5087 5088 5089 5090 5091 5092 5093 5094 5095 5096 5097 5098 5099 5100 5101 5102 5103 5104 5105 5106 5107 5108 5109 5110 5111 5112 5113 5114 5115 5116 5117 5118 5119 5120 5121 5122 5123 5124 5125 5126 5127 5128 5129 5130 5131 5132 5133 5134 5135
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;

    if (flags != 0) {
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("unsupported flags (0x%x) passed to '%s'"),
                        flags, __FUNCTION__);
        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 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;

    if (flags != 0) {
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("unsupported flags (0x%x) passed to '%s'"),
                        flags, __FUNCTION__);
        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;
    }

    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;

    if (flags != 0) {
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("unsupported flags (0x%x) passed to '%s'"),
                        flags, __FUNCTION__);
        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;
    }

    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
5136

P
Paolo Bonzini 已提交
5137 5138 5139 5140 5141 5142
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;
5143
    int ret = -1, fd = -1;
5144
    virDomainEventPtr event = NULL;
5145 5146 5147 5148
    const char *args[] = {
        "cat",
        NULL,
    };
5149
    qemuDomainObjPrivatePtr priv;
P
Paolo Bonzini 已提交
5150 5151 5152 5153 5154 5155 5156

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

    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
5157 5158
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
P
Paolo Bonzini 已提交
5159 5160
        goto cleanup;
    }
5161
    priv = vm->privateData;
P
Paolo Bonzini 已提交
5162

5163
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
5164 5165
        goto cleanup;

D
Daniel P. Berrange 已提交
5166
    if (!virDomainObjIsActive(vm)) {
5167 5168
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
5169
        goto endjob;
P
Paolo Bonzini 已提交
5170 5171
    }

5172 5173
    /* Create an empty file with appropriate ownership.  */
    if ((fd = open(path, O_CREAT|O_TRUNC|O_WRONLY, S_IRUSR|S_IWUSR)) < 0) {
5174 5175
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("failed to create '%s'"), path);
5176 5177 5178 5179
        goto endjob;
    }

    if (close(fd) < 0) {
5180
        virReportSystemError(errno,
5181 5182 5183 5184 5185 5186 5187
                             _("unable to save file %s"),
                             path);
        goto endjob;
    }

    if (driver->securityDriver &&
        driver->securityDriver->domainSetSavedStateLabel &&
5188
        driver->securityDriver->domainSetSavedStateLabel(vm, path) == -1)
5189 5190
        goto endjob;

P
Paolo Bonzini 已提交
5191 5192
    /* 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
5193 5194 5195
    resume = (vm->state == VIR_DOMAIN_RUNNING);

    /* Pause domain for non-live dump */
P
Paolo Bonzini 已提交
5196
    if (!(flags & VIR_DUMP_LIVE) && vm->state == VIR_DOMAIN_RUNNING) {
5197 5198 5199
        qemuDomainObjEnterMonitor(vm);
        if (qemuMonitorStopCPUs(priv->mon) < 0) {
            qemuDomainObjExitMonitor(vm);
5200
            goto endjob;
5201 5202
        }
        qemuDomainObjExitMonitor(vm);
P
Paolo Bonzini 已提交
5203 5204 5205
        paused = 1;
    }

5206
    qemuDomainObjEnterMonitor(vm);
5207
    ret = qemuMonitorMigrateToCommand(priv->mon, 1, args, path);
5208
    qemuDomainObjExitMonitor(vm);
5209 5210 5211 5212 5213 5214 5215 5216 5217 5218

    if (ret < 0)
        goto endjob;

    ret = qemuDomainWaitForMigrationComplete(driver, vm);

    if (ret < 0)
        goto endjob;

    paused = 1;
5219 5220 5221

    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSavedStateLabel &&
5222
        driver->securityDriver->domainRestoreSavedStateLabel(vm, path) == -1)
5223 5224 5225
        goto endjob;

endjob:
5226
    if ((ret == 0) && (flags & VIR_DUMP_CRASH)) {
5227
        qemudShutdownVMDaemon(driver, vm);
5228 5229 5230 5231 5232
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         VIR_DOMAIN_EVENT_STOPPED_CRASHED);
    }

P
Paolo Bonzini 已提交
5233 5234 5235
    /* 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.  */
5236
    else if (resume && paused) {
5237
        qemuDomainObjEnterMonitor(vm);
5238
        if (qemuMonitorStartCPUs(priv->mon, dom->conn) < 0) {
5239
            if (virGetLastError() == NULL)
5240 5241
                qemuReportError(VIR_ERR_OPERATION_FAILED,
                                "%s", _("resuming after dump failed"));
P
Paolo Bonzini 已提交
5242
        }
5243
        qemuDomainObjExitMonitor(vm);
P
Paolo Bonzini 已提交
5244
    }
5245

5246 5247
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
5248
    else if ((ret == 0) && (flags & VIR_DUMP_CRASH) && !vm->persistent) {
5249 5250 5251 5252
        virDomainRemoveInactive(&driver->domains,
                                vm);
        vm = NULL;
    }
5253 5254

cleanup:
5255 5256
    if (ret != 0)
        unlink(path);
P
Paolo Bonzini 已提交
5257 5258
    if (vm)
        virDomainObjUnlock(vm);
5259 5260
    if (event)
        qemuDomainEventQueue(driver, event);
5261
    qemuDriverUnlock(driver);
P
Paolo Bonzini 已提交
5262 5263 5264 5265
    return ret;
}


5266 5267 5268 5269 5270 5271 5272 5273 5274 5275 5276 5277
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 */
5278
            qemuDomainObjEnterMonitor(vm);
5279
            rc = qemuMonitorSetCPU(priv->mon, i, 1);
5280
            qemuDomainObjExitMonitor(vm);
5281 5282 5283 5284 5285 5286 5287 5288 5289 5290
            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 */
5291
            qemuDomainObjEnterMonitor(vm);
5292
            rc = qemuMonitorSetCPU(priv->mon, i, 0);
5293
            qemuDomainObjExitMonitor(vm);
5294 5295 5296 5297 5298 5299 5300 5301 5302 5303 5304 5305 5306 5307 5308 5309 5310 5311 5312 5313 5314 5315
            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) {
5316 5317
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
5318 5319
    const char * type;
    int max;
5320
    int ret = -1;
5321

5322
    qemuDriverLock(driver);
5323
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5324 5325
    qemuDriverUnlock(driver);

5326 5327 5328
    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

5329
    if (!vm) {
5330 5331
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
5332 5333
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
5334
        goto endjob;
5335 5336
    }

5337 5338 5339
    if (!virDomainObjIsActive(vm)) {
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                         "%s", _("domain is not running"));
5340
        goto endjob;
5341 5342
    }

5343 5344 5345 5346 5347 5348 5349 5350 5351 5352 5353 5354 5355 5356 5357 5358 5359 5360 5361 5362 5363 5364 5365 5366 5367
    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;
5368

5369
cleanup:
5370 5371
    if (vm)
        virDomainObjUnlock(vm);
5372
    return ret;
5373 5374
}

5375 5376 5377 5378 5379 5380

static int
qemudDomainPinVcpu(virDomainPtr dom,
                   unsigned int vcpu,
                   unsigned char *cpumap,
                   int maplen) {
5381 5382
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
5383
    int maxcpu, hostcpus;
5384
    virNodeInfo nodeinfo;
5385
    int ret = -1;
5386
    qemuDomainObjPrivatePtr priv;
5387

5388
    qemuDriverLock(driver);
5389
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5390 5391
    qemuDriverUnlock(driver);

5392 5393 5394
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
5395 5396
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
5397 5398 5399
        goto cleanup;
    }

D
Daniel P. Berrange 已提交
5400
    if (!virDomainObjIsActive(vm)) {
5401 5402
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s",_("cannot pin vcpus on an inactive domain"));
5403
        goto cleanup;
5404 5405
    }

5406 5407 5408
    priv = vm->privateData;

    if (vcpu > (priv->nvcpupids-1)) {
5409 5410 5411
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("vcpu number out of range %d > %d"),
                        vcpu, priv->nvcpupids);
5412
        goto cleanup;
5413 5414
    }

5415
    if (nodeGetInfo(dom->conn, &nodeinfo) < 0)
5416
        goto cleanup;
5417

5418
    hostcpus = VIR_NODEINFO_MAXCPUS(nodeinfo);
5419
    maxcpu = maplen * 8;
5420 5421
    if (maxcpu > hostcpus)
        maxcpu = hostcpus;
5422

5423 5424
    if (priv->vcpupids != NULL) {
        if (virProcessInfoSetAffinity(priv->vcpupids[vcpu],
5425
                                      cpumap, maplen, maxcpu) < 0)
5426
            goto cleanup;
5427
    } else {
5428 5429
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        "%s", _("cpu affinity is not supported"));
5430
        goto cleanup;
5431
    }
5432
    ret = 0;
5433

5434
cleanup:
5435 5436
    if (vm)
        virDomainObjUnlock(vm);
5437
    return ret;
5438 5439 5440 5441 5442 5443 5444 5445
}

static int
qemudDomainGetVcpus(virDomainPtr dom,
                    virVcpuInfoPtr info,
                    int maxinfo,
                    unsigned char *cpumaps,
                    int maplen) {
5446 5447
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
5448
    virNodeInfo nodeinfo;
5449
    int i, v, maxcpu, hostcpus;
5450
    int ret = -1;
5451
    qemuDomainObjPrivatePtr priv;
5452

5453
    qemuDriverLock(driver);
5454
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5455 5456
    qemuDriverUnlock(driver);

5457 5458 5459
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
5460 5461
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
5462 5463 5464
        goto cleanup;
    }

D
Daniel P. Berrange 已提交
5465
    if (!virDomainObjIsActive(vm)) {
5466 5467 5468
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s",
                        _("cannot list vcpu pinning for an inactive domain"));
5469
        goto cleanup;
5470 5471
    }

5472 5473
    priv = vm->privateData;

5474
    if (nodeGetInfo(dom->conn, &nodeinfo) < 0)
5475
        goto cleanup;
5476

5477
    hostcpus = VIR_NODEINFO_MAXCPUS(nodeinfo);
5478
    maxcpu = maplen * 8;
5479 5480
    if (maxcpu > hostcpus)
        maxcpu = hostcpus;
5481 5482

    /* Clamp to actual number of vcpus */
5483 5484
    if (maxinfo > priv->nvcpupids)
        maxinfo = priv->nvcpupids;
5485

5486 5487 5488 5489 5490 5491
    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;
5492

5493
                if (priv->vcpupids != NULL &&
5494 5495 5496
                    qemudGetProcessInfo(&(info[i].cpuTime),
                                        &(info[i].cpu),
                                        vm->pid,
5497
                                        priv->vcpupids[i]) < 0) {
5498
                    virReportSystemError(errno, "%s",
5499 5500 5501
                                         _("cannot get vCPU placement & pCPU time"));
                    goto cleanup;
                }
5502
            }
5503 5504
        }

5505 5506
        if (cpumaps != NULL) {
            memset(cpumaps, 0, maplen * maxinfo);
5507
            if (priv->vcpupids != NULL) {
5508 5509 5510
                for (v = 0 ; v < maxinfo ; v++) {
                    unsigned char *cpumap = VIR_GET_CPUMAP(cpumaps, maplen, v);

5511
                    if (virProcessInfoGetAffinity(priv->vcpupids[v],
5512
                                                  cpumap, maplen, maxcpu) < 0)
5513
                        goto cleanup;
5514
                }
5515
            } else {
5516 5517
                qemuReportError(VIR_ERR_NO_SUPPORT,
                                "%s", _("cpu affinity is not available"));
5518
                goto cleanup;
5519 5520 5521
            }
        }
    }
5522
    ret = maxinfo;
5523

5524
cleanup:
5525 5526
    if (vm)
        virDomainObjUnlock(vm);
5527
    return ret;
5528 5529 5530
}


5531
static int qemudDomainGetMaxVcpus(virDomainPtr dom) {
5532 5533
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
5534
    const char *type;
5535
    int ret = -1;
5536

5537
    qemuDriverLock(driver);
5538
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5539 5540
    qemuDriverUnlock(driver);

5541
    if (!vm) {
5542 5543
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
5544 5545
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
5546
        goto cleanup;
5547 5548
    }

5549
    if (!(type = virDomainVirtTypeToString(vm->def->virtType))) {
5550 5551 5552
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("unknown virt type in domain definition '%d'"),
                        vm->def->virtType);
5553
        goto cleanup;
5554 5555
    }

5556
    ret = qemudGetMaxVCPUs(NULL, type);
5557

5558
cleanup:
5559 5560
    if (vm)
        virDomainObjUnlock(vm);
5561 5562 5563
    return ret;
}

5564 5565 5566 5567 5568 5569 5570 5571 5572 5573
static int qemudDomainGetSecurityLabel(virDomainPtr dom, virSecurityLabelPtr seclabel)
{
    struct qemud_driver *driver = (struct qemud_driver *)dom->conn->privateData;
    virDomainObjPtr vm;
    const char *type;
    int ret = -1;

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

5574 5575
    memset(seclabel, 0, sizeof(*seclabel));

5576 5577 5578
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
5579 5580
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
5581 5582 5583 5584
        goto cleanup;
    }

    if (!(type = virDomainVirtTypeToString(vm->def->virtType))) {
5585 5586 5587
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("unknown virt type in domain definition '%d'"),
                        vm->def->virtType);
5588 5589 5590 5591 5592 5593 5594 5595 5596 5597 5598 5599 5600 5601 5602 5603 5604
        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
5605
    if (virDomainObjIsActive(vm)) {
5606
        if (driver->securityDriver && driver->securityDriver->domainGetSecurityProcessLabel) {
5607
            if (driver->securityDriver->domainGetSecurityProcessLabel(vm, seclabel) == -1) {
5608 5609
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                "%s", _("Failed to get security label"));
5610 5611 5612 5613 5614 5615 5616 5617 5618 5619
                goto cleanup;
            }
        }
    }

    ret = 0;

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
5620
    qemuDriverUnlock(driver);
5621 5622 5623
    return ret;
}

5624 5625
static int qemudNodeGetSecurityModel(virConnectPtr conn,
                                     virSecurityModelPtr secmodel)
5626 5627 5628
{
    struct qemud_driver *driver = (struct qemud_driver *)conn->privateData;
    char *p;
5629
    int ret = 0;
5630

5631
    qemuDriverLock(driver);
5632
    if (!driver->securityPrimaryDriver) {
5633
        memset(secmodel, 0, sizeof (*secmodel));
5634 5635
        goto cleanup;
    }
5636

5637 5638
    p = driver->caps->host.secModel.model;
    if (strlen(p) >= VIR_SECURITY_MODEL_BUFLEN-1) {
5639 5640 5641
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("security model string exceeds max %d bytes"),
                        VIR_SECURITY_MODEL_BUFLEN-1);
5642 5643
        ret = -1;
        goto cleanup;
5644 5645 5646 5647 5648
    }
    strcpy(secmodel->model, p);

    p = driver->caps->host.secModel.doi;
    if (strlen(p) >= VIR_SECURITY_DOI_BUFLEN-1) {
5649 5650 5651
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("security DOI string exceeds max %d bytes"),
                        VIR_SECURITY_DOI_BUFLEN-1);
5652 5653
        ret = -1;
        goto cleanup;
5654 5655
    }
    strcpy(secmodel->doi, p);
5656 5657 5658 5659

cleanup:
    qemuDriverUnlock(driver);
    return ret;
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 5764 5765 5766 5767 5768 5769 5770 5771 5772 5773 5774 5775 5776 5777 5778 5779 5780 5781 5782 5783 5784 5785 5786 5787 5788 5789 5790 5791 5792 5793
/* 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);
}

5794
/* TODO: check seclabel restore */
5795
static int qemudDomainRestore(virConnectPtr conn,
5796 5797 5798
                              const char *path) {
    struct qemud_driver *driver = conn->privateData;
    virDomainDefPtr def = NULL;
5799
    virDomainObjPtr vm = NULL;
5800
    int fd = -1;
5801
    pid_t read_pid = -1;
5802 5803
    int ret = -1;
    char *xml = NULL;
5804
    struct qemud_save_header header;
5805
    virDomainEventPtr event = NULL;
5806 5807 5808
    int intermediatefd = -1;
    pid_t intermediate_pid = -1;
    int childstat;
5809

5810
    qemuDriverLock(driver);
5811 5812
    /* Verify the header and read the XML */
    if ((fd = open(path, O_RDONLY)) < 0) {
5813 5814 5815 5816 5817 5818 5819 5820 5821 5822 5823 5824 5825 5826
        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;
        }
5827 5828 5829
    }

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

    if (memcmp(header.magic, QEMUD_SAVE_MAGIC, sizeof(header.magic)) != 0) {
5836 5837
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("image magic is incorrect"));
5838
        goto cleanup;
5839 5840 5841
    }

    if (header.version > QEMUD_SAVE_VERSION) {
5842 5843 5844
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("image version is not supported (%d > %d)"),
                        header.version, QEMUD_SAVE_VERSION);
5845
        goto cleanup;
5846 5847
    }

5848 5849 5850
    if (header.xml_len <= 0) {
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("invalid XML length: %d"), header.xml_len);
5851
        goto cleanup;
5852 5853
    }

5854 5855 5856 5857 5858 5859
    if (VIR_ALLOC_N(xml, header.xml_len) < 0) {
        virReportOOMError();
        goto cleanup;
    }

    if (saferead(fd, xml, header.xml_len) != header.xml_len) {
5860 5861
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to read XML"));
5862
        goto cleanup;
5863 5864 5865
    }

    /* Create a domain from this XML */
5866
    if (!(def = virDomainDefParseString(driver->caps, xml,
5867
                                        VIR_DOMAIN_XML_INACTIVE))) {
5868 5869
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to parse XML"));
5870
        goto cleanup;
5871 5872
    }

5873 5874
    if (virDomainObjIsDuplicate(&driver->domains, def, 1) < 0)
        goto cleanup;
5875

5876
    if (!(vm = virDomainAssignDef(driver->caps,
5877
                                  &driver->domains,
5878
                                  def, true))) {
5879 5880
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to assign new VM"));
5881
        goto cleanup;
5882
    }
5883
    def = NULL;
5884

5885 5886 5887
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

5888 5889
    if (header.version == 2) {
        const char *intermediate_argv[3] = { NULL, "-dc", NULL };
5890 5891
        const char *prog = qemudSaveCompressionTypeToString(header.compressed);
        if (prog == NULL) {
5892 5893 5894
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("Invalid compressed save format %d"),
                            header.compressed);
5895
            goto endjob;
5896
        }
5897

5898
        if (header.compressed != QEMUD_SAVE_FORMAT_RAW) {
5899
            intermediate_argv[0] = prog;
5900 5901
            intermediatefd = fd;
            fd = -1;
5902
            if (virExec(intermediate_argv, NULL, NULL,
5903
                        &intermediate_pid, intermediatefd, &fd, NULL, 0) < 0) {
5904 5905 5906
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("Failed to start decompression binary %s"),
                                intermediate_argv[0]);
5907
                goto endjob;
5908 5909 5910
            }
        }
    }
5911
    /* Set the migration source and start it up. */
5912
    ret = qemudStartVMDaemon(conn, driver, vm, "stdio", fd);
5913 5914 5915 5916 5917 5918 5919
    if (intermediate_pid != -1) {
        /* Wait for intermediate process to exit */
        while (waitpid(intermediate_pid, &childstat, 0) == -1 &&
               errno == EINTR);
    }
    if (intermediatefd != -1)
        close(intermediatefd);
5920
    close(fd);
5921
    fd = -1;
5922 5923 5924 5925 5926 5927 5928 5929 5930 5931 5932 5933 5934 5935 5936 5937 5938 5939 5940 5941 5942 5943 5944 5945 5946 5947 5948 5949 5950
    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;
            }
        }
    }
5951
    if (ret < 0) {
5952
        if (!vm->persistent) {
5953 5954 5955
            if (qemuDomainObjEndJob(vm) > 0)
                virDomainRemoveInactive(&driver->domains,
                                        vm);
5956 5957
            vm = NULL;
        }
5958
        goto endjob;
5959 5960
    }

5961 5962 5963
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_RESTORED);
5964

5965 5966
    /* If it was running before, resume it now. */
    if (header.was_running) {
5967
        qemuDomainObjPrivatePtr priv = vm->privateData;
5968
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
5969
        if (qemuMonitorStartCPUs(priv->mon, conn) < 0) {
5970
            if (virGetLastError() == NULL)
5971 5972
                qemuReportError(VIR_ERR_OPERATION_FAILED,
                                "%s", _("failed to resume domain"));
5973 5974
            qemuDomainObjExitMonitorWithDriver(driver,vm);
            goto endjob;
5975
        }
5976
        qemuDomainObjExitMonitorWithDriver(driver, vm);
5977
        vm->state = VIR_DOMAIN_RUNNING;
5978
        virDomainSaveStatus(driver->caps, driver->stateDir, vm);
5979
    }
5980
    ret = 0;
5981

5982
endjob:
5983 5984 5985
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
5986

5987 5988 5989 5990 5991
cleanup:
    virDomainDefFree(def);
    VIR_FREE(xml);
    if (fd != -1)
        close(fd);
5992 5993 5994 5995 5996 5997 5998
    if (read_pid != 0) {
        /* reap the process that read the file */
        while ((waitpid(read_pid, NULL, 0) == -1)
               && (errno == EINTR)) {
            /* empty */
        }
    }
5999 6000
    if (vm)
        virDomainObjUnlock(vm);
6001 6002
    if (event)
        qemuDomainEventQueue(driver, event);
6003
    qemuDriverUnlock(driver);
6004
    return ret;
D
Daniel P. Berrange 已提交
6005 6006 6007
}


6008 6009 6010 6011 6012 6013 6014 6015 6016 6017 6018 6019 6020 6021 6022 6023 6024 6025 6026 6027 6028 6029 6030 6031 6032 6033 6034 6035 6036 6037 6038 6039 6040 6041 6042 6043 6044 6045
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;
}


6046
static char *qemudDomainDumpXML(virDomainPtr dom,
6047
                                int flags) {
6048 6049 6050
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    char *ret = NULL;
6051 6052
    unsigned long balloon;
    int err;
6053

6054
    qemuDriverLock(driver);
6055
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
6056

D
Daniel P. Berrange 已提交
6057
    if (!vm) {
6058 6059
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
6060 6061
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
6062
        goto cleanup;
D
Daniel P. Berrange 已提交
6063 6064
    }

6065
    /* Refresh current memory based on balloon info */
D
Daniel P. Berrange 已提交
6066
    if (virDomainObjIsActive(vm)) {
6067
        qemuDomainObjPrivatePtr priv = vm->privateData;
6068 6069 6070
        /* Don't delay if someone's using the monitor, just use
         * existing most recent data instead */
        if (!priv->jobActive) {
6071
            if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
6072 6073
                goto cleanup;

6074
            qemuDomainObjEnterMonitorWithDriver(driver, vm);
6075
            err = qemuMonitorGetBalloonInfo(priv->mon, &balloon);
6076
            qemuDomainObjExitMonitorWithDriver(driver, vm);
6077 6078 6079 6080
            if (qemuDomainObjEndJob(vm) == 0) {
                vm = NULL;
                goto cleanup;
            }
6081 6082 6083 6084 6085 6086
            if (err < 0)
                goto cleanup;
            if (err > 0)
                vm->def->memory = balloon;
            /* err == 0 indicates no balloon support, so ignore it */
        }
6087
    }
6088

6089
    ret = qemudVMDumpXML(driver, vm, flags);
6090 6091

cleanup:
6092 6093
    if (vm)
        virDomainObjUnlock(vm);
6094
    qemuDriverUnlock(driver);
6095
    return ret;
D
Daniel P. Berrange 已提交
6096 6097 6098
}


6099 6100 6101 6102
static char *qemuDomainXMLFromNative(virConnectPtr conn,
                                     const char *format,
                                     const char *config,
                                     unsigned int flags ATTRIBUTE_UNUSED) {
6103
    struct qemud_driver *driver = conn->privateData;
6104 6105 6106 6107
    virDomainDefPtr def = NULL;
    char *xml = NULL;

    if (STRNEQ(format, QEMU_CONFIG_FORMAT_ARGV)) {
6108 6109
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("unsupported config type %s"), format);
6110 6111 6112
        goto cleanup;
    }

6113
    qemuDriverLock(driver);
6114
    def = qemuParseCommandLineString(driver->caps, config);
6115
    qemuDriverUnlock(driver);
6116 6117 6118
    if (!def)
        goto cleanup;

6119
    xml = virDomainDefFormat(def, VIR_DOMAIN_XML_INACTIVE);
6120 6121 6122 6123 6124 6125

cleanup:
    virDomainDefFree(def);
    return xml;
}

6126 6127 6128 6129 6130 6131
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;
6132
    virDomainChrDef monConfig;
6133
    const char *emulator;
6134
    unsigned long long qemuCmdFlags;
6135 6136 6137 6138 6139 6140 6141 6142
    struct stat sb;
    const char **retargv = NULL;
    const char **retenv = NULL;
    const char **tmp;
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    char *ret = NULL;
    int i;

6143 6144
    qemuDriverLock(driver);

6145
    if (STRNEQ(format, QEMU_CONFIG_FORMAT_ARGV)) {
6146 6147
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("unsupported config type %s"), format);
6148 6149 6150
        goto cleanup;
    }

6151
    def = virDomainDefParseString(driver->caps, xmlData, 0);
6152 6153 6154 6155 6156 6157 6158 6159 6160 6161 6162 6163 6164 6165 6166 6167 6168 6169 6170 6171 6172 6173 6174 6175 6176 6177 6178 6179 6180 6181 6182 6183 6184 6185 6186 6187 6188 6189 6190 6191 6192 6193 6194 6195
    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) {
6196
        virReportSystemError(errno,
6197 6198 6199 6200 6201 6202 6203 6204
                             _("Cannot find QEMU binary %s"),
                             emulator);
        goto cleanup;
    }

    if (qemudExtractVersionInfo(emulator,
                                NULL,
                                &qemuCmdFlags) < 0) {
6205 6206 6207
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Cannot determine QEMU argv syntax %s"),
                        emulator);
6208 6209 6210
        goto cleanup;
    }

6211
    if (qemuPrepareMonitorChr(driver, &monConfig, def->name) < 0)
6212
        goto cleanup;
6213 6214

    if (qemudBuildCommandLine(conn, driver, def,
D
Daniel P. Berrange 已提交
6215
                              &monConfig, 0, qemuCmdFlags,
6216 6217
                              &retargv, &retenv,
                              NULL, NULL, /* Don't want it to create TAP devices */
C
Chris Lalancette 已提交
6218
                              NULL, NULL) < 0) {
6219 6220 6221 6222 6223 6224 6225 6226 6227 6228 6229 6230 6231 6232 6233 6234
        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++;
    }

6235 6236
    if (virBufferError(&buf)) {
        virBufferFreeAndReset(&buf);
6237
        virReportOOMError();
6238
        goto cleanup;
6239
    }
6240 6241 6242 6243

    ret = virBufferContentAndReset(&buf);

cleanup:
6244
    qemuDriverUnlock(driver);
6245 6246 6247 6248 6249 6250 6251 6252 6253 6254 6255 6256 6257
    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;
}


6258
static int qemudListDefinedDomains(virConnectPtr conn,
6259
                            char **const names, int nnames) {
6260
    struct qemud_driver *driver = conn->privateData;
6261
    int n;
6262

6263
    qemuDriverLock(driver);
6264
    n = virDomainObjListGetInactiveNames(&driver->domains, names, nnames);
6265
    qemuDriverUnlock(driver);
6266
    return n;
D
Daniel P. Berrange 已提交
6267 6268
}

6269
static int qemudNumDefinedDomains(virConnectPtr conn) {
6270
    struct qemud_driver *driver = conn->privateData;
6271
    int n;
6272

6273
    qemuDriverLock(driver);
6274
    n = virDomainObjListNumOfDomains(&driver->domains, 0);
6275
    qemuDriverUnlock(driver);
6276

6277
    return n;
D
Daniel P. Berrange 已提交
6278 6279 6280
}


6281
static int qemudDomainStart(virDomainPtr dom) {
6282 6283 6284
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
6285
    virDomainEventPtr event = NULL;
6286
    char *managed_save = NULL;
6287

6288
    qemuDriverLock(driver);
6289
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
6290

6291
    if (!vm) {
6292 6293
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
6294 6295
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
6296
        goto cleanup;
6297 6298
    }

6299 6300 6301 6302
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

    if (virDomainObjIsActive(vm)) {
6303 6304
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is already running"));
6305 6306 6307
        goto endjob;
    }

6308 6309 6310 6311 6312 6313 6314 6315 6316 6317 6318 6319 6320 6321 6322 6323 6324 6325 6326 6327 6328 6329 6330 6331 6332 6333 6334 6335 6336
    /*
     * 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);
    }

6337
    ret = qemudStartVMDaemon(dom->conn, driver, vm, NULL, -1);
6338
    if (ret != -1)
6339 6340 6341
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STARTED,
                                         VIR_DOMAIN_EVENT_STARTED_BOOTED);
6342

6343
endjob:
6344 6345
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
6346

6347
cleanup:
6348
    VIR_FREE(managed_save);
6349 6350
    if (vm)
        virDomainObjUnlock(vm);
6351
    if (event)
6352
        qemuDomainEventQueue(driver, event);
6353
    qemuDriverUnlock(driver);
6354
    return ret;
D
Daniel P. Berrange 已提交
6355 6356
}

6357 6358 6359 6360 6361 6362 6363 6364 6365 6366 6367 6368 6369 6370 6371
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;

6372
        if (STRNEQ(def->os.machine, machine->name))
6373 6374 6375
            continue;

        if (!(*canonical = strdup(machine->canonical))) {
6376
            virReportOOMError();
6377 6378 6379 6380 6381 6382 6383 6384 6385
            return -1;
        }

        break;
    }

    return 0;
}

6386 6387 6388 6389 6390 6391 6392
static int
qemudCanonicalizeMachineDirect(virDomainDefPtr def, char **canonical)
{
    virCapsGuestMachinePtr *machines = NULL;
    int i, nmachines = 0;

    if (qemudProbeMachineTypes(def->emulator, &machines, &nmachines) < 0) {
6393
        virReportOOMError();
6394 6395 6396 6397 6398 6399 6400
        return -1;
    }

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

6401
        if (STRNEQ(def->os.machine, machines[i]->name))
6402 6403 6404 6405 6406 6407 6408 6409 6410 6411 6412 6413
            continue;

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

    virCapabilitiesFreeMachines(machines, nmachines);

    return 0;
}

6414 6415
int
qemudCanonicalizeMachine(struct qemud_driver *driver, virDomainDefPtr def)
6416 6417 6418 6419 6420 6421
{
    char *canonical = NULL;
    int i;

    for (i = 0; i < driver->caps->nguests; i++) {
        virCapsGuestPtr guest = driver->caps->guests[i];
6422
        virCapsGuestDomainInfoPtr info;
6423 6424 6425
        int j;

        for (j = 0; j < guest->arch.ndomains; j++) {
6426
            info = &guest->arch.domains[j]->info;
6427

6428 6429 6430 6431 6432 6433 6434 6435 6436
            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;
6437 6438
        }

6439 6440 6441 6442
        info = &guest->arch.defaultInfo;

        if (info->emulator && STREQ(info->emulator, def->emulator)) {
            if (qemudCanonicalizeMachineFromInfo(def, info, &canonical) < 0)
6443 6444 6445 6446
                return -1;
            goto out;
        }
    }
6447 6448 6449 6450

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

6451 6452 6453 6454 6455 6456 6457
out:
    if (canonical) {
        VIR_FREE(def->os.machine);
        def->os.machine = canonical;
    }
    return 0;
}
D
Daniel P. Berrange 已提交
6458

6459
static virDomainPtr qemudDomainDefine(virConnectPtr conn, const char *xml) {
6460
    struct qemud_driver *driver = conn->privateData;
6461
    virDomainDefPtr def;
6462
    virDomainObjPtr vm = NULL;
6463
    virDomainPtr dom = NULL;
6464
    virDomainEventPtr event = NULL;
6465
    int dupVM;
6466

6467
    qemuDriverLock(driver);
6468
    if (!(def = virDomainDefParseString(driver->caps, xml,
6469
                                        VIR_DOMAIN_XML_INACTIVE)))
6470
        goto cleanup;
6471

6472
    if (virSecurityDriverVerify(def) < 0)
6473 6474
        goto cleanup;

6475 6476
    if ((dupVM = virDomainObjIsDuplicate(&driver->domains, def, 0)) < 0)
        goto cleanup;
6477

6478
    if (qemudCanonicalizeMachine(driver, def) < 0)
6479 6480
        goto cleanup;

6481 6482 6483
    if (qemuAssignPCIAddresses(def) < 0)
        goto cleanup;

6484
    if (!(vm = virDomainAssignDef(driver->caps,
6485
                                  &driver->domains,
6486
                                  def, false))) {
6487
        goto cleanup;
6488
    }
6489
    def = NULL;
6490
    vm->persistent = 1;
6491

6492
    if (virDomainSaveConfig(driver->configDir,
6493
                            vm->newDef ? vm->newDef : vm->def) < 0) {
6494 6495
        virDomainRemoveInactive(&driver->domains,
                                vm);
6496
        vm = NULL;
6497
        goto cleanup;
6498 6499
    }

6500 6501
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_DEFINED,
6502
                                     !dupVM ?
6503 6504
                                     VIR_DOMAIN_EVENT_DEFINED_ADDED :
                                     VIR_DOMAIN_EVENT_DEFINED_UPDATED);
6505

6506
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
6507
    if (dom) dom->id = vm->def->id;
6508 6509

cleanup:
6510
    virDomainDefFree(def);
6511 6512
    if (vm)
        virDomainObjUnlock(vm);
6513 6514
    if (event)
        qemuDomainEventQueue(driver, event);
6515
    qemuDriverUnlock(driver);
6516
    return dom;
D
Daniel P. Berrange 已提交
6517 6518
}

6519
static int qemudDomainUndefine(virDomainPtr dom) {
6520 6521
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
6522
    virDomainEventPtr event = NULL;
6523
    int ret = -1;
D
Daniel P. Berrange 已提交
6524

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

D
Daniel P. Berrange 已提交
6528
    if (!vm) {
6529 6530
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
6531 6532
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
6533
        goto cleanup;
D
Daniel P. Berrange 已提交
6534 6535
    }

D
Daniel P. Berrange 已提交
6536
    if (virDomainObjIsActive(vm)) {
6537 6538
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot delete active domain"));
6539
        goto cleanup;
D
Daniel P. Berrange 已提交
6540 6541
    }

6542
    if (!vm->persistent) {
6543 6544
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("cannot undefine transient domain"));
6545
        goto cleanup;
6546 6547
    }

6548
    if (virDomainDeleteConfig(driver->configDir, driver->autostartDir, vm) < 0)
6549
        goto cleanup;
D
Daniel P. Berrange 已提交
6550

6551 6552 6553
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_UNDEFINED,
                                     VIR_DOMAIN_EVENT_UNDEFINED_REMOVED);
6554

6555 6556
    virDomainRemoveInactive(&driver->domains,
                            vm);
6557
    vm = NULL;
6558
    ret = 0;
D
Daniel P. Berrange 已提交
6559

6560
cleanup:
6561 6562
    if (vm)
        virDomainObjUnlock(vm);
6563 6564
    if (event)
        qemuDomainEventQueue(driver, event);
6565
    qemuDriverUnlock(driver);
6566
    return ret;
D
Daniel P. Berrange 已提交
6567 6568
}

6569

6570
static int qemudDomainChangeEjectableMedia(struct qemud_driver *driver,
6571
                                           virDomainObjPtr vm,
6572
                                           virDomainDiskDefPtr disk)
6573
{
6574
    virDomainDiskDefPtr origdisk = NULL;
6575
    int i;
6576
    int ret;
6577

6578
    origdisk = NULL;
6579
    for (i = 0 ; i < vm->def->ndisks ; i++) {
6580 6581
        if (vm->def->disks[i]->bus == disk->bus &&
            STREQ(vm->def->disks[i]->dst, disk->dst)) {
6582
            origdisk = vm->def->disks[i];
6583
            break;
6584
        }
6585 6586 6587
    }

    if (!origdisk) {
6588 6589 6590 6591
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("No device with bus '%s' and target '%s'"),
                        virDomainDiskBusTypeToString(disk->bus),
                        disk->dst);
6592 6593 6594
        return -1;
    }

6595
    if (!origdisk->info.alias) {
6596 6597
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("missing disk device alias name for %s"), origdisk->dst);
6598 6599
        return -1;
    }
6600

6601 6602
    if (origdisk->device != VIR_DOMAIN_DISK_DEVICE_FLOPPY &&
        origdisk->device != VIR_DOMAIN_DISK_DEVICE_CDROM) {
6603 6604 6605
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Removable media not supported for %s device"),
                        virDomainDiskDeviceTypeToString(disk->device));
6606
        return -1;
6607
    }
6608

6609 6610
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityImageLabel &&
6611
        driver->securityDriver->domainSetSecurityImageLabel(vm, disk) < 0)
6612 6613
        return -1;

6614
    qemuDomainObjPrivatePtr priv = vm->privateData;
6615
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
6616
    if (disk->src) {
6617
        const char *format = NULL;
6618 6619 6620
        if (disk->type != VIR_DOMAIN_DISK_TYPE_DIR) {
            if (disk->driverType)
                format = disk->driverType;
6621 6622 6623
            else if (origdisk->driverType)
                format = origdisk->driverType;
        }
6624 6625
        ret = qemuMonitorChangeMedia(priv->mon,
                                     origdisk->info.alias,
6626
                                     disk->src, format);
6627
    } else {
6628
        ret = qemuMonitorEjectMedia(priv->mon, origdisk->info.alias);
6629
    }
6630
    qemuDomainObjExitMonitorWithDriver(driver, vm);
6631

6632 6633 6634 6635 6636
    if (ret < 0)
        goto error;

    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
6637
        driver->securityDriver->domainRestoreSecurityImageLabel(vm, origdisk) < 0)
6638 6639 6640
        VIR_WARN("Unable to restore security label on ejected image %s", origdisk->src);

    VIR_FREE(origdisk->src);
6641 6642 6643 6644 6645
    origdisk->src = disk->src;
    disk->src = NULL;
    origdisk->type = disk->type;

    virDomainDiskDefFree(disk);
6646

6647
    return ret;
6648 6649 6650 6651

error:
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
6652
        driver->securityDriver->domainRestoreSecurityImageLabel(vm, disk) < 0)
6653
        VIR_WARN("Unable to restore security label on new media %s", disk->src);
6654
    return -1;
6655 6656
}

6657

6658
static int qemudDomainAttachPciDiskDevice(struct qemud_driver *driver,
6659
                                          virDomainObjPtr vm,
6660 6661
                                          virDomainDiskDefPtr disk,
                                          int qemuCmdFlags)
6662
{
6663
    int i, ret;
6664
    const char* type = virDomainDiskBusTypeToString(disk->bus);
6665
    qemuDomainObjPrivatePtr priv = vm->privateData;
6666 6667
    char *devstr = NULL;
    char *drivestr = NULL;
6668 6669

    for (i = 0 ; i < vm->def->ndisks ; i++) {
6670
        if (STREQ(vm->def->disks[i]->dst, disk->dst)) {
6671 6672
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("target %s already exists"), disk->dst);
6673 6674 6675 6676
            return -1;
        }
    }

6677 6678
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityImageLabel &&
6679
        driver->securityDriver->domainSetSecurityImageLabel(vm, disk) < 0)
6680 6681
        return -1;

6682
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
6683 6684
        if (qemuDomainPCIAddressEnsureAddr(priv->pciaddrs, &disk->info) < 0)
            goto error;
6685 6686
        if (qemuAssignDeviceDiskAlias(disk, qemuCmdFlags) < 0)
            goto error;
6687

6688 6689 6690
        if (!(drivestr = qemuBuildDriveStr(disk, 0, qemuCmdFlags)))
            goto error;

6691
        if (!(devstr = qemuBuildDriveDevStr(disk)))
6692 6693 6694
            goto error;
    }

6695
    if (VIR_REALLOC_N(vm->def->disks, vm->def->ndisks+1) < 0) {
6696
        virReportOOMError();
6697
        goto error;
6698 6699
    }

6700
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
6701 6702 6703 6704 6705 6706 6707 6708 6709 6710 6711 6712 6713 6714 6715 6716
    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));
        }
    }
6717
    qemuDomainObjExitMonitorWithDriver(driver, vm);
6718

6719 6720
    if (ret < 0)
        goto error;
6721

6722
    virDomainDiskInsertPreAlloced(vm->def, disk);
6723

6724 6725 6726
    VIR_FREE(devstr);
    VIR_FREE(drivestr);

6727 6728 6729
    return 0;

error:
6730 6731 6732
    VIR_FREE(devstr);
    VIR_FREE(drivestr);

6733 6734 6735 6736 6737
    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);

6738 6739
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
6740
        driver->securityDriver->domainRestoreSecurityImageLabel(vm, disk) < 0)
6741
        VIR_WARN("Unable to restore security label on %s", disk->src);
6742 6743

    return -1;
6744
}
6745

6746

6747
static int qemudDomainAttachPciControllerDevice(struct qemud_driver *driver,
6748
                                                virDomainObjPtr vm,
6749 6750
                                                virDomainControllerDefPtr controller,
                                                int qemuCmdFlags)
6751
{
6752 6753
    int i;
    int ret = -1;
6754
    const char* type = virDomainControllerTypeToString(controller->type);
6755
    char *devstr = NULL;
6756 6757 6758
    qemuDomainObjPrivatePtr priv = vm->privateData;

    for (i = 0 ; i < vm->def->ncontrollers ; i++) {
6759 6760
        if ((vm->def->controllers[i]->type == controller->type) &&
            (vm->def->controllers[i]->idx == controller->idx)) {
6761 6762 6763
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("target %s:%d already exists"),
                            type, controller->idx);
6764 6765 6766 6767
            return -1;
        }
    }

6768 6769 6770 6771 6772 6773
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuDomainPCIAddressEnsureAddr(priv->pciaddrs, &controller->info) < 0)
            goto cleanup;
        if (qemuAssignDeviceControllerAlias(controller) < 0)
            goto cleanup;
    }
6774

6775
    if (!(devstr = qemuBuildControllerDevStr(controller))) {
6776
        virReportOOMError();
6777 6778 6779
        goto cleanup;
    }

6780
    if (VIR_REALLOC_N(vm->def->controllers, vm->def->ncontrollers+1) < 0) {
6781
        virReportOOMError();
6782
        goto cleanup;
6783 6784 6785
    }

    qemuDomainObjEnterMonitorWithDriver(driver, vm);
6786 6787 6788 6789 6790 6791 6792
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        ret = qemuMonitorAddDevice(priv->mon, devstr);
    } else {
        ret = qemuMonitorAttachPCIDiskController(priv->mon,
                                                 type,
                                                 &controller->info.addr.pci);
    }
6793 6794 6795
    qemuDomainObjExitMonitorWithDriver(driver, vm);

    if (ret == 0) {
6796 6797
        controller->info.type = VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI;
        virDomainControllerInsertPreAlloced(vm->def, controller);
6798 6799
    }

6800
cleanup:
6801 6802 6803 6804 6805 6806
    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");

6807
    VIR_FREE(devstr);
6808 6809 6810
    return ret;
}

6811

6812
static virDomainControllerDefPtr
6813
qemuDomainFindOrCreateSCSIDiskController(struct qemud_driver *driver,
6814
                                         virDomainObjPtr vm,
6815 6816
                                         int controller,
                                         int qemuCmdFlags)
6817 6818 6819 6820 6821 6822 6823 6824 6825 6826 6827 6828 6829
{
    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
6830
    /* No SCSI controller present, for backward compatibility we
6831 6832
     * now hotplug a controller */
    if (VIR_ALLOC(cont) < 0) {
6833
        virReportOOMError();
6834 6835 6836 6837 6838 6839
        return NULL;
    }
    cont->type = VIR_DOMAIN_CONTROLLER_TYPE_SCSI;
    cont->idx = 0;

    VIR_INFO0("No SCSI controller present, hotplugging one");
6840
    if (qemudDomainAttachPciControllerDevice(driver,
6841
                                             vm, cont, qemuCmdFlags) < 0) {
6842 6843 6844 6845 6846 6847
        VIR_FREE(cont);
        return NULL;
    }
    return cont;
}

6848

6849
static int qemudDomainAttachSCSIDisk(struct qemud_driver *driver,
6850
                                     virDomainObjPtr vm,
6851
                                     virDomainDiskDefPtr disk,
6852 6853 6854 6855 6856 6857
                                     int qemuCmdFlags)
{
    int i;
    qemuDomainObjPrivatePtr priv = vm->privateData;
    virDomainControllerDefPtr cont;
    char *drivestr = NULL;
6858
    char *devstr = NULL;
6859 6860 6861
    int ret = -1;

    for (i = 0 ; i < vm->def->ndisks ; i++) {
6862
        if (STREQ(vm->def->disks[i]->dst, disk->dst)) {
6863 6864
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("target %s already exists"), disk->dst);
6865
            return -1;
6866 6867 6868
        }
    }

6869 6870 6871

    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityImageLabel &&
6872
        driver->securityDriver->domainSetSecurityImageLabel(vm, disk) < 0)
6873 6874
        return -1;

6875
    /* We should have an address already, so make sure */
6876
    if (disk->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_DRIVE) {
6877 6878 6879
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("unexpected disk address type %s"),
                        virDomainDeviceAddressTypeToString(disk->info.type));
6880
        goto error;
6881 6882
    }

6883 6884 6885
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuAssignDeviceDiskAlias(disk, qemuCmdFlags) < 0)
            goto error;
6886
        if (!(devstr = qemuBuildDriveDevStr(disk)))
6887 6888 6889 6890 6891 6892
            goto error;
    }

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

6893
    for (i = 0 ; i <= disk->info.addr.drive.controller ; i++) {
6894
        cont = qemuDomainFindOrCreateSCSIDiskController(driver, vm, i, qemuCmdFlags);
6895
        if (!cont)
6896
            goto error;
6897 6898 6899
    }

    if (cont->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI) {
6900 6901
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("SCSI controller %d was missing its PCI address"), cont->idx);
6902
        goto error;
6903 6904 6905
    }

    if (VIR_REALLOC_N(vm->def->disks, vm->def->ndisks+1) < 0) {
6906
        virReportOOMError();
6907
        goto error;
6908 6909 6910
    }

    qemuDomainObjEnterMonitorWithDriver(driver, vm);
6911 6912 6913 6914 6915 6916 6917 6918 6919 6920 6921 6922 6923 6924 6925 6926 6927 6928 6929 6930
    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));
        }
    }
6931 6932
    qemuDomainObjExitMonitorWithDriver(driver, vm);

6933 6934
    if (ret < 0)
        goto error;
6935

6936
    virDomainDiskInsertPreAlloced(vm->def, disk);
6937 6938

    VIR_FREE(devstr);
6939
    VIR_FREE(drivestr);
6940 6941 6942 6943

    return 0;

error:
6944
    VIR_FREE(devstr);
6945
    VIR_FREE(drivestr);
6946

6947 6948
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
6949
        driver->securityDriver->domainRestoreSecurityImageLabel(vm, disk) < 0)
6950
        VIR_WARN("Unable to restore security label on %s", disk->src);
6951 6952

    return -1;
6953 6954
}

6955

6956
static int qemudDomainAttachUsbMassstorageDevice(struct qemud_driver *driver,
6957
                                                 virDomainObjPtr vm,
6958 6959
                                                 virDomainDiskDefPtr disk,
                                                 int qemuCmdFlags)
6960
{
6961
    qemuDomainObjPrivatePtr priv = vm->privateData;
6962
    int i, ret;
6963 6964
    char *drivestr = NULL;
    char *devstr = NULL;
6965

6966
    for (i = 0 ; i < vm->def->ndisks ; i++) {
6967
        if (STREQ(vm->def->disks[i]->dst, disk->dst)) {
6968 6969
            qemuReportError(VIR_ERR_OPERATION_FAILED,
                            _("target %s already exists"), disk->dst);
6970 6971 6972 6973
            return -1;
        }
    }

6974 6975
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityImageLabel &&
6976
        driver->securityDriver->domainSetSecurityImageLabel(vm, disk) < 0)
6977 6978
        return -1;

6979
    if (!disk->src) {
6980 6981
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("disk source path is missing"));
6982
        goto error;
6983 6984
    }

6985
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
6986 6987
        if (qemuAssignDeviceDiskAlias(disk, qemuCmdFlags) < 0)
            goto error;
6988 6989
        if (!(drivestr = qemuBuildDriveStr(disk, 0, qemuCmdFlags)))
            goto error;
6990
        if (!(devstr = qemuBuildDriveDevStr(disk)))
6991 6992 6993
            goto error;
    }

6994
    if (VIR_REALLOC_N(vm->def->disks, vm->def->ndisks+1) < 0) {
6995
        virReportOOMError();
6996
        goto error;
6997 6998
    }

6999
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7000 7001 7002 7003 7004 7005 7006 7007 7008 7009
    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);
    }
7010
    qemuDomainObjExitMonitorWithDriver(driver, vm);
7011

7012 7013 7014
    if (ret < 0)
        goto error;

7015
    virDomainDiskInsertPreAlloced(vm->def, disk);
7016

7017 7018 7019
    VIR_FREE(devstr);
    VIR_FREE(drivestr);

7020 7021 7022
    return 0;

error:
7023 7024 7025
    VIR_FREE(devstr);
    VIR_FREE(drivestr);

7026 7027
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
7028
        driver->securityDriver->domainRestoreSecurityImageLabel(vm, disk) < 0)
7029
        VIR_WARN("Unable to restore security label on %s", disk->src);
7030 7031

    return -1;
7032 7033
}

7034

7035
/* XXX conn required for network -> bridge resolution */
M
Mark McLoughlin 已提交
7036
static int qemudDomainAttachNetDevice(virConnectPtr conn,
7037
                                      struct qemud_driver *driver,
M
Mark McLoughlin 已提交
7038
                                      virDomainObjPtr vm,
7039
                                      virDomainNetDefPtr net,
7040
                                      unsigned long long qemuCmdFlags)
M
Mark McLoughlin 已提交
7041
{
7042
    qemuDomainObjPrivatePtr priv = vm->privateData;
7043
    char *tapfd_name = NULL;
7044
    int tapfd = -1;
7045
    char *nicstr = NULL;
7046
    char *netstr = NULL;
7047
    int ret = -1;
7048
    virDomainDevicePCIAddress guestAddr;
7049
    int vlan;
M
Mark McLoughlin 已提交
7050 7051

    if (!(qemuCmdFlags & QEMUD_CMD_FLAG_HOST_NET_ADD)) {
7052 7053
        qemuReportError(VIR_ERR_NO_SUPPORT, "%s",
                        _("installed qemu version does not support host_net_add"));
M
Mark McLoughlin 已提交
7054 7055 7056 7057 7058
        return -1;
    }

    if (net->type == VIR_DOMAIN_NET_TYPE_BRIDGE ||
        net->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
7059
        if (priv->monConfig->type != VIR_DOMAIN_CHR_TYPE_UNIX) {
7060 7061 7062 7063
            qemuReportError(VIR_ERR_NO_SUPPORT,
                            _("network device type '%s' cannot be attached: "
                              "qemu is not using a unix socket monitor"),
                            virDomainNetTypeToString(net->type));
7064 7065 7066 7067 7068
            return -1;
        }

        if ((tapfd = qemudNetworkIfaceConnect(conn, driver, net, qemuCmdFlags)) < 0)
            return -1;
7069 7070 7071 7072 7073 7074 7075 7076 7077
    } 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
7078
        if ((tapfd = qemudPhysIfaceConnect(conn, driver, net,
7079
                                           net->data.direct.linkdev,
7080 7081
                                           net->data.direct.mode,
                                           qemuCmdFlags)) < 0)
7082
            return -1;
M
Mark McLoughlin 已提交
7083 7084
    }

7085 7086
    if (VIR_REALLOC_N(vm->def->nets, vm->def->nnets+1) < 0)
        goto no_memory;
M
Mark McLoughlin 已提交
7087

7088 7089 7090 7091 7092
    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
7093

7094 7095 7096 7097
    if ((qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) &&
        qemuDomainPCIAddressEnsureAddr(priv->pciaddrs, &net->info) < 0)
        goto cleanup;

7098
    vlan = qemuDomainNetVLAN(net);
M
Mark McLoughlin 已提交
7099

7100
    if (vlan < 0) {
7101 7102
        qemuReportError(VIR_ERR_NO_SUPPORT, "%s",
                        _("Unable to attach network devices without vlan"));
7103 7104 7105
        goto cleanup;
    }

7106
    if (tapfd != -1) {
7107
        if (virAsprintf(&tapfd_name, "fd-%s", net->info.alias) < 0)
7108 7109
            goto no_memory;

7110
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
7111
        if (qemuMonitorSendFileHandle(priv->mon, tapfd_name, tapfd) < 0) {
7112
            qemuDomainObjExitMonitorWithDriver(driver, vm);
7113
            goto cleanup;
7114
        }
7115
        qemuDomainObjExitMonitorWithDriver(driver, vm);
7116 7117
    }

7118
    /* FIXME - need to support vhost-net here (5th arg) */
7119
    if (!(netstr = qemuBuildHostNetStr(net, ' ',
7120
                                       vlan, tapfd_name, 0)))
7121
        goto try_tapfd_close;
M
Mark McLoughlin 已提交
7122

7123
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7124
    if (qemuMonitorAddHostNetwork(priv->mon, netstr) < 0) {
7125
        qemuDomainObjExitMonitorWithDriver(driver, vm);
7126
        goto try_tapfd_close;
7127
    }
7128
    qemuDomainObjExitMonitorWithDriver(driver, vm);
M
Mark McLoughlin 已提交
7129

7130 7131 7132
    if (tapfd != -1)
        close(tapfd);
    tapfd = -1;
M
Mark McLoughlin 已提交
7133

7134
    if (!(nicstr = qemuBuildNicStr(net, NULL, vlan)))
7135
        goto try_remove;
M
Mark McLoughlin 已提交
7136

7137
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7138
    if (qemuMonitorAddPCINetwork(priv->mon, nicstr,
7139
                                 &guestAddr) < 0) {
7140
        qemuDomainObjExitMonitorWithDriver(driver, vm);
7141
        goto try_remove;
7142
    }
7143
    net->info.type = VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI;
7144
    memcpy(&net->info.addr.pci, &guestAddr, sizeof(guestAddr));
7145
    qemuDomainObjExitMonitorWithDriver(driver, vm);
7146

7147
    ret = 0;
M
Mark McLoughlin 已提交
7148 7149 7150

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

7151
cleanup:
7152 7153 7154 7155 7156 7157
    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");

S
Stefan Berger 已提交
7158 7159
    qemuTearNWFilter(net);

7160 7161 7162 7163 7164
    VIR_FREE(nicstr);
    VIR_FREE(netstr);
    VIR_FREE(tapfd_name);
    if (tapfd != -1)
        close(tapfd);
7165

7166
    return ret;
7167

7168
try_remove:
7169
    if (vlan < 0) {
7170
        VIR_WARN0(_("Unable to remove network backend"));
7171 7172 7173 7174
    } else {
        char *hostnet_name;
        if (virAsprintf(&hostnet_name, "host%s", net->info.alias) < 0)
            goto no_memory;
7175
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
7176
        if (qemuMonitorRemoveHostNetwork(priv->mon, vlan, hostnet_name) < 0)
7177
            VIR_WARN(_("Failed to remove network backend for vlan %d, net %s"),
7178
                     vlan, hostnet_name);
7179
        qemuDomainObjExitMonitorWithDriver(driver, vm);
7180
        VIR_FREE(hostnet_name);
7181
    }
7182
    goto cleanup;
7183

7184
try_tapfd_close:
7185
    if (tapfd_name) {
7186
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
7187
        if (qemuMonitorCloseFileHandle(priv->mon, tapfd_name) < 0)
7188
            VIR_WARN(_("Failed to close tapfd with '%s'"), tapfd_name);
7189
        qemuDomainObjExitMonitorWithDriver(driver, vm);
7190
    }
7191

7192 7193
    goto cleanup;

7194
no_memory:
7195
    virReportOOMError();
7196
    goto cleanup;
M
Mark McLoughlin 已提交
7197 7198
}

7199

7200
static int qemudDomainAttachHostPciDevice(struct qemud_driver *driver,
7201
                                          virDomainObjPtr vm,
7202 7203
                                          virDomainHostdevDefPtr hostdev,
                                          int qemuCmdFlags)
7204
{
7205
    qemuDomainObjPrivatePtr priv = vm->privateData;
7206
    pciDevice *pci;
7207
    int ret;
7208
    virDomainDevicePCIAddress guestAddr;
7209
    char *devstr = NULL;
7210 7211

    if (VIR_REALLOC_N(vm->def->hostdevs, vm->def->nhostdevs+1) < 0) {
7212
        virReportOOMError();
7213 7214 7215
        return -1;
    }

7216
    pci = pciGetDevice(hostdev->source.subsys.u.pci.domain,
7217 7218 7219
                       hostdev->source.subsys.u.pci.bus,
                       hostdev->source.subsys.u.pci.slot,
                       hostdev->source.subsys.u.pci.function);
7220
    if (!pci)
7221
        return -1;
7222

7223 7224 7225 7226
    if (!pciDeviceIsAssignable(pci, !driver->relaxedACS) ||
        (hostdev->managed && pciDettachDevice(pci) < 0) ||
        pciResetDevice(pci, driver->activePciHostdevs) < 0) {
        pciFreeDevice(pci);
7227 7228 7229
        return -1;
    }

7230 7231
    if (pciDeviceListAdd(driver->activePciHostdevs, pci) < 0) {
        pciFreeDevice(pci);
7232
        return -1;
7233 7234
    }

7235
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
7236 7237
        if (qemuAssignDeviceHostdevAlias(vm->def, hostdev, -1) < 0)
            goto error;
7238 7239 7240 7241 7242 7243
        if (qemuDomainPCIAddressEnsureAddr(priv->pciaddrs, &hostdev->info) < 0)
            goto error;

        if (!(devstr = qemuBuildPCIHostdevDevStr(hostdev)))
            goto error;
    }
7244

7245
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7246 7247 7248 7249 7250 7251
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)
        ret = qemuMonitorAddDevice(priv->mon, devstr);
    else
        ret = qemuMonitorAddPCIHostDevice(priv->mon,
                                          &hostdev->source.subsys.u.pci,
                                          &guestAddr);
7252
    qemuDomainObjExitMonitorWithDriver(driver, vm);
7253
    if (ret < 0)
7254
        goto error;
7255
    hostdev->info.type = VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI;
7256
    memcpy(&hostdev->info.addr.pci, &guestAddr, sizeof(guestAddr));
7257 7258 7259

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

7260 7261
    VIR_FREE(devstr);

7262
    return 0;
7263 7264

error:
7265 7266 7267 7268 7269
    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");

7270
    VIR_FREE(devstr);
7271
    pciDeviceListDel(driver->activePciHostdevs, pci);
7272 7273

    return -1;
7274 7275
}

7276

7277
static int qemudDomainAttachHostUsbDevice(struct qemud_driver *driver,
M
Mark McLoughlin 已提交
7278
                                          virDomainObjPtr vm,
7279 7280
                                          virDomainHostdevDefPtr hostdev,
                                          int qemuCmdFlags)
7281 7282
{
    int ret;
7283
    qemuDomainObjPrivatePtr priv = vm->privateData;
7284 7285
    char *devstr = NULL;

7286 7287 7288 7289 7290 7291
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuAssignDeviceHostdevAlias(vm->def, hostdev, -1) < 0)
            goto error;
        if (!(devstr = qemuBuildUSBHostdevDevStr(hostdev)))
            goto error;
    }
7292

7293
    if (VIR_REALLOC_N(vm->def->hostdevs, vm->def->nhostdevs+1) < 0) {
7294
        virReportOOMError();
7295
        goto error;
7296
    }
7297

7298
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7299 7300 7301
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE)
        ret = qemuMonitorAddDevice(priv->mon, devstr);
    else
7302
        ret = qemuMonitorAddUSBDeviceExact(priv->mon,
7303 7304
                                           hostdev->source.subsys.u.usb.bus,
                                           hostdev->source.subsys.u.usb.device);
7305
    qemuDomainObjExitMonitorWithDriver(driver, vm);
7306 7307 7308 7309
    if (ret < 0)
        goto error;

    vm->def->hostdevs[vm->def->nhostdevs++] = hostdev;
7310

7311
    VIR_FREE(devstr);
7312

7313 7314 7315 7316 7317
    return 0;

error:
    VIR_FREE(devstr);
    return -1;
7318 7319
}

7320

7321
static int qemudDomainAttachHostDevice(struct qemud_driver *driver,
M
Mark McLoughlin 已提交
7322
                                       virDomainObjPtr vm,
7323 7324
                                       virDomainHostdevDefPtr hostdev,
                                       int qemuCmdFlags)
M
Mark McLoughlin 已提交
7325 7326
{
    if (hostdev->mode != VIR_DOMAIN_HOSTDEV_MODE_SUBSYS) {
7327 7328 7329
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        _("hostdev mode '%s' not supported"),
                        virDomainHostdevModeTypeToString(hostdev->mode));
M
Mark McLoughlin 已提交
7330 7331 7332
        return -1;
    }

7333 7334 7335 7336 7337 7338 7339 7340 7341 7342 7343 7344 7345 7346 7347 7348 7349
    /* 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);
    }


7350
    if (driver->securityDriver &&
7351
        driver->securityDriver->domainSetSecurityHostdevLabel &&
7352
        driver->securityDriver->domainSetSecurityHostdevLabel(vm, hostdev) < 0)
7353
        return -1;
M
Mark McLoughlin 已提交
7354 7355

    switch (hostdev->source.subsys.type) {
7356
    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI:
7357
        if (qemudDomainAttachHostPciDevice(driver, vm,
7358
                                           hostdev, qemuCmdFlags) < 0)
7359 7360 7361
            goto error;
        break;

M
Mark McLoughlin 已提交
7362
    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB:
7363
        if (qemudDomainAttachHostUsbDevice(driver, vm,
7364
                                           hostdev, qemuCmdFlags) < 0)
7365 7366 7367
            goto error;
        break;

M
Mark McLoughlin 已提交
7368
    default:
7369 7370 7371
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        _("hostdev subsys type '%s' not supported"),
                        virDomainHostdevSubsysTypeToString(hostdev->source.subsys.type));
7372
        goto error;
M
Mark McLoughlin 已提交
7373
    }
7374 7375 7376 7377 7378 7379

    return 0;

error:
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityHostdevLabel &&
7380
        driver->securityDriver->domainRestoreSecurityHostdevLabel(vm, hostdev) < 0)
7381 7382 7383
        VIR_WARN0("Unable to restore host device labelling on hotplug fail");

    return -1;
M
Mark McLoughlin 已提交
7384 7385
}

7386

7387
static int qemudDomainAttachDevice(virDomainPtr dom,
7388 7389
                                   const char *xml)
{
7390 7391 7392
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    virDomainDeviceDefPtr dev = NULL;
7393
    unsigned long long qemuCmdFlags;
7394
    virCgroupPtr cgroup = NULL;
7395
    int ret = -1;
7396

7397
    qemuDriverLock(driver);
7398
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
7399
    if (!vm) {
7400 7401
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
7402 7403
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
7404
        goto cleanup;
7405 7406
    }

7407 7408 7409
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
7410
    if (!virDomainObjIsActive(vm)) {
7411 7412
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot attach device on inactive domain"));
7413
        goto endjob;
7414 7415
    }

7416
    dev = virDomainDeviceDefParse(driver->caps, vm->def, xml,
7417
                                  VIR_DOMAIN_XML_INACTIVE);
7418
    if (dev == NULL)
7419
        goto endjob;
7420

7421 7422 7423
    if (qemudExtractVersionInfo(vm->def->emulator,
                                NULL,
                                &qemuCmdFlags) < 0)
7424
        goto endjob;
7425

7426
    if (dev->type == VIR_DOMAIN_DEVICE_DISK) {
7427
        if (qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_DEVICES)) {
7428
            if (virCgroupForDomain(driver->cgroup, vm->def->name, &cgroup, 0) !=0 ) {
7429 7430 7431
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("Unable to find cgroup for %s\n"),
                                vm->def->name);
7432
                goto endjob;
7433 7434 7435 7436 7437
            }
            if (dev->data.disk->src != NULL &&
                dev->data.disk->type == VIR_DOMAIN_DISK_TYPE_BLOCK &&
                virCgroupAllowDevicePath(cgroup,
                                         dev->data.disk->src) < 0) {
7438 7439 7440
                qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                _("unable to allow device %s"),
                                dev->data.disk->src);
7441
                goto endjob;
7442 7443 7444
            }
        }

7445
        switch (dev->data.disk->device) {
7446 7447
        case VIR_DOMAIN_DISK_DEVICE_CDROM:
        case VIR_DOMAIN_DISK_DEVICE_FLOPPY:
7448
            ret = qemudDomainChangeEjectableMedia(driver, vm, dev->data.disk);
7449 7450
            if (ret == 0)
                dev->data.disk = NULL;
7451
            break;
7452

7453 7454
        case VIR_DOMAIN_DISK_DEVICE_DISK:
            if (dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_USB) {
7455
                ret = qemudDomainAttachUsbMassstorageDevice(driver, vm,
7456
                                                            dev->data.disk, qemuCmdFlags);
7457 7458
                if (ret == 0)
                    dev->data.disk = NULL;
7459
            } else if (dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_VIRTIO) {
7460
                ret = qemudDomainAttachPciDiskDevice(driver, vm,
7461
                                                     dev->data.disk, qemuCmdFlags);
7462 7463
                if (ret == 0)
                    dev->data.disk = NULL;
7464
            } else if (dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_SCSI) {
7465
                ret = qemudDomainAttachSCSIDisk(driver, vm,
7466
                                                dev->data.disk, qemuCmdFlags);
7467 7468
                if (ret == 0)
                    dev->data.disk = NULL;
7469
            } else {
7470 7471 7472
                qemuReportError(VIR_ERR_NO_SUPPORT,
                                _("disk bus '%s' cannot be hotplugged."),
                                virDomainDiskBusTypeToString(dev->data.disk->bus));
7473
                /* fallthrough */
7474 7475
            }
            break;
7476

7477
        default:
7478 7479 7480
            qemuReportError(VIR_ERR_NO_SUPPORT,
                            _("disk device type '%s' cannot be hotplugged"),
                            virDomainDiskDeviceTypeToString(dev->data.disk->device));
7481 7482
            /* Fallthrough */
        }
7483
        if (ret != 0 && cgroup) {
7484 7485
            virCgroupDenyDevicePath(cgroup,
                                    dev->data.disk->src);
7486
        }
7487 7488
    } else if (dev->type == VIR_DOMAIN_DEVICE_CONTROLLER) {
        if (dev->data.controller->type == VIR_DOMAIN_CONTROLLER_TYPE_SCSI) {
7489
            ret = qemudDomainAttachPciControllerDevice(driver, vm,
7490
                                                       dev->data.controller, qemuCmdFlags);
7491 7492
            if (ret == 0)
                dev->data.controller = NULL;
7493
        } else {
7494 7495 7496
            qemuReportError(VIR_ERR_NO_SUPPORT,
                            _("disk controller bus '%s' cannot be hotplugged."),
                            virDomainControllerTypeToString(dev->data.controller->type));
7497 7498
            /* fallthrough */
        }
M
Mark McLoughlin 已提交
7499
    } else if (dev->type == VIR_DOMAIN_DEVICE_NET) {
7500 7501
        ret = qemudDomainAttachNetDevice(dom->conn, driver, vm,
                                         dev->data.net, qemuCmdFlags);
7502 7503
        if (ret == 0)
            dev->data.net = NULL;
M
Mark McLoughlin 已提交
7504
    } else if (dev->type == VIR_DOMAIN_DEVICE_HOSTDEV) {
7505
        ret = qemudDomainAttachHostDevice(driver, vm,
7506
                                          dev->data.hostdev, qemuCmdFlags);
7507 7508
        if (ret == 0)
            dev->data.hostdev = NULL;
7509
    } else {
7510 7511 7512
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        _("device type '%s' cannot be attached"),
                        virDomainDeviceTypeToString(dev->type));
7513
        goto endjob;
7514 7515
    }

7516
    if (!ret && virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
7517 7518
        ret = -1;

7519
endjob:
7520 7521
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
7522

7523
cleanup:
7524 7525 7526
    if (cgroup)
        virCgroupFree(&cgroup);

7527
    virDomainDeviceDefFree(dev);
7528 7529
    if (vm)
        virDomainObjUnlock(vm);
7530
    qemuDriverUnlock(driver);
7531 7532 7533
    return ret;
}

7534 7535 7536 7537
static int qemudDomainAttachDeviceFlags(virDomainPtr dom,
                                        const char *xml,
                                        unsigned int flags) {
    if (flags & VIR_DOMAIN_DEVICE_MODIFY_CONFIG) {
7538 7539
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot modify the persistent configuration of a domain"));
7540 7541 7542 7543 7544 7545
        return -1;
    }

    return qemudDomainAttachDevice(dom, xml);
}

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 7622 7623
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;
}


7624 7625 7626 7627 7628 7629 7630 7631 7632 7633 7634 7635 7636 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 7695 7696 7697 7698 7699 7700 7701 7702 7703 7704 7705 7706 7707 7708 7709 7710 7711
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;

    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:
            ret = qemudDomainChangeEjectableMedia(driver, vm, dev->data.disk);
            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;

7712 7713 7714 7715
    case VIR_DOMAIN_DEVICE_GRAPHICS:
        ret = qemuDomainChangeGraphics(driver, vm, dev->data.graphics);
        break;

7716 7717 7718 7719 7720 7721 7722 7723 7724 7725 7726 7727 7728 7729 7730 7731 7732 7733 7734 7735 7736 7737 7738 7739 7740 7741
    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;
}


7742
static int qemudDomainDetachPciDiskDevice(struct qemud_driver *driver,
7743
                                          virDomainObjPtr vm,
7744 7745
                                          virDomainDeviceDefPtr dev,
                                          unsigned long long qemuCmdFlags)
7746 7747 7748
{
    int i, ret = -1;
    virDomainDiskDefPtr detach = NULL;
7749
    qemuDomainObjPrivatePtr priv = vm->privateData;
7750 7751 7752 7753 7754 7755 7756 7757 7758

    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) {
7759 7760
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("disk %s not found"), dev->data.disk->dst);
7761
        goto cleanup;
7762 7763
    }

7764 7765
    if (!virDomainDeviceAddressIsValid(&detach->info,
                                       VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI)) {
7766 7767
        qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
                        _("device cannot be detached without a PCI address"));
7768
        goto cleanup;
7769 7770
    }

7771
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7772 7773 7774 7775 7776 7777 7778 7779 7780 7781 7782
    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;
        }
7783
    }
7784
    qemuDomainObjExitMonitorWithDriver(driver, vm);
7785

7786 7787 7788 7789 7790 7791 7792 7793 7794 7795
    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 {
7796
        VIR_FREE(vm->def->disks);
7797
        vm->def->ndisks = 0;
7798
    }
7799
    virDomainDiskDefFree(detach);
7800

7801 7802
    if (driver->securityDriver &&
        driver->securityDriver->domainRestoreSecurityImageLabel &&
7803
        driver->securityDriver->domainRestoreSecurityImageLabel(vm, dev->data.disk) < 0)
7804 7805
        VIR_WARN("Unable to restore security label on %s", dev->data.disk->src);

7806
    ret = 0;
7807 7808

cleanup:
7809 7810 7811
    return ret;
}

7812
static int qemudDomainDetachPciControllerDevice(struct qemud_driver *driver,
7813
                                                virDomainObjPtr vm,
7814 7815
                                                virDomainDeviceDefPtr dev,
                                                unsigned long long qemuCmdFlags)
7816 7817 7818 7819 7820 7821 7822 7823 7824 7825 7826 7827 7828 7829
{
    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) {
7830 7831 7832 7833
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("disk controller %s:%d not found"),
                        virDomainControllerTypeToString(dev->data.controller->type),
                        dev->data.controller->idx);
7834 7835 7836 7837 7838
        goto cleanup;
    }

    if (!virDomainDeviceAddressIsValid(&detach->info,
                                       VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI)) {
7839 7840
        qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
                        _("device cannot be detached without a PCI address"));
7841 7842 7843
        goto cleanup;
    }

7844 7845 7846 7847 7848
    if (qemuCmdFlags & QEMUD_CMD_FLAG_DEVICE) {
        if (qemuAssignDeviceControllerAlias(detach) < 0)
            goto cleanup;
    }

7849
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7850 7851 7852 7853 7854 7855 7856 7857 7858 7859 7860
    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;
        }
7861 7862 7863 7864 7865 7866 7867 7868 7869 7870 7871 7872 7873 7874 7875 7876
    }
    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;
    }
7877 7878 7879 7880 7881

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

7882 7883 7884 7885 7886 7887 7888 7889
    virDomainControllerDefFree(detach);

    ret = 0;

cleanup:
    return ret;
}

7890
static int
7891
qemudDomainDetachNetDevice(struct qemud_driver *driver,
7892
                           virDomainObjPtr vm,
7893 7894
                           virDomainDeviceDefPtr dev,
                           unsigned long long qemuCmdFlags)
7895 7896 7897
{
    int i, ret = -1;
    virDomainNetDefPtr detach = NULL;
7898
    qemuDomainObjPrivatePtr priv = vm->privateData;
7899 7900
    int vlan;
    char *hostnet_name = NULL;
7901 7902 7903 7904 7905 7906 7907 7908 7909 7910 7911

    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) {
7912 7913 7914 7915 7916
        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]);
7917 7918 7919
        goto cleanup;
    }

7920 7921
    if (!virDomainDeviceAddressIsValid(&detach->info,
                                       VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI)) {
7922 7923
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("device cannot be detached without a PCI address"));
7924 7925 7926
        goto cleanup;
    }

7927
    if ((vlan = qemuDomainNetVLAN(detach)) < 0) {
7928 7929
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("unable to determine original VLAN"));
7930 7931 7932 7933
        goto cleanup;
    }

    if (virAsprintf(&hostnet_name, "host%s", detach->info.alias) < 0) {
7934
        virReportOOMError();
7935 7936 7937
        goto cleanup;
    }

7938
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
7939 7940 7941 7942 7943 7944 7945 7946 7947 7948 7949
    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;
        }
7950
    }
7951

7952
    if (qemuMonitorRemoveHostNetwork(priv->mon, vlan, hostnet_name) < 0) {
7953
        qemuDomainObjExitMonitorWithDriver(driver, vm);
7954
        goto cleanup;
7955
    }
7956
    qemuDomainObjExitMonitorWithDriver(driver, vm);
7957

7958 7959
#if WITH_MACVTAP
    if (detach->type == VIR_DOMAIN_NET_TYPE_DIRECT) {
S
Stefan Berger 已提交
7960 7961
        if (detach->ifname)
            delMacvtap(detach->ifname);
7962 7963 7964
    }
#endif

7965
    if ((driver->macFilter) && (detach->ifname != NULL)) {
7966
        if ((errno = networkDisallowMacOnPort(driver,
7967 7968
                                              detach->ifname,
                                              detach->mac))) {
7969
            virReportSystemError(errno,
7970 7971 7972 7973 7974
             _("failed to remove ebtables rule on  '%s'"),
                                 detach->ifname);
        }
    }

S
Stefan Berger 已提交
7975 7976
    qemuTearNWFilter(detach);

7977 7978 7979 7980 7981 7982 7983 7984 7985 7986
    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 {
7987
        VIR_FREE(vm->def->nets);
7988
        vm->def->nnets = 0;
7989
    }
7990
    virDomainNetDefFree(detach);
7991

7992 7993 7994
    ret = 0;

cleanup:
7995
    VIR_FREE(hostnet_name);
7996 7997 7998
    return ret;
}

7999
static int qemudDomainDetachHostPciDevice(struct qemud_driver *driver,
8000
                                          virDomainObjPtr vm,
8001 8002
                                          virDomainDeviceDefPtr dev,
                                          unsigned long long qemuCmdFlags)
8003
{
8004
    virDomainHostdevDefPtr detach = NULL;
8005
    qemuDomainObjPrivatePtr priv = vm->privateData;
8006
    int i, ret;
8007
    pciDevice *pci;
8008 8009

    for (i = 0 ; i < vm->def->nhostdevs ; i++) {
8010 8011 8012 8013
        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;

8014 8015 8016 8017 8018 8019 8020 8021 8022 8023 8024 8025 8026 8027 8028
        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) {
8029 8030 8031 8032 8033 8034
        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);
8035 8036 8037
        return -1;
    }

8038 8039
    if (!virDomainDeviceAddressIsValid(&detach->info,
                                       VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI)) {
8040 8041
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("device cannot be detached without a PCI address"));
8042 8043 8044
        return -1;
    }

8045
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
8046 8047 8048 8049 8050 8051 8052 8053 8054 8055 8056
    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;
        }
8057
    }
8058
    qemuDomainObjExitMonitorWithDriver(driver, vm);
8059 8060 8061

    ret = 0;

8062
    pci = pciGetDevice(detach->source.subsys.u.pci.domain,
8063 8064 8065 8066 8067 8068
                       detach->source.subsys.u.pci.bus,
                       detach->source.subsys.u.pci.slot,
                       detach->source.subsys.u.pci.function);
    if (!pci)
        ret = -1;
    else {
8069
        pciDeviceSetManaged(pci, detach->managed);
8070 8071
        pciDeviceListDel(driver->activePciHostdevs, pci);
        if (pciResetDevice(pci, driver->activePciHostdevs) < 0)
8072
            ret = -1;
8073
        qemudReattachManagedDevice(pci);
8074
        pciFreeDevice(pci);
8075 8076
    }

8077 8078 8079 8080 8081 8082 8083 8084 8085 8086 8087 8088
    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;
8089
    }
8090
    virDomainHostdevDefFree(detach);
8091 8092 8093 8094

    return ret;
}

8095 8096 8097 8098 8099 8100 8101 8102 8103 8104 8105 8106 8107 8108 8109 8110
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;
8111 8112 8113 8114 8115 8116 8117 8118 8119 8120 8121 8122 8123 8124 8125 8126
        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;
            }
8127 8128 8129 8130 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
        }
    }

    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;
}

8177
static int qemudDomainDetachHostDevice(struct qemud_driver *driver,
8178
                                       virDomainObjPtr vm,
8179 8180
                                       virDomainDeviceDefPtr dev,
                                       unsigned long long qemuCmdFlags)
8181 8182 8183 8184 8185
{
    virDomainHostdevDefPtr hostdev = dev->data.hostdev;
    int ret;

    if (hostdev->mode != VIR_DOMAIN_HOSTDEV_MODE_SUBSYS) {
8186 8187 8188
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        _("hostdev mode '%s' not supported"),
                        virDomainHostdevModeTypeToString(hostdev->mode));
8189 8190 8191 8192 8193
        return -1;
    }

    switch (hostdev->source.subsys.type) {
    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI:
8194
        ret = qemudDomainDetachHostPciDevice(driver, vm, dev, qemuCmdFlags);
8195
        break;
8196 8197 8198
    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB:
        ret = qemudDomainDetachHostUsbDevice(driver, vm, dev, qemuCmdFlags);
        break;
8199
    default:
8200 8201 8202
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        _("hostdev subsys type '%s' not supported"),
                        virDomainHostdevSubsysTypeToString(hostdev->source.subsys.type));
8203 8204 8205
        return -1;
    }

8206
    if (driver->securityDriver &&
8207
        driver->securityDriver->domainRestoreSecurityHostdevLabel &&
8208
        driver->securityDriver->domainRestoreSecurityHostdevLabel(vm, dev->data.hostdev) < 0)
8209
        VIR_WARN0("Failed to restore host device labelling");
8210

8211 8212 8213
    return ret;
}

8214 8215
static int qemudDomainDetachDevice(virDomainPtr dom,
                                   const char *xml) {
8216 8217
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
8218
    unsigned long long qemuCmdFlags;
8219 8220
    virDomainDeviceDefPtr dev = NULL;
    int ret = -1;
8221

8222
    qemuDriverLock(driver);
8223
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
8224
    if (!vm) {
8225 8226
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
8227 8228
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
8229
        goto cleanup;
8230 8231
    }

8232 8233 8234
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
8235
    if (!virDomainObjIsActive(vm)) {
8236 8237
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot detach device on inactive domain"));
8238
        goto endjob;
8239 8240
    }

8241
    dev = virDomainDeviceDefParse(driver->caps, vm->def, xml,
8242
                                  VIR_DOMAIN_XML_INACTIVE);
8243
    if (dev == NULL)
8244
        goto endjob;
8245

8246 8247 8248 8249
    if (qemudExtractVersionInfo(vm->def->emulator,
                                NULL,
                                &qemuCmdFlags) < 0)
        goto endjob;
8250 8251 8252

    if (dev->type == VIR_DOMAIN_DEVICE_DISK &&
        dev->data.disk->device == VIR_DOMAIN_DISK_DEVICE_DISK &&
8253
        dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_VIRTIO) {
8254
        ret = qemudDomainDetachPciDiskDevice(driver, vm, dev, qemuCmdFlags);
8255
    } else if (dev->type == VIR_DOMAIN_DEVICE_NET) {
8256
        ret = qemudDomainDetachNetDevice(driver, vm, dev, qemuCmdFlags);
8257 8258
    } else if (dev->type == VIR_DOMAIN_DEVICE_CONTROLLER) {
        if (dev->data.controller->type == VIR_DOMAIN_CONTROLLER_TYPE_SCSI) {
8259 8260
            ret = qemudDomainDetachPciControllerDevice(driver, vm, dev,
                                                       qemuCmdFlags);
8261
        } else {
8262 8263 8264
            qemuReportError(VIR_ERR_NO_SUPPORT,
                            _("disk controller bus '%s' cannot be hotunplugged."),
                            virDomainControllerTypeToString(dev->data.controller->type));
8265 8266
            /* fallthrough */
        }
8267
    } else if (dev->type == VIR_DOMAIN_DEVICE_HOSTDEV) {
8268
        ret = qemudDomainDetachHostDevice(driver, vm, dev, qemuCmdFlags);
8269
    } else {
8270 8271
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        "%s", _("This type of device cannot be hot unplugged"));
8272
    }
8273

8274
    if (!ret && virDomainSaveStatus(driver->caps, driver->stateDir, vm) < 0)
8275 8276
        ret = -1;

8277
endjob:
8278 8279
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
8280

8281 8282
cleanup:
    virDomainDeviceDefFree(dev);
8283 8284
    if (vm)
        virDomainObjUnlock(vm);
8285
    qemuDriverUnlock(driver);
8286 8287 8288
    return ret;
}

8289 8290 8291 8292
static int qemudDomainDetachDeviceFlags(virDomainPtr dom,
                                        const char *xml,
                                        unsigned int flags) {
    if (flags & VIR_DOMAIN_DEVICE_MODIFY_CONFIG) {
8293 8294
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("cannot modify the persistent configuration of a domain"));
8295 8296 8297 8298 8299 8300
        return -1;
    }

    return qemudDomainDetachDevice(dom, xml);
}

8301
static int qemudDomainGetAutostart(virDomainPtr dom,
8302
                                   int *autostart) {
8303 8304 8305
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
8306

8307
    qemuDriverLock(driver);
8308
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
8309 8310
    qemuDriverUnlock(driver);

8311
    if (!vm) {
8312 8313
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
8314 8315
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
8316
        goto cleanup;
8317 8318 8319
    }

    *autostart = vm->autostart;
8320
    ret = 0;
8321

8322
cleanup:
8323 8324
    if (vm)
        virDomainObjUnlock(vm);
8325
    return ret;
8326 8327
}

8328
static int qemudDomainSetAutostart(virDomainPtr dom,
8329
                                   int autostart) {
8330 8331
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
8332 8333
    char *configFile = NULL, *autostartLink = NULL;
    int ret = -1;
8334

8335
    qemuDriverLock(driver);
8336
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
8337

8338
    if (!vm) {
8339 8340
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
8341 8342
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
8343
        goto cleanup;
8344 8345
    }

8346
    if (!vm->persistent) {
8347 8348
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("cannot set autostart for transient domain"));
8349
        goto cleanup;
8350 8351
    }

8352 8353
    autostart = (autostart != 0);

8354
    if (vm->autostart != autostart) {
8355
        if ((configFile = virDomainConfigFile(driver->configDir, vm->def->name)) == NULL)
8356
            goto cleanup;
8357
        if ((autostartLink = virDomainConfigFile(driver->autostartDir, vm->def->name)) == NULL)
8358
            goto cleanup;
8359

8360 8361
        if (autostart) {
            int err;
8362

8363
            if ((err = virFileMakePath(driver->autostartDir))) {
8364
                virReportSystemError(err,
8365 8366
                                     _("cannot create autostart directory %s"),
                                     driver->autostartDir);
8367 8368
                goto cleanup;
            }
8369

8370
            if (symlink(configFile, autostartLink) < 0) {
8371
                virReportSystemError(errno,
8372 8373
                                     _("Failed to create symlink '%s to '%s'"),
                                     autostartLink, configFile);
8374 8375 8376 8377
                goto cleanup;
            }
        } else {
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
8378
                virReportSystemError(errno,
8379 8380
                                     _("Failed to delete symlink '%s'"),
                                     autostartLink);
8381 8382
                goto cleanup;
            }
8383 8384
        }

8385
        vm->autostart = autostart;
8386
    }
8387
    ret = 0;
8388

8389 8390 8391
cleanup:
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
8392 8393
    if (vm)
        virDomainObjUnlock(vm);
8394
    qemuDriverUnlock(driver);
8395
    return ret;
8396 8397
}

8398 8399 8400 8401 8402

static char *qemuGetSchedulerType(virDomainPtr dom,
                                  int *nparams)
{
    struct qemud_driver *driver = dom->conn->privateData;
8403
    char *ret = NULL;
8404

8405
    qemuDriverLock(driver);
8406
    if (!qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_CPU)) {
8407 8408
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        __FUNCTION__);
8409
        goto cleanup;
8410 8411 8412 8413 8414 8415 8416
    }

    if (nparams)
        *nparams = 1;

    ret = strdup("posix");
    if (!ret)
8417
        virReportOOMError();
8418 8419 8420

cleanup:
    qemuDriverUnlock(driver);
8421 8422 8423 8424 8425 8426 8427 8428 8429 8430 8431 8432 8433
    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;

8434
    qemuDriverLock(driver);
8435
    if (!qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_CPU)) {
8436 8437
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        __FUNCTION__);
8438
        goto cleanup;
8439 8440 8441 8442 8443
    }

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

    if (vm == NULL) {
8444 8445
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("No such domain %s"), dom->uuid);
8446 8447 8448 8449
        goto cleanup;
    }

    if (virCgroupForDomain(driver->cgroup, vm->def->name, &group, 0) != 0) {
8450 8451
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("cannot find cgroup for domain %s"), vm->def->name);
8452 8453 8454 8455 8456 8457 8458 8459 8460
        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) {
8461 8462
                qemuReportError(VIR_ERR_INVALID_ARG, "%s",
                                _("invalid type for cpu_shares tunable, expected a 'ullong'"));
8463 8464 8465 8466 8467
                goto cleanup;
            }

            rc = virCgroupSetCpuShares(group, params[i].value.ul);
            if (rc != 0) {
8468
                virReportSystemError(-rc, "%s",
8469 8470 8471 8472
                                     _("unable to set cpu shares tunable"));
                goto cleanup;
            }
        } else {
8473 8474
            qemuReportError(VIR_ERR_INVALID_ARG,
                            _("Invalid parameter `%s'"), param->field);
8475 8476 8477 8478 8479 8480 8481 8482 8483
            goto cleanup;
        }
    }
    ret = 0;

cleanup:
    virCgroupFree(&group);
    if (vm)
        virDomainObjUnlock(vm);
8484
    qemuDriverUnlock(driver);
8485 8486 8487 8488 8489 8490 8491 8492 8493 8494 8495 8496 8497 8498
    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;

8499
    qemuDriverLock(driver);
8500
    if (!qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_CPU)) {
8501 8502
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        __FUNCTION__);
8503
        goto cleanup;
8504 8505 8506
    }

    if ((*nparams) != 1) {
8507 8508
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("Invalid parameter count"));
8509
        goto cleanup;
8510 8511 8512 8513 8514
    }

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

    if (vm == NULL) {
8515 8516
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("No such domain %s"), dom->uuid);
8517 8518 8519 8520
        goto cleanup;
    }

    if (virCgroupForDomain(driver->cgroup, vm->def->name, &group, 0) != 0) {
8521 8522
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("cannot find cgroup for domain %s"), vm->def->name);
8523 8524 8525 8526 8527
        goto cleanup;
    }

    rc = virCgroupGetCpuShares(group, &val);
    if (rc != 0) {
8528
        virReportSystemError(-rc, "%s",
8529 8530 8531 8532 8533
                             _("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
8534
    if (virStrcpyStatic(params[0].field, "cpu_shares") == NULL) {
8535 8536
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("Field cpu_shares too long for destination"));
C
Chris Lalancette 已提交
8537 8538
        goto cleanup;
    }
8539 8540 8541 8542 8543 8544 8545

    ret = 0;

cleanup:
    virCgroupFree(&group);
    if (vm)
        virDomainObjUnlock(vm);
8546
    qemuDriverUnlock(driver);
8547 8548 8549 8550
    return ret;
}


8551 8552 8553 8554 8555 8556 8557 8558 8559
/* 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)
{
8560
    struct qemud_driver *driver = dom->conn->privateData;
8561
    int i, ret = -1;
8562
    virDomainObjPtr vm;
8563
    virDomainDiskDefPtr disk = NULL;
8564

8565
    qemuDriverLock(driver);
8566
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
8567
    qemuDriverUnlock(driver);
8568
    if (!vm) {
8569 8570
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
8571 8572
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
8573
        goto cleanup;
8574
    }
8575 8576 8577 8578

    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
8579
    if (!virDomainObjIsActive (vm)) {
8580 8581
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
8582
        goto endjob;
8583 8584
    }

8585 8586 8587 8588 8589 8590 8591 8592
    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (STREQ(path, vm->def->disks[i]->dst)) {
            disk = vm->def->disks[i];
            break;
        }
    }

    if (!disk) {
8593 8594
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("invalid path: %s"), path);
8595
        goto endjob;
8596 8597
    }

8598
    if (!disk->info.alias) {
8599 8600
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("missing disk device alias name for %s"), disk->dst);
8601
        goto endjob;
8602
    }
8603

8604
    qemuDomainObjPrivatePtr priv = vm->privateData;
8605 8606
    qemuDomainObjEnterMonitor(vm);
    ret = qemuMonitorGetBlockStatsInfo(priv->mon,
8607
                                       disk->info.alias,
8608 8609 8610 8611 8612 8613
                                       &stats->rd_req,
                                       &stats->rd_bytes,
                                       &stats->wr_req,
                                       &stats->wr_bytes,
                                       &stats->errs);
    qemuDomainObjExitMonitor(vm);
8614

8615
endjob:
8616 8617
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
8618

8619
cleanup:
8620 8621
    if (vm)
        virDomainObjUnlock(vm);
8622
    return ret;
8623 8624
}

8625
#ifdef __linux__
8626 8627 8628 8629 8630
static int
qemudDomainInterfaceStats (virDomainPtr dom,
                           const char *path,
                           struct _virDomainInterfaceStats *stats)
{
8631 8632
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
8633
    int i;
8634
    int ret = -1;
8635

8636
    qemuDriverLock(driver);
8637
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
8638 8639
    qemuDriverUnlock(driver);

8640
    if (!vm) {
8641 8642
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
8643 8644
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
8645
        goto cleanup;
8646 8647
    }

D
Daniel P. Berrange 已提交
8648
    if (!virDomainObjIsActive(vm)) {
8649 8650
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
8651
        goto cleanup;
8652 8653 8654
    }

    /* Check the path is one of the domain's network interfaces. */
8655 8656
    for (i = 0 ; i < vm->def->nnets ; i++) {
        if (vm->def->nets[i]->ifname &&
8657 8658 8659 8660
            STREQ (vm->def->nets[i]->ifname, path)) {
            ret = 0;
            break;
        }
8661 8662
    }

8663
    if (ret == 0)
8664
        ret = linuxDomainInterfaceStats(path, stats);
8665
    else
8666 8667
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("invalid path, '%s' is not a known interface"), path);
8668

8669
cleanup:
8670 8671
    if (vm)
        virDomainObjUnlock(vm);
8672 8673
    return ret;
}
8674
#else
8675 8676 8677 8678
static int
qemudDomainInterfaceStats (virDomainPtr dom,
                           const char *path ATTRIBUTE_UNUSED,
                           struct _virDomainInterfaceStats *stats ATTRIBUTE_UNUSED)
8679 8680
    qemuReportError(VIR_ERR_NO_SUPPORT,
                    "%s", __FUNCTION__);
8681 8682
    return -1;
}
8683
#endif
8684

8685 8686 8687 8688 8689 8690 8691 8692 8693 8694 8695 8696 8697 8698 8699 8700
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);
8701 8702
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
8703 8704 8705
        goto cleanup;
    }

8706 8707 8708
    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

8709 8710 8711 8712 8713 8714
    if (virDomainObjIsActive(vm)) {
        qemuDomainObjPrivatePtr priv = vm->privateData;
        qemuDomainObjEnterMonitor(vm);
        ret = qemuMonitorTextGetMemoryStats(priv->mon, stats, nr_stats);
        qemuDomainObjExitMonitor(vm);
    } else {
8715 8716
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
8717 8718
    }

8719 8720 8721
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;

8722 8723 8724 8725 8726 8727
cleanup:
    if (vm)
        virDomainObjUnlock(vm);
    return ret;
}

8728 8729 8730 8731 8732 8733 8734
static int
qemudDomainBlockPeek (virDomainPtr dom,
                      const char *path,
                      unsigned long long offset, size_t size,
                      void *buffer,
                      unsigned int flags ATTRIBUTE_UNUSED)
{
8735 8736 8737
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int fd = -1, ret = -1, i;
8738

8739
    qemuDriverLock(driver);
8740
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
8741 8742
    qemuDriverUnlock(driver);

8743
    if (!vm) {
8744 8745
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
8746 8747
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
8748
        goto cleanup;
8749 8750 8751
    }

    if (!path || path[0] == '\0') {
8752 8753
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("NULL or empty path"));
8754
        goto cleanup;
8755 8756 8757
    }

    /* Check the path belongs to this domain. */
8758 8759
    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (vm->def->disks[i]->src != NULL &&
8760 8761 8762 8763
            STREQ (vm->def->disks[i]->src, path)) {
            ret = 0;
            break;
        }
8764 8765
    }

8766 8767 8768 8769 8770
    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) {
8771 8772
            virReportSystemError(errno,
                                 _("%s: failed to open"), path);
8773 8774
            goto cleanup;
        }
8775

8776 8777 8778 8779 8780 8781
        /* 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) {
8782 8783
            virReportSystemError(errno,
                                 _("%s: failed to seek or read"), path);
8784 8785 8786 8787 8788
            goto cleanup;
        }

        ret = 0;
    } else {
8789 8790
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("invalid path"));
8791 8792
    }

8793 8794 8795
cleanup:
    if (fd >= 0)
        close (fd);
8796 8797
    if (vm)
        virDomainObjUnlock(vm);
8798 8799 8800
    return ret;
}

R
Richard W.M. Jones 已提交
8801 8802 8803 8804 8805 8806
static int
qemudDomainMemoryPeek (virDomainPtr dom,
                       unsigned long long offset, size_t size,
                       void *buffer,
                       unsigned int flags)
{
8807 8808
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
8809
    char *tmp = NULL;
R
Richard W.M. Jones 已提交
8810 8811
    int fd = -1, ret = -1;

8812
    qemuDriverLock(driver);
8813
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
8814
    qemuDriverUnlock(driver);
R
Richard W.M. Jones 已提交
8815 8816

    if (!vm) {
8817 8818
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
8819 8820
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
8821 8822 8823
        goto cleanup;
    }

8824
    if (flags != VIR_MEMORY_VIRTUAL && flags != VIR_MEMORY_PHYSICAL) {
8825 8826
        qemuReportError(VIR_ERR_INVALID_ARG,
                        "%s", _("flags parameter must be VIR_MEMORY_VIRTUAL or VIR_MEMORY_PHYSICAL"));
8827
        goto cleanup;
R
Richard W.M. Jones 已提交
8828 8829
    }

8830 8831 8832
    if (qemuDomainObjBeginJob(vm) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
8833
    if (!virDomainObjIsActive(vm)) {
8834 8835
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
8836
        goto endjob;
R
Richard W.M. Jones 已提交
8837 8838
    }

8839
    if (virAsprintf(&tmp, driver->cacheDir,  "/qemu.mem.XXXXXX") < 0) {
8840
        virReportOOMError();
8841
        goto endjob;
8842 8843
    }

R
Richard W.M. Jones 已提交
8844 8845
    /* Create a temporary filename. */
    if ((fd = mkstemp (tmp)) == -1) {
8846 8847
        virReportSystemError(errno,
                             _("mkstemp(\"%s\") failed"), tmp);
8848
        goto endjob;
R
Richard W.M. Jones 已提交
8849 8850
    }

8851
    qemuDomainObjPrivatePtr priv = vm->privateData;
8852
    qemuDomainObjEnterMonitor(vm);
8853
    if (flags == VIR_MEMORY_VIRTUAL) {
8854 8855
        if (qemuMonitorSaveVirtualMemory(priv->mon, offset, size, tmp) < 0) {
            qemuDomainObjExitMonitor(vm);
8856
            goto endjob;
8857
        }
8858
    } else {
8859 8860
        if (qemuMonitorSavePhysicalMemory(priv->mon, offset, size, tmp) < 0) {
            qemuDomainObjExitMonitor(vm);
8861
            goto endjob;
8862
        }
R
Richard W.M. Jones 已提交
8863
    }
8864
    qemuDomainObjExitMonitor(vm);
R
Richard W.M. Jones 已提交
8865 8866 8867

    /* Read the memory file into buffer. */
    if (saferead (fd, buffer, size) == (ssize_t) -1) {
8868 8869 8870
        virReportSystemError(errno,
                             _("failed to read temporary file "
                               "created with template %s"), tmp);
8871
        goto endjob;
R
Richard W.M. Jones 已提交
8872 8873 8874
    }

    ret = 0;
8875

8876
endjob:
8877 8878
    if (qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
8879

8880
cleanup:
8881
    VIR_FREE(tmp);
R
Richard W.M. Jones 已提交
8882 8883
    if (fd >= 0) close (fd);
    unlink (tmp);
8884 8885
    if (vm)
        virDomainObjUnlock(vm);
R
Richard W.M. Jones 已提交
8886 8887 8888
    return ret;
}

8889

8890
static int
8891 8892 8893 8894
qemuDomainEventRegister(virConnectPtr conn,
                        virConnectDomainEventCallback callback,
                        void *opaque,
                        virFreeCallback freecb)
8895
{
8896 8897 8898
    struct qemud_driver *driver = conn->privateData;
    int ret;

8899
    qemuDriverLock(driver);
8900 8901
    ret = virDomainEventCallbackListAdd(conn, driver->domainEventCallbacks,
                                        callback, opaque, freecb);
8902
    qemuDriverUnlock(driver);
8903

8904
    return ret;
8905 8906
}

8907

8908
static int
8909 8910
qemuDomainEventDeregister(virConnectPtr conn,
                          virConnectDomainEventCallback callback)
8911
{
8912 8913 8914
    struct qemud_driver *driver = conn->privateData;
    int ret;

8915
    qemuDriverLock(driver);
8916 8917 8918 8919 8920 8921
    if (driver->domainEventDispatching)
        ret = virDomainEventCallbackListMarkDelete(conn, driver->domainEventCallbacks,
                                                   callback);
    else
        ret = virDomainEventCallbackListRemove(conn, driver->domainEventCallbacks,
                                               callback);
8922
    qemuDriverUnlock(driver);
8923

8924
    return ret;
8925 8926
}

8927 8928 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

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;
}


8970 8971
static void qemuDomainEventDispatchFunc(virConnectPtr conn,
                                        virDomainEventPtr event,
8972
                                        virConnectDomainEventGenericCallback cb,
8973 8974
                                        void *cbopaque,
                                        void *opaque)
8975
{
8976
    struct qemud_driver *driver = opaque;
8977

8978 8979 8980 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
    /* 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);
9022 9023
}

D
Daniel Veillard 已提交
9024 9025
/* Migration support. */

C
Chris Lalancette 已提交
9026 9027 9028 9029 9030 9031 9032 9033 9034 9035 9036 9037 9038 9039 9040 9041 9042 9043 9044
/* 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) {
9045 9046
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream is not open"));
C
Chris Lalancette 已提交
9047 9048 9049 9050 9051
        return -1;
    }

    qemuDriverLock(driver);
    if (qemust->watch == 0) {
9052 9053
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream does not have a callback registered"));
C
Chris Lalancette 已提交
9054 9055 9056 9057 9058 9059 9060 9061 9062 9063 9064 9065 9066 9067 9068 9069 9070 9071 9072 9073 9074 9075 9076 9077 9078 9079 9080 9081
        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) {
9082 9083
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream is not open"));
C
Chris Lalancette 已提交
9084 9085 9086 9087 9088
        return -1;
    }

    qemuDriverLock(driver);
    if (qemust->watch == 0) {
9089 9090
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream does not have a callback registered"));
C
Chris Lalancette 已提交
9091 9092 9093 9094 9095 9096 9097 9098 9099 9100 9101 9102 9103 9104 9105 9106 9107 9108 9109 9110 9111 9112 9113 9114 9115 9116 9117 9118 9119 9120 9121 9122 9123 9124 9125 9126 9127 9128 9129 9130 9131 9132 9133 9134 9135 9136 9137 9138 9139 9140 9141 9142 9143 9144 9145 9146 9147
        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) {
9148 9149
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream is not open"));
C
Chris Lalancette 已提交
9150 9151 9152 9153 9154
        return -1;
    }

    qemuDriverLock(driver);
    if (qemust->watch != 0) {
9155 9156
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream already has a callback registered"));
C
Chris Lalancette 已提交
9157 9158 9159 9160 9161 9162 9163 9164
        goto cleanup;
    }

    if ((qemust->watch = virEventAddHandle(qemust->fd,
                                           events,
                                           qemuStreamMigEvent,
                                           st,
                                           NULL)) < 0) {
9165 9166
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("cannot register file watch on stream"));
C
Chris Lalancette 已提交
9167 9168 9169 9170 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
        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;

9199
    if (VIR_ALLOC(qemust) < 0) {
9200
        virReportOOMError();
C
Chris Lalancette 已提交
9201
        return NULL;
9202
    }
C
Chris Lalancette 已提交
9203 9204 9205 9206 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

    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) {
9264 9265
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("stream is not open"));
C
Chris Lalancette 已提交
9266 9267 9268 9269 9270 9271 9272 9273 9274 9275 9276 9277 9278 9279
        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;
9280
            virReportSystemError(errno, "%s",
C
Chris Lalancette 已提交
9281 9282 9283 9284 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
                                 _("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;
9319
    unsigned long long qemuCmdFlags;
C
Chris Lalancette 已提交
9320 9321 9322 9323
    struct qemuStreamMigFile *qemust = NULL;

    qemuDriverLock(driver);
    if (!dom_xml) {
9324 9325
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("no domain XML passed"));
C
Chris Lalancette 已提交
9326 9327 9328
        goto cleanup;
    }
    if (!(flags & VIR_MIGRATE_TUNNELLED)) {
9329
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
C
Chris Lalancette 已提交
9330 9331 9332 9333
                         "%s", _("PrepareTunnel called but no TUNNELLED flag set"));
        goto cleanup;
    }
    if (st == NULL) {
9334 9335
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("tunnelled migration requested but NULL stream passed"));
C
Chris Lalancette 已提交
9336 9337 9338 9339
        goto cleanup;
    }

    /* Parse the domain XML. */
9340
    if (!(def = virDomainDefParseString(driver->caps, dom_xml,
C
Chris Lalancette 已提交
9341
                                        VIR_DOMAIN_XML_INACTIVE))) {
9342 9343
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to parse XML"));
C
Chris Lalancette 已提交
9344 9345 9346 9347
        goto cleanup;
    }

    /* Target domain name, maybe renamed. */
9348 9349 9350 9351 9352 9353
    if (dname) {
        VIR_FREE(def->name);
        def->name = strdup(dname);
        if (def->name == NULL)
            goto cleanup;
    }
C
Chris Lalancette 已提交
9354

9355 9356
    if (virDomainObjIsDuplicate(&driver->domains, def, 1) < 0)
        goto cleanup;
C
Chris Lalancette 已提交
9357

9358
    if (!(vm = virDomainAssignDef(driver->caps,
C
Chris Lalancette 已提交
9359
                                  &driver->domains,
9360
                                  def, true))) {
9361 9362
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to assign new VM"));
C
Chris Lalancette 已提交
9363 9364 9365 9366
        goto cleanup;
    }
    def = NULL;

9367 9368 9369
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

C
Chris Lalancette 已提交
9370 9371 9372 9373 9374
    /* 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) {
9375
        virReportOOMError();
9376
        goto endjob;
C
Chris Lalancette 已提交
9377 9378 9379 9380 9381
    }
    unlink(unixfile);

    /* check that this qemu version supports the interactive exec */
    if (qemudExtractVersionInfo(vm->def->emulator, NULL, &qemuCmdFlags) < 0) {
9382 9383 9384
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Cannot determine QEMU argv syntax %s"),
                        vm->def->emulator);
9385
        goto endjob;
C
Chris Lalancette 已提交
9386 9387 9388 9389 9390 9391
    }
    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 {
9392 9393
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("Destination qemu is too old to support tunnelled migration"));
9394
        goto endjob;
C
Chris Lalancette 已提交
9395 9396
    }
    if (internalret < 0) {
9397
        virReportOOMError();
9398
        goto endjob;
C
Chris Lalancette 已提交
9399 9400 9401 9402 9403 9404 9405 9406 9407 9408 9409 9410 9411 9412
    }
    /* 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;
        }
9413
        goto endjob;
C
Chris Lalancette 已提交
9414 9415 9416 9417
    }

    qemust = qemuStreamMigOpen(st, unixfile);
    if (qemust == NULL) {
9418
        qemudShutdownVMDaemon(driver, vm);
9419
        if (!vm->persistent) {
9420 9421
            if (qemuDomainObjEndJob(vm) > 0)
                virDomainRemoveInactive(&driver->domains, vm);
9422 9423
            vm = NULL;
        }
9424
        virReportSystemError(errno,
C
Chris Lalancette 已提交
9425 9426
                             _("cannot open unix socket '%s' for tunnelled migration"),
                             unixfile);
9427
        goto endjob;
C
Chris Lalancette 已提交
9428 9429 9430 9431 9432 9433 9434 9435 9436 9437
    }

    st->driver = &qemuStreamMigDrv;
    st->privateData = qemust;

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_MIGRATED);
    ret = 0;

9438
endjob:
9439 9440 9441
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
9442

C
Chris Lalancette 已提交
9443 9444
cleanup:
    virDomainDefFree(def);
9445 9446
    if (unixfile)
        unlink(unixfile);
C
Chris Lalancette 已提交
9447 9448 9449 9450 9451 9452 9453 9454 9455
    VIR_FREE(unixfile);
    if (vm)
        virDomainObjUnlock(vm);
    if (event)
        qemuDomainEventQueue(driver, event);
    qemuDriverUnlock(driver);
    return ret;
}

D
Daniel Veillard 已提交
9456 9457 9458 9459
/* Prepare is the first step, and it runs on the destination host.
 *
 * This starts an empty VM listening on a TCP port.
 */
9460
static int ATTRIBUTE_NONNULL (5)
D
Daniel Veillard 已提交
9461 9462 9463 9464 9465
qemudDomainMigratePrepare2 (virConnectPtr dconn,
                            char **cookie ATTRIBUTE_UNUSED,
                            int *cookielen ATTRIBUTE_UNUSED,
                            const char *uri_in,
                            char **uri_out,
C
Chris Lalancette 已提交
9466
                            unsigned long flags,
D
Daniel Veillard 已提交
9467 9468 9469 9470 9471
                            const char *dname,
                            unsigned long resource ATTRIBUTE_UNUSED,
                            const char *dom_xml)
{
    static int port = 0;
9472 9473
    struct qemud_driver *driver = dconn->privateData;
    virDomainDefPtr def = NULL;
D
Daniel Veillard 已提交
9474 9475
    virDomainObjPtr vm = NULL;
    int this_port;
9476
    char *hostname;
D
Daniel Veillard 已提交
9477 9478
    char migrateFrom [64];
    const char *p;
9479
    virDomainEventPtr event = NULL;
9480
    int ret = -1;
9481
    int internalret;
9482 9483

    *uri_out = NULL;
D
Daniel Veillard 已提交
9484

9485
    qemuDriverLock(driver);
C
Chris Lalancette 已提交
9486 9487 9488 9489
    if (flags & VIR_MIGRATE_TUNNELLED) {
        /* this is a logical error; we never should have gotten here with
         * VIR_MIGRATE_TUNNELLED set
         */
9490 9491
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("Tunnelled migration requested but invalid RPC method called"));
C
Chris Lalancette 已提交
9492 9493 9494
        goto cleanup;
    }

D
Daniel Veillard 已提交
9495
    if (!dom_xml) {
9496 9497
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        "%s", _("no domain XML passed"));
9498
        goto cleanup;
D
Daniel Veillard 已提交
9499 9500 9501 9502 9503 9504 9505 9506 9507 9508 9509 9510 9511 9512 9513 9514 9515
    }

    /* 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 */
9516
        if ((hostname = virGetHostnameLocalhost(0)) == NULL)
9517
            goto cleanup;
D
Daniel Veillard 已提交
9518

9519 9520 9521 9522 9523
        /* 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
9524
        /* Caller frees */
9525 9526 9527
        internalret = virAsprintf(uri_out, "tcp:%s:%d", hostname, this_port);
        VIR_FREE(hostname);
        if (internalret < 0) {
9528
            virReportOOMError();
9529
            goto cleanup;
D
Daniel Veillard 已提交
9530 9531 9532 9533 9534 9535
        }
    } 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.
         */
9536
        if (!STRPREFIX (uri_in, "tcp:")) {
9537 9538
            qemuReportError (VIR_ERR_INVALID_ARG,
                             "%s", _("only tcp URIs are supported for KVM/QEMU migrations"));
9539
            goto cleanup;
D
Daniel Veillard 已提交
9540 9541 9542 9543
        }

        /* Get the port number. */
        p = strrchr (uri_in, ':');
9544 9545 9546 9547 9548 9549 9550 9551
        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) {
9552
                virReportOOMError();
9553 9554 9555 9556 9557 9558 9559
                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)) {
9560 9561
                qemuReportError(VIR_ERR_INVALID_ARG,
                                "%s", _("URI ended with incorrect ':port'"));
9562 9563
                goto cleanup;
            }
D
Daniel Veillard 已提交
9564 9565 9566
        }
    }

9567
    if (*uri_out)
9568 9569
        VIR_DEBUG("Generated uri_out=%s", *uri_out);

D
Daniel Veillard 已提交
9570
    /* Parse the domain XML. */
9571
    if (!(def = virDomainDefParseString(driver->caps, dom_xml,
9572
                                        VIR_DOMAIN_XML_INACTIVE))) {
9573 9574
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to parse XML"));
9575
        goto cleanup;
D
Daniel Veillard 已提交
9576 9577 9578
    }

    /* Target domain name, maybe renamed. */
9579 9580 9581 9582 9583 9584
    if (dname) {
        VIR_FREE(def->name);
        def->name = strdup(dname);
        if (def->name == NULL)
            goto cleanup;
    }
D
Daniel Veillard 已提交
9585

9586 9587
    if (virDomainObjIsDuplicate(&driver->domains, def, 1) < 0)
        goto cleanup;
D
Daniel Veillard 已提交
9588

9589
    if (!(vm = virDomainAssignDef(driver->caps,
D
Daniel Veillard 已提交
9590
                                  &driver->domains,
9591
                                  def, true))) {
9592 9593
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to assign new VM"));
9594
        goto cleanup;
D
Daniel Veillard 已提交
9595
    }
9596
    def = NULL;
D
Daniel Veillard 已提交
9597

9598 9599 9600
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

D
Daniel Veillard 已提交
9601 9602 9603 9604 9605 9606 9607
    /* 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);
9608
    if (qemudStartVMDaemon (dconn, driver, vm, migrateFrom, -1) < 0) {
9609 9610 9611
        /* Note that we don't set an error here because qemudStartVMDaemon
         * should have already done that.
         */
9612
        if (!vm->persistent) {
9613 9614
            if (qemuDomainObjEndJob(vm) > 0)
                virDomainRemoveInactive(&driver->domains, vm);
9615 9616
            vm = NULL;
        }
9617
        goto endjob;
D
Daniel Veillard 已提交
9618
    }
9619 9620 9621 9622

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_MIGRATED);
9623
    ret = 0;
D
Daniel Veillard 已提交
9624

9625
endjob:
9626 9627 9628
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
9629

9630 9631 9632 9633 9634
cleanup:
    virDomainDefFree(def);
    if (ret != 0) {
        VIR_FREE(*uri_out);
    }
9635 9636
    if (vm)
        virDomainObjUnlock(vm);
9637 9638
    if (event)
        qemuDomainEventQueue(driver, event);
9639
    qemuDriverUnlock(driver);
9640
    return ret;
C
Chris Lalancette 已提交
9641 9642 9643

}

9644 9645 9646 9647

/* Perform migration using QEMU's native TCP migrate support,
 * not encrypted obviously
 */
9648
static int doNativeMigrate(struct qemud_driver *driver,
9649 9650 9651 9652 9653 9654 9655
                           virDomainObjPtr vm,
                           const char *uri,
                           unsigned long flags ATTRIBUTE_UNUSED,
                           const char *dname ATTRIBUTE_UNUSED,
                           unsigned long resource)
{
    int ret = -1;
9656
    xmlURIPtr uribits = NULL;
9657
    qemuDomainObjPrivatePtr priv = vm->privateData;
9658 9659 9660 9661 9662 9663

    /* 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) {
9664
            virReportOOMError();
9665 9666 9667 9668 9669 9670 9671 9672
            goto cleanup;
        }
        uribits = xmlParseURI(tmpuri);
        VIR_FREE(tmpuri);
    } else {
        uribits = xmlParseURI(uri);
    }
    if (!uribits) {
9673 9674
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("cannot parse URI %s"), uri);
9675 9676 9677
        goto cleanup;
    }

9678
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
9679
    if (resource > 0 &&
9680
        qemuMonitorSetMigrationSpeed(priv->mon, resource) < 0) {
9681
        qemuDomainObjExitMonitorWithDriver(driver, vm);
9682
        goto cleanup;
9683
    }
9684

9685
    if (qemuMonitorMigrateToHost(priv->mon, 1, uribits->server, uribits->port) < 0) {
9686
        qemuDomainObjExitMonitorWithDriver(driver, vm);
9687 9688
        goto cleanup;
    }
9689
    qemuDomainObjExitMonitorWithDriver(driver, vm);
9690

9691
    if (qemuDomainWaitForMigrationComplete(driver, vm) < 0)
9692 9693 9694 9695 9696 9697 9698 9699 9700 9701
        goto cleanup;

    ret = 0;

cleanup:
    xmlFreeURI(uribits);
    return ret;
}


9702 9703
#define TUNNEL_SEND_BUF_SIZE 65536

9704
static int doTunnelSendAll(virStreamPtr st,
9705 9706
                           int sock)
{
9707 9708 9709 9710 9711 9712 9713 9714
    char *buffer;
    int nbytes = TUNNEL_SEND_BUF_SIZE;

    if (VIR_ALLOC_N(buffer, TUNNEL_SEND_BUF_SIZE) < 0) {
        virReportOOMError();
        virStreamAbort(st);
        return -1;
    }
9715 9716 9717 9718 9719

    /* XXX should honour the 'resource' parameter here */
    for (;;) {
        nbytes = saferead(sock, buffer, nbytes);
        if (nbytes < 0) {
9720
            virReportSystemError(errno, "%s",
9721
                                 _("tunnelled migration failed to read from qemu"));
9722 9723
            virStreamAbort(st);
            VIR_FREE(buffer);
9724 9725 9726 9727 9728 9729 9730
            return -1;
        }
        else if (nbytes == 0)
            /* EOF; get out of here */
            break;

        if (virStreamSend(st, buffer, nbytes) < 0) {
9731 9732
            qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
                            _("Failed to write migration data to remote libvirtd"));
9733
            VIR_FREE(buffer);
9734 9735 9736 9737
            return -1;
        }
    }

9738 9739
    VIR_FREE(buffer);

9740 9741 9742 9743 9744 9745 9746
    if (virStreamFinish(st) < 0)
        /* virStreamFinish set the error for us */
        return -1;

    return 0;
}

C
Chris Lalancette 已提交
9747
static int doTunnelMigrate(virDomainPtr dom,
9748
                           struct qemud_driver *driver,
9749
                           virConnectPtr dconn,
C
Chris Lalancette 已提交
9750
                           virDomainObjPtr vm,
9751
                           const char *dom_xml,
C
Chris Lalancette 已提交
9752 9753 9754 9755 9756
                           const char *uri,
                           unsigned long flags,
                           const char *dname,
                           unsigned long resource)
{
9757
    qemuDomainObjPrivatePtr priv = vm->privateData;
9758 9759
    int client_sock = -1;
    int qemu_sock = -1;
C
Chris Lalancette 已提交
9760 9761
    struct sockaddr_un sa_qemu, sa_client;
    socklen_t addrlen;
9762
    virDomainPtr ddomain = NULL;
C
Chris Lalancette 已提交
9763
    int retval = -1;
9764
    virStreamPtr st = NULL;
C
Chris Lalancette 已提交
9765 9766
    char *unixfile = NULL;
    int internalret;
9767
    unsigned long long qemuCmdFlags;
C
Chris Lalancette 已提交
9768 9769 9770
    int status;
    unsigned long long transferred, remaining, total;

9771 9772 9773 9774 9775 9776 9777 9778
    /*
     * 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
9779 9780
     */

9781

9782
    /* Stage 1. setup local support infrastructure */
C
Chris Lalancette 已提交
9783 9784 9785

    if (virAsprintf(&unixfile, "%s/qemu.tunnelmigrate.src.%s",
                    driver->stateDir, vm->def->name) < 0) {
9786
        virReportOOMError();
9787
        goto cleanup;
C
Chris Lalancette 已提交
9788 9789 9790 9791
    }

    qemu_sock = socket(AF_UNIX, SOCK_STREAM, 0);
    if (qemu_sock < 0) {
9792
        virReportSystemError(errno, "%s",
C
Chris Lalancette 已提交
9793
                             _("cannot open tunnelled migration socket"));
9794
        goto cleanup;
C
Chris Lalancette 已提交
9795 9796 9797 9798 9799
    }
    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) {
9800 9801 9802
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Unix socket '%s' too big for destination"),
                        unixfile);
9803
        goto cleanup;
C
Chris Lalancette 已提交
9804 9805 9806
    }
    unlink(unixfile);
    if (bind(qemu_sock, (struct sockaddr *)&sa_qemu, sizeof(sa_qemu)) < 0) {
9807
        virReportSystemError(errno,
C
Chris Lalancette 已提交
9808 9809
                             _("Cannot bind to unix socket '%s' for tunnelled migration"),
                             unixfile);
9810
        goto cleanup;
C
Chris Lalancette 已提交
9811 9812
    }
    if (listen(qemu_sock, 1) < 0) {
9813
        virReportSystemError(errno,
C
Chris Lalancette 已提交
9814 9815
                             _("Cannot listen on unix socket '%s' for tunnelled migration"),
                             unixfile);
9816
        goto cleanup;
C
Chris Lalancette 已提交
9817 9818 9819 9820
    }

    /* check that this qemu version supports the unix migration */
    if (qemudExtractVersionInfo(vm->def->emulator, NULL, &qemuCmdFlags) < 0) {
9821 9822 9823
        qemuReportError(VIR_ERR_INTERNAL_ERROR,
                        _("Cannot extract Qemu version from '%s'"),
                        vm->def->emulator);
9824 9825 9826 9827 9828
        goto cleanup;
    }

    if (!(qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_UNIX) &&
        !(qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_EXEC)) {
9829 9830
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("Source qemu is too old to support tunnelled migration"));
9831
        goto cleanup;
C
Chris Lalancette 已提交
9832
    }
9833 9834 9835 9836 9837 9838 9839 9840 9841 9842 9843 9844 9845 9846 9847 9848 9849 9850 9851 9852 9853


    /* 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 */
9854
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
C
Chris Lalancette 已提交
9855
    if (qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_UNIX)
9856
        internalret = qemuMonitorMigrateToUnix(priv->mon, 1, unixfile);
C
Chris Lalancette 已提交
9857 9858
    else if (qemuCmdFlags & QEMUD_CMD_FLAG_MIGRATE_QEMU_EXEC) {
        const char *args[] = { "nc", "-U", unixfile, NULL };
9859
        internalret = qemuMonitorMigrateToCommand(priv->mon, 1, args, "/dev/null");
9860 9861
    } else {
        internalret = -1;
C
Chris Lalancette 已提交
9862
    }
9863
    qemuDomainObjExitMonitorWithDriver(driver, vm);
C
Chris Lalancette 已提交
9864
    if (internalret < 0) {
9865 9866
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("tunnelled migration monitor command failed"));
9867
        goto finish;
C
Chris Lalancette 已提交
9868 9869
    }

9870 9871 9872
    /* From this point onwards we *must* call cancel to abort the
     * migration on source if anything goes wrong */

C
Chris Lalancette 已提交
9873 9874 9875
    /* it is also possible that the migrate didn't fail initially, but
     * rather failed later on.  Check the output of "info migrate"
     */
9876
    qemuDomainObjEnterMonitorWithDriver(driver, vm);
9877 9878
    if (qemuMonitorGetMigrationStatus(priv->mon,
                                      &status,
C
Chris Lalancette 已提交
9879 9880 9881
                                      &transferred,
                                      &remaining,
                                      &total) < 0) {
9882
        qemuDomainObjExitMonitorWithDriver(driver, vm);
9883
        goto cancel;
C
Chris Lalancette 已提交
9884
    }
9885
    qemuDomainObjExitMonitorWithDriver(driver, vm);
C
Chris Lalancette 已提交
9886 9887

    if (status == QEMU_MONITOR_MIGRATION_STATUS_ERROR) {
9888 9889
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s",_("migrate failed"));
9890
        goto cancel;
C
Chris Lalancette 已提交
9891 9892 9893 9894 9895 9896
    }

    addrlen = sizeof(sa_client);
    while ((client_sock = accept(qemu_sock, (struct sockaddr *)&sa_client, &addrlen)) < 0) {
        if (errno == EAGAIN || errno == EINTR)
            continue;
9897
        virReportSystemError(errno, "%s",
C
Chris Lalancette 已提交
9898
                             _("tunnelled migration failed to accept from qemu"));
9899
        goto cancel;
C
Chris Lalancette 已提交
9900 9901
    }

9902
    retval = doTunnelSendAll(st, client_sock);
9903

9904
cancel:
9905
    if (retval != 0) {
9906
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
9907
        qemuMonitorMigrateCancel(priv->mon);
9908
        qemuDomainObjExitMonitorWithDriver(driver, vm);
9909
    }
C
Chris Lalancette 已提交
9910

9911
finish:
C
Chris Lalancette 已提交
9912 9913 9914
    dname = dname ? dname : dom->name;
    ddomain = dconn->driver->domainMigrateFinish2
        (dconn, dname, NULL, 0, uri, flags, retval);
9915 9916 9917 9918 9919 9920 9921

cleanup:
    if (client_sock != -1)
        close(client_sock);
    if (qemu_sock != -1)
        close(qemu_sock);

C
Chris Lalancette 已提交
9922 9923 9924
    if (ddomain)
        virUnrefDomain(ddomain);

9925 9926 9927 9928
    if (unixfile) {
        unlink(unixfile);
        VIR_FREE(unixfile);
    }
C
Chris Lalancette 已提交
9929

9930 9931 9932
    if (st)
        /* don't call virStreamFree(), because that resets any pending errors */
        virUnrefStream(st);
9933 9934 9935 9936
    return retval;
}


9937 9938 9939 9940
/* 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,
9941
                              struct qemud_driver *driver,
9942 9943 9944 9945 9946 9947 9948 9949 9950 9951 9952 9953 9954 9955 9956 9957 9958 9959 9960 9961 9962 9963 9964 9965 9966
                              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) {
9967 9968
        qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                        _("domainMigratePrepare2 did not set uri"));
9969
        goto cleanup;
9970 9971
    }

9972
    if (doNativeMigrate(driver, vm, uri_out, flags, dname, resource) < 0)
9973 9974 9975 9976 9977 9978 9979 9980 9981 9982 9983 9984 9985 9986 9987 9988 9989
        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;
}


9990
static int doPeer2PeerMigrate(virDomainPtr dom,
9991
                              struct qemud_driver *driver,
9992 9993 9994 9995 9996 9997 9998 9999 10000 10001 10002 10003 10004 10005 10006 10007
                              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) {
10008 10009
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        _("Failed to connect to remote libvirt URI %s"), uri);
10010 10011 10012
        return -1;
    }
    if (!VIR_DRV_SUPPORTS_FEATURE(dconn->driver, dconn,
10013
                                  VIR_DRV_FEATURE_MIGRATION_P2P)) {
10014 10015
        qemuReportError(VIR_ERR_OPERATION_FAILED, "%s",
                        _("Destination libvirt does not support peer-to-peer migration protocol"));
10016 10017 10018
        goto cleanup;
    }

10019 10020 10021
    dom_xml = qemudVMDumpXML(driver, vm,
                             VIR_DOMAIN_XML_SECURE |
                             VIR_DOMAIN_XML_UPDATE_CPU);
10022
    if (!dom_xml) {
10023 10024
        qemuReportError(VIR_ERR_OPERATION_FAILED,
                        "%s", _("failed to get domain xml"));
10025 10026 10027
        goto cleanup;
    }

10028
    if (flags & VIR_MIGRATE_TUNNELLED)
10029
        ret = doTunnelMigrate(dom, driver, dconn, vm, dom_xml, uri, flags, dname, resource);
10030
    else
10031
        ret = doNonTunnelMigrate(dom, driver, dconn, vm, dom_xml, uri, flags, dname, resource);
10032 10033 10034

cleanup:
    VIR_FREE(dom_xml);
C
Chris Lalancette 已提交
10035 10036 10037
    /* don't call virConnectClose(), because that resets any pending errors */
    virUnrefConnect(dconn);

10038
    return ret;
D
Daniel Veillard 已提交
10039 10040
}

10041

D
Daniel Veillard 已提交
10042 10043 10044 10045 10046 10047
/* 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,
10048
                           unsigned long flags,
10049
                           const char *dname,
D
Daniel Veillard 已提交
10050 10051
                           unsigned long resource)
{
10052 10053
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
10054
    virDomainEventPtr event = NULL;
10055
    int ret = -1;
10056
    int resume = 0;
10057
    qemuDomainObjPrivatePtr priv;
D
Daniel Veillard 已提交
10058

10059
    qemuDriverLock(driver);
10060
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
D
Daniel Veillard 已提交
10061
    if (!vm) {
10062 10063
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
10064 10065
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching uuid '%s'"), uuidstr);
10066
        goto cleanup;
D
Daniel Veillard 已提交
10067
    }
10068
    priv = vm->privateData;
D
Daniel Veillard 已提交
10069

10070 10071
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;
10072
    priv->jobActive = QEMU_JOB_MIGRATION;
10073

D
Daniel P. Berrange 已提交
10074
    if (!virDomainObjIsActive(vm)) {
10075 10076
        qemuReportError(VIR_ERR_OPERATION_INVALID,
                        "%s", _("domain is not running"));
10077
        goto endjob;
D
Daniel Veillard 已提交
10078 10079
    }

10080 10081 10082
    memset(&priv->jobInfo, 0, sizeof(priv->jobInfo));
    priv->jobInfo.type = VIR_DOMAIN_JOB_UNBOUNDED;

10083
    resume = vm->state == VIR_DOMAIN_RUNNING;
10084
    if (!(flags & VIR_MIGRATE_LIVE) && vm->state == VIR_DOMAIN_RUNNING) {
10085
        if (qemuDomainMigrateOffline(driver, vm) < 0)
10086
            goto endjob;
10087 10088
    }

10089
    if ((flags & (VIR_MIGRATE_TUNNELLED | VIR_MIGRATE_PEER2PEER))) {
10090
        if (doPeer2PeerMigrate(dom, driver, vm, uri, flags, dname, resource) < 0)
10091
            /* doPeer2PeerMigrate already set the error, so just get out */
10092
            goto endjob;
10093
    } else {
10094
        if (doNativeMigrate(driver, vm, uri, flags, dname, resource) < 0)
10095
            goto endjob;
10096 10097
    }

D
Daniel Veillard 已提交
10098
    /* Clean up the source domain. */
10099
    qemudShutdownVMDaemon(driver, vm);
10100
    resume = 0;
10101 10102 10103 10104

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_MIGRATED);
C
Chris Lalancette 已提交
10105
    if (!vm->persistent || (flags & VIR_MIGRATE_UNDEFINE_SOURCE)) {
10106
        virDomainDeleteConfig(driver->configDir, driver->autostartDir, vm);
10107 10108
        if (qemuDomainObjEndJob(vm) > 0)
            virDomainRemoveInactive(&driver->domains, vm);
10109 10110
        vm = NULL;
    }
10111
    ret = 0;
D
Daniel Veillard 已提交
10112

10113
endjob:
10114
    if (resume && vm->state == VIR_DOMAIN_PAUSED) {
10115
        /* we got here through some sort of failure; start the domain again */
10116
        qemuDomainObjEnterMonitorWithDriver(driver, vm);
10117
        if (qemuMonitorStartCPUs(priv->mon, dom->conn) < 0) {
10118 10119 10120 10121
            /* 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
             */
10122
            VIR_ERROR(_("Failed to resume guest %s after failure"),
10123
                      vm->def->name);
10124
        }
10125
        qemuDomainObjExitMonitorWithDriver(driver, vm);
10126

10127
        vm->state = VIR_DOMAIN_RUNNING;
10128 10129 10130 10131
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_MIGRATED);
    }
10132 10133 10134
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
10135

10136
cleanup:
10137 10138
    if (vm)
        virDomainObjUnlock(vm);
10139 10140
    if (event)
        qemuDomainEventQueue(driver, event);
10141
    qemuDriverUnlock(driver);
10142
    return ret;
D
Daniel Veillard 已提交
10143 10144 10145 10146 10147 10148 10149 10150 10151
}

/* 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
10152
                           unsigned long flags,
D
Daniel Veillard 已提交
10153 10154
                           int retcode)
{
10155 10156 10157
    struct qemud_driver *driver = dconn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
10158
    virDomainEventPtr event = NULL;
10159
    virErrorPtr orig_err;
C
Chris Lalancette 已提交
10160
    int newVM = 1;
D
Daniel Veillard 已提交
10161

10162 10163 10164
    /* Migration failed. Save the current error so nothing squashes it */
    orig_err = virSaveLastError();

10165
    qemuDriverLock(driver);
10166
    vm = virDomainFindByName(&driver->domains, dname);
D
Daniel Veillard 已提交
10167
    if (!vm) {
10168 10169
        qemuReportError(VIR_ERR_NO_DOMAIN,
                        _("no domain with matching name '%s'"), dname);
10170
        goto cleanup;
D
Daniel Veillard 已提交
10171 10172
    }

10173 10174 10175
    if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0)
        goto cleanup;

D
Daniel Veillard 已提交
10176 10177 10178 10179
    /* 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
10180 10181 10182 10183 10184
        if (flags & VIR_MIGRATE_PERSIST_DEST) {
            if (vm->persistent)
                newVM = 0;
            vm->persistent = 1;

10185
            if (virDomainSaveConfig(driver->configDir, vm->def) < 0) {
C
Chris Lalancette 已提交
10186 10187 10188 10189 10190 10191 10192 10193 10194 10195
                /* 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;
10196
                goto endjob;
C
Chris Lalancette 已提交
10197 10198 10199 10200 10201 10202 10203 10204 10205
            }

            event = virDomainEventNewFromObj(vm,
                                             VIR_DOMAIN_EVENT_DEFINED,
                                             newVM ?
                                             VIR_DOMAIN_EVENT_DEFINED_ADDED :
                                             VIR_DOMAIN_EVENT_DEFINED_UPDATED);
            if (event)
                qemuDomainEventQueue(driver, event);
10206
            event = NULL;
C
Chris Lalancette 已提交
10207 10208

        }
10209
        qemuDomainObjPrivatePtr priv = vm->privateData;
D
Daniel Veillard 已提交
10210
        dom = virGetDomain (dconn, vm->def->name, vm->def->uuid);
10211

10212 10213 10214 10215 10216 10217 10218 10219
        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)
10220 10221
                    qemuReportError(VIR_ERR_INTERNAL_ERROR,
                                    "%s", _("resume operation failed"));
10222 10223 10224
                qemuDomainObjExitMonitorWithDriver(driver, vm);
                goto endjob;
            }
10225
            qemuDomainObjExitMonitorWithDriver(driver, vm);
10226 10227

            vm->state = VIR_DOMAIN_RUNNING;
10228 10229
        }

10230 10231 10232
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_MIGRATED);
10233 10234 10235 10236 10237 10238
        if (vm->state == VIR_DOMAIN_PAUSED) {
            qemuDomainEventQueue(driver, event);
            event = virDomainEventNewFromObj(vm,
                                             VIR_DOMAIN_EVENT_SUSPENDED,
                                             VIR_DOMAIN_EVENT_SUSPENDED_PAUSED);
        }
10239
        virDomainSaveStatus(driver->caps, driver->stateDir, vm);
D
Daniel Veillard 已提交
10240
    } else {
10241
        qemudShutdownVMDaemon(driver, vm);
10242 10243 10244
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         VIR_DOMAIN_EVENT_STOPPED_FAILED);
10245
        if (!vm->persistent) {
10246 10247
            if (qemuDomainObjEndJob(vm) > 0)
                virDomainRemoveInactive(&driver->domains, vm);
10248 10249
            vm = NULL;
        }
D
Daniel Veillard 已提交
10250
    }
10251

10252
endjob:
10253 10254 10255
    if (vm &&
        qemuDomainObjEndJob(vm) == 0)
        vm = NULL;
10256

10257
cleanup:
10258 10259 10260 10261
    if (orig_err) {
        virSetError(orig_err);
        virFreeError(orig_err);
    }
10262 10263
    if (vm)
        virDomainObjUnlock(vm);
10264 10265
    if (event)
        qemuDomainEventQueue(driver, event);
10266
    qemuDriverUnlock(driver);
10267
    return dom;
D
Daniel Veillard 已提交
10268 10269
}

10270 10271 10272 10273 10274 10275 10276 10277 10278 10279 10280 10281 10282 10283 10284 10285
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;

10286
    def = virNodeDeviceDefParseString(xml, EXISTING_DEVICE);
10287 10288 10289 10290 10291 10292 10293 10294 10295 10296 10297 10298 10299 10300 10301 10302 10303
    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) {
10304 10305
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("device %s is not a PCI device"), dev->name);
10306 10307 10308 10309 10310 10311 10312 10313 10314 10315 10316 10317 10318 10319 10320 10321 10322 10323 10324 10325
        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;

10326
    pci = pciGetDevice(domain, bus, slot, function);
10327 10328 10329
    if (!pci)
        return -1;

10330
    if (pciDettachDevice(pci) < 0)
10331 10332 10333 10334
        goto out;

    ret = 0;
out:
10335
    pciFreeDevice(pci);
10336 10337 10338 10339 10340 10341 10342 10343 10344 10345 10346 10347 10348
    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;

10349
    pci = pciGetDevice(domain, bus, slot, function);
10350 10351 10352
    if (!pci)
        return -1;

10353
    if (pciReAttachDevice(pci) < 0)
10354 10355 10356 10357
        goto out;

    ret = 0;
out:
10358
    pciFreeDevice(pci);
10359 10360 10361 10362 10363 10364
    return ret;
}

static int
qemudNodeDeviceReset (virNodeDevicePtr dev)
{
10365
    struct qemud_driver *driver = dev->conn->privateData;
10366 10367 10368 10369 10370 10371 10372
    pciDevice *pci;
    unsigned domain, bus, slot, function;
    int ret = -1;

    if (qemudNodeDeviceGetPciInfo(dev, &domain, &bus, &slot, &function) < 0)
        return -1;

10373
    pci = pciGetDevice(domain, bus, slot, function);
10374 10375 10376
    if (!pci)
        return -1;

10377 10378
    qemuDriverLock(driver);

10379
    if (pciResetDevice(pci, driver->activePciHostdevs) < 0)
10380 10381 10382 10383
        goto out;

    ret = 0;
out:
10384
    qemuDriverUnlock(driver);
10385
    pciFreeDevice(pci);
10386 10387 10388
    return ret;
}

10389 10390 10391 10392 10393 10394 10395 10396 10397 10398 10399
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) {
10400 10401
        qemuReportError(VIR_ERR_NO_SUPPORT,
                        "%s", _("cannot get host CPU capabilities"));
10402 10403
    }
    else
10404
        ret = cpuCompareXML(driver->caps->host.cpu, xmlDesc);
10405 10406 10407 10408 10409 10410

    qemuDriverUnlock(driver);

    return ret;
}

10411

10412 10413 10414 10415 10416 10417 10418 10419 10420 10421 10422 10423 10424
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;
}

10425 10426 10427 10428 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

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;
}


10468 10469 10470 10471 10472 10473 10474 10475 10476 10477 10478 10479 10480 10481 10482 10483 10484 10485 10486 10487 10488 10489
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);
10490
            priv->jobSignals |= QEMU_JOB_SIGNAL_CANCEL;
10491 10492 10493 10494 10495 10496 10497 10498 10499 10500 10501 10502 10503 10504 10505 10506 10507 10508 10509 10510
        } 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;
}


10511 10512 10513 10514 10515 10516 10517 10518 10519 10520 10521 10522 10523 10524 10525 10526 10527 10528 10529 10530 10531 10532 10533 10534 10535 10536 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
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;

    if (flags != 0) {
        qemuReportError(VIR_ERR_INVALID_ARG,
                        _("unsupported flags (0x%x) passed to '%s'"), flags, __FUNCTION__);
        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 (!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 10684 10685 10686 10687 10688 10689 10690 10691 10692 10693 10694 10695 10696 10697 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 10803 10804 10805 10806 10807 10808 10809 10810 10811 10812 10813 10814 10815 10816 10817 10818 10819 10820 10821 10822 10823 10824 10825 10826 10827 10828 10829 10830 10831 10832 10833 10834 10835 10836 10837 10838 10839 10840 10841 10842 10843 10844 10845 10846 10847 10848 10849 10850 10851 10852 10853 10854 10855 10856 10857 10858 10859 10860 10861 10862 10863 10864 10865 10866 10867 10868 10869 10870 10871 10872 10873 10874 10875 10876 10877 10878 10879 10880 10881 10882 10883 10884 10885 10886 10887 10888 10889 10890 10891 10892 10893 10894 10895 10896 10897 10898 10899 10900 10901 10902 10903 10904 10905 10906 10907 10908 10909 10910 10911 10912 10913 10914 10915 10916 10917 10918 10919 10920 10921 10922 10923 10924 10925 10926 10927 10928 10929 10930 10931 10932 10933 10934 10935 10936 10937 10938 10939 10940 10941 10942 10943 10944 10945 10946 10947 10948 10949 10950 10951 10952 10953 10954 10955 10956 10957 10958 10959 10960 10961 10962 10963 10964 10965 10966 10967 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 10995 10996 10997 10998 10999 11000 11001 11002 11003 11004 11005 11006 11007 11008 11009 11010 11011 11012 11013 11014 11015 11016 11017 11018 11019 11020 11021 11022 11023 11024 11025 11026 11027 11028 11029 11030 11031 11032 11033 11034 11035 11036 11037 11038 11039 11040 11041 11042 11043 11044 11045 11046 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 11233 11234 11235 11236 11237 11238 11239 11240 11241 11242 11243 11244 11245 11246 11247 11248 11249 11250 11251 11252
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,
                                                        unsigned int flags ATTRIBUTE_UNUSED)
{
    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;

    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,
                                       unsigned int flags ATTRIBUTE_UNUSED)
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm = NULL;
    int n = -1;

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

    n = virDomainSnapshotObjListGetNames(&vm->snapshots, names, nameslen);

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);
    return n;
}

static int qemuDomainSnapshotNum(virDomainPtr domain,
                                 unsigned int flags ATTRIBUTE_UNUSED)
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm = NULL;
    int n = -1;

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

    n = virDomainSnapshotObjListNum(&vm->snapshots);

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
    qemuDriverUnlock(driver);
    return n;
}

static virDomainSnapshotPtr qemuDomainSnapshotLookupByName(virDomainPtr domain,
                                                           const char *name,
                                                           unsigned int flags ATTRIBUTE_UNUSED)
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm;
    virDomainSnapshotObjPtr snap = NULL;
    virDomainSnapshotPtr snapshot = NULL;

    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,
                                        unsigned int flags ATTRIBUTE_UNUSED)
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;

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

    ret = (vm->current_snapshot != NULL);

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

static virDomainSnapshotPtr qemuDomainSnapshotCurrent(virDomainPtr domain,
                                                      unsigned int flags ATTRIBUTE_UNUSED)
{
    struct qemud_driver *driver = domain->conn->privateData;
    virDomainObjPtr vm;
    virDomainSnapshotPtr snapshot = NULL;

    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,
                                       unsigned int flags ATTRIBUTE_UNUSED)
{
    struct qemud_driver *driver = snapshot->domain->conn->privateData;
    virDomainObjPtr vm = NULL;
    char *xml = NULL;
    virDomainSnapshotObjPtr snap = NULL;
    char uuidstr[VIR_UUID_STRING_BUFLEN];

    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,
                                      unsigned int flags ATTRIBUTE_UNUSED)
{
    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;

    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;

    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;
}
11253

11254 11255 11256 11257 11258
static virDriver qemuDriver = {
    VIR_DRV_QEMU,
    "QEMU",
    qemudOpen, /* open */
    qemudClose, /* close */
D
Daniel Veillard 已提交
11259
    qemudSupportsFeature, /* supports_feature */
11260 11261
    qemudGetType, /* type */
    qemudGetVersion, /* version */
11262
    NULL, /* libvirtVersion (impl. in libvirt.c) */
11263
    virGetHostname, /* getHostname */
11264
    qemudGetMaxVCPUs, /* getMaxVcpus */
11265
    nodeGetInfo, /* nodeGetInfo */
11266 11267 11268
    qemudGetCapabilities, /* getCapabilities */
    qemudListDomains, /* listDomains */
    qemudNumDomains, /* numOfDomains */
11269
    qemudDomainCreate, /* domainCreateXML */
11270 11271 11272 11273 11274
    qemudDomainLookupByID, /* domainLookupByID */
    qemudDomainLookupByUUID, /* domainLookupByUUID */
    qemudDomainLookupByName, /* domainLookupByName */
    qemudDomainSuspend, /* domainSuspend */
    qemudDomainResume, /* domainResume */
11275
    qemudDomainShutdown, /* domainShutdown */
11276 11277 11278
    NULL, /* domainReboot */
    qemudDomainDestroy, /* domainDestroy */
    qemudDomainGetOSType, /* domainGetOSType */
11279
    qemudDomainGetMaxMemory, /* domainGetMaxMemory */
11280
    NULL, /* domainSetMaxMemory */
11281
    qemudDomainSetMemory, /* domainSetMemory */
11282 11283 11284
    qemudDomainGetInfo, /* domainGetInfo */
    qemudDomainSave, /* domainSave */
    qemudDomainRestore, /* domainRestore */
P
Paolo Bonzini 已提交
11285
    qemudDomainCoreDump, /* domainCoreDump */
11286
    qemudDomainSetVcpus, /* domainSetVcpus */
11287 11288
    qemudDomainPinVcpu, /* domainPinVcpu */
    qemudDomainGetVcpus, /* domainGetVcpus */
11289
    qemudDomainGetMaxVcpus, /* domainGetMaxVcpus */
11290 11291
    qemudDomainGetSecurityLabel, /* domainGetSecurityLabel */
    qemudNodeGetSecurityModel, /* nodeGetSecurityModel */
11292
    qemudDomainDumpXML, /* domainDumpXML */
11293
    qemuDomainXMLFromNative, /* domainXmlFromNative */
11294
    qemuDomainXMLToNative, /* domainXMLToNative */
11295 11296
    qemudListDefinedDomains, /* listDefinedDomains */
    qemudNumDefinedDomains, /* numOfDefinedDomains */
11297 11298 11299
    qemudDomainStart, /* domainCreate */
    qemudDomainDefine, /* domainDefineXML */
    qemudDomainUndefine, /* domainUndefine */
11300
    qemudDomainAttachDevice, /* domainAttachDevice */
11301
    qemudDomainAttachDeviceFlags, /* domainAttachDeviceFlags */
11302
    qemudDomainDetachDevice, /* domainDetachDevice */
11303
    qemudDomainDetachDeviceFlags, /* domainDetachDeviceFlags */
11304
    qemuDomainUpdateDeviceFlags, /* domainUpdateDeviceFlags */
11305 11306
    qemudDomainGetAutostart, /* domainGetAutostart */
    qemudDomainSetAutostart, /* domainSetAutostart */
11307 11308 11309
    qemuGetSchedulerType, /* domainGetSchedulerType */
    qemuGetSchedulerParameters, /* domainGetSchedulerParameters */
    qemuSetSchedulerParameters, /* domainSetSchedulerParameters */
D
Daniel Veillard 已提交
11310 11311
    NULL, /* domainMigratePrepare (v1) */
    qemudDomainMigratePerform, /* domainMigratePerform */
11312
    NULL, /* domainMigrateFinish */
11313
    qemudDomainBlockStats, /* domainBlockStats */
11314
    qemudDomainInterfaceStats, /* domainInterfaceStats */
11315
    qemudDomainMemoryStats, /* domainMemoryStats */
11316
    qemudDomainBlockPeek, /* domainBlockPeek */
R
Richard W.M. Jones 已提交
11317
    qemudDomainMemoryPeek, /* domainMemoryPeek */
11318 11319
    nodeGetCellsFreeMemory, /* nodeGetCellsFreeMemory */
    nodeGetFreeMemory,  /* getFreeMemory */
11320 11321
    qemuDomainEventRegister, /* domainEventRegister */
    qemuDomainEventDeregister, /* domainEventDeregister */
D
Daniel Veillard 已提交
11322 11323
    qemudDomainMigratePrepare2, /* domainMigratePrepare2 */
    qemudDomainMigrateFinish2, /* domainMigrateFinish2 */
11324 11325 11326
    qemudNodeDeviceDettach, /* nodeDeviceDettach */
    qemudNodeDeviceReAttach, /* nodeDeviceReAttach */
    qemudNodeDeviceReset, /* nodeDeviceReset */
C
Chris Lalancette 已提交
11327
    qemudDomainMigratePrepareTunnel, /* domainMigratePrepareTunnel */
11328 11329 11330 11331
    qemuIsEncrypted,
    qemuIsSecure,
    qemuDomainIsActive,
    qemuDomainIsPersistent,
11332
    qemuCPUCompare, /* cpuCompare */
11333
    qemuCPUBaseline, /* cpuBaseline */
11334
    qemuDomainGetJobInfo, /* domainGetJobInfo */
11335
    qemuDomainAbortJob, /* domainAbortJob */
11336
    qemuDomainMigrateSetMaxDowntime, /* domainMigrateSetMaxDowntime */
11337 11338
    qemuDomainEventRegisterAny, /* domainEventRegisterAny */
    qemuDomainEventDeregisterAny, /* domainEventDeregisterAny */
11339 11340 11341
    qemuDomainManagedSave, /* domainManagedSave */
    qemuDomainHasManagedSaveImage, /* domainHasManagedSaveImage */
    qemuDomainManagedSaveRemove, /* domainManagedSaveRemove */
C
Chris Lalancette 已提交
11342 11343 11344 11345 11346 11347 11348 11349 11350
    qemuDomainSnapshotCreateXML, /* domainSnapshotCreateXML */
    qemuDomainSnapshotDumpXML, /* domainSnapshotDumpXML */
    qemuDomainSnapshotNum, /* domainSnapshotNum */
    qemuDomainSnapshotListNames, /* domainSnapshotListNames */
    qemuDomainSnapshotLookupByName, /* domainSnapshotLookupByName */
    qemuDomainHasCurrentSnapshot, /* domainHasCurrentSnapshot */
    qemuDomainSnapshotCurrent, /* domainSnapshotCurrent */
    qemuDomainRevertToSnapshot, /* domainRevertToSnapshot */
    qemuDomainSnapshotDelete, /* domainSnapshotDelete */
11351 11352 11353
};


11354
static virStateDriver qemuStateDriver = {
11355
    .name = "QEMU",
11356 11357 11358 11359
    .initialize = qemudStartup,
    .cleanup = qemudShutdown,
    .reload = qemudReload,
    .active = qemudActive,
11360
};
11361

S
Stefan Berger 已提交
11362 11363 11364 11365 11366 11367 11368 11369 11370 11371 11372 11373 11374 11375 11376
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,
};

11377
int qemuRegister(void) {
11378 11379
    virRegisterDriver(&qemuDriver);
    virRegisterStateDriver(&qemuStateDriver);
S
Stefan Berger 已提交
11380
    virNWFilterRegisterCallbackDriver(&qemuCallbackDriver);
11381 11382
    return 0;
}