qemu_driver.c 232.3 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 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 28 29 30
#include <sys/types.h>
#include <sys/poll.h>
#include <dirent.h>
#include <limits.h>
#include <string.h>
31
#include <stdbool.h>
D
Daniel P. Berrange 已提交
32 33 34 35 36 37
#include <stdio.h>
#include <strings.h>
#include <stdarg.h>
#include <stdlib.h>
#include <unistd.h>
#include <errno.h>
38
#include <sys/utsname.h>
39 40 41 42
#include <sys/stat.h>
#include <fcntl.h>
#include <signal.h>
#include <paths.h>
43 44
#include <pwd.h>
#include <stdio.h>
45
#include <sys/wait.h>
46
#include <sys/ioctl.h>
47
#include <sys/un.h>
D
Daniel P. Berrange 已提交
48

49 50 51 52
#if HAVE_SCHED_H
#include <sched.h>
#endif

53
#include "virterror_internal.h"
54
#include "logging.h"
55
#include "datatypes.h"
56 57
#include "qemu_driver.h"
#include "qemu_conf.h"
J
Jim Meyering 已提交
58
#include "c-ctype.h"
59
#include "event.h"
60
#include "buf.h"
61
#include "util.h"
62
#include "nodeinfo.h"
63
#include "stats_linux.h"
64
#include "capabilities.h"
65
#include "memory.h"
66
#include "uuid.h"
67
#include "domain_conf.h"
68 69
#include "node_device_conf.h"
#include "pci.h"
70
#include "hostusb.h"
71
#include "security/security_driver.h"
72
#include "cgroup.h"
73

74

75 76
#define VIR_FROM_THIS VIR_FROM_QEMU

77 78 79
#define QEMU_CMD_PROMPT "\n(qemu) "
#define QEMU_PASSWD_PROMPT "Password: "

80 81
static int qemudShutdown(void);

82 83
static void qemuDriverLock(struct qemud_driver *driver)
{
84
    virMutexLock(&driver->lock);
85 86 87
}
static void qemuDriverUnlock(struct qemud_driver *driver)
{
88
    virMutexUnlock(&driver->lock);
89 90
}

91 92 93 94 95 96
/* Return -1 for error, 0 for success */
typedef int qemudMonitorExtraPromptHandler(const virDomainObjPtr vm,
                                           const char *buf,
                                           const char *prompt,
                                           void *data);

97 98 99
static void qemuDomainEventFlush(int timer, void *opaque);
static void qemuDomainEventQueue(struct qemud_driver *driver,
                                 virDomainEventPtr event);
100

101 102
static void qemudDispatchVMEvent(int watch,
                                 int fd,
103 104 105
                                 int events,
                                 void *opaque);

106 107
static int qemudStartVMDaemon(virConnectPtr conn,
                              struct qemud_driver *driver,
108
                              virDomainObjPtr vm,
109 110
                              const char *migrateFrom,
                              int stdin_fd);
111

112 113
static void qemudShutdownVMDaemon(virConnectPtr conn,
                                  struct qemud_driver *driver,
114
                                  virDomainObjPtr vm);
115

116
static int qemudDomainGetMaxVcpus(virDomainPtr dom);
117

118 119 120
static int qemudMonitorCommand(const virDomainObjPtr vm,
                               const char *cmd,
                               char **reply);
121 122 123 124
static int qemudMonitorCommandWithFd(const virDomainObjPtr vm,
                                     const char *cmd,
                                     int scm_fd,
                                     char **reply);
125 126 127 128 129 130 131
static int qemudMonitorCommandWithHandler(const virDomainObjPtr vm,
                                          const char *cmd,
                                          const char *extraPrompt,
                                          qemudMonitorExtraPromptHandler extraHandler,
                                          void *handlerData,
                                          int scm_fd,
                                          char **reply);
132 133 134 135
static int qemudMonitorCommandExtra(const virDomainObjPtr vm,
                                    const char *cmd,
                                    const char *extra,
                                    const char *extraPrompt,
136
                                    int scm_fd,
137
                                    char **reply);
138 139
static int qemudMonitorSendCont(virConnectPtr conn,
                                const virDomainObjPtr vm);
140 141 142
static int qemudDomainSetMemoryBalloon(virConnectPtr conn,
                                       virDomainObjPtr vm,
                                       unsigned long newmem);
143 144
static int qemudDetectVcpuPIDs(virConnectPtr conn,
                               virDomainObjPtr vm);
145

146 147 148
static int qemuUpdateActivePciHostdevs(struct qemud_driver *driver,
                                       virDomainDefPtr def);

J
Jim Meyering 已提交
149
static struct qemud_driver *qemu_driver = NULL;
150

151 152 153 154 155 156 157 158 159
static int qemuCgroupControllerActive(struct qemud_driver *driver,
                                      int controller)
{
    if (driver->cgroup == NULL)
        return 0;
    if (driver->cgroupControllers & (1 << controller))
        return 1;
    return 0;
}
160

161
static int
162
qemudLogFD(virConnectPtr conn, struct qemud_driver *driver, const char* name)
163 164 165
{
    char logfile[PATH_MAX];
    mode_t logmode;
G
Guido Günther 已提交
166
    int ret, fd = -1;
167

168 169
    if ((ret = snprintf(logfile, sizeof(logfile), "%s/%s.log",
                        driver->logDir, name))
G
Guido Günther 已提交
170
        < 0 || ret >= sizeof(logfile)) {
171
        virReportOOMError(conn);
172 173 174 175
        return -1;
    }

    logmode = O_CREAT | O_WRONLY;
176 177
    /* Only logrotate files in /var/log, so only append if running privileged */
    if (driver->privileged)
178
        logmode |= O_APPEND;
179 180 181
    else
        logmode |= O_TRUNC;

182
    if ((fd = open(logfile, logmode, S_IRUSR | S_IWUSR)) < 0) {
183 184 185
        virReportSystemError(conn, errno,
                             _("failed to create logfile %s"),
                             logfile);
186 187
        return -1;
    }
188
    if (virSetCloseExec(fd) < 0) {
189 190
        virReportSystemError(conn, errno, "%s",
                             _("Unable to set VM logfile close-on-exec flag"));
191 192 193 194 195 196 197
        close(fd);
        return -1;
    }
    return fd;
}


198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214
static int
qemudLogReadFD(virConnectPtr conn, const char* logDir, const char* name, off_t pos)
{
    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)) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                         _("failed to build logfile name %s/%s.log"),
                         logDir, name);
        return -1;
    }


    if ((fd = open(logfile, logmode)) < 0) {
215 216 217
        virReportSystemError(conn, errno,
                             _("failed to create logfile %s"),
                             logfile);
218 219
        return -1;
    }
220
    if (virSetCloseExec(fd) < 0) {
221 222
        virReportSystemError(conn, errno, "%s",
                             _("Unable to set VM logfile close-on-exec flag"));
223 224 225 226
        close(fd);
        return -1;
    }
    if (lseek(fd, pos, SEEK_SET) < 0) {
227 228 229
        virReportSystemError(conn, errno,
                             _("Unable to seek to %lld in %s"),
                             (long long) pos, logfile);
230 231 232 233 234 235
        close(fd);
    }
    return fd;
}


236 237 238
static void
qemudAutostartConfigs(struct qemud_driver *driver) {
    unsigned int i;
239 240 241 242 243
    /* 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
     */
244 245 246
    virConnectPtr conn = virConnectOpen(driver->privileged ?
                                        "qemu:///system" :
                                        "qemu:///session");
247
    /* Ignoring NULL conn which is mostly harmless here */
248

249
    qemuDriverLock(driver);
250
    for (i = 0 ; i < driver->domains.count ; i++) {
251
        virDomainObjPtr vm = driver->domains.objs[i];
252
        virDomainObjLock(vm);
253 254
        if (vm->autostart &&
            !virDomainIsActive(vm)) {
255 256 257 258
            int ret;

            virResetLastError();
            ret = qemudStartVMDaemon(conn, driver, vm, NULL, -1);
259 260
            if (ret < 0) {
                virErrorPtr err = virGetLastError();
261 262
                VIR_ERROR(_("Failed to autostart VM '%s': %s\n"),
                          vm->def->name,
263
                          err ? err->message : "");
264
            } else {
265 266 267 268 269 270
                virDomainEventPtr event =
                    virDomainEventNewFromObj(vm,
                                             VIR_DOMAIN_EVENT_STARTED,
                                             VIR_DOMAIN_EVENT_STARTED_BOOTED);
                if (event)
                    qemuDomainEventQueue(driver, event);
271
            }
272
        }
273
        virDomainObjUnlock(vm);
274
    }
275
    qemuDriverUnlock(driver);
276

277 278
    if (conn)
        virConnectClose(conn);
279 280
}

281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297

/**
 * qemudRemoveDomainStatus
 *
 * remove all state files of a domain from statedir
 *
 * Returns 0 on success
 */
static int
qemudRemoveDomainStatus(virConnectPtr conn,
                        struct qemud_driver *driver,
                        virDomainObjPtr vm)
{
    int rc = -1;
    char *file = NULL;

    if (virAsprintf(&file, "%s/%s.xml", driver->stateDir, vm->def->name) < 0) {
298
        virReportOOMError(conn);
299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317
        goto cleanup;
    }

    if (unlink(file) < 0 && errno != ENOENT && errno != ENOTDIR) {
        qemudReportError(conn, vm, NULL, VIR_ERR_INTERNAL_ERROR,
                         _("Failed to unlink status file %s"), file);
        goto cleanup;
    }

    if(virFileDeletePid(driver->stateDir, vm->def->name))
        goto cleanup;

    rc = 0;
cleanup:
    VIR_FREE(file);
    return rc;
}


318 319 320 321 322
static int qemudOpenMonitor(virConnectPtr conn,
                            struct qemud_driver* driver,
                            virDomainObjPtr vm,
                            int reconnect);

323 324 325 326

/*
 * Open an existing VM's monitor, re-detect VCPU threads
 * and re-reserve the security labels in use
327 328
 */
static int
329 330
qemuReconnectDomain(struct qemud_driver *driver,
                    virDomainObjPtr obj)
331
{
332
    int rc;
333

334
    if ((rc = qemudOpenMonitor(NULL, driver, obj, 1)) != 0) {
335 336 337 338
        VIR_ERROR(_("Failed to reconnect monitor for %s: %d\n"),
                  obj->def->name, rc);
        goto error;
    }
339

340 341 342 343
    if (qemuUpdateActivePciHostdevs(driver, obj->def) < 0) {
        goto error;
    }

344 345 346 347 348
    if (obj->def->seclabel.type == VIR_DOMAIN_SECLABEL_DYNAMIC &&
        driver->securityDriver &&
        driver->securityDriver->domainReserveSecurityLabel &&
        driver->securityDriver->domainReserveSecurityLabel(NULL, obj) < 0)
        return -1;
349

350 351
    if (obj->def->id >= driver->nextvmid)
        driver->nextvmid = obj->def->id + 1;
352

353
    return 0;
354

355 356 357
error:
    return -1;
}
358

359 360 361 362 363 364 365 366 367 368
/**
 * qemudReconnectVMs
 *
 * Try to re-open the resources for live VMs that we care
 * about.
 */
static void
qemuReconnectDomains(struct qemud_driver *driver)
{
    int i;
369

370 371 372 373 374 375 376 377 378
    for (i = 0 ; i < driver->domains.count ; i++) {
        virDomainObjPtr obj = driver->domains.objs[i];

        virDomainObjLock(obj);
        if (qemuReconnectDomain(driver, obj) < 0) {
            /* If we can't get the monitor back, then kill the VM
             * so user has ability to start it again later without
             * danger of ending up running twice */
            qemudShutdownVMDaemon(NULL, driver, obj);
379
        }
380
        virDomainObjUnlock(obj);
381 382 383
    }
}

384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416

static int
qemudSecurityCapsInit(virSecurityDriverPtr secdrv,
                      virCapsPtr caps)
{
    const char *doi, *model;

    doi = virSecurityDriverGetDOI(secdrv);
    model = virSecurityDriverGetModel(secdrv);

    caps->host.secModel.model = strdup(model);
    if (!caps->host.secModel.model) {
        char ebuf[1024];
        VIR_ERROR(_("Failed to copy secModel model: %s"),
                  virStrerror(errno, ebuf, sizeof ebuf));
        return -1;
    }

    caps->host.secModel.doi = strdup(doi);
    if (!caps->host.secModel.doi) {
        char ebuf[1024];
        VIR_ERROR(_("Failed to copy secModel DOI: %s"),
                  virStrerror(errno, ebuf, sizeof ebuf));
        return -1;
    }

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

    return 0;
}


417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436
static int
qemudSecurityInit(struct qemud_driver *qemud_drv)
{
    int ret;
    virSecurityDriverPtr security_drv;

    ret = virSecurityDriverStartup(&security_drv,
                                   qemud_drv->securityDriverName);
    if (ret == -1) {
        VIR_ERROR0(_("Failed to start security driver"));
        return -1;
    }
    /* No security driver wanted to be enabled: just return */
    if (ret == -2) {
        VIR_INFO0(_("No security driver available"));
        return 0;
    }

    qemud_drv->securityDriver = security_drv;

437
    VIR_INFO("Initialized security driver %s", security_drv->name);
438 439 440 441 442

    /*
     * Add security policy host caps now that the security driver is
     * initialized.
     */
443 444
    return qemudSecurityCapsInit(security_drv, qemud_drv->caps);
}
445 446


447

448 449 450 451 452 453
/**
 * qemudStartup:
 *
 * Initialization function for the QEmu daemon
 */
static int
454
qemudStartup(int privileged) {
455
    char *base = NULL;
D
Daniel P. Berrange 已提交
456
    char driverConf[PATH_MAX];
457
    int rc;
458

459
    if (VIR_ALLOC(qemu_driver) < 0)
460 461
        return -1;

462
    if (virMutexInit(&qemu_driver->lock) < 0) {
463
        VIR_ERROR("%s", _("cannot initialize mutex"));
464 465 466
        VIR_FREE(qemu_driver);
        return -1;
    }
467
    qemuDriverLock(qemu_driver);
468
    qemu_driver->privileged = privileged;
469

470 471 472
    /* Don't have a dom0 so start from 1 */
    qemu_driver->nextvmid = 1;

473 474
    /* Init callback list */
    if(VIR_ALLOC(qemu_driver->domainEventCallbacks) < 0)
475
        goto out_of_memory;
476 477 478 479 480 481
    if (!(qemu_driver->domainEventQueue = virDomainEventQueueNew()))
        goto out_of_memory;

    if ((qemu_driver->domainEventTimer =
         virEventAddTimeout(-1, qemuDomainEventFlush, qemu_driver, NULL)) < 0)
        goto error;
482

483
    if (privileged) {
484 485
        if (virAsprintf(&qemu_driver->logDir,
                        "%s/log/libvirt/qemu", LOCAL_STATE_DIR) == -1)
486
            goto out_of_memory;
487

D
Daniel P. Berrange 已提交
488
        if ((base = strdup (SYSCONF_DIR "/libvirt")) == NULL)
489
            goto out_of_memory;
490 491

        if (virAsprintf(&qemu_driver->stateDir,
492
                      "%s/run/libvirt/qemu", LOCAL_STATE_DIR) == -1)
493
            goto out_of_memory;
494 495 496 497 498 499 500 501

        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;
502
    } else {
503
        uid_t uid = geteuid();
504 505
        char *userdir = virGetUserDirectory(NULL, uid);
        if (!userdir)
506
            goto error;
507

508
        if (virAsprintf(&qemu_driver->logDir,
509 510
                        "%s/.libvirt/qemu/log", userdir) == -1) {
            VIR_FREE(userdir);
511
            goto out_of_memory;
512
        }
513

514 515
        if (virAsprintf(&base, "%s/.libvirt", userdir) == -1) {
            VIR_FREE(userdir);
516
            goto out_of_memory;
517 518
        }
        VIR_FREE(userdir);
519 520 521

        if (virAsprintf(&qemu_driver->stateDir, "%s/qemu/run", base) == -1)
            goto out_of_memory;
522 523 524 525
        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;
526 527 528
    }

    if (virFileMakePath(qemu_driver->stateDir) < 0) {
529
        char ebuf[1024];
530 531
        VIR_ERROR(_("Failed to create state dir '%s': %s\n"),
                  qemu_driver->stateDir, virStrerror(errno, ebuf, sizeof ebuf));
532
        goto error;
533
    }
534 535 536 537 538 539 540 541 542 543 544 545
    if (virFileMakePath(qemu_driver->libDir) < 0) {
        char ebuf[1024];
        VIR_ERROR(_("Failed to create lib dir '%s': %s\n"),
                  qemu_driver->libDir, virStrerror(errno, ebuf, sizeof ebuf));
        goto error;
    }
    if (virFileMakePath(qemu_driver->cacheDir) < 0) {
        char ebuf[1024];
        VIR_ERROR(_("Failed to create cache dir '%s': %s\n"),
                  qemu_driver->cacheDir, virStrerror(errno, ebuf, sizeof ebuf));
        goto error;
    }
546 547 548 549

    /* Configuration paths are either ~/.libvirt/qemu/... (session) or
     * /etc/libvirt/qemu/... (system).
     */
D
Daniel P. Berrange 已提交
550
    if (snprintf (driverConf, sizeof(driverConf), "%s/qemu.conf", base) == -1)
551
        goto out_of_memory;
D
Daniel P. Berrange 已提交
552
    driverConf[sizeof(driverConf)-1] = '\0';
553

554
    if (virAsprintf(&qemu_driver->configDir, "%s/qemu", base) == -1)
555 556
        goto out_of_memory;

557
    if (virAsprintf(&qemu_driver->autostartDir, "%s/qemu/autostart", base) == -1)
558 559
        goto out_of_memory;

560
    VIR_FREE(base);
561

562 563 564 565 566 567 568
    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)));
    }

569
    if ((qemu_driver->caps = qemudCapsInit(NULL)) == NULL)
570
        goto out_of_memory;
D
Daniel P. Berrange 已提交
571

572 573 574
    if ((qemu_driver->activePciHostdevs = pciDeviceListNew(NULL)) == NULL)
        goto error;

575
    if (qemudLoadDriverConfig(qemu_driver, driverConf) < 0) {
576 577 578
        goto error;
    }

579 580 581 582 583 584 585 586 587 588 589 590 591 592 593
    if (privileged) {
        if (chown(qemu_driver->libDir, qemu_driver->user, qemu_driver->group) < 0) {
            virReportSystemError(NULL, errno,
                                 _("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) {
            virReportSystemError(NULL, errno,
                                 _("unable to set ownership of '%s' to %d:%d"),
                                 qemu_driver->cacheDir, qemu_driver->user, qemu_driver->group);
            goto error;
        }
    }

594
    if (qemudSecurityInit(qemu_driver) < 0) {
595
        goto error;
D
Daniel P. Berrange 已提交
596 597
    }

598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628
    /* 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) {
            virReportSystemError(NULL, rc,
                                 _("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) {
            virReportSystemError(NULL, errno,
                                 _("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;
    }

629 630 631 632 633 634 635 636 637 638 639 640
    /* Get all the running persistent or transient configs first */
    if (virDomainLoadAllConfigs(NULL,
                                qemu_driver->caps,
                                &qemu_driver->domains,
                                qemu_driver->stateDir,
                                NULL,
                                1, NULL, NULL) < 0)
        goto error;

    qemuReconnectDomains(qemu_driver);

    /* Then inactive persistent configs */
641 642 643 644
    if (virDomainLoadAllConfigs(NULL,
                                qemu_driver->caps,
                                &qemu_driver->domains,
                                qemu_driver->configDir,
645
                                qemu_driver->autostartDir,
646
                                0, NULL, NULL) < 0)
647
        goto error;
648 649
    qemuDriverUnlock(qemu_driver);

650 651
    qemudAutostartConfigs(qemu_driver);

652

653 654
    return 0;

655
out_of_memory:
656
    virReportOOMError(NULL);
657 658 659
error:
    if (qemu_driver)
        qemuDriverUnlock(qemu_driver);
660
    VIR_FREE(base);
661
    qemudShutdown();
662 663 664
    return -1;
}

665 666 667 668
static void qemudNotifyLoadDomain(virDomainObjPtr vm, int newVM, void *opaque)
{
    struct qemud_driver *driver = opaque;

669 670 671 672 673 674 675 676
    if (newVM) {
        virDomainEventPtr event =
            virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_DEFINED,
                                     VIR_DOMAIN_EVENT_DEFINED_ADDED);
        if (event)
            qemuDomainEventQueue(driver, event);
    }
677 678
}

679 680 681 682 683 684 685 686
/**
 * qemudReload:
 *
 * Function to restart the QEmu daemon, it will recheck the configuration
 * files and update its state and the networking
 */
static int
qemudReload(void) {
687 688 689
    if (!qemu_driver)
        return 0;

690
    qemuDriverLock(qemu_driver);
691 692 693 694
    virDomainLoadAllConfigs(NULL,
                            qemu_driver->caps,
                            &qemu_driver->domains,
                            qemu_driver->configDir,
695
                            qemu_driver->autostartDir,
696
                            0, qemudNotifyLoadDomain, qemu_driver);
697
    qemuDriverUnlock(qemu_driver);
698

699
    qemudAutostartConfigs(qemu_driver);
700 701

    return 0;
702 703
}

704 705 706 707 708 709 710 711 712 713
/**
 * 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) {
714
    unsigned int i;
715
    int active = 0;
716

717 718 719
    if (!qemu_driver)
        return 0;

720 721 722 723 724 725 726 727
    qemuDriverLock(qemu_driver);
    for (i = 0 ; i < qemu_driver->domains.count ; i++) {
        virDomainObjPtr vm = qemu_driver->domains.objs[i];
        virDomainObjLock(vm);
        if (virDomainIsActive(vm))
            active = 1;
        virDomainObjUnlock(vm);
    }
728

729 730
    qemuDriverUnlock(qemu_driver);
    return active;
731 732
}

733 734 735 736 737 738 739
/**
 * qemudShutdown:
 *
 * Shutdown the QEmu daemon, it will stop all active domains and networks
 */
static int
qemudShutdown(void) {
740

741
    if (!qemu_driver)
742
        return -1;
743

744
    qemuDriverLock(qemu_driver);
745
    pciDeviceListFree(NULL, qemu_driver->activePciHostdevs);
746 747
    virCapabilitiesFree(qemu_driver->caps);

748
    virDomainObjListFree(&qemu_driver->domains);
749

750
    VIR_FREE(qemu_driver->securityDriverName);
751
    VIR_FREE(qemu_driver->logDir);
752 753
    VIR_FREE(qemu_driver->configDir);
    VIR_FREE(qemu_driver->autostartDir);
754
    VIR_FREE(qemu_driver->stateDir);
755 756
    VIR_FREE(qemu_driver->libDir);
    VIR_FREE(qemu_driver->cacheDir);
757
    VIR_FREE(qemu_driver->vncTLSx509certdir);
J
Jim Meyering 已提交
758
    VIR_FREE(qemu_driver->vncListen);
759
    VIR_FREE(qemu_driver->vncPassword);
760
    VIR_FREE(qemu_driver->vncSASLdir);
761
    VIR_FREE(qemu_driver->saveImageFormat);
762 763
    VIR_FREE(qemu_driver->hugetlbfs_mount);
    VIR_FREE(qemu_driver->hugepage_path);
D
Daniel P. Berrange 已提交
764

765 766
    /* Free domain callback list */
    virDomainEventCallbackListFree(qemu_driver->domainEventCallbacks);
767 768 769 770
    virDomainEventQueueFree(qemu_driver->domainEventQueue);

    if (qemu_driver->domainEventTimer != -1)
        virEventRemoveTimeout(qemu_driver->domainEventTimer);
771

772 773 774
    if (qemu_driver->brctl)
        brShutdown(qemu_driver->brctl);

775 776
    virCgroupFree(&qemu_driver->cgroup);

777
    qemuDriverUnlock(qemu_driver);
778
    virMutexDestroy(&qemu_driver->lock);
779
    VIR_FREE(qemu_driver);
780 781

    return 0;
782 783 784
}

/* Return -1 for error, 1 to continue reading and 0 for success */
785
typedef int qemudHandlerMonitorOutput(virConnectPtr conn,
786
                                      virDomainObjPtr vm,
787 788 789
                                      const char *output,
                                      int fd);

790 791 792
/*
 * Returns -1 for error, 0 on end-of-file, 1 for success
 */
793
static int
794
qemudReadMonitorOutput(virConnectPtr conn,
795
                       virDomainObjPtr vm,
796 797
                       int fd,
                       char *buf,
G
Guido Günther 已提交
798
                       size_t buflen,
799
                       qemudHandlerMonitorOutput func,
800 801
                       const char *what,
                       int timeout)
802
{
G
Guido Günther 已提交
803
    size_t got = 0;
804
    buf[0] = '\0';
805
    timeout *= 1000; /* poll wants milli seconds */
806

807
    /* Consume & discard the initial greeting */
808
    while (got < (buflen-1)) {
G
Guido Günther 已提交
809
        ssize_t ret;
810 811

        ret = read(fd, buf+got, buflen-got-1);
812

813 814 815 816 817 818
        if (ret < 0) {
            struct pollfd pfd = { .fd = fd, .events = POLLIN };
            if (errno == EINTR)
                continue;

            if (errno != EAGAIN) {
819 820 821
                virReportSystemError(conn, errno,
                                     _("Failure while reading %s startup output"),
                                     what);
822 823 824
                return -1;
            }

825
            ret = poll(&pfd, 1, timeout);
826
            if (ret == 0) {
827
                qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
828
                                 _("Timed out while reading %s startup output"), what);
829 830 831
                return -1;
            } else if (ret == -1) {
                if (errno != EINTR) {
832 833 834
                    virReportSystemError(conn, errno,
                                         _("Failure while reading %s startup output"),
                                         what);
835 836 837 838 839 840 841 842
                    return -1;
                }
            } else {
                /* Make sure we continue loop & read any further data
                   available before dealing with EOF */
                if (pfd.revents & (POLLIN | POLLHUP))
                    continue;

843
                qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
844
                                 _("Failure while reading %s startup output"), what);
845 846
                return -1;
            }
847 848
        } else if (ret == 0) {
            return 0;
849 850 851
        } else {
            got += ret;
            buf[got] = '\0';
852 853 854 855 856 857
            ret = func(conn, vm, buf, fd);
            if (ret == -1)
                return -1;
            if (ret == 1)
                continue;
            return 1;
858 859 860
        }
    }

861
    qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
862
                     _("Out of space while reading %s startup output"), what);
863 864 865 866
    return -1;

}

867 868 869 870 871 872 873 874 875

/*
 * Returns -1 for error, 0 on success
 */
static int
qemudReadLogOutput(virConnectPtr conn,
                   virDomainObjPtr vm,
                   int fd,
                   char *buf,
G
Guido Günther 已提交
876
                   size_t buflen,
877 878 879 880
                   qemudHandlerMonitorOutput func,
                   const char *what,
                   int timeout)
{
881
    int retries = (timeout*10);
882
    int got = 0;
883 884 885
    buf[0] = '\0';

    while (retries) {
886
        ssize_t func_ret, ret;
887
        int isdead = 0;
G
Guido Günther 已提交
888

889 890
        func_ret = func(conn, vm, buf, fd);

891 892
        if (kill(vm->pid, 0) == -1 && errno == ESRCH)
            isdead = 1;
893

894 895
        /* Any failures should be detected before we read the log, so we
         * always have something useful to report on failure. */
896 897
        ret = saferead(fd, buf+got, buflen-got-1);
        if (ret < 0) {
898 899 900 901 902 903
            virReportSystemError(conn, errno,
                                 _("Failure while reading %s log output"),
                                 what);
            return -1;
        }

904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919
        got += ret;
        buf[got] = '\0';
        if (got == buflen-1) {
            qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                             _("Out of space while reading %s log output"),
                             what);
            return -1;
        }

        if (isdead) {
            qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                             _("Process exited while reading %s log output"),
                             what);
            return -1;
        }

920 921
        if (func_ret <= 0)
            return func_ret;
922 923 924 925 926 927 928 929 930 931

        usleep(100*1000);
        retries--;
    }
    if (retries == 0)
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                         _("Timed out while reading %s log output"), what);
    return -1;
}

932
static int
933
qemudCheckMonitorPrompt(virConnectPtr conn ATTRIBUTE_UNUSED,
934
                        virDomainObjPtr vm,
935 936 937 938 939 940 941 942 943 944 945
                        const char *output,
                        int fd)
{
    if (strstr(output, "(qemu) ") == NULL)
        return 1; /* keep reading */

    vm->monitor = fd;

    return 0;
}

946
static int
947 948 949 950 951
qemudOpenMonitorCommon(virConnectPtr conn,
                       struct qemud_driver* driver,
                       virDomainObjPtr vm,
                       int monfd,
                       int reconnect)
952
{
953
    char buf[1024];
954
    int ret;
955

956
    if (virSetCloseExec(monfd) < 0) {
957
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
958
                         "%s", _("Unable to set monitor close-on-exec flag"));
959
        return -1;
960
    }
961
    if (virSetNonBlock(monfd) < 0) {
962
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
963
                         "%s", _("Unable to put monitor into non-blocking mode"));
964
        return -1;
965 966
    }

967
    if (!reconnect) {
968 969 970 971
        if (qemudReadMonitorOutput(conn,
                                   vm, monfd,
                                   buf, sizeof(buf),
                                   qemudCheckMonitorPrompt,
972
                                   "monitor", 10) <= 0)
973 974 975
            ret = -1;
        else
            ret = 0;
976 977 978 979 980 981
    } else {
        vm->monitor = monfd;
        ret = 0;
    }

    if (ret != 0)
982
        return ret;
983

984 985
    if ((vm->monitorWatch = virEventAddHandle(vm->monitor,
                                              VIR_EVENT_HANDLE_HANGUP | VIR_EVENT_HANDLE_ERROR,
986 987
                                              qemudDispatchVMEvent,
                                              driver, NULL)) < 0)
988
        return -1;
989

990 991
    return 0;
}
992

993 994 995 996 997 998 999 1000 1001
static int
qemudOpenMonitorUnix(virConnectPtr conn,
                     struct qemud_driver* driver,
                     virDomainObjPtr vm,
                     const char *monitor,
                     int reconnect)
{
    struct sockaddr_un addr;
    int monfd;
1002
    int timeout = 3; /* In seconds */
1003
    int ret, i = 0;
1004 1005 1006 1007 1008 1009 1010 1011 1012

    if ((monfd = socket(AF_UNIX, SOCK_STREAM, 0)) < 0) {
        virReportSystemError(conn, errno,
                             "%s", _("failed to create socket"));
        return -1;
    }

    memset(&addr, 0, sizeof(addr));
    addr.sun_family = AF_UNIX;
C
Chris Lalancette 已提交
1013 1014 1015 1016 1017
    if (virStrcpyStatic(addr.sun_path, monitor) == NULL) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                         _("Monitor path %s too big for destination"), monitor);
        goto error;
    }
1018

1019 1020 1021 1022 1023 1024
    do {
        ret = connect(monfd, (struct sockaddr *) &addr, sizeof(addr));

        if (ret == 0)
            break;

1025 1026
        if (errno == ENOENT || errno == ECONNREFUSED) {
            /* ENOENT       : Socket may not have shown up yet
1027 1028 1029 1030
             * ECONNREFUSED : Leftover socket hasn't been removed yet */
            continue;
        }

1031 1032 1033
        virReportSystemError(conn, errno, "%s",
                             _("failed to connect to monitor socket"));
        goto error;
1034 1035 1036 1037 1038 1039 1040

    } while ((++i <= timeout*5) && (usleep(.2 * 1000000) <= 0));

    if (ret != 0) {
        virReportSystemError(conn, errno, "%s",
                             _("monitor socket did not show up."));
        goto error;
1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052
    }

    if (qemudOpenMonitorCommon(conn, driver, vm, monfd, reconnect) < 0)
        goto error;

    return 0;

error:
    close(monfd);
    return -1;
}

1053 1054 1055 1056 1057 1058 1059 1060
static int
qemudOpenMonitorPty(virConnectPtr conn,
                    struct qemud_driver* driver,
                    virDomainObjPtr vm,
                    const char *monitor,
                    int reconnect)
{
    int monfd;
1061

1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073
    if ((monfd = open(monitor, O_RDWR)) < 0) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                         _("Unable to open monitor path %s"), monitor);
        return -1;
    }

    if (qemudOpenMonitorCommon(conn, driver, vm, monfd, reconnect) < 0)
        goto error;

    return 0;

error:
1074
    close(monfd);
1075 1076 1077 1078 1079 1080 1081 1082 1083 1084
    return -1;
}

static int
qemudOpenMonitor(virConnectPtr conn,
                 struct qemud_driver *driver,
                 virDomainObjPtr vm,
                 int reconnect)
{
    switch (vm->monitor_chr->type) {
1085 1086 1087 1088
    case VIR_DOMAIN_CHR_TYPE_UNIX:
        return qemudOpenMonitorUnix(conn, driver, vm,
                                    vm->monitor_chr->data.nix.path,
                                    reconnect);
1089 1090 1091 1092 1093 1094 1095 1096 1097 1098
    case VIR_DOMAIN_CHR_TYPE_PTY:
        return qemudOpenMonitorPty(conn, driver, vm,
                                   vm->monitor_chr->data.file.path,
                                   reconnect);
    default:
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                         _("unable to handle monitor type: %s"),
                         virDomainChrTypeToString(vm->monitor_chr->type));
        return -1;
    }
1099 1100
}

1101
/* Returns -1 for error, 0 success, 1 continue reading */
1102 1103 1104 1105 1106 1107
static int
qemudExtractMonitorPath(virConnectPtr conn,
                        const char *haystack,
                        size_t *offset,
                        char **path)
{
1108
    static const char needle[] = "char device redirected to";
1109
    char *tmp, *dev;
1110

1111
    VIR_FREE(*path);
1112
    /* First look for our magic string */
1113 1114 1115 1116 1117
    if (!(tmp = strstr(haystack + *offset, needle))) {
        return 1;
    }
    tmp += sizeof(needle);
    dev = tmp;
1118

1119 1120 1121 1122 1123
    /*
     * And look for first whitespace character and nul terminate
     * to mark end of the pty path
     */
    while (*tmp) {
1124
        if (c_isspace(*tmp)) {
1125 1126
            *path = strndup(dev, tmp-dev);
            if (*path == NULL) {
1127
                virReportOOMError(conn);
1128 1129
                return -1;
            }
1130

1131
            /* ... now further update offset till we get EOL */
1132
            *offset = tmp - haystack;
1133 1134
            return 0;
        }
1135
        tmp++;
1136 1137 1138 1139 1140
    }

    /*
     * We found a path, but didn't find any whitespace,
     * so it must be still incomplete - we should at
1141 1142
     * least see a \n - indicate that we want to carry
     * on trying again
1143
     */
1144
    return 1;
1145 1146 1147
}

static int
1148
qemudFindCharDevicePTYs(virConnectPtr conn,
1149
                        virDomainObjPtr vm,
1150 1151
                        const char *output,
                        int fd ATTRIBUTE_UNUSED)
1152
{
1153
    size_t offset = 0;
1154
    int ret, i;
1155 1156

    /* The order in which QEMU prints out the PTY paths is
1157 1158
       the order in which it procsses its serial and parallel
       device args. This code must match that ordering.... */
1159

1160
    /* first comes the serial devices */
1161 1162
    for (i = 0 ; i < vm->def->nserials ; i++) {
        virDomainChrDefPtr chr = vm->def->serials[i];
1163 1164 1165
        if (chr->type == VIR_DOMAIN_CHR_TYPE_PTY) {
            if ((ret = qemudExtractMonitorPath(conn, output, &offset,
                                               &chr->data.file.path)) != 0)
1166
                return ret;
1167 1168 1169
        }
    }

1170
    /* then the parallel devices */
1171 1172
    for (i = 0 ; i < vm->def->nparallels ; i++) {
        virDomainChrDefPtr chr = vm->def->parallels[i];
1173 1174 1175
        if (chr->type == VIR_DOMAIN_CHR_TYPE_PTY) {
            if ((ret = qemudExtractMonitorPath(conn, output, &offset,
                                               &chr->data.file.path)) != 0)
1176
                return ret;
1177 1178 1179
        }
    }

1180
    return 0;
1181 1182
}

1183 1184 1185 1186
static int
qemudWaitForMonitor(virConnectPtr conn,
                    struct qemud_driver* driver,
                    virDomainObjPtr vm, off_t pos)
1187
{
1188
    char buf[4096]; /* Plenty of space to get startup greeting */
1189 1190 1191 1192 1193
    int logfd;
    int ret;

    if ((logfd = qemudLogReadFD(conn, driver->logDir, vm->def->name, pos))
        < 0)
1194
        return -1;
1195

1196 1197 1198
    ret = qemudReadLogOutput(conn, vm, logfd, buf, sizeof(buf),
                             qemudFindCharDevicePTYs,
                             "console", 3);
1199
    if (close(logfd) < 0) {
1200
        char ebuf[4096];
1201
        VIR_WARN(_("Unable to close logfile: %s\n"),
1202 1203
                 virStrerror(errno, ebuf, sizeof ebuf));
    }
1204

1205 1206 1207 1208 1209 1210
    if (ret < 0) {
        /* Unexpected end of file - inform user of QEMU log data */
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                         _("unable to start guest: %s"), buf);
        return -1;
    }
1211

1212 1213 1214 1215
    if (qemudOpenMonitor(conn, driver, vm, 0) < 0)
        return -1;

    return 0;
1216 1217
}

1218 1219
static int
qemudDetectVcpuPIDs(virConnectPtr conn,
1220
                    virDomainObjPtr vm) {
1221 1222 1223 1224 1225 1226
    char *qemucpus = NULL;
    char *line;
    int lastVcpu = -1;

    /* Only KVM has seperate threads for CPUs,
       others just use main QEMU process for CPU */
1227
    if (vm->def->virtType != VIR_DOMAIN_VIRT_KVM)
1228 1229 1230 1231 1232
        vm->nvcpupids = 1;
    else
        vm->nvcpupids = vm->def->vcpus;

    if (VIR_ALLOC_N(vm->vcpupids, vm->nvcpupids) < 0) {
1233
        virReportOOMError(conn);
1234 1235 1236
        return -1;
    }

1237
    if (vm->def->virtType != VIR_DOMAIN_VIRT_KVM) {
1238 1239 1240 1241
        vm->vcpupids[0] = vm->pid;
        return 0;
    }

1242
    if (qemudMonitorCommand(vm, "info cpus", &qemucpus) < 0) {
1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                         "%s", _("cannot run monitor command to fetch CPU thread info"));
        VIR_FREE(vm->vcpupids);
        vm->nvcpupids = 0;
        return -1;
    }

    /*
     * This is the gross format we're about to parse :-{
     *
     * (qemu) info cpus
     * * CPU #0: pc=0x00000000000f0c4a thread_id=30019
     *   CPU #1: pc=0x00000000fffffff0 thread_id=30020
     *   CPU #2: pc=0x00000000fffffff0 thread_id=30021
     *
     */
    line = qemucpus;
    do {
        char *offset = strchr(line, '#');
        char *end = NULL;
        int vcpu = 0, tid = 0;

        /* See if we're all done */
        if (offset == NULL)
            break;

        /* Extract VCPU number */
        if (virStrToLong_i(offset + 1, &end, 10, &vcpu) < 0)
            goto error;
        if (end == NULL || *end != ':')
            goto error;

        /* Extract host Thread ID */
        if ((offset = strstr(line, "thread_id=")) == NULL)
            goto error;
        if (virStrToLong_i(offset + strlen("thread_id="), &end, 10, &tid) < 0)
            goto error;
        if (end == NULL || !c_isspace(*end))
            goto error;

        /* Validate the VCPU is in expected range & order */
        if (vcpu > vm->nvcpupids ||
            vcpu != (lastVcpu + 1))
            goto error;

        lastVcpu = vcpu;
        vm->vcpupids[vcpu] = tid;

        /* Skip to next data line */
        line = strchr(offset, '\r');
        if (line == NULL)
            line = strchr(offset, '\n');
    } while (line != NULL);

    /* Validate we got data for all VCPUs we expected */
    if (lastVcpu != (vm->def->vcpus - 1))
        goto error;

1301
    VIR_FREE(qemucpus);
1302 1303 1304 1305
    return 0;

error:
    VIR_FREE(vm->vcpupids);
1306 1307
    vm->nvcpupids = 0;
    VIR_FREE(qemucpus);
1308 1309 1310 1311 1312 1313 1314 1315

    /* Explicitly return success, not error. Older KVM does
       not have vCPU -> Thread mapping info and we don't
       want to break its use. This merely disables ability
       to pin vCPUS with libvirt */
    return 0;
}

1316 1317
static int
qemudInitCpus(virConnectPtr conn,
D
Daniel Veillard 已提交
1318 1319
              virDomainObjPtr vm,
              const char *migrateFrom) {
1320 1321 1322 1323 1324
#if HAVE_SCHED_GETAFFINITY
    cpu_set_t mask;
    int i, maxcpu = QEMUD_CPUMASK_LEN;
    virNodeInfo nodeinfo;

1325
    if (nodeGetInfo(conn, &nodeinfo) < 0)
1326 1327 1328 1329 1330 1331 1332 1333
        return -1;

    /* setaffinity fails if you set bits for CPUs which
     * aren't present, so we have to limit ourselves */
    if (maxcpu > nodeinfo.cpus)
        maxcpu = nodeinfo.cpus;

    CPU_ZERO(&mask);
D
Daniel P. Berrange 已提交
1334 1335 1336 1337 1338 1339
    if (vm->def->cpumask) {
        for (i = 0 ; i < maxcpu ; i++)
            if (vm->def->cpumask[i])
                CPU_SET(i, &mask);
    } else {
        for (i = 0 ; i < maxcpu ; i++)
1340
            CPU_SET(i, &mask);
D
Daniel P. Berrange 已提交
1341
    }
1342 1343 1344 1345

    for (i = 0 ; i < vm->nvcpupids ; i++) {
        if (sched_setaffinity(vm->vcpupids[i],
                              sizeof(mask), &mask) < 0) {
1346 1347
            virReportSystemError(conn, errno, "%s",
                                 _("failed to set CPU affinity"));
1348 1349 1350 1351 1352
            return -1;
        }
    }
#endif /* HAVE_SCHED_GETAFFINITY */

D
Daniel Veillard 已提交
1353 1354
    if (migrateFrom == NULL) {
        /* Allow the CPUS to start executing */
1355 1356 1357 1358
        if (qemudMonitorSendCont(conn, vm) < 0) {
            if (virGetLastError() == NULL)
                qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                                 "%s", _("resume operation failed"));
D
Daniel Veillard 已提交
1359 1360
            return -1;
        }
1361 1362 1363 1364 1365 1366
    }

    return 0;
}


1367 1368 1369 1370 1371 1372
static int
qemudInitPasswords(virConnectPtr conn,
                   struct qemud_driver *driver,
                   virDomainObjPtr vm) {
    char *info = NULL;

1373 1374 1375
    if ((vm->def->ngraphics == 1) &&
        vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
        (vm->def->graphics[0]->data.vnc.passwd || driver->vncPassword)) {
1376 1377

        if (qemudMonitorCommandExtra(vm, "change vnc password",
1378 1379
                                     vm->def->graphics[0]->data.vnc.passwd ?
                                     vm->def->graphics[0]->data.vnc.passwd :
1380 1381
                                     driver->vncPassword,
                                     QEMU_PASSWD_PROMPT,
1382
                                     -1, &info) < 0) {
1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393
            qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                             "%s", _("setting VNC password failed"));
            return -1;
        }
        VIR_FREE(info);
    }

    return 0;
}


1394
static int qemudNextFreeVNCPort(struct qemud_driver *driver ATTRIBUTE_UNUSED) {
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
    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;
}

1430 1431 1432 1433 1434
static pciDeviceList *
qemuGetPciHostDeviceList(virConnectPtr conn,
                         virDomainDefPtr def)
{
    pciDeviceList *list;
1435 1436
    int i;

1437 1438
    if (!(list = pciDeviceListNew(conn)))
        return NULL;
1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453

    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;

        dev = pciGetDevice(conn,
                           hostdev->source.subsys.u.pci.domain,
                           hostdev->source.subsys.u.pci.bus,
                           hostdev->source.subsys.u.pci.slot,
                           hostdev->source.subsys.u.pci.function);
1454 1455 1456 1457
        if (!dev) {
            pciDeviceListFree(conn, list);
            return NULL;
        }
1458

1459
        if (pciDeviceListAdd(conn, list, dev) < 0) {
1460
            pciFreeDevice(conn, dev);
1461 1462
            pciDeviceListFree(conn, list);
            return NULL;
1463 1464
        }

1465
        pciDeviceSetManaged(dev, hostdev->managed);
1466 1467
    }

1468 1469 1470 1471
    return list;
}

static int
1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503
qemuUpdateActivePciHostdevs(struct qemud_driver *driver,
                            virDomainDefPtr def)
{
    pciDeviceList *pcidevs;
    int i, ret;

    if (!def->nhostdevs)
        return 0;

    if (!(pcidevs = qemuGetPciHostDeviceList(NULL, def)))
        return -1;

    ret = 0;

    for (i = 0; i < pcidevs->count; i++) {
        if (pciDeviceListAdd(NULL,
                             driver->activePciHostdevs,
                             pcidevs->devs[i]) < 0) {
            ret = -1;
            break;
        }
        pcidevs->devs[i] = NULL;
    }

    pciDeviceListFree(NULL, pcidevs);
    return ret;
}

static int
qemuPrepareHostDevices(virConnectPtr conn,
                       struct qemud_driver *driver,
                       virDomainDefPtr def)
1504 1505 1506 1507 1508 1509 1510 1511 1512 1513
{
    pciDeviceList *pcidevs;
    int i;

    if (!def->nhostdevs)
        return 0;

    if (!(pcidevs = qemuGetPciHostDeviceList(conn, def)))
        return -1;

1514
    /* We have to use 3 loops here. *All* devices must
1515 1516
     * be detached before we reset any of them, because
     * in some cases you have to reset the whole PCI,
1517 1518
     * which impacts all devices on it. Also, all devices
     * must be reset before being marked as active.
1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533
     */

    /* 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
     */

    for (i = 0; i < pcidevs->count; i++)
        if (pciDeviceGetManaged(pcidevs->devs[i]) &&
            pciDettachDevice(conn, pcidevs->devs[i]) < 0)
            goto error;

    /* Now that all the PCI hostdevs have be dettached, we can safely
     * reset them */
    for (i = 0; i < pcidevs->count; i++)
1534 1535
        if (pciResetDevice(conn, pcidevs->devs[i],
                           driver->activePciHostdevs) < 0)
1536 1537
            goto error;

1538 1539 1540 1541 1542 1543 1544 1545 1546
    /* Now mark all the devices as active */
    for (i = 0; i < pcidevs->count; i++) {
        if (pciDeviceListAdd(conn,
                             driver->activePciHostdevs,
                             pcidevs->devs[i]) < 0)
            goto error;
        pcidevs->devs[i] = NULL;
    }

1547
    pciDeviceListFree(conn, pcidevs);
1548 1549 1550
    return 0;

error:
1551
    pciDeviceListFree(conn, pcidevs);
1552 1553 1554
    return -1;
}

1555
static void
1556 1557 1558
qemuDomainReAttachHostDevices(virConnectPtr conn,
                              struct qemud_driver *driver,
                              virDomainDefPtr def)
1559
{
1560
    pciDeviceList *pcidevs;
1561 1562
    int i;

1563 1564
    if (!def->nhostdevs)
        return;
1565

1566 1567 1568 1569 1570 1571
    if (!(pcidevs = qemuGetPciHostDeviceList(conn, def))) {
        virErrorPtr err = virGetLastError();
        VIR_ERROR(_("Failed to allocate pciDeviceList: %s\n"),
                  err ? err->message : "");
        virResetError(err);
        return;
1572 1573
    }

1574 1575
    /* Again 3 loops; mark all devices as inactive before reset
     * them and reset all the devices before re-attach */
1576

1577
    for (i = 0; i < pcidevs->count; i++)
1578 1579 1580 1581 1582
        pciDeviceListDel(conn, driver->activePciHostdevs, pcidevs->devs[i]);

    for (i = 0; i < pcidevs->count; i++)
        if (pciResetDevice(conn, pcidevs->devs[i],
                           driver->activePciHostdevs) < 0) {
1583
            virErrorPtr err = virGetLastError();
1584
            VIR_ERROR(_("Failed to reset PCI device: %s\n"),
1585 1586 1587 1588
                      err ? err->message : "");
            virResetError(err);
        }

1589 1590 1591
    for (i = 0; i < pcidevs->count; i++)
        if (pciDeviceGetManaged(pcidevs->devs[i]) &&
            pciReAttachDevice(conn, pcidevs->devs[i]) < 0) {
1592
            virErrorPtr err = virGetLastError();
1593
            VIR_ERROR(_("Failed to re-attach PCI device: %s\n"),
1594 1595 1596 1597
                      err ? err->message : "");
            virResetError(err);
        }

1598
    pciDeviceListFree(conn, pcidevs);
1599 1600
}

1601 1602 1603 1604 1605 1606 1607 1608 1609 1610
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

1611 1612 1613 1614 1615 1616
static int qemuSetupCgroup(virConnectPtr conn,
                           struct qemud_driver *driver,
                           virDomainObjPtr vm)
{
    virCgroupPtr cgroup = NULL;
    int rc;
1617
    unsigned int i;
1618 1619 1620 1621
    const char *const *deviceACL =
        driver->cgroupDeviceACL ?
        (const char *const *)driver->cgroupDeviceACL :
        defaultDeviceACL;
1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633

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

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

1634 1635
    if (qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_DEVICES)) {
        rc = virCgroupDenyAllDevices(cgroup);
1636
        if (rc != 0) {
1637 1638 1639 1640 1641
            if (rc == -EPERM) {
                VIR_WARN0("Group devices ACL is not accessible, disabling whitelisting");
                goto done;
            }

1642
            virReportSystemError(conn, -rc,
1643
                                 _("Unable to deny all devices for %s"), vm->def->name);
1644 1645 1646
            goto cleanup;
        }

1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660
        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) {
                virReportSystemError(conn, -rc,
                                     _("Unable to allow device %s for %s"),
                                     vm->def->disks[i]->src, vm->def->name);
                goto cleanup;
            }
        }
1661

1662
        rc = virCgroupAllowDeviceMajor(cgroup, 'c', DEVICE_PTY_MAJOR);
1663 1664
        if (rc != 0) {
            virReportSystemError(conn, -rc, "%s",
1665
                                 _("unable to allow /dev/pts/ devices"));
1666 1667 1668
            goto cleanup;
        }

1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687
        if (vm->def->nsounds) {
            rc = virCgroupAllowDeviceMajor(cgroup, 'c', DEVICE_SND_MAJOR);
            if (rc != 0) {
                virReportSystemError(conn, -rc, "%s",
                                     _("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) {
                virReportSystemError(conn, -rc,
                                     _("unable to allow device %s"),
                                     deviceACL[i]);
                goto cleanup;
            }
1688 1689 1690 1691
        }
    }

done:
1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760
    virCgroupFree(&cgroup);
    return 0;

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


static int qemuRemoveCgroup(virConnectPtr conn,
                            struct qemud_driver *driver,
                            virDomainObjPtr vm)
{
    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) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                         _("Unable to find cgroup for %s\n"),
                         vm->def->name);
        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) {
        virReportSystemError(NULL, -rc,
                             _("unable to find cgroup for domain %s"),
                             def->name);
        goto cleanup;
    }

    rc = virCgroupAddTask(cgroup, getpid());
    if (rc != 0) {
        virReportSystemError(NULL, -rc,
                             _("unable to add domain %s task %d to cgroup"),
                             def->name, getpid());
        goto cleanup;
    }

    ret = 0;

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


1761 1762 1763 1764 1765 1766 1767 1768 1769
static int qemudDomainSetSecurityLabel(virConnectPtr conn, struct qemud_driver *driver, virDomainObjPtr vm)
{
    if (vm->def->seclabel.label != NULL)
        if (driver->securityDriver && driver->securityDriver->domainSetSecurityLabel)
            return driver->securityDriver->domainSetSecurityLabel(conn, driver->securityDriver,
                                                                 vm);
    return 0;
}

1770 1771

#ifdef __linux__
1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782
struct qemuFileOwner {
    uid_t uid;
    gid_t gid;
};

static int qemuDomainSetHostdevUSBOwnershipActor(virConnectPtr conn,
                                                 usbDevice *dev ATTRIBUTE_UNUSED,
                                                 const char *file, void *opaque)
{
    struct qemuFileOwner *owner = opaque;

1783 1784
    VIR_DEBUG("Setting ownership on %s to %d:%d", file, owner->uid, owner->gid);

1785 1786 1787 1788 1789 1790 1791 1792
    if (chown(file, owner->uid, owner->gid) < 0) {
        virReportSystemError(conn, errno, _("cannot set ownership on %s"), file);
        return -1;
    }

    return 0;
}

1793 1794 1795 1796
static int qemuDomainSetHostdevUSBOwnership(virConnectPtr conn,
                                            virDomainHostdevDefPtr def,
                                            uid_t uid, gid_t gid)
{
1797 1798
    struct qemuFileOwner owner = { uid, gid };
    int ret = -1;
1799 1800 1801 1802 1803 1804

    /* XXX what todo for USB devs assigned based on product/vendor ? Doom :-( */
    if (!def->source.subsys.u.usb.bus ||
        !def->source.subsys.u.usb.device)
        return 0;

1805 1806 1807 1808 1809 1810 1811 1812 1813
    usbDevice *dev = usbGetDevice(conn,
                                  def->source.subsys.u.usb.bus,
                                  def->source.subsys.u.usb.device);

    if (!dev)
        goto cleanup;

    ret = usbDeviceFileIterate(conn, dev,
                               qemuDomainSetHostdevUSBOwnershipActor, &owner);
1814

1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825
    usbFreeDevice(conn, dev);
cleanup:
    return ret;
}

static int qemuDomainSetHostdevPCIOwnershipActor(virConnectPtr conn,
                                                 pciDevice *dev ATTRIBUTE_UNUSED,
                                                 const char *file, void *opaque)
{
    struct qemuFileOwner *owner = opaque;

1826 1827
    VIR_DEBUG("Setting ownership on %s to %d:%d", file, owner->uid, owner->gid);

1828 1829
    if (chown(file, owner->uid, owner->gid) < 0) {
        virReportSystemError(conn, errno, _("cannot set ownership on %s"), file);
1830 1831 1832 1833 1834 1835 1836 1837 1838 1839
        return -1;
    }

    return 0;
}

static int qemuDomainSetHostdevPCIOwnership(virConnectPtr conn,
                                            virDomainHostdevDefPtr def,
                                            uid_t uid, gid_t gid)
{
1840
    struct qemuFileOwner owner = { uid, gid };
1841 1842
    int ret = -1;

1843 1844 1845 1846 1847
    pciDevice *dev = pciGetDevice(conn,
                                  def->source.subsys.u.pci.domain,
                                  def->source.subsys.u.pci.bus,
                                  def->source.subsys.u.pci.slot,
                                  def->source.subsys.u.pci.function);
1848

1849
    if (!dev)
1850 1851
        goto cleanup;

1852 1853
    ret = pciDeviceFileIterate(conn, dev,
                               qemuDomainSetHostdevPCIOwnershipActor, &owner);
1854

1855
    pciFreeDevice(conn, dev);
1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886
cleanup:
    return ret;
}
#endif


static int qemuDomainSetHostdevOwnership(virConnectPtr conn,
                                         virDomainHostdevDefPtr def,
                                         uid_t uid, gid_t gid)
{
    if (def->mode != VIR_DOMAIN_HOSTDEV_MODE_SUBSYS)
        return 0;

#ifdef __linux__
    switch (def->source.subsys.type) {
    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB:
        return qemuDomainSetHostdevUSBOwnership(conn, def, uid, gid);

    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI:
        return qemuDomainSetHostdevPCIOwnership(conn, def, uid, gid);

    }
    return 0;
#else
    qemudReportError(conn, NULL, NULL, "%s",
                     _("unable to set host device ownership on this platform"));
    return -1;
#endif

}

1887 1888
static int qemuDomainSetFileOwnership(virConnectPtr conn,
                                      const char *path,
1889 1890 1891
                                      uid_t uid, gid_t gid)
{

1892
    if (!path)
1893 1894
        return 0;

1895 1896
    VIR_DEBUG("Setting ownership on %s to %d:%d", path, uid, gid);
    if (chown(path, uid, gid) < 0) {
1897
        virReportSystemError(conn, errno, _("cannot set ownership on %s"),
1898
                             path);
1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927
        return -1;
    }
    return 0;
}

static int qemuDomainSetDeviceOwnership(virConnectPtr conn,
                                        struct qemud_driver *driver,
                                        virDomainDeviceDefPtr def,
                                        int restore)
{
    uid_t uid;
    gid_t gid;

    if (!driver->privileged)
        return 0;

    /* short circuit case of root:root */
    if (!driver->user && !driver->group)
        return 0;

    uid = restore ? 0 : driver->user;
    gid = restore ? 0 : driver->group;

    switch (def->type) {
    case VIR_DOMAIN_DEVICE_DISK:
        if (restore &&
            (def->data.disk->readonly || def->data.disk->shared))
            return 0;

1928
        return qemuDomainSetFileOwnership(conn, def->data.disk->src, uid, gid);
1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955

    case VIR_DOMAIN_DEVICE_HOSTDEV:
        return qemuDomainSetHostdevOwnership(conn, def->data.hostdev, uid, gid);
    }

    return 0;
}

static int qemuDomainSetAllDeviceOwnership(virConnectPtr conn,
                                           struct qemud_driver *driver,
                                           virDomainDefPtr def,
                                           int restore)
{
    int i;
    uid_t uid;
    gid_t gid;

    if (!driver->privileged)
        return 0;

    /* short circuit case of root:root */
    if (!driver->user && !driver->group)
        return 0;

    uid = restore ? 0 : driver->user;
    gid = restore ? 0 : driver->group;

1956 1957 1958 1959
    if (qemuDomainSetFileOwnership(conn, def->os.kernel, uid, gid) < 0 ||
        qemuDomainSetFileOwnership(conn, def->os.initrd, uid, gid) < 0)
        return -1;

1960 1961 1962 1963 1964
    for (i = 0 ; i < def->ndisks ; i++) {
        if (restore &&
            (def->disks[i]->readonly || def->disks[i]->shared))
            continue;

1965
        if (qemuDomainSetFileOwnership(conn, def->disks[i]->src, uid, gid) < 0)
1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976
            return -1;
    }

    for (i = 0 ; i < def->nhostdevs ; i++) {
        if (qemuDomainSetHostdevOwnership(conn, def->hostdevs[i], uid, gid) < 0)
            return -1;
    }

    return 0;
}

1977 1978 1979
static virDomainPtr qemudDomainLookupByName(virConnectPtr conn,
                                            const char *name);

1980 1981 1982 1983
struct qemudHookData {
    virConnectPtr conn;
    virDomainObjPtr vm;
    struct qemud_driver *driver;
1984 1985 1986
};

static int qemudSecurityHook(void *data) {
1987 1988 1989 1990
    struct qemudHookData *h = data;

    if (qemuAddToCgroup(h->driver, h->vm->def) < 0)
        return -1;
1991

1992
    if (qemudDomainSetSecurityLabel(h->conn, h->driver, h->vm) < 0)
1993 1994 1995 1996 1997
        return -1;

    if (h->driver->privileged) {
        if (qemuDomainSetAllDeviceOwnership(h->conn, h->driver, h->vm->def, 0) < 0)
            return -1;
1998

1999 2000
        DEBUG("Dropping privileges of VM to %d:%d", h->driver->user, h->driver->group);

2001 2002 2003 2004 2005
        if (h->driver->group) {
            if (setregid(h->driver->group, h->driver->group) < 0) {
                virReportSystemError(NULL, errno,
                                     _("cannot change to '%d' group"),
                                     h->driver->group);
2006
                return -1;
2007
            }
2008
        }
2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019
        if (h->driver->user) {
            if (setreuid(h->driver->user, h->driver->user) < 0) {
                virReportSystemError(NULL, errno,
                                     _("cannot change to '%d' user"),
                                     h->driver->user);
                return -1;
            }
        }
    }

    return 0;
2020 2021
}

2022 2023 2024 2025 2026 2027 2028 2029 2030 2031
static int
qemuPrepareMonitorChr(virConnectPtr conn,
                      struct qemud_driver *driver,
                      virDomainChrDefPtr monitor_chr,
                      const char *vm)
{
    monitor_chr->type = VIR_DOMAIN_CHR_TYPE_UNIX;
    monitor_chr->data.nix.listen = 1;

    if (virAsprintf(&monitor_chr->data.nix.path, "%s/%s.monitor",
2032
                    driver->libDir, vm) < 0) {
2033 2034 2035 2036 2037 2038 2039
        virReportOOMError(conn);
        return -1;
    }

    return 0;
}

2040 2041
static int qemudStartVMDaemon(virConnectPtr conn,
                              struct qemud_driver *driver,
2042
                              virDomainObjPtr vm,
2043 2044
                              const char *migrateFrom,
                              int stdin_fd) {
2045
    const char **argv = NULL, **tmp;
2046
    const char **progenv = NULL;
2047
    int i, ret;
2048
    struct stat sb;
2049 2050
    int *tapfds = NULL;
    int ntapfds = 0;
2051
    unsigned int qemuCmdFlags;
2052
    fd_set keepfd;
2053
    const char *emulator;
G
Guido Günther 已提交
2054
    pid_t child;
2055
    int pos = -1;
2056
    char ebuf[1024];
2057
    char *pidfile = NULL;
2058
    int logfile;
2059

2060
    struct qemudHookData hookData;
2061 2062 2063 2064
    hookData.conn = conn;
    hookData.vm = vm;
    hookData.driver = driver;

2065
    FD_ZERO(&keepfd);
2066

2067
    if (virDomainIsActive(vm)) {
2068
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_INVALID,
2069
                         "%s", _("VM is already active"));
2070 2071 2072
        return -1;
    }

2073 2074 2075 2076 2077 2078 2079 2080
    /* If you are using a SecurityDriver with dynamic labelling,
       then generate a security label for isolation */
    if (vm->def->seclabel.type == VIR_DOMAIN_SECLABEL_DYNAMIC &&
        driver->securityDriver &&
        driver->securityDriver->domainGenSecurityLabel &&
        driver->securityDriver->domainGenSecurityLabel(conn, vm) < 0)
        return -1;

2081 2082 2083
    /* Ensure no historical cgroup for this VM is lieing around bogus settings */
    qemuRemoveCgroup(conn, driver, vm);

2084 2085 2086
    if ((vm->def->ngraphics == 1) &&
        vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
        vm->def->graphics[0]->data.vnc.autoport) {
2087
        int port = qemudNextFreeVNCPort(driver);
2088
        if (port < 0) {
2089
            qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
2090
                             "%s", _("Unable to find an unused VNC port"));
2091
            goto cleanup;
2092
        }
2093
        vm->def->graphics[0]->data.vnc.port = port;
2094
    }
2095

2096
    if (virFileMakePath(driver->logDir) < 0) {
2097 2098 2099
        virReportSystemError(conn, errno,
                             _("cannot create log directory %s"),
                             driver->logDir);
2100
        goto cleanup;
2101 2102
    }

2103
    if ((logfile = qemudLogFD(conn, driver, vm->def->name)) < 0)
2104
        goto cleanup;
2105

2106 2107
    emulator = vm->def->emulator;

2108 2109 2110 2111
    /* 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
     */
2112
    if (stat(emulator, &sb) < 0) {
2113 2114 2115
        virReportSystemError(conn, errno,
                             _("Cannot find QEMU binary %s"),
                             emulator);
2116
        goto cleanup;
2117 2118
    }

2119
    if (qemudExtractVersionInfo(emulator,
2120
                                NULL,
2121
                                &qemuCmdFlags) < 0)
2122
        goto cleanup;
2123

2124 2125 2126
    if (qemuSetupCgroup(conn, driver, vm) < 0)
        goto cleanup;

2127
    if (qemuPrepareHostDevices(conn, driver, vm->def) < 0)
2128
        goto cleanup;
2129

2130 2131 2132 2133 2134
    if (VIR_ALLOC(vm->monitor_chr) < 0) {
        virReportOOMError(conn);
        goto cleanup;
    }

2135 2136
    if (qemuPrepareMonitorChr(conn, driver, vm->monitor_chr, vm->def->name) < 0)
        goto cleanup;
2137

D
Daniel P. Berrange 已提交
2138 2139 2140 2141 2142 2143 2144
    if ((ret = virFileDeletePid(driver->stateDir, vm->def->name)) != 0) {
        virReportSystemError(conn, ret,
                             _("Cannot remove stale PID file for %s"),
                             vm->def->name);
        goto cleanup;
    }

2145 2146 2147 2148 2149
    if (!(pidfile = virFilePid(driver->stateDir, vm->def->name))) {
        virReportSystemError(conn, errno,
                             "%s", _("Failed to build pidfile path."));
        goto cleanup;
    }
D
Daniel P. Berrange 已提交
2150

2151
    vm->def->id = driver->nextvmid++;
2152
    if (qemudBuildCommandLine(conn, driver, vm->def, vm->monitor_chr,
2153
                              qemuCmdFlags, &argv, &progenv,
2154 2155
                              &tapfds, &ntapfds, migrateFrom) < 0)
        goto cleanup;
2156

2157 2158
    tmp = progenv;
    while (*tmp) {
2159
        if (safewrite(logfile, *tmp, strlen(*tmp)) < 0)
2160
            VIR_WARN(_("Unable to write envv to logfile: %s\n"),
2161
                     virStrerror(errno, ebuf, sizeof ebuf));
2162
        if (safewrite(logfile, " ", 1) < 0)
2163
            VIR_WARN(_("Unable to write envv to logfile: %s\n"),
2164
                     virStrerror(errno, ebuf, sizeof ebuf));
2165 2166
        tmp++;
    }
2167 2168
    tmp = argv;
    while (*tmp) {
2169
        if (safewrite(logfile, *tmp, strlen(*tmp)) < 0)
2170
            VIR_WARN(_("Unable to write argv to logfile: %s\n"),
2171
                     virStrerror(errno, ebuf, sizeof ebuf));
2172
        if (safewrite(logfile, " ", 1) < 0)
2173
            VIR_WARN(_("Unable to write argv to logfile: %s\n"),
2174
                     virStrerror(errno, ebuf, sizeof ebuf));
2175 2176
        tmp++;
    }
2177
    if (safewrite(logfile, "\n", 1) < 0)
2178
        VIR_WARN(_("Unable to write argv to logfile: %s\n"),
2179
                 virStrerror(errno, ebuf, sizeof ebuf));
2180

2181
    if ((pos = lseek(logfile, 0, SEEK_END)) < 0)
2182
        VIR_WARN(_("Unable to seek to end of logfile: %s\n"),
2183
                 virStrerror(errno, ebuf, sizeof ebuf));
2184

2185 2186 2187
    for (i = 0 ; i < ntapfds ; i++)
        FD_SET(tapfds[i], &keepfd);

2188
    ret = virExecDaemonize(conn, argv, progenv, &keepfd, &child,
2189
                           stdin_fd, &logfile, &logfile,
2190
                           VIR_EXEC_NONBLOCK | VIR_EXEC_CLEAR_CAPS,
2191 2192 2193
                           qemudSecurityHook, &hookData,
                           pidfile);
    VIR_FREE(pidfile);
G
Guido Günther 已提交
2194 2195 2196

    /* wait for qemu process to to show up */
    if (ret == 0) {
2197
        if (virFileReadPid(driver->stateDir, vm->def->name, &vm->pid)) {
2198
            qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
2199
                             _("Domain %s didn't show up\n"), vm->def->name);
2200
            ret = -1;
G
Guido Günther 已提交
2201
        }
2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212
    } 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;
    }
2213 2214

    vm->state = migrateFrom ? VIR_DOMAIN_PAUSED : VIR_DOMAIN_RUNNING;
2215

2216
    for (i = 0 ; argv[i] ; i++)
2217 2218
        VIR_FREE(argv[i]);
    VIR_FREE(argv);
2219

2220 2221 2222 2223
    for (i = 0 ; progenv[i] ; i++)
        VIR_FREE(progenv[i]);
    VIR_FREE(progenv);

2224 2225 2226
    if (tapfds) {
        for (i = 0 ; i < ntapfds ; i++) {
            close(tapfds[i]);
2227
        }
2228
        VIR_FREE(tapfds);
2229 2230
    }

2231 2232 2233 2234 2235 2236 2237 2238
    if (ret == -1)
        goto cleanup;

    if ((qemudWaitForMonitor(conn, driver, vm, pos) < 0) ||
        (qemudDetectVcpuPIDs(conn, vm) < 0) ||
        (qemudInitCpus(conn, vm, migrateFrom) < 0) ||
        (qemudInitPasswords(conn, driver, vm) < 0) ||
        (qemudDomainSetMemoryBalloon(conn, vm, vm->def->memory) < 0) ||
2239
        (virDomainSaveStatus(conn, driver->stateDir, vm) < 0)) {
2240 2241 2242
        qemudShutdownVMDaemon(conn, driver, vm);
        ret = -1;
        /* No need for 'goto cleanup' now since qemudShutdownVMDaemon does enough */
2243 2244
    }

2245 2246 2247
    if (logfile != -1)
        close(logfile);

2248
    return ret;
2249 2250 2251 2252 2253 2254 2255

cleanup:
    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);
    }
2256
    qemuRemoveCgroup(conn, driver, vm);
2257 2258 2259 2260
    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;
2261 2262
    if (logfile != -1)
        close(logfile);
2263 2264
    vm->def->id = -1;
    return -1;
2265 2266 2267
}


2268
static void qemudShutdownVMDaemon(virConnectPtr conn,
2269 2270
                                  struct qemud_driver *driver,
                                  virDomainObjPtr vm) {
D
Daniel P. Berrange 已提交
2271
    int ret;
2272
    int retries = 0;
D
Daniel P. Berrange 已提交
2273

2274
    if (!virDomainIsActive(vm))
2275
        return;
2276

2277
    VIR_DEBUG(_("Shutting down VM '%s'\n"), vm->def->name);
2278

G
Guido Günther 已提交
2279 2280
    if (virKillProcess(vm->pid, 0) == 0 &&
        virKillProcess(vm->pid, SIGTERM) < 0)
2281 2282 2283
        virReportSystemError(conn, errno,
                             _("Failed to send SIGTERM to %s (%d)"),
                             vm->def->name, vm->pid);
2284

2285 2286 2287
    if (vm->monitorWatch != -1) {
        virEventRemoveHandle(vm->monitorWatch);
        vm->monitorWatch = -1;
2288
    }
2289 2290 2291 2292 2293

    if (vm->monitor != -1)
        close(vm->monitor);
    vm->monitor = -1;

2294 2295 2296 2297 2298 2299 2300
    if (vm->monitor_chr) {
        if (vm->monitor_chr->type == VIR_DOMAIN_CHR_TYPE_UNIX)
            unlink(vm->monitor_chr->data.nix.path);
        virDomainChrDefFree(vm->monitor_chr);
        vm->monitor_chr = NULL;
    }

G
Guido Günther 已提交
2301 2302
    /* shut it off for sure */
    virKillProcess(vm->pid, SIGKILL);
2303

2304 2305 2306 2307
    /* Reset Security Labels */
    if (driver->securityDriver)
        driver->securityDriver->domainRestoreSecurityLabel(conn, vm);

2308 2309 2310 2311 2312 2313 2314
    /* 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);
    }

2315 2316 2317 2318
    if (qemuDomainSetAllDeviceOwnership(conn, driver, vm->def, 1) < 0)
        VIR_WARN("Failed to restore all device ownership for %s",
                 vm->def->name);

2319
    qemuDomainReAttachHostDevices(conn, driver, vm->def);
2320

2321 2322 2323 2324 2325 2326 2327 2328 2329 2330
retry:
    if ((ret = qemuRemoveCgroup(conn, driver, vm)) < 0) {
        if (ret == -EBUSY && (retries++ < 5)) {
            usleep(200*1000);
            goto retry;
        }
        VIR_WARN("Failed to remove cgroup for %s",
                 vm->def->name);
    }

2331
    if (qemudRemoveDomainStatus(conn, driver, vm) < 0) {
2332
        VIR_WARN(_("Failed to remove domain status for %s"),
2333 2334
                 vm->def->name);
    }
D
Daniel P. Berrange 已提交
2335 2336 2337 2338 2339 2340
    if ((ret = virFileDeletePid(driver->stateDir, vm->def->name)) != 0) {
        char ebuf[1024];
        VIR_WARN(_("Failed to remove PID file for %s: %s"),
                 vm->def->name, virStrerror(errno, ebuf, sizeof ebuf));
    }

2341
    vm->pid = -1;
2342
    vm->def->id = -1;
2343
    vm->state = VIR_DOMAIN_SHUTOFF;
2344
    VIR_FREE(vm->vcpupids);
2345
    vm->nvcpupids = 0;
2346 2347

    if (vm->newDef) {
2348
        virDomainDefFree(vm->def);
2349
        vm->def = vm->newDef;
2350
        vm->def->id = -1;
2351 2352 2353 2354 2355
        vm->newDef = NULL;
    }
}


2356
static void
2357
qemudDispatchVMEvent(int watch, int fd, int events, void *opaque) {
2358
    struct qemud_driver *driver = opaque;
2359
    virDomainObjPtr vm = NULL;
2360
    virDomainEventPtr event = NULL;
2361
    unsigned int i;
2362
    int quit = 0, failed = 0;
2363

2364
    qemuDriverLock(driver);
2365
    for (i = 0 ; i < driver->domains.count ; i++) {
2366 2367 2368
        virDomainObjPtr tmpvm = driver->domains.objs[i];
        virDomainObjLock(tmpvm);
        if (virDomainIsActive(tmpvm) &&
2369
            tmpvm->monitorWatch == watch) {
2370
            vm = tmpvm;
2371
            break;
2372
        }
2373
        virDomainObjUnlock(tmpvm);
2374 2375 2376
    }

    if (!vm)
2377
        goto cleanup;
2378

2379
    if (vm->monitor != fd) {
2380 2381
        failed = 1;
    } else {
2382
        if (events & (VIR_EVENT_HANDLE_HANGUP | VIR_EVENT_HANDLE_ERROR))
2383
            quit = 1;
2384
        else {
2385 2386
            VIR_ERROR(_("unhandled fd event %d for %s"),
                      events, vm->def->name);
2387
            failed = 1;
2388
        }
2389 2390
    }

2391
    if (failed || quit) {
2392 2393 2394 2395 2396
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         quit ?
                                         VIR_DOMAIN_EVENT_STOPPED_SHUTDOWN :
                                         VIR_DOMAIN_EVENT_STOPPED_FAILED);
2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407
        qemudShutdownVMDaemon(NULL, driver, vm);
        if (!vm->persistent) {
            virDomainRemoveInactive(&driver->domains,
                                    vm);
            vm = NULL;
        }
    }

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
2408 2409
    if (event)
        qemuDomainEventQueue(driver, event);
2410
    qemuDriverUnlock(driver);
2411 2412
}

2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433

/* Throw away any data available on the monitor
 * This is done before executing a command, in order
 * to allow re-synchronization if something went badly
 * wrong in the past. it also deals with problem of
 * QEMU *sometimes* re-printing its initial greeting
 * when we reconnect to the monitor after restarts.
 */
static void
qemuMonitorDiscardPendingData(virDomainObjPtr vm) {
    char buf[1024];
    int ret = 0;

    /* Monitor is non-blocking, so just loop till we
     * get -1 or 0. Don't bother with detecting
     * errors, since we'll deal with that better later */
    do {
        ret = read(vm->monitor, buf, sizeof (buf)-1);
    } while (ret > 0);
}

2434 2435 2436
static int
qemudMonitorSendUnix(const virDomainObjPtr vm,
                     const char *cmd,
2437 2438
                     size_t cmdlen,
                     int scm_fd)
2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451
{
    struct msghdr msg;
    struct iovec iov[1];
    ssize_t ret;

    memset(&msg, 0, sizeof(msg));

    iov[0].iov_base = (void *)cmd;
    iov[0].iov_len = cmdlen;

    msg.msg_iov = iov;
    msg.msg_iovlen = 1;

2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465
    if (scm_fd != -1) {
        char control[CMSG_SPACE(sizeof(int))];
        struct cmsghdr *cmsg;

        msg.msg_control = control;
        msg.msg_controllen = sizeof(control);

        cmsg = CMSG_FIRSTHDR(&msg);
        cmsg->cmsg_len = CMSG_LEN(sizeof(int));
        cmsg->cmsg_level = SOL_SOCKET;
        cmsg->cmsg_type = SCM_RIGHTS;
        memcpy(CMSG_DATA(cmsg), &scm_fd, sizeof(int));
    }

2466 2467 2468 2469 2470 2471 2472
    do {
        ret = sendmsg(vm->monitor, &msg, 0);
    } while (ret < 0 && errno == EINTR);

    return ret == cmdlen ? 0 : -1;
}

2473 2474
static int
qemudMonitorSend(const virDomainObjPtr vm,
2475 2476
                 const char *cmd,
                 int scm_fd)
2477 2478 2479 2480 2481 2482 2483 2484 2485 2486
{
    char *full;
    size_t len;
    int ret = -1;

    if (virAsprintf(&full, "%s\r", cmd) < 0)
        return -1;

    len = strlen(full);

2487 2488
    switch (vm->monitor_chr->type) {
    case VIR_DOMAIN_CHR_TYPE_UNIX:
2489
        if (qemudMonitorSendUnix(vm, full, len, scm_fd) < 0)
2490 2491 2492 2493 2494 2495 2496 2497
            goto out;
        break;
    default:
    case VIR_DOMAIN_CHR_TYPE_PTY:
        if (safewrite(vm->monitor, full, len) != len)
            goto out;
        break;
    }
2498 2499 2500 2501 2502 2503

    ret = 0;
out:
    VIR_FREE(full);
    return ret;
}
2504

2505
static int
2506 2507 2508 2509 2510 2511 2512
qemudMonitorCommandWithHandler(const virDomainObjPtr vm,
                               const char *cmd,
                               const char *extraPrompt,
                               qemudMonitorExtraPromptHandler extraHandler,
                               void *handlerData,
                               int scm_fd,
                               char **reply) {
D
Daniel P. Berrange 已提交
2513 2514
    int size = 0;
    char *buf = NULL;
D
Daniel P. Berrange 已提交
2515

2516 2517 2518 2519 2520
    /* Should never happen, but just in case, protect
     * against null monitor (ocurrs when VM is inactive) */
    if (!vm->monitor_chr)
        return -1;

2521 2522
    qemuMonitorDiscardPendingData(vm);

2523
    VIR_DEBUG("Send '%s'", cmd);
2524
    if (qemudMonitorSend(vm, cmd, scm_fd) < 0)
D
Daniel P. Berrange 已提交
2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536
        return -1;

    *reply = NULL;

    for (;;) {
        struct pollfd fd = { vm->monitor, POLLIN | POLLERR | POLLHUP, 0 };
        char *tmp;

        /* Read all the data QEMU has sent thus far */
        for (;;) {
            char data[1024];
            int got = read(vm->monitor, data, sizeof(data));
D
Daniel P. Berrange 已提交
2537

2538 2539
            if (got == 0)
                goto error;
D
Daniel P. Berrange 已提交
2540 2541 2542 2543 2544
            if (got < 0) {
                if (errno == EINTR)
                    continue;
                if (errno == EAGAIN)
                    break;
2545
                goto error;
2546
            }
2547
            if (VIR_REALLOC_N(buf, size+got+1) < 0)
2548 2549
                goto error;

D
Daniel P. Berrange 已提交
2550 2551 2552 2553
            memmove(buf+size, data, got);
            buf[size+got] = '\0';
            size += got;
        }
2554

D
Daniel P. Berrange 已提交
2555
        /* Look for QEMU prompt to indicate completion */
2556
        if (buf) {
2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570
            char *foundPrompt;

            if (extraPrompt &&
                (foundPrompt = strstr(buf, extraPrompt)) != NULL) {
                char *promptEnd;

                if (extraHandler(vm, buf, foundPrompt, handlerData) < 0)
                    return -1;
                /* Discard output so far, necessary to detect whether
                   extraPrompt appears again.  We don't need the output between
                   original command and this prompt anyway. */
                promptEnd = foundPrompt + strlen(extraPrompt);
                memmove(buf, promptEnd, strlen(promptEnd)+1);
                size -= promptEnd - buf;
2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582
            } else if ((tmp = strstr(buf, QEMU_CMD_PROMPT)) != NULL) {
                char *commptr = NULL, *nlptr = NULL;
                /* Preserve the newline */
                tmp[1] = '\0';

                /* The monitor doesn't dump clean output after we have written to
                 * it. Every character we write dumps a bunch of useless stuff,
                 * so the result looks like "cXcoXcomXcommXcommaXcommanXcommand"
                 * Try to throw away everything before the first full command
                 * occurence, and inbetween the command and the newline starting
                 * the response
                 */
2583
                if ((commptr = strstr(buf, cmd))) {
2584
                    memmove(buf, commptr, strlen(commptr)+1);
2585 2586 2587
                    if ((nlptr = strchr(buf, '\n')))
                        memmove(buf+strlen(cmd), nlptr, strlen(nlptr)+1);
                }
C
Cole Robinson 已提交
2588

2589 2590
                break;
            }
D
Daniel P. Berrange 已提交
2591 2592 2593 2594 2595 2596
        }
    pollagain:
        /* Need to wait for more data */
        if (poll(&fd, 1, -1) < 0) {
            if (errno == EINTR)
                goto pollagain;
2597
            goto error;
D
Daniel P. Berrange 已提交
2598 2599 2600 2601
        }
    }
    *reply = buf;
    return 0;
2602 2603

 error:
2604
    VIR_FREE(buf);
2605
    return -1;
D
Daniel P. Berrange 已提交
2606 2607
}

2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645
struct extraHandlerData
{
    const char *reply;
    bool first;
};

static int
qemudMonitorCommandSimpleExtraHandler(const virDomainObjPtr vm,
                                      const char *buf ATTRIBUTE_UNUSED,
                                      const char *prompt ATTRIBUTE_UNUSED,
                                      void *data_)
{
    struct extraHandlerData *data = data_;

    if (!data->first)
        return 0;
    if (qemudMonitorSend(vm, data->reply, -1) < 0)
        return -1;
    data->first = false;
    return 0;
}

static int
qemudMonitorCommandExtra(const virDomainObjPtr vm,
                         const char *cmd,
                         const char *extra,
                         const char *extraPrompt,
                         int scm_fd,
                         char **reply) {
    struct extraHandlerData data;

    data.reply = extra;
    data.first = true;
    return qemudMonitorCommandWithHandler(vm, cmd, extraPrompt,
                                          qemudMonitorCommandSimpleExtraHandler,
                                          &data, scm_fd, reply);
}

2646 2647 2648 2649 2650 2651 2652 2653
static int
qemudMonitorCommandWithFd(const virDomainObjPtr vm,
                          const char *cmd,
                          int scm_fd,
                          char **reply) {
    return qemudMonitorCommandExtra(vm, cmd, NULL, NULL, scm_fd, reply);
}

2654 2655 2656 2657
static int
qemudMonitorCommand(const virDomainObjPtr vm,
                    const char *cmd,
                    char **reply) {
2658
    return qemudMonitorCommandWithFd(vm, cmd, -1, reply);
2659 2660
}

2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699
static virStorageEncryptionPtr
findDomainDiskEncryption(virConnectPtr conn, virDomainObjPtr vm,
                         const char *path)
{
    bool seen_volume;
    int i;

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

        disk = vm->def->disks[i];
        if (disk->src != NULL && STREQ(disk->src, path)) {
            seen_volume = true;
            if (disk->encryption != NULL)
                return disk->encryption;
        }
    }
    if (seen_volume)
        qemudReportError(conn, NULL, NULL, VIR_ERR_INVALID_DOMAIN,
                         _("missing <encryption> for volume %s"), path);
    else
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                         _("unexpected passphrase request for volume %s"),
                         path);
    return NULL;
}

static char *
findVolumeQcowPassphrase(virConnectPtr conn, virDomainObjPtr vm,
                         const char *path, size_t *passphrase_len)
{
    virStorageEncryptionPtr enc;
    virSecretPtr secret;
    char *passphrase;
    unsigned char *data;
    size_t size;

    if (conn->secretDriver == NULL ||
2700
        conn->secretDriver->lookupByUUID == NULL ||
2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719
        conn->secretDriver->getValue == NULL) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_NO_SUPPORT, "%s",
                         _("secret storage not supported"));
        return NULL;
    }

    enc = findDomainDiskEncryption(conn, vm, path);
    if (enc == NULL)
        return NULL;

    if (enc->format != VIR_STORAGE_ENCRYPTION_FORMAT_QCOW ||
        enc->nsecrets != 1 ||
        enc->secrets[0]->type !=
        VIR_STORAGE_ENCRYPTION_SECRET_TYPE_PASSPHRASE) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_INVALID_DOMAIN,
                         _("invalid <encryption> for volume %s"), path);
        return NULL;
    }

2720 2721
    secret = conn->secretDriver->lookupByUUID(conn,
                                              enc->secrets[0]->uuid);
2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794
    if (secret == NULL)
        return NULL;
    data = conn->secretDriver->getValue(secret, &size,
                                        VIR_SECRET_GET_VALUE_INTERNAL_CALL);
    virUnrefSecret(secret);
    if (data == NULL)
        return NULL;

    if (memchr(data, '\0', size) != NULL) {
        memset(data, 0, size);
        VIR_FREE(data);
        qemudReportError(conn, NULL, NULL, VIR_ERR_INVALID_SECRET,
                         _("format='qcow' passphrase for %s must not contain a "
                           "'\\0'"), path);
        return NULL;
    }

    if (VIR_ALLOC_N(passphrase, size + 1) < 0) {
        memset(data, 0, size);
        VIR_FREE(data);
        virReportOOMError(conn);
        return NULL;
    }
    memcpy(passphrase, data, size);
    passphrase[size] = '\0';

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

    *passphrase_len = size;
    return passphrase;
}

static int
qemudMonitorSendVolumePassphrase(const virDomainObjPtr vm,
                                 const char *buf,
                                 const char *prompt,
                                 void *data)
{
    virConnectPtr conn = data;
    char *passphrase, *path;
    const char *prompt_path;
    size_t path_len, passphrase_len = 0;
    int res;

    /* The complete prompt looks like this:
           ide0-hd0 (/path/to/volume) is encrypted.
           Password:
       "prompt" starts with ") is encrypted".  Extract /path/to/volume. */
    for (prompt_path = prompt; prompt_path > buf && prompt_path[-1] != '(';
         prompt_path--)
        ;
    if (prompt_path == buf)
        return -1;
    path_len = prompt - prompt_path;
    if (VIR_ALLOC_N(path, path_len + 1) < 0)
        return -1;
    memcpy(path, prompt_path, path_len);
    path[path_len] = '\0';

    passphrase = findVolumeQcowPassphrase(conn, vm, path, &passphrase_len);
    VIR_FREE(path);
    if (passphrase == NULL)
        return -1;

    res = qemudMonitorSend(vm, passphrase, -1);

    memset(passphrase, 0, passphrase_len);
    VIR_FREE(passphrase);

    return res;
}

2795
static int
2796
qemudMonitorSendCont(virConnectPtr conn,
2797 2798 2799
                     const virDomainObjPtr vm) {
    char *reply;

2800 2801 2802
    if (qemudMonitorCommandWithHandler(vm, "cont", ") is encrypted.",
                                       qemudMonitorSendVolumePassphrase, conn,
                                       -1, &reply) < 0)
2803 2804 2805 2806 2807 2808
        return -1;
    qemudDebug ("%s: cont reply: %s", vm->def->name, info);
    VIR_FREE(reply);
    return 0;
}

2809
static virDrvOpenStatus qemudOpen(virConnectPtr conn,
2810
                                  virConnectAuthPtr auth ATTRIBUTE_UNUSED,
2811
                                  int flags ATTRIBUTE_UNUSED) {
2812
    if (conn->uri == NULL) {
2813 2814 2815
        if (qemu_driver == NULL)
            return VIR_DRV_OPEN_DECLINED;

2816
        conn->uri = xmlParseURI(qemu_driver->privileged ?
2817 2818
                                "qemu:///system" :
                                "qemu:///session");
2819
        if (!conn->uri) {
2820
            virReportOOMError(conn);
2821 2822
            return VIR_DRV_OPEN_ERROR;
        }
2823 2824 2825 2826 2827 2828 2829 2830 2831 2832
    } 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;

2833 2834 2835 2836 2837 2838
        if (qemu_driver == NULL) {
            qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR, "%s",
                             _("qemu state driver is not active"));
            return VIR_DRV_OPEN_ERROR;
        }

2839
        if (qemu_driver->privileged) {
2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854
            if (STRNEQ (conn->uri->path, "/system") &&
                STRNEQ (conn->uri->path, "/session")) {
                qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                                 _("unexpected QEMU URI path '%s', try qemu:///system"),
                                 conn->uri->path);
                return VIR_DRV_OPEN_ERROR;
            }
        } else {
            if (STRNEQ (conn->uri->path, "/session")) {
                qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                                 _("unexpected QEMU URI path '%s', try qemu:///session"),
                                 conn->uri->path);
                return VIR_DRV_OPEN_ERROR;
            }
        }
2855 2856 2857 2858 2859 2860 2861
    }
    conn->privateData = qemu_driver;

    return VIR_DRV_OPEN_SUCCESS;
}

static int qemudClose(virConnectPtr conn) {
2862
    struct qemud_driver *driver = conn->privateData;
2863 2864

    /* Get rid of callbacks registered for this conn */
2865
    qemuDriverLock(driver);
2866
    virDomainEventCallbackListRemoveConn(conn, driver->domainEventCallbacks);
2867
    qemuDriverUnlock(driver);
2868 2869 2870 2871 2872 2873

    conn->privateData = NULL;

    return 0;
}

D
Daniel Veillard 已提交
2874 2875 2876 2877 2878 2879 2880 2881 2882 2883
/* Which features are supported by this driver? */
static int
qemudSupportsFeature (virConnectPtr conn ATTRIBUTE_UNUSED, int feature)
{
    switch (feature) {
    case VIR_DRV_FEATURE_MIGRATION_V2: return 1;
    default: return 0;
    }
}

2884
static const char *qemudGetType(virConnectPtr conn ATTRIBUTE_UNUSED) {
2885
    return "QEMU";
2886 2887
}

2888 2889 2890 2891 2892

static int kvmGetMaxVCPUs(void) {
    int maxvcpus = 1;

    int r, fd;
2893

2894 2895
    fd = open(KVM_DEVICE, O_RDONLY);
    if (fd < 0) {
2896 2897
        virReportSystemError(NULL, errno, _("Unable to open %s"), KVM_DEVICE);
        return -1;
2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908
    }

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

    close(fd);
    return maxvcpus;
}


2909
static int qemudGetMaxVCPUs(virConnectPtr conn, const char *type) {
2910 2911 2912
    if (!type)
        return 16;

2913
    if (STRCASEEQ(type, "qemu"))
2914 2915
        return 16;

2916
    if (STRCASEEQ(type, "kvm"))
2917
        return kvmGetMaxVCPUs();
2918

2919
    if (STRCASEEQ(type, "kqemu"))
2920
        return 1;
2921 2922 2923

    qemudReportError(conn, NULL, NULL, VIR_ERR_INVALID_ARG,
                     _("unknown type '%s'"), type);
2924 2925 2926
    return -1;
}

2927

2928
static char *qemudGetCapabilities(virConnectPtr conn) {
2929
    struct qemud_driver *driver = conn->privateData;
2930
    virCapsPtr caps;
2931
    char *xml = NULL;
2932

2933
    qemuDriverLock(driver);
2934
    if ((caps = qemudCapsInit(qemu_driver->caps)) == NULL) {
2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945
        virReportOOMError(conn);
        goto cleanup;
    }

    if (qemu_driver->securityDriver &&
        qemudSecurityCapsInit(qemu_driver->securityDriver, caps) < 0) {
        virCapabilitiesFree(caps);
        virReportOOMError(conn);
        goto cleanup;
    }

2946
    virCapabilitiesFree(qemu_driver->caps);
2947 2948 2949
    qemu_driver->caps = caps;

    if ((xml = virCapabilitiesFormatXML(driver->caps)) == NULL)
2950
        virReportOOMError(conn);
2951 2952

cleanup:
2953
    qemuDriverUnlock(driver);
2954

2955
    return xml;
2956 2957 2958
}


2959
static int qemudGetProcessInfo(unsigned long long *cpuTime, int *lastCpu, int pid, int tid) {
D
Daniel P. Berrange 已提交
2960 2961
    char proc[PATH_MAX];
    FILE *pidinfo;
2962
    unsigned long long usertime, systime;
2963 2964
    int cpu;
    int ret;
D
Daniel P. Berrange 已提交
2965

2966 2967 2968 2969 2970 2971
    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
2972 2973 2974 2975
        return -1;
    }

    if (!(pidinfo = fopen(proc, "r"))) {
2976
        /*printf("cannot read pid info");*/
D
Daniel P. Berrange 已提交
2977
        /* VM probably shut down, so fake 0 */
2978 2979 2980 2981
        if (cpuTime)
            *cpuTime = 0;
        if (lastCpu)
            *lastCpu = 0;
D
Daniel P. Berrange 已提交
2982 2983 2984
        return 0;
    }

2985 2986 2987 2988 2989 2990 2991 2992 2993 2994
    /* 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) {
2995
        fclose(pidinfo);
2996 2997
        VIR_WARN0("cannot parse process status data");
        errno = -EINVAL;
D
Daniel P. Berrange 已提交
2998 2999 3000 3001 3002 3003 3004 3005
        return -1;
    }

    /* We got jiffies
     * We want nanoseconds
     * _SC_CLK_TCK is jiffies per second
     * So calulate thus....
     */
3006 3007 3008 3009 3010
    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
3011

3012 3013
    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
3014 3015 3016 3017 3018 3019 3020

    fclose(pidinfo);

    return 0;
}


3021
static virDomainPtr qemudDomainLookupByID(virConnectPtr conn,
3022
                                          int id) {
3023 3024 3025 3026
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;

3027
    qemuDriverLock(driver);
3028
    vm  = virDomainFindByID(&driver->domains, id);
3029
    qemuDriverUnlock(driver);
3030 3031

    if (!vm) {
3032 3033
        qemudReportError(conn, NULL, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching id %d"), id);
3034
        goto cleanup;
3035 3036
    }

3037
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
3038
    if (dom) dom->id = vm->def->id;
3039 3040

cleanup:
3041 3042
    if (vm)
        virDomainObjUnlock(vm);
3043 3044
    return dom;
}
3045

3046
static virDomainPtr qemudDomainLookupByUUID(virConnectPtr conn,
3047
                                            const unsigned char *uuid) {
3048 3049 3050
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
3051

3052
    qemuDriverLock(driver);
3053
    vm = virDomainFindByUUID(&driver->domains, uuid);
3054 3055
    qemuDriverUnlock(driver);

3056
    if (!vm) {
3057 3058 3059
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(uuid, uuidstr);
        qemudReportError(conn, NULL, NULL, VIR_ERR_NO_DOMAIN,
3060
                         _("no domain with matching uuid '%s'"), uuidstr);
3061
        goto cleanup;
3062 3063
    }

3064
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
3065
    if (dom) dom->id = vm->def->id;
3066 3067

cleanup:
3068 3069
    if (vm)
        virDomainObjUnlock(vm);
3070 3071
    return dom;
}
3072

3073
static virDomainPtr qemudDomainLookupByName(virConnectPtr conn,
3074
                                            const char *name) {
3075 3076 3077
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
3078

3079
    qemuDriverLock(driver);
3080
    vm = virDomainFindByName(&driver->domains, name);
3081 3082
    qemuDriverUnlock(driver);

3083
    if (!vm) {
3084 3085
        qemudReportError(conn, NULL, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching name '%s'"), name);
3086
        goto cleanup;
3087 3088
    }

3089
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
3090
    if (dom) dom->id = vm->def->id;
3091 3092

cleanup:
3093 3094
    if (vm)
        virDomainObjUnlock(vm);
3095 3096 3097
    return dom;
}

3098
static int qemudGetVersion(virConnectPtr conn, unsigned long *version) {
3099 3100 3101
    struct qemud_driver *driver = conn->privateData;
    int ret = -1;

3102
    qemuDriverLock(driver);
3103
    if (qemudExtractVersion(conn, driver) < 0)
3104
        goto cleanup;
3105

3106
    *version = qemu_driver->qemuVersion;
3107 3108 3109
    ret = 0;

cleanup:
3110
    qemuDriverUnlock(driver);
3111
    return ret;
D
Daniel P. Berrange 已提交
3112 3113
}

3114 3115 3116
static char *
qemudGetHostname (virConnectPtr conn)
{
3117
    char *result;
3118

3119 3120
    result = virGetHostname();
    if (result == NULL) {
3121 3122
        virReportSystemError (conn, errno,
                              "%s", _("failed to determine host name"));
3123 3124 3125
        return NULL;
    }
    /* Caller frees this string. */
3126
    return result;
3127 3128
}

3129
static int qemudListDomains(virConnectPtr conn, int *ids, int nids) {
3130
    struct qemud_driver *driver = conn->privateData;
3131 3132
    int got = 0, i;

3133 3134 3135
    qemuDriverLock(driver);
    for (i = 0 ; i < driver->domains.count && got < nids ; i++) {
        virDomainObjLock(driver->domains.objs[i]);
3136 3137
        if (virDomainIsActive(driver->domains.objs[i]))
            ids[got++] = driver->domains.objs[i]->def->id;
3138 3139 3140
        virDomainObjUnlock(driver->domains.objs[i]);
    }
    qemuDriverUnlock(driver);
3141

D
Daniel P. Berrange 已提交
3142 3143
    return got;
}
3144

3145
static int qemudNumDomains(virConnectPtr conn) {
3146
    struct qemud_driver *driver = conn->privateData;
3147 3148
    int n = 0, i;

3149 3150 3151
    qemuDriverLock(driver);
    for (i = 0 ; i < driver->domains.count ; i++) {
        virDomainObjLock(driver->domains.objs[i]);
3152
        if (virDomainIsActive(driver->domains.objs[i]))
3153
            n++;
3154 3155 3156
        virDomainObjUnlock(driver->domains.objs[i]);
    }
    qemuDriverUnlock(driver);
3157

3158
    return n;
D
Daniel P. Berrange 已提交
3159
}
3160

3161
static virDomainPtr qemudDomainCreate(virConnectPtr conn, const char *xml,
3162
                                      unsigned int flags ATTRIBUTE_UNUSED) {
3163
    struct qemud_driver *driver = conn->privateData;
3164
    virDomainDefPtr def;
3165
    virDomainObjPtr vm = NULL;
3166
    virDomainPtr dom = NULL;
3167
    virDomainEventPtr event = NULL;
D
Daniel P. Berrange 已提交
3168

3169
    qemuDriverLock(driver);
3170 3171
    if (!(def = virDomainDefParseString(conn, driver->caps, xml,
                                        VIR_DOMAIN_XML_INACTIVE)))
3172
        goto cleanup;
3173

3174 3175 3176
    if (virSecurityDriverVerify(conn, def) < 0)
        goto cleanup;

3177
    /* See if a VM with matching UUID already exists */
3178
    vm = virDomainFindByUUID(&driver->domains, def->uuid);
3179
    if (vm) {
3180 3181 3182 3183 3184 3185 3186 3187 3188
        /* UUID matches, but if names don't match, refuse it */
        if (STRNEQ(vm->def->name, def->name)) {
            char uuidstr[VIR_UUID_STRING_BUFLEN];
            virUUIDFormat(vm->def->uuid, uuidstr);
            qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                             _("domain '%s' is already defined with uuid %s"),
                             vm->def->name, uuidstr);
            goto cleanup;
        }
3189

3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207
        /* UUID & name match, but if VM is already active, refuse it */
        if (virDomainIsActive(vm)) {
            qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                             _("domain is already active as '%s'"), vm->def->name);
            goto cleanup;
        }
        virDomainObjUnlock(vm);
    } else {
        /* UUID does not match, but if a name matches, refuse it */
        vm = virDomainFindByName(&driver->domains, def->name);
        if (vm) {
            char uuidstr[VIR_UUID_STRING_BUFLEN];
            virUUIDFormat(vm->def->uuid, uuidstr);
            qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                             _("domain '%s' is already defined with uuid %s"),
                             def->name, uuidstr);
            goto cleanup;
        }
3208
    }
3209

3210 3211
    if (!(vm = virDomainAssignDef(conn,
                                  &driver->domains,
3212 3213 3214 3215
                                  def)))
        goto cleanup;

    def = NULL;
D
Daniel P. Berrange 已提交
3216

3217
    if (qemudStartVMDaemon(conn, driver, vm, NULL, -1) < 0) {
3218 3219
        virDomainRemoveInactive(&driver->domains,
                                vm);
3220
        vm = NULL;
3221
        goto cleanup;
D
Daniel P. Berrange 已提交
3222
    }
3223 3224 3225 3226

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_BOOTED);
D
Daniel P. Berrange 已提交
3227

3228
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
3229
    if (dom) dom->id = vm->def->id;
3230 3231 3232

cleanup:
    virDomainDefFree(def);
3233 3234
    if (vm)
        virDomainObjUnlock(vm);
3235 3236
    if (event)
        qemuDomainEventQueue(driver, event);
3237
    qemuDriverUnlock(driver);
3238
    return dom;
D
Daniel P. Berrange 已提交
3239 3240 3241
}


3242
static int qemudDomainSuspend(virDomainPtr dom) {
3243
    struct qemud_driver *driver = dom->conn->privateData;
D
Daniel P. Berrange 已提交
3244
    char *info;
3245 3246
    virDomainObjPtr vm;
    int ret = -1;
3247
    virDomainEventPtr event = NULL;
3248

3249
    qemuDriverLock(driver);
3250
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3251

D
Daniel P. Berrange 已提交
3252
    if (!vm) {
3253 3254 3255 3256
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching uuid '%s'"), uuidstr);
3257
        goto cleanup;
D
Daniel P. Berrange 已提交
3258
    }
3259
    if (!virDomainIsActive(vm)) {
3260
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
3261
                         "%s", _("domain is not running"));
3262
        goto cleanup;
D
Daniel P. Berrange 已提交
3263
    }
3264
    if (vm->state != VIR_DOMAIN_PAUSED) {
3265
        if (qemudMonitorCommand(vm, "stop", &info) < 0) {
3266 3267 3268 3269 3270 3271
            qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                             "%s", _("suspend operation failed"));
            goto cleanup;
        }
        vm->state = VIR_DOMAIN_PAUSED;
        qemudDebug("Reply %s", info);
3272 3273 3274
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_SUSPENDED,
                                         VIR_DOMAIN_EVENT_SUSPENDED_PAUSED);
3275
        VIR_FREE(info);
D
Daniel P. Berrange 已提交
3276
    }
3277
    if (virDomainSaveStatus(dom->conn, driver->stateDir, vm) < 0)
3278
        goto cleanup;
3279 3280 3281
    ret = 0;

cleanup:
3282 3283
    if (vm)
        virDomainObjUnlock(vm);
3284

3285
    if (event)
3286
        qemuDomainEventQueue(driver, event);
3287
    qemuDriverUnlock(driver);
3288
    return ret;
D
Daniel P. Berrange 已提交
3289 3290 3291
}


3292
static int qemudDomainResume(virDomainPtr dom) {
3293 3294 3295
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
3296
    virDomainEventPtr event = NULL;
3297

3298
    qemuDriverLock(driver);
3299
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3300

D
Daniel P. Berrange 已提交
3301
    if (!vm) {
3302 3303 3304 3305
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching uuid '%s'"), uuidstr);
3306
        goto cleanup;
D
Daniel P. Berrange 已提交
3307
    }
3308
    if (!virDomainIsActive(vm)) {
3309
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
3310
                         "%s", _("domain is not running"));
3311
        goto cleanup;
D
Daniel P. Berrange 已提交
3312
    }
3313
    if (vm->state == VIR_DOMAIN_PAUSED) {
3314 3315 3316 3317
        if (qemudMonitorSendCont(dom->conn, vm) < 0) {
            if (virGetLastError() == NULL)
                qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                                 "%s", _("resume operation failed"));
3318 3319 3320
            goto cleanup;
        }
        vm->state = VIR_DOMAIN_RUNNING;
3321 3322 3323
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_UNPAUSED);
D
Daniel P. Berrange 已提交
3324
    }
3325
    if (virDomainSaveStatus(dom->conn, driver->stateDir, vm) < 0)
3326
        goto cleanup;
3327 3328 3329
    ret = 0;

cleanup:
3330 3331
    if (vm)
        virDomainObjUnlock(vm);
3332
    if (event)
3333
        qemuDomainEventQueue(driver, event);
3334
    qemuDriverUnlock(driver);
3335
    return ret;
D
Daniel P. Berrange 已提交
3336 3337 3338
}


3339
static int qemudDomainShutdown(virDomainPtr dom) {
3340 3341
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
3342
    char* info;
3343
    int ret = -1;
3344

3345
    qemuDriverLock(driver);
3346
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3347 3348
    qemuDriverUnlock(driver);

3349
    if (!vm) {
3350 3351 3352 3353
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching uuid '%s'"), uuidstr);
3354
        goto cleanup;
3355 3356
    }

3357 3358 3359 3360 3361 3362
    if (!virDomainIsActive(vm)) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
                         "%s", _("domain is not running"));
        goto cleanup;
    }

3363
    if (qemudMonitorCommand(vm, "system_powerdown", &info) < 0) {
3364
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
3365
                         "%s", _("shutdown operation failed"));
3366
        goto cleanup;
3367
    }
D
Daniel Veillard 已提交
3368
    VIR_FREE(info);
3369 3370 3371
    ret = 0;

cleanup:
3372 3373
    if (vm)
        virDomainObjUnlock(vm);
3374
    return ret;
3375 3376 3377
}


3378
static int qemudDomainDestroy(virDomainPtr dom) {
3379 3380 3381
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
3382
    virDomainEventPtr event = NULL;
3383

3384
    qemuDriverLock(driver);
3385
    vm  = virDomainFindByUUID(&driver->domains, dom->uuid);
D
Daniel P. Berrange 已提交
3386
    if (!vm) {
3387 3388 3389 3390
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching uuid '%s'"), uuidstr);
3391
        goto cleanup;
D
Daniel P. Berrange 已提交
3392
    }
3393
    if (!virDomainIsActive(vm)) {
3394
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
3395 3396 3397
                         "%s", _("domain is not running"));
        goto cleanup;
    }
3398

3399
    qemudShutdownVMDaemon(dom->conn, driver, vm);
3400 3401 3402
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_DESTROYED);
3403
    if (!vm->persistent) {
3404 3405
        virDomainRemoveInactive(&driver->domains,
                                vm);
3406 3407
        vm = NULL;
    }
3408 3409 3410
    ret = 0;

cleanup:
3411 3412
    if (vm)
        virDomainObjUnlock(vm);
3413 3414
    if (event)
        qemuDomainEventQueue(driver, event);
3415
    qemuDriverUnlock(driver);
3416
    return ret;
D
Daniel P. Berrange 已提交
3417 3418 3419
}


3420
static char *qemudDomainGetOSType(virDomainPtr dom) {
3421 3422 3423
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    char *type = NULL;
3424

3425
    qemuDriverLock(driver);
3426
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3427
    qemuDriverUnlock(driver);
3428
    if (!vm) {
3429 3430 3431 3432
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching uuid '%s'"), uuidstr);
3433
        goto cleanup;
3434 3435
    }

3436
    if (!(type = strdup(vm->def->os.type)))
3437
        virReportOOMError(dom->conn);
3438 3439

cleanup:
3440 3441
    if (vm)
        virDomainObjUnlock(vm);
3442 3443 3444
    return type;
}

3445 3446
/* Returns max memory in kb, 0 if error */
static unsigned long qemudDomainGetMaxMemory(virDomainPtr dom) {
3447 3448 3449
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    unsigned long ret = 0;
3450

3451
    qemuDriverLock(driver);
3452
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3453 3454
    qemuDriverUnlock(driver);

3455
    if (!vm) {
3456 3457
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
3458
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
3459
                         _("no domain with matching uuid '%s'"), uuidstr);
3460
        goto cleanup;
3461 3462
    }

3463 3464 3465
    ret = vm->def->maxmem;

cleanup:
3466 3467
    if (vm)
        virDomainObjUnlock(vm);
3468
    return ret;
3469 3470 3471
}

static int qemudDomainSetMaxMemory(virDomainPtr dom, unsigned long newmax) {
3472 3473 3474
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
3475

3476
    qemuDriverLock(driver);
3477
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3478 3479
    qemuDriverUnlock(driver);

3480
    if (!vm) {
3481 3482
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
3483
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
3484
                         _("no domain with matching uuid '%s'"), uuidstr);
3485
        goto cleanup;
3486 3487 3488 3489
    }

    if (newmax < vm->def->memory) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
3490
                         "%s", _("cannot set max memory lower than current memory"));
3491
        goto cleanup;;
3492 3493 3494
    }

    vm->def->maxmem = newmax;
3495 3496 3497
    ret = 0;

cleanup:
3498 3499
    if (vm)
        virDomainObjUnlock(vm);
3500
    return ret;
3501 3502
}

3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517

/* The reply from QEMU contains 'ballon: actual=421' where value is in MB */
#define BALLOON_PREFIX "balloon: actual="

/*
 * Returns: 0 if balloon not supported, +1 if balloon query worked
 * or -1 on failure
 */
static int qemudDomainGetMemoryBalloon(virConnectPtr conn,
                                       virDomainObjPtr vm,
                                       unsigned long *currmem) {
    char *reply = NULL;
    int ret = -1;
    char *offset;

3518 3519 3520
    if (!virDomainIsActive(vm))
        return 0;

3521 3522 3523 3524 3525 3526
    if (qemudMonitorCommand(vm, "info balloon", &reply) < 0) {
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                         "%s", _("could not query memory balloon allocation"));
        goto cleanup;
    }

3527
    DEBUG ("%s: balloon reply: '%s'", vm->def->name, reply);
3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580
    if ((offset = strstr(reply, BALLOON_PREFIX)) != NULL) {
        unsigned int memMB;
        char *end;
        offset += strlen(BALLOON_PREFIX);
        if (virStrToLong_ui(offset, &end, 10, &memMB) < 0) {
            qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                             "%s", _("could not parse memory balloon allocation"));
            goto cleanup;
        }
        *currmem = memMB * 1024;
        ret = 1;
    } else {
        /* We don't raise an error here, since its to be expected that
         * many QEMU's don't support ballooning
         */
        ret = 0;
    }

cleanup:
    VIR_FREE(reply);
    return ret;
}

/*
 * Returns: 0 if balloon not supported, +1 if balloon query worked
 * or -1 on failure
 */
static int qemudDomainSetMemoryBalloon(virConnectPtr conn,
                                       virDomainObjPtr vm,
                                       unsigned long newmem) {
    char *cmd;
    char *reply = NULL;
    int ret = -1;

    /*
     * 'newmem' is in KB, QEMU monitor works in MB, and we all wish
     * we just worked in bytes with unsigned long long everywhere.
     */
    if (virAsprintf(&cmd, "balloon %lu", (newmem / 1024)) < 0) {
        virReportOOMError(conn);
        goto cleanup;
    }

    if (qemudMonitorCommand(vm, cmd, &reply) < 0) {
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                         "%s", _("could not balloon memory allocation"));
        VIR_FREE(cmd);
        goto cleanup;
    }
    VIR_FREE(cmd);

    /* If the command failed qemu prints: 'unknown command'
     * No message is printed on success it seems */
3581
    DEBUG ("%s: balloon reply: %s",vm->def->name,  reply);
3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594
    if (strstr(reply, "\nunknown command:")) {
        /* Don't set error - it is expected memory balloon fails on many qemu */
        ret = 0;
    } else {
        ret = 1;
    }

cleanup:
    VIR_FREE(reply);
    return ret;
}


3595
static int qemudDomainSetMemory(virDomainPtr dom, unsigned long newmem) {
3596 3597 3598
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
3599

3600
    qemuDriverLock(driver);
3601
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3602
    qemuDriverUnlock(driver);
3603
    if (!vm) {
3604 3605
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
3606
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
3607
                         _("no domain with matching uuid '%s'"), uuidstr);
3608
        goto cleanup;
3609 3610 3611 3612
    }

    if (newmem > vm->def->maxmem) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
3613
                         "%s", _("cannot set memory higher than max memory"));
3614
        goto cleanup;
3615 3616
    }

3617 3618 3619 3620 3621 3622 3623 3624 3625
    if (virDomainIsActive(vm)) {
        ret = qemudDomainSetMemoryBalloon(dom->conn, vm, newmem);
        if (ret == 0)
            qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                             "%s", _("cannot set memory of an active domain"));
    } else {
        vm->def->memory = newmem;
        ret = 0;
    }
3626 3627

cleanup:
3628 3629
    if (vm)
        virDomainObjUnlock(vm);
3630
    return ret;
3631 3632
}

3633
static int qemudDomainGetInfo(virDomainPtr dom,
3634
                              virDomainInfoPtr info) {
3635 3636 3637
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
3638 3639
    int err;
    unsigned long balloon;
3640

3641
    qemuDriverLock(driver);
3642
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3643
    qemuDriverUnlock(driver);
D
Daniel P. Berrange 已提交
3644
    if (!vm) {
3645 3646 3647 3648
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching uuid '%s'"), uuidstr);
3649
        goto cleanup;
D
Daniel P. Berrange 已提交
3650 3651
    }

3652
    info->state = vm->state;
D
Daniel P. Berrange 已提交
3653

3654
    if (!virDomainIsActive(vm)) {
3655
        info->cpuTime = 0;
D
Daniel P. Berrange 已提交
3656
    } else {
3657
        if (qemudGetProcessInfo(&(info->cpuTime), NULL, vm->pid, 0) < 0) {
3658
            qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED, ("cannot read cputime for domain"));
3659
            goto cleanup;
D
Daniel P. Berrange 已提交
3660 3661 3662
        }
    }

3663
    info->maxMem = vm->def->maxmem;
3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678

    if (virDomainIsActive(vm)) {
        err = qemudDomainGetMemoryBalloon(dom->conn, vm, &balloon);
        if (err < 0)
            goto cleanup;

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

3679
    info->nrVirtCpu = vm->def->vcpus;
3680 3681 3682
    ret = 0;

cleanup:
3683 3684
    if (vm)
        virDomainObjUnlock(vm);
3685
    return ret;
D
Daniel P. Berrange 已提交
3686 3687 3688
}


D
Daniel P. Berrange 已提交
3689
static char *qemudEscape(const char *in, int shell)
3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710
{
    int len = 0;
    int i, j;
    char *out;

    /* To pass through the QEMU monitor, we need to use escape
       sequences: \r, \n, \", \\

       To pass through both QEMU + the shell, we need to escape
       the single character ' as the five characters '\\''
    */

    for (i = 0; in[i] != '\0'; i++) {
        switch(in[i]) {
        case '\r':
        case '\n':
        case '"':
        case '\\':
            len += 2;
            break;
        case '\'':
D
Daniel P. Berrange 已提交
3711 3712 3713 3714
            if (shell)
                len += 5;
            else
                len += 1;
3715 3716 3717 3718 3719 3720 3721
            break;
        default:
            len += 1;
            break;
        }
    }

3722
    if (VIR_ALLOC_N(out, len + 1) < 0)
3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740
        return NULL;

    for (i = j = 0; in[i] != '\0'; i++) {
        switch(in[i]) {
        case '\r':
            out[j++] = '\\';
            out[j++] = 'r';
            break;
        case '\n':
            out[j++] = '\\';
            out[j++] = 'n';
            break;
        case '"':
        case '\\':
            out[j++] = '\\';
            out[j++] = in[i];
            break;
        case '\'':
D
Daniel P. Berrange 已提交
3741 3742 3743 3744 3745 3746 3747 3748 3749
            if (shell) {
                out[j++] = '\'';
                out[j++] = '\\';
                out[j++] = '\\';
                out[j++] = '\'';
                out[j++] = '\'';
            } else {
                out[j++] = in[i];
            }
3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760
            break;
        default:
            out[j++] = in[i];
            break;
        }
    }
    out[j] = '\0';

    return out;
}

3761 3762 3763 3764 3765
static char *qemudEscapeMonitorArg(const char *in)
{
    return qemudEscape(in, 0);
}

D
Daniel P. Berrange 已提交
3766 3767 3768 3769
static char *qemudEscapeShellArg(const char *in)
{
    return qemudEscape(in, 1);
}
3770

3771
#define QEMUD_SAVE_MAGIC "LibvirtQemudSave"
3772 3773 3774
#define QEMUD_SAVE_VERSION 2

enum qemud_save_formats {
3775 3776 3777
    QEMUD_SAVE_FORMAT_RAW = 0,
    QEMUD_SAVE_FORMAT_GZIP = 1,
    QEMUD_SAVE_FORMAT_BZIP2 = 2,
3778 3779
    /*
     * Deprecated by xz and never used as part of a release
3780
     * QEMUD_SAVE_FORMAT_LZMA
3781 3782
     */
    QEMUD_SAVE_FORMAT_XZ = 3,
3783
    QEMUD_SAVE_FORMAT_LZOP = 4,
3784 3785 3786
    /* Note: add new members only at the end.
       These values are used in the on-disk format.
       Do not change or re-use numbers. */
3787 3788

    QEMUD_SAVE_FORMAT_LAST
3789
};
3790

3791 3792 3793 3794 3795
VIR_ENUM_DECL(qemudSaveCompression)
VIR_ENUM_IMPL(qemudSaveCompression, QEMUD_SAVE_FORMAT_LAST,
              "raw",
              "gzip",
              "bzip2",
3796 3797
              "xz",
              "lzop")
3798

3799 3800 3801 3802 3803
struct qemud_save_header {
    char magic[sizeof(QEMUD_SAVE_MAGIC)-1];
    int version;
    int xml_len;
    int was_running;
3804 3805
    int compressed;
    int unused[15];
3806 3807
};

3808
static int qemudDomainSave(virDomainPtr dom,
3809 3810
                           const char *path)
{
3811
    struct qemud_driver *driver = dom->conn->privateData;
3812
    virDomainObjPtr vm = NULL;
3813 3814 3815 3816 3817
    char *command = NULL;
    char *info = NULL;
    int fd = -1;
    char *safe_path = NULL;
    char *xml = NULL;
3818
    struct qemud_save_header header;
3819
    int ret = -1;
3820
    virDomainEventPtr event = NULL;
3821
    int internalret;
3822 3823 3824 3825 3826

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

3827
    qemuDriverLock(driver);
3828 3829 3830
    if (driver->saveImageFormat == NULL)
        header.compressed = QEMUD_SAVE_FORMAT_RAW;
    else {
3831 3832 3833 3834 3835 3836
        header.compressed =
            qemudSaveCompressionTypeFromString(driver->saveImageFormat);
        if (header.compressed < 0) {
            qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                             "%s", _("Invalid save image format specified "
                                     "in configuration file"));
3837
            goto cleanup;
3838
        }
3839 3840
    }

3841
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3842

D
Daniel P. Berrange 已提交
3843
    if (!vm) {
3844 3845 3846 3847
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching uuid '%s'"), uuidstr);
3848
        goto cleanup;
D
Daniel P. Berrange 已提交
3849
    }
3850

3851
    if (!virDomainIsActive(vm)) {
3852
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
3853
                         "%s", _("domain is not running"));
3854
        goto cleanup;
D
Daniel P. Berrange 已提交
3855
    }
3856 3857 3858 3859

    /* Pause */
    if (vm->state == VIR_DOMAIN_RUNNING) {
        header.was_running = 1;
3860
        if (qemudMonitorCommand(vm, "stop", &info) < 0) {
3861
            qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
3862
                             "%s", _("suspend operation failed"));
3863
            goto cleanup;
3864
        }
3865 3866 3867
        vm->state = VIR_DOMAIN_PAUSED;
        qemudDebug("Reply %s", info);
        VIR_FREE(info);
3868 3869 3870
    }

    /* Get XML for the domain */
3871
    xml = virDomainDefFormat(dom->conn, vm->def, VIR_DOMAIN_XML_SECURE);
3872 3873
    if (!xml) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
3874
                         "%s", _("failed to get domain xml"));
3875
        goto cleanup;
3876 3877 3878 3879 3880 3881
    }
    header.xml_len = strlen(xml) + 1;

    /* Write header to file, followed by XML */
    if ((fd = open(path, O_CREAT|O_TRUNC|O_WRONLY, S_IRUSR|S_IWUSR)) < 0) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
3882
                         _("failed to create '%s'"), path);
3883
        goto cleanup;
3884 3885 3886 3887
    }

    if (safewrite(fd, &header, sizeof(header)) != sizeof(header)) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
3888
                         "%s", _("failed to write save header"));
3889
        goto cleanup;
3890 3891 3892 3893
    }

    if (safewrite(fd, xml, header.xml_len) != header.xml_len) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
3894
                         "%s", _("failed to write xml"));
3895
        goto cleanup;
3896 3897
    }

3898
    if (close(fd) < 0) {
3899 3900 3901
        virReportSystemError(dom->conn, errno,
                             _("unable to save file %s"),
                             path);
3902 3903 3904
        goto cleanup;
    }
    fd = -1;
3905 3906 3907 3908

    /* Migrate to file */
    safe_path = qemudEscapeShellArg(path);
    if (!safe_path) {
3909
        virReportOOMError(dom->conn);
3910
        goto cleanup;
3911
    }
3912

3913 3914 3915 3916 3917 3918 3919 3920 3921
    {
        const char *prog = qemudSaveCompressionTypeToString(header.compressed);
        const char *args;

        if (prog == NULL) {
            qemudReportError(dom->conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
                             _("Invalid compress format %d"), header.compressed);
            goto cleanup;
        }
3922

3923 3924 3925 3926 3927 3928 3929 3930 3931 3932
        if (STREQ (prog, "raw")) {
            prog = "cat";
            args = "";
        } else {
            args = "-c";
        }
        internalret = virAsprintf(&command, "migrate \"exec:"
                                  "%s %s >> '%s' 2>/dev/null\"", prog, args,
                                  safe_path);
    }
3933

3934
    if (internalret < 0) {
3935
        virReportOOMError(dom->conn);
3936
        goto cleanup;
3937 3938
    }

3939
    if (qemudMonitorCommand(vm, command, &info) < 0) {
3940
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
3941
                         "%s", _("migrate operation failed"));
3942
        goto cleanup;
3943 3944
    }

3945
    DEBUG ("%s: migrate reply: %s", vm->def->name, info);
3946 3947 3948 3949 3950 3951 3952

    /* If the command isn't supported then qemu prints:
     * unknown command: migrate" */
    if (strstr(info, "unknown command:")) {
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                          "%s",
                          _("'migrate' not supported by this qemu"));
3953
        goto cleanup;
3954 3955
    }

3956 3957
    /* Shut it down */
    qemudShutdownVMDaemon(dom->conn, driver, vm);
3958 3959 3960
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_SAVED);
3961
    if (!vm->persistent) {
3962 3963
        virDomainRemoveInactive(&driver->domains,
                                vm);
3964 3965
        vm = NULL;
    }
3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976
    ret = 0;

cleanup:
    if (fd != -1)
        close(fd);
    VIR_FREE(xml);
    VIR_FREE(safe_path);
    VIR_FREE(command);
    VIR_FREE(info);
    if (ret != 0)
        unlink(path);
3977 3978
    if (vm)
        virDomainObjUnlock(vm);
3979 3980
    if (event)
        qemuDomainEventQueue(driver, event);
3981
    qemuDriverUnlock(driver);
3982
    return ret;
D
Daniel P. Berrange 已提交
3983 3984 3985
}


P
Paolo Bonzini 已提交
3986 3987 3988 3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 4005 4006 4007 4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064 4065 4066 4067 4068 4069 4070 4071 4072 4073
static int qemudDomainCoreDump(virDomainPtr dom,
                               const char *path,
                               int flags ATTRIBUTE_UNUSED) {
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    char *command = NULL;
    char *info = NULL;
    char *safe_path = NULL;
    int resume = 0, paused = 0;
    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);
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching uuid '%s'"), uuidstr);
        goto cleanup;
    }

    if (!virDomainIsActive(vm)) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
                         "%s", _("domain is not running"));
        goto cleanup;
    }

    /* Migrate will always stop the VM, so once we support live dumping
       the resume condition will stay the same, independent of whether
       the stop command is issued.  */
    resume = (vm->state == VIR_DOMAIN_RUNNING);

    /* Pause domain for non-live dump */
    if (vm->state == VIR_DOMAIN_RUNNING) {
        if (qemudMonitorCommand (vm, "stop", &info) < 0) {
            qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                             "%s", _("suspending before dump failed"));
            goto cleanup;
        }
        DEBUG ("%s: stop reply: %s", vm->def->name, info);
        VIR_FREE(info);
        paused = 1;
    }

    /* Migrate to file */
    safe_path = qemudEscapeShellArg(path);
    if (!safe_path) {
        virReportOOMError(dom->conn);
        goto cleanup;
    }
    if (virAsprintf(&command, "migrate \"exec:"
                  "dd of='%s' 2>/dev/null"
                  "\"", safe_path) == -1) {
        virReportOOMError(dom->conn);
        command = NULL;
        goto cleanup;
    }

    if (qemudMonitorCommand(vm, command, &info) < 0) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                         "%s", _("migrate operation failed"));
        goto cleanup;
    }

    DEBUG ("%s: migrate reply: %s", vm->def->name, info);

    /* If the command isn't supported then qemu prints:
     * unknown command: migrate" */
    if (strstr(info, "unknown command:")) {
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                          "%s",
                          _("'migrate' not supported by this qemu"));
        goto cleanup;
    }

    paused = 1;
    ret = 0;
cleanup:
    VIR_FREE(safe_path);
    VIR_FREE(command);
    VIR_FREE(info);

    /* 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.  */
    if (resume && paused) {
4074 4075 4076 4077
        if (qemudMonitorSendCont(dom->conn, vm) < 0) {
            if (virGetLastError() == NULL)
                qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                                 "%s", _("resuming after dump failed"));
P
Paolo Bonzini 已提交
4078 4079 4080 4081 4082 4083 4084 4085
        }
    }
    if (vm)
        virDomainObjUnlock(vm);
    return ret;
}


4086
static int qemudDomainSetVcpus(virDomainPtr dom, unsigned int nvcpus) {
4087 4088
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
4089
    int max;
4090
    int ret = -1;
4091
    const char *type;
4092

4093
    qemuDriverLock(driver);
4094
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4095 4096
    qemuDriverUnlock(driver);

4097
    if (!vm) {
4098 4099
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4100
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
4101
                         _("no domain with matching uuid '%s'"), uuidstr);
4102
        goto cleanup;
4103 4104
    }

4105
    if (virDomainIsActive(vm)) {
4106
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID, "%s",
4107
                         _("cannot change vcpu count of an active domain"));
4108
        goto cleanup;
4109 4110
    }

4111 4112 4113 4114 4115 4116 4117 4118
    if (!(type = virDomainVirtTypeToString(vm->def->virtType))) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
                         _("unknown virt type in domain definition '%d'"),
                         vm->def->virtType);
        goto cleanup;
    }

    if ((max = qemudGetMaxVCPUs(dom->conn, type)) < 0) {
4119 4120
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INTERNAL_ERROR, "%s",
                         _("could not determine max vcpus for the domain"));
4121
        goto cleanup;
4122 4123 4124 4125 4126 4127
    }

    if (nvcpus > max) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
                         _("requested vcpus is greater than max allowable"
                           " vcpus for the domain: %d > %d"), nvcpus, max);
4128
        goto cleanup;
4129 4130 4131
    }

    vm->def->vcpus = nvcpus;
4132 4133 4134
    ret = 0;

cleanup:
4135 4136
    if (vm)
        virDomainObjUnlock(vm);
4137
    return ret;
4138 4139
}

4140 4141 4142 4143 4144 4145 4146

#if HAVE_SCHED_GETAFFINITY
static int
qemudDomainPinVcpu(virDomainPtr dom,
                   unsigned int vcpu,
                   unsigned char *cpumap,
                   int maplen) {
4147 4148
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
4149 4150 4151
    cpu_set_t mask;
    int i, maxcpu;
    virNodeInfo nodeinfo;
4152
    int ret = -1;
4153

4154
    qemuDriverLock(driver);
4155
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4156 4157
    qemuDriverUnlock(driver);

4158 4159 4160 4161 4162 4163 4164 4165
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching uuid '%s'"), uuidstr);
        goto cleanup;
    }

4166
    if (!virDomainIsActive(vm)) {
4167
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
4168
                         "%s",_("cannot pin vcpus on an inactive domain"));
4169
        goto cleanup;
4170 4171 4172 4173 4174 4175
    }

    if (vcpu > (vm->nvcpupids-1)) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
                         _("vcpu number out of range %d > %d"),
                         vcpu, vm->nvcpupids);
4176
        goto cleanup;
4177 4178
    }

4179
    if (nodeGetInfo(dom->conn, &nodeinfo) < 0)
4180
        goto cleanup;
4181 4182 4183 4184 4185 4186 4187 4188 4189 4190 4191 4192 4193

    maxcpu = maplen * 8;
    if (maxcpu > nodeinfo.cpus)
        maxcpu = nodeinfo.cpus;

    CPU_ZERO(&mask);
    for (i = 0 ; i < maxcpu ; i++) {
        if ((cpumap[i/8] >> (i % 8)) & 1)
            CPU_SET(i, &mask);
    }

    if (vm->vcpupids != NULL) {
        if (sched_setaffinity(vm->vcpupids[vcpu], sizeof(mask), &mask) < 0) {
4194 4195
            virReportSystemError(dom->conn, errno, "%s",
                                 _("cannot set affinity"));
4196
            goto cleanup;
4197 4198 4199 4200
        }
    } else {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                         "%s", _("cpu affinity is not supported"));
4201
        goto cleanup;
4202
    }
4203
    ret = 0;
4204

4205
cleanup:
4206 4207
    if (vm)
        virDomainObjUnlock(vm);
4208
    return ret;
4209 4210 4211 4212 4213 4214 4215 4216
}

static int
qemudDomainGetVcpus(virDomainPtr dom,
                    virVcpuInfoPtr info,
                    int maxinfo,
                    unsigned char *cpumaps,
                    int maplen) {
4217 4218
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
4219 4220
    virNodeInfo nodeinfo;
    int i, v, maxcpu;
4221
    int ret = -1;
4222

4223
    qemuDriverLock(driver);
4224
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4225 4226
    qemuDriverUnlock(driver);

4227 4228 4229 4230 4231 4232 4233 4234
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching uuid '%s'"), uuidstr);
        goto cleanup;
    }

4235
    if (!virDomainIsActive(vm)) {
4236
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
4237
                         "%s",_("cannot pin vcpus on an inactive domain"));
4238
        goto cleanup;
4239 4240
    }

4241
    if (nodeGetInfo(dom->conn, &nodeinfo) < 0)
4242
        goto cleanup;
4243 4244 4245 4246 4247 4248 4249 4250 4251

    maxcpu = maplen * 8;
    if (maxcpu > nodeinfo.cpus)
        maxcpu = nodeinfo.cpus;

    /* Clamp to actual number of vcpus */
    if (maxinfo > vm->nvcpupids)
        maxinfo = vm->nvcpupids;

4252 4253 4254 4255 4256 4257
    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;
4258 4259 4260 4261 4262 4263 4264 4265 4266 4267

                if (vm->vcpupids != NULL &&
                    qemudGetProcessInfo(&(info[i].cpuTime),
                                        &(info[i].cpu),
                                        vm->pid,
                                        vm->vcpupids[i]) < 0) {
                    virReportSystemError(dom->conn, errno, "%s",
                                         _("cannot get vCPU placement & pCPU time"));
                    goto cleanup;
                }
4268
            }
4269 4270
        }

4271 4272 4273 4274 4275 4276 4277 4278 4279
        if (cpumaps != NULL) {
            memset(cpumaps, 0, maplen * maxinfo);
            if (vm->vcpupids != NULL) {
                for (v = 0 ; v < maxinfo ; v++) {
                    cpu_set_t mask;
                    unsigned char *cpumap = VIR_GET_CPUMAP(cpumaps, maplen, v);
                    CPU_ZERO(&mask);

                    if (sched_getaffinity(vm->vcpupids[v], sizeof(mask), &mask) < 0) {
4280 4281
                        virReportSystemError(dom->conn, errno, "%s",
                                             _("cannot get affinity"));
4282 4283 4284 4285 4286 4287
                        goto cleanup;
                    }

                    for (i = 0 ; i < maxcpu ; i++)
                        if (CPU_ISSET(i, &mask))
                            VIR_USE_CPU(cpumap, i);
4288
                }
4289 4290 4291 4292
            } else {
                qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                                 "%s", _("cpu affinity is not available"));
                goto cleanup;
4293 4294 4295
            }
        }
    }
4296
    ret = maxinfo;
4297

4298
cleanup:
4299 4300
    if (vm)
        virDomainObjUnlock(vm);
4301
    return ret;
4302 4303 4304 4305
}
#endif /* HAVE_SCHED_GETAFFINITY */


4306
static int qemudDomainGetMaxVcpus(virDomainPtr dom) {
4307 4308
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
4309
    const char *type;
4310
    int ret = -1;
4311

4312
    qemuDriverLock(driver);
4313
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4314 4315
    qemuDriverUnlock(driver);

4316
    if (!vm) {
4317 4318
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4319
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
4320
                         _("no domain with matching uuid '%s'"), uuidstr);
4321
        goto cleanup;
4322 4323
    }

4324
    if (!(type = virDomainVirtTypeToString(vm->def->virtType))) {
4325 4326 4327
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
                         _("unknown virt type in domain definition '%d'"),
                         vm->def->virtType);
4328
        goto cleanup;
4329 4330
    }

4331
    ret = qemudGetMaxVCPUs(dom->conn, type);
4332

4333
cleanup:
4334 4335
    if (vm)
        virDomainObjUnlock(vm);
4336 4337 4338
    return ret;
}

4339 4340 4341 4342 4343 4344 4345 4346 4347 4348
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);

4349 4350
    memset(seclabel, 0, sizeof(*seclabel));

4351 4352 4353
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
4354
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
4355 4356 4357 4358 4359 4360 4361 4362 4363 4364 4365 4366 4367 4368 4369 4370 4371 4372 4373 4374 4375 4376 4377 4378 4379 4380 4381 4382 4383
                         _("no domain with matching uuid '%s'"), uuidstr);
        goto cleanup;
    }

    if (!(type = virDomainVirtTypeToString(vm->def->virtType))) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
                         _("unknown virt type in domain definition '%d'"),
                         vm->def->virtType);
        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().
     */
    if (virDomainIsActive(vm)) {
        if (driver->securityDriver && driver->securityDriver->domainGetSecurityLabel) {
            if (driver->securityDriver->domainGetSecurityLabel(dom->conn, vm, seclabel) == -1) {
                qemudReportError(dom->conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
4384
                                 "%s", _("Failed to get security label"));
4385 4386 4387 4388 4389 4390 4391 4392 4393 4394
                goto cleanup;
            }
        }
    }

    ret = 0;

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
4395
    qemuDriverUnlock(driver);
4396 4397 4398
    return ret;
}

4399 4400
static int qemudNodeGetSecurityModel(virConnectPtr conn,
                                     virSecurityModelPtr secmodel)
4401 4402 4403
{
    struct qemud_driver *driver = (struct qemud_driver *)conn->privateData;
    char *p;
4404
    int ret = 0;
4405

4406 4407
    qemuDriverLock(driver);
    if (!driver->securityDriver) {
4408
        memset(secmodel, 0, sizeof (*secmodel));
4409 4410
        goto cleanup;
    }
4411

4412 4413 4414 4415 4416
    p = driver->caps->host.secModel.model;
    if (strlen(p) >= VIR_SECURITY_MODEL_BUFLEN-1) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                         _("security model string exceeds max %d bytes"),
                         VIR_SECURITY_MODEL_BUFLEN-1);
4417 4418
        ret = -1;
        goto cleanup;
4419 4420 4421 4422 4423 4424 4425 4426
    }
    strcpy(secmodel->model, p);

    p = driver->caps->host.secModel.doi;
    if (strlen(p) >= VIR_SECURITY_DOI_BUFLEN-1) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                         _("security DOI string exceeds max %d bytes"),
                         VIR_SECURITY_DOI_BUFLEN-1);
4427 4428
        ret = -1;
        goto cleanup;
4429 4430
    }
    strcpy(secmodel->doi, p);
4431 4432 4433 4434

cleanup:
    qemuDriverUnlock(driver);
    return ret;
4435 4436 4437
}

/* TODO: check seclabel restore */
4438
static int qemudDomainRestore(virConnectPtr conn,
4439 4440 4441
                              const char *path) {
    struct qemud_driver *driver = conn->privateData;
    virDomainDefPtr def = NULL;
4442
    virDomainObjPtr vm = NULL;
4443 4444 4445
    int fd = -1;
    int ret = -1;
    char *xml = NULL;
4446
    struct qemud_save_header header;
4447
    virDomainEventPtr event = NULL;
4448 4449 4450
    int intermediatefd = -1;
    pid_t intermediate_pid = -1;
    int childstat;
4451

4452
    qemuDriverLock(driver);
4453 4454 4455
    /* Verify the header and read the XML */
    if ((fd = open(path, O_RDONLY)) < 0) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
4456
                         "%s", _("cannot read domain image"));
4457
        goto cleanup;
4458 4459 4460 4461
    }

    if (saferead(fd, &header, sizeof(header)) != sizeof(header)) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
4462
                         "%s", _("failed to read qemu header"));
4463
        goto cleanup;
4464 4465 4466 4467
    }

    if (memcmp(header.magic, QEMUD_SAVE_MAGIC, sizeof(header.magic)) != 0) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
4468
                         "%s", _("image magic is incorrect"));
4469
        goto cleanup;
4470 4471 4472 4473
    }

    if (header.version > QEMUD_SAVE_VERSION) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
4474
                         _("image version is not supported (%d > %d)"),
4475
                         header.version, QEMUD_SAVE_VERSION);
4476
        goto cleanup;
4477 4478
    }

4479
    if (VIR_ALLOC_N(xml, header.xml_len) < 0) {
4480
        virReportOOMError(conn);
4481
        goto cleanup;
4482 4483 4484 4485
    }

    if (saferead(fd, xml, header.xml_len) != header.xml_len) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
4486
                         "%s", _("failed to read XML"));
4487
        goto cleanup;
4488 4489 4490
    }

    /* Create a domain from this XML */
4491 4492
    if (!(def = virDomainDefParseString(conn, driver->caps, xml,
                                        VIR_DOMAIN_XML_INACTIVE))) {
4493
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
4494
                         "%s", _("failed to parse XML"));
4495
        goto cleanup;
4496 4497
    }

4498
    /* See if a VM with matching UUID already exists */
4499
    vm = virDomainFindByUUID(&driver->domains, def->uuid);
4500
    if (vm) {
4501 4502 4503 4504 4505 4506 4507 4508 4509 4510 4511
        /* UUID matches, but if names don't match, refuse it */
        if (STRNEQ(vm->def->name, def->name)) {
            char uuidstr[VIR_UUID_STRING_BUFLEN];
            virUUIDFormat(vm->def->uuid, uuidstr);
            qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                             _("domain '%s' is already defined with uuid %s"),
                             vm->def->name, uuidstr);
            goto cleanup;
        }

        /* UUID & name match, but if VM is already active, refuse it */
4512
        if (virDomainIsActive(vm)) {
4513
            qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_INVALID,
4514 4515
                             _("domain is already active as '%s'"), vm->def->name);
            goto cleanup;
4516 4517 4518 4519 4520 4521 4522 4523 4524 4525 4526 4527
        }
        virDomainObjUnlock(vm);
    } else {
        /* UUID does not match, but if a name matches, refuse it */
        vm = virDomainFindByName(&driver->domains, def->name);
        if (vm) {
            char uuidstr[VIR_UUID_STRING_BUFLEN];
            virUUIDFormat(vm->def->uuid, uuidstr);
            qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                             _("domain '%s' is already defined with uuid %s"),
                             def->name, uuidstr);
            goto cleanup;
4528
        }
4529 4530
    }

4531 4532 4533
    if (!(vm = virDomainAssignDef(conn,
                                  &driver->domains,
                                  def))) {
4534
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
4535
                         "%s", _("failed to assign new VM"));
4536
        goto cleanup;
4537
    }
4538
    def = NULL;
4539

4540 4541
    if (header.version == 2) {
        const char *intermediate_argv[3] = { NULL, "-dc", NULL };
4542 4543
        const char *prog = qemudSaveCompressionTypeToString(header.compressed);
        if (prog == NULL) {
4544
            qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
4545
                             _("Invalid compressed save format %d"),
4546 4547 4548
                             header.compressed);
            goto cleanup;
        }
4549 4550 4551 4552

        if (header.compressed != QEMUD_SAVE_FORMAT_RAW)
            intermediate_argv[0] = prog;
        else {
4553 4554 4555 4556 4557 4558 4559 4560 4561 4562 4563
            intermediatefd = fd;
            fd = -1;
            if (virExec(conn, intermediate_argv, NULL, NULL,
                        &intermediate_pid, intermediatefd, &fd, NULL, 0) < 0) {
                qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                                 _("Failed to start decompression binary %s"),
                                 intermediate_argv[0]);
                goto cleanup;
            }
        }
    }
4564
    /* Set the migration source and start it up. */
4565
    ret = qemudStartVMDaemon(conn, driver, vm, "stdio", fd);
4566 4567 4568 4569 4570 4571 4572
    if (intermediate_pid != -1) {
        /* Wait for intermediate process to exit */
        while (waitpid(intermediate_pid, &childstat, 0) == -1 &&
               errno == EINTR);
    }
    if (intermediatefd != -1)
        close(intermediatefd);
4573
    close(fd);
4574
    fd = -1;
4575
    if (ret < 0) {
4576
        if (!vm->persistent) {
4577 4578
            virDomainRemoveInactive(&driver->domains,
                                    vm);
4579 4580
            vm = NULL;
        }
4581
        goto cleanup;
4582 4583
    }

4584 4585 4586
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_RESTORED);
4587

4588 4589
    /* If it was running before, resume it now. */
    if (header.was_running) {
4590 4591 4592 4593
        if (qemudMonitorSendCont(conn, vm) < 0) {
            if (virGetLastError() == NULL)
                qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                                 "%s", _("failed to resume domain"));
4594
            goto cleanup;
4595 4596
        }
        vm->state = VIR_DOMAIN_RUNNING;
4597
        virDomainSaveStatus(conn, driver->stateDir, vm);
4598
    }
4599
    ret = 0;
4600

4601 4602 4603 4604 4605
cleanup:
    virDomainDefFree(def);
    VIR_FREE(xml);
    if (fd != -1)
        close(fd);
4606 4607
    if (vm)
        virDomainObjUnlock(vm);
4608 4609
    if (event)
        qemuDomainEventQueue(driver, event);
4610
    qemuDriverUnlock(driver);
4611
    return ret;
D
Daniel P. Berrange 已提交
4612 4613 4614
}


4615
static char *qemudDomainDumpXML(virDomainPtr dom,
4616
                                int flags) {
4617 4618 4619
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    char *ret = NULL;
4620 4621
    unsigned long balloon;
    int err;
4622

4623
    qemuDriverLock(driver);
4624
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4625 4626
    qemuDriverUnlock(driver);

D
Daniel P. Berrange 已提交
4627
    if (!vm) {
4628 4629 4630 4631
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching uuid '%s'"), uuidstr);
4632
        goto cleanup;
D
Daniel P. Berrange 已提交
4633 4634
    }

4635 4636 4637 4638 4639 4640 4641
    /* Refresh current memory based on balloon info */
    err = qemudDomainGetMemoryBalloon(dom->conn, vm, &balloon);
    if (err < 0)
        goto cleanup;
    if (err > 0)
        vm->def->memory = balloon;

4642 4643 4644 4645 4646 4647
    ret = virDomainDefFormat(dom->conn,
                             (flags & VIR_DOMAIN_XML_INACTIVE) && vm->newDef ?
                             vm->newDef : vm->def,
                             flags);

cleanup:
4648 4649
    if (vm)
        virDomainObjUnlock(vm);
4650
    return ret;
D
Daniel P. Berrange 已提交
4651 4652 4653
}


4654 4655 4656 4657
static char *qemuDomainXMLFromNative(virConnectPtr conn,
                                     const char *format,
                                     const char *config,
                                     unsigned int flags ATTRIBUTE_UNUSED) {
4658
    struct qemud_driver *driver = conn->privateData;
4659 4660 4661 4662 4663 4664 4665 4666 4667
    virDomainDefPtr def = NULL;
    char *xml = NULL;

    if (STRNEQ(format, QEMU_CONFIG_FORMAT_ARGV)) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_INVALID_ARG,
                         _("unsupported config type %s"), format);
        goto cleanup;
    }

4668
    qemuDriverLock(driver);
4669
    def = qemuParseCommandLineString(conn, driver->caps, config);
4670
    qemuDriverUnlock(driver);
4671 4672 4673 4674 4675 4676 4677 4678 4679 4680
    if (!def)
        goto cleanup;

    xml = virDomainDefFormat(conn, def, VIR_DOMAIN_XML_INACTIVE);

cleanup:
    virDomainDefFree(def);
    return xml;
}

4681 4682 4683 4684 4685 4686
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;
4687
    virDomainChrDef monitor_chr;
4688 4689 4690 4691 4692 4693 4694 4695 4696 4697
    const char *emulator;
    unsigned int qemuCmdFlags;
    struct stat sb;
    const char **retargv = NULL;
    const char **retenv = NULL;
    const char **tmp;
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    char *ret = NULL;
    int i;

4698 4699
    qemuDriverLock(driver);

4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711 4712 4713 4714 4715 4716 4717 4718 4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731 4732 4733 4734 4735 4736 4737 4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758 4759 4760 4761 4762 4763 4764 4765
    if (STRNEQ(format, QEMU_CONFIG_FORMAT_ARGV)) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_INVALID_ARG,
                         _("unsupported config type %s"), format);
        goto cleanup;
    }

    def = virDomainDefParseString(conn, driver->caps, xmlData, 0);
    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) {
        virReportSystemError(conn, errno,
                             _("Cannot find QEMU binary %s"),
                             emulator);
        goto cleanup;
    }

    if (qemudExtractVersionInfo(emulator,
                                NULL,
                                &qemuCmdFlags) < 0) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                         _("Cannot determine QEMU argv syntax %s"),
                         emulator);
        goto cleanup;
    }

4766 4767
    if (qemuPrepareMonitorChr(conn, driver, &monitor_chr, def->name) < 0)
        goto cleanup;
4768 4769

    if (qemudBuildCommandLine(conn, driver, def,
4770
                              &monitor_chr, qemuCmdFlags,
4771 4772 4773 4774 4775 4776 4777 4778 4779 4780 4781 4782 4783 4784 4785 4786 4787 4788 4789 4790 4791 4792 4793 4794 4795
                              &retargv, &retenv,
                              NULL, NULL, /* Don't want it to create TAP devices */
                              NULL) < 0) {
        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++;
    }

    if (virBufferError(&buf))
        goto cleanup;

    ret = virBufferContentAndReset(&buf);

cleanup:
4796
    qemuDriverUnlock(driver);
4797 4798 4799 4800 4801 4802 4803 4804 4805 4806 4807 4808 4809
    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;
}


4810
static int qemudListDefinedDomains(virConnectPtr conn,
4811
                            char **const names, int nnames) {
4812
    struct qemud_driver *driver = conn->privateData;
4813
    int got = 0, i;
4814

4815
    qemuDriverLock(driver);
4816
    for (i = 0 ; i < driver->domains.count && got < nnames ; i++) {
4817
        virDomainObjLock(driver->domains.objs[i]);
4818 4819
        if (!virDomainIsActive(driver->domains.objs[i])) {
            if (!(names[got++] = strdup(driver->domains.objs[i]->def->name))) {
4820
                virReportOOMError(conn);
4821
                virDomainObjUnlock(driver->domains.objs[i]);
4822 4823
                goto cleanup;
            }
4824
        }
4825
        virDomainObjUnlock(driver->domains.objs[i]);
D
Daniel P. Berrange 已提交
4826
    }
4827

4828
    qemuDriverUnlock(driver);
D
Daniel P. Berrange 已提交
4829
    return got;
4830 4831 4832

 cleanup:
    for (i = 0 ; i < got ; i++)
4833
        VIR_FREE(names[i]);
4834
    qemuDriverUnlock(driver);
4835
    return -1;
D
Daniel P. Berrange 已提交
4836 4837
}

4838
static int qemudNumDefinedDomains(virConnectPtr conn) {
4839
    struct qemud_driver *driver = conn->privateData;
4840 4841
    int n = 0, i;

4842
    qemuDriverLock(driver);
4843 4844
    for (i = 0 ; i < driver->domains.count ; i++)
        if (!virDomainIsActive(driver->domains.objs[i]))
4845
            n++;
4846
    qemuDriverUnlock(driver);
4847

4848
    return n;
D
Daniel P. Berrange 已提交
4849 4850 4851
}


4852
static int qemudDomainStart(virDomainPtr dom) {
4853 4854 4855
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4856
    virDomainEventPtr event = NULL;
4857

4858
    qemuDriverLock(driver);
4859
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4860

4861
    if (!vm) {
4862 4863 4864 4865
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching uuid '%s'"), uuidstr);
4866
        goto cleanup;
4867 4868
    }

4869
    ret = qemudStartVMDaemon(dom->conn, driver, vm, NULL, -1);
4870
    if (ret != -1)
4871 4872 4873
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STARTED,
                                         VIR_DOMAIN_EVENT_STARTED_BOOTED);
4874 4875

cleanup:
4876 4877
    if (vm)
        virDomainObjUnlock(vm);
4878
    if (event)
4879
        qemuDomainEventQueue(driver, event);
4880
    qemuDriverUnlock(driver);
4881
    return ret;
D
Daniel P. Berrange 已提交
4882 4883
}

4884 4885 4886 4887 4888 4889 4890 4891 4892 4893 4894 4895 4896 4897 4898 4899 4900 4901 4902 4903 4904 4905 4906 4907 4908 4909 4910 4911 4912
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;

        if (strcmp(def->os.machine, machine->name) != 0)
            continue;

        if (!(*canonical = strdup(machine->canonical))) {
            virReportOOMError(NULL);
            return -1;
        }

        break;
    }

    return 0;
}

4913 4914 4915 4916 4917 4918 4919 4920 4921 4922 4923 4924 4925 4926 4927 4928 4929 4930 4931 4932 4933 4934 4935 4936 4937 4938 4939 4940
static int
qemudCanonicalizeMachineDirect(virDomainDefPtr def, char **canonical)
{
    virCapsGuestMachinePtr *machines = NULL;
    int i, nmachines = 0;

    if (qemudProbeMachineTypes(def->emulator, &machines, &nmachines) < 0) {
        virReportOOMError(NULL);
        return -1;
    }

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

        if (strcmp(def->os.machine, machines[i]->name) != 0)
            continue;

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

    virCapabilitiesFreeMachines(machines, nmachines);

    return 0;
}

4941 4942
int
qemudCanonicalizeMachine(struct qemud_driver *driver, virDomainDefPtr def)
4943 4944 4945 4946 4947 4948
{
    char *canonical = NULL;
    int i;

    for (i = 0; i < driver->caps->nguests; i++) {
        virCapsGuestPtr guest = driver->caps->guests[i];
4949
        virCapsGuestDomainInfoPtr info;
4950 4951 4952
        int j;

        for (j = 0; j < guest->arch.ndomains; j++) {
4953
            info = &guest->arch.domains[j]->info;
4954

4955 4956 4957 4958 4959 4960 4961 4962 4963
            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;
4964 4965
        }

4966 4967 4968 4969
        info = &guest->arch.defaultInfo;

        if (info->emulator && STREQ(info->emulator, def->emulator)) {
            if (qemudCanonicalizeMachineFromInfo(def, info, &canonical) < 0)
4970 4971 4972 4973
                return -1;
            goto out;
        }
    }
4974 4975 4976 4977

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

4978 4979 4980 4981 4982 4983 4984
out:
    if (canonical) {
        VIR_FREE(def->os.machine);
        def->os.machine = canonical;
    }
    return 0;
}
D
Daniel P. Berrange 已提交
4985

4986
static virDomainPtr qemudDomainDefine(virConnectPtr conn, const char *xml) {
4987
    struct qemud_driver *driver = conn->privateData;
4988
    virDomainDefPtr def;
4989
    virDomainObjPtr vm = NULL;
4990
    virDomainPtr dom = NULL;
4991
    virDomainEventPtr event = NULL;
4992
    int newVM = 1;
4993

4994
    qemuDriverLock(driver);
4995 4996
    if (!(def = virDomainDefParseString(conn, driver->caps, xml,
                                        VIR_DOMAIN_XML_INACTIVE)))
4997
        goto cleanup;
4998

4999 5000 5001
    if (virSecurityDriverVerify(conn, def) < 0)
        goto cleanup;

5002 5003
    /* See if a VM with matching UUID already exists */
    vm = virDomainFindByUUID(&driver->domains, def->uuid);
5004
    if (vm) {
5005 5006 5007 5008 5009 5010 5011 5012 5013 5014 5015
        /* UUID matches, but if names don't match, refuse it */
        if (STRNEQ(vm->def->name, def->name)) {
            char uuidstr[VIR_UUID_STRING_BUFLEN];
            virUUIDFormat(vm->def->uuid, uuidstr);
            qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                             _("domain '%s' is already defined with uuid %s"),
                             vm->def->name, uuidstr);
            goto cleanup;
        }

        /* UUID & name match */
5016
        virDomainObjUnlock(vm);
5017
        newVM = 0;
5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028
    } else {
        /* UUID does not match, but if a name matches, refuse it */
        vm = virDomainFindByName(&driver->domains, def->name);
        if (vm) {
            char uuidstr[VIR_UUID_STRING_BUFLEN];
            virUUIDFormat(vm->def->uuid, uuidstr);
            qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                             _("domain '%s' is already defined with uuid %s"),
                             def->name, uuidstr);
            goto cleanup;
        }
5029
    }
5030

5031
    if (qemudCanonicalizeMachine(driver, def) < 0)
5032 5033
        goto cleanup;

5034 5035 5036
    if (!(vm = virDomainAssignDef(conn,
                                  &driver->domains,
                                  def))) {
5037
        goto cleanup;
5038
    }
5039
    def = NULL;
5040
    vm->persistent = 1;
5041

5042 5043
    if (virDomainSaveConfig(conn,
                            driver->configDir,
5044
                            vm->newDef ? vm->newDef : vm->def) < 0) {
5045 5046
        virDomainRemoveInactive(&driver->domains,
                                vm);
5047
        vm = NULL;
5048
        goto cleanup;
5049 5050
    }

5051 5052 5053 5054 5055
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_DEFINED,
                                     newVM ?
                                     VIR_DOMAIN_EVENT_DEFINED_ADDED :
                                     VIR_DOMAIN_EVENT_DEFINED_UPDATED);
5056

5057
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
5058
    if (dom) dom->id = vm->def->id;
5059 5060

cleanup:
5061
    virDomainDefFree(def);
5062 5063
    if (vm)
        virDomainObjUnlock(vm);
5064 5065
    if (event)
        qemuDomainEventQueue(driver, event);
5066
    qemuDriverUnlock(driver);
5067
    return dom;
D
Daniel P. Berrange 已提交
5068 5069
}

5070
static int qemudDomainUndefine(virDomainPtr dom) {
5071 5072
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
5073
    virDomainEventPtr event = NULL;
5074
    int ret = -1;
D
Daniel P. Berrange 已提交
5075

5076
    qemuDriverLock(driver);
5077
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5078

D
Daniel P. Berrange 已提交
5079
    if (!vm) {
5080 5081 5082 5083
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching uuid '%s'"), uuidstr);
5084
        goto cleanup;
D
Daniel P. Berrange 已提交
5085 5086
    }

5087
    if (virDomainIsActive(vm)) {
5088
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
5089
                         "%s", _("cannot delete active domain"));
5090
        goto cleanup;
D
Daniel P. Berrange 已提交
5091 5092
    }

5093 5094 5095
    if (!vm->persistent) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
                         "%s", _("cannot undefine transient domain"));
5096
        goto cleanup;
5097 5098 5099
    }

    if (virDomainDeleteConfig(dom->conn, driver->configDir, driver->autostartDir, vm) < 0)
5100
        goto cleanup;
D
Daniel P. Berrange 已提交
5101

5102 5103 5104
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_UNDEFINED,
                                     VIR_DOMAIN_EVENT_UNDEFINED_REMOVED);
5105

5106 5107
    virDomainRemoveInactive(&driver->domains,
                            vm);
5108
    vm = NULL;
5109
    ret = 0;
D
Daniel P. Berrange 已提交
5110

5111
cleanup:
5112 5113
    if (vm)
        virDomainObjUnlock(vm);
5114 5115
    if (event)
        qemuDomainEventQueue(driver, event);
5116
    qemuDriverUnlock(driver);
5117
    return ret;
D
Daniel P. Berrange 已提交
5118 5119
}

5120
/* Return the disks name for use in monitor commands */
5121
static char *qemudDiskDeviceName(const virConnectPtr conn,
5122
                                 const virDomainDiskDefPtr disk) {
5123 5124 5125 5126 5127 5128

    int busid, devid;
    int ret;
    char *devname;

    if (virDiskNameToBusDeviceIndex(disk, &busid, &devid) < 0) {
5129
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
5130 5131 5132 5133 5134 5135 5136
                         _("cannot convert disk '%s' to bus/device index"),
                         disk->dst);
        return NULL;
    }

    switch (disk->bus) {
        case VIR_DOMAIN_DISK_BUS_IDE:
5137
            if (disk->device== VIR_DOMAIN_DISK_DEVICE_DISK)
5138
                ret = virAsprintf(&devname, "ide%d-hd%d", busid, devid);
5139
            else
5140
                ret = virAsprintf(&devname, "ide%d-cd%d", busid, devid);
5141 5142
            break;
        case VIR_DOMAIN_DISK_BUS_SCSI:
5143
            if (disk->device == VIR_DOMAIN_DISK_DEVICE_DISK)
5144
                ret = virAsprintf(&devname, "scsi%d-hd%d", busid, devid);
5145
            else
5146
                ret = virAsprintf(&devname, "scsi%d-cd%d", busid, devid);
5147 5148
            break;
        case VIR_DOMAIN_DISK_BUS_FDC:
5149
            ret = virAsprintf(&devname, "floppy%d", devid);
5150 5151
            break;
        case VIR_DOMAIN_DISK_BUS_VIRTIO:
5152
            ret = virAsprintf(&devname, "virtio%d", devid);
5153 5154
            break;
        default:
5155
            qemudReportError(conn, NULL, NULL, VIR_ERR_NO_SUPPORT,
5156 5157 5158 5159 5160 5161
                             _("Unsupported disk name mapping for bus '%s'"),
                             virDomainDiskBusTypeToString(disk->bus));
            return NULL;
    }

    if (ret == -1) {
5162
        virReportOOMError(conn);
5163 5164 5165 5166 5167 5168
        return NULL;
    }

    return devname;
}

5169 5170
static int qemudDomainChangeEjectableMedia(virConnectPtr conn,
                                           virDomainObjPtr vm,
5171 5172
                                           virDomainDeviceDefPtr dev,
                                           unsigned int qemuCmdFlags)
5173
{
5174
    virDomainDiskDefPtr origdisk = NULL, newdisk;
5175
    char *cmd, *reply, *safe_path;
5176
    char *devname = NULL;
5177
    int i;
5178

5179
    origdisk = NULL;
5180
    newdisk = dev->data.disk;
5181 5182 5183 5184
    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (vm->def->disks[i]->bus == newdisk->bus &&
            STREQ(vm->def->disks[i]->dst, newdisk->dst)) {
            origdisk = vm->def->disks[i];
5185
            break;
5186
        }
5187 5188 5189
    }

    if (!origdisk) {
5190
        qemudReportError(conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
5191 5192 5193 5194 5195 5196 5197
                         _("No device with bus '%s' and target '%s'"),
                         virDomainDiskBusTypeToString(newdisk->bus),
                         newdisk->dst);
        return -1;
    }

    if (qemuCmdFlags & QEMUD_CMD_FLAG_DRIVE) {
5198
        if (!(devname = qemudDiskDeviceName(conn, newdisk)))
5199 5200 5201 5202 5203 5204 5205 5206 5207
            return -1;
    } else {
        /* Back compat for no -drive option */
        if (newdisk->device == VIR_DOMAIN_DISK_DEVICE_FLOPPY)
            devname = strdup(newdisk->dst);
        else if (newdisk->device == VIR_DOMAIN_DISK_DEVICE_CDROM &&
                 STREQ(newdisk->dst, "hdc"))
            devname = strdup("cdrom");
        else {
5208
            qemudReportError(conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
5209 5210 5211 5212 5213 5214 5215 5216
                             _("Emulator version does not support removable "
                               "media for device '%s' and target '%s'"),
                               virDomainDiskDeviceTypeToString(newdisk->device),
                               newdisk->dst);
            return -1;
        }

        if (!devname) {
5217
            virReportOOMError(conn);
5218 5219 5220
            return -1;
        }
    }
5221

5222
    if (newdisk->src) {
5223 5224
        safe_path = qemudEscapeMonitorArg(newdisk->src);
        if (!safe_path) {
5225
            virReportOOMError(conn);
5226
            VIR_FREE(devname);
5227 5228
            return -1;
        }
5229
        if (virAsprintf(&cmd, "change %s \"%s\"", devname, safe_path) == -1) {
5230
            virReportOOMError(conn);
5231
            VIR_FREE(safe_path);
5232
            VIR_FREE(devname);
5233 5234
            return -1;
        }
5235
        VIR_FREE(safe_path);
5236

5237
    } else if (virAsprintf(&cmd, "eject %s", devname) == -1) {
5238
        virReportOOMError(conn);
5239
        VIR_FREE(devname);
5240 5241
        return -1;
    }
5242
    VIR_FREE(devname);
5243

5244
    if (qemudMonitorCommand(vm, cmd, &reply) < 0) {
5245
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
C
Cole Robinson 已提交
5246
                         "%s", _("could not change cdrom media"));
5247
        VIR_FREE(cmd);
5248 5249
        return -1;
    }
5250 5251 5252 5253

    /* If the command failed qemu prints:
     * device not found, device is locked ...
     * No message is printed on success it seems */
5254
    DEBUG ("%s: ejectable media change reply: %s", vm->def->name, reply);
5255
    if (strstr(reply, "\ndevice ")) {
5256
        qemudReportError (conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
C
Cole Robinson 已提交
5257
                          _("changing cdrom media failed: %s"), reply);
5258 5259 5260 5261
        VIR_FREE(reply);
        VIR_FREE(cmd);
        return -1;
    }
5262 5263
    VIR_FREE(reply);
    VIR_FREE(cmd);
5264

5265 5266
    VIR_FREE(origdisk->src);
    origdisk->src = newdisk->src;
5267
    newdisk->src = NULL;
5268
    origdisk->type = newdisk->type;
5269 5270 5271
    return 0;
}

5272 5273 5274 5275 5276 5277 5278 5279 5280 5281 5282 5283 5284 5285 5286 5287 5288 5289 5290 5291 5292 5293 5294 5295 5296 5297 5298 5299 5300 5301 5302 5303 5304 5305 5306 5307 5308 5309 5310 5311 5312 5313 5314 5315 5316 5317 5318 5319 5320 5321 5322 5323 5324 5325 5326 5327 5328 5329 5330 5331 5332 5333 5334 5335 5336 5337 5338
static int
qemudParsePciAddReply(virDomainObjPtr vm,
                      const char *reply,
                      unsigned *domain,
                      unsigned *bus,
                      unsigned *slot)
{
    char *s, *e;

    DEBUG("%s: pci_add reply: %s", vm->def->name, reply);

    /* If the command succeeds qemu prints:
     * OK bus 0, slot XXX...
     * or
     * OK domain 0, bus 0, slot XXX
     */
    if (!(s = strstr(reply, "OK ")))
        return -1;

    s += 3;

    if (STRPREFIX(s, "domain ")) {
        s += strlen("domain ");

        if (virStrToLong_ui(s, &e, 10, domain) == -1) {
            VIR_WARN(_("Unable to parse domain number '%s'\n"), s);
            return -1;
        }

        if (!STRPREFIX(e, ", ")) {
            VIR_WARN(_("Expected ', ' parsing pci_add reply '%s'\n"), s);
            return -1;
        }
        s = e + 2;
    }

    if (!STRPREFIX(s, "bus ")) {
        VIR_WARN(_("Expected 'bus ' parsing pci_add reply '%s'\n"), s);
        return -1;
    }
    s += strlen("bus ");

    if (virStrToLong_ui(s, &e, 10, bus) == -1) {
        VIR_WARN(_("Unable to parse bus number '%s'\n"), s);
        return -1;
    }

    if (!STRPREFIX(e, ", ")) {
        VIR_WARN(_("Expected ', ' parsing pci_add reply '%s'\n"), s);
        return -1;
    }
    s = e + 2;

    if (!STRPREFIX(s, "slot ")) {
        VIR_WARN(_("Expected 'slot ' parsing pci_add reply '%s'\n"), s);
        return -1;
    }
    s += strlen("slot ");

    if (virStrToLong_ui(s, &e, 10, slot) == -1) {
        VIR_WARN(_("Unable to parse slot number '%s'\n"), s);
        return -1;
    }

    return 0;
}

5339 5340 5341
static int qemudDomainAttachPciDiskDevice(virConnectPtr conn,
                                          virDomainObjPtr vm,
                                          virDomainDeviceDefPtr dev)
5342 5343
{
    int ret, i;
5344
    char *cmd, *reply;
5345 5346
    char *safe_path;
    const char* type = virDomainDiskBusTypeToString(dev->data.disk->bus);
5347
    int tryOldSyntax = 0;
5348
    unsigned domain, bus, slot;
5349 5350 5351

    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (STREQ(vm->def->disks[i]->dst, dev->data.disk->dst)) {
5352
            qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
5353 5354 5355 5356 5357 5358
                           _("target %s already exists"), dev->data.disk->dst);
            return -1;
        }
    }

    if (VIR_REALLOC_N(vm->def->disks, vm->def->ndisks+1) < 0) {
5359
        virReportOOMError(conn);
5360 5361 5362
        return -1;
    }

5363
try_command:
5364 5365
    safe_path = qemudEscapeMonitorArg(dev->data.disk->src);
    if (!safe_path) {
5366
        virReportOOMError(conn);
5367 5368 5369
        return -1;
    }

5370 5371
    ret = virAsprintf(&cmd, "pci_add %s storage file=%s,if=%s",
                      (tryOldSyntax ? "0": "pci_addr=auto"), safe_path, type);
5372 5373
    VIR_FREE(safe_path);
    if (ret == -1) {
5374
        virReportOOMError(conn);
5375 5376 5377
        return ret;
    }

5378
    if (qemudMonitorCommand(vm, cmd, &reply) < 0) {
5379
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
5380 5381 5382 5383 5384
                         _("cannot attach %s disk"), type);
        VIR_FREE(cmd);
        return -1;
    }

5385 5386
    VIR_FREE(cmd);

5387 5388 5389 5390 5391 5392
    if (qemudParsePciAddReply(vm, reply, &domain, &bus, &slot) < 0) {
        if (!tryOldSyntax && strstr(reply, "invalid char in expression")) {
            VIR_FREE(reply);
            tryOldSyntax = 1;
            goto try_command;
        }
5393

5394
        qemudReportError (conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
5395
                          _("adding %s disk failed: %s"), type, reply);
5396 5397 5398 5399
        VIR_FREE(reply);
        return -1;
    }

5400 5401 5402 5403 5404 5405
    VIR_FREE(reply);

    dev->data.disk->pci_addr.domain = domain;
    dev->data.disk->pci_addr.bus    = bus;
    dev->data.disk->pci_addr.slot   = slot;

5406
    virDomainDiskInsertPreAlloced(vm->def, dev->data.disk);
5407 5408 5409

    return 0;
}
5410

5411 5412 5413
static int qemudDomainAttachUsbMassstorageDevice(virConnectPtr conn,
                                                 virDomainObjPtr vm,
                                                 virDomainDeviceDefPtr dev)
5414
{
5415 5416
    int ret, i;
    char *safe_path;
5417 5418
    char *cmd, *reply;

5419 5420
    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (STREQ(vm->def->disks[i]->dst, dev->data.disk->dst)) {
5421
            qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
5422 5423 5424 5425 5426
                           _("target %s already exists"), dev->data.disk->dst);
            return -1;
        }
    }

5427
    if (VIR_REALLOC_N(vm->def->disks, vm->def->ndisks+1) < 0) {
5428
        virReportOOMError(conn);
5429
        return -1;
5430 5431
    }

5432 5433
    safe_path = qemudEscapeMonitorArg(dev->data.disk->src);
    if (!safe_path) {
5434
        virReportOOMError(conn);
5435 5436 5437
        return -1;
    }

5438
    ret = virAsprintf(&cmd, "usb_add disk:%s", safe_path);
5439
    VIR_FREE(safe_path);
5440
    if (ret == -1) {
5441
        virReportOOMError(conn);
5442 5443 5444
        return ret;
    }

5445
    if (qemudMonitorCommand(vm, cmd, &reply) < 0) {
5446
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
5447
                         "%s", _("cannot attach usb disk"));
5448 5449 5450 5451
        VIR_FREE(cmd);
        return -1;
    }

5452
    DEBUG ("%s: attach_usb reply: %s",vm->def->name,  reply);
5453 5454 5455
    /* If the command failed qemu prints:
     * Could not add ... */
    if (strstr(reply, "Could not add ")) {
5456
        qemudReportError (conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
5457
                          "%s",
5458
                          _("adding usb disk failed"));
5459 5460 5461 5462
        VIR_FREE(reply);
        VIR_FREE(cmd);
        return -1;
    }
5463

5464
    virDomainDiskInsertPreAlloced(vm->def, dev->data.disk);
5465

5466 5467 5468 5469 5470
    VIR_FREE(reply);
    VIR_FREE(cmd);
    return 0;
}

M
Mark McLoughlin 已提交
5471
static int qemudDomainAttachNetDevice(virConnectPtr conn,
5472
                                      struct qemud_driver *driver,
M
Mark McLoughlin 已提交
5473 5474 5475 5476 5477
                                      virDomainObjPtr vm,
                                      virDomainDeviceDefPtr dev,
                                      unsigned int qemuCmdFlags)
{
    virDomainNetDefPtr net = dev->data.net;
5478
    char *cmd = NULL, *reply = NULL, *remove_cmd = NULL;
5479 5480
    char *tapfd_name = NULL, *tapfd_close = NULL;
    int i, tapfd = -1;
5481
    unsigned domain, bus, slot;
M
Mark McLoughlin 已提交
5482 5483 5484 5485 5486 5487 5488 5489 5490

    if (!(qemuCmdFlags & QEMUD_CMD_FLAG_HOST_NET_ADD)) {
        qemudReportError(conn, dom, NULL, VIR_ERR_NO_SUPPORT, "%s",
                         _("installed qemu version does not support host_net_add"));
        return -1;
    }

    if (net->type == VIR_DOMAIN_NET_TYPE_BRIDGE ||
        net->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
5491 5492 5493 5494 5495 5496 5497 5498 5499 5500
        if (vm->monitor_chr->type != VIR_DOMAIN_CHR_TYPE_UNIX) {
            qemudReportError(conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                             _("network device type '%s' cannot be attached: "
                               "qemu is not using a unix socket monitor"),
                             virDomainNetTypeToString(net->type));
            return -1;
        }

        if ((tapfd = qemudNetworkIfaceConnect(conn, driver, net, qemuCmdFlags)) < 0)
            return -1;
M
Mark McLoughlin 已提交
5501 5502
    }

5503 5504
    if (VIR_REALLOC_N(vm->def->nets, vm->def->nnets+1) < 0)
        goto no_memory;
M
Mark McLoughlin 已提交
5505 5506

    if ((qemuCmdFlags & QEMUD_CMD_FLAG_NET_NAME) &&
5507 5508
        qemuAssignNetNames(vm->def, net) < 0)
        goto no_memory;
M
Mark McLoughlin 已提交
5509 5510 5511 5512 5513 5514 5515 5516 5517

    /* Choose a vlan value greater than all other values since
     * older versions did not store the value in the state file.
     */
    net->vlan = vm->def->nnets;
    for (i = 0; i < vm->def->nnets; i++)
        if (vm->def->nets[i]->vlan >= net->vlan)
            net->vlan = vm->def->nets[i]->vlan;

5518 5519 5520 5521 5522 5523 5524 5525 5526 5527 5528 5529 5530 5531 5532 5533 5534 5535 5536 5537 5538 5539 5540 5541 5542 5543 5544 5545 5546 5547 5548 5549 5550 5551 5552
    if (tapfd != -1) {
        if (virAsprintf(&tapfd_name, "fd-%s", net->hostnet_name) < 0)
            goto no_memory;

        if (virAsprintf(&tapfd_close, "closefd %s", tapfd_name) < 0)
            goto no_memory;

        if (virAsprintf(&cmd, "getfd %s", tapfd_name) < 0)
            goto no_memory;

        if (qemudMonitorCommandWithFd(vm, cmd, tapfd, &reply) < 0) {
            qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                             _("failed to pass fd to qemu with '%s'"), cmd);
            goto cleanup;
        }

        DEBUG("%s: getfd reply: %s", vm->def->name, reply);

        /* If the command isn't supported then qemu prints:
         * unknown command: getfd" */
        if (strstr(reply, "unknown command:")) {
            qemudReportError(conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                             "%s",
                             _("bridge/network interface attach not supported: "
                               "qemu 'getfd' monitor command not available"));
            goto cleanup;
        }

        VIR_FREE(reply);
        VIR_FREE(cmd);
    }

    if (qemuBuildHostNetStr(conn, net, "host_net_add ", ' ',
                            net->vlan, tapfd_name, &cmd) < 0)
        goto try_tapfd_close;
M
Mark McLoughlin 已提交
5553

5554 5555 5556 5557
    remove_cmd = NULL;
    if (net->vlan >= 0 && net->hostnet_name &&
        virAsprintf(&remove_cmd, "host_net_remove %d %s",
                    net->vlan, net->hostnet_name) < 0) {
5558 5559
        virReportOOMError(conn);
        goto try_tapfd_close;
5560 5561
    }

M
Mark McLoughlin 已提交
5562 5563 5564
    if (qemudMonitorCommand(vm, cmd, &reply) < 0) {
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                         _("failed to add network backend with '%s'"), cmd);
5565
        goto try_tapfd_close;
M
Mark McLoughlin 已提交
5566 5567 5568 5569 5570 5571
    }

    DEBUG("%s: host_net_add reply: %s", vm->def->name, reply);

    VIR_FREE(reply);
    VIR_FREE(cmd);
5572 5573 5574 5575 5576
    VIR_FREE(tapfd_name);
    VIR_FREE(tapfd_close);
    if (tapfd != -1)
        close(tapfd);
    tapfd = -1;
M
Mark McLoughlin 已提交
5577 5578

    if (qemuBuildNicStr(conn, net,
5579 5580
                        "pci_add pci_addr=auto ", ' ', net->vlan, &cmd) < 0)
        goto try_remove;
M
Mark McLoughlin 已提交
5581 5582 5583 5584

    if (qemudMonitorCommand(vm, cmd, &reply) < 0) {
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                         _("failed to add NIC with '%s'"), cmd);
5585
        goto try_remove;
M
Mark McLoughlin 已提交
5586 5587
    }

5588 5589 5590 5591 5592 5593
    if (qemudParsePciAddReply(vm, reply, &domain, &bus, &slot) < 0) {
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                         _("parsing pci_add reply failed: %s"), reply);
        goto try_remove;
    }

5594
    VIR_FREE(cmd);
5595
    VIR_FREE(reply);
5596
    VIR_FREE(remove_cmd);
M
Mark McLoughlin 已提交
5597

5598 5599 5600 5601
    net->pci_addr.domain = domain;
    net->pci_addr.bus    = bus;
    net->pci_addr.slot   = slot;

M
Mark McLoughlin 已提交
5602 5603 5604
    vm->def->nets[vm->def->nnets++] = net;

    return 0;
5605 5606

try_remove:
5607
    VIR_FREE(reply);
5608 5609 5610 5611 5612 5613 5614

    if (!remove_cmd)
        VIR_WARN0(_("Unable to remove network backend\n"));
    else if (qemudMonitorCommand(vm, remove_cmd, &reply) < 0)
        VIR_WARN(_("Failed to remove network backend with '%s'\n"), remove_cmd);
    else
        VIR_DEBUG("%s: host_net_remove reply: %s\n", vm->def->name, reply);
5615
    goto cleanup;
5616

5617 5618 5619 5620 5621 5622 5623 5624 5625 5626 5627
try_tapfd_close:
    VIR_FREE(reply);

    if (tapfd_close) {
        if (qemudMonitorCommand(vm, tapfd_close, &reply) < 0)
            VIR_WARN(_("Failed to close tapfd with '%s'\n"), tapfd_close);
        else
            VIR_DEBUG("%s: closefd: %s\n", vm->def->name, reply);
    }
    goto cleanup;

5628 5629 5630 5631
no_memory:
    virReportOOMError(conn);
cleanup:
    VIR_FREE(cmd);
5632 5633
    VIR_FREE(reply);
    VIR_FREE(remove_cmd);
5634 5635 5636 5637
    VIR_FREE(tapfd_close);
    VIR_FREE(tapfd_name);
    if (tapfd != -1)
        close(tapfd);
5638
    return -1;
M
Mark McLoughlin 已提交
5639 5640
}

5641
static int qemudDomainAttachHostPciDevice(virConnectPtr conn,
5642
                                          struct qemud_driver *driver,
5643 5644 5645 5646 5647 5648
                                          virDomainObjPtr vm,
                                          virDomainDeviceDefPtr dev)
{
    virDomainHostdevDefPtr hostdev = dev->data.hostdev;
    char *cmd, *reply;
    unsigned domain, bus, slot;
5649
    pciDevice *pci;
5650 5651 5652 5653 5654 5655

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

5656 5657 5658 5659 5660 5661 5662
    pci = pciGetDevice(conn,
                       hostdev->source.subsys.u.pci.domain,
                       hostdev->source.subsys.u.pci.bus,
                       hostdev->source.subsys.u.pci.slot,
                       hostdev->source.subsys.u.pci.function);
    if (!dev)
        return -1;
5663

5664
    if ((hostdev->managed && pciDettachDevice(conn, pci) < 0) ||
5665 5666 5667 5668 5669 5670
        pciResetDevice(conn, pci, driver->activePciHostdevs) < 0) {
        pciFreeDevice(conn, pci);
        return -1;
    }

    if (pciDeviceListAdd(conn, driver->activePciHostdevs, pci) < 0) {
5671
        pciFreeDevice(conn, pci);
5672
        return -1;
5673 5674
    }

5675
    cmd = reply = NULL;
5676

5677
    if (virAsprintf(&cmd, "pci_add pci_addr=auto host host=%.2x:%.2x.%.1x",
5678 5679 5680 5681
                    hostdev->source.subsys.u.pci.bus,
                    hostdev->source.subsys.u.pci.slot,
                    hostdev->source.subsys.u.pci.function) < 0) {
        virReportOOMError(conn);
5682
        goto error;
5683 5684 5685 5686 5687
    }

    if (qemudMonitorCommand(vm, cmd, &reply) < 0) {
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                         "%s", _("cannot attach host pci device"));
5688
        goto error;
5689 5690 5691 5692 5693
    }

    if (strstr(reply, "invalid type: host")) {
        qemudReportError(conn, dom, NULL, VIR_ERR_NO_SUPPORT, "%s",
                         _("PCI device assignment is not supported by this version of qemu"));
5694
        goto error;
5695 5696 5697 5698 5699
    }

    if (qemudParsePciAddReply(vm, reply, &domain, &bus, &slot) < 0) {
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                         _("parsing pci_add reply failed: %s"), reply);
5700
        goto error;
5701 5702 5703 5704 5705 5706 5707 5708 5709 5710 5711 5712
    }

    hostdev->source.subsys.u.pci.guest_addr.domain = domain;
    hostdev->source.subsys.u.pci.guest_addr.bus    = bus;
    hostdev->source.subsys.u.pci.guest_addr.slot   = slot;

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

    VIR_FREE(reply);
    VIR_FREE(cmd);

    return 0;
5713 5714 5715 5716 5717 5718 5719 5720

error:
    pciDeviceListDel(conn, driver->activePciHostdevs, pci);

    VIR_FREE(reply);
    VIR_FREE(cmd);

    return -1;
5721 5722
}

M
Mark McLoughlin 已提交
5723 5724 5725
static int qemudDomainAttachHostUsbDevice(virConnectPtr conn,
                                          virDomainObjPtr vm,
                                          virDomainDeviceDefPtr dev)
5726 5727 5728 5729
{
    int ret;
    char *cmd, *reply;

5730
    if (VIR_REALLOC_N(vm->def->hostdevs, vm->def->nhostdevs+1) < 0) {
5731
        virReportOOMError(conn);
5732 5733
        return -1;
    }
5734

5735
    if (dev->data.hostdev->source.subsys.u.usb.vendor) {
5736 5737 5738
        ret = virAsprintf(&cmd, "usb_add host:%.4x:%.4x",
                          dev->data.hostdev->source.subsys.u.usb.vendor,
                          dev->data.hostdev->source.subsys.u.usb.product);
5739
    } else {
5740 5741 5742
        ret = virAsprintf(&cmd, "usb_add host:%.3d.%.3d",
                          dev->data.hostdev->source.subsys.u.usb.bus,
                          dev->data.hostdev->source.subsys.u.usb.device);
5743 5744
    }
    if (ret == -1) {
5745
        virReportOOMError(conn);
5746 5747 5748
        return -1;
    }

5749
    if (qemudMonitorCommand(vm, cmd, &reply) < 0) {
5750
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
5751 5752 5753 5754 5755
                         "%s", _("cannot attach usb device"));
        VIR_FREE(cmd);
        return -1;
    }

5756
    DEBUG ("%s: attach_usb reply: %s", vm->def->name, reply);
5757 5758 5759
    /* If the command failed qemu prints:
     * Could not add ... */
    if (strstr(reply, "Could not add ")) {
5760
        qemudReportError (conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
5761 5762 5763 5764 5765 5766
                          "%s",
                          _("adding usb device failed"));
        VIR_FREE(reply);
        VIR_FREE(cmd);
        return -1;
    }
5767

5768
    vm->def->hostdevs[vm->def->nhostdevs++] = dev->data.hostdev;
5769

5770 5771
    VIR_FREE(reply);
    VIR_FREE(cmd);
5772 5773 5774
    return 0;
}

M
Mark McLoughlin 已提交
5775 5776 5777 5778 5779 5780 5781 5782 5783 5784 5785 5786 5787 5788 5789 5790
static int qemudDomainAttachHostDevice(virConnectPtr conn,
                                       struct qemud_driver *driver,
                                       virDomainObjPtr vm,
                                       virDomainDeviceDefPtr dev)
{
    virDomainHostdevDefPtr hostdev = dev->data.hostdev;

    if (hostdev->mode != VIR_DOMAIN_HOSTDEV_MODE_SUBSYS) {
        qemudReportError(conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                         _("hostdev mode '%s' not supported"),
                         virDomainHostdevModeTypeToString(hostdev->mode));
        return -1;
    }

    if (qemuDomainSetDeviceOwnership(conn, driver, dev, 0) < 0)
        return -1;
5791 5792 5793
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityHostdevLabel(conn, vm, dev->data.hostdev) < 0)
        return -1;
M
Mark McLoughlin 已提交
5794 5795

    switch (hostdev->source.subsys.type) {
5796
    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI:
5797
        return qemudDomainAttachHostPciDevice(conn, driver, vm, dev);
M
Mark McLoughlin 已提交
5798 5799 5800 5801 5802 5803 5804 5805 5806 5807
    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB:
        return qemudDomainAttachHostUsbDevice(conn, vm, dev);
    default:
        qemudReportError(conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                         _("hostdev subsys type '%s' not supported"),
                         virDomainHostdevSubsysTypeToString(hostdev->source.subsys.type));
        return -1;
    }
}

5808 5809
static int qemudDomainAttachDevice(virDomainPtr dom,
                                   const char *xml) {
5810 5811 5812
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    virDomainDeviceDefPtr dev = NULL;
5813
    unsigned int qemuCmdFlags;
5814
    virCgroupPtr cgroup = NULL;
5815
    int ret = -1;
5816

5817
    qemuDriverLock(driver);
5818
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5819
    if (!vm) {
5820 5821 5822 5823
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching uuid '%s'"), uuidstr);
5824
        goto cleanup;
5825 5826 5827
    }

    if (!virDomainIsActive(vm)) {
5828
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
5829
                         "%s", _("cannot attach device on inactive domain"));
5830
        goto cleanup;
5831 5832
    }

5833 5834
    dev = virDomainDeviceDefParse(dom->conn, driver->caps, vm->def, xml,
                                  VIR_DOMAIN_XML_INACTIVE);
5835 5836 5837
    if (dev == NULL)
        goto cleanup;

5838 5839 5840 5841
    if (qemudExtractVersionInfo(vm->def->emulator,
                                NULL,
                                &qemuCmdFlags) < 0)
        goto cleanup;
5842

5843
    if (dev->type == VIR_DOMAIN_DEVICE_DISK) {
5844
        if (qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_DEVICES)) {
5845 5846 5847 5848 5849 5850 5851 5852 5853 5854 5855 5856 5857 5858 5859 5860 5861
            if (virCgroupForDomain(driver->cgroup, vm->def->name, &cgroup, 0) !=0 ) {
                qemudReportError(dom->conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                                 _("Unable to find cgroup for %s\n"),
                                 vm->def->name);
                goto cleanup;
            }
            if (dev->data.disk->src != NULL &&
                dev->data.disk->type == VIR_DOMAIN_DISK_TYPE_BLOCK &&
                virCgroupAllowDevicePath(cgroup,
                                         dev->data.disk->src) < 0) {
                qemudReportError(dom->conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                                 _("unable to allow device %s"),
                                 dev->data.disk->src);
                goto cleanup;
            }
        }

5862
        switch (dev->data.disk->device) {
5863 5864
        case VIR_DOMAIN_DISK_DEVICE_CDROM:
        case VIR_DOMAIN_DISK_DEVICE_FLOPPY:
5865 5866
            if (driver->securityDriver)
                driver->securityDriver->domainSetSecurityImageLabel(dom->conn, vm, dev->data.disk);
5867 5868 5869 5870

            if (qemuDomainSetDeviceOwnership(dom->conn, driver, dev, 0) < 0)
                goto cleanup;

5871
            ret = qemudDomainChangeEjectableMedia(dom->conn, vm, dev, qemuCmdFlags);
5872
            break;
5873

5874
        case VIR_DOMAIN_DISK_DEVICE_DISK:
5875 5876
            if (driver->securityDriver)
                driver->securityDriver->domainSetSecurityImageLabel(dom->conn, vm, dev->data.disk);
5877 5878 5879 5880

            if (qemuDomainSetDeviceOwnership(dom->conn, driver, dev, 0) < 0)
                goto cleanup;

5881
            if (dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_USB) {
5882
                ret = qemudDomainAttachUsbMassstorageDevice(dom->conn, vm, dev);
5883 5884
            } else if (dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_SCSI ||
                       dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_VIRTIO) {
5885
                ret = qemudDomainAttachPciDiskDevice(dom->conn, vm, dev);
5886 5887 5888 5889
            } else {
                qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                                 _("disk bus '%s' cannot be hotplugged."),
                                 virDomainDiskBusTypeToString(dev->data.disk->bus));
5890
                /* fallthrough */
5891 5892
            }
            break;
5893

5894 5895
        default:
            qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
5896 5897
                             _("disk device type '%s' cannot be hotplugged"),
                             virDomainDiskDeviceTypeToString(dev->data.disk->device));
5898 5899 5900 5901 5902
            /* Fallthrough */
        }
        if (ret != 0) {
            virCgroupDenyDevicePath(cgroup,
                                    dev->data.disk->src);
5903
        }
M
Mark McLoughlin 已提交
5904
    } else if (dev->type == VIR_DOMAIN_DEVICE_NET) {
5905
        ret = qemudDomainAttachNetDevice(dom->conn, driver, vm, dev, qemuCmdFlags);
M
Mark McLoughlin 已提交
5906 5907
    } else if (dev->type == VIR_DOMAIN_DEVICE_HOSTDEV) {
        ret = qemudDomainAttachHostDevice(dom->conn, driver, vm, dev);
5908
    } else {
5909
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
5910 5911
                         _("device type '%s' cannot be attached"),
                         virDomainDeviceTypeToString(dev->type));
5912
        goto cleanup;
5913 5914
    }

5915
    if (!ret && virDomainSaveStatus(dom->conn, driver->stateDir, vm) < 0)
5916 5917
        ret = -1;

5918
cleanup:
5919 5920 5921
    if (cgroup)
        virCgroupFree(&cgroup);

5922 5923 5924
    if (ret < 0) {
        if (qemuDomainSetDeviceOwnership(dom->conn, driver, dev, 1) < 0)
            VIR_WARN0("Fail to restore disk device ownership");
G
Guido Günther 已提交
5925
        virDomainDeviceDefFree(dev);
5926
    }
5927 5928
    if (vm)
        virDomainObjUnlock(vm);
5929
    qemuDriverUnlock(driver);
5930 5931 5932
    return ret;
}

5933 5934
static int qemudDomainDetachPciDiskDevice(virConnectPtr conn,
                                          virDomainObjPtr vm, virDomainDeviceDefPtr dev)
5935 5936
{
    int i, ret = -1;
5937 5938
    char *cmd = NULL;
    char *reply = NULL;
5939
    virDomainDiskDefPtr detach = NULL;
5940
    int tryOldSyntax = 0;
5941 5942 5943 5944 5945 5946 5947 5948 5949

    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) {
5950
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
5951
                         _("disk %s not found"), dev->data.disk->dst);
5952
        goto cleanup;
5953 5954
    }

5955
    if (!virDiskHasValidPciAddr(detach)) {
5956
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
5957 5958
                         _("disk %s cannot be detached - no PCI address for device"),
                           detach->dst);
5959
        goto cleanup;
5960 5961
    }

5962 5963
try_command:
    if (tryOldSyntax) {
5964
        if (virAsprintf(&cmd, "pci_del 0 %.2x", detach->pci_addr.slot) < 0) {
5965 5966 5967 5968
            virReportOOMError(conn);
            goto cleanup;
        }
    } else {
5969 5970 5971 5972
        if (virAsprintf(&cmd, "pci_del pci_addr=%.4x:%.2x:%.2x",
                        detach->pci_addr.domain,
                        detach->pci_addr.bus,
                        detach->pci_addr.slot) < 0) {
5973 5974 5975
            virReportOOMError(conn);
            goto cleanup;
        }
5976 5977
    }

5978
    if (qemudMonitorCommand(vm, cmd, &reply) < 0) {
5979
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
5980
                          _("failed to execute detach disk %s command"), detach->dst);
5981
        goto cleanup;
5982 5983
    }

5984
    DEBUG ("%s: pci_del reply: %s",vm->def->name,  reply);
5985 5986 5987 5988 5989 5990

    if (!tryOldSyntax &&
        strstr(reply, "extraneous characters")) {
        tryOldSyntax = 1;
        goto try_command;
    }
5991 5992
    /* If the command fails due to a wrong slot qemu prints: invalid slot,
     * nothing is printed on success */
5993 5994
    if (strstr(reply, "invalid slot") ||
        strstr(reply, "Invalid pci address")) {
5995
        qemudReportError (conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
5996 5997 5998 5999 6000 6001
                          _("failed to detach disk %s: invalid PCI address %.4x:%.2x:%.2x: %s"),
                          detach->dst,
                          detach->pci_addr.domain,
                          detach->pci_addr.bus,
                          detach->pci_addr.slot,
                          reply);
6002
        goto cleanup;
6003 6004
    }

6005 6006 6007 6008 6009 6010 6011 6012 6013 6014
    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 {
6015
        VIR_FREE(vm->def->disks);
6016
        vm->def->ndisks = 0;
6017
    }
6018
    virDomainDiskDefFree(detach);
6019

6020
    ret = 0;
6021 6022

cleanup:
6023 6024 6025 6026 6027
    VIR_FREE(reply);
    VIR_FREE(cmd);
    return ret;
}

6028 6029 6030 6031 6032 6033 6034 6035 6036 6037 6038 6039 6040 6041 6042 6043 6044 6045 6046 6047 6048 6049 6050 6051 6052 6053 6054 6055 6056 6057 6058 6059 6060 6061 6062 6063 6064 6065 6066 6067 6068 6069 6070 6071 6072 6073 6074 6075 6076 6077 6078 6079 6080 6081 6082 6083 6084 6085 6086 6087 6088 6089 6090 6091 6092 6093 6094 6095 6096 6097 6098 6099 6100 6101 6102 6103 6104 6105 6106
static int
qemudDomainDetachNetDevice(virConnectPtr conn,
                           virDomainObjPtr vm,
                           virDomainDeviceDefPtr dev)
{
    int i, ret = -1;
    char *cmd = NULL;
    char *reply = NULL;
    virDomainNetDefPtr detach = NULL;

    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) {
        qemudReportError(conn, NULL, NULL, 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]);
        goto cleanup;
    }

    if (!virNetHasValidPciAddr(detach) || detach->vlan < 0 || !detach->hostnet_name) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                         "%s", _("network device cannot be detached - device state missing"));
        goto cleanup;
    }

    if (virAsprintf(&cmd, "pci_del pci_addr=%.4x:%.2x:%.2x",
                    detach->pci_addr.domain,
                    detach->pci_addr.bus,
                    detach->pci_addr.slot) < 0) {
        virReportOOMError(conn);
        goto cleanup;
    }

    if (qemudMonitorCommand(vm, cmd, &reply) < 0) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                          _("network device dettach command '%s' failed"), cmd);
        goto cleanup;
    }

    DEBUG("%s: pci_del reply: %s", vm->def->name,  reply);

    /* If the command fails due to a wrong PCI address qemu prints
     * 'invalid pci address'; nothing is printed on success */
    if (strstr(reply, "Invalid pci address")) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                         _("failed to detach network device: invalid PCI address %.4x:%.2x:%.2x: %s"),
                         detach->pci_addr.domain,
                         detach->pci_addr.bus,
                         detach->pci_addr.slot,
                         reply);
        goto cleanup;
    }

    VIR_FREE(reply);
    VIR_FREE(cmd);

    if (virAsprintf(&cmd, "host_net_remove %d %s",
                    detach->vlan, detach->hostnet_name) < 0) {
        virReportOOMError(conn);
        goto cleanup;
    }

    if (qemudMonitorCommand(vm, cmd, &reply) < 0) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                          _("network device dettach command '%s' failed"), cmd);
        goto cleanup;
    }

    DEBUG("%s: host_net_remove reply: %s", vm->def->name,  reply);

6107 6108 6109 6110 6111 6112 6113 6114 6115 6116
    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 {
6117
        VIR_FREE(vm->def->nets);
6118
        vm->def->nnets = 0;
6119
    }
6120
    virDomainNetDefFree(detach);
6121

6122 6123 6124 6125 6126 6127 6128 6129
    ret = 0;

cleanup:
    VIR_FREE(reply);
    VIR_FREE(cmd);
    return ret;
}

6130
static int qemudDomainDetachHostPciDevice(virConnectPtr conn,
6131
                                          struct qemud_driver *driver,
6132 6133 6134 6135 6136 6137
                                          virDomainObjPtr vm,
                                          virDomainDeviceDefPtr dev)
{
    virDomainHostdevDefPtr detach;
    char *cmd, *reply;
    int i, ret;
6138
    pciDevice *pci;
6139 6140 6141 6142 6143 6144 6145 6146 6147 6148 6149 6150 6151 6152 6153 6154 6155 6156 6157 6158 6159 6160 6161 6162 6163 6164 6165 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 6196 6197 6198 6199 6200 6201 6202 6203 6204 6205 6206

    for (i = 0 ; i < vm->def->nhostdevs ; i++) {
        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) {
        qemudReportError(conn, NULL, NULL, 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);
        return -1;
    }

    if (!virHostdevHasValidGuestAddr(detach)) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                         "%s", _("hostdev cannot be detached - device state missing"));
        return -1;
    }

    if (virAsprintf(&cmd, "pci_del pci_addr=%.4x:%.2x:%.2x",
                    detach->source.subsys.u.pci.guest_addr.domain,
                    detach->source.subsys.u.pci.guest_addr.bus,
                    detach->source.subsys.u.pci.guest_addr.slot) < 0) {
        virReportOOMError(conn);
        return -1;
    }

    if (qemudMonitorCommand(vm, cmd, &reply) < 0) {
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                         "%s", _("cannot detach host pci device"));
        VIR_FREE(cmd);
        return -1;
    }

    DEBUG("%s: pci_del reply: %s", vm->def->name,  reply);

    /* If the command fails due to a wrong PCI address qemu prints
     * 'invalid pci address'; nothing is printed on success */
    if (strstr(reply, "Invalid pci address")) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                         _("failed to detach host pci device: invalid PCI address %.4x:%.2x:%.2x: %s"),
                         detach->source.subsys.u.pci.guest_addr.domain,
                         detach->source.subsys.u.pci.guest_addr.bus,
                         detach->source.subsys.u.pci.guest_addr.slot,
                         reply);
        VIR_FREE(reply);
        VIR_FREE(cmd);
        return -1;
    }

    VIR_FREE(reply);
    VIR_FREE(cmd);

    ret = 0;

6207 6208 6209 6210 6211 6212 6213 6214
    pci = pciGetDevice(conn,
                       detach->source.subsys.u.pci.domain,
                       detach->source.subsys.u.pci.bus,
                       detach->source.subsys.u.pci.slot,
                       detach->source.subsys.u.pci.function);
    if (!pci)
        ret = -1;
    else {
6215 6216
        pciDeviceListDel(conn, driver->activePciHostdevs, pci);
        if (pciResetDevice(conn, pci, driver->activePciHostdevs) < 0)
6217 6218
            ret = -1;
        if (detach->managed && pciReAttachDevice(conn, pci) < 0)
6219
            ret = -1;
6220
        pciFreeDevice(conn, pci);
6221 6222
    }

6223 6224 6225 6226 6227 6228 6229 6230 6231 6232 6233 6234
    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;
6235
    }
6236
    virDomainHostdevDefFree(detach);
6237 6238 6239 6240 6241 6242 6243 6244 6245 6246 6247 6248 6249 6250 6251 6252 6253 6254 6255 6256 6257

    return ret;
}

static int qemudDomainDetachHostDevice(virConnectPtr conn,
                                       struct qemud_driver *driver,
                                       virDomainObjPtr vm,
                                       virDomainDeviceDefPtr dev)
{
    virDomainHostdevDefPtr hostdev = dev->data.hostdev;
    int ret;

    if (hostdev->mode != VIR_DOMAIN_HOSTDEV_MODE_SUBSYS) {
        qemudReportError(conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                         _("hostdev mode '%s' not supported"),
                         virDomainHostdevModeTypeToString(hostdev->mode));
        return -1;
    }

    switch (hostdev->source.subsys.type) {
    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI:
6258
        ret = qemudDomainDetachHostPciDevice(conn, driver, vm, dev);
6259
        break;
6260 6261 6262 6263 6264 6265 6266
    default:
        qemudReportError(conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                         _("hostdev subsys type '%s' not supported"),
                         virDomainHostdevSubsysTypeToString(hostdev->source.subsys.type));
        return -1;
    }

6267 6268 6269 6270
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityHostdevLabel(conn, vm, dev->data.hostdev) < 0)
        VIR_WARN0("Failed to restore device labelling");

6271
    if (qemuDomainSetDeviceOwnership(conn, driver, dev, 1) < 0)
6272
        VIR_WARN0("Failed to restore device ownership");
6273 6274 6275 6276

    return ret;
}

6277 6278
static int qemudDomainDetachDevice(virDomainPtr dom,
                                   const char *xml) {
6279 6280 6281 6282
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    virDomainDeviceDefPtr dev = NULL;
    int ret = -1;
6283

6284
    qemuDriverLock(driver);
6285
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
6286
    if (!vm) {
6287 6288 6289 6290
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching uuid '%s'"), uuidstr);
6291
        goto cleanup;
6292 6293 6294
    }

    if (!virDomainIsActive(vm)) {
6295
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
G
Guido Günther 已提交
6296
                         "%s", _("cannot detach device on inactive domain"));
6297
        goto cleanup;
6298 6299
    }

6300 6301
    dev = virDomainDeviceDefParse(dom->conn, driver->caps, vm->def, xml,
                                  VIR_DOMAIN_XML_INACTIVE);
6302 6303 6304
    if (dev == NULL)
        goto cleanup;

6305 6306 6307 6308

    if (dev->type == VIR_DOMAIN_DEVICE_DISK &&
        dev->data.disk->device == VIR_DOMAIN_DISK_DEVICE_DISK &&
        (dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_SCSI ||
6309
         dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_VIRTIO)) {
6310
        ret = qemudDomainDetachPciDiskDevice(dom->conn, vm, dev);
6311
        if (driver->securityDriver)
6312
            driver->securityDriver->domainRestoreSecurityImageLabel(dom->conn, dev->data.disk);
6313 6314
        if (qemuDomainSetDeviceOwnership(dom->conn, driver, dev, 1) < 0)
            VIR_WARN0("Fail to restore disk device ownership");
6315 6316
    } else if (dev->type == VIR_DOMAIN_DEVICE_NET) {
        ret = qemudDomainDetachNetDevice(dom->conn, vm, dev);
6317 6318
    } else if (dev->type == VIR_DOMAIN_DEVICE_HOSTDEV) {
        ret = qemudDomainDetachHostDevice(dom->conn, driver, vm, dev);
6319
    } else
6320 6321 6322
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                         "%s", _("only SCSI or virtio disk device can be detached dynamically"));

6323
    if (!ret && virDomainSaveStatus(dom->conn, driver->stateDir, vm) < 0)
6324 6325
        ret = -1;

6326 6327
cleanup:
    virDomainDeviceDefFree(dev);
6328 6329
    if (vm)
        virDomainObjUnlock(vm);
6330
    qemuDriverUnlock(driver);
6331 6332 6333
    return ret;
}

6334
static int qemudDomainGetAutostart(virDomainPtr dom,
6335
                                   int *autostart) {
6336 6337 6338
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
6339

6340
    qemuDriverLock(driver);
6341
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
6342 6343
    qemuDriverUnlock(driver);

6344
    if (!vm) {
6345 6346 6347 6348
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching uuid '%s'"), uuidstr);
6349
        goto cleanup;
6350 6351 6352
    }

    *autostart = vm->autostart;
6353
    ret = 0;
6354

6355
cleanup:
6356 6357
    if (vm)
        virDomainObjUnlock(vm);
6358
    return ret;
6359 6360
}

6361
static int qemudDomainSetAutostart(virDomainPtr dom,
6362
                                   int autostart) {
6363 6364
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
6365 6366
    char *configFile = NULL, *autostartLink = NULL;
    int ret = -1;
6367

6368
    qemuDriverLock(driver);
6369
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
6370

6371
    if (!vm) {
6372 6373 6374 6375
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching uuid '%s'"), uuidstr);
6376
        goto cleanup;
6377 6378
    }

6379 6380 6381
    if (!vm->persistent) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
                         "%s", _("cannot set autostart for transient domain"));
6382
        goto cleanup;
6383 6384
    }

6385 6386
    autostart = (autostart != 0);

6387 6388 6389 6390 6391
    if (vm->autostart != autostart) {
        if ((configFile = virDomainConfigFile(dom->conn, driver->configDir, vm->def->name)) == NULL)
            goto cleanup;
        if ((autostartLink = virDomainConfigFile(dom->conn, driver->autostartDir, vm->def->name)) == NULL)
            goto cleanup;
6392

6393 6394
        if (autostart) {
            int err;
6395

6396
            if ((err = virFileMakePath(driver->autostartDir))) {
6397 6398 6399
                virReportSystemError(dom->conn, err,
                                     _("cannot create autostart directory %s"),
                                     driver->autostartDir);
6400 6401
                goto cleanup;
            }
6402

6403
            if (symlink(configFile, autostartLink) < 0) {
6404 6405 6406
                virReportSystemError(dom->conn, errno,
                                     _("Failed to create symlink '%s to '%s'"),
                                     autostartLink, configFile);
6407 6408 6409 6410
                goto cleanup;
            }
        } else {
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
6411 6412 6413
                virReportSystemError(dom->conn, errno,
                                     _("Failed to delete symlink '%s'"),
                                     autostartLink);
6414 6415
                goto cleanup;
            }
6416 6417
        }

6418
        vm->autostart = autostart;
6419
    }
6420
    ret = 0;
6421

6422 6423 6424
cleanup:
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
6425 6426
    if (vm)
        virDomainObjUnlock(vm);
6427
    qemuDriverUnlock(driver);
6428
    return ret;
6429 6430
}

6431 6432 6433 6434 6435

static char *qemuGetSchedulerType(virDomainPtr dom,
                                  int *nparams)
{
    struct qemud_driver *driver = dom->conn->privateData;
6436
    char *ret = NULL;
6437

6438
    qemuDriverLock(driver);
6439
    if (!qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_CPU)) {
6440 6441
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                         __FUNCTION__);
6442
        goto cleanup;
6443 6444 6445 6446 6447 6448 6449 6450
    }

    if (nparams)
        *nparams = 1;

    ret = strdup("posix");
    if (!ret)
        virReportOOMError(dom->conn);
6451 6452 6453

cleanup:
    qemuDriverUnlock(driver);
6454 6455 6456 6457 6458 6459 6460 6461 6462 6463 6464 6465 6466
    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;

6467
    qemuDriverLock(driver);
6468
    if (!qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_CPU)) {
6469 6470
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                         __FUNCTION__);
6471
        goto cleanup;
6472 6473 6474 6475 6476 6477 6478 6479 6480 6481 6482 6483 6484 6485 6486 6487 6488 6489 6490 6491 6492 6493 6494 6495 6496 6497 6498 6499 6500 6501 6502 6503 6504 6505 6506 6507 6508 6509 6510 6511 6512 6513 6514 6515 6516
    }

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

    if (vm == NULL) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
                         _("No such domain %s"), dom->uuid);
        goto cleanup;
    }

    if (virCgroupForDomain(driver->cgroup, vm->def->name, &group, 0) != 0) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
                         _("cannot find cgroup for domain %s"), vm->def->name);
        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) {
                qemudReportError(dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
                                 _("invalid type for cpu_shares tunable, expected a 'ullong'"));
                goto cleanup;
            }

            rc = virCgroupSetCpuShares(group, params[i].value.ul);
            if (rc != 0) {
                virReportSystemError(dom->conn, -rc, "%s",
                                     _("unable to set cpu shares tunable"));
                goto cleanup;
            }
        } else {
            qemudReportError(dom->conn, domain, NULL, VIR_ERR_INVALID_ARG,
                             _("Invalid parameter `%s'"), param->field);
            goto cleanup;
        }
    }
    ret = 0;

cleanup:
    virCgroupFree(&group);
    if (vm)
        virDomainObjUnlock(vm);
6517
    qemuDriverUnlock(driver);
6518 6519 6520 6521 6522 6523 6524 6525 6526 6527 6528 6529 6530 6531
    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;

6532
    qemuDriverLock(driver);
6533
    if (!qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_CPU)) {
6534 6535
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                         __FUNCTION__);
6536
        goto cleanup;
6537 6538 6539 6540 6541
    }

    if ((*nparams) != 1) {
        qemudReportError(dom->conn, domain, NULL, VIR_ERR_INVALID_ARG,
                         "%s", _("Invalid parameter count"));
6542
        goto cleanup;
6543 6544 6545 6546 6547 6548 6549 6550 6551 6552 6553 6554 6555 6556 6557 6558 6559 6560 6561 6562 6563 6564 6565 6566
    }

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

    if (vm == NULL) {
        qemudReportError(dom->conn, domain, NULL, VIR_ERR_INTERNAL_ERROR,
                         _("No such domain %s"), dom->uuid);
        goto cleanup;
    }

    if (virCgroupForDomain(driver->cgroup, vm->def->name, &group, 0) != 0) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
                         _("cannot find cgroup for domain %s"), vm->def->name);
        goto cleanup;
    }

    rc = virCgroupGetCpuShares(group, &val);
    if (rc != 0) {
        virReportSystemError(dom->conn, -rc, "%s",
                             _("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
6567 6568 6569 6570 6571
    if (virStrcpyStatic(params[0].field, "cpu_shares") == NULL) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
                         "%s", _("Field cpu_shares too long for destination"));
        goto cleanup;
    }
6572 6573 6574 6575 6576 6577 6578

    ret = 0;

cleanup:
    virCgroupFree(&group);
    if (vm)
        virDomainObjUnlock(vm);
6579
    qemuDriverUnlock(driver);
6580 6581 6582 6583
    return ret;
}


6584 6585 6586 6587 6588 6589 6590 6591 6592
/* 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)
{
6593
    struct qemud_driver *driver = dom->conn->privateData;
6594
    char *dummy, *info = NULL;
6595
    const char *p, *eol;
6596
    const char *qemu_dev_name = NULL;
6597
    size_t len;
6598
    int i, ret = -1;
6599
    virDomainObjPtr vm;
6600
    virDomainDiskDefPtr disk = NULL;
6601

6602
    qemuDriverLock(driver);
6603
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
6604
    qemuDriverUnlock(driver);
6605
    if (!vm) {
6606 6607 6608 6609
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching uuid '%s'"), uuidstr);
6610
        goto cleanup;
6611
    }
6612
    if (!virDomainIsActive (vm)) {
6613
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
6614
                          "%s", _("domain is not running"));
6615
        goto cleanup;
6616 6617
    }

6618 6619 6620 6621 6622 6623 6624 6625
    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (STREQ(path, vm->def->disks[i]->dst)) {
            disk = vm->def->disks[i];
            break;
        }
    }

    if (!disk) {
6626 6627
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
                          _("invalid path: %s"), path);
6628
        goto cleanup;
6629 6630
    }

6631
    qemu_dev_name = qemudDiskDeviceName(dom->conn, disk);
6632
    if (!qemu_dev_name)
6633
        goto cleanup;
6634 6635
    len = strlen (qemu_dev_name);

6636
    if (qemudMonitorCommand (vm, "info blockstats", &info) < 0) {
6637 6638
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                          "%s", _("'info blockstats' command failed"));
6639
        goto cleanup;
6640
    }
6641
    DEBUG ("%s: info blockstats reply: %s", vm->def->name, info);
6642 6643 6644 6645 6646 6647

    /* If the command isn't supported then qemu prints the supported
     * info commands, so the output starts "info ".  Since this is
     * unlikely to be the name of a block device, we can use this
     * to detect if qemu supports the command.
     */
C
Cole Robinson 已提交
6648
    if (strstr(info, "\ninfo ")) {
6649 6650 6651
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                          "%s",
                          _("'info blockstats' not supported by this qemu"));
6652
        goto cleanup;
6653 6654 6655 6656 6657 6658 6659 6660 6661 6662 6663 6664 6665 6666 6667 6668 6669 6670 6671 6672 6673 6674 6675 6676 6677 6678
    }

    stats->rd_req = -1;
    stats->rd_bytes = -1;
    stats->wr_req = -1;
    stats->wr_bytes = -1;
    stats->errs = -1;

    /* The output format for both qemu & KVM is:
     *   blockdevice: rd_bytes=% wr_bytes=% rd_operations=% wr_operations=%
     *   (repeated for each block device)
     * where '%' is a 64 bit number.
     */
    p = info;

    while (*p) {
        if (STREQLEN (p, qemu_dev_name, len)
            && p[len] == ':' && p[len+1] == ' ') {

            eol = strchr (p, '\n');
            if (!eol)
                eol = p + strlen (p);

            p += len+2;         /* Skip to first label. */

            while (*p) {
6679
                if (STRPREFIX (p, "rd_bytes=")) {
6680 6681
                    p += 9;
                    if (virStrToLong_ll (p, &dummy, 10, &stats->rd_bytes) == -1)
6682 6683
                        DEBUG ("%s: error reading rd_bytes: %s",
                               vm->def->name, p);
6684
                } else if (STRPREFIX (p, "wr_bytes=")) {
6685 6686
                    p += 9;
                    if (virStrToLong_ll (p, &dummy, 10, &stats->wr_bytes) == -1)
6687 6688
                        DEBUG ("%s: error reading wr_bytes: %s",
                               vm->def->name, p);
6689
                } else if (STRPREFIX (p, "rd_operations=")) {
6690 6691
                    p += 14;
                    if (virStrToLong_ll (p, &dummy, 10, &stats->rd_req) == -1)
6692 6693
                        DEBUG ("%s: error reading rd_req: %s",
                               vm->def->name, p);
6694
                } else if (STRPREFIX (p, "wr_operations=")) {
6695 6696
                    p += 14;
                    if (virStrToLong_ll (p, &dummy, 10, &stats->wr_req) == -1)
6697 6698
                        DEBUG ("%s: error reading wr_req: %s",
                               vm->def->name, p);
6699
                } else
6700
                    DEBUG ("%s: unknown block stat near %s", vm->def->name, p);
6701 6702 6703 6704 6705 6706

                /* Skip to next label. */
                p = strchr (p, ' ');
                if (!p || p >= eol) break;
                p++;
            }
6707
            ret = 0;
6708
            goto cleanup;
6709 6710 6711 6712 6713 6714 6715 6716 6717 6718 6719
        }

        /* Skip to next line. */
        p = strchr (p, '\n');
        if (!p) break;
        p++;
    }

    /* If we reach here then the device was not found. */
    qemudReportError (dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
                      _("device not found: %s (%s)"), path, qemu_dev_name);
6720
 cleanup:
6721 6722
    VIR_FREE(qemu_dev_name);
    VIR_FREE(info);
6723 6724
    if (vm)
        virDomainObjUnlock(vm);
6725
    return ret;
6726 6727
}

6728
#ifdef __linux__
6729 6730 6731 6732 6733
static int
qemudDomainInterfaceStats (virDomainPtr dom,
                           const char *path,
                           struct _virDomainInterfaceStats *stats)
{
6734 6735
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
6736
    int i;
6737
    int ret = -1;
6738

6739
    qemuDriverLock(driver);
6740
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
6741 6742
    qemuDriverUnlock(driver);

6743
    if (!vm) {
6744 6745 6746 6747
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching uuid '%s'"), uuidstr);
6748
        goto cleanup;
6749 6750
    }

6751
    if (!virDomainIsActive(vm)) {
6752
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
6753
                         "%s", _("domain is not running"));
6754
        goto cleanup;
6755 6756 6757
    }

    /* Check the path is one of the domain's network interfaces. */
6758 6759
    for (i = 0 ; i < vm->def->nnets ; i++) {
        if (vm->def->nets[i]->ifname &&
6760 6761 6762 6763
            STREQ (vm->def->nets[i]->ifname, path)) {
            ret = 0;
            break;
        }
6764 6765
    }

6766 6767 6768 6769 6770
    if (ret == 0)
        ret = linuxDomainInterfaceStats (dom->conn, path, stats);
    else
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
                          _("invalid path, '%s' is not a known interface"), path);
6771

6772
cleanup:
6773 6774
    if (vm)
        virDomainObjUnlock(vm);
6775 6776
    return ret;
}
6777
#else
6778 6779 6780 6781
static int
qemudDomainInterfaceStats (virDomainPtr dom,
                           const char *path ATTRIBUTE_UNUSED,
                           struct _virDomainInterfaceStats *stats ATTRIBUTE_UNUSED)
6782 6783 6784 6785
    qemudReportError (dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                      "%s", __FUNCTION__);
    return -1;
}
6786
#endif
6787

6788 6789 6790 6791 6792 6793 6794
static int
qemudDomainBlockPeek (virDomainPtr dom,
                      const char *path,
                      unsigned long long offset, size_t size,
                      void *buffer,
                      unsigned int flags ATTRIBUTE_UNUSED)
{
6795 6796 6797
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int fd = -1, ret = -1, i;
6798

6799
    qemuDriverLock(driver);
6800
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
6801 6802
    qemuDriverUnlock(driver);

6803
    if (!vm) {
6804 6805 6806 6807
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching uuid '%s'"), uuidstr);
6808
        goto cleanup;
6809 6810 6811 6812
    }

    if (!path || path[0] == '\0') {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
J
Jim Meyering 已提交
6813
                         "%s", _("NULL or empty path"));
6814
        goto cleanup;
6815 6816 6817
    }

    /* Check the path belongs to this domain. */
6818 6819
    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (vm->def->disks[i]->src != NULL &&
6820 6821 6822 6823
            STREQ (vm->def->disks[i]->src, path)) {
            ret = 0;
            break;
        }
6824 6825
    }

6826 6827 6828 6829 6830
    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) {
6831 6832
            virReportSystemError (dom->conn, errno,
                                  _("%s: failed to open"), path);
6833 6834
            goto cleanup;
        }
6835

6836 6837 6838 6839 6840 6841
        /* 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) {
6842 6843
            virReportSystemError (dom->conn, errno,
                                  _("%s: failed to seek or read"), path);
6844 6845 6846 6847 6848 6849 6850
            goto cleanup;
        }

        ret = 0;
    } else {
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
                          "%s", _("invalid path"));
6851 6852
    }

6853 6854 6855
cleanup:
    if (fd >= 0)
        close (fd);
6856 6857
    if (vm)
        virDomainObjUnlock(vm);
6858 6859 6860
    return ret;
}

R
Richard W.M. Jones 已提交
6861 6862 6863 6864 6865 6866
static int
qemudDomainMemoryPeek (virDomainPtr dom,
                       unsigned long long offset, size_t size,
                       void *buffer,
                       unsigned int flags)
{
6867 6868 6869
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    char cmd[256], *info = NULL;
6870
    char *tmp = NULL;
R
Richard W.M. Jones 已提交
6871 6872
    int fd = -1, ret = -1;

6873
    qemuDriverLock(driver);
6874
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
6875
    qemuDriverUnlock(driver);
R
Richard W.M. Jones 已提交
6876 6877

    if (!vm) {
6878 6879 6880 6881
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching uuid '%s'"), uuidstr);
6882 6883 6884
        goto cleanup;
    }

6885
    if (flags != VIR_MEMORY_VIRTUAL && flags != VIR_MEMORY_PHYSICAL) {
6886
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
6887
                     "%s", _("flags parameter must be VIR_MEMORY_VIRTUAL or VIR_MEMORY_PHYSICAL"));
6888
        goto cleanup;
R
Richard W.M. Jones 已提交
6889 6890
    }

6891
    if (!virDomainIsActive(vm)) {
6892
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
R
Richard W.M. Jones 已提交
6893
                         "%s", _("domain is not running"));
6894
        goto cleanup;
R
Richard W.M. Jones 已提交
6895 6896
    }

6897 6898 6899 6900 6901
    if (virAsprintf(&tmp, driver->cacheDir,  "/qemu.mem.XXXXXX") < 0) {
        virReportOOMError(dom->conn);
        goto cleanup;
    }

R
Richard W.M. Jones 已提交
6902 6903
    /* Create a temporary filename. */
    if ((fd = mkstemp (tmp)) == -1) {
6904 6905
        virReportSystemError (dom->conn, errno,
                              _("mkstemp(\"%s\") failed"), tmp);
6906
        goto cleanup;
R
Richard W.M. Jones 已提交
6907 6908
    }

6909 6910 6911 6912 6913 6914 6915
    if (flags == VIR_MEMORY_VIRTUAL)
        /* Issue the memsave command. */
        snprintf (cmd, sizeof cmd, "memsave %llu %zi \"%s\"", offset, size, tmp);
    else
        /* Issue the pmemsave command. */
        snprintf (cmd, sizeof cmd, "pmemsave %llu %zi \"%s\"", offset, size, tmp);

6916
    if (qemudMonitorCommand (vm, cmd, &info) < 0) {
R
Richard W.M. Jones 已提交
6917
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
6918
                          "%s", _("'memsave' command failed"));
6919
        goto cleanup;
R
Richard W.M. Jones 已提交
6920 6921
    }

6922
    DEBUG ("%s: (p)memsave reply: %s", vm->def->name, info);
R
Richard W.M. Jones 已提交
6923 6924 6925

    /* Read the memory file into buffer. */
    if (saferead (fd, buffer, size) == (ssize_t) -1) {
6926 6927 6928
        virReportSystemError (dom->conn, errno,
                              _("failed to read temporary file "
                                "created with template %s"), tmp);
6929
        goto cleanup;
R
Richard W.M. Jones 已提交
6930 6931 6932
    }

    ret = 0;
6933 6934

cleanup:
6935
    VIR_FREE(tmp);
6936
    VIR_FREE(info);
R
Richard W.M. Jones 已提交
6937 6938
    if (fd >= 0) close (fd);
    unlink (tmp);
6939 6940
    if (vm)
        virDomainObjUnlock(vm);
R
Richard W.M. Jones 已提交
6941 6942 6943
    return ret;
}

6944

6945 6946
static int
qemudDomainEventRegister (virConnectPtr conn,
6947
                          virConnectDomainEventCallback callback,
6948 6949
                          void *opaque,
                          virFreeCallback freecb)
6950
{
6951 6952 6953
    struct qemud_driver *driver = conn->privateData;
    int ret;

6954
    qemuDriverLock(driver);
6955 6956
    ret = virDomainEventCallbackListAdd(conn, driver->domainEventCallbacks,
                                        callback, opaque, freecb);
6957
    qemuDriverUnlock(driver);
6958

6959
    return ret;
6960 6961 6962 6963
}

static int
qemudDomainEventDeregister (virConnectPtr conn,
6964
                            virConnectDomainEventCallback callback)
6965
{
6966 6967 6968
    struct qemud_driver *driver = conn->privateData;
    int ret;

6969
    qemuDriverLock(driver);
6970 6971 6972 6973 6974 6975
    if (driver->domainEventDispatching)
        ret = virDomainEventCallbackListMarkDelete(conn, driver->domainEventCallbacks,
                                                   callback);
    else
        ret = virDomainEventCallbackListRemove(conn, driver->domainEventCallbacks,
                                               callback);
6976
    qemuDriverUnlock(driver);
6977

6978
    return ret;
6979 6980
}

6981 6982 6983 6984 6985
static void qemuDomainEventDispatchFunc(virConnectPtr conn,
                                        virDomainEventPtr event,
                                        virConnectDomainEventCallback cb,
                                        void *cbopaque,
                                        void *opaque)
6986
{
6987
    struct qemud_driver *driver = opaque;
6988

6989 6990 6991 6992 6993 6994 6995 6996 6997 6998 6999 7000 7001 7002 7003 7004 7005 7006 7007 7008 7009 7010 7011 7012 7013 7014 7015 7016 7017 7018 7019 7020 7021 7022 7023 7024 7025 7026 7027 7028 7029 7030 7031 7032
    /* 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);
7033 7034
}

D
Daniel Veillard 已提交
7035 7036 7037 7038 7039 7040 7041 7042 7043 7044 7045 7046 7047 7048 7049 7050 7051 7052
/* Migration support. */

/* Prepare is the first step, and it runs on the destination host.
 *
 * This starts an empty VM listening on a TCP port.
 */
static int
qemudDomainMigratePrepare2 (virConnectPtr dconn,
                            char **cookie ATTRIBUTE_UNUSED,
                            int *cookielen ATTRIBUTE_UNUSED,
                            const char *uri_in,
                            char **uri_out,
                            unsigned long flags ATTRIBUTE_UNUSED,
                            const char *dname,
                            unsigned long resource ATTRIBUTE_UNUSED,
                            const char *dom_xml)
{
    static int port = 0;
7053 7054
    struct qemud_driver *driver = dconn->privateData;
    virDomainDefPtr def = NULL;
D
Daniel Veillard 已提交
7055 7056
    virDomainObjPtr vm = NULL;
    int this_port;
7057
    char *hostname;
D
Daniel Veillard 已提交
7058 7059
    char migrateFrom [64];
    const char *p;
7060
    virDomainEventPtr event = NULL;
7061
    int ret = -1;
7062
    int internalret;
7063 7064

    *uri_out = NULL;
D
Daniel Veillard 已提交
7065

7066
    qemuDriverLock(driver);
D
Daniel Veillard 已提交
7067 7068 7069
    if (!dom_xml) {
        qemudReportError (dconn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                          "%s", _("no domain XML passed"));
7070
        goto cleanup;
D
Daniel Veillard 已提交
7071 7072 7073 7074 7075 7076 7077 7078 7079 7080 7081 7082 7083 7084 7085 7086 7087
    }

    /* 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 */
7088
        if ((hostname = virGetHostname()) == NULL) {
7089 7090
            virReportSystemError (dconn, errno,
                                  "%s", _("failed to determine host name"));
7091
            goto cleanup;
D
Daniel Veillard 已提交
7092 7093 7094
        }

        /* Caller frees */
7095 7096 7097
        internalret = virAsprintf(uri_out, "tcp:%s:%d", hostname, this_port);
        VIR_FREE(hostname);
        if (internalret < 0) {
7098
            virReportOOMError (dconn);
7099
            goto cleanup;
D
Daniel Veillard 已提交
7100 7101 7102 7103 7104 7105
        }
    } 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.
         */
7106
        if (!STRPREFIX (uri_in, "tcp:")) {
D
Daniel Veillard 已提交
7107
            qemudReportError (dconn, NULL, NULL, VIR_ERR_INVALID_ARG,
7108
                  "%s", _("only tcp URIs are supported for KVM/QEMU migrations"));
7109
            goto cleanup;
D
Daniel Veillard 已提交
7110 7111 7112 7113 7114 7115 7116 7117 7118
        }

        /* Get the port number. */
        p = strrchr (uri_in, ':');
        p++; /* definitely has a ':' in it, see above */
        this_port = virParseNumber (&p);
        if (this_port == -1 || p-uri_in != strlen (uri_in)) {
            qemudReportError (dconn, NULL, NULL, VIR_ERR_INVALID_ARG,
                              "%s", _("URI did not have ':port' at the end"));
7119
            goto cleanup;
D
Daniel Veillard 已提交
7120 7121 7122 7123
        }
    }

    /* Parse the domain XML. */
7124 7125
    if (!(def = virDomainDefParseString(dconn, driver->caps, dom_xml,
                                        VIR_DOMAIN_XML_INACTIVE))) {
D
Daniel Veillard 已提交
7126 7127
        qemudReportError (dconn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                          "%s", _("failed to parse XML"));
7128
        goto cleanup;
D
Daniel Veillard 已提交
7129 7130 7131 7132 7133 7134 7135 7136 7137 7138 7139 7140 7141 7142 7143 7144
    }

    /* Target domain name, maybe renamed. */
    dname = dname ? dname : def->name;

#if 1
    /* Ensure the name and UUID don't already exist in an active VM */
    vm = virDomainFindByUUID(&driver->domains, def->uuid);
#else
    /* For TESTING ONLY you can change #if 1 -> #if 0 above and use
     * this code which lets you do localhost migrations.  You must still
     * supply a fresh 'dname' but this code assigns a random UUID.
     */
    if (virUUIDGenerate (def->uuid) == -1) {
        qemudReportError (dconn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
            _("could not generate random UUID"));
7145
        goto cleanup;
D
Daniel Veillard 已提交
7146 7147 7148 7149 7150 7151 7152 7153 7154
    }
#endif

    if (!vm) vm = virDomainFindByName(&driver->domains, dname);
    if (vm) {
        if (virDomainIsActive(vm)) {
            qemudReportError (dconn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                              _("domain with the same name or UUID already exists as '%s'"),
                              vm->def->name);
7155
            goto cleanup;
D
Daniel Veillard 已提交
7156
        }
7157
        virDomainObjUnlock(vm);
D
Daniel Veillard 已提交
7158 7159 7160 7161 7162 7163 7164
    }

    if (!(vm = virDomainAssignDef(dconn,
                                  &driver->domains,
                                  def))) {
        qemudReportError (dconn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                          "%s", _("failed to assign new VM"));
7165
        goto cleanup;
D
Daniel Veillard 已提交
7166
    }
7167
    def = NULL;
D
Daniel Veillard 已提交
7168 7169 7170 7171 7172 7173 7174 7175

    /* 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);
7176
    if (qemudStartVMDaemon (dconn, driver, vm, migrateFrom, -1) < 0) {
7177 7178 7179
        /* Note that we don't set an error here because qemudStartVMDaemon
         * should have already done that.
         */
7180
        if (!vm->persistent) {
D
Daniel Veillard 已提交
7181
            virDomainRemoveInactive(&driver->domains, vm);
7182 7183
            vm = NULL;
        }
7184
        goto cleanup;
D
Daniel Veillard 已提交
7185
    }
7186 7187 7188 7189

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_MIGRATED);
7190
    ret = 0;
D
Daniel Veillard 已提交
7191

7192 7193 7194 7195 7196
cleanup:
    virDomainDefFree(def);
    if (ret != 0) {
        VIR_FREE(*uri_out);
    }
7197 7198
    if (vm)
        virDomainObjUnlock(vm);
7199 7200
    if (event)
        qemuDomainEventQueue(driver, event);
7201
    qemuDriverUnlock(driver);
7202
    return ret;
D
Daniel Veillard 已提交
7203 7204 7205 7206 7207 7208 7209 7210
}

/* 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,
7211
                           unsigned long flags,
D
Daniel Veillard 已提交
7212 7213 7214
                           const char *dname ATTRIBUTE_UNUSED,
                           unsigned long resource)
{
7215 7216
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
7217
    virDomainEventPtr event = NULL;
D
Daniel Veillard 已提交
7218 7219
    char *safe_uri;
    char cmd[HOST_NAME_MAX+50];
7220 7221
    char *info = NULL;
    int ret = -1;
7222
    int paused = 0;
D
Daniel Veillard 已提交
7223

7224
    qemuDriverLock(driver);
7225
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
D
Daniel Veillard 已提交
7226
    if (!vm) {
7227 7228 7229 7230
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching uuid '%s'"), uuidstr);
7231
        goto cleanup;
D
Daniel Veillard 已提交
7232 7233 7234
    }

    if (!virDomainIsActive(vm)) {
7235
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
D
Daniel Veillard 已提交
7236
                          "%s", _("domain is not running"));
7237
        goto cleanup;
D
Daniel Veillard 已提交
7238 7239
    }

7240 7241
    if (!(flags & VIR_MIGRATE_LIVE)) {
        /* Pause domain for non-live migration */
7242 7243 7244 7245 7246
        if (qemudMonitorCommand (vm, "stop", &info) < 0) {
            qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                             "%s", _("off-line migration specified, but suspend operation failed"));
            goto cleanup;
        }
7247
        DEBUG ("%s: stop reply: %s", vm->def->name, info);
7248
        VIR_FREE(info);
7249
        paused = 1;
7250

7251 7252 7253 7254 7255 7256
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_SUSPENDED,
                                         VIR_DOMAIN_EVENT_SUSPENDED_MIGRATED);
        if (event)
            qemuDomainEventQueue(driver, event);
        event = NULL;
7257 7258
    }

D
Daniel Veillard 已提交
7259 7260 7261
    if (resource > 0) {
        /* Issue migrate_set_speed command.  Don't worry if it fails. */
        snprintf (cmd, sizeof cmd, "migrate_set_speed %lum", resource);
7262
        qemudMonitorCommand (vm, cmd, &info);
D
Daniel Veillard 已提交
7263

7264
        DEBUG ("%s: migrate_set_speed reply: %s", vm->def->name, info);
D
Daniel Veillard 已提交
7265 7266 7267 7268 7269 7270
        VIR_FREE (info);
    }

    /* Issue the migrate command. */
    safe_uri = qemudEscapeMonitorArg (uri);
    if (!safe_uri) {
7271
        virReportOOMError (dom->conn);
7272
        goto cleanup;
D
Daniel Veillard 已提交
7273 7274 7275 7276
    }
    snprintf (cmd, sizeof cmd, "migrate \"%s\"", safe_uri);
    VIR_FREE (safe_uri);

7277
    if (qemudMonitorCommand (vm, cmd, &info) < 0) {
D
Daniel Veillard 已提交
7278 7279
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                          "%s", _("migrate operation failed"));
7280
        goto cleanup;
D
Daniel Veillard 已提交
7281 7282
    }

7283
    DEBUG ("%s: migrate reply: %s", vm->def->name, info);
D
Daniel Veillard 已提交
7284 7285 7286 7287 7288

    /* Now check for "fail" in the output string */
    if (strstr(info, "fail") != NULL) {
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                          _("migrate failed: %s"), info);
7289
        goto cleanup;
D
Daniel Veillard 已提交
7290 7291
    }

7292 7293 7294 7295 7296 7297 7298 7299 7300 7301 7302 7303 7304 7305 7306
    /* it is also possible that the migrate didn't fail initially, but
     * rather failed later on.  Check the output of "info migrate"
     */
    VIR_FREE(info);
    if (qemudMonitorCommand(vm, "info migrate", &info) < 0) {
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                          "%s", _("could not get info about migration"));
        goto cleanup;
    }
    if (strstr(info, "fail") != NULL) {
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                          _("migrate failed: %s"), info);
        goto cleanup;
    }

D
Daniel Veillard 已提交
7307 7308
    /* Clean up the source domain. */
    qemudShutdownVMDaemon (dom->conn, driver, vm);
7309
    paused = 0;
7310 7311 7312 7313

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_MIGRATED);
7314
    if (!vm->persistent) {
D
Daniel Veillard 已提交
7315
        virDomainRemoveInactive(&driver->domains, vm);
7316 7317
        vm = NULL;
    }
7318
    ret = 0;
D
Daniel Veillard 已提交
7319

7320
cleanup:
7321 7322
    if (paused) {
        /* we got here through some sort of failure; start the domain again */
7323
        if (qemudMonitorSendCont(dom->conn, vm) < 0) {
7324 7325 7326 7327
            /* 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
             */
7328 7329
            VIR_ERROR(_("Failed to resume guest %s after failure\n"),
                      vm->def->name);
7330 7331 7332 7333 7334 7335 7336
        }

        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_MIGRATED);
    }

7337
    VIR_FREE(info);
7338 7339
    if (vm)
        virDomainObjUnlock(vm);
7340 7341
    if (event)
        qemuDomainEventQueue(driver, event);
7342
    qemuDriverUnlock(driver);
7343
    return ret;
D
Daniel Veillard 已提交
7344 7345 7346 7347 7348 7349 7350 7351 7352 7353 7354 7355
}

/* 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,
                           unsigned long flags ATTRIBUTE_UNUSED,
                           int retcode)
{
7356 7357 7358
    struct qemud_driver *driver = dconn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
7359
    virDomainEventPtr event = NULL;
D
Daniel Veillard 已提交
7360

7361
    qemuDriverLock(driver);
7362
    vm = virDomainFindByName(&driver->domains, dname);
D
Daniel Veillard 已提交
7363
    if (!vm) {
7364 7365
        qemudReportError (dconn, NULL, NULL, VIR_ERR_NO_DOMAIN,
                          _("no domain with matching name '%s'"), dname);
7366
        goto cleanup;
D
Daniel Veillard 已提交
7367 7368 7369 7370 7371 7372 7373
    }

    /* Did the migration go as planned?  If yes, return the domain
     * object, but if no, clean up the empty qemu process.
     */
    if (retcode == 0) {
        dom = virGetDomain (dconn, vm->def->name, vm->def->uuid);
7374 7375 7376 7377 7378

        /* 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
         */
7379 7380 7381 7382
        if (qemudMonitorSendCont(dconn, vm) < 0) {
            if (virGetLastError() == NULL)
                qemudReportError(dconn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                                 "%s", _("resume operation failed"));
7383 7384 7385
            goto cleanup;
        }

D
Daniel Veillard 已提交
7386
        vm->state = VIR_DOMAIN_RUNNING;
7387 7388 7389
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_MIGRATED);
D
Daniel Veillard 已提交
7390
        virDomainSaveStatus(dconn, driver->stateDir, vm);
D
Daniel Veillard 已提交
7391 7392
    } else {
        qemudShutdownVMDaemon (dconn, driver, vm);
7393 7394 7395
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         VIR_DOMAIN_EVENT_STOPPED_FAILED);
7396
        if (!vm->persistent) {
D
Daniel Veillard 已提交
7397
            virDomainRemoveInactive(&driver->domains, vm);
7398 7399
            vm = NULL;
        }
D
Daniel Veillard 已提交
7400
    }
7401 7402

cleanup:
7403 7404
    if (vm)
        virDomainObjUnlock(vm);
7405 7406
    if (event)
        qemuDomainEventQueue(driver, event);
7407
    qemuDriverUnlock(driver);
7408
    return dom;
D
Daniel Veillard 已提交
7409 7410
}

7411 7412 7413 7414 7415 7416 7417 7418 7419 7420 7421 7422 7423 7424 7425 7426
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;

7427
    def = virNodeDeviceDefParseString(dev->conn, xml, EXISTING_DEVICE);
7428 7429 7430 7431 7432 7433 7434 7435 7436 7437 7438 7439 7440 7441 7442 7443 7444 7445 7446 7447 7448 7449 7450 7451 7452 7453 7454 7455 7456 7457 7458 7459 7460 7461 7462 7463 7464 7465 7466 7467 7468 7469 7470 7471 7472 7473 7474 7475 7476 7477 7478 7479 7480 7481 7482 7483 7484 7485 7486 7487 7488 7489 7490 7491 7492 7493 7494 7495 7496 7497 7498 7499 7500 7501 7502 7503 7504 7505
    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) {
        qemudReportError(dev->conn, NULL, NULL, VIR_ERR_INVALID_ARG,
                         _("device %s is not a PCI device"), dev->name);
        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;

    pci = pciGetDevice(dev->conn, domain, bus, slot, function);
    if (!pci)
        return -1;

    if (pciDettachDevice(dev->conn, pci) < 0)
        goto out;

    ret = 0;
out:
    pciFreeDevice(dev->conn, pci);
    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;

    pci = pciGetDevice(dev->conn, domain, bus, slot, function);
    if (!pci)
        return -1;

    if (pciReAttachDevice(dev->conn, pci) < 0)
        goto out;

    ret = 0;
out:
    pciFreeDevice(dev->conn, pci);
    return ret;
}

static int
qemudNodeDeviceReset (virNodeDevicePtr dev)
{
7506
    struct qemud_driver *driver = dev->conn->privateData;
7507 7508 7509 7510 7511 7512 7513 7514 7515 7516 7517
    pciDevice *pci;
    unsigned domain, bus, slot, function;
    int ret = -1;

    if (qemudNodeDeviceGetPciInfo(dev, &domain, &bus, &slot, &function) < 0)
        return -1;

    pci = pciGetDevice(dev->conn, domain, bus, slot, function);
    if (!pci)
        return -1;

7518 7519 7520
    qemuDriverLock(driver);

    if (pciResetDevice(dev->conn, pci, driver->activePciHostdevs) < 0)
7521 7522 7523 7524
        goto out;

    ret = 0;
out:
7525
    qemuDriverUnlock(driver);
7526 7527 7528 7529
    pciFreeDevice(dev->conn, pci);
    return ret;
}

7530 7531 7532 7533 7534
static virDriver qemuDriver = {
    VIR_DRV_QEMU,
    "QEMU",
    qemudOpen, /* open */
    qemudClose, /* close */
D
Daniel Veillard 已提交
7535
    qemudSupportsFeature, /* supports_feature */
7536 7537
    qemudGetType, /* type */
    qemudGetVersion, /* version */
7538
    qemudGetHostname, /* getHostname */
7539
    qemudGetMaxVCPUs, /* getMaxVcpus */
7540
    nodeGetInfo, /* nodeGetInfo */
7541 7542 7543
    qemudGetCapabilities, /* getCapabilities */
    qemudListDomains, /* listDomains */
    qemudNumDomains, /* numOfDomains */
7544
    qemudDomainCreate, /* domainCreateXML */
7545 7546 7547 7548 7549
    qemudDomainLookupByID, /* domainLookupByID */
    qemudDomainLookupByUUID, /* domainLookupByUUID */
    qemudDomainLookupByName, /* domainLookupByName */
    qemudDomainSuspend, /* domainSuspend */
    qemudDomainResume, /* domainResume */
7550
    qemudDomainShutdown, /* domainShutdown */
7551 7552 7553
    NULL, /* domainReboot */
    qemudDomainDestroy, /* domainDestroy */
    qemudDomainGetOSType, /* domainGetOSType */
7554 7555 7556
    qemudDomainGetMaxMemory, /* domainGetMaxMemory */
    qemudDomainSetMaxMemory, /* domainSetMaxMemory */
    qemudDomainSetMemory, /* domainSetMemory */
7557 7558 7559
    qemudDomainGetInfo, /* domainGetInfo */
    qemudDomainSave, /* domainSave */
    qemudDomainRestore, /* domainRestore */
P
Paolo Bonzini 已提交
7560
    qemudDomainCoreDump, /* domainCoreDump */
7561
    qemudDomainSetVcpus, /* domainSetVcpus */
7562 7563 7564 7565
#if HAVE_SCHED_GETAFFINITY
    qemudDomainPinVcpu, /* domainPinVcpu */
    qemudDomainGetVcpus, /* domainGetVcpus */
#else
7566 7567
    NULL, /* domainPinVcpu */
    NULL, /* domainGetVcpus */
7568
#endif
7569
    qemudDomainGetMaxVcpus, /* domainGetMaxVcpus */
7570 7571
    qemudDomainGetSecurityLabel, /* domainGetSecurityLabel */
    qemudNodeGetSecurityModel, /* nodeGetSecurityModel */
7572
    qemudDomainDumpXML, /* domainDumpXML */
7573
    qemuDomainXMLFromNative, /* domainXmlFromNative */
7574
    qemuDomainXMLToNative, /* domainXMLToNative */
7575 7576
    qemudListDefinedDomains, /* listDefinedDomains */
    qemudNumDefinedDomains, /* numOfDefinedDomains */
7577 7578 7579
    qemudDomainStart, /* domainCreate */
    qemudDomainDefine, /* domainDefineXML */
    qemudDomainUndefine, /* domainUndefine */
7580
    qemudDomainAttachDevice, /* domainAttachDevice */
7581
    qemudDomainDetachDevice, /* domainDetachDevice */
7582 7583
    qemudDomainGetAutostart, /* domainGetAutostart */
    qemudDomainSetAutostart, /* domainSetAutostart */
7584 7585 7586
    qemuGetSchedulerType, /* domainGetSchedulerType */
    qemuGetSchedulerParameters, /* domainGetSchedulerParameters */
    qemuSetSchedulerParameters, /* domainSetSchedulerParameters */
D
Daniel Veillard 已提交
7587 7588
    NULL, /* domainMigratePrepare (v1) */
    qemudDomainMigratePerform, /* domainMigratePerform */
7589
    NULL, /* domainMigrateFinish */
7590
    qemudDomainBlockStats, /* domainBlockStats */
7591
    qemudDomainInterfaceStats, /* domainInterfaceStats */
7592
    qemudDomainBlockPeek, /* domainBlockPeek */
R
Richard W.M. Jones 已提交
7593
    qemudDomainMemoryPeek, /* domainMemoryPeek */
7594 7595
    nodeGetCellsFreeMemory, /* nodeGetCellsFreeMemory */
    nodeGetFreeMemory,  /* getFreeMemory */
7596 7597
    qemudDomainEventRegister, /* domainEventRegister */
    qemudDomainEventDeregister, /* domainEventDeregister */
D
Daniel Veillard 已提交
7598 7599
    qemudDomainMigratePrepare2, /* domainMigratePrepare2 */
    qemudDomainMigrateFinish2, /* domainMigrateFinish2 */
7600 7601 7602
    qemudNodeDeviceDettach, /* nodeDeviceDettach */
    qemudNodeDeviceReAttach, /* nodeDeviceReAttach */
    qemudNodeDeviceReset, /* nodeDeviceReset */
7603 7604 7605
};


7606
static virStateDriver qemuStateDriver = {
7607 7608 7609 7610
    .initialize = qemudStartup,
    .cleanup = qemudShutdown,
    .reload = qemudReload,
    .active = qemudActive,
7611
};
7612

7613
int qemuRegister(void) {
7614 7615 7616 7617
    virRegisterDriver(&qemuDriver);
    virRegisterStateDriver(&qemuStateDriver);
    return 0;
}