qemu_driver.c 194.6 KB
Newer Older
D
Daniel P. Berrange 已提交
1 2 3
/*
 * driver.c: core driver methods for managing qemu guests
 *
4
 * Copyright (C) 2006, 2007, 2008, 2009 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"
58
#include "qemu_monitor_text.h"
J
Jim Meyering 已提交
59
#include "c-ctype.h"
60
#include "event.h"
61
#include "buf.h"
62
#include "util.h"
63
#include "nodeinfo.h"
64
#include "stats_linux.h"
65
#include "capabilities.h"
66
#include "memory.h"
67
#include "uuid.h"
68
#include "domain_conf.h"
69 70
#include "node_device_conf.h"
#include "pci.h"
71
#include "hostusb.h"
72
#include "security/security_driver.h"
73
#include "cgroup.h"
74

75

76 77
#define VIR_FROM_THIS VIR_FROM_QEMU

78 79
static int qemudShutdown(void);

80 81
static void qemuDriverLock(struct qemud_driver *driver)
{
82
    virMutexLock(&driver->lock);
83 84 85
}
static void qemuDriverUnlock(struct qemud_driver *driver)
{
86
    virMutexUnlock(&driver->lock);
87 88
}

89 90 91
static void qemuDomainEventFlush(int timer, void *opaque);
static void qemuDomainEventQueue(struct qemud_driver *driver,
                                 virDomainEventPtr event);
92

93 94
static void qemudDispatchVMEvent(int watch,
                                 int fd,
95 96 97
                                 int events,
                                 void *opaque);

98 99
static int qemudStartVMDaemon(virConnectPtr conn,
                              struct qemud_driver *driver,
100
                              virDomainObjPtr vm,
101 102
                              const char *migrateFrom,
                              int stdin_fd);
103

104 105
static void qemudShutdownVMDaemon(virConnectPtr conn,
                                  struct qemud_driver *driver,
106
                                  virDomainObjPtr vm);
107

108
static int qemudDomainGetMaxVcpus(virDomainPtr dom);
109

110 111
static int qemuDetectVcpuPIDs(virConnectPtr conn,
                              virDomainObjPtr vm);
112

113 114 115
static int qemuUpdateActivePciHostdevs(struct qemud_driver *driver,
                                       virDomainDefPtr def);

J
Jim Meyering 已提交
116
static struct qemud_driver *qemu_driver = NULL;
117

118 119 120 121 122 123 124 125 126
static int qemuCgroupControllerActive(struct qemud_driver *driver,
                                      int controller)
{
    if (driver->cgroup == NULL)
        return 0;
    if (driver->cgroupControllers & (1 << controller))
        return 1;
    return 0;
}
127

128
static int
129
qemudLogFD(virConnectPtr conn, struct qemud_driver *driver, const char* name)
130 131 132
{
    char logfile[PATH_MAX];
    mode_t logmode;
G
Guido Günther 已提交
133
    int ret, fd = -1;
134

135 136
    if ((ret = snprintf(logfile, sizeof(logfile), "%s/%s.log",
                        driver->logDir, name))
G
Guido Günther 已提交
137
        < 0 || ret >= sizeof(logfile)) {
138
        virReportOOMError(conn);
139 140 141 142
        return -1;
    }

    logmode = O_CREAT | O_WRONLY;
143 144
    /* Only logrotate files in /var/log, so only append if running privileged */
    if (driver->privileged)
145
        logmode |= O_APPEND;
146 147 148
    else
        logmode |= O_TRUNC;

149
    if ((fd = open(logfile, logmode, S_IRUSR | S_IWUSR)) < 0) {
150 151 152
        virReportSystemError(conn, errno,
                             _("failed to create logfile %s"),
                             logfile);
153 154
        return -1;
    }
155
    if (virSetCloseExec(fd) < 0) {
156 157
        virReportSystemError(conn, errno, "%s",
                             _("Unable to set VM logfile close-on-exec flag"));
158 159 160 161 162 163 164
        close(fd);
        return -1;
    }
    return fd;
}


165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181
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) {
182 183 184
        virReportSystemError(conn, errno,
                             _("failed to create logfile %s"),
                             logfile);
185 186
        return -1;
    }
187
    if (virSetCloseExec(fd) < 0) {
188 189
        virReportSystemError(conn, errno, "%s",
                             _("Unable to set VM logfile close-on-exec flag"));
190 191 192 193
        close(fd);
        return -1;
    }
    if (lseek(fd, pos, SEEK_SET) < 0) {
194 195 196
        virReportSystemError(conn, errno,
                             _("Unable to seek to %lld in %s"),
                             (long long) pos, logfile);
197 198 199 200 201 202
        close(fd);
    }
    return fd;
}


203 204 205
static void
qemudAutostartConfigs(struct qemud_driver *driver) {
    unsigned int i;
206 207 208 209 210
    /* 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
     */
211 212 213
    virConnectPtr conn = virConnectOpen(driver->privileged ?
                                        "qemu:///system" :
                                        "qemu:///session");
214
    /* Ignoring NULL conn which is mostly harmless here */
215

216
    qemuDriverLock(driver);
217
    for (i = 0 ; i < driver->domains.count ; i++) {
218
        virDomainObjPtr vm = driver->domains.objs[i];
219
        virDomainObjLock(vm);
220 221
        if (vm->autostart &&
            !virDomainIsActive(vm)) {
222 223 224 225
            int ret;

            virResetLastError();
            ret = qemudStartVMDaemon(conn, driver, vm, NULL, -1);
226 227
            if (ret < 0) {
                virErrorPtr err = virGetLastError();
228 229
                VIR_ERROR(_("Failed to autostart VM '%s': %s\n"),
                          vm->def->name,
230
                          err ? err->message : "");
231
            } else {
232 233 234 235 236 237
                virDomainEventPtr event =
                    virDomainEventNewFromObj(vm,
                                             VIR_DOMAIN_EVENT_STARTED,
                                             VIR_DOMAIN_EVENT_STARTED_BOOTED);
                if (event)
                    qemuDomainEventQueue(driver, event);
238
            }
239
        }
240
        virDomainObjUnlock(vm);
241
    }
242
    qemuDriverUnlock(driver);
243

244 245
    if (conn)
        virConnectClose(conn);
246 247
}

248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264

/**
 * 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) {
265
        virReportOOMError(conn);
266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284
        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;
}


285 286 287 288 289
static int qemudOpenMonitor(virConnectPtr conn,
                            struct qemud_driver* driver,
                            virDomainObjPtr vm,
                            int reconnect);

290 291 292 293

/*
 * Open an existing VM's monitor, re-detect VCPU threads
 * and re-reserve the security labels in use
294 295
 */
static int
296 297
qemuReconnectDomain(struct qemud_driver *driver,
                    virDomainObjPtr obj)
298
{
299
    int rc;
300

301
    if ((rc = qemudOpenMonitor(NULL, driver, obj, 1)) != 0) {
302 303 304 305
        VIR_ERROR(_("Failed to reconnect monitor for %s: %d\n"),
                  obj->def->name, rc);
        goto error;
    }
306

307 308 309 310
    if (qemuUpdateActivePciHostdevs(driver, obj->def) < 0) {
        goto error;
    }

311 312 313 314 315
    if (obj->def->seclabel.type == VIR_DOMAIN_SECLABEL_DYNAMIC &&
        driver->securityDriver &&
        driver->securityDriver->domainReserveSecurityLabel &&
        driver->securityDriver->domainReserveSecurityLabel(NULL, obj) < 0)
        return -1;
316

317 318
    if (obj->def->id >= driver->nextvmid)
        driver->nextvmid = obj->def->id + 1;
319

320
    return 0;
321

322 323 324
error:
    return -1;
}
325

326 327 328 329 330 331 332 333 334 335
/**
 * qemudReconnectVMs
 *
 * Try to re-open the resources for live VMs that we care
 * about.
 */
static void
qemuReconnectDomains(struct qemud_driver *driver)
{
    int i;
336

337 338 339 340 341 342 343 344 345
    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);
346
        }
347
        virDomainObjUnlock(obj);
348 349 350
    }
}

351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383

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;
}


384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403
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;

404
    VIR_INFO("Initialized security driver %s", security_drv->name);
405 406 407 408 409

    /*
     * Add security policy host caps now that the security driver is
     * initialized.
     */
410 411
    return qemudSecurityCapsInit(security_drv, qemud_drv->caps);
}
412 413


414

415 416 417 418 419 420
/**
 * qemudStartup:
 *
 * Initialization function for the QEmu daemon
 */
static int
421
qemudStartup(int privileged) {
422
    char *base = NULL;
D
Daniel P. Berrange 已提交
423
    char driverConf[PATH_MAX];
424
    int rc;
425

426
    if (VIR_ALLOC(qemu_driver) < 0)
427 428
        return -1;

429
    if (virMutexInit(&qemu_driver->lock) < 0) {
430
        VIR_ERROR("%s", _("cannot initialize mutex"));
431 432 433
        VIR_FREE(qemu_driver);
        return -1;
    }
434
    qemuDriverLock(qemu_driver);
435
    qemu_driver->privileged = privileged;
436

437 438 439
    /* Don't have a dom0 so start from 1 */
    qemu_driver->nextvmid = 1;

440 441
    /* Init callback list */
    if(VIR_ALLOC(qemu_driver->domainEventCallbacks) < 0)
442
        goto out_of_memory;
443 444 445 446 447 448
    if (!(qemu_driver->domainEventQueue = virDomainEventQueueNew()))
        goto out_of_memory;

    if ((qemu_driver->domainEventTimer =
         virEventAddTimeout(-1, qemuDomainEventFlush, qemu_driver, NULL)) < 0)
        goto error;
449

450
    if (privileged) {
451 452
        if (virAsprintf(&qemu_driver->logDir,
                        "%s/log/libvirt/qemu", LOCAL_STATE_DIR) == -1)
453
            goto out_of_memory;
454

D
Daniel P. Berrange 已提交
455
        if ((base = strdup (SYSCONF_DIR "/libvirt")) == NULL)
456
            goto out_of_memory;
457 458

        if (virAsprintf(&qemu_driver->stateDir,
459
                      "%s/run/libvirt/qemu", LOCAL_STATE_DIR) == -1)
460
            goto out_of_memory;
461 462 463 464 465 466 467 468

        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;
469
    } else {
470
        uid_t uid = geteuid();
471 472
        char *userdir = virGetUserDirectory(NULL, uid);
        if (!userdir)
473
            goto error;
474

475
        if (virAsprintf(&qemu_driver->logDir,
476 477
                        "%s/.libvirt/qemu/log", userdir) == -1) {
            VIR_FREE(userdir);
478
            goto out_of_memory;
479
        }
480

481 482
        if (virAsprintf(&base, "%s/.libvirt", userdir) == -1) {
            VIR_FREE(userdir);
483
            goto out_of_memory;
484 485
        }
        VIR_FREE(userdir);
486 487 488

        if (virAsprintf(&qemu_driver->stateDir, "%s/qemu/run", base) == -1)
            goto out_of_memory;
489 490 491 492
        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;
493 494 495
    }

    if (virFileMakePath(qemu_driver->stateDir) < 0) {
496
        char ebuf[1024];
497 498
        VIR_ERROR(_("Failed to create state dir '%s': %s\n"),
                  qemu_driver->stateDir, virStrerror(errno, ebuf, sizeof ebuf));
499
        goto error;
500
    }
501 502 503 504 505 506 507 508 509 510 511 512
    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;
    }
513 514 515 516

    /* Configuration paths are either ~/.libvirt/qemu/... (session) or
     * /etc/libvirt/qemu/... (system).
     */
D
Daniel P. Berrange 已提交
517
    if (snprintf (driverConf, sizeof(driverConf), "%s/qemu.conf", base) == -1)
518
        goto out_of_memory;
D
Daniel P. Berrange 已提交
519
    driverConf[sizeof(driverConf)-1] = '\0';
520

521
    if (virAsprintf(&qemu_driver->configDir, "%s/qemu", base) == -1)
522 523
        goto out_of_memory;

524
    if (virAsprintf(&qemu_driver->autostartDir, "%s/qemu/autostart", base) == -1)
525 526
        goto out_of_memory;

527
    VIR_FREE(base);
528

529 530 531 532 533 534 535
    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)));
    }

536
    if ((qemu_driver->caps = qemudCapsInit(NULL)) == NULL)
537
        goto out_of_memory;
D
Daniel P. Berrange 已提交
538

539 540 541
    if ((qemu_driver->activePciHostdevs = pciDeviceListNew(NULL)) == NULL)
        goto error;

542
    if (qemudLoadDriverConfig(qemu_driver, driverConf) < 0) {
543 544 545
        goto error;
    }

546 547 548 549 550 551 552 553 554 555 556 557 558 559 560
    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;
        }
    }

561
    if (qemudSecurityInit(qemu_driver) < 0) {
562
        goto error;
D
Daniel P. Berrange 已提交
563 564
    }

565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595
    /* 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;
    }

596 597 598 599 600 601 602 603 604 605 606 607
    /* 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 */
608 609 610 611
    if (virDomainLoadAllConfigs(NULL,
                                qemu_driver->caps,
                                &qemu_driver->domains,
                                qemu_driver->configDir,
612
                                qemu_driver->autostartDir,
613
                                0, NULL, NULL) < 0)
614
        goto error;
615 616
    qemuDriverUnlock(qemu_driver);

617 618
    qemudAutostartConfigs(qemu_driver);

619

620 621
    return 0;

622
out_of_memory:
623
    virReportOOMError(NULL);
624 625 626
error:
    if (qemu_driver)
        qemuDriverUnlock(qemu_driver);
627
    VIR_FREE(base);
628
    qemudShutdown();
629 630 631
    return -1;
}

632 633 634 635
static void qemudNotifyLoadDomain(virDomainObjPtr vm, int newVM, void *opaque)
{
    struct qemud_driver *driver = opaque;

636 637 638 639 640 641 642 643
    if (newVM) {
        virDomainEventPtr event =
            virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_DEFINED,
                                     VIR_DOMAIN_EVENT_DEFINED_ADDED);
        if (event)
            qemuDomainEventQueue(driver, event);
    }
644 645
}

646 647 648 649 650 651 652 653
/**
 * qemudReload:
 *
 * Function to restart the QEmu daemon, it will recheck the configuration
 * files and update its state and the networking
 */
static int
qemudReload(void) {
654 655 656
    if (!qemu_driver)
        return 0;

657
    qemuDriverLock(qemu_driver);
658 659 660 661
    virDomainLoadAllConfigs(NULL,
                            qemu_driver->caps,
                            &qemu_driver->domains,
                            qemu_driver->configDir,
662
                            qemu_driver->autostartDir,
663
                            0, qemudNotifyLoadDomain, qemu_driver);
664
    qemuDriverUnlock(qemu_driver);
665

666
    qemudAutostartConfigs(qemu_driver);
667 668

    return 0;
669 670
}

671 672 673 674 675 676 677 678 679 680
/**
 * 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) {
681
    unsigned int i;
682
    int active = 0;
683

684 685 686
    if (!qemu_driver)
        return 0;

687 688 689 690 691 692 693 694
    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);
    }
695

696 697
    qemuDriverUnlock(qemu_driver);
    return active;
698 699
}

700 701 702 703 704 705 706
/**
 * qemudShutdown:
 *
 * Shutdown the QEmu daemon, it will stop all active domains and networks
 */
static int
qemudShutdown(void) {
707

708
    if (!qemu_driver)
709
        return -1;
710

711
    qemuDriverLock(qemu_driver);
712
    pciDeviceListFree(NULL, qemu_driver->activePciHostdevs);
713 714
    virCapabilitiesFree(qemu_driver->caps);

715
    virDomainObjListFree(&qemu_driver->domains);
716

717
    VIR_FREE(qemu_driver->securityDriverName);
718
    VIR_FREE(qemu_driver->logDir);
719 720
    VIR_FREE(qemu_driver->configDir);
    VIR_FREE(qemu_driver->autostartDir);
721
    VIR_FREE(qemu_driver->stateDir);
722 723
    VIR_FREE(qemu_driver->libDir);
    VIR_FREE(qemu_driver->cacheDir);
724
    VIR_FREE(qemu_driver->vncTLSx509certdir);
J
Jim Meyering 已提交
725
    VIR_FREE(qemu_driver->vncListen);
726
    VIR_FREE(qemu_driver->vncPassword);
727
    VIR_FREE(qemu_driver->vncSASLdir);
728
    VIR_FREE(qemu_driver->saveImageFormat);
729 730
    VIR_FREE(qemu_driver->hugetlbfs_mount);
    VIR_FREE(qemu_driver->hugepage_path);
D
Daniel P. Berrange 已提交
731

732 733
    /* Free domain callback list */
    virDomainEventCallbackListFree(qemu_driver->domainEventCallbacks);
734 735 736 737
    virDomainEventQueueFree(qemu_driver->domainEventQueue);

    if (qemu_driver->domainEventTimer != -1)
        virEventRemoveTimeout(qemu_driver->domainEventTimer);
738

739 740 741
    if (qemu_driver->brctl)
        brShutdown(qemu_driver->brctl);

742 743
    virCgroupFree(&qemu_driver->cgroup);

744
    qemuDriverUnlock(qemu_driver);
745
    virMutexDestroy(&qemu_driver->lock);
746
    VIR_FREE(qemu_driver);
747 748

    return 0;
749 750 751
}

/* Return -1 for error, 1 to continue reading and 0 for success */
752
typedef int qemudHandlerMonitorOutput(virConnectPtr conn,
753
                                      virDomainObjPtr vm,
754 755 756
                                      const char *output,
                                      int fd);

757 758 759
/*
 * Returns -1 for error, 0 on end-of-file, 1 for success
 */
760
static int
761
qemudReadMonitorOutput(virConnectPtr conn,
762
                       virDomainObjPtr vm,
763 764
                       int fd,
                       char *buf,
G
Guido Günther 已提交
765
                       size_t buflen,
766
                       qemudHandlerMonitorOutput func,
767 768
                       const char *what,
                       int timeout)
769
{
G
Guido Günther 已提交
770
    size_t got = 0;
771
    buf[0] = '\0';
772
    timeout *= 1000; /* poll wants milli seconds */
773

774
    /* Consume & discard the initial greeting */
775
    while (got < (buflen-1)) {
G
Guido Günther 已提交
776
        ssize_t ret;
777 778

        ret = read(fd, buf+got, buflen-got-1);
779

780 781 782 783 784 785
        if (ret < 0) {
            struct pollfd pfd = { .fd = fd, .events = POLLIN };
            if (errno == EINTR)
                continue;

            if (errno != EAGAIN) {
786 787 788
                virReportSystemError(conn, errno,
                                     _("Failure while reading %s startup output"),
                                     what);
789 790 791
                return -1;
            }

792
            ret = poll(&pfd, 1, timeout);
793
            if (ret == 0) {
794
                qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
795
                                 _("Timed out while reading %s startup output"), what);
796 797 798
                return -1;
            } else if (ret == -1) {
                if (errno != EINTR) {
799 800 801
                    virReportSystemError(conn, errno,
                                         _("Failure while reading %s startup output"),
                                         what);
802 803 804 805 806 807 808 809
                    return -1;
                }
            } else {
                /* Make sure we continue loop & read any further data
                   available before dealing with EOF */
                if (pfd.revents & (POLLIN | POLLHUP))
                    continue;

810
                qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
811
                                 _("Failure while reading %s startup output"), what);
812 813
                return -1;
            }
814 815
        } else if (ret == 0) {
            return 0;
816 817 818
        } else {
            got += ret;
            buf[got] = '\0';
819 820 821 822 823 824
            ret = func(conn, vm, buf, fd);
            if (ret == -1)
                return -1;
            if (ret == 1)
                continue;
            return 1;
825 826 827
        }
    }

828
    qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
829
                     _("Out of space while reading %s startup output"), what);
830 831 832 833
    return -1;

}

834 835 836 837 838 839 840 841 842

/*
 * Returns -1 for error, 0 on success
 */
static int
qemudReadLogOutput(virConnectPtr conn,
                   virDomainObjPtr vm,
                   int fd,
                   char *buf,
G
Guido Günther 已提交
843
                   size_t buflen,
844 845 846 847
                   qemudHandlerMonitorOutput func,
                   const char *what,
                   int timeout)
{
848
    int retries = (timeout*10);
849
    int got = 0;
850 851 852
    buf[0] = '\0';

    while (retries) {
853
        ssize_t func_ret, ret;
854
        int isdead = 0;
G
Guido Günther 已提交
855

856 857
        func_ret = func(conn, vm, buf, fd);

858 859
        if (kill(vm->pid, 0) == -1 && errno == ESRCH)
            isdead = 1;
860

861 862
        /* Any failures should be detected before we read the log, so we
         * always have something useful to report on failure. */
863 864
        ret = saferead(fd, buf+got, buflen-got-1);
        if (ret < 0) {
865 866 867 868 869 870
            virReportSystemError(conn, errno,
                                 _("Failure while reading %s log output"),
                                 what);
            return -1;
        }

871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886
        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;
        }

887 888
        if (func_ret <= 0)
            return func_ret;
889 890 891 892 893 894 895 896 897 898

        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;
}

899
static int
900
qemudCheckMonitorPrompt(virConnectPtr conn ATTRIBUTE_UNUSED,
901
                        virDomainObjPtr vm,
902 903 904 905 906 907 908 909 910 911 912
                        const char *output,
                        int fd)
{
    if (strstr(output, "(qemu) ") == NULL)
        return 1; /* keep reading */

    vm->monitor = fd;

    return 0;
}

913
static int
914 915 916 917 918
qemudOpenMonitorCommon(virConnectPtr conn,
                       struct qemud_driver* driver,
                       virDomainObjPtr vm,
                       int monfd,
                       int reconnect)
919
{
920
    char buf[1024];
921
    int ret;
922

923
    if (virSetCloseExec(monfd) < 0) {
924
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
925
                         "%s", _("Unable to set monitor close-on-exec flag"));
926
        return -1;
927
    }
928
    if (virSetNonBlock(monfd) < 0) {
929
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
930
                         "%s", _("Unable to put monitor into non-blocking mode"));
931
        return -1;
932 933
    }

934
    if (!reconnect) {
935 936 937 938
        if (qemudReadMonitorOutput(conn,
                                   vm, monfd,
                                   buf, sizeof(buf),
                                   qemudCheckMonitorPrompt,
939
                                   "monitor", 10) <= 0)
940 941 942
            ret = -1;
        else
            ret = 0;
943 944 945 946 947 948
    } else {
        vm->monitor = monfd;
        ret = 0;
    }

    if (ret != 0)
949
        return ret;
950

951 952
    if ((vm->monitorWatch = virEventAddHandle(vm->monitor,
                                              VIR_EVENT_HANDLE_HANGUP | VIR_EVENT_HANDLE_ERROR,
953 954
                                              qemudDispatchVMEvent,
                                              driver, NULL)) < 0)
955
        return -1;
956

957 958
    return 0;
}
959

960 961 962 963 964 965 966 967 968
static int
qemudOpenMonitorUnix(virConnectPtr conn,
                     struct qemud_driver* driver,
                     virDomainObjPtr vm,
                     const char *monitor,
                     int reconnect)
{
    struct sockaddr_un addr;
    int monfd;
969
    int timeout = 3; /* In seconds */
970
    int ret, i = 0;
971 972 973 974 975 976 977 978 979

    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
980 981 982 983 984
    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;
    }
985

986 987 988 989 990 991
    do {
        ret = connect(monfd, (struct sockaddr *) &addr, sizeof(addr));

        if (ret == 0)
            break;

992 993
        if (errno == ENOENT || errno == ECONNREFUSED) {
            /* ENOENT       : Socket may not have shown up yet
994 995 996 997
             * ECONNREFUSED : Leftover socket hasn't been removed yet */
            continue;
        }

998 999 1000
        virReportSystemError(conn, errno, "%s",
                             _("failed to connect to monitor socket"));
        goto error;
1001 1002 1003 1004 1005 1006 1007

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

    if (ret != 0) {
        virReportSystemError(conn, errno, "%s",
                             _("monitor socket did not show up."));
        goto error;
1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019
    }

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

    return 0;

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

1020 1021 1022 1023 1024 1025 1026 1027
static int
qemudOpenMonitorPty(virConnectPtr conn,
                    struct qemud_driver* driver,
                    virDomainObjPtr vm,
                    const char *monitor,
                    int reconnect)
{
    int monfd;
1028

1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040
    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:
1041
    close(monfd);
1042 1043 1044 1045 1046 1047 1048 1049 1050 1051
    return -1;
}

static int
qemudOpenMonitor(virConnectPtr conn,
                 struct qemud_driver *driver,
                 virDomainObjPtr vm,
                 int reconnect)
{
    switch (vm->monitor_chr->type) {
1052 1053 1054 1055
    case VIR_DOMAIN_CHR_TYPE_UNIX:
        return qemudOpenMonitorUnix(conn, driver, vm,
                                    vm->monitor_chr->data.nix.path,
                                    reconnect);
1056 1057 1058 1059 1060 1061 1062 1063 1064 1065
    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;
    }
1066 1067
}

1068
/* Returns -1 for error, 0 success, 1 continue reading */
1069 1070 1071 1072 1073 1074
static int
qemudExtractMonitorPath(virConnectPtr conn,
                        const char *haystack,
                        size_t *offset,
                        char **path)
{
1075
    static const char needle[] = "char device redirected to";
1076
    char *tmp, *dev;
1077

1078
    VIR_FREE(*path);
1079
    /* First look for our magic string */
1080 1081 1082 1083 1084
    if (!(tmp = strstr(haystack + *offset, needle))) {
        return 1;
    }
    tmp += sizeof(needle);
    dev = tmp;
1085

1086 1087 1088 1089 1090
    /*
     * And look for first whitespace character and nul terminate
     * to mark end of the pty path
     */
    while (*tmp) {
1091
        if (c_isspace(*tmp)) {
1092 1093
            *path = strndup(dev, tmp-dev);
            if (*path == NULL) {
1094
                virReportOOMError(conn);
1095 1096
                return -1;
            }
1097

1098
            /* ... now further update offset till we get EOL */
1099
            *offset = tmp - haystack;
1100 1101
            return 0;
        }
1102
        tmp++;
1103 1104 1105 1106 1107
    }

    /*
     * We found a path, but didn't find any whitespace,
     * so it must be still incomplete - we should at
1108 1109
     * least see a \n - indicate that we want to carry
     * on trying again
1110
     */
1111
    return 1;
1112 1113 1114
}

static int
1115
qemudFindCharDevicePTYs(virConnectPtr conn,
1116
                        virDomainObjPtr vm,
1117 1118
                        const char *output,
                        int fd ATTRIBUTE_UNUSED)
1119
{
1120
    size_t offset = 0;
1121
    int ret, i;
1122 1123

    /* The order in which QEMU prints out the PTY paths is
1124 1125
       the order in which it procsses its serial and parallel
       device args. This code must match that ordering.... */
1126

1127
    /* first comes the serial devices */
1128 1129
    for (i = 0 ; i < vm->def->nserials ; i++) {
        virDomainChrDefPtr chr = vm->def->serials[i];
1130 1131 1132
        if (chr->type == VIR_DOMAIN_CHR_TYPE_PTY) {
            if ((ret = qemudExtractMonitorPath(conn, output, &offset,
                                               &chr->data.file.path)) != 0)
1133
                return ret;
1134 1135 1136
        }
    }

1137
    /* then the parallel devices */
1138 1139
    for (i = 0 ; i < vm->def->nparallels ; i++) {
        virDomainChrDefPtr chr = vm->def->parallels[i];
1140 1141 1142
        if (chr->type == VIR_DOMAIN_CHR_TYPE_PTY) {
            if ((ret = qemudExtractMonitorPath(conn, output, &offset,
                                               &chr->data.file.path)) != 0)
1143
                return ret;
1144 1145 1146
        }
    }

1147
    return 0;
1148 1149
}

1150 1151 1152 1153
static int
qemudWaitForMonitor(virConnectPtr conn,
                    struct qemud_driver* driver,
                    virDomainObjPtr vm, off_t pos)
1154
{
1155
    char buf[4096]; /* Plenty of space to get startup greeting */
1156 1157 1158 1159 1160
    int logfd;
    int ret;

    if ((logfd = qemudLogReadFD(conn, driver->logDir, vm->def->name, pos))
        < 0)
1161
        return -1;
1162

1163 1164 1165
    ret = qemudReadLogOutput(conn, vm, logfd, buf, sizeof(buf),
                             qemudFindCharDevicePTYs,
                             "console", 3);
1166
    if (close(logfd) < 0) {
1167
        char ebuf[4096];
1168
        VIR_WARN(_("Unable to close logfile: %s\n"),
1169 1170
                 virStrerror(errno, ebuf, sizeof ebuf));
    }
1171

1172 1173 1174 1175 1176 1177
    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;
    }
1178

1179 1180 1181 1182
    if (qemudOpenMonitor(conn, driver, vm, 0) < 0)
        return -1;

    return 0;
1183 1184
}

1185
static int
1186 1187 1188 1189
qemuDetectVcpuPIDs(virConnectPtr conn,
                   virDomainObjPtr vm) {
    pid_t *cpupids = NULL;
    int ncpupids;
1190

1191
    if (vm->def->virtType != VIR_DOMAIN_VIRT_KVM) {
1192 1193 1194 1195 1196
        vm->nvcpupids = 1;
        if (VIR_ALLOC_N(vm->vcpupids, vm->nvcpupids) < 0) {
            virReportOOMError(conn);
            return -1;
        }
1197 1198 1199 1200
        vm->vcpupids[0] = vm->pid;
        return 0;
    }

1201
    /* What follows is now all KVM specific */
1202

1203 1204
    if ((ncpupids = qemuMonitorGetCPUInfo(vm, &cpupids)) < 0)
        return -1;
1205

1206 1207 1208
    /* Treat failure to get VCPU<->PID mapping as non-fatal */
    if (ncpupids == 0)
        return 0;
1209

1210 1211 1212 1213 1214 1215 1216
    if (ncpupids != vm->def->vcpus) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                         _("got wrong number of vCPU pids from QEMU monitor. got %d, wanted %d"),
                         ncpupids, (int)vm->def->vcpus);
        VIR_FREE(cpupids);
        return -1;
    }
1217

1218 1219
    vm->nvcpupids = ncpupids;
    vm->vcpupids = cpupids;
1220 1221 1222
    return 0;
}

1223 1224
static int
qemudInitCpus(virConnectPtr conn,
D
Daniel Veillard 已提交
1225 1226
              virDomainObjPtr vm,
              const char *migrateFrom) {
1227 1228 1229 1230 1231
#if HAVE_SCHED_GETAFFINITY
    cpu_set_t mask;
    int i, maxcpu = QEMUD_CPUMASK_LEN;
    virNodeInfo nodeinfo;

1232
    if (nodeGetInfo(conn, &nodeinfo) < 0)
1233 1234 1235 1236 1237 1238 1239 1240
        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
1241 1242 1243 1244 1245 1246
    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++)
1247
            CPU_SET(i, &mask);
D
Daniel P. Berrange 已提交
1248
    }
1249 1250 1251 1252

    for (i = 0 ; i < vm->nvcpupids ; i++) {
        if (sched_setaffinity(vm->vcpupids[i],
                              sizeof(mask), &mask) < 0) {
1253 1254
            virReportSystemError(conn, errno, "%s",
                                 _("failed to set CPU affinity"));
1255 1256 1257 1258 1259
            return -1;
        }
    }
#endif /* HAVE_SCHED_GETAFFINITY */

D
Daniel Veillard 已提交
1260 1261
    if (migrateFrom == NULL) {
        /* Allow the CPUS to start executing */
1262
        if (qemuMonitorStartCPUs(conn, vm) < 0) {
1263 1264 1265
            if (virGetLastError() == NULL)
                qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                                 "%s", _("resume operation failed"));
D
Daniel Veillard 已提交
1266 1267
            return -1;
        }
1268 1269 1270 1271 1272 1273
    }

    return 0;
}


1274
static int
1275 1276
qemuInitPasswords(struct qemud_driver *driver,
                  virDomainObjPtr vm) {
1277
    int ret = 0;
1278

1279 1280 1281
    if ((vm->def->ngraphics == 1) &&
        vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
        (vm->def->graphics[0]->data.vnc.passwd || driver->vncPassword)) {
1282

1283 1284 1285 1286
        ret = qemuMonitorSetVNCPassword(vm,
                                        vm->def->graphics[0]->data.vnc.passwd ?
                                        vm->def->graphics[0]->data.vnc.passwd :
                                        driver->vncPassword);
1287 1288
    }

1289
    return ret;
1290 1291 1292
}


1293
static int qemudNextFreeVNCPort(struct qemud_driver *driver ATTRIBUTE_UNUSED) {
1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328
    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;
}

1329 1330 1331 1332 1333
static pciDeviceList *
qemuGetPciHostDeviceList(virConnectPtr conn,
                         virDomainDefPtr def)
{
    pciDeviceList *list;
1334 1335
    int i;

1336 1337
    if (!(list = pciDeviceListNew(conn)))
        return NULL;
1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352

    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);
1353 1354 1355 1356
        if (!dev) {
            pciDeviceListFree(conn, list);
            return NULL;
        }
1357

1358
        if (pciDeviceListAdd(conn, list, dev) < 0) {
1359
            pciFreeDevice(conn, dev);
1360 1361
            pciDeviceListFree(conn, list);
            return NULL;
1362 1363
        }

1364
        pciDeviceSetManaged(dev, hostdev->managed);
1365 1366
    }

1367 1368 1369 1370
    return list;
}

static int
1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402
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)
1403 1404 1405 1406 1407 1408 1409 1410 1411 1412
{
    pciDeviceList *pcidevs;
    int i;

    if (!def->nhostdevs)
        return 0;

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

1413
    /* We have to use 3 loops here. *All* devices must
1414 1415
     * be detached before we reset any of them, because
     * in some cases you have to reset the whole PCI,
1416 1417
     * which impacts all devices on it. Also, all devices
     * must be reset before being marked as active.
1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432
     */

    /* 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++)
1433 1434
        if (pciResetDevice(conn, pcidevs->devs[i],
                           driver->activePciHostdevs) < 0)
1435 1436
            goto error;

1437 1438 1439 1440 1441 1442 1443 1444 1445
    /* 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;
    }

1446
    pciDeviceListFree(conn, pcidevs);
1447 1448 1449
    return 0;

error:
1450
    pciDeviceListFree(conn, pcidevs);
1451 1452 1453
    return -1;
}

1454
static void
1455 1456 1457
qemuDomainReAttachHostDevices(virConnectPtr conn,
                              struct qemud_driver *driver,
                              virDomainDefPtr def)
1458
{
1459
    pciDeviceList *pcidevs;
1460 1461
    int i;

1462 1463
    if (!def->nhostdevs)
        return;
1464

1465 1466 1467 1468 1469 1470
    if (!(pcidevs = qemuGetPciHostDeviceList(conn, def))) {
        virErrorPtr err = virGetLastError();
        VIR_ERROR(_("Failed to allocate pciDeviceList: %s\n"),
                  err ? err->message : "");
        virResetError(err);
        return;
1471 1472
    }

1473 1474
    /* Again 3 loops; mark all devices as inactive before reset
     * them and reset all the devices before re-attach */
1475

1476
    for (i = 0; i < pcidevs->count; i++)
1477 1478 1479 1480 1481
        pciDeviceListDel(conn, driver->activePciHostdevs, pcidevs->devs[i]);

    for (i = 0; i < pcidevs->count; i++)
        if (pciResetDevice(conn, pcidevs->devs[i],
                           driver->activePciHostdevs) < 0) {
1482
            virErrorPtr err = virGetLastError();
1483
            VIR_ERROR(_("Failed to reset PCI device: %s\n"),
1484 1485 1486 1487
                      err ? err->message : "");
            virResetError(err);
        }

1488 1489 1490
    for (i = 0; i < pcidevs->count; i++)
        if (pciDeviceGetManaged(pcidevs->devs[i]) &&
            pciReAttachDevice(conn, pcidevs->devs[i]) < 0) {
1491
            virErrorPtr err = virGetLastError();
1492
            VIR_ERROR(_("Failed to re-attach PCI device: %s\n"),
1493 1494 1495 1496
                      err ? err->message : "");
            virResetError(err);
        }

1497
    pciDeviceListFree(conn, pcidevs);
1498 1499
}

1500 1501 1502 1503 1504 1505 1506 1507 1508 1509
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

1510 1511 1512 1513 1514 1515
static int qemuSetupCgroup(virConnectPtr conn,
                           struct qemud_driver *driver,
                           virDomainObjPtr vm)
{
    virCgroupPtr cgroup = NULL;
    int rc;
1516
    unsigned int i;
1517 1518 1519 1520
    const char *const *deviceACL =
        driver->cgroupDeviceACL ?
        (const char *const *)driver->cgroupDeviceACL :
        defaultDeviceACL;
1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532

    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;
    }

1533 1534
    if (qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_DEVICES)) {
        rc = virCgroupDenyAllDevices(cgroup);
1535
        if (rc != 0) {
1536 1537 1538 1539 1540
            if (rc == -EPERM) {
                VIR_WARN0("Group devices ACL is not accessible, disabling whitelisting");
                goto done;
            }

1541
            virReportSystemError(conn, -rc,
1542
                                 _("Unable to deny all devices for %s"), vm->def->name);
1543 1544 1545
            goto cleanup;
        }

1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559
        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;
            }
        }
1560

1561
        rc = virCgroupAllowDeviceMajor(cgroup, 'c', DEVICE_PTY_MAJOR);
1562 1563
        if (rc != 0) {
            virReportSystemError(conn, -rc, "%s",
1564
                                 _("unable to allow /dev/pts/ devices"));
1565 1566 1567
            goto cleanup;
        }

1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586
        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;
            }
1587 1588 1589 1590
        }
    }

done:
1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659
    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;
}


1660 1661 1662 1663 1664 1665 1666 1667 1668
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;
}

1669 1670

#ifdef __linux__
1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681
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;

1682 1683
    VIR_DEBUG("Setting ownership on %s to %d:%d", file, owner->uid, owner->gid);

1684 1685 1686 1687 1688 1689 1690 1691
    if (chown(file, owner->uid, owner->gid) < 0) {
        virReportSystemError(conn, errno, _("cannot set ownership on %s"), file);
        return -1;
    }

    return 0;
}

1692 1693 1694 1695
static int qemuDomainSetHostdevUSBOwnership(virConnectPtr conn,
                                            virDomainHostdevDefPtr def,
                                            uid_t uid, gid_t gid)
{
1696 1697
    struct qemuFileOwner owner = { uid, gid };
    int ret = -1;
1698 1699 1700 1701 1702 1703

    /* 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;

1704 1705 1706 1707 1708 1709 1710 1711 1712
    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);
1713

1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724
    usbFreeDevice(conn, dev);
cleanup:
    return ret;
}

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

1725 1726
    VIR_DEBUG("Setting ownership on %s to %d:%d", file, owner->uid, owner->gid);

1727 1728
    if (chown(file, owner->uid, owner->gid) < 0) {
        virReportSystemError(conn, errno, _("cannot set ownership on %s"), file);
1729 1730 1731 1732 1733 1734 1735 1736 1737 1738
        return -1;
    }

    return 0;
}

static int qemuDomainSetHostdevPCIOwnership(virConnectPtr conn,
                                            virDomainHostdevDefPtr def,
                                            uid_t uid, gid_t gid)
{
1739
    struct qemuFileOwner owner = { uid, gid };
1740 1741
    int ret = -1;

1742 1743 1744 1745 1746
    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);
1747

1748
    if (!dev)
1749 1750
        goto cleanup;

1751 1752
    ret = pciDeviceFileIterate(conn, dev,
                               qemuDomainSetHostdevPCIOwnershipActor, &owner);
1753

1754
    pciFreeDevice(conn, dev);
1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785
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

}

1786 1787
static int qemuDomainSetFileOwnership(virConnectPtr conn,
                                      const char *path,
1788 1789 1790
                                      uid_t uid, gid_t gid)
{

1791
    if (!path)
1792 1793
        return 0;

1794 1795
    VIR_DEBUG("Setting ownership on %s to %d:%d", path, uid, gid);
    if (chown(path, uid, gid) < 0) {
1796
        virReportSystemError(conn, errno, _("cannot set ownership on %s"),
1797
                             path);
1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826
        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;

1827
        return qemuDomainSetFileOwnership(conn, def->data.disk->src, uid, gid);
1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854

    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;

1855 1856 1857 1858
    if (qemuDomainSetFileOwnership(conn, def->os.kernel, uid, gid) < 0 ||
        qemuDomainSetFileOwnership(conn, def->os.initrd, uid, gid) < 0)
        return -1;

1859 1860 1861 1862 1863
    for (i = 0 ; i < def->ndisks ; i++) {
        if (restore &&
            (def->disks[i]->readonly || def->disks[i]->shared))
            continue;

1864
        if (qemuDomainSetFileOwnership(conn, def->disks[i]->src, uid, gid) < 0)
1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875
            return -1;
    }

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

    return 0;
}

1876 1877 1878
static virDomainPtr qemudDomainLookupByName(virConnectPtr conn,
                                            const char *name);

1879 1880 1881 1882
struct qemudHookData {
    virConnectPtr conn;
    virDomainObjPtr vm;
    struct qemud_driver *driver;
1883 1884 1885
};

static int qemudSecurityHook(void *data) {
1886 1887 1888 1889
    struct qemudHookData *h = data;

    if (qemuAddToCgroup(h->driver, h->vm->def) < 0)
        return -1;
1890

1891
    if (qemudDomainSetSecurityLabel(h->conn, h->driver, h->vm) < 0)
1892 1893 1894 1895 1896
        return -1;

    if (h->driver->privileged) {
        if (qemuDomainSetAllDeviceOwnership(h->conn, h->driver, h->vm->def, 0) < 0)
            return -1;
1897

1898 1899
        DEBUG("Dropping privileges of VM to %d:%d", h->driver->user, h->driver->group);

1900 1901 1902 1903 1904
        if (h->driver->group) {
            if (setregid(h->driver->group, h->driver->group) < 0) {
                virReportSystemError(NULL, errno,
                                     _("cannot change to '%d' group"),
                                     h->driver->group);
1905
                return -1;
1906
            }
1907
        }
1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918
        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;
1919 1920
}

1921 1922 1923 1924 1925 1926 1927 1928 1929 1930
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",
1931
                    driver->libDir, vm) < 0) {
1932 1933 1934 1935 1936 1937 1938
        virReportOOMError(conn);
        return -1;
    }

    return 0;
}

1939 1940
static int qemudStartVMDaemon(virConnectPtr conn,
                              struct qemud_driver *driver,
1941
                              virDomainObjPtr vm,
1942 1943
                              const char *migrateFrom,
                              int stdin_fd) {
1944
    const char **argv = NULL, **tmp;
1945
    const char **progenv = NULL;
1946
    int i, ret;
1947
    struct stat sb;
1948 1949
    int *tapfds = NULL;
    int ntapfds = 0;
1950
    unsigned int qemuCmdFlags;
1951
    fd_set keepfd;
1952
    const char *emulator;
G
Guido Günther 已提交
1953
    pid_t child;
1954
    int pos = -1;
1955
    char ebuf[1024];
1956
    char *pidfile = NULL;
1957
    int logfile;
1958

1959
    struct qemudHookData hookData;
1960 1961 1962 1963
    hookData.conn = conn;
    hookData.vm = vm;
    hookData.driver = driver;

1964
    FD_ZERO(&keepfd);
1965

1966
    if (virDomainIsActive(vm)) {
1967
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_INVALID,
1968
                         "%s", _("VM is already active"));
1969 1970 1971
        return -1;
    }

1972 1973 1974 1975 1976 1977 1978 1979
    /* 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;

1980 1981 1982
    /* Ensure no historical cgroup for this VM is lieing around bogus settings */
    qemuRemoveCgroup(conn, driver, vm);

1983 1984 1985
    if ((vm->def->ngraphics == 1) &&
        vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
        vm->def->graphics[0]->data.vnc.autoport) {
1986
        int port = qemudNextFreeVNCPort(driver);
1987
        if (port < 0) {
1988
            qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
1989
                             "%s", _("Unable to find an unused VNC port"));
1990
            goto cleanup;
1991
        }
1992
        vm->def->graphics[0]->data.vnc.port = port;
1993
    }
1994

1995
    if (virFileMakePath(driver->logDir) < 0) {
1996 1997 1998
        virReportSystemError(conn, errno,
                             _("cannot create log directory %s"),
                             driver->logDir);
1999
        goto cleanup;
2000 2001
    }

2002
    if ((logfile = qemudLogFD(conn, driver, vm->def->name)) < 0)
2003
        goto cleanup;
2004

2005 2006
    emulator = vm->def->emulator;

2007 2008 2009 2010
    /* 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
     */
2011
    if (stat(emulator, &sb) < 0) {
2012 2013 2014
        virReportSystemError(conn, errno,
                             _("Cannot find QEMU binary %s"),
                             emulator);
2015
        goto cleanup;
2016 2017
    }

2018
    if (qemudExtractVersionInfo(emulator,
2019
                                NULL,
2020
                                &qemuCmdFlags) < 0)
2021
        goto cleanup;
2022

2023 2024 2025
    if (qemuSetupCgroup(conn, driver, vm) < 0)
        goto cleanup;

2026
    if (qemuPrepareHostDevices(conn, driver, vm->def) < 0)
2027
        goto cleanup;
2028

2029 2030 2031 2032 2033
    if (VIR_ALLOC(vm->monitor_chr) < 0) {
        virReportOOMError(conn);
        goto cleanup;
    }

2034 2035
    if (qemuPrepareMonitorChr(conn, driver, vm->monitor_chr, vm->def->name) < 0)
        goto cleanup;
2036

D
Daniel P. Berrange 已提交
2037 2038 2039 2040 2041 2042 2043
    if ((ret = virFileDeletePid(driver->stateDir, vm->def->name)) != 0) {
        virReportSystemError(conn, ret,
                             _("Cannot remove stale PID file for %s"),
                             vm->def->name);
        goto cleanup;
    }

2044 2045 2046 2047 2048
    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
2049

2050
    vm->def->id = driver->nextvmid++;
2051
    if (qemudBuildCommandLine(conn, driver, vm->def, vm->monitor_chr,
2052
                              qemuCmdFlags, &argv, &progenv,
2053 2054
                              &tapfds, &ntapfds, migrateFrom) < 0)
        goto cleanup;
2055

2056 2057
    tmp = progenv;
    while (*tmp) {
2058
        if (safewrite(logfile, *tmp, strlen(*tmp)) < 0)
2059
            VIR_WARN(_("Unable to write envv to logfile: %s\n"),
2060
                     virStrerror(errno, ebuf, sizeof ebuf));
2061
        if (safewrite(logfile, " ", 1) < 0)
2062
            VIR_WARN(_("Unable to write envv to logfile: %s\n"),
2063
                     virStrerror(errno, ebuf, sizeof ebuf));
2064 2065
        tmp++;
    }
2066 2067
    tmp = argv;
    while (*tmp) {
2068
        if (safewrite(logfile, *tmp, strlen(*tmp)) < 0)
2069
            VIR_WARN(_("Unable to write argv to logfile: %s\n"),
2070
                     virStrerror(errno, ebuf, sizeof ebuf));
2071
        if (safewrite(logfile, " ", 1) < 0)
2072
            VIR_WARN(_("Unable to write argv to logfile: %s\n"),
2073
                     virStrerror(errno, ebuf, sizeof ebuf));
2074 2075
        tmp++;
    }
2076
    if (safewrite(logfile, "\n", 1) < 0)
2077
        VIR_WARN(_("Unable to write argv to logfile: %s\n"),
2078
                 virStrerror(errno, ebuf, sizeof ebuf));
2079

2080
    if ((pos = lseek(logfile, 0, SEEK_END)) < 0)
2081
        VIR_WARN(_("Unable to seek to end of logfile: %s\n"),
2082
                 virStrerror(errno, ebuf, sizeof ebuf));
2083

2084 2085 2086
    for (i = 0 ; i < ntapfds ; i++)
        FD_SET(tapfds[i], &keepfd);

2087
    ret = virExecDaemonize(conn, argv, progenv, &keepfd, &child,
2088
                           stdin_fd, &logfile, &logfile,
2089
                           VIR_EXEC_NONBLOCK | VIR_EXEC_CLEAR_CAPS,
2090 2091 2092
                           qemudSecurityHook, &hookData,
                           pidfile);
    VIR_FREE(pidfile);
G
Guido Günther 已提交
2093 2094 2095

    /* wait for qemu process to to show up */
    if (ret == 0) {
2096
        if (virFileReadPid(driver->stateDir, vm->def->name, &vm->pid)) {
2097
            qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
2098
                             _("Domain %s didn't show up\n"), vm->def->name);
2099
            ret = -1;
G
Guido Günther 已提交
2100
        }
2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111
    } 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;
    }
2112 2113

    vm->state = migrateFrom ? VIR_DOMAIN_PAUSED : VIR_DOMAIN_RUNNING;
2114

2115
    for (i = 0 ; argv[i] ; i++)
2116 2117
        VIR_FREE(argv[i]);
    VIR_FREE(argv);
2118

2119 2120 2121 2122
    for (i = 0 ; progenv[i] ; i++)
        VIR_FREE(progenv[i]);
    VIR_FREE(progenv);

2123 2124 2125
    if (tapfds) {
        for (i = 0 ; i < ntapfds ; i++) {
            close(tapfds[i]);
2126
        }
2127
        VIR_FREE(tapfds);
2128 2129
    }

2130 2131 2132 2133
    if (ret == -1)
        goto cleanup;

    if ((qemudWaitForMonitor(conn, driver, vm, pos) < 0) ||
2134
        (qemuDetectVcpuPIDs(conn, vm) < 0) ||
2135
        (qemudInitCpus(conn, vm, migrateFrom) < 0) ||
2136
        (qemuInitPasswords(driver, vm) < 0) ||
2137
        (qemuMonitorSetBalloon(vm, vm->def->memory) < 0) ||
2138
        (virDomainSaveStatus(conn, driver->stateDir, vm) < 0)) {
2139 2140 2141
        qemudShutdownVMDaemon(conn, driver, vm);
        ret = -1;
        /* No need for 'goto cleanup' now since qemudShutdownVMDaemon does enough */
2142 2143
    }

2144 2145 2146
    if (logfile != -1)
        close(logfile);

2147
    return ret;
2148 2149 2150 2151 2152 2153 2154

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);
    }
2155
    qemuRemoveCgroup(conn, driver, vm);
2156 2157 2158 2159
    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;
2160 2161
    if (logfile != -1)
        close(logfile);
2162 2163
    vm->def->id = -1;
    return -1;
2164 2165 2166
}


2167
static void qemudShutdownVMDaemon(virConnectPtr conn,
2168 2169
                                  struct qemud_driver *driver,
                                  virDomainObjPtr vm) {
D
Daniel P. Berrange 已提交
2170
    int ret;
2171
    int retries = 0;
D
Daniel P. Berrange 已提交
2172

2173
    if (!virDomainIsActive(vm))
2174
        return;
2175

2176
    VIR_DEBUG(_("Shutting down VM '%s'\n"), vm->def->name);
2177

G
Guido Günther 已提交
2178 2179
    if (virKillProcess(vm->pid, 0) == 0 &&
        virKillProcess(vm->pid, SIGTERM) < 0)
2180 2181 2182
        virReportSystemError(conn, errno,
                             _("Failed to send SIGTERM to %s (%d)"),
                             vm->def->name, vm->pid);
2183

2184 2185 2186
    if (vm->monitorWatch != -1) {
        virEventRemoveHandle(vm->monitorWatch);
        vm->monitorWatch = -1;
2187
    }
2188 2189 2190 2191 2192

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

2193 2194 2195 2196 2197 2198 2199
    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
2200 2201
    /* shut it off for sure */
    virKillProcess(vm->pid, SIGKILL);
2202

2203 2204 2205 2206
    /* Reset Security Labels */
    if (driver->securityDriver)
        driver->securityDriver->domainRestoreSecurityLabel(conn, vm);

2207 2208 2209 2210 2211 2212 2213
    /* 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);
    }

2214 2215 2216 2217
    if (qemuDomainSetAllDeviceOwnership(conn, driver, vm->def, 1) < 0)
        VIR_WARN("Failed to restore all device ownership for %s",
                 vm->def->name);

2218
    qemuDomainReAttachHostDevices(conn, driver, vm->def);
2219

2220 2221 2222 2223 2224 2225 2226 2227 2228 2229
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);
    }

2230
    if (qemudRemoveDomainStatus(conn, driver, vm) < 0) {
2231
        VIR_WARN(_("Failed to remove domain status for %s"),
2232 2233
                 vm->def->name);
    }
D
Daniel P. Berrange 已提交
2234 2235 2236 2237 2238 2239
    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));
    }

2240
    vm->pid = -1;
2241
    vm->def->id = -1;
2242
    vm->state = VIR_DOMAIN_SHUTOFF;
2243
    VIR_FREE(vm->vcpupids);
2244
    vm->nvcpupids = 0;
2245 2246

    if (vm->newDef) {
2247
        virDomainDefFree(vm->def);
2248
        vm->def = vm->newDef;
2249
        vm->def->id = -1;
2250 2251 2252 2253 2254
        vm->newDef = NULL;
    }
}


2255
static void
2256
qemudDispatchVMEvent(int watch, int fd, int events, void *opaque) {
2257
    struct qemud_driver *driver = opaque;
2258
    virDomainObjPtr vm = NULL;
2259
    virDomainEventPtr event = NULL;
2260
    unsigned int i;
2261
    int quit = 0, failed = 0;
2262

2263
    qemuDriverLock(driver);
2264
    for (i = 0 ; i < driver->domains.count ; i++) {
2265 2266 2267
        virDomainObjPtr tmpvm = driver->domains.objs[i];
        virDomainObjLock(tmpvm);
        if (virDomainIsActive(tmpvm) &&
2268
            tmpvm->monitorWatch == watch) {
2269
            vm = tmpvm;
2270
            break;
2271
        }
2272
        virDomainObjUnlock(tmpvm);
2273 2274 2275
    }

    if (!vm)
2276
        goto cleanup;
2277

2278
    if (vm->monitor != fd) {
2279 2280
        failed = 1;
    } else {
2281
        if (events & (VIR_EVENT_HANDLE_HANGUP | VIR_EVENT_HANDLE_ERROR))
2282
            quit = 1;
2283
        else {
2284 2285
            VIR_ERROR(_("unhandled fd event %d for %s"),
                      events, vm->def->name);
2286
            failed = 1;
2287
        }
2288 2289
    }

2290
    if (failed || quit) {
2291 2292 2293 2294 2295
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         quit ?
                                         VIR_DOMAIN_EVENT_STOPPED_SHUTDOWN :
                                         VIR_DOMAIN_EVENT_STOPPED_FAILED);
2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306
        qemudShutdownVMDaemon(NULL, driver, vm);
        if (!vm->persistent) {
            virDomainRemoveInactive(&driver->domains,
                                    vm);
            vm = NULL;
        }
    }

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
2307 2308
    if (event)
        qemuDomainEventQueue(driver, event);
2309
    qemuDriverUnlock(driver);
2310 2311
}

2312

2313

2314
static virDrvOpenStatus qemudOpen(virConnectPtr conn,
2315
                                  virConnectAuthPtr auth ATTRIBUTE_UNUSED,
2316
                                  int flags ATTRIBUTE_UNUSED) {
2317
    if (conn->uri == NULL) {
2318 2319 2320
        if (qemu_driver == NULL)
            return VIR_DRV_OPEN_DECLINED;

2321
        conn->uri = xmlParseURI(qemu_driver->privileged ?
2322 2323
                                "qemu:///system" :
                                "qemu:///session");
2324
        if (!conn->uri) {
2325
            virReportOOMError(conn);
2326 2327
            return VIR_DRV_OPEN_ERROR;
        }
2328 2329 2330 2331 2332 2333 2334 2335 2336 2337
    } 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;

2338 2339 2340 2341 2342 2343
        if (qemu_driver == NULL) {
            qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR, "%s",
                             _("qemu state driver is not active"));
            return VIR_DRV_OPEN_ERROR;
        }

2344
        if (qemu_driver->privileged) {
2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359
            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;
            }
        }
2360 2361 2362 2363 2364 2365 2366
    }
    conn->privateData = qemu_driver;

    return VIR_DRV_OPEN_SUCCESS;
}

static int qemudClose(virConnectPtr conn) {
2367
    struct qemud_driver *driver = conn->privateData;
2368 2369

    /* Get rid of callbacks registered for this conn */
2370
    qemuDriverLock(driver);
2371
    virDomainEventCallbackListRemoveConn(conn, driver->domainEventCallbacks);
2372
    qemuDriverUnlock(driver);
2373 2374 2375 2376 2377 2378

    conn->privateData = NULL;

    return 0;
}

D
Daniel Veillard 已提交
2379 2380 2381 2382 2383 2384 2385 2386 2387 2388
/* 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;
    }
}

2389
static const char *qemudGetType(virConnectPtr conn ATTRIBUTE_UNUSED) {
2390
    return "QEMU";
2391 2392
}

2393 2394 2395 2396 2397

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

    int r, fd;
2398

2399 2400
    fd = open(KVM_DEVICE, O_RDONLY);
    if (fd < 0) {
2401 2402
        virReportSystemError(NULL, errno, _("Unable to open %s"), KVM_DEVICE);
        return -1;
2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413
    }

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

    close(fd);
    return maxvcpus;
}


2414
static int qemudGetMaxVCPUs(virConnectPtr conn, const char *type) {
2415 2416 2417
    if (!type)
        return 16;

2418
    if (STRCASEEQ(type, "qemu"))
2419 2420
        return 16;

2421
    if (STRCASEEQ(type, "kvm"))
2422
        return kvmGetMaxVCPUs();
2423

2424
    if (STRCASEEQ(type, "kqemu"))
2425
        return 1;
2426 2427 2428

    qemudReportError(conn, NULL, NULL, VIR_ERR_INVALID_ARG,
                     _("unknown type '%s'"), type);
2429 2430 2431
    return -1;
}

2432

2433
static char *qemudGetCapabilities(virConnectPtr conn) {
2434
    struct qemud_driver *driver = conn->privateData;
2435
    virCapsPtr caps;
2436
    char *xml = NULL;
2437

2438
    qemuDriverLock(driver);
2439
    if ((caps = qemudCapsInit(qemu_driver->caps)) == NULL) {
2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450
        virReportOOMError(conn);
        goto cleanup;
    }

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

2451
    virCapabilitiesFree(qemu_driver->caps);
2452 2453 2454
    qemu_driver->caps = caps;

    if ((xml = virCapabilitiesFormatXML(driver->caps)) == NULL)
2455
        virReportOOMError(conn);
2456 2457

cleanup:
2458
    qemuDriverUnlock(driver);
2459

2460
    return xml;
2461 2462 2463
}


2464
static int qemudGetProcessInfo(unsigned long long *cpuTime, int *lastCpu, int pid, int tid) {
D
Daniel P. Berrange 已提交
2465 2466
    char proc[PATH_MAX];
    FILE *pidinfo;
2467
    unsigned long long usertime, systime;
2468 2469
    int cpu;
    int ret;
D
Daniel P. Berrange 已提交
2470

2471 2472 2473 2474 2475 2476
    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
2477 2478 2479 2480
        return -1;
    }

    if (!(pidinfo = fopen(proc, "r"))) {
2481
        /*printf("cannot read pid info");*/
D
Daniel P. Berrange 已提交
2482
        /* VM probably shut down, so fake 0 */
2483 2484 2485 2486
        if (cpuTime)
            *cpuTime = 0;
        if (lastCpu)
            *lastCpu = 0;
D
Daniel P. Berrange 已提交
2487 2488 2489
        return 0;
    }

2490 2491 2492 2493 2494 2495 2496 2497 2498 2499
    /* 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) {
2500
        fclose(pidinfo);
2501 2502
        VIR_WARN0("cannot parse process status data");
        errno = -EINVAL;
D
Daniel P. Berrange 已提交
2503 2504 2505 2506 2507 2508 2509 2510
        return -1;
    }

    /* We got jiffies
     * We want nanoseconds
     * _SC_CLK_TCK is jiffies per second
     * So calulate thus....
     */
2511 2512 2513 2514 2515
    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
2516

2517 2518
    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
2519 2520 2521 2522 2523 2524 2525

    fclose(pidinfo);

    return 0;
}


2526
static virDomainPtr qemudDomainLookupByID(virConnectPtr conn,
2527
                                          int id) {
2528 2529 2530 2531
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;

2532
    qemuDriverLock(driver);
2533
    vm  = virDomainFindByID(&driver->domains, id);
2534
    qemuDriverUnlock(driver);
2535 2536

    if (!vm) {
2537 2538
        qemudReportError(conn, NULL, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching id %d"), id);
2539
        goto cleanup;
2540 2541
    }

2542
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
2543
    if (dom) dom->id = vm->def->id;
2544 2545

cleanup:
2546 2547
    if (vm)
        virDomainObjUnlock(vm);
2548 2549
    return dom;
}
2550

2551
static virDomainPtr qemudDomainLookupByUUID(virConnectPtr conn,
2552
                                            const unsigned char *uuid) {
2553 2554 2555
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
2556

2557
    qemuDriverLock(driver);
2558
    vm = virDomainFindByUUID(&driver->domains, uuid);
2559 2560
    qemuDriverUnlock(driver);

2561
    if (!vm) {
2562 2563 2564
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(uuid, uuidstr);
        qemudReportError(conn, NULL, NULL, VIR_ERR_NO_DOMAIN,
2565
                         _("no domain with matching uuid '%s'"), uuidstr);
2566
        goto cleanup;
2567 2568
    }

2569
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
2570
    if (dom) dom->id = vm->def->id;
2571 2572

cleanup:
2573 2574
    if (vm)
        virDomainObjUnlock(vm);
2575 2576
    return dom;
}
2577

2578
static virDomainPtr qemudDomainLookupByName(virConnectPtr conn,
2579
                                            const char *name) {
2580 2581 2582
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
2583

2584
    qemuDriverLock(driver);
2585
    vm = virDomainFindByName(&driver->domains, name);
2586 2587
    qemuDriverUnlock(driver);

2588
    if (!vm) {
2589 2590
        qemudReportError(conn, NULL, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching name '%s'"), name);
2591
        goto cleanup;
2592 2593
    }

2594
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
2595
    if (dom) dom->id = vm->def->id;
2596 2597

cleanup:
2598 2599
    if (vm)
        virDomainObjUnlock(vm);
2600 2601 2602
    return dom;
}

2603
static int qemudGetVersion(virConnectPtr conn, unsigned long *version) {
2604 2605 2606
    struct qemud_driver *driver = conn->privateData;
    int ret = -1;

2607
    qemuDriverLock(driver);
2608
    if (qemudExtractVersion(conn, driver) < 0)
2609
        goto cleanup;
2610

2611
    *version = qemu_driver->qemuVersion;
2612 2613 2614
    ret = 0;

cleanup:
2615
    qemuDriverUnlock(driver);
2616
    return ret;
D
Daniel P. Berrange 已提交
2617 2618
}

2619 2620 2621
static char *
qemudGetHostname (virConnectPtr conn)
{
2622
    char *result;
2623

2624 2625
    result = virGetHostname();
    if (result == NULL) {
2626 2627
        virReportSystemError (conn, errno,
                              "%s", _("failed to determine host name"));
2628 2629 2630
        return NULL;
    }
    /* Caller frees this string. */
2631
    return result;
2632 2633
}

2634
static int qemudListDomains(virConnectPtr conn, int *ids, int nids) {
2635
    struct qemud_driver *driver = conn->privateData;
2636 2637
    int got = 0, i;

2638 2639 2640
    qemuDriverLock(driver);
    for (i = 0 ; i < driver->domains.count && got < nids ; i++) {
        virDomainObjLock(driver->domains.objs[i]);
2641 2642
        if (virDomainIsActive(driver->domains.objs[i]))
            ids[got++] = driver->domains.objs[i]->def->id;
2643 2644 2645
        virDomainObjUnlock(driver->domains.objs[i]);
    }
    qemuDriverUnlock(driver);
2646

D
Daniel P. Berrange 已提交
2647 2648
    return got;
}
2649

2650
static int qemudNumDomains(virConnectPtr conn) {
2651
    struct qemud_driver *driver = conn->privateData;
2652 2653
    int n = 0, i;

2654 2655 2656
    qemuDriverLock(driver);
    for (i = 0 ; i < driver->domains.count ; i++) {
        virDomainObjLock(driver->domains.objs[i]);
2657
        if (virDomainIsActive(driver->domains.objs[i]))
2658
            n++;
2659 2660 2661
        virDomainObjUnlock(driver->domains.objs[i]);
    }
    qemuDriverUnlock(driver);
2662

2663
    return n;
D
Daniel P. Berrange 已提交
2664
}
2665

2666
static virDomainPtr qemudDomainCreate(virConnectPtr conn, const char *xml,
2667
                                      unsigned int flags ATTRIBUTE_UNUSED) {
2668
    struct qemud_driver *driver = conn->privateData;
2669
    virDomainDefPtr def;
2670
    virDomainObjPtr vm = NULL;
2671
    virDomainPtr dom = NULL;
2672
    virDomainEventPtr event = NULL;
D
Daniel P. Berrange 已提交
2673

2674
    qemuDriverLock(driver);
2675 2676
    if (!(def = virDomainDefParseString(conn, driver->caps, xml,
                                        VIR_DOMAIN_XML_INACTIVE)))
2677
        goto cleanup;
2678

2679 2680 2681
    if (virSecurityDriverVerify(conn, def) < 0)
        goto cleanup;

2682
    /* See if a VM with matching UUID already exists */
2683
    vm = virDomainFindByUUID(&driver->domains, def->uuid);
2684
    if (vm) {
2685 2686 2687 2688 2689 2690 2691 2692 2693
        /* 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;
        }
2694

2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712
        /* 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;
        }
2713
    }
2714

2715 2716
    if (!(vm = virDomainAssignDef(conn,
                                  &driver->domains,
2717 2718 2719 2720
                                  def)))
        goto cleanup;

    def = NULL;
D
Daniel P. Berrange 已提交
2721

2722
    if (qemudStartVMDaemon(conn, driver, vm, NULL, -1) < 0) {
2723 2724
        virDomainRemoveInactive(&driver->domains,
                                vm);
2725
        vm = NULL;
2726
        goto cleanup;
D
Daniel P. Berrange 已提交
2727
    }
2728 2729 2730 2731

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_BOOTED);
D
Daniel P. Berrange 已提交
2732

2733
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
2734
    if (dom) dom->id = vm->def->id;
2735 2736 2737

cleanup:
    virDomainDefFree(def);
2738 2739
    if (vm)
        virDomainObjUnlock(vm);
2740 2741
    if (event)
        qemuDomainEventQueue(driver, event);
2742
    qemuDriverUnlock(driver);
2743
    return dom;
D
Daniel P. Berrange 已提交
2744 2745 2746
}


2747
static int qemudDomainSuspend(virDomainPtr dom) {
2748 2749 2750
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
2751
    virDomainEventPtr event = NULL;
2752

2753
    qemuDriverLock(driver);
2754
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
2755

D
Daniel P. Berrange 已提交
2756
    if (!vm) {
2757 2758 2759 2760
        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);
2761
        goto cleanup;
D
Daniel P. Berrange 已提交
2762
    }
2763
    if (!virDomainIsActive(vm)) {
2764
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
2765
                         "%s", _("domain is not running"));
2766
        goto cleanup;
D
Daniel P. Berrange 已提交
2767
    }
2768
    if (vm->state != VIR_DOMAIN_PAUSED) {
2769
        if (qemuMonitorStopCPUs(vm) < 0)
2770 2771
            goto cleanup;
        vm->state = VIR_DOMAIN_PAUSED;
2772 2773 2774
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_SUSPENDED,
                                         VIR_DOMAIN_EVENT_SUSPENDED_PAUSED);
D
Daniel P. Berrange 已提交
2775
    }
2776
    if (virDomainSaveStatus(dom->conn, driver->stateDir, vm) < 0)
2777
        goto cleanup;
2778 2779 2780
    ret = 0;

cleanup:
2781 2782
    if (vm)
        virDomainObjUnlock(vm);
2783

2784
    if (event)
2785
        qemuDomainEventQueue(driver, event);
2786
    qemuDriverUnlock(driver);
2787
    return ret;
D
Daniel P. Berrange 已提交
2788 2789 2790
}


2791
static int qemudDomainResume(virDomainPtr dom) {
2792 2793 2794
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
2795
    virDomainEventPtr event = NULL;
2796

2797
    qemuDriverLock(driver);
2798
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
2799

D
Daniel P. Berrange 已提交
2800
    if (!vm) {
2801 2802 2803 2804
        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);
2805
        goto cleanup;
D
Daniel P. Berrange 已提交
2806
    }
2807
    if (!virDomainIsActive(vm)) {
2808
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
2809
                         "%s", _("domain is not running"));
2810
        goto cleanup;
D
Daniel P. Berrange 已提交
2811
    }
2812
    if (vm->state == VIR_DOMAIN_PAUSED) {
2813
        if (qemuMonitorStartCPUs(dom->conn, vm) < 0) {
2814 2815 2816
            if (virGetLastError() == NULL)
                qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                                 "%s", _("resume operation failed"));
2817 2818 2819
            goto cleanup;
        }
        vm->state = VIR_DOMAIN_RUNNING;
2820 2821 2822
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_UNPAUSED);
D
Daniel P. Berrange 已提交
2823
    }
2824
    if (virDomainSaveStatus(dom->conn, driver->stateDir, vm) < 0)
2825
        goto cleanup;
2826 2827 2828
    ret = 0;

cleanup:
2829 2830
    if (vm)
        virDomainObjUnlock(vm);
2831
    if (event)
2832
        qemuDomainEventQueue(driver, event);
2833
    qemuDriverUnlock(driver);
2834
    return ret;
D
Daniel P. Berrange 已提交
2835 2836 2837
}


2838
static int qemudDomainShutdown(virDomainPtr dom) {
2839 2840 2841
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
2842

2843
    qemuDriverLock(driver);
2844
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
2845 2846
    qemuDriverUnlock(driver);

2847
    if (!vm) {
2848 2849 2850 2851
        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);
2852
        goto cleanup;
2853 2854
    }

2855 2856 2857 2858 2859 2860
    if (!virDomainIsActive(vm)) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
                         "%s", _("domain is not running"));
        goto cleanup;
    }

2861
    if (qemuMonitorSystemPowerdown(vm) < 0)
2862
        goto cleanup;
2863

2864 2865 2866
    ret = 0;

cleanup:
2867 2868
    if (vm)
        virDomainObjUnlock(vm);
2869
    return ret;
2870 2871 2872
}


2873
static int qemudDomainDestroy(virDomainPtr dom) {
2874 2875 2876
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
2877
    virDomainEventPtr event = NULL;
2878

2879
    qemuDriverLock(driver);
2880
    vm  = virDomainFindByUUID(&driver->domains, dom->uuid);
D
Daniel P. Berrange 已提交
2881
    if (!vm) {
2882 2883 2884 2885
        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);
2886
        goto cleanup;
D
Daniel P. Berrange 已提交
2887
    }
2888
    if (!virDomainIsActive(vm)) {
2889
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
2890 2891 2892
                         "%s", _("domain is not running"));
        goto cleanup;
    }
2893

2894
    qemudShutdownVMDaemon(dom->conn, driver, vm);
2895 2896 2897
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_DESTROYED);
2898
    if (!vm->persistent) {
2899 2900
        virDomainRemoveInactive(&driver->domains,
                                vm);
2901 2902
        vm = NULL;
    }
2903 2904 2905
    ret = 0;

cleanup:
2906 2907
    if (vm)
        virDomainObjUnlock(vm);
2908 2909
    if (event)
        qemuDomainEventQueue(driver, event);
2910
    qemuDriverUnlock(driver);
2911
    return ret;
D
Daniel P. Berrange 已提交
2912 2913 2914
}


2915
static char *qemudDomainGetOSType(virDomainPtr dom) {
2916 2917 2918
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    char *type = NULL;
2919

2920
    qemuDriverLock(driver);
2921
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
2922
    qemuDriverUnlock(driver);
2923
    if (!vm) {
2924 2925 2926 2927
        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);
2928
        goto cleanup;
2929 2930
    }

2931
    if (!(type = strdup(vm->def->os.type)))
2932
        virReportOOMError(dom->conn);
2933 2934

cleanup:
2935 2936
    if (vm)
        virDomainObjUnlock(vm);
2937 2938 2939
    return type;
}

2940 2941
/* Returns max memory in kb, 0 if error */
static unsigned long qemudDomainGetMaxMemory(virDomainPtr dom) {
2942 2943 2944
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    unsigned long ret = 0;
2945

2946
    qemuDriverLock(driver);
2947
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
2948 2949
    qemuDriverUnlock(driver);

2950
    if (!vm) {
2951 2952
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
2953
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
2954
                         _("no domain with matching uuid '%s'"), uuidstr);
2955
        goto cleanup;
2956 2957
    }

2958 2959 2960
    ret = vm->def->maxmem;

cleanup:
2961 2962
    if (vm)
        virDomainObjUnlock(vm);
2963
    return ret;
2964 2965 2966
}

static int qemudDomainSetMaxMemory(virDomainPtr dom, unsigned long newmax) {
2967 2968 2969
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
2970

2971
    qemuDriverLock(driver);
2972
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
2973 2974
    qemuDriverUnlock(driver);

2975
    if (!vm) {
2976 2977
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
2978
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
2979
                         _("no domain with matching uuid '%s'"), uuidstr);
2980
        goto cleanup;
2981 2982 2983 2984
    }

    if (newmax < vm->def->memory) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
2985
                         "%s", _("cannot set max memory lower than current memory"));
2986
        goto cleanup;;
2987 2988 2989
    }

    vm->def->maxmem = newmax;
2990 2991 2992
    ret = 0;

cleanup:
2993 2994
    if (vm)
        virDomainObjUnlock(vm);
2995
    return ret;
2996 2997
}

2998

2999
static int qemudDomainSetMemory(virDomainPtr dom, unsigned long newmem) {
3000 3001 3002
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
3003

3004
    qemuDriverLock(driver);
3005
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3006
    qemuDriverUnlock(driver);
3007
    if (!vm) {
3008 3009
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
3010
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
3011
                         _("no domain with matching uuid '%s'"), uuidstr);
3012
        goto cleanup;
3013 3014 3015 3016
    }

    if (newmem > vm->def->maxmem) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
3017
                         "%s", _("cannot set memory higher than max memory"));
3018
        goto cleanup;
3019 3020
    }

3021
    if (virDomainIsActive(vm)) {
3022 3023 3024
        ret = qemuMonitorSetBalloon(vm, newmem);
        /* Turn lack of balloon support into a fatal error */
        if (ret == 0) {
3025 3026
            qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                             "%s", _("cannot set memory of an active domain"));
3027 3028
            ret = -1;
        }
3029 3030 3031 3032
    } else {
        vm->def->memory = newmem;
        ret = 0;
    }
3033 3034

cleanup:
3035 3036
    if (vm)
        virDomainObjUnlock(vm);
3037
    return ret;
3038 3039
}

3040
static int qemudDomainGetInfo(virDomainPtr dom,
3041
                              virDomainInfoPtr info) {
3042 3043 3044
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
3045 3046
    int err;
    unsigned long balloon;
3047

3048
    qemuDriverLock(driver);
3049
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3050
    qemuDriverUnlock(driver);
D
Daniel P. Berrange 已提交
3051
    if (!vm) {
3052 3053 3054 3055
        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);
3056
        goto cleanup;
D
Daniel P. Berrange 已提交
3057 3058
    }

3059
    info->state = vm->state;
D
Daniel P. Berrange 已提交
3060

3061
    if (!virDomainIsActive(vm)) {
3062
        info->cpuTime = 0;
D
Daniel P. Berrange 已提交
3063
    } else {
3064
        if (qemudGetProcessInfo(&(info->cpuTime), NULL, vm->pid, 0) < 0) {
3065
            qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED, ("cannot read cputime for domain"));
3066
            goto cleanup;
D
Daniel P. Berrange 已提交
3067 3068 3069
        }
    }

3070
    info->maxMem = vm->def->maxmem;
3071 3072

    if (virDomainIsActive(vm)) {
3073
        err = qemuMonitorGetBalloonInfo(vm, &balloon);
3074 3075 3076 3077 3078 3079 3080 3081 3082 3083 3084 3085
        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;
    }

3086
    info->nrVirtCpu = vm->def->vcpus;
3087 3088 3089
    ret = 0;

cleanup:
3090 3091
    if (vm)
        virDomainObjUnlock(vm);
3092
    return ret;
D
Daniel P. Berrange 已提交
3093 3094 3095
}


3096
#define QEMUD_SAVE_MAGIC "LibvirtQemudSave"
3097 3098 3099
#define QEMUD_SAVE_VERSION 2

enum qemud_save_formats {
3100 3101 3102
    QEMUD_SAVE_FORMAT_RAW = 0,
    QEMUD_SAVE_FORMAT_GZIP = 1,
    QEMUD_SAVE_FORMAT_BZIP2 = 2,
3103 3104
    /*
     * Deprecated by xz and never used as part of a release
3105
     * QEMUD_SAVE_FORMAT_LZMA
3106 3107
     */
    QEMUD_SAVE_FORMAT_XZ = 3,
3108
    QEMUD_SAVE_FORMAT_LZOP = 4,
3109 3110 3111
    /* Note: add new members only at the end.
       These values are used in the on-disk format.
       Do not change or re-use numbers. */
3112 3113

    QEMUD_SAVE_FORMAT_LAST
3114
};
3115

3116 3117 3118 3119 3120
VIR_ENUM_DECL(qemudSaveCompression)
VIR_ENUM_IMPL(qemudSaveCompression, QEMUD_SAVE_FORMAT_LAST,
              "raw",
              "gzip",
              "bzip2",
3121 3122
              "xz",
              "lzop")
3123

3124 3125 3126 3127 3128
struct qemud_save_header {
    char magic[sizeof(QEMUD_SAVE_MAGIC)-1];
    int version;
    int xml_len;
    int was_running;
3129 3130
    int compressed;
    int unused[15];
3131 3132
};

3133
static int qemudDomainSave(virDomainPtr dom,
3134 3135
                           const char *path)
{
3136
    struct qemud_driver *driver = dom->conn->privateData;
3137
    virDomainObjPtr vm = NULL;
3138 3139
    int fd = -1;
    char *xml = NULL;
3140
    struct qemud_save_header header;
3141
    int ret = -1;
3142
    virDomainEventPtr event = NULL;
3143 3144 3145 3146 3147

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

3148
    qemuDriverLock(driver);
3149 3150 3151
    if (driver->saveImageFormat == NULL)
        header.compressed = QEMUD_SAVE_FORMAT_RAW;
    else {
3152 3153 3154 3155 3156 3157
        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"));
3158
            goto cleanup;
3159
        }
3160 3161
    }

3162
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3163

D
Daniel P. Berrange 已提交
3164
    if (!vm) {
3165 3166 3167 3168
        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);
3169
        goto cleanup;
D
Daniel P. Berrange 已提交
3170
    }
3171

3172
    if (!virDomainIsActive(vm)) {
3173
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
3174
                         "%s", _("domain is not running"));
3175
        goto cleanup;
D
Daniel P. Berrange 已提交
3176
    }
3177 3178 3179 3180

    /* Pause */
    if (vm->state == VIR_DOMAIN_RUNNING) {
        header.was_running = 1;
3181
        if (qemuMonitorStopCPUs(vm) < 0)
3182
            goto cleanup;
3183
        vm->state = VIR_DOMAIN_PAUSED;
3184 3185 3186
    }

    /* Get XML for the domain */
3187
    xml = virDomainDefFormat(dom->conn, vm->def, VIR_DOMAIN_XML_SECURE);
3188 3189
    if (!xml) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
3190
                         "%s", _("failed to get domain xml"));
3191
        goto cleanup;
3192 3193 3194 3195 3196 3197
    }
    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,
3198
                         _("failed to create '%s'"), path);
3199
        goto cleanup;
3200 3201 3202 3203
    }

    if (safewrite(fd, &header, sizeof(header)) != sizeof(header)) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
3204
                         "%s", _("failed to write save header"));
3205
        goto cleanup;
3206 3207 3208 3209
    }

    if (safewrite(fd, xml, header.xml_len) != header.xml_len) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
3210
                         "%s", _("failed to write xml"));
3211
        goto cleanup;
3212 3213
    }

3214
    if (close(fd) < 0) {
3215 3216 3217
        virReportSystemError(dom->conn, errno,
                             _("unable to save file %s"),
                             path);
3218 3219 3220
        goto cleanup;
    }
    fd = -1;
3221

3222 3223 3224 3225
    if (header.compressed == QEMUD_SAVE_FORMAT_RAW) {
        const char *args[] = { "cat", NULL };
        ret = qemuMonitorMigrateToCommand(vm, args, path);
    } else {
3226
        const char *prog = qemudSaveCompressionTypeToString(header.compressed);
3227 3228 3229 3230 3231 3232
        const char *args[] = {
            prog,
            "-c",
            NULL
        };
        ret = qemuMonitorMigrateToCommand(vm, args, path);
3233 3234
    }

3235
    if (ret < 0)
3236
        goto cleanup;
3237

3238 3239
    /* Shut it down */
    qemudShutdownVMDaemon(dom->conn, driver, vm);
3240 3241 3242
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_SAVED);
3243
    if (!vm->persistent) {
3244 3245
        virDomainRemoveInactive(&driver->domains,
                                vm);
3246 3247
        vm = NULL;
    }
3248 3249 3250 3251 3252 3253 3254

cleanup:
    if (fd != -1)
        close(fd);
    VIR_FREE(xml);
    if (ret != 0)
        unlink(path);
3255 3256
    if (vm)
        virDomainObjUnlock(vm);
3257 3258
    if (event)
        qemuDomainEventQueue(driver, event);
3259
    qemuDriverUnlock(driver);
3260
    return ret;
D
Daniel P. Berrange 已提交
3261 3262 3263
}


P
Paolo Bonzini 已提交
3264 3265 3266 3267 3268 3269 3270
static int qemudDomainCoreDump(virDomainPtr dom,
                               const char *path,
                               int flags ATTRIBUTE_UNUSED) {
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int resume = 0, paused = 0;
    int ret = -1;
3271 3272 3273 3274
    const char *args[] = {
        "cat",
        NULL,
    };
P
Paolo Bonzini 已提交
3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300

    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) {
3301
        if (qemuMonitorStopCPUs(vm) < 0)
P
Paolo Bonzini 已提交
3302 3303 3304 3305
            goto cleanup;
        paused = 1;
    }

3306
    ret = qemuMonitorMigrateToCommand(vm, args, path);
P
Paolo Bonzini 已提交
3307 3308 3309 3310 3311 3312 3313
    paused = 1;
cleanup:

    /* 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) {
3314
        if (qemuMonitorStartCPUs(dom->conn, vm) < 0) {
3315 3316 3317
            if (virGetLastError() == NULL)
                qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                                 "%s", _("resuming after dump failed"));
P
Paolo Bonzini 已提交
3318 3319 3320 3321 3322 3323 3324 3325
        }
    }
    if (vm)
        virDomainObjUnlock(vm);
    return ret;
}


3326
static int qemudDomainSetVcpus(virDomainPtr dom, unsigned int nvcpus) {
3327 3328
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
3329
    int max;
3330
    int ret = -1;
3331
    const char *type;
3332

3333
    qemuDriverLock(driver);
3334
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3335 3336
    qemuDriverUnlock(driver);

3337
    if (!vm) {
3338 3339
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
3340
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
3341
                         _("no domain with matching uuid '%s'"), uuidstr);
3342
        goto cleanup;
3343 3344
    }

3345
    if (virDomainIsActive(vm)) {
3346
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID, "%s",
3347
                         _("cannot change vcpu count of an active domain"));
3348
        goto cleanup;
3349 3350
    }

3351 3352 3353 3354 3355 3356 3357 3358
    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) {
3359 3360
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INTERNAL_ERROR, "%s",
                         _("could not determine max vcpus for the domain"));
3361
        goto cleanup;
3362 3363 3364 3365 3366 3367
    }

    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);
3368
        goto cleanup;
3369 3370 3371
    }

    vm->def->vcpus = nvcpus;
3372 3373 3374
    ret = 0;

cleanup:
3375 3376
    if (vm)
        virDomainObjUnlock(vm);
3377
    return ret;
3378 3379
}

3380 3381 3382 3383 3384 3385 3386

#if HAVE_SCHED_GETAFFINITY
static int
qemudDomainPinVcpu(virDomainPtr dom,
                   unsigned int vcpu,
                   unsigned char *cpumap,
                   int maplen) {
3387 3388
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
3389 3390 3391
    cpu_set_t mask;
    int i, maxcpu;
    virNodeInfo nodeinfo;
3392
    int ret = -1;
3393

3394
    qemuDriverLock(driver);
3395
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3396 3397
    qemuDriverUnlock(driver);

3398 3399 3400 3401 3402 3403 3404 3405
    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;
    }

3406
    if (!virDomainIsActive(vm)) {
3407
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
3408
                         "%s",_("cannot pin vcpus on an inactive domain"));
3409
        goto cleanup;
3410 3411 3412 3413 3414 3415
    }

    if (vcpu > (vm->nvcpupids-1)) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
                         _("vcpu number out of range %d > %d"),
                         vcpu, vm->nvcpupids);
3416
        goto cleanup;
3417 3418
    }

3419
    if (nodeGetInfo(dom->conn, &nodeinfo) < 0)
3420
        goto cleanup;
3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433

    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) {
3434 3435
            virReportSystemError(dom->conn, errno, "%s",
                                 _("cannot set affinity"));
3436
            goto cleanup;
3437 3438 3439 3440
        }
    } else {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                         "%s", _("cpu affinity is not supported"));
3441
        goto cleanup;
3442
    }
3443
    ret = 0;
3444

3445
cleanup:
3446 3447
    if (vm)
        virDomainObjUnlock(vm);
3448
    return ret;
3449 3450 3451 3452 3453 3454 3455 3456
}

static int
qemudDomainGetVcpus(virDomainPtr dom,
                    virVcpuInfoPtr info,
                    int maxinfo,
                    unsigned char *cpumaps,
                    int maplen) {
3457 3458
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
3459 3460
    virNodeInfo nodeinfo;
    int i, v, maxcpu;
3461
    int ret = -1;
3462

3463
    qemuDriverLock(driver);
3464
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3465 3466
    qemuDriverUnlock(driver);

3467 3468 3469 3470 3471 3472 3473 3474
    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;
    }

3475
    if (!virDomainIsActive(vm)) {
3476
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
3477
                         "%s",_("cannot pin vcpus on an inactive domain"));
3478
        goto cleanup;
3479 3480
    }

3481
    if (nodeGetInfo(dom->conn, &nodeinfo) < 0)
3482
        goto cleanup;
3483 3484 3485 3486 3487 3488 3489 3490 3491

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

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

3492 3493 3494 3495 3496 3497
    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;
3498 3499 3500 3501 3502 3503 3504 3505 3506 3507

                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;
                }
3508
            }
3509 3510
        }

3511 3512 3513 3514 3515 3516 3517 3518 3519
        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) {
3520 3521
                        virReportSystemError(dom->conn, errno, "%s",
                                             _("cannot get affinity"));
3522 3523 3524 3525 3526 3527
                        goto cleanup;
                    }

                    for (i = 0 ; i < maxcpu ; i++)
                        if (CPU_ISSET(i, &mask))
                            VIR_USE_CPU(cpumap, i);
3528
                }
3529 3530 3531 3532
            } else {
                qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                                 "%s", _("cpu affinity is not available"));
                goto cleanup;
3533 3534 3535
            }
        }
    }
3536
    ret = maxinfo;
3537

3538
cleanup:
3539 3540
    if (vm)
        virDomainObjUnlock(vm);
3541
    return ret;
3542 3543 3544 3545
}
#endif /* HAVE_SCHED_GETAFFINITY */


3546
static int qemudDomainGetMaxVcpus(virDomainPtr dom) {
3547 3548
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
3549
    const char *type;
3550
    int ret = -1;
3551

3552
    qemuDriverLock(driver);
3553
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3554 3555
    qemuDriverUnlock(driver);

3556
    if (!vm) {
3557 3558
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
3559
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
3560
                         _("no domain with matching uuid '%s'"), uuidstr);
3561
        goto cleanup;
3562 3563
    }

3564
    if (!(type = virDomainVirtTypeToString(vm->def->virtType))) {
3565 3566 3567
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
                         _("unknown virt type in domain definition '%d'"),
                         vm->def->virtType);
3568
        goto cleanup;
3569 3570
    }

3571
    ret = qemudGetMaxVCPUs(dom->conn, type);
3572

3573
cleanup:
3574 3575
    if (vm)
        virDomainObjUnlock(vm);
3576 3577 3578
    return ret;
}

3579 3580 3581 3582 3583 3584 3585 3586 3587 3588
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);

3589 3590
    memset(seclabel, 0, sizeof(*seclabel));

3591 3592 3593
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
3594
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623
                         _("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,
3624
                                 "%s", _("Failed to get security label"));
3625 3626 3627 3628 3629 3630 3631 3632 3633 3634
                goto cleanup;
            }
        }
    }

    ret = 0;

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
3635
    qemuDriverUnlock(driver);
3636 3637 3638
    return ret;
}

3639 3640
static int qemudNodeGetSecurityModel(virConnectPtr conn,
                                     virSecurityModelPtr secmodel)
3641 3642 3643
{
    struct qemud_driver *driver = (struct qemud_driver *)conn->privateData;
    char *p;
3644
    int ret = 0;
3645

3646 3647
    qemuDriverLock(driver);
    if (!driver->securityDriver) {
3648
        memset(secmodel, 0, sizeof (*secmodel));
3649 3650
        goto cleanup;
    }
3651

3652 3653 3654 3655 3656
    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);
3657 3658
        ret = -1;
        goto cleanup;
3659 3660 3661 3662 3663 3664 3665 3666
    }
    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);
3667 3668
        ret = -1;
        goto cleanup;
3669 3670
    }
    strcpy(secmodel->doi, p);
3671 3672 3673 3674

cleanup:
    qemuDriverUnlock(driver);
    return ret;
3675 3676 3677
}

/* TODO: check seclabel restore */
3678
static int qemudDomainRestore(virConnectPtr conn,
3679 3680 3681
                              const char *path) {
    struct qemud_driver *driver = conn->privateData;
    virDomainDefPtr def = NULL;
3682
    virDomainObjPtr vm = NULL;
3683 3684 3685
    int fd = -1;
    int ret = -1;
    char *xml = NULL;
3686
    struct qemud_save_header header;
3687
    virDomainEventPtr event = NULL;
3688 3689 3690
    int intermediatefd = -1;
    pid_t intermediate_pid = -1;
    int childstat;
3691

3692
    qemuDriverLock(driver);
3693 3694 3695
    /* Verify the header and read the XML */
    if ((fd = open(path, O_RDONLY)) < 0) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
3696
                         "%s", _("cannot read domain image"));
3697
        goto cleanup;
3698 3699 3700 3701
    }

    if (saferead(fd, &header, sizeof(header)) != sizeof(header)) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
3702
                         "%s", _("failed to read qemu header"));
3703
        goto cleanup;
3704 3705 3706 3707
    }

    if (memcmp(header.magic, QEMUD_SAVE_MAGIC, sizeof(header.magic)) != 0) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
3708
                         "%s", _("image magic is incorrect"));
3709
        goto cleanup;
3710 3711 3712 3713
    }

    if (header.version > QEMUD_SAVE_VERSION) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
3714
                         _("image version is not supported (%d > %d)"),
3715
                         header.version, QEMUD_SAVE_VERSION);
3716
        goto cleanup;
3717 3718
    }

3719
    if (VIR_ALLOC_N(xml, header.xml_len) < 0) {
3720
        virReportOOMError(conn);
3721
        goto cleanup;
3722 3723 3724 3725
    }

    if (saferead(fd, xml, header.xml_len) != header.xml_len) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
3726
                         "%s", _("failed to read XML"));
3727
        goto cleanup;
3728 3729 3730
    }

    /* Create a domain from this XML */
3731 3732
    if (!(def = virDomainDefParseString(conn, driver->caps, xml,
                                        VIR_DOMAIN_XML_INACTIVE))) {
3733
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
3734
                         "%s", _("failed to parse XML"));
3735
        goto cleanup;
3736 3737
    }

3738
    /* See if a VM with matching UUID already exists */
3739
    vm = virDomainFindByUUID(&driver->domains, def->uuid);
3740
    if (vm) {
3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751
        /* 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 */
3752
        if (virDomainIsActive(vm)) {
3753
            qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_INVALID,
3754 3755
                             _("domain is already active as '%s'"), vm->def->name);
            goto cleanup;
3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767
        }
        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;
3768
        }
3769 3770
    }

3771 3772 3773
    if (!(vm = virDomainAssignDef(conn,
                                  &driver->domains,
                                  def))) {
3774
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
3775
                         "%s", _("failed to assign new VM"));
3776
        goto cleanup;
3777
    }
3778
    def = NULL;
3779

3780 3781
    if (header.version == 2) {
        const char *intermediate_argv[3] = { NULL, "-dc", NULL };
3782 3783
        const char *prog = qemudSaveCompressionTypeToString(header.compressed);
        if (prog == NULL) {
3784
            qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
3785
                             _("Invalid compressed save format %d"),
3786 3787 3788
                             header.compressed);
            goto cleanup;
        }
3789 3790 3791 3792

        if (header.compressed != QEMUD_SAVE_FORMAT_RAW)
            intermediate_argv[0] = prog;
        else {
3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803
            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;
            }
        }
    }
3804
    /* Set the migration source and start it up. */
3805
    ret = qemudStartVMDaemon(conn, driver, vm, "stdio", fd);
3806 3807 3808 3809 3810 3811 3812
    if (intermediate_pid != -1) {
        /* Wait for intermediate process to exit */
        while (waitpid(intermediate_pid, &childstat, 0) == -1 &&
               errno == EINTR);
    }
    if (intermediatefd != -1)
        close(intermediatefd);
3813
    close(fd);
3814
    fd = -1;
3815
    if (ret < 0) {
3816
        if (!vm->persistent) {
3817 3818
            virDomainRemoveInactive(&driver->domains,
                                    vm);
3819 3820
            vm = NULL;
        }
3821
        goto cleanup;
3822 3823
    }

3824 3825 3826
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_RESTORED);
3827

3828 3829
    /* If it was running before, resume it now. */
    if (header.was_running) {
3830
        if (qemuMonitorStartCPUs(conn, vm) < 0) {
3831 3832 3833
            if (virGetLastError() == NULL)
                qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                                 "%s", _("failed to resume domain"));
3834
            goto cleanup;
3835 3836
        }
        vm->state = VIR_DOMAIN_RUNNING;
3837
        virDomainSaveStatus(conn, driver->stateDir, vm);
3838
    }
3839
    ret = 0;
3840

3841 3842 3843 3844 3845
cleanup:
    virDomainDefFree(def);
    VIR_FREE(xml);
    if (fd != -1)
        close(fd);
3846 3847
    if (vm)
        virDomainObjUnlock(vm);
3848 3849
    if (event)
        qemuDomainEventQueue(driver, event);
3850
    qemuDriverUnlock(driver);
3851
    return ret;
D
Daniel P. Berrange 已提交
3852 3853 3854
}


3855
static char *qemudDomainDumpXML(virDomainPtr dom,
3856
                                int flags) {
3857 3858 3859
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    char *ret = NULL;
3860 3861
    unsigned long balloon;
    int err;
3862

3863
    qemuDriverLock(driver);
3864
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3865 3866
    qemuDriverUnlock(driver);

D
Daniel P. Berrange 已提交
3867
    if (!vm) {
3868 3869 3870 3871
        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);
3872
        goto cleanup;
D
Daniel P. Berrange 已提交
3873 3874
    }

3875
    /* Refresh current memory based on balloon info */
3876 3877 3878 3879 3880 3881 3882 3883
    if (virDomainIsActive(vm)) {
        err = qemuMonitorGetBalloonInfo(vm, &balloon);
        if (err < 0)
            goto cleanup;
        if (err > 0)
            vm->def->memory = balloon;
        /* err == 0 indicates no balloon support, so ignore it */
    }
3884

3885 3886 3887 3888 3889 3890
    ret = virDomainDefFormat(dom->conn,
                             (flags & VIR_DOMAIN_XML_INACTIVE) && vm->newDef ?
                             vm->newDef : vm->def,
                             flags);

cleanup:
3891 3892
    if (vm)
        virDomainObjUnlock(vm);
3893
    return ret;
D
Daniel P. Berrange 已提交
3894 3895 3896
}


3897 3898 3899 3900
static char *qemuDomainXMLFromNative(virConnectPtr conn,
                                     const char *format,
                                     const char *config,
                                     unsigned int flags ATTRIBUTE_UNUSED) {
3901
    struct qemud_driver *driver = conn->privateData;
3902 3903 3904 3905 3906 3907 3908 3909 3910
    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;
    }

3911
    qemuDriverLock(driver);
3912
    def = qemuParseCommandLineString(conn, driver->caps, config);
3913
    qemuDriverUnlock(driver);
3914 3915 3916 3917 3918 3919 3920 3921 3922 3923
    if (!def)
        goto cleanup;

    xml = virDomainDefFormat(conn, def, VIR_DOMAIN_XML_INACTIVE);

cleanup:
    virDomainDefFree(def);
    return xml;
}

3924 3925 3926 3927 3928 3929
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;
3930
    virDomainChrDef monitor_chr;
3931 3932 3933 3934 3935 3936 3937 3938 3939 3940
    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;

3941 3942
    qemuDriverLock(driver);

3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985 3986 3987 3988 3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 4005 4006 4007 4008
    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;
    }

4009 4010
    if (qemuPrepareMonitorChr(conn, driver, &monitor_chr, def->name) < 0)
        goto cleanup;
4011 4012

    if (qemudBuildCommandLine(conn, driver, def,
4013
                              &monitor_chr, qemuCmdFlags,
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
                              &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:
4039
    qemuDriverUnlock(driver);
4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052
    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;
}


4053
static int qemudListDefinedDomains(virConnectPtr conn,
4054
                            char **const names, int nnames) {
4055
    struct qemud_driver *driver = conn->privateData;
4056
    int got = 0, i;
4057

4058
    qemuDriverLock(driver);
4059
    for (i = 0 ; i < driver->domains.count && got < nnames ; i++) {
4060
        virDomainObjLock(driver->domains.objs[i]);
4061 4062
        if (!virDomainIsActive(driver->domains.objs[i])) {
            if (!(names[got++] = strdup(driver->domains.objs[i]->def->name))) {
4063
                virReportOOMError(conn);
4064
                virDomainObjUnlock(driver->domains.objs[i]);
4065 4066
                goto cleanup;
            }
4067
        }
4068
        virDomainObjUnlock(driver->domains.objs[i]);
D
Daniel P. Berrange 已提交
4069
    }
4070

4071
    qemuDriverUnlock(driver);
D
Daniel P. Berrange 已提交
4072
    return got;
4073 4074 4075

 cleanup:
    for (i = 0 ; i < got ; i++)
4076
        VIR_FREE(names[i]);
4077
    qemuDriverUnlock(driver);
4078
    return -1;
D
Daniel P. Berrange 已提交
4079 4080
}

4081
static int qemudNumDefinedDomains(virConnectPtr conn) {
4082
    struct qemud_driver *driver = conn->privateData;
4083 4084
    int n = 0, i;

4085
    qemuDriverLock(driver);
4086 4087
    for (i = 0 ; i < driver->domains.count ; i++)
        if (!virDomainIsActive(driver->domains.objs[i]))
4088
            n++;
4089
    qemuDriverUnlock(driver);
4090

4091
    return n;
D
Daniel P. Berrange 已提交
4092 4093 4094
}


4095
static int qemudDomainStart(virDomainPtr dom) {
4096 4097 4098
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4099
    virDomainEventPtr event = NULL;
4100

4101
    qemuDriverLock(driver);
4102
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4103

4104
    if (!vm) {
4105 4106 4107 4108
        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);
4109
        goto cleanup;
4110 4111
    }

4112
    ret = qemudStartVMDaemon(dom->conn, driver, vm, NULL, -1);
4113
    if (ret != -1)
4114 4115 4116
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STARTED,
                                         VIR_DOMAIN_EVENT_STARTED_BOOTED);
4117 4118

cleanup:
4119 4120
    if (vm)
        virDomainObjUnlock(vm);
4121
    if (event)
4122
        qemuDomainEventQueue(driver, event);
4123
    qemuDriverUnlock(driver);
4124
    return ret;
D
Daniel P. Berrange 已提交
4125 4126
}

4127 4128 4129 4130 4131 4132 4133 4134 4135 4136 4137 4138 4139 4140 4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155
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;
}

4156 4157 4158 4159 4160 4161 4162 4163 4164 4165 4166 4167 4168 4169 4170 4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 4181 4182 4183
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;
}

4184 4185
int
qemudCanonicalizeMachine(struct qemud_driver *driver, virDomainDefPtr def)
4186 4187 4188 4189 4190 4191
{
    char *canonical = NULL;
    int i;

    for (i = 0; i < driver->caps->nguests; i++) {
        virCapsGuestPtr guest = driver->caps->guests[i];
4192
        virCapsGuestDomainInfoPtr info;
4193 4194 4195
        int j;

        for (j = 0; j < guest->arch.ndomains; j++) {
4196
            info = &guest->arch.domains[j]->info;
4197

4198 4199 4200 4201 4202 4203 4204 4205 4206
            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;
4207 4208
        }

4209 4210 4211 4212
        info = &guest->arch.defaultInfo;

        if (info->emulator && STREQ(info->emulator, def->emulator)) {
            if (qemudCanonicalizeMachineFromInfo(def, info, &canonical) < 0)
4213 4214 4215 4216
                return -1;
            goto out;
        }
    }
4217 4218 4219 4220

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

4221 4222 4223 4224 4225 4226 4227
out:
    if (canonical) {
        VIR_FREE(def->os.machine);
        def->os.machine = canonical;
    }
    return 0;
}
D
Daniel P. Berrange 已提交
4228

4229
static virDomainPtr qemudDomainDefine(virConnectPtr conn, const char *xml) {
4230
    struct qemud_driver *driver = conn->privateData;
4231
    virDomainDefPtr def;
4232
    virDomainObjPtr vm = NULL;
4233
    virDomainPtr dom = NULL;
4234
    virDomainEventPtr event = NULL;
4235
    int newVM = 1;
4236

4237
    qemuDriverLock(driver);
4238 4239
    if (!(def = virDomainDefParseString(conn, driver->caps, xml,
                                        VIR_DOMAIN_XML_INACTIVE)))
4240
        goto cleanup;
4241

4242 4243 4244
    if (virSecurityDriverVerify(conn, def) < 0)
        goto cleanup;

4245 4246
    /* See if a VM with matching UUID already exists */
    vm = virDomainFindByUUID(&driver->domains, def->uuid);
4247
    if (vm) {
4248 4249 4250 4251 4252 4253 4254 4255 4256 4257 4258
        /* 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 */
4259
        virDomainObjUnlock(vm);
4260
        newVM = 0;
4261 4262 4263 4264 4265 4266 4267 4268 4269 4270 4271
    } 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;
        }
4272
    }
4273

4274
    if (qemudCanonicalizeMachine(driver, def) < 0)
4275 4276
        goto cleanup;

4277 4278 4279
    if (!(vm = virDomainAssignDef(conn,
                                  &driver->domains,
                                  def))) {
4280
        goto cleanup;
4281
    }
4282
    def = NULL;
4283
    vm->persistent = 1;
4284

4285 4286
    if (virDomainSaveConfig(conn,
                            driver->configDir,
4287
                            vm->newDef ? vm->newDef : vm->def) < 0) {
4288 4289
        virDomainRemoveInactive(&driver->domains,
                                vm);
4290
        vm = NULL;
4291
        goto cleanup;
4292 4293
    }

4294 4295 4296 4297 4298
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_DEFINED,
                                     newVM ?
                                     VIR_DOMAIN_EVENT_DEFINED_ADDED :
                                     VIR_DOMAIN_EVENT_DEFINED_UPDATED);
4299

4300
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
4301
    if (dom) dom->id = vm->def->id;
4302 4303

cleanup:
4304
    virDomainDefFree(def);
4305 4306
    if (vm)
        virDomainObjUnlock(vm);
4307 4308
    if (event)
        qemuDomainEventQueue(driver, event);
4309
    qemuDriverUnlock(driver);
4310
    return dom;
D
Daniel P. Berrange 已提交
4311 4312
}

4313
static int qemudDomainUndefine(virDomainPtr dom) {
4314 4315
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
4316
    virDomainEventPtr event = NULL;
4317
    int ret = -1;
D
Daniel P. Berrange 已提交
4318

4319
    qemuDriverLock(driver);
4320
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4321

D
Daniel P. Berrange 已提交
4322
    if (!vm) {
4323 4324 4325 4326
        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);
4327
        goto cleanup;
D
Daniel P. Berrange 已提交
4328 4329
    }

4330
    if (virDomainIsActive(vm)) {
4331
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
4332
                         "%s", _("cannot delete active domain"));
4333
        goto cleanup;
D
Daniel P. Berrange 已提交
4334 4335
    }

4336 4337 4338
    if (!vm->persistent) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
                         "%s", _("cannot undefine transient domain"));
4339
        goto cleanup;
4340 4341 4342
    }

    if (virDomainDeleteConfig(dom->conn, driver->configDir, driver->autostartDir, vm) < 0)
4343
        goto cleanup;
D
Daniel P. Berrange 已提交
4344

4345 4346 4347
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_UNDEFINED,
                                     VIR_DOMAIN_EVENT_UNDEFINED_REMOVED);
4348

4349 4350
    virDomainRemoveInactive(&driver->domains,
                            vm);
4351
    vm = NULL;
4352
    ret = 0;
D
Daniel P. Berrange 已提交
4353

4354
cleanup:
4355 4356
    if (vm)
        virDomainObjUnlock(vm);
4357 4358
    if (event)
        qemuDomainEventQueue(driver, event);
4359
    qemuDriverUnlock(driver);
4360
    return ret;
D
Daniel P. Berrange 已提交
4361 4362
}

4363
/* Return the disks name for use in monitor commands */
4364
static char *qemudDiskDeviceName(const virConnectPtr conn,
4365
                                 const virDomainDiskDefPtr disk) {
4366 4367 4368 4369 4370 4371

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

    if (virDiskNameToBusDeviceIndex(disk, &busid, &devid) < 0) {
4372
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
4373 4374 4375 4376 4377 4378 4379
                         _("cannot convert disk '%s' to bus/device index"),
                         disk->dst);
        return NULL;
    }

    switch (disk->bus) {
        case VIR_DOMAIN_DISK_BUS_IDE:
4380
            if (disk->device== VIR_DOMAIN_DISK_DEVICE_DISK)
4381
                ret = virAsprintf(&devname, "ide%d-hd%d", busid, devid);
4382
            else
4383
                ret = virAsprintf(&devname, "ide%d-cd%d", busid, devid);
4384 4385
            break;
        case VIR_DOMAIN_DISK_BUS_SCSI:
4386
            if (disk->device == VIR_DOMAIN_DISK_DEVICE_DISK)
4387
                ret = virAsprintf(&devname, "scsi%d-hd%d", busid, devid);
4388
            else
4389
                ret = virAsprintf(&devname, "scsi%d-cd%d", busid, devid);
4390 4391
            break;
        case VIR_DOMAIN_DISK_BUS_FDC:
4392
            ret = virAsprintf(&devname, "floppy%d", devid);
4393 4394
            break;
        case VIR_DOMAIN_DISK_BUS_VIRTIO:
4395
            ret = virAsprintf(&devname, "virtio%d", devid);
4396 4397
            break;
        default:
4398
            qemudReportError(conn, NULL, NULL, VIR_ERR_NO_SUPPORT,
4399 4400 4401 4402 4403 4404
                             _("Unsupported disk name mapping for bus '%s'"),
                             virDomainDiskBusTypeToString(disk->bus));
            return NULL;
    }

    if (ret == -1) {
4405
        virReportOOMError(conn);
4406 4407 4408 4409 4410 4411
        return NULL;
    }

    return devname;
}

4412 4413
static int qemudDomainChangeEjectableMedia(virConnectPtr conn,
                                           virDomainObjPtr vm,
4414 4415
                                           virDomainDeviceDefPtr dev,
                                           unsigned int qemuCmdFlags)
4416
{
4417
    virDomainDiskDefPtr origdisk = NULL, newdisk;
4418
    char *devname = NULL;
4419
    int i;
4420
    int ret;
4421

4422
    origdisk = NULL;
4423
    newdisk = dev->data.disk;
4424 4425 4426 4427
    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];
4428
            break;
4429
        }
4430 4431 4432
    }

    if (!origdisk) {
4433
        qemudReportError(conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
4434 4435 4436 4437 4438 4439 4440
                         _("No device with bus '%s' and target '%s'"),
                         virDomainDiskBusTypeToString(newdisk->bus),
                         newdisk->dst);
        return -1;
    }

    if (qemuCmdFlags & QEMUD_CMD_FLAG_DRIVE) {
4441
        if (!(devname = qemudDiskDeviceName(conn, newdisk)))
4442 4443 4444 4445 4446 4447 4448 4449 4450
            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 {
4451
            qemudReportError(conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
4452 4453 4454 4455 4456 4457 4458 4459
                             _("Emulator version does not support removable "
                               "media for device '%s' and target '%s'"),
                               virDomainDiskDeviceTypeToString(newdisk->device),
                               newdisk->dst);
            return -1;
        }

        if (!devname) {
4460
            virReportOOMError(conn);
4461 4462 4463
            return -1;
        }
    }
4464

4465
    if (newdisk->src) {
4466 4467 4468
        ret = qemuMonitorChangeMedia(vm, devname, newdisk->src);
    } else {
        ret = qemuMonitorEjectMedia(vm, devname);
4469
    }
4470

4471 4472 4473 4474 4475
    if (ret == 0) {
        VIR_FREE(origdisk->src);
        origdisk->src = newdisk->src;
        newdisk->src = NULL;
        origdisk->type = newdisk->type;
4476
    }
4477

4478
    return ret;
4479 4480
}

4481

4482 4483 4484
static int qemudDomainAttachPciDiskDevice(virConnectPtr conn,
                                          virDomainObjPtr vm,
                                          virDomainDeviceDefPtr dev)
4485
{
4486
    int i;
4487 4488 4489 4490
    const char* type = virDomainDiskBusTypeToString(dev->data.disk->bus);

    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (STREQ(vm->def->disks[i]->dst, dev->data.disk->dst)) {
4491
            qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
4492 4493 4494 4495 4496 4497
                           _("target %s already exists"), dev->data.disk->dst);
            return -1;
        }
    }

    if (VIR_REALLOC_N(vm->def->disks, vm->def->ndisks+1) < 0) {
4498
        virReportOOMError(conn);
4499 4500 4501
        return -1;
    }

4502 4503 4504 4505 4506 4507
    if (qemuMonitorAddPCIDisk(vm,
                              dev->data.disk->src,
                              type,
                              &dev->data.disk->pci_addr.domain,
                              &dev->data.disk->pci_addr.bus,
                              &dev->data.disk->pci_addr.slot) < 0)
4508
        return -1;
4509

4510
    virDomainDiskInsertPreAlloced(vm->def, dev->data.disk);
4511 4512 4513

    return 0;
}
4514

4515 4516 4517
static int qemudDomainAttachUsbMassstorageDevice(virConnectPtr conn,
                                                 virDomainObjPtr vm,
                                                 virDomainDeviceDefPtr dev)
4518
{
4519
    int i;
4520

4521 4522
    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (STREQ(vm->def->disks[i]->dst, dev->data.disk->dst)) {
4523
            qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
4524 4525 4526 4527 4528
                           _("target %s already exists"), dev->data.disk->dst);
            return -1;
        }
    }

4529 4530 4531
    if (!dev->data.disk->src) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                         "%s", _("disk source path is missing"));
4532 4533 4534
        return -1;
    }

4535
    if (VIR_REALLOC_N(vm->def->disks, vm->def->ndisks+1) < 0) {
4536
        virReportOOMError(conn);
4537 4538 4539
        return -1;
    }

4540
    if (qemuMonitorAddUSBDisk(vm, dev->data.disk->src) < 0)
4541
        return -1;
4542

4543
    virDomainDiskInsertPreAlloced(vm->def, dev->data.disk);
4544

4545 4546 4547
    return 0;
}

M
Mark McLoughlin 已提交
4548
static int qemudDomainAttachNetDevice(virConnectPtr conn,
4549
                                      struct qemud_driver *driver,
M
Mark McLoughlin 已提交
4550 4551 4552 4553 4554
                                      virDomainObjPtr vm,
                                      virDomainDeviceDefPtr dev,
                                      unsigned int qemuCmdFlags)
{
    virDomainNetDefPtr net = dev->data.net;
4555
    char *tapfd_name = NULL;
4556
    int i, tapfd = -1;
4557
    char *nicstr = NULL;
4558
    char *netstr = NULL;
4559
    int ret = -1;
M
Mark McLoughlin 已提交
4560 4561 4562 4563 4564 4565 4566 4567 4568

    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) {
4569 4570 4571 4572 4573 4574 4575 4576 4577 4578
        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
4579 4580
    }

4581 4582
    if (VIR_REALLOC_N(vm->def->nets, vm->def->nnets+1) < 0)
        goto no_memory;
M
Mark McLoughlin 已提交
4583 4584

    if ((qemuCmdFlags & QEMUD_CMD_FLAG_NET_NAME) &&
4585 4586
        qemuAssignNetNames(vm->def, net) < 0)
        goto no_memory;
M
Mark McLoughlin 已提交
4587 4588 4589 4590 4591 4592 4593 4594 4595

    /* 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;

4596 4597 4598 4599
    if (tapfd != -1) {
        if (virAsprintf(&tapfd_name, "fd-%s", net->hostnet_name) < 0)
            goto no_memory;

4600
        if (qemuMonitorSendFileHandle(vm, tapfd_name, tapfd) < 0)
4601 4602 4603
            goto cleanup;
    }

4604 4605
    if (qemuBuildHostNetStr(conn, net, ' ',
                            net->vlan, tapfd_name, &netstr) < 0)
4606
        goto try_tapfd_close;
M
Mark McLoughlin 已提交
4607

4608
    if (qemuMonitorAddHostNetwork(vm, netstr) < 0)
4609
        goto try_tapfd_close;
M
Mark McLoughlin 已提交
4610

4611 4612 4613
    if (tapfd != -1)
        close(tapfd);
    tapfd = -1;
M
Mark McLoughlin 已提交
4614

4615
    if (qemuBuildNicStr(conn, net, NULL, net->vlan, &nicstr) < 0)
4616
        goto try_remove;
M
Mark McLoughlin 已提交
4617

4618 4619 4620 4621
    if (qemuMonitorAddPCINetwork(vm, nicstr,
                                 &net->pci_addr.domain,
                                 &net->pci_addr.bus,
                                 &net->pci_addr.slot) < 0)
4622 4623
        goto try_remove;

4624
    ret = 0;
M
Mark McLoughlin 已提交
4625 4626 4627

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

4628 4629 4630 4631 4632 4633
cleanup:
    VIR_FREE(nicstr);
    VIR_FREE(netstr);
    VIR_FREE(tapfd_name);
    if (tapfd != -1)
        close(tapfd);
4634

4635
    return ret;
4636

4637 4638
try_remove:
    if (!net->hostnet_name || net->vlan == 0)
4639
        VIR_WARN0(_("Unable to remove network backend\n"));
4640 4641 4642
    else if (qemuMonitorRemoveHostNetwork(vm, net->vlan, net->hostnet_name) < 0)
        VIR_WARN(_("Failed to remove network backend for vlan %d, net %s"),
                 net->vlan, net->hostnet_name);
4643
    goto cleanup;
4644

4645
try_tapfd_close:
4646 4647 4648 4649
    if (tapfd_name &&
        qemuMonitorCloseFileHandle(vm, tapfd_name) < 0)
        VIR_WARN(_("Failed to close tapfd with '%s'\n"), tapfd_name);

4650 4651
    goto cleanup;

4652 4653
no_memory:
    virReportOOMError(conn);
4654
    goto cleanup;
M
Mark McLoughlin 已提交
4655 4656
}

4657
static int qemudDomainAttachHostPciDevice(virConnectPtr conn,
4658
                                          struct qemud_driver *driver,
4659 4660 4661 4662
                                          virDomainObjPtr vm,
                                          virDomainDeviceDefPtr dev)
{
    virDomainHostdevDefPtr hostdev = dev->data.hostdev;
4663
    pciDevice *pci;
4664 4665 4666 4667 4668 4669

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

4670 4671 4672 4673 4674 4675 4676
    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;
4677

4678
    if ((hostdev->managed && pciDettachDevice(conn, pci) < 0) ||
4679 4680 4681 4682 4683 4684
        pciResetDevice(conn, pci, driver->activePciHostdevs) < 0) {
        pciFreeDevice(conn, pci);
        return -1;
    }

    if (pciDeviceListAdd(conn, driver->activePciHostdevs, pci) < 0) {
4685
        pciFreeDevice(conn, pci);
4686
        return -1;
4687 4688
    }

4689 4690 4691 4692 4693 4694 4695 4696
    if (qemuMonitorAddPCIHostDevice(vm,
                                    hostdev->source.subsys.u.pci.domain,
                                    hostdev->source.subsys.u.pci.bus,
                                    hostdev->source.subsys.u.pci.slot,
                                    hostdev->source.subsys.u.pci.function,
                                    &hostdev->source.subsys.u.pci.guest_addr.domain,
                                    &hostdev->source.subsys.u.pci.guest_addr.bus,
                                    &hostdev->source.subsys.u.pci.guest_addr.slot) < 0)
4697
        goto error;
4698 4699 4700 4701

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

    return 0;
4702 4703 4704 4705 4706

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

    return -1;
4707 4708
}

M
Mark McLoughlin 已提交
4709 4710 4711
static int qemudDomainAttachHostUsbDevice(virConnectPtr conn,
                                          virDomainObjPtr vm,
                                          virDomainDeviceDefPtr dev)
4712 4713 4714
{
    int ret;

4715
    if (VIR_REALLOC_N(vm->def->hostdevs, vm->def->nhostdevs+1) < 0) {
4716
        virReportOOMError(conn);
4717 4718
        return -1;
    }
4719

4720
    if (dev->data.hostdev->source.subsys.u.usb.vendor) {
4721 4722 4723
        ret = qemuMonitorAddUSBDeviceMatch(vm,
                                           dev->data.hostdev->source.subsys.u.usb.vendor,
                                           dev->data.hostdev->source.subsys.u.usb.product);
4724
    } else {
4725 4726 4727
        ret = qemuMonitorAddUSBDeviceExact(vm,
                                           dev->data.hostdev->source.subsys.u.usb.bus,
                                           dev->data.hostdev->source.subsys.u.usb.device);
4728 4729
    }

4730 4731
    if (ret != -1)
        vm->def->hostdevs[vm->def->nhostdevs++] = dev->data.hostdev;
4732

4733
    return ret;
4734 4735
}

M
Mark McLoughlin 已提交
4736 4737 4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749 4750 4751
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;
4752 4753 4754
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityHostdevLabel(conn, vm, dev->data.hostdev) < 0)
        return -1;
M
Mark McLoughlin 已提交
4755 4756

    switch (hostdev->source.subsys.type) {
4757
    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI:
4758
        return qemudDomainAttachHostPciDevice(conn, driver, vm, dev);
M
Mark McLoughlin 已提交
4759 4760 4761 4762 4763 4764 4765 4766 4767 4768
    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;
    }
}

4769 4770
static int qemudDomainAttachDevice(virDomainPtr dom,
                                   const char *xml) {
4771 4772 4773
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    virDomainDeviceDefPtr dev = NULL;
4774
    unsigned int qemuCmdFlags;
4775
    virCgroupPtr cgroup = NULL;
4776
    int ret = -1;
4777

4778
    qemuDriverLock(driver);
4779
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4780
    if (!vm) {
4781 4782 4783 4784
        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);
4785
        goto cleanup;
4786 4787 4788
    }

    if (!virDomainIsActive(vm)) {
4789
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
4790
                         "%s", _("cannot attach device on inactive domain"));
4791
        goto cleanup;
4792 4793
    }

4794 4795
    dev = virDomainDeviceDefParse(dom->conn, driver->caps, vm->def, xml,
                                  VIR_DOMAIN_XML_INACTIVE);
4796 4797 4798
    if (dev == NULL)
        goto cleanup;

4799 4800 4801 4802
    if (qemudExtractVersionInfo(vm->def->emulator,
                                NULL,
                                &qemuCmdFlags) < 0)
        goto cleanup;
4803

4804
    if (dev->type == VIR_DOMAIN_DEVICE_DISK) {
4805
        if (qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_DEVICES)) {
4806 4807 4808 4809 4810 4811 4812 4813 4814 4815 4816 4817 4818 4819 4820 4821 4822
            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;
            }
        }

4823
        switch (dev->data.disk->device) {
4824 4825
        case VIR_DOMAIN_DISK_DEVICE_CDROM:
        case VIR_DOMAIN_DISK_DEVICE_FLOPPY:
4826 4827
            if (driver->securityDriver)
                driver->securityDriver->domainSetSecurityImageLabel(dom->conn, vm, dev->data.disk);
4828 4829 4830 4831

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

4832
            ret = qemudDomainChangeEjectableMedia(dom->conn, vm, dev, qemuCmdFlags);
4833
            break;
4834

4835
        case VIR_DOMAIN_DISK_DEVICE_DISK:
4836 4837
            if (driver->securityDriver)
                driver->securityDriver->domainSetSecurityImageLabel(dom->conn, vm, dev->data.disk);
4838 4839 4840 4841

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

4842
            if (dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_USB) {
4843
                ret = qemudDomainAttachUsbMassstorageDevice(dom->conn, vm, dev);
4844 4845
            } else if (dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_SCSI ||
                       dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_VIRTIO) {
4846
                ret = qemudDomainAttachPciDiskDevice(dom->conn, vm, dev);
4847 4848 4849 4850
            } else {
                qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                                 _("disk bus '%s' cannot be hotplugged."),
                                 virDomainDiskBusTypeToString(dev->data.disk->bus));
4851
                /* fallthrough */
4852 4853
            }
            break;
4854

4855 4856
        default:
            qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
4857 4858
                             _("disk device type '%s' cannot be hotplugged"),
                             virDomainDiskDeviceTypeToString(dev->data.disk->device));
4859 4860 4861 4862 4863
            /* Fallthrough */
        }
        if (ret != 0) {
            virCgroupDenyDevicePath(cgroup,
                                    dev->data.disk->src);
4864
        }
M
Mark McLoughlin 已提交
4865
    } else if (dev->type == VIR_DOMAIN_DEVICE_NET) {
4866
        ret = qemudDomainAttachNetDevice(dom->conn, driver, vm, dev, qemuCmdFlags);
M
Mark McLoughlin 已提交
4867 4868
    } else if (dev->type == VIR_DOMAIN_DEVICE_HOSTDEV) {
        ret = qemudDomainAttachHostDevice(dom->conn, driver, vm, dev);
4869
    } else {
4870
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
4871 4872
                         _("device type '%s' cannot be attached"),
                         virDomainDeviceTypeToString(dev->type));
4873
        goto cleanup;
4874 4875
    }

4876
    if (!ret && virDomainSaveStatus(dom->conn, driver->stateDir, vm) < 0)
4877 4878
        ret = -1;

4879
cleanup:
4880 4881 4882
    if (cgroup)
        virCgroupFree(&cgroup);

4883
    if (ret < 0 && dev != NULL) {
4884 4885
        if (qemuDomainSetDeviceOwnership(dom->conn, driver, dev, 1) < 0)
            VIR_WARN0("Fail to restore disk device ownership");
G
Guido Günther 已提交
4886
        virDomainDeviceDefFree(dev);
4887
    }
4888 4889
    if (vm)
        virDomainObjUnlock(vm);
4890
    qemuDriverUnlock(driver);
4891 4892 4893
    return ret;
}

4894 4895
static int qemudDomainDetachPciDiskDevice(virConnectPtr conn,
                                          virDomainObjPtr vm, virDomainDeviceDefPtr dev)
4896 4897 4898 4899 4900 4901 4902 4903 4904 4905 4906 4907
{
    int i, ret = -1;
    virDomainDiskDefPtr detach = NULL;

    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) {
4908
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
4909
                         _("disk %s not found"), dev->data.disk->dst);
4910
        goto cleanup;
4911 4912
    }

4913
    if (!virDiskHasValidPciAddr(detach)) {
4914
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
4915 4916
                         _("disk %s cannot be detached - no PCI address for device"),
                           detach->dst);
4917
        goto cleanup;
4918 4919
    }

4920 4921 4922 4923
    if (qemuMonitorRemovePCIDevice(vm,
                                   detach->pci_addr.domain,
                                   detach->pci_addr.bus,
                                   detach->pci_addr.slot) < 0)
4924
        goto cleanup;
4925

4926 4927 4928 4929 4930 4931 4932 4933 4934 4935
    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 {
4936
        VIR_FREE(vm->def->disks);
4937
        vm->def->ndisks = 0;
4938
    }
4939
    virDomainDiskDefFree(detach);
4940

4941
    ret = 0;
4942 4943

cleanup:
4944 4945 4946
    return ret;
}

4947 4948 4949 4950 4951 4952 4953 4954 4955 4956 4957 4958 4959 4960 4961 4962 4963 4964 4965 4966 4967 4968 4969 4970 4971 4972 4973 4974 4975 4976 4977 4978
static int
qemudDomainDetachNetDevice(virConnectPtr conn,
                           virDomainObjPtr vm,
                           virDomainDeviceDefPtr dev)
{
    int i, ret = -1;
    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;
    }

4979 4980 4981 4982
    if (qemuMonitorRemovePCIDevice(vm,
                                   detach->pci_addr.domain,
                                   detach->pci_addr.bus,
                                   detach->pci_addr.slot) < 0)
4983 4984
        goto cleanup;

4985
    if (qemuMonitorRemoveHostNetwork(vm, detach->vlan, detach->hostnet_name) < 0)
4986 4987
        goto cleanup;

4988 4989 4990 4991 4992 4993 4994 4995 4996 4997
    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 {
4998
        VIR_FREE(vm->def->nets);
4999
        vm->def->nnets = 0;
5000
    }
5001
    virDomainNetDefFree(detach);
5002

5003 5004 5005 5006 5007 5008
    ret = 0;

cleanup:
    return ret;
}

5009
static int qemudDomainDetachHostPciDevice(virConnectPtr conn,
5010
                                          struct qemud_driver *driver,
5011 5012 5013
                                          virDomainObjPtr vm,
                                          virDomainDeviceDefPtr dev)
{
5014
    virDomainHostdevDefPtr detach = NULL;
5015
    int i, ret;
5016
    pciDevice *pci;
5017 5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047 5048

    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;
    }

5049 5050 5051 5052
    if (qemuMonitorRemovePCIDevice(vm,
                                   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)
5053 5054 5055 5056
        return -1;

    ret = 0;

5057 5058 5059 5060 5061 5062 5063 5064
    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 {
5065 5066
        pciDeviceListDel(conn, driver->activePciHostdevs, pci);
        if (pciResetDevice(conn, pci, driver->activePciHostdevs) < 0)
5067 5068
            ret = -1;
        if (detach->managed && pciReAttachDevice(conn, pci) < 0)
5069
            ret = -1;
5070
        pciFreeDevice(conn, pci);
5071 5072
    }

5073 5074 5075 5076 5077 5078 5079 5080 5081 5082 5083 5084
    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;
5085
    }
5086
    virDomainHostdevDefFree(detach);
5087 5088 5089 5090 5091 5092 5093 5094 5095 5096 5097 5098 5099 5100 5101 5102 5103 5104 5105 5106 5107

    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:
5108
        ret = qemudDomainDetachHostPciDevice(conn, driver, vm, dev);
5109
        break;
5110 5111 5112 5113 5114 5115 5116
    default:
        qemudReportError(conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                         _("hostdev subsys type '%s' not supported"),
                         virDomainHostdevSubsysTypeToString(hostdev->source.subsys.type));
        return -1;
    }

5117 5118 5119 5120
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityHostdevLabel(conn, vm, dev->data.hostdev) < 0)
        VIR_WARN0("Failed to restore device labelling");

5121
    if (qemuDomainSetDeviceOwnership(conn, driver, dev, 1) < 0)
5122
        VIR_WARN0("Failed to restore device ownership");
5123 5124 5125 5126

    return ret;
}

5127 5128
static int qemudDomainDetachDevice(virDomainPtr dom,
                                   const char *xml) {
5129 5130 5131 5132
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    virDomainDeviceDefPtr dev = NULL;
    int ret = -1;
5133

5134
    qemuDriverLock(driver);
5135
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5136
    if (!vm) {
5137 5138 5139 5140
        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);
5141
        goto cleanup;
5142 5143 5144
    }

    if (!virDomainIsActive(vm)) {
5145
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
G
Guido Günther 已提交
5146
                         "%s", _("cannot detach device on inactive domain"));
5147
        goto cleanup;
5148 5149
    }

5150 5151
    dev = virDomainDeviceDefParse(dom->conn, driver->caps, vm->def, xml,
                                  VIR_DOMAIN_XML_INACTIVE);
5152 5153 5154
    if (dev == NULL)
        goto cleanup;

5155 5156 5157 5158

    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 ||
5159
         dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_VIRTIO)) {
5160
        ret = qemudDomainDetachPciDiskDevice(dom->conn, vm, dev);
5161
        if (driver->securityDriver)
5162
            driver->securityDriver->domainRestoreSecurityImageLabel(dom->conn, dev->data.disk);
5163 5164
        if (qemuDomainSetDeviceOwnership(dom->conn, driver, dev, 1) < 0)
            VIR_WARN0("Fail to restore disk device ownership");
5165 5166
    } else if (dev->type == VIR_DOMAIN_DEVICE_NET) {
        ret = qemudDomainDetachNetDevice(dom->conn, vm, dev);
5167 5168
    } else if (dev->type == VIR_DOMAIN_DEVICE_HOSTDEV) {
        ret = qemudDomainDetachHostDevice(dom->conn, driver, vm, dev);
5169
    } else
5170 5171 5172
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                         "%s", _("only SCSI or virtio disk device can be detached dynamically"));

5173
    if (!ret && virDomainSaveStatus(dom->conn, driver->stateDir, vm) < 0)
5174 5175
        ret = -1;

5176 5177
cleanup:
    virDomainDeviceDefFree(dev);
5178 5179
    if (vm)
        virDomainObjUnlock(vm);
5180
    qemuDriverUnlock(driver);
5181 5182 5183
    return ret;
}

5184
static int qemudDomainGetAutostart(virDomainPtr dom,
5185
                                   int *autostart) {
5186 5187 5188
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
5189

5190
    qemuDriverLock(driver);
5191
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5192 5193
    qemuDriverUnlock(driver);

5194
    if (!vm) {
5195 5196 5197 5198
        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);
5199
        goto cleanup;
5200 5201 5202
    }

    *autostart = vm->autostart;
5203
    ret = 0;
5204

5205
cleanup:
5206 5207
    if (vm)
        virDomainObjUnlock(vm);
5208
    return ret;
5209 5210
}

5211
static int qemudDomainSetAutostart(virDomainPtr dom,
5212
                                   int autostart) {
5213 5214
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
5215 5216
    char *configFile = NULL, *autostartLink = NULL;
    int ret = -1;
5217

5218
    qemuDriverLock(driver);
5219
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5220

5221
    if (!vm) {
5222 5223 5224 5225
        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);
5226
        goto cleanup;
5227 5228
    }

5229 5230 5231
    if (!vm->persistent) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
                         "%s", _("cannot set autostart for transient domain"));
5232
        goto cleanup;
5233 5234
    }

5235 5236
    autostart = (autostart != 0);

5237 5238 5239 5240 5241
    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;
5242

5243 5244
        if (autostart) {
            int err;
5245

5246
            if ((err = virFileMakePath(driver->autostartDir))) {
5247 5248 5249
                virReportSystemError(dom->conn, err,
                                     _("cannot create autostart directory %s"),
                                     driver->autostartDir);
5250 5251
                goto cleanup;
            }
5252

5253
            if (symlink(configFile, autostartLink) < 0) {
5254 5255 5256
                virReportSystemError(dom->conn, errno,
                                     _("Failed to create symlink '%s to '%s'"),
                                     autostartLink, configFile);
5257 5258 5259 5260
                goto cleanup;
            }
        } else {
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
5261 5262 5263
                virReportSystemError(dom->conn, errno,
                                     _("Failed to delete symlink '%s'"),
                                     autostartLink);
5264 5265
                goto cleanup;
            }
5266 5267
        }

5268
        vm->autostart = autostart;
5269
    }
5270
    ret = 0;
5271

5272 5273 5274
cleanup:
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
5275 5276
    if (vm)
        virDomainObjUnlock(vm);
5277
    qemuDriverUnlock(driver);
5278
    return ret;
5279 5280
}

5281 5282 5283 5284 5285

static char *qemuGetSchedulerType(virDomainPtr dom,
                                  int *nparams)
{
    struct qemud_driver *driver = dom->conn->privateData;
5286
    char *ret = NULL;
5287

5288
    qemuDriverLock(driver);
5289
    if (!qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_CPU)) {
5290 5291
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                         __FUNCTION__);
5292
        goto cleanup;
5293 5294 5295 5296 5297 5298 5299 5300
    }

    if (nparams)
        *nparams = 1;

    ret = strdup("posix");
    if (!ret)
        virReportOOMError(dom->conn);
5301 5302 5303

cleanup:
    qemuDriverUnlock(driver);
5304 5305 5306 5307 5308 5309 5310 5311 5312 5313 5314 5315 5316
    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;

5317
    qemuDriverLock(driver);
5318
    if (!qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_CPU)) {
5319 5320
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                         __FUNCTION__);
5321
        goto cleanup;
5322 5323 5324 5325 5326 5327 5328 5329 5330 5331 5332 5333 5334 5335 5336 5337 5338 5339 5340 5341 5342 5343 5344 5345 5346 5347 5348 5349 5350 5351 5352 5353 5354 5355 5356 5357 5358 5359 5360 5361 5362 5363 5364 5365 5366
    }

    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);
5367
    qemuDriverUnlock(driver);
5368 5369 5370 5371 5372 5373 5374 5375 5376 5377 5378 5379 5380 5381
    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;

5382
    qemuDriverLock(driver);
5383
    if (!qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_CPU)) {
5384 5385
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                         __FUNCTION__);
5386
        goto cleanup;
5387 5388 5389 5390 5391
    }

    if ((*nparams) != 1) {
        qemudReportError(dom->conn, domain, NULL, VIR_ERR_INVALID_ARG,
                         "%s", _("Invalid parameter count"));
5392
        goto cleanup;
5393 5394 5395 5396 5397 5398 5399 5400 5401 5402 5403 5404 5405 5406 5407 5408 5409 5410 5411 5412 5413 5414 5415 5416
    }

    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
5417 5418 5419 5420 5421
    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;
    }
5422 5423 5424 5425 5426 5427 5428

    ret = 0;

cleanup:
    virCgroupFree(&group);
    if (vm)
        virDomainObjUnlock(vm);
5429
    qemuDriverUnlock(driver);
5430 5431 5432 5433
    return ret;
}


5434 5435 5436 5437 5438 5439 5440 5441 5442
/* 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)
{
5443
    struct qemud_driver *driver = dom->conn->privateData;
5444 5445
    const char *qemu_dev_name = NULL;
    int i, ret = -1;
5446
    virDomainObjPtr vm;
5447
    virDomainDiskDefPtr disk = NULL;
5448

5449
    qemuDriverLock(driver);
5450
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5451
    qemuDriverUnlock(driver);
5452
    if (!vm) {
5453 5454 5455 5456
        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);
5457
        goto cleanup;
5458
    }
5459
    if (!virDomainIsActive (vm)) {
5460
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
5461
                          "%s", _("domain is not running"));
5462
        goto cleanup;
5463 5464
    }

5465 5466 5467 5468 5469 5470 5471 5472
    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (STREQ(path, vm->def->disks[i]->dst)) {
            disk = vm->def->disks[i];
            break;
        }
    }

    if (!disk) {
5473 5474
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
                          _("invalid path: %s"), path);
5475
        goto cleanup;
5476 5477
    }

5478
    qemu_dev_name = qemudDiskDeviceName(dom->conn, disk);
5479
    if (!qemu_dev_name)
5480
        goto cleanup;
5481

5482 5483 5484 5485 5486 5487
    if (qemuMonitorGetBlockStatsInfo(vm, qemu_dev_name,
                                     &stats->rd_req,
                                     &stats->rd_bytes,
                                     &stats->wr_req,
                                     &stats->wr_bytes,
                                     &stats->errs) < 0)
5488
        goto cleanup;
5489

5490
    ret = 0;
5491

5492
cleanup:
5493
    VIR_FREE(qemu_dev_name);
5494 5495
    if (vm)
        virDomainObjUnlock(vm);
5496
    return ret;
5497 5498
}

5499
#ifdef __linux__
5500 5501 5502 5503 5504
static int
qemudDomainInterfaceStats (virDomainPtr dom,
                           const char *path,
                           struct _virDomainInterfaceStats *stats)
{
5505 5506
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
5507
    int i;
5508
    int ret = -1;
5509

5510
    qemuDriverLock(driver);
5511
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5512 5513
    qemuDriverUnlock(driver);

5514
    if (!vm) {
5515 5516 5517 5518
        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);
5519
        goto cleanup;
5520 5521
    }

5522
    if (!virDomainIsActive(vm)) {
5523
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
5524
                         "%s", _("domain is not running"));
5525
        goto cleanup;
5526 5527 5528
    }

    /* Check the path is one of the domain's network interfaces. */
5529 5530
    for (i = 0 ; i < vm->def->nnets ; i++) {
        if (vm->def->nets[i]->ifname &&
5531 5532 5533 5534
            STREQ (vm->def->nets[i]->ifname, path)) {
            ret = 0;
            break;
        }
5535 5536
    }

5537 5538 5539 5540 5541
    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);
5542

5543
cleanup:
5544 5545
    if (vm)
        virDomainObjUnlock(vm);
5546 5547
    return ret;
}
5548
#else
5549 5550 5551 5552
static int
qemudDomainInterfaceStats (virDomainPtr dom,
                           const char *path ATTRIBUTE_UNUSED,
                           struct _virDomainInterfaceStats *stats ATTRIBUTE_UNUSED)
5553 5554 5555 5556
    qemudReportError (dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                      "%s", __FUNCTION__);
    return -1;
}
5557
#endif
5558

5559 5560 5561 5562 5563 5564 5565
static int
qemudDomainBlockPeek (virDomainPtr dom,
                      const char *path,
                      unsigned long long offset, size_t size,
                      void *buffer,
                      unsigned int flags ATTRIBUTE_UNUSED)
{
5566 5567 5568
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int fd = -1, ret = -1, i;
5569

5570
    qemuDriverLock(driver);
5571
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5572 5573
    qemuDriverUnlock(driver);

5574
    if (!vm) {
5575 5576 5577 5578
        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);
5579
        goto cleanup;
5580 5581 5582 5583
    }

    if (!path || path[0] == '\0') {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
J
Jim Meyering 已提交
5584
                         "%s", _("NULL or empty path"));
5585
        goto cleanup;
5586 5587 5588
    }

    /* Check the path belongs to this domain. */
5589 5590
    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (vm->def->disks[i]->src != NULL &&
5591 5592 5593 5594
            STREQ (vm->def->disks[i]->src, path)) {
            ret = 0;
            break;
        }
5595 5596
    }

5597 5598 5599 5600 5601
    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) {
5602 5603
            virReportSystemError (dom->conn, errno,
                                  _("%s: failed to open"), path);
5604 5605
            goto cleanup;
        }
5606

5607 5608 5609 5610 5611 5612
        /* 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) {
5613 5614
            virReportSystemError (dom->conn, errno,
                                  _("%s: failed to seek or read"), path);
5615 5616 5617 5618 5619 5620 5621
            goto cleanup;
        }

        ret = 0;
    } else {
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
                          "%s", _("invalid path"));
5622 5623
    }

5624 5625 5626
cleanup:
    if (fd >= 0)
        close (fd);
5627 5628
    if (vm)
        virDomainObjUnlock(vm);
5629 5630 5631
    return ret;
}

R
Richard W.M. Jones 已提交
5632 5633 5634 5635 5636 5637
static int
qemudDomainMemoryPeek (virDomainPtr dom,
                       unsigned long long offset, size_t size,
                       void *buffer,
                       unsigned int flags)
{
5638 5639
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
5640
    char *tmp = NULL;
R
Richard W.M. Jones 已提交
5641 5642
    int fd = -1, ret = -1;

5643
    qemuDriverLock(driver);
5644
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5645
    qemuDriverUnlock(driver);
R
Richard W.M. Jones 已提交
5646 5647

    if (!vm) {
5648 5649 5650 5651
        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);
5652 5653 5654
        goto cleanup;
    }

5655
    if (flags != VIR_MEMORY_VIRTUAL && flags != VIR_MEMORY_PHYSICAL) {
5656
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
5657
                     "%s", _("flags parameter must be VIR_MEMORY_VIRTUAL or VIR_MEMORY_PHYSICAL"));
5658
        goto cleanup;
R
Richard W.M. Jones 已提交
5659 5660
    }

5661
    if (!virDomainIsActive(vm)) {
5662
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
R
Richard W.M. Jones 已提交
5663
                         "%s", _("domain is not running"));
5664
        goto cleanup;
R
Richard W.M. Jones 已提交
5665 5666
    }

5667 5668 5669 5670 5671
    if (virAsprintf(&tmp, driver->cacheDir,  "/qemu.mem.XXXXXX") < 0) {
        virReportOOMError(dom->conn);
        goto cleanup;
    }

R
Richard W.M. Jones 已提交
5672 5673
    /* Create a temporary filename. */
    if ((fd = mkstemp (tmp)) == -1) {
5674 5675
        virReportSystemError (dom->conn, errno,
                              _("mkstemp(\"%s\") failed"), tmp);
5676
        goto cleanup;
R
Richard W.M. Jones 已提交
5677 5678
    }

5679 5680 5681 5682 5683 5684
    if (flags == VIR_MEMORY_VIRTUAL) {
        if (qemuMonitorSaveVirtualMemory(vm, offset, size, tmp) < 0)
            goto cleanup;
    } else {
        if (qemuMonitorSavePhysicalMemory(vm, offset, size, tmp) < 0)
            goto cleanup;
R
Richard W.M. Jones 已提交
5685 5686 5687 5688
    }

    /* Read the memory file into buffer. */
    if (saferead (fd, buffer, size) == (ssize_t) -1) {
5689 5690 5691
        virReportSystemError (dom->conn, errno,
                              _("failed to read temporary file "
                                "created with template %s"), tmp);
5692
        goto cleanup;
R
Richard W.M. Jones 已提交
5693 5694 5695
    }

    ret = 0;
5696 5697

cleanup:
5698
    VIR_FREE(tmp);
R
Richard W.M. Jones 已提交
5699 5700
    if (fd >= 0) close (fd);
    unlink (tmp);
5701 5702
    if (vm)
        virDomainObjUnlock(vm);
R
Richard W.M. Jones 已提交
5703 5704 5705
    return ret;
}

5706

5707 5708
static int
qemudDomainEventRegister (virConnectPtr conn,
5709
                          virConnectDomainEventCallback callback,
5710 5711
                          void *opaque,
                          virFreeCallback freecb)
5712
{
5713 5714 5715
    struct qemud_driver *driver = conn->privateData;
    int ret;

5716
    qemuDriverLock(driver);
5717 5718
    ret = virDomainEventCallbackListAdd(conn, driver->domainEventCallbacks,
                                        callback, opaque, freecb);
5719
    qemuDriverUnlock(driver);
5720

5721
    return ret;
5722 5723 5724 5725
}

static int
qemudDomainEventDeregister (virConnectPtr conn,
5726
                            virConnectDomainEventCallback callback)
5727
{
5728 5729 5730
    struct qemud_driver *driver = conn->privateData;
    int ret;

5731
    qemuDriverLock(driver);
5732 5733 5734 5735 5736 5737
    if (driver->domainEventDispatching)
        ret = virDomainEventCallbackListMarkDelete(conn, driver->domainEventCallbacks,
                                                   callback);
    else
        ret = virDomainEventCallbackListRemove(conn, driver->domainEventCallbacks,
                                               callback);
5738
    qemuDriverUnlock(driver);
5739

5740
    return ret;
5741 5742
}

5743 5744 5745 5746 5747
static void qemuDomainEventDispatchFunc(virConnectPtr conn,
                                        virDomainEventPtr event,
                                        virConnectDomainEventCallback cb,
                                        void *cbopaque,
                                        void *opaque)
5748
{
5749
    struct qemud_driver *driver = opaque;
5750

5751 5752 5753 5754 5755 5756 5757 5758 5759 5760 5761 5762 5763 5764 5765 5766 5767 5768 5769 5770 5771 5772 5773 5774 5775 5776 5777 5778 5779 5780 5781 5782 5783 5784 5785 5786 5787 5788 5789 5790 5791 5792 5793 5794
    /* 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);
5795 5796
}

D
Daniel Veillard 已提交
5797 5798 5799 5800 5801 5802 5803 5804 5805 5806 5807 5808 5809 5810 5811 5812 5813 5814
/* 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;
5815 5816
    struct qemud_driver *driver = dconn->privateData;
    virDomainDefPtr def = NULL;
D
Daniel Veillard 已提交
5817 5818
    virDomainObjPtr vm = NULL;
    int this_port;
5819
    char *hostname;
D
Daniel Veillard 已提交
5820 5821
    char migrateFrom [64];
    const char *p;
5822
    virDomainEventPtr event = NULL;
5823
    int ret = -1;
5824
    int internalret;
5825 5826

    *uri_out = NULL;
D
Daniel Veillard 已提交
5827

5828
    qemuDriverLock(driver);
D
Daniel Veillard 已提交
5829 5830 5831
    if (!dom_xml) {
        qemudReportError (dconn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                          "%s", _("no domain XML passed"));
5832
        goto cleanup;
D
Daniel Veillard 已提交
5833 5834 5835 5836 5837 5838 5839 5840 5841 5842 5843 5844 5845 5846 5847 5848 5849
    }

    /* 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 */
5850
        if ((hostname = virGetHostname()) == NULL) {
5851 5852
            virReportSystemError (dconn, errno,
                                  "%s", _("failed to determine host name"));
5853
            goto cleanup;
D
Daniel Veillard 已提交
5854 5855
        }

5856 5857 5858 5859 5860
        /* XXX this really should have been a properly well-formed
         * URI, but we can't add in tcp:// now without breaking
         * compatability with old targets. We at least make the
         * new targets accept both syntaxes though.
         */
D
Daniel Veillard 已提交
5861
        /* Caller frees */
5862 5863 5864
        internalret = virAsprintf(uri_out, "tcp:%s:%d", hostname, this_port);
        VIR_FREE(hostname);
        if (internalret < 0) {
5865
            virReportOOMError (dconn);
5866
            goto cleanup;
D
Daniel Veillard 已提交
5867 5868 5869 5870 5871 5872
        }
    } 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.
         */
5873
        if (!STRPREFIX (uri_in, "tcp:")) {
D
Daniel Veillard 已提交
5874
            qemudReportError (dconn, NULL, NULL, VIR_ERR_INVALID_ARG,
5875
                  "%s", _("only tcp URIs are supported for KVM/QEMU migrations"));
5876
            goto cleanup;
D
Daniel Veillard 已提交
5877 5878 5879 5880 5881 5882 5883 5884 5885
        }

        /* 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"));
5886
            goto cleanup;
D
Daniel Veillard 已提交
5887 5888 5889 5890
        }
    }

    /* Parse the domain XML. */
5891 5892
    if (!(def = virDomainDefParseString(dconn, driver->caps, dom_xml,
                                        VIR_DOMAIN_XML_INACTIVE))) {
D
Daniel Veillard 已提交
5893 5894
        qemudReportError (dconn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                          "%s", _("failed to parse XML"));
5895
        goto cleanup;
D
Daniel Veillard 已提交
5896 5897 5898 5899 5900 5901 5902 5903 5904 5905 5906 5907 5908 5909 5910 5911
    }

    /* 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"));
5912
        goto cleanup;
D
Daniel Veillard 已提交
5913 5914 5915 5916 5917 5918 5919 5920 5921
    }
#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);
5922
            goto cleanup;
D
Daniel Veillard 已提交
5923
        }
5924
        virDomainObjUnlock(vm);
D
Daniel Veillard 已提交
5925 5926 5927 5928 5929 5930 5931
    }

    if (!(vm = virDomainAssignDef(dconn,
                                  &driver->domains,
                                  def))) {
        qemudReportError (dconn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                          "%s", _("failed to assign new VM"));
5932
        goto cleanup;
D
Daniel Veillard 已提交
5933
    }
5934
    def = NULL;
D
Daniel Veillard 已提交
5935 5936 5937 5938 5939 5940 5941 5942

    /* 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);
5943
    if (qemudStartVMDaemon (dconn, driver, vm, migrateFrom, -1) < 0) {
5944 5945 5946
        /* Note that we don't set an error here because qemudStartVMDaemon
         * should have already done that.
         */
5947
        if (!vm->persistent) {
D
Daniel Veillard 已提交
5948
            virDomainRemoveInactive(&driver->domains, vm);
5949 5950
            vm = NULL;
        }
5951
        goto cleanup;
D
Daniel Veillard 已提交
5952
    }
5953 5954 5955 5956

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_MIGRATED);
5957
    ret = 0;
D
Daniel Veillard 已提交
5958

5959 5960 5961 5962 5963
cleanup:
    virDomainDefFree(def);
    if (ret != 0) {
        VIR_FREE(*uri_out);
    }
5964 5965
    if (vm)
        virDomainObjUnlock(vm);
5966 5967
    if (event)
        qemuDomainEventQueue(driver, event);
5968
    qemuDriverUnlock(driver);
5969
    return ret;
D
Daniel Veillard 已提交
5970 5971 5972 5973 5974 5975 5976 5977
}

/* 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,
5978
                           unsigned long flags,
D
Daniel Veillard 已提交
5979 5980 5981
                           const char *dname ATTRIBUTE_UNUSED,
                           unsigned long resource)
{
5982 5983
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
5984
    virDomainEventPtr event = NULL;
5985
    int ret = -1;
5986
    int paused = 0;
5987
    int status;
5988
    xmlURIPtr uribits = NULL;
5989
    unsigned long long transferred, remaining, total;
D
Daniel Veillard 已提交
5990

5991
    qemuDriverLock(driver);
5992
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
D
Daniel Veillard 已提交
5993
    if (!vm) {
5994 5995 5996 5997
        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);
5998
        goto cleanup;
D
Daniel Veillard 已提交
5999 6000 6001
    }

    if (!virDomainIsActive(vm)) {
6002
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
D
Daniel Veillard 已提交
6003
                          "%s", _("domain is not running"));
6004
        goto cleanup;
D
Daniel Veillard 已提交
6005 6006
    }

6007 6008
    if (!(flags & VIR_MIGRATE_LIVE)) {
        /* Pause domain for non-live migration */
6009
        if (qemuMonitorStopCPUs(vm) < 0)
6010 6011
            goto cleanup;
        paused = 1;
6012

6013 6014 6015 6016 6017 6018
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_SUSPENDED,
                                         VIR_DOMAIN_EVENT_SUSPENDED_MIGRATED);
        if (event)
            qemuDomainEventQueue(driver, event);
        event = NULL;
6019 6020
    }

6021 6022 6023
    if (resource > 0 &&
        qemuMonitorSetMigrationSpeed(vm, resource) < 0)
        goto cleanup;
D
Daniel Veillard 已提交
6024 6025

    /* Issue the migrate command. */
6026
    if (STRPREFIX(uri, "tcp:") && !STRPREFIX(uri, "tcp://")) {
6027
        /* HACK: source host generates bogus URIs, so fix them up */
6028 6029 6030 6031 6032 6033 6034 6035 6036
        char *tmpuri;
        if (virAsprintf(&tmpuri, "tcp://%s", uri + strlen("tcp:")) < 0) {
            virReportOOMError(dom->conn);
            goto cleanup;
        }
        uribits = xmlParseURI(tmpuri);
        VIR_FREE(tmpuri);
    } else {
        uribits = xmlParseURI(uri);
D
Daniel Veillard 已提交
6037
    }
6038 6039 6040
    if (!uribits) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
                         _("cannot parse URI %s"), uri);
6041
        goto cleanup;
D
Daniel Veillard 已提交
6042 6043
    }

6044
    if (qemuMonitorMigrateToHost(vm, uribits->server, uribits->port) < 0)
6045
        goto cleanup;
D
Daniel Veillard 已提交
6046

6047 6048 6049
    /* it is also possible that the migrate didn't fail initially, but
     * rather failed later on.  Check the output of "info migrate"
     */
6050 6051 6052 6053
    if (qemuMonitorGetMigrationStatus(vm, &status,
                                      &transferred,
                                      &remaining,
                                      &total) < 0) {
6054 6055
        goto cleanup;
    }
6056 6057

    if (status != QEMU_MONITOR_MIGRATION_STATUS_COMPLETED) {
6058
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
6059
                          "%s", _("migrate did not successfully complete"));
6060 6061 6062
        goto cleanup;
    }

D
Daniel Veillard 已提交
6063 6064
    /* Clean up the source domain. */
    qemudShutdownVMDaemon (dom->conn, driver, vm);
6065
    paused = 0;
6066 6067 6068 6069

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_MIGRATED);
6070
    if (!vm->persistent) {
D
Daniel Veillard 已提交
6071
        virDomainRemoveInactive(&driver->domains, vm);
6072 6073
        vm = NULL;
    }
6074
    ret = 0;
D
Daniel Veillard 已提交
6075

6076
cleanup:
6077 6078
    if (paused) {
        /* we got here through some sort of failure; start the domain again */
6079
        if (qemuMonitorStartCPUs(dom->conn, vm) < 0) {
6080 6081 6082 6083
            /* 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
             */
6084 6085
            VIR_ERROR(_("Failed to resume guest %s after failure\n"),
                      vm->def->name);
6086 6087 6088 6089 6090 6091 6092
        }

        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_MIGRATED);
    }

6093 6094
    if (uribits)
        xmlFreeURI(uribits);
6095 6096
    if (vm)
        virDomainObjUnlock(vm);
6097 6098
    if (event)
        qemuDomainEventQueue(driver, event);
6099
    qemuDriverUnlock(driver);
6100
    return ret;
D
Daniel Veillard 已提交
6101 6102 6103 6104 6105 6106 6107 6108 6109 6110 6111 6112
}

/* 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)
{
6113 6114 6115
    struct qemud_driver *driver = dconn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
6116
    virDomainEventPtr event = NULL;
D
Daniel Veillard 已提交
6117

6118
    qemuDriverLock(driver);
6119
    vm = virDomainFindByName(&driver->domains, dname);
D
Daniel Veillard 已提交
6120
    if (!vm) {
6121 6122
        qemudReportError (dconn, NULL, NULL, VIR_ERR_NO_DOMAIN,
                          _("no domain with matching name '%s'"), dname);
6123
        goto cleanup;
D
Daniel Veillard 已提交
6124 6125 6126 6127 6128 6129 6130
    }

    /* 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);
6131 6132 6133 6134 6135

        /* 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
         */
6136
        if (qemuMonitorStartCPUs(dconn, vm) < 0) {
6137 6138 6139
            if (virGetLastError() == NULL)
                qemudReportError(dconn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                                 "%s", _("resume operation failed"));
6140 6141 6142
            goto cleanup;
        }

D
Daniel Veillard 已提交
6143
        vm->state = VIR_DOMAIN_RUNNING;
6144 6145 6146
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_MIGRATED);
D
Daniel Veillard 已提交
6147
        virDomainSaveStatus(dconn, driver->stateDir, vm);
D
Daniel Veillard 已提交
6148 6149
    } else {
        qemudShutdownVMDaemon (dconn, driver, vm);
6150 6151 6152
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         VIR_DOMAIN_EVENT_STOPPED_FAILED);
6153
        if (!vm->persistent) {
D
Daniel Veillard 已提交
6154
            virDomainRemoveInactive(&driver->domains, vm);
6155 6156
            vm = NULL;
        }
D
Daniel Veillard 已提交
6157
    }
6158 6159

cleanup:
6160 6161
    if (vm)
        virDomainObjUnlock(vm);
6162 6163
    if (event)
        qemuDomainEventQueue(driver, event);
6164
    qemuDriverUnlock(driver);
6165
    return dom;
D
Daniel Veillard 已提交
6166 6167
}

6168 6169 6170 6171 6172 6173 6174 6175 6176 6177 6178 6179 6180 6181 6182 6183
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;

6184
    def = virNodeDeviceDefParseString(dev->conn, xml, EXISTING_DEVICE);
6185 6186 6187 6188 6189 6190 6191 6192 6193 6194 6195 6196 6197 6198 6199 6200 6201 6202 6203 6204 6205 6206 6207 6208 6209 6210 6211 6212 6213 6214 6215 6216 6217 6218 6219 6220 6221 6222 6223 6224 6225 6226 6227 6228 6229 6230 6231 6232 6233 6234 6235 6236 6237 6238 6239 6240 6241 6242 6243 6244 6245 6246 6247 6248 6249 6250 6251 6252 6253 6254 6255 6256 6257 6258 6259 6260 6261 6262
    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)
{
6263
    struct qemud_driver *driver = dev->conn->privateData;
6264 6265 6266 6267 6268 6269 6270 6271 6272 6273 6274
    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;

6275 6276 6277
    qemuDriverLock(driver);

    if (pciResetDevice(dev->conn, pci, driver->activePciHostdevs) < 0)
6278 6279 6280 6281
        goto out;

    ret = 0;
out:
6282
    qemuDriverUnlock(driver);
6283 6284 6285 6286
    pciFreeDevice(dev->conn, pci);
    return ret;
}

6287 6288 6289 6290 6291
static virDriver qemuDriver = {
    VIR_DRV_QEMU,
    "QEMU",
    qemudOpen, /* open */
    qemudClose, /* close */
D
Daniel Veillard 已提交
6292
    qemudSupportsFeature, /* supports_feature */
6293 6294
    qemudGetType, /* type */
    qemudGetVersion, /* version */
6295
    qemudGetHostname, /* getHostname */
6296
    qemudGetMaxVCPUs, /* getMaxVcpus */
6297
    nodeGetInfo, /* nodeGetInfo */
6298 6299 6300
    qemudGetCapabilities, /* getCapabilities */
    qemudListDomains, /* listDomains */
    qemudNumDomains, /* numOfDomains */
6301
    qemudDomainCreate, /* domainCreateXML */
6302 6303 6304 6305 6306
    qemudDomainLookupByID, /* domainLookupByID */
    qemudDomainLookupByUUID, /* domainLookupByUUID */
    qemudDomainLookupByName, /* domainLookupByName */
    qemudDomainSuspend, /* domainSuspend */
    qemudDomainResume, /* domainResume */
6307
    qemudDomainShutdown, /* domainShutdown */
6308 6309 6310
    NULL, /* domainReboot */
    qemudDomainDestroy, /* domainDestroy */
    qemudDomainGetOSType, /* domainGetOSType */
6311 6312 6313
    qemudDomainGetMaxMemory, /* domainGetMaxMemory */
    qemudDomainSetMaxMemory, /* domainSetMaxMemory */
    qemudDomainSetMemory, /* domainSetMemory */
6314 6315 6316
    qemudDomainGetInfo, /* domainGetInfo */
    qemudDomainSave, /* domainSave */
    qemudDomainRestore, /* domainRestore */
P
Paolo Bonzini 已提交
6317
    qemudDomainCoreDump, /* domainCoreDump */
6318
    qemudDomainSetVcpus, /* domainSetVcpus */
6319 6320 6321 6322
#if HAVE_SCHED_GETAFFINITY
    qemudDomainPinVcpu, /* domainPinVcpu */
    qemudDomainGetVcpus, /* domainGetVcpus */
#else
6323 6324
    NULL, /* domainPinVcpu */
    NULL, /* domainGetVcpus */
6325
#endif
6326
    qemudDomainGetMaxVcpus, /* domainGetMaxVcpus */
6327 6328
    qemudDomainGetSecurityLabel, /* domainGetSecurityLabel */
    qemudNodeGetSecurityModel, /* nodeGetSecurityModel */
6329
    qemudDomainDumpXML, /* domainDumpXML */
6330
    qemuDomainXMLFromNative, /* domainXmlFromNative */
6331
    qemuDomainXMLToNative, /* domainXMLToNative */
6332 6333
    qemudListDefinedDomains, /* listDefinedDomains */
    qemudNumDefinedDomains, /* numOfDefinedDomains */
6334 6335 6336
    qemudDomainStart, /* domainCreate */
    qemudDomainDefine, /* domainDefineXML */
    qemudDomainUndefine, /* domainUndefine */
6337
    qemudDomainAttachDevice, /* domainAttachDevice */
6338
    qemudDomainDetachDevice, /* domainDetachDevice */
6339 6340
    qemudDomainGetAutostart, /* domainGetAutostart */
    qemudDomainSetAutostart, /* domainSetAutostart */
6341 6342 6343
    qemuGetSchedulerType, /* domainGetSchedulerType */
    qemuGetSchedulerParameters, /* domainGetSchedulerParameters */
    qemuSetSchedulerParameters, /* domainSetSchedulerParameters */
D
Daniel Veillard 已提交
6344 6345
    NULL, /* domainMigratePrepare (v1) */
    qemudDomainMigratePerform, /* domainMigratePerform */
6346
    NULL, /* domainMigrateFinish */
6347
    qemudDomainBlockStats, /* domainBlockStats */
6348
    qemudDomainInterfaceStats, /* domainInterfaceStats */
6349
    qemudDomainBlockPeek, /* domainBlockPeek */
R
Richard W.M. Jones 已提交
6350
    qemudDomainMemoryPeek, /* domainMemoryPeek */
6351 6352
    nodeGetCellsFreeMemory, /* nodeGetCellsFreeMemory */
    nodeGetFreeMemory,  /* getFreeMemory */
6353 6354
    qemudDomainEventRegister, /* domainEventRegister */
    qemudDomainEventDeregister, /* domainEventDeregister */
D
Daniel Veillard 已提交
6355 6356
    qemudDomainMigratePrepare2, /* domainMigratePrepare2 */
    qemudDomainMigrateFinish2, /* domainMigrateFinish2 */
6357 6358 6359
    qemudNodeDeviceDettach, /* nodeDeviceDettach */
    qemudNodeDeviceReAttach, /* nodeDeviceReAttach */
    qemudNodeDeviceReset, /* nodeDeviceReset */
6360 6361 6362
};


6363
static virStateDriver qemuStateDriver = {
6364 6365 6366 6367
    .initialize = qemudStartup,
    .cleanup = qemudShutdown,
    .reload = qemudReload,
    .active = qemudActive,
6368
};
6369

6370
int qemuRegister(void) {
6371 6372 6373 6374
    virRegisterDriver(&qemuDriver);
    virRegisterStateDriver(&qemuStateDriver);
    return 0;
}