qemu_driver.c 219.3 KB
Newer Older
D
Daniel P. Berrange 已提交
1 2 3
/*
 * driver.c: core driver methods for managing qemu guests
 *
4
 * Copyright (C) 2006, 2007, 2008, 2009 Red Hat, Inc.
D
Daniel P. Berrange 已提交
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
 * Copyright (C) 2006 Daniel P. Berrange
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with this library; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307  USA
 *
 * Author: Daniel P. Berrange <berrange@redhat.com>
 */

24
#include <config.h>
25

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

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

53
#include "virterror_internal.h"
54
#include "logging.h"
55
#include "datatypes.h"
56 57
#include "qemu_driver.h"
#include "qemu_conf.h"
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

110 111 112
static int qemudDomainSetMemoryBalloon(virConnectPtr conn,
                                       virDomainObjPtr vm,
                                       unsigned long newmem);
113 114
static int qemudDetectVcpuPIDs(virConnectPtr conn,
                               virDomainObjPtr vm);
115

116 117 118
static int qemuUpdateActivePciHostdevs(struct qemud_driver *driver,
                                       virDomainDefPtr def);

J
Jim Meyering 已提交
119
static struct qemud_driver *qemu_driver = NULL;
120

121 122 123 124 125 126 127 128 129
static int qemuCgroupControllerActive(struct qemud_driver *driver,
                                      int controller)
{
    if (driver->cgroup == NULL)
        return 0;
    if (driver->cgroupControllers & (1 << controller))
        return 1;
    return 0;
}
130

131
static int
132
qemudLogFD(virConnectPtr conn, struct qemud_driver *driver, const char* name)
133 134 135
{
    char logfile[PATH_MAX];
    mode_t logmode;
G
Guido Günther 已提交
136
    int ret, fd = -1;
137

138 139
    if ((ret = snprintf(logfile, sizeof(logfile), "%s/%s.log",
                        driver->logDir, name))
G
Guido Günther 已提交
140
        < 0 || ret >= sizeof(logfile)) {
141
        virReportOOMError(conn);
142 143 144 145
        return -1;
    }

    logmode = O_CREAT | O_WRONLY;
146 147
    /* Only logrotate files in /var/log, so only append if running privileged */
    if (driver->privileged)
148
        logmode |= O_APPEND;
149 150 151
    else
        logmode |= O_TRUNC;

152
    if ((fd = open(logfile, logmode, S_IRUSR | S_IWUSR)) < 0) {
153 154 155
        virReportSystemError(conn, errno,
                             _("failed to create logfile %s"),
                             logfile);
156 157
        return -1;
    }
158
    if (virSetCloseExec(fd) < 0) {
159 160
        virReportSystemError(conn, errno, "%s",
                             _("Unable to set VM logfile close-on-exec flag"));
161 162 163 164 165 166 167
        close(fd);
        return -1;
    }
    return fd;
}


168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184
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) {
185 186 187
        virReportSystemError(conn, errno,
                             _("failed to create logfile %s"),
                             logfile);
188 189
        return -1;
    }
190
    if (virSetCloseExec(fd) < 0) {
191 192
        virReportSystemError(conn, errno, "%s",
                             _("Unable to set VM logfile close-on-exec flag"));
193 194 195 196
        close(fd);
        return -1;
    }
    if (lseek(fd, pos, SEEK_SET) < 0) {
197 198 199
        virReportSystemError(conn, errno,
                             _("Unable to seek to %lld in %s"),
                             (long long) pos, logfile);
200 201 202 203 204 205
        close(fd);
    }
    return fd;
}


206 207 208
static void
qemudAutostartConfigs(struct qemud_driver *driver) {
    unsigned int i;
209 210 211 212 213
    /* 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
     */
214 215 216
    virConnectPtr conn = virConnectOpen(driver->privileged ?
                                        "qemu:///system" :
                                        "qemu:///session");
217
    /* Ignoring NULL conn which is mostly harmless here */
218

219
    qemuDriverLock(driver);
220
    for (i = 0 ; i < driver->domains.count ; i++) {
221
        virDomainObjPtr vm = driver->domains.objs[i];
222
        virDomainObjLock(vm);
223 224
        if (vm->autostart &&
            !virDomainIsActive(vm)) {
225 226 227 228
            int ret;

            virResetLastError();
            ret = qemudStartVMDaemon(conn, driver, vm, NULL, -1);
229 230
            if (ret < 0) {
                virErrorPtr err = virGetLastError();
231 232
                VIR_ERROR(_("Failed to autostart VM '%s': %s\n"),
                          vm->def->name,
233
                          err ? err->message : "");
234
            } else {
235 236 237 238 239 240
                virDomainEventPtr event =
                    virDomainEventNewFromObj(vm,
                                             VIR_DOMAIN_EVENT_STARTED,
                                             VIR_DOMAIN_EVENT_STARTED_BOOTED);
                if (event)
                    qemuDomainEventQueue(driver, event);
241
            }
242
        }
243
        virDomainObjUnlock(vm);
244
    }
245
    qemuDriverUnlock(driver);
246

247 248
    if (conn)
        virConnectClose(conn);
249 250
}

251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267

/**
 * 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) {
268
        virReportOOMError(conn);
269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287
        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;
}


288 289 290 291 292
static int qemudOpenMonitor(virConnectPtr conn,
                            struct qemud_driver* driver,
                            virDomainObjPtr vm,
                            int reconnect);

293 294 295 296

/*
 * Open an existing VM's monitor, re-detect VCPU threads
 * and re-reserve the security labels in use
297 298
 */
static int
299 300
qemuReconnectDomain(struct qemud_driver *driver,
                    virDomainObjPtr obj)
301
{
302
    int rc;
303

304
    if ((rc = qemudOpenMonitor(NULL, driver, obj, 1)) != 0) {
305 306 307 308
        VIR_ERROR(_("Failed to reconnect monitor for %s: %d\n"),
                  obj->def->name, rc);
        goto error;
    }
309

310 311 312 313
    if (qemuUpdateActivePciHostdevs(driver, obj->def) < 0) {
        goto error;
    }

314 315 316 317 318
    if (obj->def->seclabel.type == VIR_DOMAIN_SECLABEL_DYNAMIC &&
        driver->securityDriver &&
        driver->securityDriver->domainReserveSecurityLabel &&
        driver->securityDriver->domainReserveSecurityLabel(NULL, obj) < 0)
        return -1;
319

320 321
    if (obj->def->id >= driver->nextvmid)
        driver->nextvmid = obj->def->id + 1;
322

323
    return 0;
324

325 326 327
error:
    return -1;
}
328

329 330 331 332 333 334 335 336 337 338
/**
 * qemudReconnectVMs
 *
 * Try to re-open the resources for live VMs that we care
 * about.
 */
static void
qemuReconnectDomains(struct qemud_driver *driver)
{
    int i;
339

340 341 342 343 344 345 346 347 348
    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);
349
        }
350
        virDomainObjUnlock(obj);
351 352 353
    }
}

354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386

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;
}


387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406
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;

407
    VIR_INFO("Initialized security driver %s", security_drv->name);
408 409 410 411 412

    /*
     * Add security policy host caps now that the security driver is
     * initialized.
     */
413 414
    return qemudSecurityCapsInit(security_drv, qemud_drv->caps);
}
415 416


417

418 419 420 421 422 423
/**
 * qemudStartup:
 *
 * Initialization function for the QEmu daemon
 */
static int
424
qemudStartup(int privileged) {
425
    char *base = NULL;
D
Daniel P. Berrange 已提交
426
    char driverConf[PATH_MAX];
427
    int rc;
428

429
    if (VIR_ALLOC(qemu_driver) < 0)
430 431
        return -1;

432
    if (virMutexInit(&qemu_driver->lock) < 0) {
433
        VIR_ERROR("%s", _("cannot initialize mutex"));
434 435 436
        VIR_FREE(qemu_driver);
        return -1;
    }
437
    qemuDriverLock(qemu_driver);
438
    qemu_driver->privileged = privileged;
439

440 441 442
    /* Don't have a dom0 so start from 1 */
    qemu_driver->nextvmid = 1;

443 444
    /* Init callback list */
    if(VIR_ALLOC(qemu_driver->domainEventCallbacks) < 0)
445
        goto out_of_memory;
446 447 448 449 450 451
    if (!(qemu_driver->domainEventQueue = virDomainEventQueueNew()))
        goto out_of_memory;

    if ((qemu_driver->domainEventTimer =
         virEventAddTimeout(-1, qemuDomainEventFlush, qemu_driver, NULL)) < 0)
        goto error;
452

453
    if (privileged) {
454 455
        if (virAsprintf(&qemu_driver->logDir,
                        "%s/log/libvirt/qemu", LOCAL_STATE_DIR) == -1)
456
            goto out_of_memory;
457

D
Daniel P. Berrange 已提交
458
        if ((base = strdup (SYSCONF_DIR "/libvirt")) == NULL)
459
            goto out_of_memory;
460 461

        if (virAsprintf(&qemu_driver->stateDir,
462
                      "%s/run/libvirt/qemu", LOCAL_STATE_DIR) == -1)
463
            goto out_of_memory;
464 465 466 467 468 469 470 471

        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;
472
    } else {
473
        uid_t uid = geteuid();
474 475
        char *userdir = virGetUserDirectory(NULL, uid);
        if (!userdir)
476
            goto error;
477

478
        if (virAsprintf(&qemu_driver->logDir,
479 480
                        "%s/.libvirt/qemu/log", userdir) == -1) {
            VIR_FREE(userdir);
481
            goto out_of_memory;
482
        }
483

484 485
        if (virAsprintf(&base, "%s/.libvirt", userdir) == -1) {
            VIR_FREE(userdir);
486
            goto out_of_memory;
487 488
        }
        VIR_FREE(userdir);
489 490 491

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

    if (virFileMakePath(qemu_driver->stateDir) < 0) {
499
        char ebuf[1024];
500 501
        VIR_ERROR(_("Failed to create state dir '%s': %s\n"),
                  qemu_driver->stateDir, virStrerror(errno, ebuf, sizeof ebuf));
502
        goto error;
503
    }
504 505 506 507 508 509 510 511 512 513 514 515
    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;
    }
516 517 518 519

    /* Configuration paths are either ~/.libvirt/qemu/... (session) or
     * /etc/libvirt/qemu/... (system).
     */
D
Daniel P. Berrange 已提交
520
    if (snprintf (driverConf, sizeof(driverConf), "%s/qemu.conf", base) == -1)
521
        goto out_of_memory;
D
Daniel P. Berrange 已提交
522
    driverConf[sizeof(driverConf)-1] = '\0';
523

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

527
    if (virAsprintf(&qemu_driver->autostartDir, "%s/qemu/autostart", base) == -1)
528 529
        goto out_of_memory;

530
    VIR_FREE(base);
531

532 533 534 535 536 537 538
    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)));
    }

539
    if ((qemu_driver->caps = qemudCapsInit(NULL)) == NULL)
540
        goto out_of_memory;
D
Daniel P. Berrange 已提交
541

542 543 544
    if ((qemu_driver->activePciHostdevs = pciDeviceListNew(NULL)) == NULL)
        goto error;

545
    if (qemudLoadDriverConfig(qemu_driver, driverConf) < 0) {
546 547 548
        goto error;
    }

549 550 551 552 553 554 555 556 557 558 559 560 561 562 563
    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;
        }
    }

564
    if (qemudSecurityInit(qemu_driver) < 0) {
565
        goto error;
D
Daniel P. Berrange 已提交
566 567
    }

568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598
    /* 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;
    }

599 600 601 602 603 604 605 606 607 608 609 610
    /* 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 */
611 612 613 614
    if (virDomainLoadAllConfigs(NULL,
                                qemu_driver->caps,
                                &qemu_driver->domains,
                                qemu_driver->configDir,
615
                                qemu_driver->autostartDir,
616
                                0, NULL, NULL) < 0)
617
        goto error;
618 619
    qemuDriverUnlock(qemu_driver);

620 621
    qemudAutostartConfigs(qemu_driver);

622

623 624
    return 0;

625
out_of_memory:
626
    virReportOOMError(NULL);
627 628 629
error:
    if (qemu_driver)
        qemuDriverUnlock(qemu_driver);
630
    VIR_FREE(base);
631
    qemudShutdown();
632 633 634
    return -1;
}

635 636 637 638
static void qemudNotifyLoadDomain(virDomainObjPtr vm, int newVM, void *opaque)
{
    struct qemud_driver *driver = opaque;

639 640 641 642 643 644 645 646
    if (newVM) {
        virDomainEventPtr event =
            virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_DEFINED,
                                     VIR_DOMAIN_EVENT_DEFINED_ADDED);
        if (event)
            qemuDomainEventQueue(driver, event);
    }
647 648
}

649 650 651 652 653 654 655 656
/**
 * qemudReload:
 *
 * Function to restart the QEmu daemon, it will recheck the configuration
 * files and update its state and the networking
 */
static int
qemudReload(void) {
657 658 659
    if (!qemu_driver)
        return 0;

660
    qemuDriverLock(qemu_driver);
661 662 663 664
    virDomainLoadAllConfigs(NULL,
                            qemu_driver->caps,
                            &qemu_driver->domains,
                            qemu_driver->configDir,
665
                            qemu_driver->autostartDir,
666
                            0, qemudNotifyLoadDomain, qemu_driver);
667
    qemuDriverUnlock(qemu_driver);
668

669
    qemudAutostartConfigs(qemu_driver);
670 671

    return 0;
672 673
}

674 675 676 677 678 679 680 681 682 683
/**
 * 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) {
684
    unsigned int i;
685
    int active = 0;
686

687 688 689
    if (!qemu_driver)
        return 0;

690 691 692 693 694 695 696 697
    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);
    }
698

699 700
    qemuDriverUnlock(qemu_driver);
    return active;
701 702
}

703 704 705 706 707 708 709
/**
 * qemudShutdown:
 *
 * Shutdown the QEmu daemon, it will stop all active domains and networks
 */
static int
qemudShutdown(void) {
710

711
    if (!qemu_driver)
712
        return -1;
713

714
    qemuDriverLock(qemu_driver);
715
    pciDeviceListFree(NULL, qemu_driver->activePciHostdevs);
716 717
    virCapabilitiesFree(qemu_driver->caps);

718
    virDomainObjListFree(&qemu_driver->domains);
719

720
    VIR_FREE(qemu_driver->securityDriverName);
721
    VIR_FREE(qemu_driver->logDir);
722 723
    VIR_FREE(qemu_driver->configDir);
    VIR_FREE(qemu_driver->autostartDir);
724
    VIR_FREE(qemu_driver->stateDir);
725 726
    VIR_FREE(qemu_driver->libDir);
    VIR_FREE(qemu_driver->cacheDir);
727
    VIR_FREE(qemu_driver->vncTLSx509certdir);
J
Jim Meyering 已提交
728
    VIR_FREE(qemu_driver->vncListen);
729
    VIR_FREE(qemu_driver->vncPassword);
730
    VIR_FREE(qemu_driver->vncSASLdir);
731
    VIR_FREE(qemu_driver->saveImageFormat);
732 733
    VIR_FREE(qemu_driver->hugetlbfs_mount);
    VIR_FREE(qemu_driver->hugepage_path);
D
Daniel P. Berrange 已提交
734

735 736
    /* Free domain callback list */
    virDomainEventCallbackListFree(qemu_driver->domainEventCallbacks);
737 738 739 740
    virDomainEventQueueFree(qemu_driver->domainEventQueue);

    if (qemu_driver->domainEventTimer != -1)
        virEventRemoveTimeout(qemu_driver->domainEventTimer);
741

742 743 744
    if (qemu_driver->brctl)
        brShutdown(qemu_driver->brctl);

745 746
    virCgroupFree(&qemu_driver->cgroup);

747
    qemuDriverUnlock(qemu_driver);
748
    virMutexDestroy(&qemu_driver->lock);
749
    VIR_FREE(qemu_driver);
750 751

    return 0;
752 753 754
}

/* Return -1 for error, 1 to continue reading and 0 for success */
755
typedef int qemudHandlerMonitorOutput(virConnectPtr conn,
756
                                      virDomainObjPtr vm,
757 758 759
                                      const char *output,
                                      int fd);

760 761 762
/*
 * Returns -1 for error, 0 on end-of-file, 1 for success
 */
763
static int
764
qemudReadMonitorOutput(virConnectPtr conn,
765
                       virDomainObjPtr vm,
766 767
                       int fd,
                       char *buf,
G
Guido Günther 已提交
768
                       size_t buflen,
769
                       qemudHandlerMonitorOutput func,
770 771
                       const char *what,
                       int timeout)
772
{
G
Guido Günther 已提交
773
    size_t got = 0;
774
    buf[0] = '\0';
775
    timeout *= 1000; /* poll wants milli seconds */
776

777
    /* Consume & discard the initial greeting */
778
    while (got < (buflen-1)) {
G
Guido Günther 已提交
779
        ssize_t ret;
780 781

        ret = read(fd, buf+got, buflen-got-1);
782

783 784 785 786 787 788
        if (ret < 0) {
            struct pollfd pfd = { .fd = fd, .events = POLLIN };
            if (errno == EINTR)
                continue;

            if (errno != EAGAIN) {
789 790 791
                virReportSystemError(conn, errno,
                                     _("Failure while reading %s startup output"),
                                     what);
792 793 794
                return -1;
            }

795
            ret = poll(&pfd, 1, timeout);
796
            if (ret == 0) {
797
                qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
798
                                 _("Timed out while reading %s startup output"), what);
799 800 801
                return -1;
            } else if (ret == -1) {
                if (errno != EINTR) {
802 803 804
                    virReportSystemError(conn, errno,
                                         _("Failure while reading %s startup output"),
                                         what);
805 806 807 808 809 810 811 812
                    return -1;
                }
            } else {
                /* Make sure we continue loop & read any further data
                   available before dealing with EOF */
                if (pfd.revents & (POLLIN | POLLHUP))
                    continue;

813
                qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
814
                                 _("Failure while reading %s startup output"), what);
815 816
                return -1;
            }
817 818
        } else if (ret == 0) {
            return 0;
819 820 821
        } else {
            got += ret;
            buf[got] = '\0';
822 823 824 825 826 827
            ret = func(conn, vm, buf, fd);
            if (ret == -1)
                return -1;
            if (ret == 1)
                continue;
            return 1;
828 829 830
        }
    }

831
    qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
832
                     _("Out of space while reading %s startup output"), what);
833 834 835 836
    return -1;

}

837 838 839 840 841 842 843 844 845

/*
 * Returns -1 for error, 0 on success
 */
static int
qemudReadLogOutput(virConnectPtr conn,
                   virDomainObjPtr vm,
                   int fd,
                   char *buf,
G
Guido Günther 已提交
846
                   size_t buflen,
847 848 849 850
                   qemudHandlerMonitorOutput func,
                   const char *what,
                   int timeout)
{
851
    int retries = (timeout*10);
852
    int got = 0;
853 854 855
    buf[0] = '\0';

    while (retries) {
856
        ssize_t func_ret, ret;
857
        int isdead = 0;
G
Guido Günther 已提交
858

859 860
        func_ret = func(conn, vm, buf, fd);

861 862
        if (kill(vm->pid, 0) == -1 && errno == ESRCH)
            isdead = 1;
863

864 865
        /* Any failures should be detected before we read the log, so we
         * always have something useful to report on failure. */
866 867
        ret = saferead(fd, buf+got, buflen-got-1);
        if (ret < 0) {
868 869 870 871 872 873
            virReportSystemError(conn, errno,
                                 _("Failure while reading %s log output"),
                                 what);
            return -1;
        }

874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889
        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;
        }

890 891
        if (func_ret <= 0)
            return func_ret;
892 893 894 895 896 897 898 899 900 901

        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;
}

902
static int
903
qemudCheckMonitorPrompt(virConnectPtr conn ATTRIBUTE_UNUSED,
904
                        virDomainObjPtr vm,
905 906 907 908 909 910 911 912 913 914 915
                        const char *output,
                        int fd)
{
    if (strstr(output, "(qemu) ") == NULL)
        return 1; /* keep reading */

    vm->monitor = fd;

    return 0;
}

916
static int
917 918 919 920 921
qemudOpenMonitorCommon(virConnectPtr conn,
                       struct qemud_driver* driver,
                       virDomainObjPtr vm,
                       int monfd,
                       int reconnect)
922
{
923
    char buf[1024];
924
    int ret;
925

926
    if (virSetCloseExec(monfd) < 0) {
927
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
928
                         "%s", _("Unable to set monitor close-on-exec flag"));
929
        return -1;
930
    }
931
    if (virSetNonBlock(monfd) < 0) {
932
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
933
                         "%s", _("Unable to put monitor into non-blocking mode"));
934
        return -1;
935 936
    }

937
    if (!reconnect) {
938 939 940 941
        if (qemudReadMonitorOutput(conn,
                                   vm, monfd,
                                   buf, sizeof(buf),
                                   qemudCheckMonitorPrompt,
942
                                   "monitor", 10) <= 0)
943 944 945
            ret = -1;
        else
            ret = 0;
946 947 948 949 950 951
    } else {
        vm->monitor = monfd;
        ret = 0;
    }

    if (ret != 0)
952
        return ret;
953

954 955
    if ((vm->monitorWatch = virEventAddHandle(vm->monitor,
                                              VIR_EVENT_HANDLE_HANGUP | VIR_EVENT_HANDLE_ERROR,
956 957
                                              qemudDispatchVMEvent,
                                              driver, NULL)) < 0)
958
        return -1;
959

960 961
    return 0;
}
962

963 964 965 966 967 968 969 970 971
static int
qemudOpenMonitorUnix(virConnectPtr conn,
                     struct qemud_driver* driver,
                     virDomainObjPtr vm,
                     const char *monitor,
                     int reconnect)
{
    struct sockaddr_un addr;
    int monfd;
972
    int timeout = 3; /* In seconds */
973
    int ret, i = 0;
974 975 976 977 978 979 980 981 982

    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
983 984 985 986 987
    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;
    }
988

989 990 991 992 993 994
    do {
        ret = connect(monfd, (struct sockaddr *) &addr, sizeof(addr));

        if (ret == 0)
            break;

995 996
        if (errno == ENOENT || errno == ECONNREFUSED) {
            /* ENOENT       : Socket may not have shown up yet
997 998 999 1000
             * ECONNREFUSED : Leftover socket hasn't been removed yet */
            continue;
        }

1001 1002 1003
        virReportSystemError(conn, errno, "%s",
                             _("failed to connect to monitor socket"));
        goto error;
1004 1005 1006 1007 1008 1009 1010

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

    if (ret != 0) {
        virReportSystemError(conn, errno, "%s",
                             _("monitor socket did not show up."));
        goto error;
1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022
    }

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

    return 0;

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

1023 1024 1025 1026 1027 1028 1029 1030
static int
qemudOpenMonitorPty(virConnectPtr conn,
                    struct qemud_driver* driver,
                    virDomainObjPtr vm,
                    const char *monitor,
                    int reconnect)
{
    int monfd;
1031

1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043
    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:
1044
    close(monfd);
1045 1046 1047 1048 1049 1050 1051 1052 1053 1054
    return -1;
}

static int
qemudOpenMonitor(virConnectPtr conn,
                 struct qemud_driver *driver,
                 virDomainObjPtr vm,
                 int reconnect)
{
    switch (vm->monitor_chr->type) {
1055 1056 1057 1058
    case VIR_DOMAIN_CHR_TYPE_UNIX:
        return qemudOpenMonitorUnix(conn, driver, vm,
                                    vm->monitor_chr->data.nix.path,
                                    reconnect);
1059 1060 1061 1062 1063 1064 1065 1066 1067 1068
    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;
    }
1069 1070
}

1071
/* Returns -1 for error, 0 success, 1 continue reading */
1072 1073 1074 1075 1076 1077
static int
qemudExtractMonitorPath(virConnectPtr conn,
                        const char *haystack,
                        size_t *offset,
                        char **path)
{
1078
    static const char needle[] = "char device redirected to";
1079
    char *tmp, *dev;
1080

1081
    VIR_FREE(*path);
1082
    /* First look for our magic string */
1083 1084 1085 1086 1087
    if (!(tmp = strstr(haystack + *offset, needle))) {
        return 1;
    }
    tmp += sizeof(needle);
    dev = tmp;
1088

1089 1090 1091 1092 1093
    /*
     * And look for first whitespace character and nul terminate
     * to mark end of the pty path
     */
    while (*tmp) {
1094
        if (c_isspace(*tmp)) {
1095 1096
            *path = strndup(dev, tmp-dev);
            if (*path == NULL) {
1097
                virReportOOMError(conn);
1098 1099
                return -1;
            }
1100

1101
            /* ... now further update offset till we get EOL */
1102
            *offset = tmp - haystack;
1103 1104
            return 0;
        }
1105
        tmp++;
1106 1107 1108 1109 1110
    }

    /*
     * We found a path, but didn't find any whitespace,
     * so it must be still incomplete - we should at
1111 1112
     * least see a \n - indicate that we want to carry
     * on trying again
1113
     */
1114
    return 1;
1115 1116 1117
}

static int
1118
qemudFindCharDevicePTYs(virConnectPtr conn,
1119
                        virDomainObjPtr vm,
1120 1121
                        const char *output,
                        int fd ATTRIBUTE_UNUSED)
1122
{
1123
    size_t offset = 0;
1124
    int ret, i;
1125 1126

    /* The order in which QEMU prints out the PTY paths is
1127 1128
       the order in which it procsses its serial and parallel
       device args. This code must match that ordering.... */
1129

1130
    /* first comes the serial devices */
1131 1132
    for (i = 0 ; i < vm->def->nserials ; i++) {
        virDomainChrDefPtr chr = vm->def->serials[i];
1133 1134 1135
        if (chr->type == VIR_DOMAIN_CHR_TYPE_PTY) {
            if ((ret = qemudExtractMonitorPath(conn, output, &offset,
                                               &chr->data.file.path)) != 0)
1136
                return ret;
1137 1138 1139
        }
    }

1140
    /* then the parallel devices */
1141 1142
    for (i = 0 ; i < vm->def->nparallels ; i++) {
        virDomainChrDefPtr chr = vm->def->parallels[i];
1143 1144 1145
        if (chr->type == VIR_DOMAIN_CHR_TYPE_PTY) {
            if ((ret = qemudExtractMonitorPath(conn, output, &offset,
                                               &chr->data.file.path)) != 0)
1146
                return ret;
1147 1148 1149
        }
    }

1150
    return 0;
1151 1152
}

1153 1154 1155 1156
static int
qemudWaitForMonitor(virConnectPtr conn,
                    struct qemud_driver* driver,
                    virDomainObjPtr vm, off_t pos)
1157
{
1158
    char buf[4096]; /* Plenty of space to get startup greeting */
1159 1160 1161 1162 1163
    int logfd;
    int ret;

    if ((logfd = qemudLogReadFD(conn, driver->logDir, vm->def->name, pos))
        < 0)
1164
        return -1;
1165

1166 1167 1168
    ret = qemudReadLogOutput(conn, vm, logfd, buf, sizeof(buf),
                             qemudFindCharDevicePTYs,
                             "console", 3);
1169
    if (close(logfd) < 0) {
1170
        char ebuf[4096];
1171
        VIR_WARN(_("Unable to close logfile: %s\n"),
1172 1173
                 virStrerror(errno, ebuf, sizeof ebuf));
    }
1174

1175 1176 1177 1178 1179 1180
    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;
    }
1181

1182 1183 1184 1185
    if (qemudOpenMonitor(conn, driver, vm, 0) < 0)
        return -1;

    return 0;
1186 1187
}

1188 1189
static int
qemudDetectVcpuPIDs(virConnectPtr conn,
1190
                    virDomainObjPtr vm) {
1191 1192 1193 1194 1195 1196
    char *qemucpus = NULL;
    char *line;
    int lastVcpu = -1;

    /* Only KVM has seperate threads for CPUs,
       others just use main QEMU process for CPU */
1197
    if (vm->def->virtType != VIR_DOMAIN_VIRT_KVM)
1198 1199 1200 1201 1202
        vm->nvcpupids = 1;
    else
        vm->nvcpupids = vm->def->vcpus;

    if (VIR_ALLOC_N(vm->vcpupids, vm->nvcpupids) < 0) {
1203
        virReportOOMError(conn);
1204 1205 1206
        return -1;
    }

1207
    if (vm->def->virtType != VIR_DOMAIN_VIRT_KVM) {
1208 1209 1210 1211
        vm->vcpupids[0] = vm->pid;
        return 0;
    }

1212
    if (qemudMonitorCommand(vm, "info cpus", &qemucpus) < 0) {
1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                         "%s", _("cannot run monitor command to fetch CPU thread info"));
        VIR_FREE(vm->vcpupids);
        vm->nvcpupids = 0;
        return -1;
    }

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

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

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

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

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

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

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

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

1271
    VIR_FREE(qemucpus);
1272 1273 1274 1275
    return 0;

error:
    VIR_FREE(vm->vcpupids);
1276 1277
    vm->nvcpupids = 0;
    VIR_FREE(qemucpus);
1278 1279 1280 1281 1282 1283 1284 1285

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

1286 1287
static int
qemudInitCpus(virConnectPtr conn,
D
Daniel Veillard 已提交
1288 1289
              virDomainObjPtr vm,
              const char *migrateFrom) {
1290 1291 1292 1293 1294
#if HAVE_SCHED_GETAFFINITY
    cpu_set_t mask;
    int i, maxcpu = QEMUD_CPUMASK_LEN;
    virNodeInfo nodeinfo;

1295
    if (nodeGetInfo(conn, &nodeinfo) < 0)
1296 1297 1298 1299 1300 1301 1302 1303
        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
1304 1305 1306 1307 1308 1309
    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++)
1310
            CPU_SET(i, &mask);
D
Daniel P. Berrange 已提交
1311
    }
1312 1313 1314 1315

    for (i = 0 ; i < vm->nvcpupids ; i++) {
        if (sched_setaffinity(vm->vcpupids[i],
                              sizeof(mask), &mask) < 0) {
1316 1317
            virReportSystemError(conn, errno, "%s",
                                 _("failed to set CPU affinity"));
1318 1319 1320 1321 1322
            return -1;
        }
    }
#endif /* HAVE_SCHED_GETAFFINITY */

D
Daniel Veillard 已提交
1323 1324
    if (migrateFrom == NULL) {
        /* Allow the CPUS to start executing */
1325 1326 1327 1328
        if (qemudMonitorSendCont(conn, vm) < 0) {
            if (virGetLastError() == NULL)
                qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                                 "%s", _("resume operation failed"));
D
Daniel Veillard 已提交
1329 1330
            return -1;
        }
1331 1332 1333 1334 1335 1336
    }

    return 0;
}


1337 1338 1339 1340 1341 1342
static int
qemudInitPasswords(virConnectPtr conn,
                   struct qemud_driver *driver,
                   virDomainObjPtr vm) {
    char *info = NULL;

1343 1344 1345
    if ((vm->def->ngraphics == 1) &&
        vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
        (vm->def->graphics[0]->data.vnc.passwd || driver->vncPassword)) {
1346 1347

        if (qemudMonitorCommandExtra(vm, "change vnc password",
1348 1349
                                     vm->def->graphics[0]->data.vnc.passwd ?
                                     vm->def->graphics[0]->data.vnc.passwd :
1350 1351
                                     driver->vncPassword,
                                     QEMU_PASSWD_PROMPT,
1352
                                     -1, &info) < 0) {
1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363
            qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                             "%s", _("setting VNC password failed"));
            return -1;
        }
        VIR_FREE(info);
    }

    return 0;
}


1364
static int qemudNextFreeVNCPort(struct qemud_driver *driver ATTRIBUTE_UNUSED) {
1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399
    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;
}

1400 1401 1402 1403 1404
static pciDeviceList *
qemuGetPciHostDeviceList(virConnectPtr conn,
                         virDomainDefPtr def)
{
    pciDeviceList *list;
1405 1406
    int i;

1407 1408
    if (!(list = pciDeviceListNew(conn)))
        return NULL;
1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423

    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);
1424 1425 1426 1427
        if (!dev) {
            pciDeviceListFree(conn, list);
            return NULL;
        }
1428

1429
        if (pciDeviceListAdd(conn, list, dev) < 0) {
1430
            pciFreeDevice(conn, dev);
1431 1432
            pciDeviceListFree(conn, list);
            return NULL;
1433 1434
        }

1435
        pciDeviceSetManaged(dev, hostdev->managed);
1436 1437
    }

1438 1439 1440 1441
    return list;
}

static int
1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473
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)
1474 1475 1476 1477 1478 1479 1480 1481 1482 1483
{
    pciDeviceList *pcidevs;
    int i;

    if (!def->nhostdevs)
        return 0;

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

1484
    /* We have to use 3 loops here. *All* devices must
1485 1486
     * be detached before we reset any of them, because
     * in some cases you have to reset the whole PCI,
1487 1488
     * which impacts all devices on it. Also, all devices
     * must be reset before being marked as active.
1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503
     */

    /* 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++)
1504 1505
        if (pciResetDevice(conn, pcidevs->devs[i],
                           driver->activePciHostdevs) < 0)
1506 1507
            goto error;

1508 1509 1510 1511 1512 1513 1514 1515 1516
    /* 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;
    }

1517
    pciDeviceListFree(conn, pcidevs);
1518 1519 1520
    return 0;

error:
1521
    pciDeviceListFree(conn, pcidevs);
1522 1523 1524
    return -1;
}

1525
static void
1526 1527 1528
qemuDomainReAttachHostDevices(virConnectPtr conn,
                              struct qemud_driver *driver,
                              virDomainDefPtr def)
1529
{
1530
    pciDeviceList *pcidevs;
1531 1532
    int i;

1533 1534
    if (!def->nhostdevs)
        return;
1535

1536 1537 1538 1539 1540 1541
    if (!(pcidevs = qemuGetPciHostDeviceList(conn, def))) {
        virErrorPtr err = virGetLastError();
        VIR_ERROR(_("Failed to allocate pciDeviceList: %s\n"),
                  err ? err->message : "");
        virResetError(err);
        return;
1542 1543
    }

1544 1545
    /* Again 3 loops; mark all devices as inactive before reset
     * them and reset all the devices before re-attach */
1546

1547
    for (i = 0; i < pcidevs->count; i++)
1548 1549 1550 1551 1552
        pciDeviceListDel(conn, driver->activePciHostdevs, pcidevs->devs[i]);

    for (i = 0; i < pcidevs->count; i++)
        if (pciResetDevice(conn, pcidevs->devs[i],
                           driver->activePciHostdevs) < 0) {
1553
            virErrorPtr err = virGetLastError();
1554
            VIR_ERROR(_("Failed to reset PCI device: %s\n"),
1555 1556 1557 1558
                      err ? err->message : "");
            virResetError(err);
        }

1559 1560 1561
    for (i = 0; i < pcidevs->count; i++)
        if (pciDeviceGetManaged(pcidevs->devs[i]) &&
            pciReAttachDevice(conn, pcidevs->devs[i]) < 0) {
1562
            virErrorPtr err = virGetLastError();
1563
            VIR_ERROR(_("Failed to re-attach PCI device: %s\n"),
1564 1565 1566 1567
                      err ? err->message : "");
            virResetError(err);
        }

1568
    pciDeviceListFree(conn, pcidevs);
1569 1570
}

1571 1572 1573 1574 1575 1576 1577 1578 1579 1580
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

1581 1582 1583 1584 1585 1586
static int qemuSetupCgroup(virConnectPtr conn,
                           struct qemud_driver *driver,
                           virDomainObjPtr vm)
{
    virCgroupPtr cgroup = NULL;
    int rc;
1587
    unsigned int i;
1588 1589 1590 1591
    const char *const *deviceACL =
        driver->cgroupDeviceACL ?
        (const char *const *)driver->cgroupDeviceACL :
        defaultDeviceACL;
1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603

    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;
    }

1604 1605
    if (qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_DEVICES)) {
        rc = virCgroupDenyAllDevices(cgroup);
1606
        if (rc != 0) {
1607 1608 1609 1610 1611
            if (rc == -EPERM) {
                VIR_WARN0("Group devices ACL is not accessible, disabling whitelisting");
                goto done;
            }

1612
            virReportSystemError(conn, -rc,
1613
                                 _("Unable to deny all devices for %s"), vm->def->name);
1614 1615 1616
            goto cleanup;
        }

1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630
        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;
            }
        }
1631

1632
        rc = virCgroupAllowDeviceMajor(cgroup, 'c', DEVICE_PTY_MAJOR);
1633 1634
        if (rc != 0) {
            virReportSystemError(conn, -rc, "%s",
1635
                                 _("unable to allow /dev/pts/ devices"));
1636 1637 1638
            goto cleanup;
        }

1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657
        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;
            }
1658 1659 1660 1661
        }
    }

done:
1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730
    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;
}


1731 1732 1733 1734 1735 1736 1737 1738 1739
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;
}

1740 1741

#ifdef __linux__
1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752
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;

1753 1754
    VIR_DEBUG("Setting ownership on %s to %d:%d", file, owner->uid, owner->gid);

1755 1756 1757 1758 1759 1760 1761 1762
    if (chown(file, owner->uid, owner->gid) < 0) {
        virReportSystemError(conn, errno, _("cannot set ownership on %s"), file);
        return -1;
    }

    return 0;
}

1763 1764 1765 1766
static int qemuDomainSetHostdevUSBOwnership(virConnectPtr conn,
                                            virDomainHostdevDefPtr def,
                                            uid_t uid, gid_t gid)
{
1767 1768
    struct qemuFileOwner owner = { uid, gid };
    int ret = -1;
1769 1770 1771 1772 1773 1774

    /* 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;

1775 1776 1777 1778 1779 1780 1781 1782 1783
    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);
1784

1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795
    usbFreeDevice(conn, dev);
cleanup:
    return ret;
}

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

1796 1797
    VIR_DEBUG("Setting ownership on %s to %d:%d", file, owner->uid, owner->gid);

1798 1799
    if (chown(file, owner->uid, owner->gid) < 0) {
        virReportSystemError(conn, errno, _("cannot set ownership on %s"), file);
1800 1801 1802 1803 1804 1805 1806 1807 1808 1809
        return -1;
    }

    return 0;
}

static int qemuDomainSetHostdevPCIOwnership(virConnectPtr conn,
                                            virDomainHostdevDefPtr def,
                                            uid_t uid, gid_t gid)
{
1810
    struct qemuFileOwner owner = { uid, gid };
1811 1812
    int ret = -1;

1813 1814 1815 1816 1817
    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);
1818

1819
    if (!dev)
1820 1821
        goto cleanup;

1822 1823
    ret = pciDeviceFileIterate(conn, dev,
                               qemuDomainSetHostdevPCIOwnershipActor, &owner);
1824

1825
    pciFreeDevice(conn, dev);
1826 1827 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 1855 1856
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

}

1857 1858
static int qemuDomainSetFileOwnership(virConnectPtr conn,
                                      const char *path,
1859 1860 1861
                                      uid_t uid, gid_t gid)
{

1862
    if (!path)
1863 1864
        return 0;

1865 1866
    VIR_DEBUG("Setting ownership on %s to %d:%d", path, uid, gid);
    if (chown(path, uid, gid) < 0) {
1867
        virReportSystemError(conn, errno, _("cannot set ownership on %s"),
1868
                             path);
1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897
        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;

1898
        return qemuDomainSetFileOwnership(conn, def->data.disk->src, uid, gid);
1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925

    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;

1926 1927 1928 1929
    if (qemuDomainSetFileOwnership(conn, def->os.kernel, uid, gid) < 0 ||
        qemuDomainSetFileOwnership(conn, def->os.initrd, uid, gid) < 0)
        return -1;

1930 1931 1932 1933 1934
    for (i = 0 ; i < def->ndisks ; i++) {
        if (restore &&
            (def->disks[i]->readonly || def->disks[i]->shared))
            continue;

1935
        if (qemuDomainSetFileOwnership(conn, def->disks[i]->src, uid, gid) < 0)
1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946
            return -1;
    }

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

    return 0;
}

1947 1948 1949
static virDomainPtr qemudDomainLookupByName(virConnectPtr conn,
                                            const char *name);

1950 1951 1952 1953
struct qemudHookData {
    virConnectPtr conn;
    virDomainObjPtr vm;
    struct qemud_driver *driver;
1954 1955 1956
};

static int qemudSecurityHook(void *data) {
1957 1958 1959 1960
    struct qemudHookData *h = data;

    if (qemuAddToCgroup(h->driver, h->vm->def) < 0)
        return -1;
1961

1962
    if (qemudDomainSetSecurityLabel(h->conn, h->driver, h->vm) < 0)
1963 1964 1965 1966 1967
        return -1;

    if (h->driver->privileged) {
        if (qemuDomainSetAllDeviceOwnership(h->conn, h->driver, h->vm->def, 0) < 0)
            return -1;
1968

1969 1970
        DEBUG("Dropping privileges of VM to %d:%d", h->driver->user, h->driver->group);

1971 1972 1973 1974 1975
        if (h->driver->group) {
            if (setregid(h->driver->group, h->driver->group) < 0) {
                virReportSystemError(NULL, errno,
                                     _("cannot change to '%d' group"),
                                     h->driver->group);
1976
                return -1;
1977
            }
1978
        }
1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989
        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;
1990 1991
}

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001
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",
2002
                    driver->libDir, vm) < 0) {
2003 2004 2005 2006 2007 2008 2009
        virReportOOMError(conn);
        return -1;
    }

    return 0;
}

2010 2011
static int qemudStartVMDaemon(virConnectPtr conn,
                              struct qemud_driver *driver,
2012
                              virDomainObjPtr vm,
2013 2014
                              const char *migrateFrom,
                              int stdin_fd) {
2015
    const char **argv = NULL, **tmp;
2016
    const char **progenv = NULL;
2017
    int i, ret;
2018
    struct stat sb;
2019 2020
    int *tapfds = NULL;
    int ntapfds = 0;
2021
    unsigned int qemuCmdFlags;
2022
    fd_set keepfd;
2023
    const char *emulator;
G
Guido Günther 已提交
2024
    pid_t child;
2025
    int pos = -1;
2026
    char ebuf[1024];
2027
    char *pidfile = NULL;
2028
    int logfile;
2029

2030
    struct qemudHookData hookData;
2031 2032 2033 2034
    hookData.conn = conn;
    hookData.vm = vm;
    hookData.driver = driver;

2035
    FD_ZERO(&keepfd);
2036

2037
    if (virDomainIsActive(vm)) {
2038
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_INVALID,
2039
                         "%s", _("VM is already active"));
2040 2041 2042
        return -1;
    }

2043 2044 2045 2046 2047 2048 2049 2050
    /* 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;

2051 2052 2053
    /* Ensure no historical cgroup for this VM is lieing around bogus settings */
    qemuRemoveCgroup(conn, driver, vm);

2054 2055 2056
    if ((vm->def->ngraphics == 1) &&
        vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
        vm->def->graphics[0]->data.vnc.autoport) {
2057
        int port = qemudNextFreeVNCPort(driver);
2058
        if (port < 0) {
2059
            qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
2060
                             "%s", _("Unable to find an unused VNC port"));
2061
            goto cleanup;
2062
        }
2063
        vm->def->graphics[0]->data.vnc.port = port;
2064
    }
2065

2066
    if (virFileMakePath(driver->logDir) < 0) {
2067 2068 2069
        virReportSystemError(conn, errno,
                             _("cannot create log directory %s"),
                             driver->logDir);
2070
        goto cleanup;
2071 2072
    }

2073
    if ((logfile = qemudLogFD(conn, driver, vm->def->name)) < 0)
2074
        goto cleanup;
2075

2076 2077
    emulator = vm->def->emulator;

2078 2079 2080 2081
    /* 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
     */
2082
    if (stat(emulator, &sb) < 0) {
2083 2084 2085
        virReportSystemError(conn, errno,
                             _("Cannot find QEMU binary %s"),
                             emulator);
2086
        goto cleanup;
2087 2088
    }

2089
    if (qemudExtractVersionInfo(emulator,
2090
                                NULL,
2091
                                &qemuCmdFlags) < 0)
2092
        goto cleanup;
2093

2094 2095 2096
    if (qemuSetupCgroup(conn, driver, vm) < 0)
        goto cleanup;

2097
    if (qemuPrepareHostDevices(conn, driver, vm->def) < 0)
2098
        goto cleanup;
2099

2100 2101 2102 2103 2104
    if (VIR_ALLOC(vm->monitor_chr) < 0) {
        virReportOOMError(conn);
        goto cleanup;
    }

2105 2106
    if (qemuPrepareMonitorChr(conn, driver, vm->monitor_chr, vm->def->name) < 0)
        goto cleanup;
2107

D
Daniel P. Berrange 已提交
2108 2109 2110 2111 2112 2113 2114
    if ((ret = virFileDeletePid(driver->stateDir, vm->def->name)) != 0) {
        virReportSystemError(conn, ret,
                             _("Cannot remove stale PID file for %s"),
                             vm->def->name);
        goto cleanup;
    }

2115 2116 2117 2118 2119
    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
2120

2121
    vm->def->id = driver->nextvmid++;
2122
    if (qemudBuildCommandLine(conn, driver, vm->def, vm->monitor_chr,
2123
                              qemuCmdFlags, &argv, &progenv,
2124 2125
                              &tapfds, &ntapfds, migrateFrom) < 0)
        goto cleanup;
2126

2127 2128
    tmp = progenv;
    while (*tmp) {
2129
        if (safewrite(logfile, *tmp, strlen(*tmp)) < 0)
2130
            VIR_WARN(_("Unable to write envv to logfile: %s\n"),
2131
                     virStrerror(errno, ebuf, sizeof ebuf));
2132
        if (safewrite(logfile, " ", 1) < 0)
2133
            VIR_WARN(_("Unable to write envv to logfile: %s\n"),
2134
                     virStrerror(errno, ebuf, sizeof ebuf));
2135 2136
        tmp++;
    }
2137 2138
    tmp = argv;
    while (*tmp) {
2139
        if (safewrite(logfile, *tmp, strlen(*tmp)) < 0)
2140
            VIR_WARN(_("Unable to write argv to logfile: %s\n"),
2141
                     virStrerror(errno, ebuf, sizeof ebuf));
2142
        if (safewrite(logfile, " ", 1) < 0)
2143
            VIR_WARN(_("Unable to write argv to logfile: %s\n"),
2144
                     virStrerror(errno, ebuf, sizeof ebuf));
2145 2146
        tmp++;
    }
2147
    if (safewrite(logfile, "\n", 1) < 0)
2148
        VIR_WARN(_("Unable to write argv to logfile: %s\n"),
2149
                 virStrerror(errno, ebuf, sizeof ebuf));
2150

2151
    if ((pos = lseek(logfile, 0, SEEK_END)) < 0)
2152
        VIR_WARN(_("Unable to seek to end of logfile: %s\n"),
2153
                 virStrerror(errno, ebuf, sizeof ebuf));
2154

2155 2156 2157
    for (i = 0 ; i < ntapfds ; i++)
        FD_SET(tapfds[i], &keepfd);

2158
    ret = virExecDaemonize(conn, argv, progenv, &keepfd, &child,
2159
                           stdin_fd, &logfile, &logfile,
2160
                           VIR_EXEC_NONBLOCK | VIR_EXEC_CLEAR_CAPS,
2161 2162 2163
                           qemudSecurityHook, &hookData,
                           pidfile);
    VIR_FREE(pidfile);
G
Guido Günther 已提交
2164 2165 2166

    /* wait for qemu process to to show up */
    if (ret == 0) {
2167
        if (virFileReadPid(driver->stateDir, vm->def->name, &vm->pid)) {
2168
            qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
2169
                             _("Domain %s didn't show up\n"), vm->def->name);
2170
            ret = -1;
G
Guido Günther 已提交
2171
        }
2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182
    } 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;
    }
2183 2184

    vm->state = migrateFrom ? VIR_DOMAIN_PAUSED : VIR_DOMAIN_RUNNING;
2185

2186
    for (i = 0 ; argv[i] ; i++)
2187 2188
        VIR_FREE(argv[i]);
    VIR_FREE(argv);
2189

2190 2191 2192 2193
    for (i = 0 ; progenv[i] ; i++)
        VIR_FREE(progenv[i]);
    VIR_FREE(progenv);

2194 2195 2196
    if (tapfds) {
        for (i = 0 ; i < ntapfds ; i++) {
            close(tapfds[i]);
2197
        }
2198
        VIR_FREE(tapfds);
2199 2200
    }

2201 2202 2203 2204 2205 2206 2207 2208
    if (ret == -1)
        goto cleanup;

    if ((qemudWaitForMonitor(conn, driver, vm, pos) < 0) ||
        (qemudDetectVcpuPIDs(conn, vm) < 0) ||
        (qemudInitCpus(conn, vm, migrateFrom) < 0) ||
        (qemudInitPasswords(conn, driver, vm) < 0) ||
        (qemudDomainSetMemoryBalloon(conn, vm, vm->def->memory) < 0) ||
2209
        (virDomainSaveStatus(conn, driver->stateDir, vm) < 0)) {
2210 2211 2212
        qemudShutdownVMDaemon(conn, driver, vm);
        ret = -1;
        /* No need for 'goto cleanup' now since qemudShutdownVMDaemon does enough */
2213 2214
    }

2215 2216 2217
    if (logfile != -1)
        close(logfile);

2218
    return ret;
2219 2220 2221 2222 2223 2224 2225

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);
    }
2226
    qemuRemoveCgroup(conn, driver, vm);
2227 2228 2229 2230
    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;
2231 2232
    if (logfile != -1)
        close(logfile);
2233 2234
    vm->def->id = -1;
    return -1;
2235 2236 2237
}


2238
static void qemudShutdownVMDaemon(virConnectPtr conn,
2239 2240
                                  struct qemud_driver *driver,
                                  virDomainObjPtr vm) {
D
Daniel P. Berrange 已提交
2241
    int ret;
2242
    int retries = 0;
D
Daniel P. Berrange 已提交
2243

2244
    if (!virDomainIsActive(vm))
2245
        return;
2246

2247
    VIR_DEBUG(_("Shutting down VM '%s'\n"), vm->def->name);
2248

G
Guido Günther 已提交
2249 2250
    if (virKillProcess(vm->pid, 0) == 0 &&
        virKillProcess(vm->pid, SIGTERM) < 0)
2251 2252 2253
        virReportSystemError(conn, errno,
                             _("Failed to send SIGTERM to %s (%d)"),
                             vm->def->name, vm->pid);
2254

2255 2256 2257
    if (vm->monitorWatch != -1) {
        virEventRemoveHandle(vm->monitorWatch);
        vm->monitorWatch = -1;
2258
    }
2259 2260 2261 2262 2263

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

2264 2265 2266 2267 2268 2269 2270
    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
2271 2272
    /* shut it off for sure */
    virKillProcess(vm->pid, SIGKILL);
2273

2274 2275 2276 2277
    /* Reset Security Labels */
    if (driver->securityDriver)
        driver->securityDriver->domainRestoreSecurityLabel(conn, vm);

2278 2279 2280 2281 2282 2283 2284
    /* 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);
    }

2285 2286 2287 2288
    if (qemuDomainSetAllDeviceOwnership(conn, driver, vm->def, 1) < 0)
        VIR_WARN("Failed to restore all device ownership for %s",
                 vm->def->name);

2289
    qemuDomainReAttachHostDevices(conn, driver, vm->def);
2290

2291 2292 2293 2294 2295 2296 2297 2298 2299 2300
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);
    }

2301
    if (qemudRemoveDomainStatus(conn, driver, vm) < 0) {
2302
        VIR_WARN(_("Failed to remove domain status for %s"),
2303 2304
                 vm->def->name);
    }
D
Daniel P. Berrange 已提交
2305 2306 2307 2308 2309 2310
    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));
    }

2311
    vm->pid = -1;
2312
    vm->def->id = -1;
2313
    vm->state = VIR_DOMAIN_SHUTOFF;
2314
    VIR_FREE(vm->vcpupids);
2315
    vm->nvcpupids = 0;
2316 2317

    if (vm->newDef) {
2318
        virDomainDefFree(vm->def);
2319
        vm->def = vm->newDef;
2320
        vm->def->id = -1;
2321 2322 2323 2324 2325
        vm->newDef = NULL;
    }
}


2326
static void
2327
qemudDispatchVMEvent(int watch, int fd, int events, void *opaque) {
2328
    struct qemud_driver *driver = opaque;
2329
    virDomainObjPtr vm = NULL;
2330
    virDomainEventPtr event = NULL;
2331
    unsigned int i;
2332
    int quit = 0, failed = 0;
2333

2334
    qemuDriverLock(driver);
2335
    for (i = 0 ; i < driver->domains.count ; i++) {
2336 2337 2338
        virDomainObjPtr tmpvm = driver->domains.objs[i];
        virDomainObjLock(tmpvm);
        if (virDomainIsActive(tmpvm) &&
2339
            tmpvm->monitorWatch == watch) {
2340
            vm = tmpvm;
2341
            break;
2342
        }
2343
        virDomainObjUnlock(tmpvm);
2344 2345 2346
    }

    if (!vm)
2347
        goto cleanup;
2348

2349
    if (vm->monitor != fd) {
2350 2351
        failed = 1;
    } else {
2352
        if (events & (VIR_EVENT_HANDLE_HANGUP | VIR_EVENT_HANDLE_ERROR))
2353
            quit = 1;
2354
        else {
2355 2356
            VIR_ERROR(_("unhandled fd event %d for %s"),
                      events, vm->def->name);
2357
            failed = 1;
2358
        }
2359 2360
    }

2361
    if (failed || quit) {
2362 2363 2364 2365 2366
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         quit ?
                                         VIR_DOMAIN_EVENT_STOPPED_SHUTDOWN :
                                         VIR_DOMAIN_EVENT_STOPPED_FAILED);
2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377
        qemudShutdownVMDaemon(NULL, driver, vm);
        if (!vm->persistent) {
            virDomainRemoveInactive(&driver->domains,
                                    vm);
            vm = NULL;
        }
    }

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
2378 2379
    if (event)
        qemuDomainEventQueue(driver, event);
2380
    qemuDriverUnlock(driver);
2381 2382
}

2383

2384

2385
static virDrvOpenStatus qemudOpen(virConnectPtr conn,
2386
                                  virConnectAuthPtr auth ATTRIBUTE_UNUSED,
2387
                                  int flags ATTRIBUTE_UNUSED) {
2388
    if (conn->uri == NULL) {
2389 2390 2391
        if (qemu_driver == NULL)
            return VIR_DRV_OPEN_DECLINED;

2392
        conn->uri = xmlParseURI(qemu_driver->privileged ?
2393 2394
                                "qemu:///system" :
                                "qemu:///session");
2395
        if (!conn->uri) {
2396
            virReportOOMError(conn);
2397 2398
            return VIR_DRV_OPEN_ERROR;
        }
2399 2400 2401 2402 2403 2404 2405 2406 2407 2408
    } 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;

2409 2410 2411 2412 2413 2414
        if (qemu_driver == NULL) {
            qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR, "%s",
                             _("qemu state driver is not active"));
            return VIR_DRV_OPEN_ERROR;
        }

2415
        if (qemu_driver->privileged) {
2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430
            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;
            }
        }
2431 2432 2433 2434 2435 2436 2437
    }
    conn->privateData = qemu_driver;

    return VIR_DRV_OPEN_SUCCESS;
}

static int qemudClose(virConnectPtr conn) {
2438
    struct qemud_driver *driver = conn->privateData;
2439 2440

    /* Get rid of callbacks registered for this conn */
2441
    qemuDriverLock(driver);
2442
    virDomainEventCallbackListRemoveConn(conn, driver->domainEventCallbacks);
2443
    qemuDriverUnlock(driver);
2444 2445 2446 2447 2448 2449

    conn->privateData = NULL;

    return 0;
}

D
Daniel Veillard 已提交
2450 2451 2452 2453 2454 2455 2456 2457 2458 2459
/* 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;
    }
}

2460
static const char *qemudGetType(virConnectPtr conn ATTRIBUTE_UNUSED) {
2461
    return "QEMU";
2462 2463
}

2464 2465 2466 2467 2468

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

    int r, fd;
2469

2470 2471
    fd = open(KVM_DEVICE, O_RDONLY);
    if (fd < 0) {
2472 2473
        virReportSystemError(NULL, errno, _("Unable to open %s"), KVM_DEVICE);
        return -1;
2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484
    }

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

    close(fd);
    return maxvcpus;
}


2485
static int qemudGetMaxVCPUs(virConnectPtr conn, const char *type) {
2486 2487 2488
    if (!type)
        return 16;

2489
    if (STRCASEEQ(type, "qemu"))
2490 2491
        return 16;

2492
    if (STRCASEEQ(type, "kvm"))
2493
        return kvmGetMaxVCPUs();
2494

2495
    if (STRCASEEQ(type, "kqemu"))
2496
        return 1;
2497 2498 2499

    qemudReportError(conn, NULL, NULL, VIR_ERR_INVALID_ARG,
                     _("unknown type '%s'"), type);
2500 2501 2502
    return -1;
}

2503

2504
static char *qemudGetCapabilities(virConnectPtr conn) {
2505
    struct qemud_driver *driver = conn->privateData;
2506
    virCapsPtr caps;
2507
    char *xml = NULL;
2508

2509
    qemuDriverLock(driver);
2510
    if ((caps = qemudCapsInit(qemu_driver->caps)) == NULL) {
2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521
        virReportOOMError(conn);
        goto cleanup;
    }

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

2522
    virCapabilitiesFree(qemu_driver->caps);
2523 2524 2525
    qemu_driver->caps = caps;

    if ((xml = virCapabilitiesFormatXML(driver->caps)) == NULL)
2526
        virReportOOMError(conn);
2527 2528

cleanup:
2529
    qemuDriverUnlock(driver);
2530

2531
    return xml;
2532 2533 2534
}


2535
static int qemudGetProcessInfo(unsigned long long *cpuTime, int *lastCpu, int pid, int tid) {
D
Daniel P. Berrange 已提交
2536 2537
    char proc[PATH_MAX];
    FILE *pidinfo;
2538
    unsigned long long usertime, systime;
2539 2540
    int cpu;
    int ret;
D
Daniel P. Berrange 已提交
2541

2542 2543 2544 2545 2546 2547
    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
2548 2549 2550 2551
        return -1;
    }

    if (!(pidinfo = fopen(proc, "r"))) {
2552
        /*printf("cannot read pid info");*/
D
Daniel P. Berrange 已提交
2553
        /* VM probably shut down, so fake 0 */
2554 2555 2556 2557
        if (cpuTime)
            *cpuTime = 0;
        if (lastCpu)
            *lastCpu = 0;
D
Daniel P. Berrange 已提交
2558 2559 2560
        return 0;
    }

2561 2562 2563 2564 2565 2566 2567 2568 2569 2570
    /* 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) {
2571
        fclose(pidinfo);
2572 2573
        VIR_WARN0("cannot parse process status data");
        errno = -EINVAL;
D
Daniel P. Berrange 已提交
2574 2575 2576 2577 2578 2579 2580 2581
        return -1;
    }

    /* We got jiffies
     * We want nanoseconds
     * _SC_CLK_TCK is jiffies per second
     * So calulate thus....
     */
2582 2583 2584 2585 2586
    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
2587

2588 2589
    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
2590 2591 2592 2593 2594 2595 2596

    fclose(pidinfo);

    return 0;
}


2597
static virDomainPtr qemudDomainLookupByID(virConnectPtr conn,
2598
                                          int id) {
2599 2600 2601 2602
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;

2603
    qemuDriverLock(driver);
2604
    vm  = virDomainFindByID(&driver->domains, id);
2605
    qemuDriverUnlock(driver);
2606 2607

    if (!vm) {
2608 2609
        qemudReportError(conn, NULL, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching id %d"), id);
2610
        goto cleanup;
2611 2612
    }

2613
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
2614
    if (dom) dom->id = vm->def->id;
2615 2616

cleanup:
2617 2618
    if (vm)
        virDomainObjUnlock(vm);
2619 2620
    return dom;
}
2621

2622
static virDomainPtr qemudDomainLookupByUUID(virConnectPtr conn,
2623
                                            const unsigned char *uuid) {
2624 2625 2626
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
2627

2628
    qemuDriverLock(driver);
2629
    vm = virDomainFindByUUID(&driver->domains, uuid);
2630 2631
    qemuDriverUnlock(driver);

2632
    if (!vm) {
2633 2634 2635
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(uuid, uuidstr);
        qemudReportError(conn, NULL, NULL, VIR_ERR_NO_DOMAIN,
2636
                         _("no domain with matching uuid '%s'"), uuidstr);
2637
        goto cleanup;
2638 2639
    }

2640
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
2641
    if (dom) dom->id = vm->def->id;
2642 2643

cleanup:
2644 2645
    if (vm)
        virDomainObjUnlock(vm);
2646 2647
    return dom;
}
2648

2649
static virDomainPtr qemudDomainLookupByName(virConnectPtr conn,
2650
                                            const char *name) {
2651 2652 2653
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
2654

2655
    qemuDriverLock(driver);
2656
    vm = virDomainFindByName(&driver->domains, name);
2657 2658
    qemuDriverUnlock(driver);

2659
    if (!vm) {
2660 2661
        qemudReportError(conn, NULL, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching name '%s'"), name);
2662
        goto cleanup;
2663 2664
    }

2665
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
2666
    if (dom) dom->id = vm->def->id;
2667 2668

cleanup:
2669 2670
    if (vm)
        virDomainObjUnlock(vm);
2671 2672 2673
    return dom;
}

2674
static int qemudGetVersion(virConnectPtr conn, unsigned long *version) {
2675 2676 2677
    struct qemud_driver *driver = conn->privateData;
    int ret = -1;

2678
    qemuDriverLock(driver);
2679
    if (qemudExtractVersion(conn, driver) < 0)
2680
        goto cleanup;
2681

2682
    *version = qemu_driver->qemuVersion;
2683 2684 2685
    ret = 0;

cleanup:
2686
    qemuDriverUnlock(driver);
2687
    return ret;
D
Daniel P. Berrange 已提交
2688 2689
}

2690 2691 2692
static char *
qemudGetHostname (virConnectPtr conn)
{
2693
    char *result;
2694

2695 2696
    result = virGetHostname();
    if (result == NULL) {
2697 2698
        virReportSystemError (conn, errno,
                              "%s", _("failed to determine host name"));
2699 2700 2701
        return NULL;
    }
    /* Caller frees this string. */
2702
    return result;
2703 2704
}

2705
static int qemudListDomains(virConnectPtr conn, int *ids, int nids) {
2706
    struct qemud_driver *driver = conn->privateData;
2707 2708
    int got = 0, i;

2709 2710 2711
    qemuDriverLock(driver);
    for (i = 0 ; i < driver->domains.count && got < nids ; i++) {
        virDomainObjLock(driver->domains.objs[i]);
2712 2713
        if (virDomainIsActive(driver->domains.objs[i]))
            ids[got++] = driver->domains.objs[i]->def->id;
2714 2715 2716
        virDomainObjUnlock(driver->domains.objs[i]);
    }
    qemuDriverUnlock(driver);
2717

D
Daniel P. Berrange 已提交
2718 2719
    return got;
}
2720

2721
static int qemudNumDomains(virConnectPtr conn) {
2722
    struct qemud_driver *driver = conn->privateData;
2723 2724
    int n = 0, i;

2725 2726 2727
    qemuDriverLock(driver);
    for (i = 0 ; i < driver->domains.count ; i++) {
        virDomainObjLock(driver->domains.objs[i]);
2728
        if (virDomainIsActive(driver->domains.objs[i]))
2729
            n++;
2730 2731 2732
        virDomainObjUnlock(driver->domains.objs[i]);
    }
    qemuDriverUnlock(driver);
2733

2734
    return n;
D
Daniel P. Berrange 已提交
2735
}
2736

2737
static virDomainPtr qemudDomainCreate(virConnectPtr conn, const char *xml,
2738
                                      unsigned int flags ATTRIBUTE_UNUSED) {
2739
    struct qemud_driver *driver = conn->privateData;
2740
    virDomainDefPtr def;
2741
    virDomainObjPtr vm = NULL;
2742
    virDomainPtr dom = NULL;
2743
    virDomainEventPtr event = NULL;
D
Daniel P. Berrange 已提交
2744

2745
    qemuDriverLock(driver);
2746 2747
    if (!(def = virDomainDefParseString(conn, driver->caps, xml,
                                        VIR_DOMAIN_XML_INACTIVE)))
2748
        goto cleanup;
2749

2750 2751 2752
    if (virSecurityDriverVerify(conn, def) < 0)
        goto cleanup;

2753
    /* See if a VM with matching UUID already exists */
2754
    vm = virDomainFindByUUID(&driver->domains, def->uuid);
2755
    if (vm) {
2756 2757 2758 2759 2760 2761 2762 2763 2764
        /* 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;
        }
2765

2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783
        /* 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;
        }
2784
    }
2785

2786 2787
    if (!(vm = virDomainAssignDef(conn,
                                  &driver->domains,
2788 2789 2790 2791
                                  def)))
        goto cleanup;

    def = NULL;
D
Daniel P. Berrange 已提交
2792

2793
    if (qemudStartVMDaemon(conn, driver, vm, NULL, -1) < 0) {
2794 2795
        virDomainRemoveInactive(&driver->domains,
                                vm);
2796
        vm = NULL;
2797
        goto cleanup;
D
Daniel P. Berrange 已提交
2798
    }
2799 2800 2801 2802

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_BOOTED);
D
Daniel P. Berrange 已提交
2803

2804
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
2805
    if (dom) dom->id = vm->def->id;
2806 2807 2808

cleanup:
    virDomainDefFree(def);
2809 2810
    if (vm)
        virDomainObjUnlock(vm);
2811 2812
    if (event)
        qemuDomainEventQueue(driver, event);
2813
    qemuDriverUnlock(driver);
2814
    return dom;
D
Daniel P. Berrange 已提交
2815 2816 2817
}


2818
static int qemudDomainSuspend(virDomainPtr dom) {
2819
    struct qemud_driver *driver = dom->conn->privateData;
D
Daniel P. Berrange 已提交
2820
    char *info;
2821 2822
    virDomainObjPtr vm;
    int ret = -1;
2823
    virDomainEventPtr event = NULL;
2824

2825
    qemuDriverLock(driver);
2826
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
2827

D
Daniel P. Berrange 已提交
2828
    if (!vm) {
2829 2830 2831 2832
        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);
2833
        goto cleanup;
D
Daniel P. Berrange 已提交
2834
    }
2835
    if (!virDomainIsActive(vm)) {
2836
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
2837
                         "%s", _("domain is not running"));
2838
        goto cleanup;
D
Daniel P. Berrange 已提交
2839
    }
2840
    if (vm->state != VIR_DOMAIN_PAUSED) {
2841
        if (qemudMonitorCommand(vm, "stop", &info) < 0) {
2842 2843 2844 2845 2846 2847
            qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                             "%s", _("suspend operation failed"));
            goto cleanup;
        }
        vm->state = VIR_DOMAIN_PAUSED;
        qemudDebug("Reply %s", info);
2848 2849 2850
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_SUSPENDED,
                                         VIR_DOMAIN_EVENT_SUSPENDED_PAUSED);
2851
        VIR_FREE(info);
D
Daniel P. Berrange 已提交
2852
    }
2853
    if (virDomainSaveStatus(dom->conn, driver->stateDir, vm) < 0)
2854
        goto cleanup;
2855 2856 2857
    ret = 0;

cleanup:
2858 2859
    if (vm)
        virDomainObjUnlock(vm);
2860

2861
    if (event)
2862
        qemuDomainEventQueue(driver, event);
2863
    qemuDriverUnlock(driver);
2864
    return ret;
D
Daniel P. Berrange 已提交
2865 2866 2867
}


2868
static int qemudDomainResume(virDomainPtr dom) {
2869 2870 2871
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
2872
    virDomainEventPtr event = NULL;
2873

2874
    qemuDriverLock(driver);
2875
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
2876

D
Daniel P. Berrange 已提交
2877
    if (!vm) {
2878 2879 2880 2881
        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);
2882
        goto cleanup;
D
Daniel P. Berrange 已提交
2883
    }
2884
    if (!virDomainIsActive(vm)) {
2885
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
2886
                         "%s", _("domain is not running"));
2887
        goto cleanup;
D
Daniel P. Berrange 已提交
2888
    }
2889
    if (vm->state == VIR_DOMAIN_PAUSED) {
2890 2891 2892 2893
        if (qemudMonitorSendCont(dom->conn, vm) < 0) {
            if (virGetLastError() == NULL)
                qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                                 "%s", _("resume operation failed"));
2894 2895 2896
            goto cleanup;
        }
        vm->state = VIR_DOMAIN_RUNNING;
2897 2898 2899
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_UNPAUSED);
D
Daniel P. Berrange 已提交
2900
    }
2901
    if (virDomainSaveStatus(dom->conn, driver->stateDir, vm) < 0)
2902
        goto cleanup;
2903 2904 2905
    ret = 0;

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


2915
static int qemudDomainShutdown(virDomainPtr dom) {
2916 2917
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
2918
    char* info;
2919
    int ret = -1;
2920

2921
    qemuDriverLock(driver);
2922
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
2923 2924
    qemuDriverUnlock(driver);

2925
    if (!vm) {
2926 2927 2928 2929
        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);
2930
        goto cleanup;
2931 2932
    }

2933 2934 2935 2936 2937 2938
    if (!virDomainIsActive(vm)) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
                         "%s", _("domain is not running"));
        goto cleanup;
    }

2939
    if (qemudMonitorCommand(vm, "system_powerdown", &info) < 0) {
2940
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
2941
                         "%s", _("shutdown operation failed"));
2942
        goto cleanup;
2943
    }
D
Daniel Veillard 已提交
2944
    VIR_FREE(info);
2945 2946 2947
    ret = 0;

cleanup:
2948 2949
    if (vm)
        virDomainObjUnlock(vm);
2950
    return ret;
2951 2952 2953
}


2954
static int qemudDomainDestroy(virDomainPtr dom) {
2955 2956 2957
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
2958
    virDomainEventPtr event = NULL;
2959

2960
    qemuDriverLock(driver);
2961
    vm  = virDomainFindByUUID(&driver->domains, dom->uuid);
D
Daniel P. Berrange 已提交
2962
    if (!vm) {
2963 2964 2965 2966
        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);
2967
        goto cleanup;
D
Daniel P. Berrange 已提交
2968
    }
2969
    if (!virDomainIsActive(vm)) {
2970
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
2971 2972 2973
                         "%s", _("domain is not running"));
        goto cleanup;
    }
2974

2975
    qemudShutdownVMDaemon(dom->conn, driver, vm);
2976 2977 2978
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_DESTROYED);
2979
    if (!vm->persistent) {
2980 2981
        virDomainRemoveInactive(&driver->domains,
                                vm);
2982 2983
        vm = NULL;
    }
2984 2985 2986
    ret = 0;

cleanup:
2987 2988
    if (vm)
        virDomainObjUnlock(vm);
2989 2990
    if (event)
        qemuDomainEventQueue(driver, event);
2991
    qemuDriverUnlock(driver);
2992
    return ret;
D
Daniel P. Berrange 已提交
2993 2994 2995
}


2996
static char *qemudDomainGetOSType(virDomainPtr dom) {
2997 2998 2999
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    char *type = NULL;
3000

3001
    qemuDriverLock(driver);
3002
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3003
    qemuDriverUnlock(driver);
3004
    if (!vm) {
3005 3006 3007 3008
        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);
3009
        goto cleanup;
3010 3011
    }

3012
    if (!(type = strdup(vm->def->os.type)))
3013
        virReportOOMError(dom->conn);
3014 3015

cleanup:
3016 3017
    if (vm)
        virDomainObjUnlock(vm);
3018 3019 3020
    return type;
}

3021 3022
/* Returns max memory in kb, 0 if error */
static unsigned long qemudDomainGetMaxMemory(virDomainPtr dom) {
3023 3024 3025
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    unsigned long ret = 0;
3026

3027
    qemuDriverLock(driver);
3028
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3029 3030
    qemuDriverUnlock(driver);

3031
    if (!vm) {
3032 3033
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
3034
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
3035
                         _("no domain with matching uuid '%s'"), uuidstr);
3036
        goto cleanup;
3037 3038
    }

3039 3040 3041
    ret = vm->def->maxmem;

cleanup:
3042 3043
    if (vm)
        virDomainObjUnlock(vm);
3044
    return ret;
3045 3046 3047
}

static int qemudDomainSetMaxMemory(virDomainPtr dom, unsigned long newmax) {
3048 3049 3050
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
3051

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

3056
    if (!vm) {
3057 3058
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
3059
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
3060
                         _("no domain with matching uuid '%s'"), uuidstr);
3061
        goto cleanup;
3062 3063 3064 3065
    }

    if (newmax < vm->def->memory) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
3066
                         "%s", _("cannot set max memory lower than current memory"));
3067
        goto cleanup;;
3068 3069 3070
    }

    vm->def->maxmem = newmax;
3071 3072 3073
    ret = 0;

cleanup:
3074 3075
    if (vm)
        virDomainObjUnlock(vm);
3076
    return ret;
3077 3078
}

3079 3080 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093

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

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

3094 3095 3096
    if (!virDomainIsActive(vm))
        return 0;

3097 3098 3099 3100 3101 3102
    if (qemudMonitorCommand(vm, "info balloon", &reply) < 0) {
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                         "%s", _("could not query memory balloon allocation"));
        goto cleanup;
    }

3103
    DEBUG ("%s: balloon reply: '%s'", vm->def->name, reply);
3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156
    if ((offset = strstr(reply, BALLOON_PREFIX)) != NULL) {
        unsigned int memMB;
        char *end;
        offset += strlen(BALLOON_PREFIX);
        if (virStrToLong_ui(offset, &end, 10, &memMB) < 0) {
            qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                             "%s", _("could not parse memory balloon allocation"));
            goto cleanup;
        }
        *currmem = memMB * 1024;
        ret = 1;
    } else {
        /* We don't raise an error here, since its to be expected that
         * many QEMU's don't support ballooning
         */
        ret = 0;
    }

cleanup:
    VIR_FREE(reply);
    return ret;
}

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

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

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

    /* If the command failed qemu prints: 'unknown command'
     * No message is printed on success it seems */
3157
    DEBUG ("%s: balloon reply: %s",vm->def->name,  reply);
3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170
    if (strstr(reply, "\nunknown command:")) {
        /* Don't set error - it is expected memory balloon fails on many qemu */
        ret = 0;
    } else {
        ret = 1;
    }

cleanup:
    VIR_FREE(reply);
    return ret;
}


3171
static int qemudDomainSetMemory(virDomainPtr dom, unsigned long newmem) {
3172 3173 3174
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
3175

3176
    qemuDriverLock(driver);
3177
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3178
    qemuDriverUnlock(driver);
3179
    if (!vm) {
3180 3181
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
3182
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
3183
                         _("no domain with matching uuid '%s'"), uuidstr);
3184
        goto cleanup;
3185 3186 3187 3188
    }

    if (newmem > vm->def->maxmem) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
3189
                         "%s", _("cannot set memory higher than max memory"));
3190
        goto cleanup;
3191 3192
    }

3193 3194 3195 3196 3197 3198 3199 3200 3201
    if (virDomainIsActive(vm)) {
        ret = qemudDomainSetMemoryBalloon(dom->conn, vm, newmem);
        if (ret == 0)
            qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                             "%s", _("cannot set memory of an active domain"));
    } else {
        vm->def->memory = newmem;
        ret = 0;
    }
3202 3203

cleanup:
3204 3205
    if (vm)
        virDomainObjUnlock(vm);
3206
    return ret;
3207 3208
}

3209
static int qemudDomainGetInfo(virDomainPtr dom,
3210
                              virDomainInfoPtr info) {
3211 3212 3213
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
3214 3215
    int err;
    unsigned long balloon;
3216

3217
    qemuDriverLock(driver);
3218
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3219
    qemuDriverUnlock(driver);
D
Daniel P. Berrange 已提交
3220
    if (!vm) {
3221 3222 3223 3224
        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);
3225
        goto cleanup;
D
Daniel P. Berrange 已提交
3226 3227
    }

3228
    info->state = vm->state;
D
Daniel P. Berrange 已提交
3229

3230
    if (!virDomainIsActive(vm)) {
3231
        info->cpuTime = 0;
D
Daniel P. Berrange 已提交
3232
    } else {
3233
        if (qemudGetProcessInfo(&(info->cpuTime), NULL, vm->pid, 0) < 0) {
3234
            qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED, ("cannot read cputime for domain"));
3235
            goto cleanup;
D
Daniel P. Berrange 已提交
3236 3237 3238
        }
    }

3239
    info->maxMem = vm->def->maxmem;
3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254

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

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

3255
    info->nrVirtCpu = vm->def->vcpus;
3256 3257 3258
    ret = 0;

cleanup:
3259 3260
    if (vm)
        virDomainObjUnlock(vm);
3261
    return ret;
D
Daniel P. Berrange 已提交
3262 3263 3264
}


D
Daniel P. Berrange 已提交
3265
static char *qemudEscape(const char *in, int shell)
3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286
{
    int len = 0;
    int i, j;
    char *out;

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

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

    for (i = 0; in[i] != '\0'; i++) {
        switch(in[i]) {
        case '\r':
        case '\n':
        case '"':
        case '\\':
            len += 2;
            break;
        case '\'':
D
Daniel P. Berrange 已提交
3287 3288 3289 3290
            if (shell)
                len += 5;
            else
                len += 1;
3291 3292 3293 3294 3295 3296 3297
            break;
        default:
            len += 1;
            break;
        }
    }

3298
    if (VIR_ALLOC_N(out, len + 1) < 0)
3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316
        return NULL;

    for (i = j = 0; in[i] != '\0'; i++) {
        switch(in[i]) {
        case '\r':
            out[j++] = '\\';
            out[j++] = 'r';
            break;
        case '\n':
            out[j++] = '\\';
            out[j++] = 'n';
            break;
        case '"':
        case '\\':
            out[j++] = '\\';
            out[j++] = in[i];
            break;
        case '\'':
D
Daniel P. Berrange 已提交
3317 3318 3319 3320 3321 3322 3323 3324 3325
            if (shell) {
                out[j++] = '\'';
                out[j++] = '\\';
                out[j++] = '\\';
                out[j++] = '\'';
                out[j++] = '\'';
            } else {
                out[j++] = in[i];
            }
3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336
            break;
        default:
            out[j++] = in[i];
            break;
        }
    }
    out[j] = '\0';

    return out;
}

3337 3338 3339 3340 3341
static char *qemudEscapeMonitorArg(const char *in)
{
    return qemudEscape(in, 0);
}

D
Daniel P. Berrange 已提交
3342 3343 3344 3345
static char *qemudEscapeShellArg(const char *in)
{
    return qemudEscape(in, 1);
}
3346

3347
#define QEMUD_SAVE_MAGIC "LibvirtQemudSave"
3348 3349 3350
#define QEMUD_SAVE_VERSION 2

enum qemud_save_formats {
3351 3352 3353
    QEMUD_SAVE_FORMAT_RAW = 0,
    QEMUD_SAVE_FORMAT_GZIP = 1,
    QEMUD_SAVE_FORMAT_BZIP2 = 2,
3354 3355
    /*
     * Deprecated by xz and never used as part of a release
3356
     * QEMUD_SAVE_FORMAT_LZMA
3357 3358
     */
    QEMUD_SAVE_FORMAT_XZ = 3,
3359
    QEMUD_SAVE_FORMAT_LZOP = 4,
3360 3361 3362
    /* Note: add new members only at the end.
       These values are used in the on-disk format.
       Do not change or re-use numbers. */
3363 3364

    QEMUD_SAVE_FORMAT_LAST
3365
};
3366

3367 3368 3369 3370 3371
VIR_ENUM_DECL(qemudSaveCompression)
VIR_ENUM_IMPL(qemudSaveCompression, QEMUD_SAVE_FORMAT_LAST,
              "raw",
              "gzip",
              "bzip2",
3372 3373
              "xz",
              "lzop")
3374

3375 3376 3377 3378 3379
struct qemud_save_header {
    char magic[sizeof(QEMUD_SAVE_MAGIC)-1];
    int version;
    int xml_len;
    int was_running;
3380 3381
    int compressed;
    int unused[15];
3382 3383
};

3384
static int qemudDomainSave(virDomainPtr dom,
3385 3386
                           const char *path)
{
3387
    struct qemud_driver *driver = dom->conn->privateData;
3388
    virDomainObjPtr vm = NULL;
3389 3390 3391 3392 3393
    char *command = NULL;
    char *info = NULL;
    int fd = -1;
    char *safe_path = NULL;
    char *xml = NULL;
3394
    struct qemud_save_header header;
3395
    int ret = -1;
3396
    virDomainEventPtr event = NULL;
3397
    int internalret;
3398 3399 3400 3401 3402

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

3403
    qemuDriverLock(driver);
3404 3405 3406
    if (driver->saveImageFormat == NULL)
        header.compressed = QEMUD_SAVE_FORMAT_RAW;
    else {
3407 3408 3409 3410 3411 3412
        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"));
3413
            goto cleanup;
3414
        }
3415 3416
    }

3417
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3418

D
Daniel P. Berrange 已提交
3419
    if (!vm) {
3420 3421 3422 3423
        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);
3424
        goto cleanup;
D
Daniel P. Berrange 已提交
3425
    }
3426

3427
    if (!virDomainIsActive(vm)) {
3428
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
3429
                         "%s", _("domain is not running"));
3430
        goto cleanup;
D
Daniel P. Berrange 已提交
3431
    }
3432 3433 3434 3435

    /* Pause */
    if (vm->state == VIR_DOMAIN_RUNNING) {
        header.was_running = 1;
3436
        if (qemudMonitorCommand(vm, "stop", &info) < 0) {
3437
            qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
3438
                             "%s", _("suspend operation failed"));
3439
            goto cleanup;
3440
        }
3441 3442 3443
        vm->state = VIR_DOMAIN_PAUSED;
        qemudDebug("Reply %s", info);
        VIR_FREE(info);
3444 3445 3446
    }

    /* Get XML for the domain */
3447
    xml = virDomainDefFormat(dom->conn, vm->def, VIR_DOMAIN_XML_SECURE);
3448 3449
    if (!xml) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
3450
                         "%s", _("failed to get domain xml"));
3451
        goto cleanup;
3452 3453 3454 3455 3456 3457
    }
    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,
3458
                         _("failed to create '%s'"), path);
3459
        goto cleanup;
3460 3461 3462 3463
    }

    if (safewrite(fd, &header, sizeof(header)) != sizeof(header)) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
3464
                         "%s", _("failed to write save header"));
3465
        goto cleanup;
3466 3467 3468 3469
    }

    if (safewrite(fd, xml, header.xml_len) != header.xml_len) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
3470
                         "%s", _("failed to write xml"));
3471
        goto cleanup;
3472 3473
    }

3474
    if (close(fd) < 0) {
3475 3476 3477
        virReportSystemError(dom->conn, errno,
                             _("unable to save file %s"),
                             path);
3478 3479 3480
        goto cleanup;
    }
    fd = -1;
3481 3482 3483 3484

    /* Migrate to file */
    safe_path = qemudEscapeShellArg(path);
    if (!safe_path) {
3485
        virReportOOMError(dom->conn);
3486
        goto cleanup;
3487
    }
3488

3489 3490 3491 3492 3493 3494 3495 3496 3497
    {
        const char *prog = qemudSaveCompressionTypeToString(header.compressed);
        const char *args;

        if (prog == NULL) {
            qemudReportError(dom->conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
                             _("Invalid compress format %d"), header.compressed);
            goto cleanup;
        }
3498

3499 3500 3501 3502 3503 3504 3505 3506 3507 3508
        if (STREQ (prog, "raw")) {
            prog = "cat";
            args = "";
        } else {
            args = "-c";
        }
        internalret = virAsprintf(&command, "migrate \"exec:"
                                  "%s %s >> '%s' 2>/dev/null\"", prog, args,
                                  safe_path);
    }
3509

3510
    if (internalret < 0) {
3511
        virReportOOMError(dom->conn);
3512
        goto cleanup;
3513 3514
    }

3515
    if (qemudMonitorCommand(vm, command, &info) < 0) {
3516
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
3517
                         "%s", _("migrate operation failed"));
3518
        goto cleanup;
3519 3520
    }

3521
    DEBUG ("%s: migrate reply: %s", vm->def->name, info);
3522 3523 3524 3525 3526 3527 3528

    /* If the command isn't supported then qemu prints:
     * unknown command: migrate" */
    if (strstr(info, "unknown command:")) {
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                          "%s",
                          _("'migrate' not supported by this qemu"));
3529
        goto cleanup;
3530 3531
    }

3532 3533
    /* Shut it down */
    qemudShutdownVMDaemon(dom->conn, driver, vm);
3534 3535 3536
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_SAVED);
3537
    if (!vm->persistent) {
3538 3539
        virDomainRemoveInactive(&driver->domains,
                                vm);
3540 3541
        vm = NULL;
    }
3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552
    ret = 0;

cleanup:
    if (fd != -1)
        close(fd);
    VIR_FREE(xml);
    VIR_FREE(safe_path);
    VIR_FREE(command);
    VIR_FREE(info);
    if (ret != 0)
        unlink(path);
3553 3554
    if (vm)
        virDomainObjUnlock(vm);
3555 3556
    if (event)
        qemuDomainEventQueue(driver, event);
3557
    qemuDriverUnlock(driver);
3558
    return ret;
D
Daniel P. Berrange 已提交
3559 3560 3561
}


P
Paolo Bonzini 已提交
3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 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 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649
static int qemudDomainCoreDump(virDomainPtr dom,
                               const char *path,
                               int flags ATTRIBUTE_UNUSED) {
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    char *command = NULL;
    char *info = NULL;
    char *safe_path = NULL;
    int resume = 0, paused = 0;
    int ret = -1;

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

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

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

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

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

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

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

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

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

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

    /* Since the monitor is always attached to a pty for libvirt, it
       will support synchronous operations so we always get here after
       the migration is complete.  */
    if (resume && paused) {
3650 3651 3652 3653
        if (qemudMonitorSendCont(dom->conn, vm) < 0) {
            if (virGetLastError() == NULL)
                qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                                 "%s", _("resuming after dump failed"));
P
Paolo Bonzini 已提交
3654 3655 3656 3657 3658 3659 3660 3661
        }
    }
    if (vm)
        virDomainObjUnlock(vm);
    return ret;
}


3662
static int qemudDomainSetVcpus(virDomainPtr dom, unsigned int nvcpus) {
3663 3664
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
3665
    int max;
3666
    int ret = -1;
3667
    const char *type;
3668

3669
    qemuDriverLock(driver);
3670
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3671 3672
    qemuDriverUnlock(driver);

3673
    if (!vm) {
3674 3675
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
3676
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
3677
                         _("no domain with matching uuid '%s'"), uuidstr);
3678
        goto cleanup;
3679 3680
    }

3681
    if (virDomainIsActive(vm)) {
3682
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID, "%s",
3683
                         _("cannot change vcpu count of an active domain"));
3684
        goto cleanup;
3685 3686
    }

3687 3688 3689 3690 3691 3692 3693 3694
    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) {
3695 3696
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INTERNAL_ERROR, "%s",
                         _("could not determine max vcpus for the domain"));
3697
        goto cleanup;
3698 3699 3700 3701 3702 3703
    }

    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);
3704
        goto cleanup;
3705 3706 3707
    }

    vm->def->vcpus = nvcpus;
3708 3709 3710
    ret = 0;

cleanup:
3711 3712
    if (vm)
        virDomainObjUnlock(vm);
3713
    return ret;
3714 3715
}

3716 3717 3718 3719 3720 3721 3722

#if HAVE_SCHED_GETAFFINITY
static int
qemudDomainPinVcpu(virDomainPtr dom,
                   unsigned int vcpu,
                   unsigned char *cpumap,
                   int maplen) {
3723 3724
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
3725 3726 3727
    cpu_set_t mask;
    int i, maxcpu;
    virNodeInfo nodeinfo;
3728
    int ret = -1;
3729

3730
    qemuDriverLock(driver);
3731
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3732 3733
    qemuDriverUnlock(driver);

3734 3735 3736 3737 3738 3739 3740 3741
    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;
    }

3742
    if (!virDomainIsActive(vm)) {
3743
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
3744
                         "%s",_("cannot pin vcpus on an inactive domain"));
3745
        goto cleanup;
3746 3747 3748 3749 3750 3751
    }

    if (vcpu > (vm->nvcpupids-1)) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
                         _("vcpu number out of range %d > %d"),
                         vcpu, vm->nvcpupids);
3752
        goto cleanup;
3753 3754
    }

3755
    if (nodeGetInfo(dom->conn, &nodeinfo) < 0)
3756
        goto cleanup;
3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769

    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) {
3770 3771
            virReportSystemError(dom->conn, errno, "%s",
                                 _("cannot set affinity"));
3772
            goto cleanup;
3773 3774 3775 3776
        }
    } else {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                         "%s", _("cpu affinity is not supported"));
3777
        goto cleanup;
3778
    }
3779
    ret = 0;
3780

3781
cleanup:
3782 3783
    if (vm)
        virDomainObjUnlock(vm);
3784
    return ret;
3785 3786 3787 3788 3789 3790 3791 3792
}

static int
qemudDomainGetVcpus(virDomainPtr dom,
                    virVcpuInfoPtr info,
                    int maxinfo,
                    unsigned char *cpumaps,
                    int maplen) {
3793 3794
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
3795 3796
    virNodeInfo nodeinfo;
    int i, v, maxcpu;
3797
    int ret = -1;
3798

3799
    qemuDriverLock(driver);
3800
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3801 3802
    qemuDriverUnlock(driver);

3803 3804 3805 3806 3807 3808 3809 3810
    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;
    }

3811
    if (!virDomainIsActive(vm)) {
3812
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
3813
                         "%s",_("cannot pin vcpus on an inactive domain"));
3814
        goto cleanup;
3815 3816
    }

3817
    if (nodeGetInfo(dom->conn, &nodeinfo) < 0)
3818
        goto cleanup;
3819 3820 3821 3822 3823 3824 3825 3826 3827

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

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

3828 3829 3830 3831 3832 3833
    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;
3834 3835 3836 3837 3838 3839 3840 3841 3842 3843

                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;
                }
3844
            }
3845 3846
        }

3847 3848 3849 3850 3851 3852 3853 3854 3855
        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) {
3856 3857
                        virReportSystemError(dom->conn, errno, "%s",
                                             _("cannot get affinity"));
3858 3859 3860 3861 3862 3863
                        goto cleanup;
                    }

                    for (i = 0 ; i < maxcpu ; i++)
                        if (CPU_ISSET(i, &mask))
                            VIR_USE_CPU(cpumap, i);
3864
                }
3865 3866 3867 3868
            } else {
                qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                                 "%s", _("cpu affinity is not available"));
                goto cleanup;
3869 3870 3871
            }
        }
    }
3872
    ret = maxinfo;
3873

3874
cleanup:
3875 3876
    if (vm)
        virDomainObjUnlock(vm);
3877
    return ret;
3878 3879 3880 3881
}
#endif /* HAVE_SCHED_GETAFFINITY */


3882
static int qemudDomainGetMaxVcpus(virDomainPtr dom) {
3883 3884
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
3885
    const char *type;
3886
    int ret = -1;
3887

3888
    qemuDriverLock(driver);
3889
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3890 3891
    qemuDriverUnlock(driver);

3892
    if (!vm) {
3893 3894
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
3895
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
3896
                         _("no domain with matching uuid '%s'"), uuidstr);
3897
        goto cleanup;
3898 3899
    }

3900
    if (!(type = virDomainVirtTypeToString(vm->def->virtType))) {
3901 3902 3903
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
                         _("unknown virt type in domain definition '%d'"),
                         vm->def->virtType);
3904
        goto cleanup;
3905 3906
    }

3907
    ret = qemudGetMaxVCPUs(dom->conn, type);
3908

3909
cleanup:
3910 3911
    if (vm)
        virDomainObjUnlock(vm);
3912 3913 3914
    return ret;
}

3915 3916 3917 3918 3919 3920 3921 3922 3923 3924
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);

3925 3926
    memset(seclabel, 0, sizeof(*seclabel));

3927 3928 3929
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
3930
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959
                         _("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,
3960
                                 "%s", _("Failed to get security label"));
3961 3962 3963 3964 3965 3966 3967 3968 3969 3970
                goto cleanup;
            }
        }
    }

    ret = 0;

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
3971
    qemuDriverUnlock(driver);
3972 3973 3974
    return ret;
}

3975 3976
static int qemudNodeGetSecurityModel(virConnectPtr conn,
                                     virSecurityModelPtr secmodel)
3977 3978 3979
{
    struct qemud_driver *driver = (struct qemud_driver *)conn->privateData;
    char *p;
3980
    int ret = 0;
3981

3982 3983
    qemuDriverLock(driver);
    if (!driver->securityDriver) {
3984
        memset(secmodel, 0, sizeof (*secmodel));
3985 3986
        goto cleanup;
    }
3987

3988 3989 3990 3991 3992
    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);
3993 3994
        ret = -1;
        goto cleanup;
3995 3996 3997 3998 3999 4000 4001 4002
    }
    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);
4003 4004
        ret = -1;
        goto cleanup;
4005 4006
    }
    strcpy(secmodel->doi, p);
4007 4008 4009 4010

cleanup:
    qemuDriverUnlock(driver);
    return ret;
4011 4012 4013
}

/* TODO: check seclabel restore */
4014
static int qemudDomainRestore(virConnectPtr conn,
4015 4016 4017
                              const char *path) {
    struct qemud_driver *driver = conn->privateData;
    virDomainDefPtr def = NULL;
4018
    virDomainObjPtr vm = NULL;
4019 4020 4021
    int fd = -1;
    int ret = -1;
    char *xml = NULL;
4022
    struct qemud_save_header header;
4023
    virDomainEventPtr event = NULL;
4024 4025 4026
    int intermediatefd = -1;
    pid_t intermediate_pid = -1;
    int childstat;
4027

4028
    qemuDriverLock(driver);
4029 4030 4031
    /* Verify the header and read the XML */
    if ((fd = open(path, O_RDONLY)) < 0) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
4032
                         "%s", _("cannot read domain image"));
4033
        goto cleanup;
4034 4035 4036 4037
    }

    if (saferead(fd, &header, sizeof(header)) != sizeof(header)) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
4038
                         "%s", _("failed to read qemu header"));
4039
        goto cleanup;
4040 4041 4042 4043
    }

    if (memcmp(header.magic, QEMUD_SAVE_MAGIC, sizeof(header.magic)) != 0) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
4044
                         "%s", _("image magic is incorrect"));
4045
        goto cleanup;
4046 4047 4048 4049
    }

    if (header.version > QEMUD_SAVE_VERSION) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
4050
                         _("image version is not supported (%d > %d)"),
4051
                         header.version, QEMUD_SAVE_VERSION);
4052
        goto cleanup;
4053 4054
    }

4055
    if (VIR_ALLOC_N(xml, header.xml_len) < 0) {
4056
        virReportOOMError(conn);
4057
        goto cleanup;
4058 4059 4060 4061
    }

    if (saferead(fd, xml, header.xml_len) != header.xml_len) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
4062
                         "%s", _("failed to read XML"));
4063
        goto cleanup;
4064 4065 4066
    }

    /* Create a domain from this XML */
4067 4068
    if (!(def = virDomainDefParseString(conn, driver->caps, xml,
                                        VIR_DOMAIN_XML_INACTIVE))) {
4069
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
4070
                         "%s", _("failed to parse XML"));
4071
        goto cleanup;
4072 4073
    }

4074
    /* See if a VM with matching UUID already exists */
4075
    vm = virDomainFindByUUID(&driver->domains, def->uuid);
4076
    if (vm) {
4077 4078 4079 4080 4081 4082 4083 4084 4085 4086 4087
        /* 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 */
4088
        if (virDomainIsActive(vm)) {
4089
            qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_INVALID,
4090 4091
                             _("domain is already active as '%s'"), vm->def->name);
            goto cleanup;
4092 4093 4094 4095 4096 4097 4098 4099 4100 4101 4102 4103
        }
        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;
4104
        }
4105 4106
    }

4107 4108 4109
    if (!(vm = virDomainAssignDef(conn,
                                  &driver->domains,
                                  def))) {
4110
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
4111
                         "%s", _("failed to assign new VM"));
4112
        goto cleanup;
4113
    }
4114
    def = NULL;
4115

4116 4117
    if (header.version == 2) {
        const char *intermediate_argv[3] = { NULL, "-dc", NULL };
4118 4119
        const char *prog = qemudSaveCompressionTypeToString(header.compressed);
        if (prog == NULL) {
4120
            qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
4121
                             _("Invalid compressed save format %d"),
4122 4123 4124
                             header.compressed);
            goto cleanup;
        }
4125 4126 4127 4128

        if (header.compressed != QEMUD_SAVE_FORMAT_RAW)
            intermediate_argv[0] = prog;
        else {
4129 4130 4131 4132 4133 4134 4135 4136 4137 4138 4139
            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;
            }
        }
    }
4140
    /* Set the migration source and start it up. */
4141
    ret = qemudStartVMDaemon(conn, driver, vm, "stdio", fd);
4142 4143 4144 4145 4146 4147 4148
    if (intermediate_pid != -1) {
        /* Wait for intermediate process to exit */
        while (waitpid(intermediate_pid, &childstat, 0) == -1 &&
               errno == EINTR);
    }
    if (intermediatefd != -1)
        close(intermediatefd);
4149
    close(fd);
4150
    fd = -1;
4151
    if (ret < 0) {
4152
        if (!vm->persistent) {
4153 4154
            virDomainRemoveInactive(&driver->domains,
                                    vm);
4155 4156
            vm = NULL;
        }
4157
        goto cleanup;
4158 4159
    }

4160 4161 4162
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_RESTORED);
4163

4164 4165
    /* If it was running before, resume it now. */
    if (header.was_running) {
4166 4167 4168 4169
        if (qemudMonitorSendCont(conn, vm) < 0) {
            if (virGetLastError() == NULL)
                qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                                 "%s", _("failed to resume domain"));
4170
            goto cleanup;
4171 4172
        }
        vm->state = VIR_DOMAIN_RUNNING;
4173
        virDomainSaveStatus(conn, driver->stateDir, vm);
4174
    }
4175
    ret = 0;
4176

4177 4178 4179 4180 4181
cleanup:
    virDomainDefFree(def);
    VIR_FREE(xml);
    if (fd != -1)
        close(fd);
4182 4183
    if (vm)
        virDomainObjUnlock(vm);
4184 4185
    if (event)
        qemuDomainEventQueue(driver, event);
4186
    qemuDriverUnlock(driver);
4187
    return ret;
D
Daniel P. Berrange 已提交
4188 4189 4190
}


4191
static char *qemudDomainDumpXML(virDomainPtr dom,
4192
                                int flags) {
4193 4194 4195
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    char *ret = NULL;
4196 4197
    unsigned long balloon;
    int err;
4198

4199
    qemuDriverLock(driver);
4200
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4201 4202
    qemuDriverUnlock(driver);

D
Daniel P. Berrange 已提交
4203
    if (!vm) {
4204 4205 4206 4207
        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);
4208
        goto cleanup;
D
Daniel P. Berrange 已提交
4209 4210
    }

4211 4212 4213 4214 4215 4216 4217
    /* Refresh current memory based on balloon info */
    err = qemudDomainGetMemoryBalloon(dom->conn, vm, &balloon);
    if (err < 0)
        goto cleanup;
    if (err > 0)
        vm->def->memory = balloon;

4218 4219 4220 4221 4222 4223
    ret = virDomainDefFormat(dom->conn,
                             (flags & VIR_DOMAIN_XML_INACTIVE) && vm->newDef ?
                             vm->newDef : vm->def,
                             flags);

cleanup:
4224 4225
    if (vm)
        virDomainObjUnlock(vm);
4226
    return ret;
D
Daniel P. Berrange 已提交
4227 4228 4229
}


4230 4231 4232 4233
static char *qemuDomainXMLFromNative(virConnectPtr conn,
                                     const char *format,
                                     const char *config,
                                     unsigned int flags ATTRIBUTE_UNUSED) {
4234
    struct qemud_driver *driver = conn->privateData;
4235 4236 4237 4238 4239 4240 4241 4242 4243
    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;
    }

4244
    qemuDriverLock(driver);
4245
    def = qemuParseCommandLineString(conn, driver->caps, config);
4246
    qemuDriverUnlock(driver);
4247 4248 4249 4250 4251 4252 4253 4254 4255 4256
    if (!def)
        goto cleanup;

    xml = virDomainDefFormat(conn, def, VIR_DOMAIN_XML_INACTIVE);

cleanup:
    virDomainDefFree(def);
    return xml;
}

4257 4258 4259 4260 4261 4262
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;
4263
    virDomainChrDef monitor_chr;
4264 4265 4266 4267 4268 4269 4270 4271 4272 4273
    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;

4274 4275
    qemuDriverLock(driver);

4276 4277 4278 4279 4280 4281 4282 4283 4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302 4303 4304 4305 4306 4307 4308 4309 4310 4311 4312 4313 4314 4315 4316 4317 4318 4319 4320 4321 4322 4323 4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334 4335 4336 4337 4338 4339 4340 4341
    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;
    }

4342 4343
    if (qemuPrepareMonitorChr(conn, driver, &monitor_chr, def->name) < 0)
        goto cleanup;
4344 4345

    if (qemudBuildCommandLine(conn, driver, def,
4346
                              &monitor_chr, qemuCmdFlags,
4347 4348 4349 4350 4351 4352 4353 4354 4355 4356 4357 4358 4359 4360 4361 4362 4363 4364 4365 4366 4367 4368 4369 4370 4371
                              &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:
4372
    qemuDriverUnlock(driver);
4373 4374 4375 4376 4377 4378 4379 4380 4381 4382 4383 4384 4385
    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;
}


4386
static int qemudListDefinedDomains(virConnectPtr conn,
4387
                            char **const names, int nnames) {
4388
    struct qemud_driver *driver = conn->privateData;
4389
    int got = 0, i;
4390

4391
    qemuDriverLock(driver);
4392
    for (i = 0 ; i < driver->domains.count && got < nnames ; i++) {
4393
        virDomainObjLock(driver->domains.objs[i]);
4394 4395
        if (!virDomainIsActive(driver->domains.objs[i])) {
            if (!(names[got++] = strdup(driver->domains.objs[i]->def->name))) {
4396
                virReportOOMError(conn);
4397
                virDomainObjUnlock(driver->domains.objs[i]);
4398 4399
                goto cleanup;
            }
4400
        }
4401
        virDomainObjUnlock(driver->domains.objs[i]);
D
Daniel P. Berrange 已提交
4402
    }
4403

4404
    qemuDriverUnlock(driver);
D
Daniel P. Berrange 已提交
4405
    return got;
4406 4407 4408

 cleanup:
    for (i = 0 ; i < got ; i++)
4409
        VIR_FREE(names[i]);
4410
    qemuDriverUnlock(driver);
4411
    return -1;
D
Daniel P. Berrange 已提交
4412 4413
}

4414
static int qemudNumDefinedDomains(virConnectPtr conn) {
4415
    struct qemud_driver *driver = conn->privateData;
4416 4417
    int n = 0, i;

4418
    qemuDriverLock(driver);
4419 4420
    for (i = 0 ; i < driver->domains.count ; i++)
        if (!virDomainIsActive(driver->domains.objs[i]))
4421
            n++;
4422
    qemuDriverUnlock(driver);
4423

4424
    return n;
D
Daniel P. Berrange 已提交
4425 4426 4427
}


4428
static int qemudDomainStart(virDomainPtr dom) {
4429 4430 4431
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4432
    virDomainEventPtr event = NULL;
4433

4434
    qemuDriverLock(driver);
4435
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4436

4437
    if (!vm) {
4438 4439 4440 4441
        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);
4442
        goto cleanup;
4443 4444
    }

4445
    ret = qemudStartVMDaemon(dom->conn, driver, vm, NULL, -1);
4446
    if (ret != -1)
4447 4448 4449
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STARTED,
                                         VIR_DOMAIN_EVENT_STARTED_BOOTED);
4450 4451

cleanup:
4452 4453
    if (vm)
        virDomainObjUnlock(vm);
4454
    if (event)
4455
        qemuDomainEventQueue(driver, event);
4456
    qemuDriverUnlock(driver);
4457
    return ret;
D
Daniel P. Berrange 已提交
4458 4459
}

4460 4461 4462 4463 4464 4465 4466 4467 4468 4469 4470 4471 4472 4473 4474 4475 4476 4477 4478 4479 4480 4481 4482 4483 4484 4485 4486 4487 4488
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;
}

4489 4490 4491 4492 4493 4494 4495 4496 4497 4498 4499 4500 4501 4502 4503 4504 4505 4506 4507 4508 4509 4510 4511 4512 4513 4514 4515 4516
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;
}

4517 4518
int
qemudCanonicalizeMachine(struct qemud_driver *driver, virDomainDefPtr def)
4519 4520 4521 4522 4523 4524
{
    char *canonical = NULL;
    int i;

    for (i = 0; i < driver->caps->nguests; i++) {
        virCapsGuestPtr guest = driver->caps->guests[i];
4525
        virCapsGuestDomainInfoPtr info;
4526 4527 4528
        int j;

        for (j = 0; j < guest->arch.ndomains; j++) {
4529
            info = &guest->arch.domains[j]->info;
4530

4531 4532 4533 4534 4535 4536 4537 4538 4539
            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;
4540 4541
        }

4542 4543 4544 4545
        info = &guest->arch.defaultInfo;

        if (info->emulator && STREQ(info->emulator, def->emulator)) {
            if (qemudCanonicalizeMachineFromInfo(def, info, &canonical) < 0)
4546 4547 4548 4549
                return -1;
            goto out;
        }
    }
4550 4551 4552 4553

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

4554 4555 4556 4557 4558 4559 4560
out:
    if (canonical) {
        VIR_FREE(def->os.machine);
        def->os.machine = canonical;
    }
    return 0;
}
D
Daniel P. Berrange 已提交
4561

4562
static virDomainPtr qemudDomainDefine(virConnectPtr conn, const char *xml) {
4563
    struct qemud_driver *driver = conn->privateData;
4564
    virDomainDefPtr def;
4565
    virDomainObjPtr vm = NULL;
4566
    virDomainPtr dom = NULL;
4567
    virDomainEventPtr event = NULL;
4568
    int newVM = 1;
4569

4570
    qemuDriverLock(driver);
4571 4572
    if (!(def = virDomainDefParseString(conn, driver->caps, xml,
                                        VIR_DOMAIN_XML_INACTIVE)))
4573
        goto cleanup;
4574

4575 4576 4577
    if (virSecurityDriverVerify(conn, def) < 0)
        goto cleanup;

4578 4579
    /* See if a VM with matching UUID already exists */
    vm = virDomainFindByUUID(&driver->domains, def->uuid);
4580
    if (vm) {
4581 4582 4583 4584 4585 4586 4587 4588 4589 4590 4591
        /* 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 */
4592
        virDomainObjUnlock(vm);
4593
        newVM = 0;
4594 4595 4596 4597 4598 4599 4600 4601 4602 4603 4604
    } 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;
        }
4605
    }
4606

4607
    if (qemudCanonicalizeMachine(driver, def) < 0)
4608 4609
        goto cleanup;

4610 4611 4612
    if (!(vm = virDomainAssignDef(conn,
                                  &driver->domains,
                                  def))) {
4613
        goto cleanup;
4614
    }
4615
    def = NULL;
4616
    vm->persistent = 1;
4617

4618 4619
    if (virDomainSaveConfig(conn,
                            driver->configDir,
4620
                            vm->newDef ? vm->newDef : vm->def) < 0) {
4621 4622
        virDomainRemoveInactive(&driver->domains,
                                vm);
4623
        vm = NULL;
4624
        goto cleanup;
4625 4626
    }

4627 4628 4629 4630 4631
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_DEFINED,
                                     newVM ?
                                     VIR_DOMAIN_EVENT_DEFINED_ADDED :
                                     VIR_DOMAIN_EVENT_DEFINED_UPDATED);
4632

4633
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
4634
    if (dom) dom->id = vm->def->id;
4635 4636

cleanup:
4637
    virDomainDefFree(def);
4638 4639
    if (vm)
        virDomainObjUnlock(vm);
4640 4641
    if (event)
        qemuDomainEventQueue(driver, event);
4642
    qemuDriverUnlock(driver);
4643
    return dom;
D
Daniel P. Berrange 已提交
4644 4645
}

4646
static int qemudDomainUndefine(virDomainPtr dom) {
4647 4648
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
4649
    virDomainEventPtr event = NULL;
4650
    int ret = -1;
D
Daniel P. Berrange 已提交
4651

4652
    qemuDriverLock(driver);
4653
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4654

D
Daniel P. Berrange 已提交
4655
    if (!vm) {
4656 4657 4658 4659
        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);
4660
        goto cleanup;
D
Daniel P. Berrange 已提交
4661 4662
    }

4663
    if (virDomainIsActive(vm)) {
4664
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
4665
                         "%s", _("cannot delete active domain"));
4666
        goto cleanup;
D
Daniel P. Berrange 已提交
4667 4668
    }

4669 4670 4671
    if (!vm->persistent) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
                         "%s", _("cannot undefine transient domain"));
4672
        goto cleanup;
4673 4674 4675
    }

    if (virDomainDeleteConfig(dom->conn, driver->configDir, driver->autostartDir, vm) < 0)
4676
        goto cleanup;
D
Daniel P. Berrange 已提交
4677

4678 4679 4680
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_UNDEFINED,
                                     VIR_DOMAIN_EVENT_UNDEFINED_REMOVED);
4681

4682 4683
    virDomainRemoveInactive(&driver->domains,
                            vm);
4684
    vm = NULL;
4685
    ret = 0;
D
Daniel P. Berrange 已提交
4686

4687
cleanup:
4688 4689
    if (vm)
        virDomainObjUnlock(vm);
4690 4691
    if (event)
        qemuDomainEventQueue(driver, event);
4692
    qemuDriverUnlock(driver);
4693
    return ret;
D
Daniel P. Berrange 已提交
4694 4695
}

4696
/* Return the disks name for use in monitor commands */
4697
static char *qemudDiskDeviceName(const virConnectPtr conn,
4698
                                 const virDomainDiskDefPtr disk) {
4699 4700 4701 4702 4703 4704

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

    if (virDiskNameToBusDeviceIndex(disk, &busid, &devid) < 0) {
4705
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
4706 4707 4708 4709 4710 4711 4712
                         _("cannot convert disk '%s' to bus/device index"),
                         disk->dst);
        return NULL;
    }

    switch (disk->bus) {
        case VIR_DOMAIN_DISK_BUS_IDE:
4713
            if (disk->device== VIR_DOMAIN_DISK_DEVICE_DISK)
4714
                ret = virAsprintf(&devname, "ide%d-hd%d", busid, devid);
4715
            else
4716
                ret = virAsprintf(&devname, "ide%d-cd%d", busid, devid);
4717 4718
            break;
        case VIR_DOMAIN_DISK_BUS_SCSI:
4719
            if (disk->device == VIR_DOMAIN_DISK_DEVICE_DISK)
4720
                ret = virAsprintf(&devname, "scsi%d-hd%d", busid, devid);
4721
            else
4722
                ret = virAsprintf(&devname, "scsi%d-cd%d", busid, devid);
4723 4724
            break;
        case VIR_DOMAIN_DISK_BUS_FDC:
4725
            ret = virAsprintf(&devname, "floppy%d", devid);
4726 4727
            break;
        case VIR_DOMAIN_DISK_BUS_VIRTIO:
4728
            ret = virAsprintf(&devname, "virtio%d", devid);
4729 4730
            break;
        default:
4731
            qemudReportError(conn, NULL, NULL, VIR_ERR_NO_SUPPORT,
4732 4733 4734 4735 4736 4737
                             _("Unsupported disk name mapping for bus '%s'"),
                             virDomainDiskBusTypeToString(disk->bus));
            return NULL;
    }

    if (ret == -1) {
4738
        virReportOOMError(conn);
4739 4740 4741 4742 4743 4744
        return NULL;
    }

    return devname;
}

4745 4746
static int qemudDomainChangeEjectableMedia(virConnectPtr conn,
                                           virDomainObjPtr vm,
4747 4748
                                           virDomainDeviceDefPtr dev,
                                           unsigned int qemuCmdFlags)
4749
{
4750
    virDomainDiskDefPtr origdisk = NULL, newdisk;
4751
    char *cmd, *reply, *safe_path;
4752
    char *devname = NULL;
4753
    int i;
4754

4755
    origdisk = NULL;
4756
    newdisk = dev->data.disk;
4757 4758 4759 4760
    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];
4761
            break;
4762
        }
4763 4764 4765
    }

    if (!origdisk) {
4766
        qemudReportError(conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
4767 4768 4769 4770 4771 4772 4773
                         _("No device with bus '%s' and target '%s'"),
                         virDomainDiskBusTypeToString(newdisk->bus),
                         newdisk->dst);
        return -1;
    }

    if (qemuCmdFlags & QEMUD_CMD_FLAG_DRIVE) {
4774
        if (!(devname = qemudDiskDeviceName(conn, newdisk)))
4775 4776 4777 4778 4779 4780 4781 4782 4783
            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 {
4784
            qemudReportError(conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
4785 4786 4787 4788 4789 4790 4791 4792
                             _("Emulator version does not support removable "
                               "media for device '%s' and target '%s'"),
                               virDomainDiskDeviceTypeToString(newdisk->device),
                               newdisk->dst);
            return -1;
        }

        if (!devname) {
4793
            virReportOOMError(conn);
4794 4795 4796
            return -1;
        }
    }
4797

4798
    if (newdisk->src) {
4799 4800
        safe_path = qemudEscapeMonitorArg(newdisk->src);
        if (!safe_path) {
4801
            virReportOOMError(conn);
4802
            VIR_FREE(devname);
4803 4804
            return -1;
        }
4805
        if (virAsprintf(&cmd, "change %s \"%s\"", devname, safe_path) == -1) {
4806
            virReportOOMError(conn);
4807
            VIR_FREE(safe_path);
4808
            VIR_FREE(devname);
4809 4810
            return -1;
        }
4811
        VIR_FREE(safe_path);
4812

4813
    } else if (virAsprintf(&cmd, "eject %s", devname) == -1) {
4814
        virReportOOMError(conn);
4815
        VIR_FREE(devname);
4816 4817
        return -1;
    }
4818
    VIR_FREE(devname);
4819

4820
    if (qemudMonitorCommand(vm, cmd, &reply) < 0) {
4821
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
C
Cole Robinson 已提交
4822
                         "%s", _("could not change cdrom media"));
4823
        VIR_FREE(cmd);
4824 4825
        return -1;
    }
4826 4827 4828 4829

    /* If the command failed qemu prints:
     * device not found, device is locked ...
     * No message is printed on success it seems */
4830
    DEBUG ("%s: ejectable media change reply: %s", vm->def->name, reply);
4831
    if (strstr(reply, "\ndevice ")) {
4832
        qemudReportError (conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
C
Cole Robinson 已提交
4833
                          _("changing cdrom media failed: %s"), reply);
4834 4835 4836 4837
        VIR_FREE(reply);
        VIR_FREE(cmd);
        return -1;
    }
4838 4839
    VIR_FREE(reply);
    VIR_FREE(cmd);
4840

4841 4842
    VIR_FREE(origdisk->src);
    origdisk->src = newdisk->src;
4843
    newdisk->src = NULL;
4844
    origdisk->type = newdisk->type;
4845 4846 4847
    return 0;
}

4848 4849 4850 4851 4852 4853 4854 4855 4856 4857 4858 4859 4860 4861 4862 4863 4864 4865 4866 4867 4868 4869 4870 4871 4872 4873 4874 4875 4876 4877 4878 4879 4880 4881 4882 4883 4884 4885 4886 4887 4888 4889 4890 4891 4892 4893 4894 4895 4896 4897 4898 4899 4900 4901 4902 4903 4904 4905 4906 4907 4908 4909 4910 4911 4912 4913 4914
static int
qemudParsePciAddReply(virDomainObjPtr vm,
                      const char *reply,
                      unsigned *domain,
                      unsigned *bus,
                      unsigned *slot)
{
    char *s, *e;

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

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

    s += 3;

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

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

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

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

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

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

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

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

    return 0;
}

4915 4916 4917
static int qemudDomainAttachPciDiskDevice(virConnectPtr conn,
                                          virDomainObjPtr vm,
                                          virDomainDeviceDefPtr dev)
4918 4919
{
    int ret, i;
4920
    char *cmd, *reply;
4921 4922
    char *safe_path;
    const char* type = virDomainDiskBusTypeToString(dev->data.disk->bus);
4923
    int tryOldSyntax = 0;
4924
    unsigned domain, bus, slot;
4925 4926 4927

    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (STREQ(vm->def->disks[i]->dst, dev->data.disk->dst)) {
4928
            qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
4929 4930 4931 4932 4933 4934
                           _("target %s already exists"), dev->data.disk->dst);
            return -1;
        }
    }

    if (VIR_REALLOC_N(vm->def->disks, vm->def->ndisks+1) < 0) {
4935
        virReportOOMError(conn);
4936 4937 4938
        return -1;
    }

4939
try_command:
4940 4941
    safe_path = qemudEscapeMonitorArg(dev->data.disk->src);
    if (!safe_path) {
4942
        virReportOOMError(conn);
4943 4944 4945
        return -1;
    }

4946 4947
    ret = virAsprintf(&cmd, "pci_add %s storage file=%s,if=%s",
                      (tryOldSyntax ? "0": "pci_addr=auto"), safe_path, type);
4948 4949
    VIR_FREE(safe_path);
    if (ret == -1) {
4950
        virReportOOMError(conn);
4951 4952 4953
        return ret;
    }

4954
    if (qemudMonitorCommand(vm, cmd, &reply) < 0) {
4955
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
4956 4957 4958 4959 4960
                         _("cannot attach %s disk"), type);
        VIR_FREE(cmd);
        return -1;
    }

4961 4962
    VIR_FREE(cmd);

4963 4964 4965 4966 4967 4968
    if (qemudParsePciAddReply(vm, reply, &domain, &bus, &slot) < 0) {
        if (!tryOldSyntax && strstr(reply, "invalid char in expression")) {
            VIR_FREE(reply);
            tryOldSyntax = 1;
            goto try_command;
        }
4969

4970
        qemudReportError (conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
4971
                          _("adding %s disk failed: %s"), type, reply);
4972 4973 4974 4975
        VIR_FREE(reply);
        return -1;
    }

4976 4977 4978 4979 4980 4981
    VIR_FREE(reply);

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

4982
    virDomainDiskInsertPreAlloced(vm->def, dev->data.disk);
4983 4984 4985

    return 0;
}
4986

4987 4988 4989
static int qemudDomainAttachUsbMassstorageDevice(virConnectPtr conn,
                                                 virDomainObjPtr vm,
                                                 virDomainDeviceDefPtr dev)
4990
{
4991 4992
    int ret, i;
    char *safe_path;
4993 4994
    char *cmd, *reply;

4995 4996
    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (STREQ(vm->def->disks[i]->dst, dev->data.disk->dst)) {
4997
            qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
4998 4999 5000 5001 5002
                           _("target %s already exists"), dev->data.disk->dst);
            return -1;
        }
    }

5003
    if (VIR_REALLOC_N(vm->def->disks, vm->def->ndisks+1) < 0) {
5004
        virReportOOMError(conn);
5005
        return -1;
5006 5007
    }

5008 5009
    safe_path = qemudEscapeMonitorArg(dev->data.disk->src);
    if (!safe_path) {
5010
        virReportOOMError(conn);
5011 5012 5013
        return -1;
    }

5014
    ret = virAsprintf(&cmd, "usb_add disk:%s", safe_path);
5015
    VIR_FREE(safe_path);
5016
    if (ret == -1) {
5017
        virReportOOMError(conn);
5018 5019 5020
        return ret;
    }

5021
    if (qemudMonitorCommand(vm, cmd, &reply) < 0) {
5022
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
5023
                         "%s", _("cannot attach usb disk"));
5024 5025 5026 5027
        VIR_FREE(cmd);
        return -1;
    }

5028
    DEBUG ("%s: attach_usb reply: %s",vm->def->name,  reply);
5029 5030 5031
    /* If the command failed qemu prints:
     * Could not add ... */
    if (strstr(reply, "Could not add ")) {
5032
        qemudReportError (conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
5033
                          "%s",
5034
                          _("adding usb disk failed"));
5035 5036 5037 5038
        VIR_FREE(reply);
        VIR_FREE(cmd);
        return -1;
    }
5039

5040
    virDomainDiskInsertPreAlloced(vm->def, dev->data.disk);
5041

5042 5043 5044 5045 5046
    VIR_FREE(reply);
    VIR_FREE(cmd);
    return 0;
}

M
Mark McLoughlin 已提交
5047
static int qemudDomainAttachNetDevice(virConnectPtr conn,
5048
                                      struct qemud_driver *driver,
M
Mark McLoughlin 已提交
5049 5050 5051 5052 5053
                                      virDomainObjPtr vm,
                                      virDomainDeviceDefPtr dev,
                                      unsigned int qemuCmdFlags)
{
    virDomainNetDefPtr net = dev->data.net;
5054
    char *cmd = NULL, *reply = NULL, *remove_cmd = NULL;
5055 5056
    char *tapfd_name = NULL, *tapfd_close = NULL;
    int i, tapfd = -1;
5057
    unsigned domain, bus, slot;
M
Mark McLoughlin 已提交
5058 5059 5060 5061 5062 5063 5064 5065 5066

    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) {
5067 5068 5069 5070 5071 5072 5073 5074 5075 5076
        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
5077 5078
    }

5079 5080
    if (VIR_REALLOC_N(vm->def->nets, vm->def->nnets+1) < 0)
        goto no_memory;
M
Mark McLoughlin 已提交
5081 5082

    if ((qemuCmdFlags & QEMUD_CMD_FLAG_NET_NAME) &&
5083 5084
        qemuAssignNetNames(vm->def, net) < 0)
        goto no_memory;
M
Mark McLoughlin 已提交
5085 5086 5087 5088 5089 5090 5091 5092 5093

    /* 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;

5094 5095 5096 5097 5098 5099 5100 5101 5102 5103 5104 5105 5106 5107 5108 5109 5110 5111 5112 5113 5114 5115 5116 5117 5118 5119 5120 5121 5122 5123 5124 5125 5126 5127 5128
    if (tapfd != -1) {
        if (virAsprintf(&tapfd_name, "fd-%s", net->hostnet_name) < 0)
            goto no_memory;

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

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

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

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

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

        VIR_FREE(reply);
        VIR_FREE(cmd);
    }

    if (qemuBuildHostNetStr(conn, net, "host_net_add ", ' ',
                            net->vlan, tapfd_name, &cmd) < 0)
        goto try_tapfd_close;
M
Mark McLoughlin 已提交
5129

5130 5131 5132 5133
    remove_cmd = NULL;
    if (net->vlan >= 0 && net->hostnet_name &&
        virAsprintf(&remove_cmd, "host_net_remove %d %s",
                    net->vlan, net->hostnet_name) < 0) {
5134 5135
        virReportOOMError(conn);
        goto try_tapfd_close;
5136 5137
    }

M
Mark McLoughlin 已提交
5138 5139 5140
    if (qemudMonitorCommand(vm, cmd, &reply) < 0) {
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                         _("failed to add network backend with '%s'"), cmd);
5141
        goto try_tapfd_close;
M
Mark McLoughlin 已提交
5142 5143 5144 5145 5146 5147
    }

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

    VIR_FREE(reply);
    VIR_FREE(cmd);
5148 5149 5150 5151 5152
    VIR_FREE(tapfd_name);
    VIR_FREE(tapfd_close);
    if (tapfd != -1)
        close(tapfd);
    tapfd = -1;
M
Mark McLoughlin 已提交
5153 5154

    if (qemuBuildNicStr(conn, net,
5155 5156
                        "pci_add pci_addr=auto ", ' ', net->vlan, &cmd) < 0)
        goto try_remove;
M
Mark McLoughlin 已提交
5157 5158 5159 5160

    if (qemudMonitorCommand(vm, cmd, &reply) < 0) {
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                         _("failed to add NIC with '%s'"), cmd);
5161
        goto try_remove;
M
Mark McLoughlin 已提交
5162 5163
    }

5164 5165 5166 5167 5168 5169
    if (qemudParsePciAddReply(vm, reply, &domain, &bus, &slot) < 0) {
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                         _("parsing pci_add reply failed: %s"), reply);
        goto try_remove;
    }

5170
    VIR_FREE(cmd);
5171
    VIR_FREE(reply);
5172
    VIR_FREE(remove_cmd);
M
Mark McLoughlin 已提交
5173

5174 5175 5176 5177
    net->pci_addr.domain = domain;
    net->pci_addr.bus    = bus;
    net->pci_addr.slot   = slot;

M
Mark McLoughlin 已提交
5178 5179 5180
    vm->def->nets[vm->def->nnets++] = net;

    return 0;
5181 5182

try_remove:
5183
    VIR_FREE(reply);
5184 5185 5186 5187 5188 5189 5190

    if (!remove_cmd)
        VIR_WARN0(_("Unable to remove network backend\n"));
    else if (qemudMonitorCommand(vm, remove_cmd, &reply) < 0)
        VIR_WARN(_("Failed to remove network backend with '%s'\n"), remove_cmd);
    else
        VIR_DEBUG("%s: host_net_remove reply: %s\n", vm->def->name, reply);
5191
    goto cleanup;
5192

5193 5194 5195 5196 5197 5198 5199 5200 5201 5202 5203
try_tapfd_close:
    VIR_FREE(reply);

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

5204 5205 5206 5207
no_memory:
    virReportOOMError(conn);
cleanup:
    VIR_FREE(cmd);
5208 5209
    VIR_FREE(reply);
    VIR_FREE(remove_cmd);
5210 5211 5212 5213
    VIR_FREE(tapfd_close);
    VIR_FREE(tapfd_name);
    if (tapfd != -1)
        close(tapfd);
5214
    return -1;
M
Mark McLoughlin 已提交
5215 5216
}

5217
static int qemudDomainAttachHostPciDevice(virConnectPtr conn,
5218
                                          struct qemud_driver *driver,
5219 5220 5221 5222 5223 5224
                                          virDomainObjPtr vm,
                                          virDomainDeviceDefPtr dev)
{
    virDomainHostdevDefPtr hostdev = dev->data.hostdev;
    char *cmd, *reply;
    unsigned domain, bus, slot;
5225
    pciDevice *pci;
5226 5227 5228 5229 5230 5231

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

5232 5233 5234 5235 5236 5237 5238
    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;
5239

5240
    if ((hostdev->managed && pciDettachDevice(conn, pci) < 0) ||
5241 5242 5243 5244 5245 5246
        pciResetDevice(conn, pci, driver->activePciHostdevs) < 0) {
        pciFreeDevice(conn, pci);
        return -1;
    }

    if (pciDeviceListAdd(conn, driver->activePciHostdevs, pci) < 0) {
5247
        pciFreeDevice(conn, pci);
5248
        return -1;
5249 5250
    }

5251
    cmd = reply = NULL;
5252

5253
    if (virAsprintf(&cmd, "pci_add pci_addr=auto host host=%.2x:%.2x.%.1x",
5254 5255 5256 5257
                    hostdev->source.subsys.u.pci.bus,
                    hostdev->source.subsys.u.pci.slot,
                    hostdev->source.subsys.u.pci.function) < 0) {
        virReportOOMError(conn);
5258
        goto error;
5259 5260 5261 5262 5263
    }

    if (qemudMonitorCommand(vm, cmd, &reply) < 0) {
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                         "%s", _("cannot attach host pci device"));
5264
        goto error;
5265 5266 5267 5268 5269
    }

    if (strstr(reply, "invalid type: host")) {
        qemudReportError(conn, dom, NULL, VIR_ERR_NO_SUPPORT, "%s",
                         _("PCI device assignment is not supported by this version of qemu"));
5270
        goto error;
5271 5272 5273 5274 5275
    }

    if (qemudParsePciAddReply(vm, reply, &domain, &bus, &slot) < 0) {
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                         _("parsing pci_add reply failed: %s"), reply);
5276
        goto error;
5277 5278 5279 5280 5281 5282 5283 5284 5285 5286 5287 5288
    }

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

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

    VIR_FREE(reply);
    VIR_FREE(cmd);

    return 0;
5289 5290 5291 5292 5293 5294 5295 5296

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

    VIR_FREE(reply);
    VIR_FREE(cmd);

    return -1;
5297 5298
}

M
Mark McLoughlin 已提交
5299 5300 5301
static int qemudDomainAttachHostUsbDevice(virConnectPtr conn,
                                          virDomainObjPtr vm,
                                          virDomainDeviceDefPtr dev)
5302 5303 5304 5305
{
    int ret;
    char *cmd, *reply;

5306
    if (VIR_REALLOC_N(vm->def->hostdevs, vm->def->nhostdevs+1) < 0) {
5307
        virReportOOMError(conn);
5308 5309
        return -1;
    }
5310

5311
    if (dev->data.hostdev->source.subsys.u.usb.vendor) {
5312 5313 5314
        ret = virAsprintf(&cmd, "usb_add host:%.4x:%.4x",
                          dev->data.hostdev->source.subsys.u.usb.vendor,
                          dev->data.hostdev->source.subsys.u.usb.product);
5315
    } else {
5316 5317 5318
        ret = virAsprintf(&cmd, "usb_add host:%.3d.%.3d",
                          dev->data.hostdev->source.subsys.u.usb.bus,
                          dev->data.hostdev->source.subsys.u.usb.device);
5319 5320
    }
    if (ret == -1) {
5321
        virReportOOMError(conn);
5322 5323 5324
        return -1;
    }

5325
    if (qemudMonitorCommand(vm, cmd, &reply) < 0) {
5326
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
5327 5328 5329 5330 5331
                         "%s", _("cannot attach usb device"));
        VIR_FREE(cmd);
        return -1;
    }

5332
    DEBUG ("%s: attach_usb reply: %s", vm->def->name, reply);
5333 5334 5335
    /* If the command failed qemu prints:
     * Could not add ... */
    if (strstr(reply, "Could not add ")) {
5336
        qemudReportError (conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
5337 5338 5339 5340 5341 5342
                          "%s",
                          _("adding usb device failed"));
        VIR_FREE(reply);
        VIR_FREE(cmd);
        return -1;
    }
5343

5344
    vm->def->hostdevs[vm->def->nhostdevs++] = dev->data.hostdev;
5345

5346 5347
    VIR_FREE(reply);
    VIR_FREE(cmd);
5348 5349 5350
    return 0;
}

M
Mark McLoughlin 已提交
5351 5352 5353 5354 5355 5356 5357 5358 5359 5360 5361 5362 5363 5364 5365 5366
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;
5367 5368 5369
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityHostdevLabel(conn, vm, dev->data.hostdev) < 0)
        return -1;
M
Mark McLoughlin 已提交
5370 5371

    switch (hostdev->source.subsys.type) {
5372
    case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI:
5373
        return qemudDomainAttachHostPciDevice(conn, driver, vm, dev);
M
Mark McLoughlin 已提交
5374 5375 5376 5377 5378 5379 5380 5381 5382 5383
    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;
    }
}

5384 5385
static int qemudDomainAttachDevice(virDomainPtr dom,
                                   const char *xml) {
5386 5387 5388
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    virDomainDeviceDefPtr dev = NULL;
5389
    unsigned int qemuCmdFlags;
5390
    virCgroupPtr cgroup = NULL;
5391
    int ret = -1;
5392

5393
    qemuDriverLock(driver);
5394
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5395
    if (!vm) {
5396 5397 5398 5399
        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);
5400
        goto cleanup;
5401 5402 5403
    }

    if (!virDomainIsActive(vm)) {
5404
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
5405
                         "%s", _("cannot attach device on inactive domain"));
5406
        goto cleanup;
5407 5408
    }

5409 5410
    dev = virDomainDeviceDefParse(dom->conn, driver->caps, vm->def, xml,
                                  VIR_DOMAIN_XML_INACTIVE);
5411 5412 5413
    if (dev == NULL)
        goto cleanup;

5414 5415 5416 5417
    if (qemudExtractVersionInfo(vm->def->emulator,
                                NULL,
                                &qemuCmdFlags) < 0)
        goto cleanup;
5418

5419
    if (dev->type == VIR_DOMAIN_DEVICE_DISK) {
5420
        if (qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_DEVICES)) {
5421 5422 5423 5424 5425 5426 5427 5428 5429 5430 5431 5432 5433 5434 5435 5436 5437
            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;
            }
        }

5438
        switch (dev->data.disk->device) {
5439 5440
        case VIR_DOMAIN_DISK_DEVICE_CDROM:
        case VIR_DOMAIN_DISK_DEVICE_FLOPPY:
5441 5442
            if (driver->securityDriver)
                driver->securityDriver->domainSetSecurityImageLabel(dom->conn, vm, dev->data.disk);
5443 5444 5445 5446

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

5447
            ret = qemudDomainChangeEjectableMedia(dom->conn, vm, dev, qemuCmdFlags);
5448
            break;
5449

5450
        case VIR_DOMAIN_DISK_DEVICE_DISK:
5451 5452
            if (driver->securityDriver)
                driver->securityDriver->domainSetSecurityImageLabel(dom->conn, vm, dev->data.disk);
5453 5454 5455 5456

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

5457
            if (dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_USB) {
5458
                ret = qemudDomainAttachUsbMassstorageDevice(dom->conn, vm, dev);
5459 5460
            } else if (dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_SCSI ||
                       dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_VIRTIO) {
5461
                ret = qemudDomainAttachPciDiskDevice(dom->conn, vm, dev);
5462 5463 5464 5465
            } else {
                qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                                 _("disk bus '%s' cannot be hotplugged."),
                                 virDomainDiskBusTypeToString(dev->data.disk->bus));
5466
                /* fallthrough */
5467 5468
            }
            break;
5469

5470 5471
        default:
            qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
5472 5473
                             _("disk device type '%s' cannot be hotplugged"),
                             virDomainDiskDeviceTypeToString(dev->data.disk->device));
5474 5475 5476 5477 5478
            /* Fallthrough */
        }
        if (ret != 0) {
            virCgroupDenyDevicePath(cgroup,
                                    dev->data.disk->src);
5479
        }
M
Mark McLoughlin 已提交
5480
    } else if (dev->type == VIR_DOMAIN_DEVICE_NET) {
5481
        ret = qemudDomainAttachNetDevice(dom->conn, driver, vm, dev, qemuCmdFlags);
M
Mark McLoughlin 已提交
5482 5483
    } else if (dev->type == VIR_DOMAIN_DEVICE_HOSTDEV) {
        ret = qemudDomainAttachHostDevice(dom->conn, driver, vm, dev);
5484
    } else {
5485
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
5486 5487
                         _("device type '%s' cannot be attached"),
                         virDomainDeviceTypeToString(dev->type));
5488
        goto cleanup;
5489 5490
    }

5491
    if (!ret && virDomainSaveStatus(dom->conn, driver->stateDir, vm) < 0)
5492 5493
        ret = -1;

5494
cleanup:
5495 5496 5497
    if (cgroup)
        virCgroupFree(&cgroup);

5498 5499 5500
    if (ret < 0) {
        if (qemuDomainSetDeviceOwnership(dom->conn, driver, dev, 1) < 0)
            VIR_WARN0("Fail to restore disk device ownership");
G
Guido Günther 已提交
5501
        virDomainDeviceDefFree(dev);
5502
    }
5503 5504
    if (vm)
        virDomainObjUnlock(vm);
5505
    qemuDriverUnlock(driver);
5506 5507 5508
    return ret;
}

5509 5510
static int qemudDomainDetachPciDiskDevice(virConnectPtr conn,
                                          virDomainObjPtr vm, virDomainDeviceDefPtr dev)
5511 5512
{
    int i, ret = -1;
5513 5514
    char *cmd = NULL;
    char *reply = NULL;
5515
    virDomainDiskDefPtr detach = NULL;
5516
    int tryOldSyntax = 0;
5517 5518 5519 5520 5521 5522 5523 5524 5525

    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) {
5526
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
5527
                         _("disk %s not found"), dev->data.disk->dst);
5528
        goto cleanup;
5529 5530
    }

5531
    if (!virDiskHasValidPciAddr(detach)) {
5532
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
5533 5534
                         _("disk %s cannot be detached - no PCI address for device"),
                           detach->dst);
5535
        goto cleanup;
5536 5537
    }

5538 5539
try_command:
    if (tryOldSyntax) {
5540
        if (virAsprintf(&cmd, "pci_del 0 %.2x", detach->pci_addr.slot) < 0) {
5541 5542 5543 5544
            virReportOOMError(conn);
            goto cleanup;
        }
    } else {
5545 5546 5547 5548
        if (virAsprintf(&cmd, "pci_del pci_addr=%.4x:%.2x:%.2x",
                        detach->pci_addr.domain,
                        detach->pci_addr.bus,
                        detach->pci_addr.slot) < 0) {
5549 5550 5551
            virReportOOMError(conn);
            goto cleanup;
        }
5552 5553
    }

5554
    if (qemudMonitorCommand(vm, cmd, &reply) < 0) {
5555
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
5556
                          _("failed to execute detach disk %s command"), detach->dst);
5557
        goto cleanup;
5558 5559
    }

5560
    DEBUG ("%s: pci_del reply: %s",vm->def->name,  reply);
5561 5562 5563 5564 5565 5566

    if (!tryOldSyntax &&
        strstr(reply, "extraneous characters")) {
        tryOldSyntax = 1;
        goto try_command;
    }
5567 5568
    /* If the command fails due to a wrong slot qemu prints: invalid slot,
     * nothing is printed on success */
5569 5570
    if (strstr(reply, "invalid slot") ||
        strstr(reply, "Invalid pci address")) {
5571
        qemudReportError (conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
5572 5573 5574 5575 5576 5577
                          _("failed to detach disk %s: invalid PCI address %.4x:%.2x:%.2x: %s"),
                          detach->dst,
                          detach->pci_addr.domain,
                          detach->pci_addr.bus,
                          detach->pci_addr.slot,
                          reply);
5578
        goto cleanup;
5579 5580
    }

5581 5582 5583 5584 5585 5586 5587 5588 5589 5590
    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 {
5591
        VIR_FREE(vm->def->disks);
5592
        vm->def->ndisks = 0;
5593
    }
5594
    virDomainDiskDefFree(detach);
5595

5596
    ret = 0;
5597 5598

cleanup:
5599 5600 5601 5602 5603
    VIR_FREE(reply);
    VIR_FREE(cmd);
    return ret;
}

5604 5605 5606 5607 5608 5609 5610 5611 5612 5613 5614 5615 5616 5617 5618 5619 5620 5621 5622 5623 5624 5625 5626 5627 5628 5629 5630 5631 5632 5633 5634 5635 5636 5637 5638 5639 5640 5641 5642 5643 5644 5645 5646 5647 5648 5649 5650 5651 5652 5653 5654 5655 5656 5657 5658 5659 5660 5661 5662 5663 5664 5665 5666 5667 5668 5669 5670 5671 5672 5673 5674 5675 5676 5677 5678 5679 5680 5681 5682
static int
qemudDomainDetachNetDevice(virConnectPtr conn,
                           virDomainObjPtr vm,
                           virDomainDeviceDefPtr dev)
{
    int i, ret = -1;
    char *cmd = NULL;
    char *reply = NULL;
    virDomainNetDefPtr detach = NULL;

    for (i = 0 ; i < vm->def->nnets ; i++) {
        virDomainNetDefPtr net = vm->def->nets[i];

        if (!memcmp(net->mac, dev->data.net->mac,  sizeof(net->mac))) {
            detach = net;
            break;
        }
    }

    if (!detach) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                         _("network device %02x:%02x:%02x:%02x:%02x:%02x not found"),
                         dev->data.net->mac[0], dev->data.net->mac[1],
                         dev->data.net->mac[2], dev->data.net->mac[3],
                         dev->data.net->mac[4], dev->data.net->mac[5]);
        goto cleanup;
    }

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

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

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

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

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

    VIR_FREE(reply);
    VIR_FREE(cmd);

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

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

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

5683 5684 5685 5686 5687 5688 5689 5690 5691 5692
    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 {
5693
        VIR_FREE(vm->def->nets);
5694
        vm->def->nnets = 0;
5695
    }
5696
    virDomainNetDefFree(detach);
5697

5698 5699 5700 5701 5702 5703 5704 5705
    ret = 0;

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

5706
static int qemudDomainDetachHostPciDevice(virConnectPtr conn,
5707
                                          struct qemud_driver *driver,
5708 5709 5710
                                          virDomainObjPtr vm,
                                          virDomainDeviceDefPtr dev)
{
5711
    virDomainHostdevDefPtr detach = NULL;
5712 5713
    char *cmd, *reply;
    int i, ret;
5714
    pciDevice *pci;
5715 5716 5717 5718 5719 5720 5721 5722 5723 5724 5725 5726 5727 5728 5729 5730 5731 5732 5733 5734 5735 5736 5737 5738 5739 5740 5741 5742 5743 5744 5745 5746 5747 5748 5749 5750 5751 5752 5753 5754 5755 5756 5757 5758 5759 5760 5761 5762 5763 5764 5765 5766 5767 5768 5769 5770 5771 5772 5773 5774 5775 5776 5777 5778 5779 5780 5781 5782

    for (i = 0 ; i < vm->def->nhostdevs ; i++) {
        unsigned domain   = vm->def->hostdevs[i]->source.subsys.u.pci.domain;
        unsigned bus      = vm->def->hostdevs[i]->source.subsys.u.pci.bus;
        unsigned slot     = vm->def->hostdevs[i]->source.subsys.u.pci.slot;
        unsigned function = vm->def->hostdevs[i]->source.subsys.u.pci.function;

        if (dev->data.hostdev->source.subsys.u.pci.domain   == domain &&
            dev->data.hostdev->source.subsys.u.pci.bus      == bus &&
            dev->data.hostdev->source.subsys.u.pci.slot     == slot &&
            dev->data.hostdev->source.subsys.u.pci.function == function) {
            detach = vm->def->hostdevs[i];
            break;
        }
    }

    if (!detach) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                         _("host pci device %.4x:%.2x:%.2x.%.1x not found"),
                         dev->data.hostdev->source.subsys.u.pci.domain,
                         dev->data.hostdev->source.subsys.u.pci.bus,
                         dev->data.hostdev->source.subsys.u.pci.slot,
                         dev->data.hostdev->source.subsys.u.pci.function);
        return -1;
    }

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

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

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

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

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

    VIR_FREE(reply);
    VIR_FREE(cmd);

    ret = 0;

5783 5784 5785 5786 5787 5788 5789 5790
    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 {
5791 5792
        pciDeviceListDel(conn, driver->activePciHostdevs, pci);
        if (pciResetDevice(conn, pci, driver->activePciHostdevs) < 0)
5793 5794
            ret = -1;
        if (detach->managed && pciReAttachDevice(conn, pci) < 0)
5795
            ret = -1;
5796
        pciFreeDevice(conn, pci);
5797 5798
    }

5799 5800 5801 5802 5803 5804 5805 5806 5807 5808 5809 5810
    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;
5811
    }
5812
    virDomainHostdevDefFree(detach);
5813 5814 5815 5816 5817 5818 5819 5820 5821 5822 5823 5824 5825 5826 5827 5828 5829 5830 5831 5832 5833

    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:
5834
        ret = qemudDomainDetachHostPciDevice(conn, driver, vm, dev);
5835
        break;
5836 5837 5838 5839 5840 5841 5842
    default:
        qemudReportError(conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                         _("hostdev subsys type '%s' not supported"),
                         virDomainHostdevSubsysTypeToString(hostdev->source.subsys.type));
        return -1;
    }

5843 5844 5845 5846
    if (driver->securityDriver &&
        driver->securityDriver->domainSetSecurityHostdevLabel(conn, vm, dev->data.hostdev) < 0)
        VIR_WARN0("Failed to restore device labelling");

5847
    if (qemuDomainSetDeviceOwnership(conn, driver, dev, 1) < 0)
5848
        VIR_WARN0("Failed to restore device ownership");
5849 5850 5851 5852

    return ret;
}

5853 5854
static int qemudDomainDetachDevice(virDomainPtr dom,
                                   const char *xml) {
5855 5856 5857 5858
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    virDomainDeviceDefPtr dev = NULL;
    int ret = -1;
5859

5860
    qemuDriverLock(driver);
5861
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5862
    if (!vm) {
5863 5864 5865 5866
        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);
5867
        goto cleanup;
5868 5869 5870
    }

    if (!virDomainIsActive(vm)) {
5871
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
G
Guido Günther 已提交
5872
                         "%s", _("cannot detach device on inactive domain"));
5873
        goto cleanup;
5874 5875
    }

5876 5877
    dev = virDomainDeviceDefParse(dom->conn, driver->caps, vm->def, xml,
                                  VIR_DOMAIN_XML_INACTIVE);
5878 5879 5880
    if (dev == NULL)
        goto cleanup;

5881 5882 5883 5884

    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 ||
5885
         dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_VIRTIO)) {
5886
        ret = qemudDomainDetachPciDiskDevice(dom->conn, vm, dev);
5887
        if (driver->securityDriver)
5888
            driver->securityDriver->domainRestoreSecurityImageLabel(dom->conn, dev->data.disk);
5889 5890
        if (qemuDomainSetDeviceOwnership(dom->conn, driver, dev, 1) < 0)
            VIR_WARN0("Fail to restore disk device ownership");
5891 5892
    } else if (dev->type == VIR_DOMAIN_DEVICE_NET) {
        ret = qemudDomainDetachNetDevice(dom->conn, vm, dev);
5893 5894
    } else if (dev->type == VIR_DOMAIN_DEVICE_HOSTDEV) {
        ret = qemudDomainDetachHostDevice(dom->conn, driver, vm, dev);
5895
    } else
5896 5897 5898
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                         "%s", _("only SCSI or virtio disk device can be detached dynamically"));

5899
    if (!ret && virDomainSaveStatus(dom->conn, driver->stateDir, vm) < 0)
5900 5901
        ret = -1;

5902 5903
cleanup:
    virDomainDeviceDefFree(dev);
5904 5905
    if (vm)
        virDomainObjUnlock(vm);
5906
    qemuDriverUnlock(driver);
5907 5908 5909
    return ret;
}

5910
static int qemudDomainGetAutostart(virDomainPtr dom,
5911
                                   int *autostart) {
5912 5913 5914
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
5915

5916
    qemuDriverLock(driver);
5917
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5918 5919
    qemuDriverUnlock(driver);

5920
    if (!vm) {
5921 5922 5923 5924
        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);
5925
        goto cleanup;
5926 5927 5928
    }

    *autostart = vm->autostart;
5929
    ret = 0;
5930

5931
cleanup:
5932 5933
    if (vm)
        virDomainObjUnlock(vm);
5934
    return ret;
5935 5936
}

5937
static int qemudDomainSetAutostart(virDomainPtr dom,
5938
                                   int autostart) {
5939 5940
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
5941 5942
    char *configFile = NULL, *autostartLink = NULL;
    int ret = -1;
5943

5944
    qemuDriverLock(driver);
5945
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5946

5947
    if (!vm) {
5948 5949 5950 5951
        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);
5952
        goto cleanup;
5953 5954
    }

5955 5956 5957
    if (!vm->persistent) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
                         "%s", _("cannot set autostart for transient domain"));
5958
        goto cleanup;
5959 5960
    }

5961 5962
    autostart = (autostart != 0);

5963 5964 5965 5966 5967
    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;
5968

5969 5970
        if (autostart) {
            int err;
5971

5972
            if ((err = virFileMakePath(driver->autostartDir))) {
5973 5974 5975
                virReportSystemError(dom->conn, err,
                                     _("cannot create autostart directory %s"),
                                     driver->autostartDir);
5976 5977
                goto cleanup;
            }
5978

5979
            if (symlink(configFile, autostartLink) < 0) {
5980 5981 5982
                virReportSystemError(dom->conn, errno,
                                     _("Failed to create symlink '%s to '%s'"),
                                     autostartLink, configFile);
5983 5984 5985 5986
                goto cleanup;
            }
        } else {
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
5987 5988 5989
                virReportSystemError(dom->conn, errno,
                                     _("Failed to delete symlink '%s'"),
                                     autostartLink);
5990 5991
                goto cleanup;
            }
5992 5993
        }

5994
        vm->autostart = autostart;
5995
    }
5996
    ret = 0;
5997

5998 5999 6000
cleanup:
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
6001 6002
    if (vm)
        virDomainObjUnlock(vm);
6003
    qemuDriverUnlock(driver);
6004
    return ret;
6005 6006
}

6007 6008 6009 6010 6011

static char *qemuGetSchedulerType(virDomainPtr dom,
                                  int *nparams)
{
    struct qemud_driver *driver = dom->conn->privateData;
6012
    char *ret = NULL;
6013

6014
    qemuDriverLock(driver);
6015
    if (!qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_CPU)) {
6016 6017
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                         __FUNCTION__);
6018
        goto cleanup;
6019 6020 6021 6022 6023 6024 6025 6026
    }

    if (nparams)
        *nparams = 1;

    ret = strdup("posix");
    if (!ret)
        virReportOOMError(dom->conn);
6027 6028 6029

cleanup:
    qemuDriverUnlock(driver);
6030 6031 6032 6033 6034 6035 6036 6037 6038 6039 6040 6041 6042
    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;

6043
    qemuDriverLock(driver);
6044
    if (!qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_CPU)) {
6045 6046
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                         __FUNCTION__);
6047
        goto cleanup;
6048 6049 6050 6051 6052 6053 6054 6055 6056 6057 6058 6059 6060 6061 6062 6063 6064 6065 6066 6067 6068 6069 6070 6071 6072 6073 6074 6075 6076 6077 6078 6079 6080 6081 6082 6083 6084 6085 6086 6087 6088 6089 6090 6091 6092
    }

    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);
6093
    qemuDriverUnlock(driver);
6094 6095 6096 6097 6098 6099 6100 6101 6102 6103 6104 6105 6106 6107
    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;

6108
    qemuDriverLock(driver);
6109
    if (!qemuCgroupControllerActive(driver, VIR_CGROUP_CONTROLLER_CPU)) {
6110 6111
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                         __FUNCTION__);
6112
        goto cleanup;
6113 6114 6115 6116 6117
    }

    if ((*nparams) != 1) {
        qemudReportError(dom->conn, domain, NULL, VIR_ERR_INVALID_ARG,
                         "%s", _("Invalid parameter count"));
6118
        goto cleanup;
6119 6120 6121 6122 6123 6124 6125 6126 6127 6128 6129 6130 6131 6132 6133 6134 6135 6136 6137 6138 6139 6140 6141 6142
    }

    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
6143 6144 6145 6146 6147
    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;
    }
6148 6149 6150 6151 6152 6153 6154

    ret = 0;

cleanup:
    virCgroupFree(&group);
    if (vm)
        virDomainObjUnlock(vm);
6155
    qemuDriverUnlock(driver);
6156 6157 6158 6159
    return ret;
}


6160 6161 6162 6163 6164 6165 6166 6167 6168
/* 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)
{
6169
    struct qemud_driver *driver = dom->conn->privateData;
6170
    char *dummy, *info = NULL;
6171
    const char *p, *eol;
6172
    const char *qemu_dev_name = NULL;
6173
    size_t len;
6174
    int i, ret = -1;
6175
    virDomainObjPtr vm;
6176
    virDomainDiskDefPtr disk = NULL;
6177

6178
    qemuDriverLock(driver);
6179
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
6180
    qemuDriverUnlock(driver);
6181
    if (!vm) {
6182 6183 6184 6185
        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);
6186
        goto cleanup;
6187
    }
6188
    if (!virDomainIsActive (vm)) {
6189
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
6190
                          "%s", _("domain is not running"));
6191
        goto cleanup;
6192 6193
    }

6194 6195 6196 6197 6198 6199 6200 6201
    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (STREQ(path, vm->def->disks[i]->dst)) {
            disk = vm->def->disks[i];
            break;
        }
    }

    if (!disk) {
6202 6203
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
                          _("invalid path: %s"), path);
6204
        goto cleanup;
6205 6206
    }

6207
    qemu_dev_name = qemudDiskDeviceName(dom->conn, disk);
6208
    if (!qemu_dev_name)
6209
        goto cleanup;
6210 6211
    len = strlen (qemu_dev_name);

6212
    if (qemudMonitorCommand (vm, "info blockstats", &info) < 0) {
6213 6214
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                          "%s", _("'info blockstats' command failed"));
6215
        goto cleanup;
6216
    }
6217
    DEBUG ("%s: info blockstats reply: %s", vm->def->name, info);
6218 6219 6220 6221 6222 6223

    /* If the command isn't supported then qemu prints the supported
     * info commands, so the output starts "info ".  Since this is
     * unlikely to be the name of a block device, we can use this
     * to detect if qemu supports the command.
     */
C
Cole Robinson 已提交
6224
    if (strstr(info, "\ninfo ")) {
6225 6226 6227
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                          "%s",
                          _("'info blockstats' not supported by this qemu"));
6228
        goto cleanup;
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
    }

    stats->rd_req = -1;
    stats->rd_bytes = -1;
    stats->wr_req = -1;
    stats->wr_bytes = -1;
    stats->errs = -1;

    /* The output format for both qemu & KVM is:
     *   blockdevice: rd_bytes=% wr_bytes=% rd_operations=% wr_operations=%
     *   (repeated for each block device)
     * where '%' is a 64 bit number.
     */
    p = info;

    while (*p) {
        if (STREQLEN (p, qemu_dev_name, len)
            && p[len] == ':' && p[len+1] == ' ') {

            eol = strchr (p, '\n');
            if (!eol)
                eol = p + strlen (p);

            p += len+2;         /* Skip to first label. */

            while (*p) {
6255
                if (STRPREFIX (p, "rd_bytes=")) {
6256 6257
                    p += 9;
                    if (virStrToLong_ll (p, &dummy, 10, &stats->rd_bytes) == -1)
6258 6259
                        DEBUG ("%s: error reading rd_bytes: %s",
                               vm->def->name, p);
6260
                } else if (STRPREFIX (p, "wr_bytes=")) {
6261 6262
                    p += 9;
                    if (virStrToLong_ll (p, &dummy, 10, &stats->wr_bytes) == -1)
6263 6264
                        DEBUG ("%s: error reading wr_bytes: %s",
                               vm->def->name, p);
6265
                } else if (STRPREFIX (p, "rd_operations=")) {
6266 6267
                    p += 14;
                    if (virStrToLong_ll (p, &dummy, 10, &stats->rd_req) == -1)
6268 6269
                        DEBUG ("%s: error reading rd_req: %s",
                               vm->def->name, p);
6270
                } else if (STRPREFIX (p, "wr_operations=")) {
6271 6272
                    p += 14;
                    if (virStrToLong_ll (p, &dummy, 10, &stats->wr_req) == -1)
6273 6274
                        DEBUG ("%s: error reading wr_req: %s",
                               vm->def->name, p);
6275
                } else
6276
                    DEBUG ("%s: unknown block stat near %s", vm->def->name, p);
6277 6278 6279 6280 6281 6282

                /* Skip to next label. */
                p = strchr (p, ' ');
                if (!p || p >= eol) break;
                p++;
            }
6283
            ret = 0;
6284
            goto cleanup;
6285 6286 6287 6288 6289 6290 6291 6292 6293 6294 6295
        }

        /* Skip to next line. */
        p = strchr (p, '\n');
        if (!p) break;
        p++;
    }

    /* If we reach here then the device was not found. */
    qemudReportError (dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
                      _("device not found: %s (%s)"), path, qemu_dev_name);
6296
 cleanup:
6297 6298
    VIR_FREE(qemu_dev_name);
    VIR_FREE(info);
6299 6300
    if (vm)
        virDomainObjUnlock(vm);
6301
    return ret;
6302 6303
}

6304
#ifdef __linux__
6305 6306 6307 6308 6309
static int
qemudDomainInterfaceStats (virDomainPtr dom,
                           const char *path,
                           struct _virDomainInterfaceStats *stats)
{
6310 6311
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
6312
    int i;
6313
    int ret = -1;
6314

6315
    qemuDriverLock(driver);
6316
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
6317 6318
    qemuDriverUnlock(driver);

6319
    if (!vm) {
6320 6321 6322 6323
        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);
6324
        goto cleanup;
6325 6326
    }

6327
    if (!virDomainIsActive(vm)) {
6328
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
6329
                         "%s", _("domain is not running"));
6330
        goto cleanup;
6331 6332 6333
    }

    /* Check the path is one of the domain's network interfaces. */
6334 6335
    for (i = 0 ; i < vm->def->nnets ; i++) {
        if (vm->def->nets[i]->ifname &&
6336 6337 6338 6339
            STREQ (vm->def->nets[i]->ifname, path)) {
            ret = 0;
            break;
        }
6340 6341
    }

6342 6343 6344 6345 6346
    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);
6347

6348
cleanup:
6349 6350
    if (vm)
        virDomainObjUnlock(vm);
6351 6352
    return ret;
}
6353
#else
6354 6355 6356 6357
static int
qemudDomainInterfaceStats (virDomainPtr dom,
                           const char *path ATTRIBUTE_UNUSED,
                           struct _virDomainInterfaceStats *stats ATTRIBUTE_UNUSED)
6358 6359 6360 6361
    qemudReportError (dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                      "%s", __FUNCTION__);
    return -1;
}
6362
#endif
6363

6364 6365 6366 6367 6368 6369 6370
static int
qemudDomainBlockPeek (virDomainPtr dom,
                      const char *path,
                      unsigned long long offset, size_t size,
                      void *buffer,
                      unsigned int flags ATTRIBUTE_UNUSED)
{
6371 6372 6373
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int fd = -1, ret = -1, i;
6374

6375
    qemuDriverLock(driver);
6376
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
6377 6378
    qemuDriverUnlock(driver);

6379
    if (!vm) {
6380 6381 6382 6383
        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);
6384
        goto cleanup;
6385 6386 6387 6388
    }

    if (!path || path[0] == '\0') {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
J
Jim Meyering 已提交
6389
                         "%s", _("NULL or empty path"));
6390
        goto cleanup;
6391 6392 6393
    }

    /* Check the path belongs to this domain. */
6394 6395
    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (vm->def->disks[i]->src != NULL &&
6396 6397 6398 6399
            STREQ (vm->def->disks[i]->src, path)) {
            ret = 0;
            break;
        }
6400 6401
    }

6402 6403 6404 6405 6406
    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) {
6407 6408
            virReportSystemError (dom->conn, errno,
                                  _("%s: failed to open"), path);
6409 6410
            goto cleanup;
        }
6411

6412 6413 6414 6415 6416 6417
        /* 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) {
6418 6419
            virReportSystemError (dom->conn, errno,
                                  _("%s: failed to seek or read"), path);
6420 6421 6422 6423 6424 6425 6426
            goto cleanup;
        }

        ret = 0;
    } else {
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
                          "%s", _("invalid path"));
6427 6428
    }

6429 6430 6431
cleanup:
    if (fd >= 0)
        close (fd);
6432 6433
    if (vm)
        virDomainObjUnlock(vm);
6434 6435 6436
    return ret;
}

R
Richard W.M. Jones 已提交
6437 6438 6439 6440 6441 6442
static int
qemudDomainMemoryPeek (virDomainPtr dom,
                       unsigned long long offset, size_t size,
                       void *buffer,
                       unsigned int flags)
{
6443 6444 6445
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    char cmd[256], *info = NULL;
6446
    char *tmp = NULL;
R
Richard W.M. Jones 已提交
6447 6448
    int fd = -1, ret = -1;

6449
    qemuDriverLock(driver);
6450
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
6451
    qemuDriverUnlock(driver);
R
Richard W.M. Jones 已提交
6452 6453

    if (!vm) {
6454 6455 6456 6457
        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);
6458 6459 6460
        goto cleanup;
    }

6461
    if (flags != VIR_MEMORY_VIRTUAL && flags != VIR_MEMORY_PHYSICAL) {
6462
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
6463
                     "%s", _("flags parameter must be VIR_MEMORY_VIRTUAL or VIR_MEMORY_PHYSICAL"));
6464
        goto cleanup;
R
Richard W.M. Jones 已提交
6465 6466
    }

6467
    if (!virDomainIsActive(vm)) {
6468
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
R
Richard W.M. Jones 已提交
6469
                         "%s", _("domain is not running"));
6470
        goto cleanup;
R
Richard W.M. Jones 已提交
6471 6472
    }

6473 6474 6475 6476 6477
    if (virAsprintf(&tmp, driver->cacheDir,  "/qemu.mem.XXXXXX") < 0) {
        virReportOOMError(dom->conn);
        goto cleanup;
    }

R
Richard W.M. Jones 已提交
6478 6479
    /* Create a temporary filename. */
    if ((fd = mkstemp (tmp)) == -1) {
6480 6481
        virReportSystemError (dom->conn, errno,
                              _("mkstemp(\"%s\") failed"), tmp);
6482
        goto cleanup;
R
Richard W.M. Jones 已提交
6483 6484
    }

6485 6486 6487 6488 6489 6490 6491
    if (flags == VIR_MEMORY_VIRTUAL)
        /* Issue the memsave command. */
        snprintf (cmd, sizeof cmd, "memsave %llu %zi \"%s\"", offset, size, tmp);
    else
        /* Issue the pmemsave command. */
        snprintf (cmd, sizeof cmd, "pmemsave %llu %zi \"%s\"", offset, size, tmp);

6492
    if (qemudMonitorCommand (vm, cmd, &info) < 0) {
R
Richard W.M. Jones 已提交
6493
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
6494
                          "%s", _("'memsave' command failed"));
6495
        goto cleanup;
R
Richard W.M. Jones 已提交
6496 6497
    }

6498
    DEBUG ("%s: (p)memsave reply: %s", vm->def->name, info);
R
Richard W.M. Jones 已提交
6499 6500 6501

    /* Read the memory file into buffer. */
    if (saferead (fd, buffer, size) == (ssize_t) -1) {
6502 6503 6504
        virReportSystemError (dom->conn, errno,
                              _("failed to read temporary file "
                                "created with template %s"), tmp);
6505
        goto cleanup;
R
Richard W.M. Jones 已提交
6506 6507 6508
    }

    ret = 0;
6509 6510

cleanup:
6511
    VIR_FREE(tmp);
6512
    VIR_FREE(info);
R
Richard W.M. Jones 已提交
6513 6514
    if (fd >= 0) close (fd);
    unlink (tmp);
6515 6516
    if (vm)
        virDomainObjUnlock(vm);
R
Richard W.M. Jones 已提交
6517 6518 6519
    return ret;
}

6520

6521 6522
static int
qemudDomainEventRegister (virConnectPtr conn,
6523
                          virConnectDomainEventCallback callback,
6524 6525
                          void *opaque,
                          virFreeCallback freecb)
6526
{
6527 6528 6529
    struct qemud_driver *driver = conn->privateData;
    int ret;

6530
    qemuDriverLock(driver);
6531 6532
    ret = virDomainEventCallbackListAdd(conn, driver->domainEventCallbacks,
                                        callback, opaque, freecb);
6533
    qemuDriverUnlock(driver);
6534

6535
    return ret;
6536 6537 6538 6539
}

static int
qemudDomainEventDeregister (virConnectPtr conn,
6540
                            virConnectDomainEventCallback callback)
6541
{
6542 6543 6544
    struct qemud_driver *driver = conn->privateData;
    int ret;

6545
    qemuDriverLock(driver);
6546 6547 6548 6549 6550 6551
    if (driver->domainEventDispatching)
        ret = virDomainEventCallbackListMarkDelete(conn, driver->domainEventCallbacks,
                                                   callback);
    else
        ret = virDomainEventCallbackListRemove(conn, driver->domainEventCallbacks,
                                               callback);
6552
    qemuDriverUnlock(driver);
6553

6554
    return ret;
6555 6556
}

6557 6558 6559 6560 6561
static void qemuDomainEventDispatchFunc(virConnectPtr conn,
                                        virDomainEventPtr event,
                                        virConnectDomainEventCallback cb,
                                        void *cbopaque,
                                        void *opaque)
6562
{
6563
    struct qemud_driver *driver = opaque;
6564

6565 6566 6567 6568 6569 6570 6571 6572 6573 6574 6575 6576 6577 6578 6579 6580 6581 6582 6583 6584 6585 6586 6587 6588 6589 6590 6591 6592 6593 6594 6595 6596 6597 6598 6599 6600 6601 6602 6603 6604 6605 6606 6607 6608
    /* 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);
6609 6610
}

D
Daniel Veillard 已提交
6611 6612 6613 6614 6615 6616 6617 6618 6619 6620 6621 6622 6623 6624 6625 6626 6627 6628
/* 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;
6629 6630
    struct qemud_driver *driver = dconn->privateData;
    virDomainDefPtr def = NULL;
D
Daniel Veillard 已提交
6631 6632
    virDomainObjPtr vm = NULL;
    int this_port;
6633
    char *hostname;
D
Daniel Veillard 已提交
6634 6635
    char migrateFrom [64];
    const char *p;
6636
    virDomainEventPtr event = NULL;
6637
    int ret = -1;
6638
    int internalret;
6639 6640

    *uri_out = NULL;
D
Daniel Veillard 已提交
6641

6642
    qemuDriverLock(driver);
D
Daniel Veillard 已提交
6643 6644 6645
    if (!dom_xml) {
        qemudReportError (dconn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                          "%s", _("no domain XML passed"));
6646
        goto cleanup;
D
Daniel Veillard 已提交
6647 6648 6649 6650 6651 6652 6653 6654 6655 6656 6657 6658 6659 6660 6661 6662 6663
    }

    /* 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 */
6664
        if ((hostname = virGetHostname()) == NULL) {
6665 6666
            virReportSystemError (dconn, errno,
                                  "%s", _("failed to determine host name"));
6667
            goto cleanup;
D
Daniel Veillard 已提交
6668 6669 6670
        }

        /* Caller frees */
6671 6672 6673
        internalret = virAsprintf(uri_out, "tcp:%s:%d", hostname, this_port);
        VIR_FREE(hostname);
        if (internalret < 0) {
6674
            virReportOOMError (dconn);
6675
            goto cleanup;
D
Daniel Veillard 已提交
6676 6677 6678 6679 6680 6681
        }
    } 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.
         */
6682
        if (!STRPREFIX (uri_in, "tcp:")) {
D
Daniel Veillard 已提交
6683
            qemudReportError (dconn, NULL, NULL, VIR_ERR_INVALID_ARG,
6684
                  "%s", _("only tcp URIs are supported for KVM/QEMU migrations"));
6685
            goto cleanup;
D
Daniel Veillard 已提交
6686 6687 6688 6689 6690 6691 6692 6693 6694
        }

        /* 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"));
6695
            goto cleanup;
D
Daniel Veillard 已提交
6696 6697 6698 6699
        }
    }

    /* Parse the domain XML. */
6700 6701
    if (!(def = virDomainDefParseString(dconn, driver->caps, dom_xml,
                                        VIR_DOMAIN_XML_INACTIVE))) {
D
Daniel Veillard 已提交
6702 6703
        qemudReportError (dconn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                          "%s", _("failed to parse XML"));
6704
        goto cleanup;
D
Daniel Veillard 已提交
6705 6706 6707 6708 6709 6710 6711 6712 6713 6714 6715 6716 6717 6718 6719 6720
    }

    /* 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"));
6721
        goto cleanup;
D
Daniel Veillard 已提交
6722 6723 6724 6725 6726 6727 6728 6729 6730
    }
#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);
6731
            goto cleanup;
D
Daniel Veillard 已提交
6732
        }
6733
        virDomainObjUnlock(vm);
D
Daniel Veillard 已提交
6734 6735 6736 6737 6738 6739 6740
    }

    if (!(vm = virDomainAssignDef(dconn,
                                  &driver->domains,
                                  def))) {
        qemudReportError (dconn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                          "%s", _("failed to assign new VM"));
6741
        goto cleanup;
D
Daniel Veillard 已提交
6742
    }
6743
    def = NULL;
D
Daniel Veillard 已提交
6744 6745 6746 6747 6748 6749 6750 6751

    /* 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);
6752
    if (qemudStartVMDaemon (dconn, driver, vm, migrateFrom, -1) < 0) {
6753 6754 6755
        /* Note that we don't set an error here because qemudStartVMDaemon
         * should have already done that.
         */
6756
        if (!vm->persistent) {
D
Daniel Veillard 已提交
6757
            virDomainRemoveInactive(&driver->domains, vm);
6758 6759
            vm = NULL;
        }
6760
        goto cleanup;
D
Daniel Veillard 已提交
6761
    }
6762 6763 6764 6765

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_MIGRATED);
6766
    ret = 0;
D
Daniel Veillard 已提交
6767

6768 6769 6770 6771 6772
cleanup:
    virDomainDefFree(def);
    if (ret != 0) {
        VIR_FREE(*uri_out);
    }
6773 6774
    if (vm)
        virDomainObjUnlock(vm);
6775 6776
    if (event)
        qemuDomainEventQueue(driver, event);
6777
    qemuDriverUnlock(driver);
6778
    return ret;
D
Daniel Veillard 已提交
6779 6780 6781 6782 6783 6784 6785 6786
}

/* 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,
6787
                           unsigned long flags,
D
Daniel Veillard 已提交
6788 6789 6790
                           const char *dname ATTRIBUTE_UNUSED,
                           unsigned long resource)
{
6791 6792
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
6793
    virDomainEventPtr event = NULL;
D
Daniel Veillard 已提交
6794 6795
    char *safe_uri;
    char cmd[HOST_NAME_MAX+50];
6796 6797
    char *info = NULL;
    int ret = -1;
6798
    int paused = 0;
D
Daniel Veillard 已提交
6799

6800
    qemuDriverLock(driver);
6801
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
D
Daniel Veillard 已提交
6802
    if (!vm) {
6803 6804 6805 6806
        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);
6807
        goto cleanup;
D
Daniel Veillard 已提交
6808 6809 6810
    }

    if (!virDomainIsActive(vm)) {
6811
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
D
Daniel Veillard 已提交
6812
                          "%s", _("domain is not running"));
6813
        goto cleanup;
D
Daniel Veillard 已提交
6814 6815
    }

6816 6817
    if (!(flags & VIR_MIGRATE_LIVE)) {
        /* Pause domain for non-live migration */
6818 6819 6820 6821 6822
        if (qemudMonitorCommand (vm, "stop", &info) < 0) {
            qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                             "%s", _("off-line migration specified, but suspend operation failed"));
            goto cleanup;
        }
6823
        DEBUG ("%s: stop reply: %s", vm->def->name, info);
6824
        VIR_FREE(info);
6825
        paused = 1;
6826

6827 6828 6829 6830 6831 6832
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_SUSPENDED,
                                         VIR_DOMAIN_EVENT_SUSPENDED_MIGRATED);
        if (event)
            qemuDomainEventQueue(driver, event);
        event = NULL;
6833 6834
    }

D
Daniel Veillard 已提交
6835 6836 6837
    if (resource > 0) {
        /* Issue migrate_set_speed command.  Don't worry if it fails. */
        snprintf (cmd, sizeof cmd, "migrate_set_speed %lum", resource);
6838
        qemudMonitorCommand (vm, cmd, &info);
D
Daniel Veillard 已提交
6839

6840
        DEBUG ("%s: migrate_set_speed reply: %s", vm->def->name, info);
D
Daniel Veillard 已提交
6841 6842 6843 6844 6845 6846
        VIR_FREE (info);
    }

    /* Issue the migrate command. */
    safe_uri = qemudEscapeMonitorArg (uri);
    if (!safe_uri) {
6847
        virReportOOMError (dom->conn);
6848
        goto cleanup;
D
Daniel Veillard 已提交
6849 6850 6851 6852
    }
    snprintf (cmd, sizeof cmd, "migrate \"%s\"", safe_uri);
    VIR_FREE (safe_uri);

6853
    if (qemudMonitorCommand (vm, cmd, &info) < 0) {
D
Daniel Veillard 已提交
6854 6855
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                          "%s", _("migrate operation failed"));
6856
        goto cleanup;
D
Daniel Veillard 已提交
6857 6858
    }

6859
    DEBUG ("%s: migrate reply: %s", vm->def->name, info);
D
Daniel Veillard 已提交
6860 6861 6862 6863 6864

    /* Now check for "fail" in the output string */
    if (strstr(info, "fail") != NULL) {
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                          _("migrate failed: %s"), info);
6865
        goto cleanup;
D
Daniel Veillard 已提交
6866 6867
    }

6868 6869 6870 6871 6872 6873 6874 6875 6876 6877 6878 6879 6880 6881 6882
    /* it is also possible that the migrate didn't fail initially, but
     * rather failed later on.  Check the output of "info migrate"
     */
    VIR_FREE(info);
    if (qemudMonitorCommand(vm, "info migrate", &info) < 0) {
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                          "%s", _("could not get info about migration"));
        goto cleanup;
    }
    if (strstr(info, "fail") != NULL) {
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                          _("migrate failed: %s"), info);
        goto cleanup;
    }

D
Daniel Veillard 已提交
6883 6884
    /* Clean up the source domain. */
    qemudShutdownVMDaemon (dom->conn, driver, vm);
6885
    paused = 0;
6886 6887 6888 6889

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_MIGRATED);
6890
    if (!vm->persistent) {
D
Daniel Veillard 已提交
6891
        virDomainRemoveInactive(&driver->domains, vm);
6892 6893
        vm = NULL;
    }
6894
    ret = 0;
D
Daniel Veillard 已提交
6895

6896
cleanup:
6897 6898
    if (paused) {
        /* we got here through some sort of failure; start the domain again */
6899
        if (qemudMonitorSendCont(dom->conn, vm) < 0) {
6900 6901 6902 6903
            /* 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
             */
6904 6905
            VIR_ERROR(_("Failed to resume guest %s after failure\n"),
                      vm->def->name);
6906 6907 6908 6909 6910 6911 6912
        }

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

6913
    VIR_FREE(info);
6914 6915
    if (vm)
        virDomainObjUnlock(vm);
6916 6917
    if (event)
        qemuDomainEventQueue(driver, event);
6918
    qemuDriverUnlock(driver);
6919
    return ret;
D
Daniel Veillard 已提交
6920 6921 6922 6923 6924 6925 6926 6927 6928 6929 6930 6931
}

/* 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)
{
6932 6933 6934
    struct qemud_driver *driver = dconn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
6935
    virDomainEventPtr event = NULL;
D
Daniel Veillard 已提交
6936

6937
    qemuDriverLock(driver);
6938
    vm = virDomainFindByName(&driver->domains, dname);
D
Daniel Veillard 已提交
6939
    if (!vm) {
6940 6941
        qemudReportError (dconn, NULL, NULL, VIR_ERR_NO_DOMAIN,
                          _("no domain with matching name '%s'"), dname);
6942
        goto cleanup;
D
Daniel Veillard 已提交
6943 6944 6945 6946 6947 6948 6949
    }

    /* 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);
6950 6951 6952 6953 6954

        /* 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
         */
6955 6956 6957 6958
        if (qemudMonitorSendCont(dconn, vm) < 0) {
            if (virGetLastError() == NULL)
                qemudReportError(dconn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                                 "%s", _("resume operation failed"));
6959 6960 6961
            goto cleanup;
        }

D
Daniel Veillard 已提交
6962
        vm->state = VIR_DOMAIN_RUNNING;
6963 6964 6965
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_MIGRATED);
D
Daniel Veillard 已提交
6966
        virDomainSaveStatus(dconn, driver->stateDir, vm);
D
Daniel Veillard 已提交
6967 6968
    } else {
        qemudShutdownVMDaemon (dconn, driver, vm);
6969 6970 6971
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         VIR_DOMAIN_EVENT_STOPPED_FAILED);
6972
        if (!vm->persistent) {
D
Daniel Veillard 已提交
6973
            virDomainRemoveInactive(&driver->domains, vm);
6974 6975
            vm = NULL;
        }
D
Daniel Veillard 已提交
6976
    }
6977 6978

cleanup:
6979 6980
    if (vm)
        virDomainObjUnlock(vm);
6981 6982
    if (event)
        qemuDomainEventQueue(driver, event);
6983
    qemuDriverUnlock(driver);
6984
    return dom;
D
Daniel Veillard 已提交
6985 6986
}

6987 6988 6989 6990 6991 6992 6993 6994 6995 6996 6997 6998 6999 7000 7001 7002
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;

7003
    def = virNodeDeviceDefParseString(dev->conn, xml, EXISTING_DEVICE);
7004 7005 7006 7007 7008 7009 7010 7011 7012 7013 7014 7015 7016 7017 7018 7019 7020 7021 7022 7023 7024 7025 7026 7027 7028 7029 7030 7031 7032 7033 7034 7035 7036 7037 7038 7039 7040 7041 7042 7043 7044 7045 7046 7047 7048 7049 7050 7051 7052 7053 7054 7055 7056 7057 7058 7059 7060 7061 7062 7063 7064 7065 7066 7067 7068 7069 7070 7071 7072 7073 7074 7075 7076 7077 7078 7079 7080 7081
    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)
{
7082
    struct qemud_driver *driver = dev->conn->privateData;
7083 7084 7085 7086 7087 7088 7089 7090 7091 7092 7093
    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;

7094 7095 7096
    qemuDriverLock(driver);

    if (pciResetDevice(dev->conn, pci, driver->activePciHostdevs) < 0)
7097 7098 7099 7100
        goto out;

    ret = 0;
out:
7101
    qemuDriverUnlock(driver);
7102 7103 7104 7105
    pciFreeDevice(dev->conn, pci);
    return ret;
}

7106 7107 7108 7109 7110
static virDriver qemuDriver = {
    VIR_DRV_QEMU,
    "QEMU",
    qemudOpen, /* open */
    qemudClose, /* close */
D
Daniel Veillard 已提交
7111
    qemudSupportsFeature, /* supports_feature */
7112 7113
    qemudGetType, /* type */
    qemudGetVersion, /* version */
7114
    qemudGetHostname, /* getHostname */
7115
    qemudGetMaxVCPUs, /* getMaxVcpus */
7116
    nodeGetInfo, /* nodeGetInfo */
7117 7118 7119
    qemudGetCapabilities, /* getCapabilities */
    qemudListDomains, /* listDomains */
    qemudNumDomains, /* numOfDomains */
7120
    qemudDomainCreate, /* domainCreateXML */
7121 7122 7123 7124 7125
    qemudDomainLookupByID, /* domainLookupByID */
    qemudDomainLookupByUUID, /* domainLookupByUUID */
    qemudDomainLookupByName, /* domainLookupByName */
    qemudDomainSuspend, /* domainSuspend */
    qemudDomainResume, /* domainResume */
7126
    qemudDomainShutdown, /* domainShutdown */
7127 7128 7129
    NULL, /* domainReboot */
    qemudDomainDestroy, /* domainDestroy */
    qemudDomainGetOSType, /* domainGetOSType */
7130 7131 7132
    qemudDomainGetMaxMemory, /* domainGetMaxMemory */
    qemudDomainSetMaxMemory, /* domainSetMaxMemory */
    qemudDomainSetMemory, /* domainSetMemory */
7133 7134 7135
    qemudDomainGetInfo, /* domainGetInfo */
    qemudDomainSave, /* domainSave */
    qemudDomainRestore, /* domainRestore */
P
Paolo Bonzini 已提交
7136
    qemudDomainCoreDump, /* domainCoreDump */
7137
    qemudDomainSetVcpus, /* domainSetVcpus */
7138 7139 7140 7141
#if HAVE_SCHED_GETAFFINITY
    qemudDomainPinVcpu, /* domainPinVcpu */
    qemudDomainGetVcpus, /* domainGetVcpus */
#else
7142 7143
    NULL, /* domainPinVcpu */
    NULL, /* domainGetVcpus */
7144
#endif
7145
    qemudDomainGetMaxVcpus, /* domainGetMaxVcpus */
7146 7147
    qemudDomainGetSecurityLabel, /* domainGetSecurityLabel */
    qemudNodeGetSecurityModel, /* nodeGetSecurityModel */
7148
    qemudDomainDumpXML, /* domainDumpXML */
7149
    qemuDomainXMLFromNative, /* domainXmlFromNative */
7150
    qemuDomainXMLToNative, /* domainXMLToNative */
7151 7152
    qemudListDefinedDomains, /* listDefinedDomains */
    qemudNumDefinedDomains, /* numOfDefinedDomains */
7153 7154 7155
    qemudDomainStart, /* domainCreate */
    qemudDomainDefine, /* domainDefineXML */
    qemudDomainUndefine, /* domainUndefine */
7156
    qemudDomainAttachDevice, /* domainAttachDevice */
7157
    qemudDomainDetachDevice, /* domainDetachDevice */
7158 7159
    qemudDomainGetAutostart, /* domainGetAutostart */
    qemudDomainSetAutostart, /* domainSetAutostart */
7160 7161 7162
    qemuGetSchedulerType, /* domainGetSchedulerType */
    qemuGetSchedulerParameters, /* domainGetSchedulerParameters */
    qemuSetSchedulerParameters, /* domainSetSchedulerParameters */
D
Daniel Veillard 已提交
7163 7164
    NULL, /* domainMigratePrepare (v1) */
    qemudDomainMigratePerform, /* domainMigratePerform */
7165
    NULL, /* domainMigrateFinish */
7166
    qemudDomainBlockStats, /* domainBlockStats */
7167
    qemudDomainInterfaceStats, /* domainInterfaceStats */
7168
    qemudDomainBlockPeek, /* domainBlockPeek */
R
Richard W.M. Jones 已提交
7169
    qemudDomainMemoryPeek, /* domainMemoryPeek */
7170 7171
    nodeGetCellsFreeMemory, /* nodeGetCellsFreeMemory */
    nodeGetFreeMemory,  /* getFreeMemory */
7172 7173
    qemudDomainEventRegister, /* domainEventRegister */
    qemudDomainEventDeregister, /* domainEventDeregister */
D
Daniel Veillard 已提交
7174 7175
    qemudDomainMigratePrepare2, /* domainMigratePrepare2 */
    qemudDomainMigrateFinish2, /* domainMigrateFinish2 */
7176 7177 7178
    qemudNodeDeviceDettach, /* nodeDeviceDettach */
    qemudNodeDeviceReAttach, /* nodeDeviceReAttach */
    qemudNodeDeviceReset, /* nodeDeviceReset */
7179 7180 7181
};


7182
static virStateDriver qemuStateDriver = {
7183 7184 7185 7186
    .initialize = qemudStartup,
    .cleanup = qemudShutdown,
    .reload = qemudReload,
    .active = qemudActive,
7187
};
7188

7189
int qemuRegister(void) {
7190 7191 7192 7193
    virRegisterDriver(&qemuDriver);
    virRegisterStateDriver(&qemuStateDriver);
    return 0;
}