qemu_driver.c 177.8 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 31 32 33 34 35 36
#include <sys/types.h>
#include <sys/poll.h>
#include <dirent.h>
#include <limits.h>
#include <string.h>
#include <stdio.h>
#include <strings.h>
#include <stdarg.h>
#include <stdlib.h>
#include <unistd.h>
#include <errno.h>
37
#include <sys/utsname.h>
38 39 40 41
#include <sys/stat.h>
#include <fcntl.h>
#include <signal.h>
#include <paths.h>
42 43
#include <pwd.h>
#include <stdio.h>
44
#include <sys/wait.h>
45
#include <sys/ioctl.h>
D
Daniel P. Berrange 已提交
46

47 48 49 50
#if HAVE_SCHED_H
#include <sched.h>
#endif

51
#include "virterror_internal.h"
52
#include "logging.h"
53
#include "datatypes.h"
54 55
#include "qemu_driver.h"
#include "qemu_conf.h"
J
Jim Meyering 已提交
56
#include "c-ctype.h"
57
#include "event.h"
58
#include "buf.h"
59
#include "util.h"
60
#include "nodeinfo.h"
61
#include "stats_linux.h"
62
#include "capabilities.h"
63
#include "memory.h"
64
#include "uuid.h"
65
#include "domain_conf.h"
66 67
#include "node_device_conf.h"
#include "pci.h"
68 69
#include "security.h"

70

71 72
#define VIR_FROM_THIS VIR_FROM_QEMU

R
Richard W.M. Jones 已提交
73
/* For storing short-lived temporary files. */
74
#define TEMPDIR LOCAL_STATE_DIR "/cache/libvirt/qemu"
R
Richard W.M. Jones 已提交
75

76 77 78
#define QEMU_CMD_PROMPT "\n(qemu) "
#define QEMU_PASSWD_PROMPT "Password: "

79 80
static int qemudShutdown(void);

81 82
static void qemuDriverLock(struct qemud_driver *driver)
{
83
    virMutexLock(&driver->lock);
84 85 86
}
static void qemuDriverUnlock(struct qemud_driver *driver)
{
87
    virMutexUnlock(&driver->lock);
88 89
}

90 91 92
static void qemuDomainEventFlush(int timer, void *opaque);
static void qemuDomainEventQueue(struct qemud_driver *driver,
                                 virDomainEventPtr event);
93

94 95
static void qemudDispatchVMEvent(int watch,
                                 int fd,
96 97 98
                                 int events,
                                 void *opaque);

99 100
static int qemudStartVMDaemon(virConnectPtr conn,
                              struct qemud_driver *driver,
101
                              virDomainObjPtr vm,
102 103
                              const char *migrateFrom,
                              int stdin_fd);
104

105 106
static void qemudShutdownVMDaemon(virConnectPtr conn,
                                  struct qemud_driver *driver,
107
                                  virDomainObjPtr vm);
108

109
static int qemudDomainGetMaxVcpus(virDomainPtr dom);
110

111 112 113 114 115 116 117 118
static int qemudMonitorCommand(const virDomainObjPtr vm,
                               const char *cmd,
                               char **reply);
static int qemudMonitorCommandExtra(const virDomainObjPtr vm,
                                    const char *cmd,
                                    const char *extra,
                                    const char *extraPrompt,
                                    char **reply);
119 120 121
static int qemudDomainSetMemoryBalloon(virConnectPtr conn,
                                       virDomainObjPtr vm,
                                       unsigned long newmem);
122 123
static int qemudDetectVcpuPIDs(virConnectPtr conn,
                               virDomainObjPtr vm);
124

J
Jim Meyering 已提交
125
static struct qemud_driver *qemu_driver = NULL;
126 127


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

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

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

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


165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181
static int
qemudLogReadFD(virConnectPtr conn, const char* logDir, const char* name, off_t pos)
{
    char logfile[PATH_MAX];
    mode_t logmode = O_RDONLY;
    int ret, fd = -1;

    if ((ret = snprintf(logfile, sizeof(logfile), "%s/%s.log", logDir, name))
        < 0 || ret >= sizeof(logfile)) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                         _("failed to build logfile name %s/%s.log"),
                         logDir, name);
        return -1;
    }


    if ((fd = open(logfile, logmode)) < 0) {
182 183 184
        virReportSystemError(conn, errno,
                             _("failed to create logfile %s"),
                             logfile);
185 186
        return -1;
    }
187
    if (virSetCloseExec(fd) < 0) {
188 189
        virReportSystemError(conn, errno, "%s",
                             _("Unable to set VM logfile close-on-exec flag"));
190 191 192 193
        close(fd);
        return -1;
    }
    if (lseek(fd, pos, SEEK_SET) < 0) {
194 195 196
        virReportSystemError(conn, errno,
                             _("Unable to seek to %lld in %s"),
                             (long long) pos, logfile);
197 198 199 200 201 202
        close(fd);
    }
    return fd;
}


203 204 205
static void
qemudAutostartConfigs(struct qemud_driver *driver) {
    unsigned int i;
206 207 208 209 210
    /* XXX: Figure out a better way todo this. The domain
     * startup code needs a connection handle in order
     * to lookup the bridge associated with a virtual
     * network
     */
211 212 213
    virConnectPtr conn = virConnectOpen(driver->privileged ?
                                        "qemu:///system" :
                                        "qemu:///session");
214
    /* Ignoring NULL conn which is mostly harmless here */
215

216
    qemuDriverLock(driver);
217
    for (i = 0 ; i < driver->domains.count ; i++) {
218
        virDomainObjPtr vm = driver->domains.objs[i];
219
        virDomainObjLock(vm);
220 221
        if (vm->autostart &&
            !virDomainIsActive(vm)) {
222
            int ret = qemudStartVMDaemon(conn, driver, vm, NULL, -1);
223 224
            if (ret < 0) {
                virErrorPtr err = virGetLastError();
225 226
                VIR_ERROR(_("Failed to autostart VM '%s': %s\n"),
                          vm->def->name,
227
                          err ? err->message : "");
228
            } else {
229 230 231 232 233 234
                virDomainEventPtr event =
                    virDomainEventNewFromObj(vm,
                                             VIR_DOMAIN_EVENT_STARTED,
                                             VIR_DOMAIN_EVENT_STARTED_BOOTED);
                if (event)
                    qemuDomainEventQueue(driver, event);
235
            }
236
        }
237
        virDomainObjUnlock(vm);
238
    }
239
    qemuDriverUnlock(driver);
240

241 242
    if (conn)
        virConnectClose(conn);
243 244
}

245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261

/**
 * 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) {
262
        virReportOOMError(conn);
263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281
        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;
}


282 283 284 285 286
static int qemudOpenMonitor(virConnectPtr conn,
                            struct qemud_driver* driver,
                            virDomainObjPtr vm,
                            int reconnect);

287 288 289 290

/*
 * Open an existing VM's monitor, re-detect VCPU threads
 * and re-reserve the security labels in use
291 292
 */
static int
293 294
qemuReconnectDomain(struct qemud_driver *driver,
                    virDomainObjPtr obj)
295
{
296
    int rc;
297

298
    if ((rc = qemudOpenMonitor(NULL, driver, obj, 1)) != 0) {
299 300 301 302
        VIR_ERROR(_("Failed to reconnect monitor for %s: %d\n"),
                  obj->def->name, rc);
        goto error;
    }
303

304 305 306
    if (qemudDetectVcpuPIDs(NULL, obj) < 0) {
        goto error;
    }
307

308 309 310 311 312
    if (obj->def->seclabel.type == VIR_DOMAIN_SECLABEL_DYNAMIC &&
        driver->securityDriver &&
        driver->securityDriver->domainReserveSecurityLabel &&
        driver->securityDriver->domainReserveSecurityLabel(NULL, obj) < 0)
        return -1;
313

314 315
    if (obj->def->id >= driver->nextvmid)
        driver->nextvmid = obj->def->id + 1;
316

317
    return 0;
318

319 320 321
error:
    return -1;
}
322

323 324 325 326 327 328 329 330 331 332
/**
 * qemudReconnectVMs
 *
 * Try to re-open the resources for live VMs that we care
 * about.
 */
static void
qemuReconnectDomains(struct qemud_driver *driver)
{
    int i;
333

334 335 336 337 338 339 340 341 342
    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);
343
        }
344
        virDomainObjUnlock(obj);
345 346 347
    }
}

348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380

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;
}


381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400
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;

401
    VIR_INFO("Initialized security driver %s", security_drv->name);
402 403 404 405 406

    /*
     * Add security policy host caps now that the security driver is
     * initialized.
     */
407 408
    return qemudSecurityCapsInit(security_drv, qemud_drv->caps);
}
409 410


411

412 413 414 415 416 417
/**
 * qemudStartup:
 *
 * Initialization function for the QEmu daemon
 */
static int
418
qemudStartup(int privileged) {
419
    char *base = NULL;
D
Daniel P. Berrange 已提交
420
    char driverConf[PATH_MAX];
421

422
    if (VIR_ALLOC(qemu_driver) < 0)
423 424
        return -1;

425
    if (virMutexInit(&qemu_driver->lock) < 0) {
426
        VIR_ERROR("%s", _("cannot initialize mutex"));
427 428 429
        VIR_FREE(qemu_driver);
        return -1;
    }
430
    qemuDriverLock(qemu_driver);
431
    qemu_driver->privileged = privileged;
432

433 434 435
    /* Don't have a dom0 so start from 1 */
    qemu_driver->nextvmid = 1;

436 437
    /* Init callback list */
    if(VIR_ALLOC(qemu_driver->domainEventCallbacks) < 0)
438
        goto out_of_memory;
439 440 441 442 443 444
    if (!(qemu_driver->domainEventQueue = virDomainEventQueueNew()))
        goto out_of_memory;

    if ((qemu_driver->domainEventTimer =
         virEventAddTimeout(-1, qemuDomainEventFlush, qemu_driver, NULL)) < 0)
        goto error;
445

446
    if (privileged) {
447 448
        if (virAsprintf(&qemu_driver->logDir,
                        "%s/log/libvirt/qemu", LOCAL_STATE_DIR) == -1)
449
            goto out_of_memory;
450

D
Daniel P. Berrange 已提交
451
        if ((base = strdup (SYSCONF_DIR "/libvirt")) == NULL)
452
            goto out_of_memory;
453 454 455 456

        if (virAsprintf(&qemu_driver->stateDir,
                      "%s/run/libvirt/qemu/", LOCAL_STATE_DIR) == -1)
            goto out_of_memory;
457
    } else {
458
        uid_t uid = geteuid();
459 460
        char *userdir = virGetUserDirectory(NULL, uid);
        if (!userdir)
461
            goto error;
462

463
        if (virAsprintf(&qemu_driver->logDir,
464 465
                        "%s/.libvirt/qemu/log", userdir) == -1) {
            VIR_FREE(userdir);
466
            goto out_of_memory;
467
        }
468

469 470
        if (virAsprintf(&base, "%s/.libvirt", userdir) == -1) {
            VIR_FREE(userdir);
471
            goto out_of_memory;
472 473
        }
        VIR_FREE(userdir);
474 475 476 477 478 479

        if (virAsprintf(&qemu_driver->stateDir, "%s/qemu/run", base) == -1)
            goto out_of_memory;
    }

    if (virFileMakePath(qemu_driver->stateDir) < 0) {
480
        char ebuf[1024];
481 482
        VIR_ERROR(_("Failed to create state dir '%s': %s\n"),
                  qemu_driver->stateDir, virStrerror(errno, ebuf, sizeof ebuf));
483
        goto error;
484 485 486 487 488
    }

    /* Configuration paths are either ~/.libvirt/qemu/... (session) or
     * /etc/libvirt/qemu/... (system).
     */
D
Daniel P. Berrange 已提交
489
    if (snprintf (driverConf, sizeof(driverConf), "%s/qemu.conf", base) == -1)
490
        goto out_of_memory;
D
Daniel P. Berrange 已提交
491
    driverConf[sizeof(driverConf)-1] = '\0';
492

493
    if (virAsprintf(&qemu_driver->configDir, "%s/qemu", base) == -1)
494 495
        goto out_of_memory;

496
    if (virAsprintf(&qemu_driver->autostartDir, "%s/qemu/autostart", base) == -1)
497 498
        goto out_of_memory;

499
    VIR_FREE(base);
500 501 502

    if ((qemu_driver->caps = qemudCapsInit()) == NULL)
        goto out_of_memory;
D
Daniel P. Berrange 已提交
503

504
    if (qemudLoadDriverConfig(qemu_driver, driverConf) < 0) {
505 506 507
        goto error;
    }

508
    if (qemudSecurityInit(qemu_driver) < 0) {
509
        goto error;
D
Daniel P. Berrange 已提交
510 511
    }

512 513 514 515 516 517 518 519 520 521 522 523
    /* 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 */
524 525 526 527
    if (virDomainLoadAllConfigs(NULL,
                                qemu_driver->caps,
                                &qemu_driver->domains,
                                qemu_driver->configDir,
528
                                qemu_driver->autostartDir,
529
                                0, NULL, NULL) < 0)
530
        goto error;
531 532
    qemuDriverUnlock(qemu_driver);

533 534
    qemudAutostartConfigs(qemu_driver);

535

536 537
    return 0;

538
out_of_memory:
539
    virReportOOMError(NULL);
540 541 542
error:
    if (qemu_driver)
        qemuDriverUnlock(qemu_driver);
543
    VIR_FREE(base);
544
    qemudShutdown();
545 546 547
    return -1;
}

548 549 550 551
static void qemudNotifyLoadDomain(virDomainObjPtr vm, int newVM, void *opaque)
{
    struct qemud_driver *driver = opaque;

552 553 554 555 556 557 558 559
    if (newVM) {
        virDomainEventPtr event =
            virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_DEFINED,
                                     VIR_DOMAIN_EVENT_DEFINED_ADDED);
        if (event)
            qemuDomainEventQueue(driver, event);
    }
560 561
}

562 563 564 565 566 567 568 569
/**
 * qemudReload:
 *
 * Function to restart the QEmu daemon, it will recheck the configuration
 * files and update its state and the networking
 */
static int
qemudReload(void) {
570 571 572
    if (!qemu_driver)
        return 0;

573
    qemuDriverLock(qemu_driver);
574 575 576 577
    virDomainLoadAllConfigs(NULL,
                            qemu_driver->caps,
                            &qemu_driver->domains,
                            qemu_driver->configDir,
578
                            qemu_driver->autostartDir,
579
                            0, qemudNotifyLoadDomain, qemu_driver);
580
    qemuDriverUnlock(qemu_driver);
581

582
    qemudAutostartConfigs(qemu_driver);
583 584

    return 0;
585 586
}

587 588 589 590 591 592 593 594 595 596
/**
 * 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) {
597
    unsigned int i;
598
    int active = 0;
599

600 601 602
    if (!qemu_driver)
        return 0;

603 604 605 606 607 608 609 610
    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);
    }
611

612 613
    qemuDriverUnlock(qemu_driver);
    return active;
614 615
}

616 617 618 619 620 621 622
/**
 * qemudShutdown:
 *
 * Shutdown the QEmu daemon, it will stop all active domains and networks
 */
static int
qemudShutdown(void) {
623

624
    if (!qemu_driver)
625
        return -1;
626

627
    qemuDriverLock(qemu_driver);
628 629
    virCapabilitiesFree(qemu_driver->caps);

630
    virDomainObjListFree(&qemu_driver->domains);
631

632
    VIR_FREE(qemu_driver->securityDriverName);
633
    VIR_FREE(qemu_driver->logDir);
634 635
    VIR_FREE(qemu_driver->configDir);
    VIR_FREE(qemu_driver->autostartDir);
636
    VIR_FREE(qemu_driver->stateDir);
637
    VIR_FREE(qemu_driver->vncTLSx509certdir);
J
Jim Meyering 已提交
638
    VIR_FREE(qemu_driver->vncListen);
639
    VIR_FREE(qemu_driver->vncPassword);
640
    VIR_FREE(qemu_driver->vncSASLdir);
D
Daniel P. Berrange 已提交
641

642 643
    /* Free domain callback list */
    virDomainEventCallbackListFree(qemu_driver->domainEventCallbacks);
644 645 646 647
    virDomainEventQueueFree(qemu_driver->domainEventQueue);

    if (qemu_driver->domainEventTimer != -1)
        virEventRemoveTimeout(qemu_driver->domainEventTimer);
648

649 650 651
    if (qemu_driver->brctl)
        brShutdown(qemu_driver->brctl);

652
    qemuDriverUnlock(qemu_driver);
653
    virMutexDestroy(&qemu_driver->lock);
654
    VIR_FREE(qemu_driver);
655 656

    return 0;
657 658 659
}

/* Return -1 for error, 1 to continue reading and 0 for success */
660
typedef int qemudHandlerMonitorOutput(virConnectPtr conn,
661
                                      virDomainObjPtr vm,
662 663 664
                                      const char *output,
                                      int fd);

665 666 667
/*
 * Returns -1 for error, 0 on end-of-file, 1 for success
 */
668
static int
669
qemudReadMonitorOutput(virConnectPtr conn,
670
                       virDomainObjPtr vm,
671 672
                       int fd,
                       char *buf,
G
Guido Günther 已提交
673
                       size_t buflen,
674
                       qemudHandlerMonitorOutput func,
675 676
                       const char *what,
                       int timeout)
677
{
G
Guido Günther 已提交
678
    size_t got = 0;
679
    buf[0] = '\0';
680
    timeout *= 1000; /* poll wants milli seconds */
681

682
    /* Consume & discard the initial greeting */
683
    while (got < (buflen-1)) {
G
Guido Günther 已提交
684
        ssize_t ret;
685 686

        ret = read(fd, buf+got, buflen-got-1);
687

688 689 690 691 692 693
        if (ret < 0) {
            struct pollfd pfd = { .fd = fd, .events = POLLIN };
            if (errno == EINTR)
                continue;

            if (errno != EAGAIN) {
694 695 696
                virReportSystemError(conn, errno,
                                     _("Failure while reading %s startup output"),
                                     what);
697 698 699
                return -1;
            }

700
            ret = poll(&pfd, 1, timeout);
701
            if (ret == 0) {
702
                qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
703
                                 _("Timed out while reading %s startup output"), what);
704 705 706
                return -1;
            } else if (ret == -1) {
                if (errno != EINTR) {
707 708 709
                    virReportSystemError(conn, errno,
                                         _("Failure while reading %s startup output"),
                                         what);
710 711 712 713 714 715 716 717
                    return -1;
                }
            } else {
                /* Make sure we continue loop & read any further data
                   available before dealing with EOF */
                if (pfd.revents & (POLLIN | POLLHUP))
                    continue;

718
                qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
719
                                 _("Failure while reading %s startup output"), what);
720 721
                return -1;
            }
722 723
        } else if (ret == 0) {
            return 0;
724 725 726
        } else {
            got += ret;
            buf[got] = '\0';
727 728 729 730 731 732
            ret = func(conn, vm, buf, fd);
            if (ret == -1)
                return -1;
            if (ret == 1)
                continue;
            return 1;
733 734 735
        }
    }

736
    qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
737
                     _("Out of space while reading %s startup output"), what);
738 739 740 741
    return -1;

}

742 743 744 745 746 747 748 749 750

/*
 * Returns -1 for error, 0 on success
 */
static int
qemudReadLogOutput(virConnectPtr conn,
                   virDomainObjPtr vm,
                   int fd,
                   char *buf,
G
Guido Günther 已提交
751
                   size_t buflen,
752 753 754 755
                   qemudHandlerMonitorOutput func,
                   const char *what,
                   int timeout)
{
756
    int retries = (timeout*10);
757
    int got = 0;
758 759 760
    buf[0] = '\0';

    while (retries) {
761
        ssize_t func_ret, ret;
762
        int isdead = 0;
G
Guido Günther 已提交
763

764 765
        func_ret = func(conn, vm, buf, fd);

766 767
        if (kill(vm->pid, 0) == -1 && errno == ESRCH)
            isdead = 1;
768

769 770
        /* Any failures should be detected before we read the log, so we
         * always have something useful to report on failure. */
771 772
        ret = saferead(fd, buf+got, buflen-got-1);
        if (ret < 0) {
773 774 775 776 777 778
            virReportSystemError(conn, errno,
                                 _("Failure while reading %s log output"),
                                 what);
            return -1;
        }

779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794
        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;
        }

795 796
        if (func_ret <= 0)
            return func_ret;
797 798 799 800 801 802 803 804 805 806

        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;
}

807
static int
808
qemudCheckMonitorPrompt(virConnectPtr conn ATTRIBUTE_UNUSED,
809
                        virDomainObjPtr vm,
810 811 812 813 814 815 816 817 818 819 820
                        const char *output,
                        int fd)
{
    if (strstr(output, "(qemu) ") == NULL)
        return 1; /* keep reading */

    vm->monitor = fd;

    return 0;
}

821
static int
822 823 824 825 826
qemudOpenMonitorCommon(virConnectPtr conn,
                       struct qemud_driver* driver,
                       virDomainObjPtr vm,
                       int monfd,
                       int reconnect)
827
{
828
    char buf[1024];
829
    int ret;
830

831
    if (virSetCloseExec(monfd) < 0) {
832
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
833
                         "%s", _("Unable to set monitor close-on-exec flag"));
834
        return -1;
835
    }
836
    if (virSetNonBlock(monfd) < 0) {
837
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
838
                         "%s", _("Unable to put monitor into non-blocking mode"));
839
        return -1;
840 841
    }

842
    if (!reconnect) {
843 844 845 846
        if (qemudReadMonitorOutput(conn,
                                   vm, monfd,
                                   buf, sizeof(buf),
                                   qemudCheckMonitorPrompt,
847
                                   "monitor", 10) <= 0)
848 849 850
            ret = -1;
        else
            ret = 0;
851 852 853 854 855 856
    } else {
        vm->monitor = monfd;
        ret = 0;
    }

    if (ret != 0)
857
        return ret;
858

859 860 861
    if ((vm->monitorWatch = virEventAddHandle(vm->monitor, 0,
                                              qemudDispatchVMEvent,
                                              driver, NULL)) < 0)
862
        return -1;
863

864 865
    return 0;
}
866

867 868 869 870 871 872 873 874 875
static int
qemudOpenMonitorUnix(virConnectPtr conn,
                     struct qemud_driver* driver,
                     virDomainObjPtr vm,
                     const char *monitor,
                     int reconnect)
{
    struct sockaddr_un addr;
    int monfd;
876
    int timeout = 3; /* In seconds */
877
    int ret, i = 0;
878 879 880 881 882 883 884 885 886 887 888

    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;
    strncpy(addr.sun_path, monitor, sizeof(addr.sun_path));

889 890 891 892 893 894 895 896 897 898 899 900
    do {
        ret = connect(monfd, (struct sockaddr *) &addr, sizeof(addr));

        if (ret == 0)
            break;

        if (errno == EACCES || errno == ECONNREFUSED) {
            /* EACCES       : Socket may not have shown up yet
             * ECONNREFUSED : Leftover socket hasn't been removed yet */
            continue;
        }

901 902 903
        virReportSystemError(conn, errno, "%s",
                             _("failed to connect to monitor socket"));
        goto error;
904 905 906 907 908 909 910

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

    if (ret != 0) {
        virReportSystemError(conn, errno, "%s",
                             _("monitor socket did not show up."));
        goto error;
911 912 913 914 915 916 917 918 919 920 921 922
    }

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

    return 0;

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

923 924 925 926 927 928 929 930
static int
qemudOpenMonitorPty(virConnectPtr conn,
                    struct qemud_driver* driver,
                    virDomainObjPtr vm,
                    const char *monitor,
                    int reconnect)
{
    int monfd;
931

932 933 934 935 936 937 938 939 940 941 942 943
    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:
944
    close(monfd);
945 946 947 948 949 950 951 952 953 954
    return -1;
}

static int
qemudOpenMonitor(virConnectPtr conn,
                 struct qemud_driver *driver,
                 virDomainObjPtr vm,
                 int reconnect)
{
    switch (vm->monitor_chr->type) {
955 956 957 958
    case VIR_DOMAIN_CHR_TYPE_UNIX:
        return qemudOpenMonitorUnix(conn, driver, vm,
                                    vm->monitor_chr->data.nix.path,
                                    reconnect);
959 960 961 962 963 964 965 966 967 968
    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;
    }
969 970
}

971
/* Returns -1 for error, 0 success, 1 continue reading */
972 973 974 975 976 977
static int
qemudExtractMonitorPath(virConnectPtr conn,
                        const char *haystack,
                        size_t *offset,
                        char **path)
{
978
    static const char needle[] = "char device redirected to";
979
    char *tmp, *dev;
980

981
    VIR_FREE(*path);
982
    /* First look for our magic string */
983 984 985 986 987
    if (!(tmp = strstr(haystack + *offset, needle))) {
        return 1;
    }
    tmp += sizeof(needle);
    dev = tmp;
988

989 990 991 992 993
    /*
     * And look for first whitespace character and nul terminate
     * to mark end of the pty path
     */
    while (*tmp) {
994
        if (c_isspace(*tmp)) {
995
            if (VIR_ALLOC_N(*path, (tmp-dev)+1) < 0) {
996
                virReportOOMError(conn);
997 998 999 1000
                return -1;
            }
            strncpy(*path, dev, (tmp-dev));
            (*path)[(tmp-dev)] = '\0';
1001
            /* ... now further update offset till we get EOL */
1002
            *offset = tmp - haystack;
1003 1004
            return 0;
        }
1005
        tmp++;
1006 1007 1008 1009 1010
    }

    /*
     * We found a path, but didn't find any whitespace,
     * so it must be still incomplete - we should at
1011 1012
     * least see a \n - indicate that we want to carry
     * on trying again
1013
     */
1014
    return 1;
1015 1016 1017
}

static int
1018
qemudFindCharDevicePTYs(virConnectPtr conn,
1019
                        virDomainObjPtr vm,
1020 1021
                        const char *output,
                        int fd ATTRIBUTE_UNUSED)
1022
{
1023
    size_t offset = 0;
1024
    int ret, i;
1025 1026

    /* The order in which QEMU prints out the PTY paths is
1027 1028
       the order in which it procsses its serial and parallel
       device args. This code must match that ordering.... */
1029

1030
    /* first comes the serial devices */
1031 1032
    for (i = 0 ; i < vm->def->nserials ; i++) {
        virDomainChrDefPtr chr = vm->def->serials[i];
1033 1034 1035
        if (chr->type == VIR_DOMAIN_CHR_TYPE_PTY) {
            if ((ret = qemudExtractMonitorPath(conn, output, &offset,
                                               &chr->data.file.path)) != 0)
1036
                return ret;
1037 1038 1039
        }
    }

1040
    /* then the parallel devices */
1041 1042
    for (i = 0 ; i < vm->def->nparallels ; i++) {
        virDomainChrDefPtr chr = vm->def->parallels[i];
1043 1044 1045
        if (chr->type == VIR_DOMAIN_CHR_TYPE_PTY) {
            if ((ret = qemudExtractMonitorPath(conn, output, &offset,
                                               &chr->data.file.path)) != 0)
1046
                return ret;
1047 1048 1049
        }
    }

1050
    return 0;
1051 1052
}

1053 1054 1055 1056
static int
qemudWaitForMonitor(virConnectPtr conn,
                    struct qemud_driver* driver,
                    virDomainObjPtr vm, off_t pos)
1057
{
1058
    char buf[4096]; /* Plenty of space to get startup greeting */
1059 1060 1061 1062 1063
    int logfd;
    int ret;

    if ((logfd = qemudLogReadFD(conn, driver->logDir, vm->def->name, pos))
        < 0)
1064
        return -1;
1065

1066 1067 1068
    ret = qemudReadLogOutput(conn, vm, logfd, buf, sizeof(buf),
                             qemudFindCharDevicePTYs,
                             "console", 3);
1069
    if (close(logfd) < 0) {
1070
        char ebuf[4096];
1071
        VIR_WARN(_("Unable to close logfile: %s\n"),
1072 1073
                 virStrerror(errno, ebuf, sizeof ebuf));
    }
1074

1075 1076 1077 1078 1079 1080
    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;
    }
1081

1082 1083 1084 1085
    if (qemudOpenMonitor(conn, driver, vm, 0) < 0)
        return -1;

    return 0;
1086 1087
}

1088 1089
static int
qemudDetectVcpuPIDs(virConnectPtr conn,
1090
                    virDomainObjPtr vm) {
1091 1092 1093 1094 1095 1096
    char *qemucpus = NULL;
    char *line;
    int lastVcpu = -1;

    /* Only KVM has seperate threads for CPUs,
       others just use main QEMU process for CPU */
1097
    if (vm->def->virtType != VIR_DOMAIN_VIRT_KVM)
1098 1099 1100 1101 1102
        vm->nvcpupids = 1;
    else
        vm->nvcpupids = vm->def->vcpus;

    if (VIR_ALLOC_N(vm->vcpupids, vm->nvcpupids) < 0) {
1103
        virReportOOMError(conn);
1104 1105 1106
        return -1;
    }

1107
    if (vm->def->virtType != VIR_DOMAIN_VIRT_KVM) {
1108 1109 1110 1111
        vm->vcpupids[0] = vm->pid;
        return 0;
    }

1112
    if (qemudMonitorCommand(vm, "info cpus", &qemucpus) < 0) {
1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170
        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;

1171
    VIR_FREE(qemucpus);
1172 1173 1174 1175
    return 0;

error:
    VIR_FREE(vm->vcpupids);
1176 1177
    vm->nvcpupids = 0;
    VIR_FREE(qemucpus);
1178 1179 1180 1181 1182 1183 1184 1185

    /* 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;
}

1186 1187
static int
qemudInitCpus(virConnectPtr conn,
D
Daniel Veillard 已提交
1188 1189
              virDomainObjPtr vm,
              const char *migrateFrom) {
1190 1191 1192 1193 1194 1195
    char *info = NULL;
#if HAVE_SCHED_GETAFFINITY
    cpu_set_t mask;
    int i, maxcpu = QEMUD_CPUMASK_LEN;
    virNodeInfo nodeinfo;

1196
    if (nodeGetInfo(conn, &nodeinfo) < 0)
1197 1198 1199 1200 1201 1202 1203 1204
        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
1205 1206 1207 1208 1209 1210
    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++)
1211
            CPU_SET(i, &mask);
D
Daniel P. Berrange 已提交
1212
    }
1213 1214 1215 1216

    for (i = 0 ; i < vm->nvcpupids ; i++) {
        if (sched_setaffinity(vm->vcpupids[i],
                              sizeof(mask), &mask) < 0) {
1217 1218
            virReportSystemError(conn, errno, "%s",
                                 _("failed to set CPU affinity"));
1219 1220 1221 1222 1223
            return -1;
        }
    }
#endif /* HAVE_SCHED_GETAFFINITY */

D
Daniel Veillard 已提交
1224 1225
    if (migrateFrom == NULL) {
        /* Allow the CPUS to start executing */
1226
        if (qemudMonitorCommand(vm, "cont", &info) < 0) {
D
Daniel Veillard 已提交
1227 1228 1229 1230 1231
            qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                             "%s", _("resume operation failed"));
            return -1;
        }
        VIR_FREE(info);
1232 1233 1234 1235 1236 1237
    }

    return 0;
}


1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249
static int
qemudInitPasswords(virConnectPtr conn,
                   struct qemud_driver *driver,
                   virDomainObjPtr vm) {
    char *info = NULL;

    /*
     * NB: Might have more passwords to set in the future. eg a qcow
     * disk decryption password, but there's no monitor command
     * for that yet...
     */

1250 1251 1252
    if ((vm->def->ngraphics == 1) &&
        vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
        (vm->def->graphics[0]->data.vnc.passwd || driver->vncPassword)) {
1253 1254

        if (qemudMonitorCommandExtra(vm, "change vnc password",
1255 1256
                                     vm->def->graphics[0]->data.vnc.passwd ?
                                     vm->def->graphics[0]->data.vnc.passwd :
1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270
                                     driver->vncPassword,
                                     QEMU_PASSWD_PROMPT,
                                     &info) < 0) {
            qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                             "%s", _("setting VNC password failed"));
            return -1;
        }
        VIR_FREE(info);
    }

    return 0;
}


1271
static int qemudNextFreeVNCPort(struct qemud_driver *driver ATTRIBUTE_UNUSED) {
1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306
    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;
}

1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324
static int qemuPrepareHostDevices(virConnectPtr conn,
                                  virDomainDefPtr def) {
    int i;

    /* We have to use 2 loops here. *All* devices must
     * be detached before we reset any of them, because
     * in some cases you have to reset the whole PCI,
     * which impacts all devices on it
     */

    for (i = 0 ; i < def->nhostdevs ; i++) {
        virDomainHostdevDefPtr hostdev = def->hostdevs[i];

        if (hostdev->mode != VIR_DOMAIN_HOSTDEV_MODE_SUBSYS)
            continue;
        if (hostdev->source.subsys.type != VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI)
            continue;

1325
        if (hostdev->managed) {
1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379
            pciDevice *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);
            if (!dev)
                goto error;

            if (pciDettachDevice(conn, dev) < 0) {
                pciFreeDevice(conn, dev);
                goto error;
            }

            pciFreeDevice(conn, dev);
        } /* else {
             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
        } */
    }

    /* Now that all the PCI hostdevs have be dettached, we can safely
     * reset them */
    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);
        if (!dev)
            goto error;

        if (pciResetDevice(conn, dev) < 0) {
            pciFreeDevice(conn, dev);
            goto error;
        }

        pciFreeDevice(conn, dev);
    }

    return 0;

error:
    return -1;
}

1380 1381 1382 1383 1384 1385 1386 1387 1388
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;
}

1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 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 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587

#ifdef __linux__
static int qemuDomainSetHostdevUSBOwnership(virConnectPtr conn,
                                            virDomainHostdevDefPtr def,
                                            uid_t uid, gid_t gid)
{
    char *usbpath = NULL;

    /* 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;

    if (virAsprintf(&usbpath, "/dev/bus/usb/%03d/%03d",
                    def->source.subsys.u.usb.bus,
                    def->source.subsys.u.usb.device) < 0) {
        virReportOOMError(conn);
        return -1;
    }

    VIR_DEBUG("Setting ownership on %s to %d:%d", usbpath, uid, gid);
    if (chown(usbpath, uid, gid) < 0) {
        virReportSystemError(conn, errno, _("cannot set ownership on %s"), usbpath);
        VIR_FREE(usbpath);
        return -1;
    }
    VIR_FREE(usbpath);

    return 0;
}

static int qemuDomainSetHostdevPCIOwnership(virConnectPtr conn,
                                            virDomainHostdevDefPtr def,
                                            uid_t uid, gid_t gid)
{
    char *pcidir = NULL;
    char *file = NULL;
    DIR *dir = NULL;
    int ret = -1;
    struct dirent *ent;

    if (virAsprintf(&pcidir, "/sys/bus/pci/devices/%04x:%02x:%02x.%x",
                    def->source.subsys.u.pci.domain,
                    def->source.subsys.u.pci.bus,
                    def->source.subsys.u.pci.slot,
                    def->source.subsys.u.pci.function) < 0) {
        virReportOOMError(conn);
        goto cleanup;
    }

    if (!(dir = opendir(pcidir))) {
        virReportSystemError(conn, errno,
                             _("cannot open %s"), pcidir);
        goto cleanup;
    }

    while ((ent = readdir(dir)) != NULL) {
        /* QEMU device assignment requires:
         *   $PCIDIR/config, $PCIDIR/resource, $PCIDIR/resourceNNN, $PCIDIR/rom
         */
        if (STREQ(ent->d_name, "config") ||
            STRPREFIX(ent->d_name, "resource") ||
            STREQ(ent->d_name, "rom")) {
            if (virAsprintf(&file, "%s/%s", pcidir, ent->d_name) < 0) {
                virReportOOMError(conn);
                goto cleanup;
            }
            VIR_DEBUG("Setting ownership on %s to %d:%d", file, uid, gid);
            if (chown(file, uid, gid) < 0) {
                virReportSystemError(conn, errno, _("cannot set ownership on %s"), file);
                goto cleanup;
            }
            VIR_FREE(file);
        }
    }

    ret = 0;

cleanup:
    if (dir)
        closedir(dir);
    VIR_FREE(file);
    VIR_FREE(pcidir);
    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

}

static int qemuDomainSetDiskOwnership(virConnectPtr conn,
                                      virDomainDiskDefPtr def,
                                      uid_t uid, gid_t gid)
{

    if (!def->src)
        return 0;

    VIR_DEBUG("Setting ownership on %s to %d:%d", def->src, uid, gid);
    if (chown(def->src, uid, gid) < 0) {
        virReportSystemError(conn, errno, _("cannot set ownership on %s"),
                             def->src);
        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;

        return qemuDomainSetDiskOwnership(conn, def->data.disk, uid, gid);

    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;

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

        if (qemuDomainSetDiskOwnership(conn, def->disks[i], uid, gid) < 0)
            return -1;
    }

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

    return 0;
}

1588 1589 1590
static virDomainPtr qemudDomainLookupByName(virConnectPtr conn,
                                            const char *name);

1591 1592 1593 1594 1595 1596 1597
struct gemudHookData {
        virConnectPtr conn;
        virDomainObjPtr vm;
        struct qemud_driver *driver;
};

static int qemudSecurityHook(void *data) {
1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610
    struct gemudHookData *h = (struct gemudHookData *) data;

    if (qemudDomainSetSecurityLabel(h->conn, h->driver, h->vm) < 0) {
        qemudReportError(h->conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                         _("Failed to set security label"));
        return -1;
    }

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

        if (qemuDomainSetAllDeviceOwnership(h->conn, h->driver, h->vm->def, 0) < 0)
            return -1;
1611

1612 1613 1614 1615 1616
        if (h->driver->group) {
            if (setregid(h->driver->group, h->driver->group) < 0) {
                virReportSystemError(NULL, errno,
                                     _("cannot change to '%d' group"),
                                     h->driver->group);
1617
                return -1;
1618
            }
1619
        }
1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630
        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;
1631 1632
}

1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650
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",
                    driver->stateDir, vm) < 0) {
        virReportOOMError(conn);
        return -1;
    }

    return 0;
}

1651 1652
static int qemudStartVMDaemon(virConnectPtr conn,
                              struct qemud_driver *driver,
1653
                              virDomainObjPtr vm,
1654 1655
                              const char *migrateFrom,
                              int stdin_fd) {
1656
    const char **argv = NULL, **tmp;
1657
    const char **progenv = NULL;
1658
    int i, ret;
1659
    struct stat sb;
1660 1661
    int *tapfds = NULL;
    int ntapfds = 0;
1662
    unsigned int qemuCmdFlags;
1663
    fd_set keepfd;
1664
    const char *emulator;
G
Guido Günther 已提交
1665
    pid_t child;
1666
    int pos = -1;
1667
    char ebuf[1024];
1668
    char *pidfile = NULL;
1669
    int logfile;
1670

1671 1672 1673 1674 1675
    struct gemudHookData hookData;
    hookData.conn = conn;
    hookData.vm = vm;
    hookData.driver = driver;

1676
    FD_ZERO(&keepfd);
1677

1678
    if (virDomainIsActive(vm)) {
1679
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_INVALID,
1680
                         "%s", _("VM is already active"));
1681 1682 1683
        return -1;
    }

1684 1685 1686 1687 1688 1689 1690 1691
    /* 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;

1692 1693 1694
    if ((vm->def->ngraphics == 1) &&
        vm->def->graphics[0]->type == VIR_DOMAIN_GRAPHICS_TYPE_VNC &&
        vm->def->graphics[0]->data.vnc.autoport) {
1695
        int port = qemudNextFreeVNCPort(driver);
1696
        if (port < 0) {
1697
            qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
1698
                             "%s", _("Unable to find an unused VNC port"));
1699
            goto cleanup;
1700
        }
1701
        vm->def->graphics[0]->data.vnc.port = port;
1702
    }
1703

1704
    if (virFileMakePath(driver->logDir) < 0) {
1705 1706 1707
        virReportSystemError(conn, errno,
                             _("cannot create log directory %s"),
                             driver->logDir);
1708
        goto cleanup;
1709 1710
    }

1711
    if ((logfile = qemudLogFD(conn, driver, vm->def->name)) < 0)
1712
        goto cleanup;
1713

1714 1715
    emulator = vm->def->emulator;

1716 1717 1718 1719
    /* 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
     */
1720
    if (stat(emulator, &sb) < 0) {
1721 1722 1723
        virReportSystemError(conn, errno,
                             _("Cannot find QEMU binary %s"),
                             emulator);
1724
        goto cleanup;
1725 1726
    }

1727
    if (qemudExtractVersionInfo(emulator,
1728
                                NULL,
1729
                                &qemuCmdFlags) < 0)
1730
        goto cleanup;
1731

1732
    if (qemuPrepareHostDevices(conn, vm->def) < 0)
1733
        goto cleanup;
1734

1735 1736 1737 1738 1739
    if (VIR_ALLOC(vm->monitor_chr) < 0) {
        virReportOOMError(conn);
        goto cleanup;
    }

1740 1741
    if (qemuPrepareMonitorChr(conn, driver, vm->monitor_chr, vm->def->name) < 0)
        goto cleanup;
1742

D
Daniel P. Berrange 已提交
1743 1744 1745 1746 1747 1748 1749
    if ((ret = virFileDeletePid(driver->stateDir, vm->def->name)) != 0) {
        virReportSystemError(conn, ret,
                             _("Cannot remove stale PID file for %s"),
                             vm->def->name);
        goto cleanup;
    }

1750 1751 1752 1753 1754
    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
1755

1756
    vm->def->id = driver->nextvmid++;
1757
    if (qemudBuildCommandLine(conn, driver, vm->def, vm->monitor_chr,
1758
                              qemuCmdFlags, &argv, &progenv,
1759 1760
                              &tapfds, &ntapfds, migrateFrom) < 0)
        goto cleanup;
1761

1762 1763
    tmp = progenv;
    while (*tmp) {
1764
        if (safewrite(logfile, *tmp, strlen(*tmp)) < 0)
1765
            VIR_WARN(_("Unable to write envv to logfile: %s\n"),
1766
                     virStrerror(errno, ebuf, sizeof ebuf));
1767
        if (safewrite(logfile, " ", 1) < 0)
1768
            VIR_WARN(_("Unable to write envv to logfile: %s\n"),
1769
                     virStrerror(errno, ebuf, sizeof ebuf));
1770 1771
        tmp++;
    }
1772 1773
    tmp = argv;
    while (*tmp) {
1774
        if (safewrite(logfile, *tmp, strlen(*tmp)) < 0)
1775
            VIR_WARN(_("Unable to write argv to logfile: %s\n"),
1776
                     virStrerror(errno, ebuf, sizeof ebuf));
1777
        if (safewrite(logfile, " ", 1) < 0)
1778
            VIR_WARN(_("Unable to write argv to logfile: %s\n"),
1779
                     virStrerror(errno, ebuf, sizeof ebuf));
1780 1781
        tmp++;
    }
1782
    if (safewrite(logfile, "\n", 1) < 0)
1783
        VIR_WARN(_("Unable to write argv to logfile: %s\n"),
1784
                 virStrerror(errno, ebuf, sizeof ebuf));
1785

1786
    if ((pos = lseek(logfile, 0, SEEK_END)) < 0)
1787
        VIR_WARN(_("Unable to seek to end of logfile: %s\n"),
1788
                 virStrerror(errno, ebuf, sizeof ebuf));
1789

1790 1791 1792
    for (i = 0 ; i < ntapfds ; i++)
        FD_SET(tapfds[i], &keepfd);

1793
    ret = virExecDaemonize(conn, argv, progenv, &keepfd, &child,
1794
                           stdin_fd, &logfile, &logfile,
1795
                           VIR_EXEC_NONBLOCK | VIR_EXEC_CLEAR_CAPS,
1796 1797 1798
                           qemudSecurityHook, &hookData,
                           pidfile);
    VIR_FREE(pidfile);
G
Guido Günther 已提交
1799 1800 1801

    /* wait for qemu process to to show up */
    if (ret == 0) {
1802
        if (virFileReadPid(driver->stateDir, vm->def->name, &vm->pid)) {
1803
            qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
1804
                             _("Domain %s didn't show up\n"), vm->def->name);
1805
            ret = -1;
G
Guido Günther 已提交
1806
        }
1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817
    } 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;
    }
1818 1819

    vm->state = migrateFrom ? VIR_DOMAIN_PAUSED : VIR_DOMAIN_RUNNING;
1820

1821
    for (i = 0 ; argv[i] ; i++)
1822 1823
        VIR_FREE(argv[i]);
    VIR_FREE(argv);
1824

1825 1826 1827 1828
    for (i = 0 ; progenv[i] ; i++)
        VIR_FREE(progenv[i]);
    VIR_FREE(progenv);

1829 1830 1831
    if (tapfds) {
        for (i = 0 ; i < ntapfds ; i++) {
            close(tapfds[i]);
1832
        }
1833
        VIR_FREE(tapfds);
1834 1835
    }

1836 1837 1838 1839 1840 1841 1842 1843
    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) ||
1844
        (virDomainSaveStatus(conn, driver->stateDir, vm) < 0)) {
1845 1846 1847
        qemudShutdownVMDaemon(conn, driver, vm);
        ret = -1;
        /* No need for 'goto cleanup' now since qemudShutdownVMDaemon does enough */
1848 1849
    }

1850
    return ret;
1851 1852 1853 1854 1855 1856 1857

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);
    }
1858 1859 1860 1861
    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;
1862 1863
    if (logfile != -1)
        close(logfile);
1864 1865
    vm->def->id = -1;
    return -1;
1866 1867 1868
}


1869
static void qemudShutdownVMDaemon(virConnectPtr conn ATTRIBUTE_UNUSED,
1870 1871
                                  struct qemud_driver *driver,
                                  virDomainObjPtr vm) {
D
Daniel P. Berrange 已提交
1872 1873
    int ret;

1874
    if (!virDomainIsActive(vm))
1875
        return;
1876

1877
    VIR_DEBUG(_("Shutting down VM '%s'\n"), vm->def->name);
1878

G
Guido Günther 已提交
1879 1880
    if (virKillProcess(vm->pid, 0) == 0 &&
        virKillProcess(vm->pid, SIGTERM) < 0)
1881 1882 1883
        virReportSystemError(conn, errno,
                             _("Failed to send SIGTERM to %s (%d)"),
                             vm->def->name, vm->pid);
1884

1885 1886 1887
    if (vm->monitorWatch != -1) {
        virEventRemoveHandle(vm->monitorWatch);
        vm->monitorWatch = -1;
1888
    }
1889 1890 1891 1892 1893

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

G
Guido Günther 已提交
1894 1895
    /* shut it off for sure */
    virKillProcess(vm->pid, SIGKILL);
1896

1897 1898 1899 1900
    /* Reset Security Labels */
    if (driver->securityDriver)
        driver->securityDriver->domainRestoreSecurityLabel(conn, vm);

1901 1902 1903 1904 1905 1906 1907
    /* 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);
    }

1908 1909 1910 1911
    if (qemuDomainSetAllDeviceOwnership(conn, driver, vm->def, 1) < 0)
        VIR_WARN("Failed to restore all device ownership for %s",
                 vm->def->name);

1912
    if (qemudRemoveDomainStatus(conn, driver, vm) < 0) {
1913
        VIR_WARN(_("Failed to remove domain status for %s"),
1914 1915
                 vm->def->name);
    }
D
Daniel P. Berrange 已提交
1916 1917 1918 1919 1920 1921
    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));
    }

1922
    vm->pid = -1;
1923
    vm->def->id = -1;
1924
    vm->state = VIR_DOMAIN_SHUTOFF;
1925
    VIR_FREE(vm->vcpupids);
1926
    vm->nvcpupids = 0;
1927 1928

    if (vm->newDef) {
1929
        virDomainDefFree(vm->def);
1930
        vm->def = vm->newDef;
1931
        vm->def->id = -1;
1932 1933 1934 1935 1936
        vm->newDef = NULL;
    }
}


1937
static void
1938
qemudDispatchVMEvent(int watch, int fd, int events, void *opaque) {
1939
    struct qemud_driver *driver = opaque;
1940
    virDomainObjPtr vm = NULL;
1941
    virDomainEventPtr event = NULL;
1942
    unsigned int i;
1943
    int quit = 0, failed = 0;
1944

1945
    qemuDriverLock(driver);
1946
    for (i = 0 ; i < driver->domains.count ; i++) {
1947 1948 1949
        virDomainObjPtr tmpvm = driver->domains.objs[i];
        virDomainObjLock(tmpvm);
        if (virDomainIsActive(tmpvm) &&
1950
            tmpvm->monitorWatch == watch) {
1951
            vm = tmpvm;
1952
            break;
1953
        }
1954
        virDomainObjUnlock(tmpvm);
1955 1956 1957
    }

    if (!vm)
1958
        goto cleanup;
1959

1960
    if (vm->monitor != fd) {
1961 1962
        failed = 1;
    } else {
1963
        if (events & (VIR_EVENT_HANDLE_HANGUP | VIR_EVENT_HANDLE_ERROR))
1964
            quit = 1;
1965
        else {
1966 1967
            VIR_ERROR(_("unhandled fd event %d for %s"),
                      events, vm->def->name);
1968
            failed = 1;
1969
        }
1970 1971
    }

1972
    if (failed || quit) {
1973 1974 1975 1976 1977
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         quit ?
                                         VIR_DOMAIN_EVENT_STOPPED_SHUTDOWN :
                                         VIR_DOMAIN_EVENT_STOPPED_FAILED);
1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988
        qemudShutdownVMDaemon(NULL, driver, vm);
        if (!vm->persistent) {
            virDomainRemoveInactive(&driver->domains,
                                    vm);
            vm = NULL;
        }
    }

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
1989 1990
    if (event)
        qemuDomainEventQueue(driver, event);
1991
    qemuDriverUnlock(driver);
1992 1993
}

1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015

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

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


2016
static int
2017 2018 2019 2020 2021
qemudMonitorCommandExtra(const virDomainObjPtr vm,
                         const char *cmd,
                         const char *extra,
                         const char *extraPrompt,
                         char **reply) {
D
Daniel P. Berrange 已提交
2022 2023
    int size = 0;
    char *buf = NULL;
2024
    size_t cmdlen = strlen(cmd);
2025
    size_t extralen = extra ? strlen(extra) : 0;
D
Daniel P. Berrange 已提交
2026

2027 2028
    qemuMonitorDiscardPendingData(vm);

2029 2030 2031
    if (safewrite(vm->monitor, cmd, cmdlen) != cmdlen)
        return -1;
    if (safewrite(vm->monitor, "\r", 1) != 1)
D
Daniel P. Berrange 已提交
2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043
        return -1;

    *reply = NULL;

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

        /* Read all the data QEMU has sent thus far */
        for (;;) {
            char data[1024];
            int got = read(vm->monitor, data, sizeof(data));
D
Daniel P. Berrange 已提交
2044

2045 2046
            if (got == 0)
                goto error;
D
Daniel P. Berrange 已提交
2047 2048 2049 2050 2051
            if (got < 0) {
                if (errno == EINTR)
                    continue;
                if (errno == EAGAIN)
                    break;
2052
                goto error;
2053
            }
2054
            if (VIR_REALLOC_N(buf, size+got+1) < 0)
2055 2056
                goto error;

D
Daniel P. Berrange 已提交
2057 2058 2059 2060
            memmove(buf+size, data, got);
            buf[size+got] = '\0';
            size += got;
        }
2061

D
Daniel P. Berrange 已提交
2062
        /* Look for QEMU prompt to indicate completion */
2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087
        if (buf) {
            if (extra) {
                if (strstr(buf, extraPrompt) != NULL) {
                    if (safewrite(vm->monitor, extra, extralen) != extralen)
                        return -1;
                    if (safewrite(vm->monitor, "\r", 1) != 1)
                        return -1;
                    extra = NULL;
                }
            } else if ((tmp = strstr(buf, QEMU_CMD_PROMPT)) != NULL) {
                char *commptr = NULL, *nlptr = NULL;
                /* Preserve the newline */
                tmp[1] = '\0';

                /* The monitor doesn't dump clean output after we have written to
                 * it. Every character we write dumps a bunch of useless stuff,
                 * so the result looks like "cXcoXcomXcommXcommaXcommanXcommand"
                 * Try to throw away everything before the first full command
                 * occurence, and inbetween the command and the newline starting
                 * the response
                 */
                if ((commptr = strstr(buf, cmd)))
                    memmove(buf, commptr, strlen(commptr)+1);
                if ((nlptr = strchr(buf, '\n')))
                    memmove(buf+strlen(cmd), nlptr, strlen(nlptr)+1);
C
Cole Robinson 已提交
2088

2089 2090
                break;
            }
D
Daniel P. Berrange 已提交
2091 2092 2093 2094 2095 2096
        }
    pollagain:
        /* Need to wait for more data */
        if (poll(&fd, 1, -1) < 0) {
            if (errno == EINTR)
                goto pollagain;
2097
            goto error;
D
Daniel P. Berrange 已提交
2098 2099 2100 2101
        }
    }
    *reply = buf;
    return 0;
2102 2103

 error:
2104
    VIR_FREE(buf);
2105
    return -1;
D
Daniel P. Berrange 已提交
2106 2107
}

2108 2109 2110 2111 2112 2113 2114 2115
static int
qemudMonitorCommand(const virDomainObjPtr vm,
                    const char *cmd,
                    char **reply) {
    return qemudMonitorCommandExtra(vm, cmd, NULL, NULL, reply);
}


2116

2117
static virDrvOpenStatus qemudOpen(virConnectPtr conn,
2118
                                  virConnectAuthPtr auth ATTRIBUTE_UNUSED,
2119
                                  int flags ATTRIBUTE_UNUSED) {
2120
    if (conn->uri == NULL) {
2121 2122 2123
        if (qemu_driver == NULL)
            return VIR_DRV_OPEN_DECLINED;

2124
        conn->uri = xmlParseURI(qemu_driver->privileged ?
2125 2126
                                "qemu:///system" :
                                "qemu:///session");
2127
        if (!conn->uri) {
2128
            virReportOOMError(conn);
2129 2130
            return VIR_DRV_OPEN_ERROR;
        }
2131 2132 2133 2134 2135 2136 2137 2138 2139 2140
    } 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;

2141 2142 2143 2144 2145 2146
        if (qemu_driver == NULL) {
            qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR, "%s",
                             _("qemu state driver is not active"));
            return VIR_DRV_OPEN_ERROR;
        }

2147
        if (qemu_driver->privileged) {
2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162
            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;
            }
        }
2163 2164 2165 2166 2167 2168 2169
    }
    conn->privateData = qemu_driver;

    return VIR_DRV_OPEN_SUCCESS;
}

static int qemudClose(virConnectPtr conn) {
2170
    struct qemud_driver *driver = conn->privateData;
2171 2172

    /* Get rid of callbacks registered for this conn */
2173
    qemuDriverLock(driver);
2174
    virDomainEventCallbackListRemoveConn(conn, driver->domainEventCallbacks);
2175
    qemuDriverUnlock(driver);
2176 2177 2178 2179 2180 2181

    conn->privateData = NULL;

    return 0;
}

D
Daniel Veillard 已提交
2182 2183 2184 2185 2186 2187 2188 2189 2190 2191
/* 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;
    }
}

2192
static const char *qemudGetType(virConnectPtr conn ATTRIBUTE_UNUSED) {
2193
    return "QEMU";
2194 2195
}

2196 2197 2198 2199 2200

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

    int r, fd;
2201

2202 2203
    fd = open(KVM_DEVICE, O_RDONLY);
    if (fd < 0) {
2204 2205
        virReportSystemError(NULL, errno, _("Unable to open %s"), KVM_DEVICE);
        return -1;
2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216
    }

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

    close(fd);
    return maxvcpus;
}


2217
static int qemudGetMaxVCPUs(virConnectPtr conn, const char *type) {
2218 2219 2220
    if (!type)
        return 16;

2221
    if (STRCASEEQ(type, "qemu"))
2222 2223
        return 16;

2224
    if (STRCASEEQ(type, "kvm"))
2225
        return kvmGetMaxVCPUs();
2226

2227
    if (STRCASEEQ(type, "kqemu"))
2228
        return 1;
2229 2230 2231

    qemudReportError(conn, NULL, NULL, VIR_ERR_INVALID_ARG,
                     _("unknown type '%s'"), type);
2232 2233 2234
    return -1;
}

2235

2236
static char *qemudGetCapabilities(virConnectPtr conn) {
2237
    struct qemud_driver *driver = conn->privateData;
2238
    virCapsPtr caps;
2239
    char *xml = NULL;
2240

2241
    qemuDriverLock(driver);
2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253
    if ((caps = qemudCapsInit()) == NULL) {
        virReportOOMError(conn);
        goto cleanup;
    }

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

2254
    virCapabilitiesFree(qemu_driver->caps);
2255 2256 2257
    qemu_driver->caps = caps;

    if ((xml = virCapabilitiesFormatXML(driver->caps)) == NULL)
2258
        virReportOOMError(conn);
2259 2260

cleanup:
2261
    qemuDriverUnlock(driver);
2262

2263
    return xml;
2264 2265 2266
}


D
Daniel P. Berrange 已提交
2267 2268 2269
static int qemudGetProcessInfo(unsigned long long *cpuTime, int pid) {
    char proc[PATH_MAX];
    FILE *pidinfo;
2270
    unsigned long long usertime, systime;
D
Daniel P. Berrange 已提交
2271 2272 2273 2274 2275 2276

    if (snprintf(proc, sizeof(proc), "/proc/%d/stat", pid) >= (int)sizeof(proc)) {
        return -1;
    }

    if (!(pidinfo = fopen(proc, "r"))) {
2277
        /*printf("cannot read pid info");*/
D
Daniel P. Berrange 已提交
2278 2279 2280 2281 2282
        /* VM probably shut down, so fake 0 */
        *cpuTime = 0;
        return 0;
    }

2283
    if (fscanf(pidinfo, "%*d %*s %*c %*d %*d %*d %*d %*d %*u %*u %*u %*u %*u %llu %llu", &usertime, &systime) != 2) {
2284
        qemudDebug("not enough arg");
D
Daniel P. Berrange 已提交
2285 2286 2287 2288 2289 2290 2291 2292
        return -1;
    }

    /* We got jiffies
     * We want nanoseconds
     * _SC_CLK_TCK is jiffies per second
     * So calulate thus....
     */
2293
    *cpuTime = 1000ull * 1000ull * 1000ull * (usertime + systime) / (unsigned long long)sysconf(_SC_CLK_TCK);
D
Daniel P. Berrange 已提交
2294

2295
    qemudDebug("Got %llu %llu %llu", usertime, systime, *cpuTime);
D
Daniel P. Berrange 已提交
2296 2297 2298 2299 2300 2301 2302

    fclose(pidinfo);

    return 0;
}


2303
static virDomainPtr qemudDomainLookupByID(virConnectPtr conn,
2304
                                          int id) {
2305 2306 2307 2308
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;

2309
    qemuDriverLock(driver);
2310
    vm  = virDomainFindByID(&driver->domains, id);
2311
    qemuDriverUnlock(driver);
2312 2313

    if (!vm) {
2314 2315
        qemudReportError(conn, NULL, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching id %d"), id);
2316
        goto cleanup;
2317 2318
    }

2319
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
2320
    if (dom) dom->id = vm->def->id;
2321 2322

cleanup:
2323 2324
    if (vm)
        virDomainObjUnlock(vm);
2325 2326
    return dom;
}
2327

2328
static virDomainPtr qemudDomainLookupByUUID(virConnectPtr conn,
2329
                                            const unsigned char *uuid) {
2330 2331 2332
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
2333

2334
    qemuDriverLock(driver);
2335
    vm = virDomainFindByUUID(&driver->domains, uuid);
2336 2337
    qemuDriverUnlock(driver);

2338
    if (!vm) {
2339 2340 2341
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(uuid, uuidstr);
        qemudReportError(conn, NULL, NULL, VIR_ERR_NO_DOMAIN,
2342
                         _("no domain with matching uuid '%s'"), uuidstr);
2343
        goto cleanup;
2344 2345
    }

2346
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
2347
    if (dom) dom->id = vm->def->id;
2348 2349

cleanup:
2350 2351
    if (vm)
        virDomainObjUnlock(vm);
2352 2353
    return dom;
}
2354

2355
static virDomainPtr qemudDomainLookupByName(virConnectPtr conn,
2356
                                            const char *name) {
2357 2358 2359
    struct qemud_driver *driver = conn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
2360

2361
    qemuDriverLock(driver);
2362
    vm = virDomainFindByName(&driver->domains, name);
2363 2364
    qemuDriverUnlock(driver);

2365
    if (!vm) {
2366 2367
        qemudReportError(conn, NULL, NULL, VIR_ERR_NO_DOMAIN,
                         _("no domain with matching name '%s'"), name);
2368
        goto cleanup;
2369 2370
    }

2371
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
2372
    if (dom) dom->id = vm->def->id;
2373 2374

cleanup:
2375 2376
    if (vm)
        virDomainObjUnlock(vm);
2377 2378 2379
    return dom;
}

2380
static int qemudGetVersion(virConnectPtr conn, unsigned long *version) {
2381 2382 2383
    struct qemud_driver *driver = conn->privateData;
    int ret = -1;

2384
    qemuDriverLock(driver);
2385
    if (qemudExtractVersion(conn, driver) < 0)
2386
        goto cleanup;
2387

2388
    *version = qemu_driver->qemuVersion;
2389 2390 2391
    ret = 0;

cleanup:
2392
    qemuDriverUnlock(driver);
2393
    return ret;
D
Daniel P. Berrange 已提交
2394 2395
}

2396 2397 2398
static char *
qemudGetHostname (virConnectPtr conn)
{
2399
    char *result;
2400

2401 2402
    result = virGetHostname();
    if (result == NULL) {
2403 2404
        virReportSystemError (conn, errno,
                              "%s", _("failed to determine host name"));
2405 2406 2407
        return NULL;
    }
    /* Caller frees this string. */
2408
    return result;
2409 2410
}

2411
static int qemudListDomains(virConnectPtr conn, int *ids, int nids) {
2412
    struct qemud_driver *driver = conn->privateData;
2413 2414
    int got = 0, i;

2415 2416 2417
    qemuDriverLock(driver);
    for (i = 0 ; i < driver->domains.count && got < nids ; i++) {
        virDomainObjLock(driver->domains.objs[i]);
2418 2419
        if (virDomainIsActive(driver->domains.objs[i]))
            ids[got++] = driver->domains.objs[i]->def->id;
2420 2421 2422
        virDomainObjUnlock(driver->domains.objs[i]);
    }
    qemuDriverUnlock(driver);
2423

D
Daniel P. Berrange 已提交
2424 2425
    return got;
}
2426

2427
static int qemudNumDomains(virConnectPtr conn) {
2428
    struct qemud_driver *driver = conn->privateData;
2429 2430
    int n = 0, i;

2431 2432 2433
    qemuDriverLock(driver);
    for (i = 0 ; i < driver->domains.count ; i++) {
        virDomainObjLock(driver->domains.objs[i]);
2434
        if (virDomainIsActive(driver->domains.objs[i]))
2435
            n++;
2436 2437 2438
        virDomainObjUnlock(driver->domains.objs[i]);
    }
    qemuDriverUnlock(driver);
2439

2440
    return n;
D
Daniel P. Berrange 已提交
2441
}
2442

2443
static virDomainPtr qemudDomainCreate(virConnectPtr conn, const char *xml,
2444
                                      unsigned int flags ATTRIBUTE_UNUSED) {
2445
    struct qemud_driver *driver = conn->privateData;
2446
    virDomainDefPtr def;
2447
    virDomainObjPtr vm = NULL;
2448
    virDomainPtr dom = NULL;
2449
    virDomainEventPtr event = NULL;
D
Daniel P. Berrange 已提交
2450

2451
    qemuDriverLock(driver);
2452 2453
    if (!(def = virDomainDefParseString(conn, driver->caps, xml,
                                        VIR_DOMAIN_XML_INACTIVE)))
2454
        goto cleanup;
2455

2456 2457 2458
    if (virSecurityDriverVerify(conn, def) < 0)
        goto cleanup;

2459
    /* See if a VM with matching UUID already exists */
2460
    vm = virDomainFindByUUID(&driver->domains, def->uuid);
2461
    if (vm) {
2462 2463 2464 2465 2466 2467 2468 2469 2470
        /* 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;
        }
2471

2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489
        /* 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;
        }
2490
    }
2491

2492 2493
    if (!(vm = virDomainAssignDef(conn,
                                  &driver->domains,
2494 2495 2496 2497
                                  def)))
        goto cleanup;

    def = NULL;
D
Daniel P. Berrange 已提交
2498

2499
    if (qemudStartVMDaemon(conn, driver, vm, NULL, -1) < 0) {
2500 2501
        virDomainRemoveInactive(&driver->domains,
                                vm);
2502
        vm = NULL;
2503
        goto cleanup;
D
Daniel P. Berrange 已提交
2504
    }
2505 2506 2507 2508

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_BOOTED);
D
Daniel P. Berrange 已提交
2509

2510
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
2511
    if (dom) dom->id = vm->def->id;
2512 2513 2514

cleanup:
    virDomainDefFree(def);
2515 2516
    if (vm)
        virDomainObjUnlock(vm);
2517 2518
    if (event)
        qemuDomainEventQueue(driver, event);
2519
    qemuDriverUnlock(driver);
2520
    return dom;
D
Daniel P. Berrange 已提交
2521 2522 2523
}


2524
static int qemudDomainSuspend(virDomainPtr dom) {
2525
    struct qemud_driver *driver = dom->conn->privateData;
D
Daniel P. Berrange 已提交
2526
    char *info;
2527 2528
    virDomainObjPtr vm;
    int ret = -1;
2529
    virDomainEventPtr event = NULL;
2530

2531
    qemuDriverLock(driver);
2532
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
2533

D
Daniel P. Berrange 已提交
2534
    if (!vm) {
2535 2536 2537 2538
        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);
2539
        goto cleanup;
D
Daniel P. Berrange 已提交
2540
    }
2541
    if (!virDomainIsActive(vm)) {
2542
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
2543
                         "%s", _("domain is not running"));
2544
        goto cleanup;
D
Daniel P. Berrange 已提交
2545
    }
2546
    if (vm->state != VIR_DOMAIN_PAUSED) {
2547
        if (qemudMonitorCommand(vm, "stop", &info) < 0) {
2548 2549 2550 2551 2552 2553
            qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                             "%s", _("suspend operation failed"));
            goto cleanup;
        }
        vm->state = VIR_DOMAIN_PAUSED;
        qemudDebug("Reply %s", info);
2554 2555 2556
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_SUSPENDED,
                                         VIR_DOMAIN_EVENT_SUSPENDED_PAUSED);
2557
        VIR_FREE(info);
D
Daniel P. Berrange 已提交
2558
    }
2559
    if (virDomainSaveStatus(dom->conn, driver->stateDir, vm) < 0)
2560
        goto cleanup;
2561 2562 2563
    ret = 0;

cleanup:
2564 2565
    if (vm)
        virDomainObjUnlock(vm);
2566

2567
    if (event)
2568
        qemuDomainEventQueue(driver, event);
2569
    qemuDriverUnlock(driver);
2570
    return ret;
D
Daniel P. Berrange 已提交
2571 2572 2573
}


2574
static int qemudDomainResume(virDomainPtr dom) {
2575
    struct qemud_driver *driver = dom->conn->privateData;
D
Daniel P. Berrange 已提交
2576
    char *info;
2577 2578
    virDomainObjPtr vm;
    int ret = -1;
2579
    virDomainEventPtr event = NULL;
2580

2581
    qemuDriverLock(driver);
2582
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
2583

D
Daniel P. Berrange 已提交
2584
    if (!vm) {
2585 2586 2587 2588
        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);
2589
        goto cleanup;
D
Daniel P. Berrange 已提交
2590
    }
2591
    if (!virDomainIsActive(vm)) {
2592
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
2593
                         "%s", _("domain is not running"));
2594
        goto cleanup;
D
Daniel P. Berrange 已提交
2595
    }
2596
    if (vm->state == VIR_DOMAIN_PAUSED) {
2597
        if (qemudMonitorCommand(vm, "cont", &info) < 0) {
2598 2599 2600 2601 2602 2603
            qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                             "%s", _("resume operation failed"));
            goto cleanup;
        }
        vm->state = VIR_DOMAIN_RUNNING;
        qemudDebug("Reply %s", info);
2604 2605 2606
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_UNPAUSED);
2607
        VIR_FREE(info);
D
Daniel P. Berrange 已提交
2608
    }
2609
    if (virDomainSaveStatus(dom->conn, driver->stateDir, vm) < 0)
2610
        goto cleanup;
2611 2612 2613
    ret = 0;

cleanup:
2614 2615
    if (vm)
        virDomainObjUnlock(vm);
2616
    if (event)
2617
        qemuDomainEventQueue(driver, event);
2618
    qemuDriverUnlock(driver);
2619
    return ret;
D
Daniel P. Berrange 已提交
2620 2621 2622
}


2623
static int qemudDomainShutdown(virDomainPtr dom) {
2624 2625
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
2626
    char* info;
2627
    int ret = -1;
2628

2629
    qemuDriverLock(driver);
2630
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
2631 2632
    qemuDriverUnlock(driver);

2633
    if (!vm) {
2634 2635 2636 2637
        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);
2638
        goto cleanup;
2639 2640
    }

2641
    if (qemudMonitorCommand(vm, "system_powerdown", &info) < 0) {
2642
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
2643
                         "%s", _("shutdown operation failed"));
2644
        goto cleanup;
2645
    }
D
Daniel Veillard 已提交
2646
    VIR_FREE(info);
2647 2648 2649
    ret = 0;

cleanup:
2650 2651
    if (vm)
        virDomainObjUnlock(vm);
2652
    return ret;
2653 2654 2655
}


2656
static int qemudDomainDestroy(virDomainPtr dom) {
2657 2658 2659
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
2660
    virDomainEventPtr event = NULL;
2661

2662
    qemuDriverLock(driver);
2663
    vm  = virDomainFindByUUID(&driver->domains, dom->uuid);
D
Daniel P. Berrange 已提交
2664
    if (!vm) {
2665 2666 2667 2668
        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);
2669
        goto cleanup;
D
Daniel P. Berrange 已提交
2670
    }
2671 2672 2673 2674 2675
    if (!virDomainIsActive(vm)) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                         "%s", _("domain is not running"));
        goto cleanup;
    }
2676

2677
    qemudShutdownVMDaemon(dom->conn, driver, vm);
2678 2679 2680
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_DESTROYED);
2681
    if (!vm->persistent) {
2682 2683
        virDomainRemoveInactive(&driver->domains,
                                vm);
2684 2685
        vm = NULL;
    }
2686 2687 2688
    ret = 0;

cleanup:
2689 2690
    if (vm)
        virDomainObjUnlock(vm);
2691 2692
    if (event)
        qemuDomainEventQueue(driver, event);
2693
    qemuDriverUnlock(driver);
2694
    return ret;
D
Daniel P. Berrange 已提交
2695 2696 2697
}


2698
static char *qemudDomainGetOSType(virDomainPtr dom) {
2699 2700 2701
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    char *type = NULL;
2702

2703
    qemuDriverLock(driver);
2704
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
2705
    qemuDriverUnlock(driver);
2706
    if (!vm) {
2707 2708 2709 2710
        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);
2711
        goto cleanup;
2712 2713
    }

2714
    if (!(type = strdup(vm->def->os.type)))
2715
        virReportOOMError(dom->conn);
2716 2717

cleanup:
2718 2719
    if (vm)
        virDomainObjUnlock(vm);
2720 2721 2722
    return type;
}

2723 2724
/* Returns max memory in kb, 0 if error */
static unsigned long qemudDomainGetMaxMemory(virDomainPtr dom) {
2725 2726 2727
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    unsigned long ret = 0;
2728

2729
    qemuDriverLock(driver);
2730
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
2731 2732
    qemuDriverUnlock(driver);

2733
    if (!vm) {
2734 2735
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
2736
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
2737
                         _("no domain with matching uuid '%s'"), uuidstr);
2738
        goto cleanup;
2739 2740
    }

2741 2742 2743
    ret = vm->def->maxmem;

cleanup:
2744 2745
    if (vm)
        virDomainObjUnlock(vm);
2746
    return ret;
2747 2748 2749
}

static int qemudDomainSetMaxMemory(virDomainPtr dom, unsigned long newmax) {
2750 2751 2752
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
2753

2754
    qemuDriverLock(driver);
2755
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
2756 2757
    qemuDriverUnlock(driver);

2758
    if (!vm) {
2759 2760
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
2761
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
2762
                         _("no domain with matching uuid '%s'"), uuidstr);
2763
        goto cleanup;
2764 2765 2766 2767
    }

    if (newmax < vm->def->memory) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
2768
                         "%s", _("cannot set max memory lower than current memory"));
2769
        goto cleanup;;
2770 2771 2772
    }

    vm->def->maxmem = newmax;
2773 2774 2775
    ret = 0;

cleanup:
2776 2777
    if (vm)
        virDomainObjUnlock(vm);
2778
    return ret;
2779 2780
}

2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795

/* 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;

2796 2797 2798
    if (!virDomainIsActive(vm))
        return 0;

2799 2800 2801 2802 2803 2804
    if (qemudMonitorCommand(vm, "info balloon", &reply) < 0) {
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                         "%s", _("could not query memory balloon allocation"));
        goto cleanup;
    }

2805
    DEBUG ("%s: balloon reply: '%s'", vm->def->name, reply);
2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858
    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 */
2859
    DEBUG ("%s: balloon reply: %s",vm->def->name,  reply);
2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872
    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;
}


2873
static int qemudDomainSetMemory(virDomainPtr dom, unsigned long newmem) {
2874 2875 2876
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
2877

2878
    qemuDriverLock(driver);
2879
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
2880
    qemuDriverUnlock(driver);
2881
    if (!vm) {
2882 2883
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
2884
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
2885
                         _("no domain with matching uuid '%s'"), uuidstr);
2886
        goto cleanup;
2887 2888 2889 2890
    }

    if (newmem > vm->def->maxmem) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
2891
                         "%s", _("cannot set memory higher than max memory"));
2892
        goto cleanup;
2893 2894
    }

2895 2896 2897 2898 2899 2900 2901 2902 2903
    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;
    }
2904 2905

cleanup:
2906 2907
    if (vm)
        virDomainObjUnlock(vm);
2908
    return ret;
2909 2910
}

2911
static int qemudDomainGetInfo(virDomainPtr dom,
2912
                              virDomainInfoPtr info) {
2913 2914 2915
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
2916 2917
    int err;
    unsigned long balloon;
2918

2919
    qemuDriverLock(driver);
2920
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
2921
    qemuDriverUnlock(driver);
D
Daniel P. Berrange 已提交
2922
    if (!vm) {
2923 2924 2925 2926
        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);
2927
        goto cleanup;
D
Daniel P. Berrange 已提交
2928 2929
    }

2930
    info->state = vm->state;
D
Daniel P. Berrange 已提交
2931

2932
    if (!virDomainIsActive(vm)) {
2933
        info->cpuTime = 0;
D
Daniel P. Berrange 已提交
2934
    } else {
2935
        if (qemudGetProcessInfo(&(info->cpuTime), vm->pid) < 0) {
2936
            qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED, ("cannot read cputime for domain"));
2937
            goto cleanup;
D
Daniel P. Berrange 已提交
2938 2939 2940
        }
    }

2941
    info->maxMem = vm->def->maxmem;
2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956

    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;
    }

2957
    info->nrVirtCpu = vm->def->vcpus;
2958 2959 2960
    ret = 0;

cleanup:
2961 2962
    if (vm)
        virDomainObjUnlock(vm);
2963
    return ret;
D
Daniel P. Berrange 已提交
2964 2965 2966
}


D
Daniel P. Berrange 已提交
2967
static char *qemudEscape(const char *in, int shell)
2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988
{
    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
2989 2990 2991 2992
            if (shell)
                len += 5;
            else
                len += 1;
2993 2994 2995 2996 2997 2998 2999
            break;
        default:
            len += 1;
            break;
        }
    }

3000
    if (VIR_ALLOC_N(out, len + 1) < 0)
3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018
        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
3019 3020 3021 3022 3023 3024 3025 3026 3027
            if (shell) {
                out[j++] = '\'';
                out[j++] = '\\';
                out[j++] = '\\';
                out[j++] = '\'';
                out[j++] = '\'';
            } else {
                out[j++] = in[i];
            }
3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038
            break;
        default:
            out[j++] = in[i];
            break;
        }
    }
    out[j] = '\0';

    return out;
}

3039 3040 3041 3042 3043
static char *qemudEscapeMonitorArg(const char *in)
{
    return qemudEscape(in, 0);
}

D
Daniel P. Berrange 已提交
3044 3045 3046 3047
static char *qemudEscapeShellArg(const char *in)
{
    return qemudEscape(in, 1);
}
3048

3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059
#define QEMUD_SAVE_MAGIC "LibvirtQemudSave"
#define QEMUD_SAVE_VERSION 1

struct qemud_save_header {
    char magic[sizeof(QEMUD_SAVE_MAGIC)-1];
    int version;
    int xml_len;
    int was_running;
    int unused[16];
};

3060
static int qemudDomainSave(virDomainPtr dom,
3061
                           const char *path) {
3062 3063 3064 3065 3066 3067 3068
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    char *command = NULL;
    char *info = NULL;
    int fd = -1;
    char *safe_path = NULL;
    char *xml = NULL;
3069
    struct qemud_save_header header;
3070
    int ret = -1;
3071
    virDomainEventPtr event = NULL;
3072 3073 3074 3075 3076

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

3077
    qemuDriverLock(driver);
3078
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3079

D
Daniel P. Berrange 已提交
3080
    if (!vm) {
3081 3082 3083 3084
        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);
3085
        goto cleanup;
D
Daniel P. Berrange 已提交
3086
    }
3087

3088
    if (!virDomainIsActive(vm)) {
3089
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
3090
                         "%s", _("domain is not running"));
3091
        goto cleanup;
D
Daniel P. Berrange 已提交
3092
    }
3093 3094 3095 3096

    /* Pause */
    if (vm->state == VIR_DOMAIN_RUNNING) {
        header.was_running = 1;
3097
        if (qemudMonitorCommand(vm, "stop", &info) < 0) {
3098
            qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
3099
                             "%s", _("suspend operation failed"));
3100
            goto cleanup;
3101
        }
3102 3103 3104
        vm->state = VIR_DOMAIN_PAUSED;
        qemudDebug("Reply %s", info);
        VIR_FREE(info);
3105 3106 3107
    }

    /* Get XML for the domain */
3108
    xml = virDomainDefFormat(dom->conn, vm->def, VIR_DOMAIN_XML_SECURE);
3109 3110
    if (!xml) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
3111
                         "%s", _("failed to get domain xml"));
3112
        goto cleanup;
3113 3114 3115 3116 3117 3118
    }
    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,
3119
                         _("failed to create '%s'"), path);
3120
        goto cleanup;
3121 3122 3123 3124
    }

    if (safewrite(fd, &header, sizeof(header)) != sizeof(header)) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
3125
                         "%s", _("failed to write save header"));
3126
        goto cleanup;
3127 3128 3129 3130
    }

    if (safewrite(fd, xml, header.xml_len) != header.xml_len) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
3131
                         "%s", _("failed to write xml"));
3132
        goto cleanup;
3133 3134
    }

3135
    if (close(fd) < 0) {
3136 3137 3138
        virReportSystemError(dom->conn, errno,
                             _("unable to save file %s"),
                             path);
3139 3140 3141
        goto cleanup;
    }
    fd = -1;
3142 3143 3144 3145

    /* Migrate to file */
    safe_path = qemudEscapeShellArg(path);
    if (!safe_path) {
3146
        virReportOOMError(dom->conn);
3147
        goto cleanup;
3148
    }
3149
    if (virAsprintf(&command, "migrate \"exec:"
3150
                  "dd of='%s' oflag=append conv=notrunc 2>/dev/null"
3151
                  "\"", safe_path) == -1) {
3152
        virReportOOMError(dom->conn);
3153 3154
        command = NULL;
        goto cleanup;
3155 3156
    }

3157
    if (qemudMonitorCommand(vm, command, &info) < 0) {
3158
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
3159
                         "%s", _("migrate operation failed"));
3160
        goto cleanup;
3161 3162
    }

3163
    DEBUG ("%s: migrate reply: %s", vm->def->name, info);
3164 3165 3166 3167 3168 3169 3170

    /* 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"));
3171
        goto cleanup;
3172 3173
    }

3174 3175
    /* Shut it down */
    qemudShutdownVMDaemon(dom->conn, driver, vm);
3176 3177 3178
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_SAVED);
3179
    if (!vm->persistent) {
3180 3181
        virDomainRemoveInactive(&driver->domains,
                                vm);
3182 3183
        vm = NULL;
    }
3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194
    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);
3195 3196
    if (vm)
        virDomainObjUnlock(vm);
3197 3198
    if (event)
        qemuDomainEventQueue(driver, event);
3199
    qemuDriverUnlock(driver);
3200
    return ret;
D
Daniel P. Berrange 已提交
3201 3202 3203
}


P
Paolo Bonzini 已提交
3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305
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) {
        if (qemudMonitorCommand(vm, "cont", &info) < 0) {
            qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                             "%s", _("resuming after dump failed"));
            goto cleanup;
        }
        DEBUG ("%s: cont reply: %s", vm->def->name, info);
        VIR_FREE(info);
    }
    if (vm)
        virDomainObjUnlock(vm);
    return ret;
}


3306
static int qemudDomainSetVcpus(virDomainPtr dom, unsigned int nvcpus) {
3307 3308
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
3309
    int max;
3310
    int ret = -1;
3311
    const char *type;
3312

3313
    qemuDriverLock(driver);
3314
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3315 3316
    qemuDriverUnlock(driver);

3317
    if (!vm) {
3318 3319
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
3320
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
3321
                         _("no domain with matching uuid '%s'"), uuidstr);
3322
        goto cleanup;
3323 3324
    }

3325
    if (virDomainIsActive(vm)) {
3326
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID, "%s",
3327
                         _("cannot change vcpu count of an active domain"));
3328
        goto cleanup;
3329 3330
    }

3331 3332 3333 3334 3335 3336 3337 3338
    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) {
3339 3340
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INTERNAL_ERROR, "%s",
                         _("could not determine max vcpus for the domain"));
3341
        goto cleanup;
3342 3343 3344 3345 3346 3347
    }

    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);
3348
        goto cleanup;
3349 3350 3351
    }

    vm->def->vcpus = nvcpus;
3352 3353 3354
    ret = 0;

cleanup:
3355 3356
    if (vm)
        virDomainObjUnlock(vm);
3357
    return ret;
3358 3359
}

3360 3361 3362 3363 3364 3365 3366

#if HAVE_SCHED_GETAFFINITY
static int
qemudDomainPinVcpu(virDomainPtr dom,
                   unsigned int vcpu,
                   unsigned char *cpumap,
                   int maplen) {
3367 3368
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
3369 3370 3371
    cpu_set_t mask;
    int i, maxcpu;
    virNodeInfo nodeinfo;
3372
    int ret = -1;
3373

3374
    qemuDriverLock(driver);
3375
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3376 3377
    qemuDriverUnlock(driver);

3378 3379 3380 3381 3382 3383 3384 3385
    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;
    }

3386
    if (!virDomainIsActive(vm)) {
3387
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
3388
                         "%s",_("cannot pin vcpus on an inactive domain"));
3389
        goto cleanup;
3390 3391 3392 3393 3394 3395
    }

    if (vcpu > (vm->nvcpupids-1)) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
                         _("vcpu number out of range %d > %d"),
                         vcpu, vm->nvcpupids);
3396
        goto cleanup;
3397 3398
    }

3399
    if (nodeGetInfo(dom->conn, &nodeinfo) < 0)
3400
        goto cleanup;
3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413

    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) {
3414 3415
            virReportSystemError(dom->conn, errno, "%s",
                                 _("cannot set affinity"));
3416
            goto cleanup;
3417 3418 3419 3420
        }
    } else {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                         "%s", _("cpu affinity is not supported"));
3421
        goto cleanup;
3422
    }
3423
    ret = 0;
3424

3425
cleanup:
3426 3427
    if (vm)
        virDomainObjUnlock(vm);
3428
    return ret;
3429 3430 3431 3432 3433 3434 3435 3436
}

static int
qemudDomainGetVcpus(virDomainPtr dom,
                    virVcpuInfoPtr info,
                    int maxinfo,
                    unsigned char *cpumaps,
                    int maplen) {
3437 3438
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
3439 3440
    virNodeInfo nodeinfo;
    int i, v, maxcpu;
3441
    int ret = -1;
3442

3443
    qemuDriverLock(driver);
3444
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3445 3446
    qemuDriverUnlock(driver);

3447 3448 3449 3450 3451 3452 3453 3454
    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;
    }

3455
    if (!virDomainIsActive(vm)) {
3456
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
3457
                         "%s",_("cannot pin vcpus on an inactive domain"));
3458
        goto cleanup;
3459 3460
    }

3461
    if (nodeGetInfo(dom->conn, &nodeinfo) < 0)
3462
        goto cleanup;
3463 3464 3465 3466 3467 3468 3469 3470 3471

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

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

3472 3473 3474 3475 3476 3477 3478 3479
    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;
                /* XXX cpu time, current pCPU mapping */
            }
3480 3481
        }

3482 3483 3484 3485 3486 3487 3488 3489 3490
        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) {
3491 3492
                        virReportSystemError(dom->conn, errno, "%s",
                                             _("cannot get affinity"));
3493 3494 3495 3496 3497 3498
                        goto cleanup;
                    }

                    for (i = 0 ; i < maxcpu ; i++)
                        if (CPU_ISSET(i, &mask))
                            VIR_USE_CPU(cpumap, i);
3499
                }
3500 3501 3502 3503
            } else {
                qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                                 "%s", _("cpu affinity is not available"));
                goto cleanup;
3504 3505 3506
            }
        }
    }
3507
    ret = maxinfo;
3508

3509
cleanup:
3510 3511
    if (vm)
        virDomainObjUnlock(vm);
3512
    return ret;
3513 3514 3515 3516
}
#endif /* HAVE_SCHED_GETAFFINITY */


3517
static int qemudDomainGetMaxVcpus(virDomainPtr dom) {
3518 3519
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
3520
    const char *type;
3521
    int ret = -1;
3522

3523
    qemuDriverLock(driver);
3524
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3525 3526
    qemuDriverUnlock(driver);

3527
    if (!vm) {
3528 3529
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
3530
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
3531
                         _("no domain with matching uuid '%s'"), uuidstr);
3532
        goto cleanup;
3533 3534
    }

3535
    if (!(type = virDomainVirtTypeToString(vm->def->virtType))) {
3536 3537 3538
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
                         _("unknown virt type in domain definition '%d'"),
                         vm->def->virtType);
3539
        goto cleanup;
3540 3541
    }

3542
    ret = qemudGetMaxVCPUs(dom->conn, type);
3543

3544
cleanup:
3545 3546
    if (vm)
        virDomainObjUnlock(vm);
3547 3548 3549
    return ret;
}

3550 3551 3552 3553 3554 3555 3556 3557 3558 3559
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);

3560 3561
    memset(seclabel, 0, sizeof(*seclabel));

3562 3563 3564
    if (!vm) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(dom->uuid, uuidstr);
3565
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_DOMAIN,
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
                         _("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,
                                 _("Failed to get security label"));
                goto cleanup;
            }
        }
    }

    ret = 0;

cleanup:
    if (vm)
        virDomainObjUnlock(vm);
3606
    qemuDriverUnlock(driver);
3607 3608 3609
    return ret;
}

3610 3611
static int qemudNodeGetSecurityModel(virConnectPtr conn,
                                     virSecurityModelPtr secmodel)
3612 3613 3614
{
    struct qemud_driver *driver = (struct qemud_driver *)conn->privateData;
    char *p;
3615
    int ret = 0;
3616

3617 3618
    qemuDriverLock(driver);
    if (!driver->securityDriver) {
3619
        memset(secmodel, 0, sizeof (*secmodel));
3620 3621
        goto cleanup;
    }
3622

3623 3624 3625 3626 3627
    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);
3628 3629
        ret = -1;
        goto cleanup;
3630 3631 3632 3633 3634 3635 3636 3637
    }
    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);
3638 3639
        ret = -1;
        goto cleanup;
3640 3641
    }
    strcpy(secmodel->doi, p);
3642 3643 3644 3645

cleanup:
    qemuDriverUnlock(driver);
    return ret;
3646 3647 3648
}

/* TODO: check seclabel restore */
3649
static int qemudDomainRestore(virConnectPtr conn,
3650 3651 3652
                              const char *path) {
    struct qemud_driver *driver = conn->privateData;
    virDomainDefPtr def = NULL;
3653
    virDomainObjPtr vm = NULL;
3654 3655 3656
    int fd = -1;
    int ret = -1;
    char *xml = NULL;
3657
    struct qemud_save_header header;
3658
    virDomainEventPtr event = NULL;
3659

3660
    qemuDriverLock(driver);
3661 3662 3663
    /* Verify the header and read the XML */
    if ((fd = open(path, O_RDONLY)) < 0) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
3664
                         "%s", _("cannot read domain image"));
3665
        goto cleanup;
3666 3667 3668 3669
    }

    if (saferead(fd, &header, sizeof(header)) != sizeof(header)) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
3670
                         "%s", _("failed to read qemu header"));
3671
        goto cleanup;
3672 3673 3674 3675
    }

    if (memcmp(header.magic, QEMUD_SAVE_MAGIC, sizeof(header.magic)) != 0) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
3676
                         "%s", _("image magic is incorrect"));
3677
        goto cleanup;
3678 3679 3680 3681
    }

    if (header.version > QEMUD_SAVE_VERSION) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
3682
                         _("image version is not supported (%d > %d)"),
3683
                         header.version, QEMUD_SAVE_VERSION);
3684
        goto cleanup;
3685 3686
    }

3687
    if (VIR_ALLOC_N(xml, header.xml_len) < 0) {
3688
        virReportOOMError(conn);
3689
        goto cleanup;
3690 3691 3692 3693
    }

    if (saferead(fd, xml, header.xml_len) != header.xml_len) {
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
3694
                         "%s", _("failed to read XML"));
3695
        goto cleanup;
3696 3697 3698
    }

    /* Create a domain from this XML */
3699 3700
    if (!(def = virDomainDefParseString(conn, driver->caps, xml,
                                        VIR_DOMAIN_XML_INACTIVE))) {
3701
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
3702
                         "%s", _("failed to parse XML"));
3703
        goto cleanup;
3704 3705
    }

3706
    /* See if a VM with matching UUID already exists */
3707
    vm = virDomainFindByUUID(&driver->domains, def->uuid);
3708
    if (vm) {
3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719
        /* 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 */
3720
        if (virDomainIsActive(vm)) {
3721
            qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_INVALID,
3722 3723
                             _("domain is already active as '%s'"), vm->def->name);
            goto cleanup;
3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735
        }
        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;
3736
        }
3737 3738
    }

3739 3740 3741
    if (!(vm = virDomainAssignDef(conn,
                                  &driver->domains,
                                  def))) {
3742
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
3743
                         "%s", _("failed to assign new VM"));
3744
        goto cleanup;
3745
    }
3746
    def = NULL;
3747 3748

    /* Set the migration source and start it up. */
3749
    ret = qemudStartVMDaemon(conn, driver, vm, "stdio", fd);
3750
    close(fd);
3751
    fd = -1;
3752
    if (ret < 0) {
3753
        if (!vm->persistent) {
3754 3755
            virDomainRemoveInactive(&driver->domains,
                                    vm);
3756 3757
            vm = NULL;
        }
3758
        goto cleanup;
3759 3760
    }

3761 3762 3763
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_RESTORED);
3764

3765 3766 3767
    /* If it was running before, resume it now. */
    if (header.was_running) {
        char *info;
3768
        if (qemudMonitorCommand(vm, "cont", &info) < 0) {
3769
            qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
3770
                             "%s", _("failed to resume domain"));
3771
            goto cleanup;
3772
        }
3773
        VIR_FREE(info);
3774 3775
        vm->state = VIR_DOMAIN_RUNNING;
    }
3776
    ret = 0;
3777

3778 3779 3780 3781 3782
cleanup:
    virDomainDefFree(def);
    VIR_FREE(xml);
    if (fd != -1)
        close(fd);
3783 3784
    if (vm)
        virDomainObjUnlock(vm);
3785 3786
    if (event)
        qemuDomainEventQueue(driver, event);
3787
    qemuDriverUnlock(driver);
3788
    return ret;
D
Daniel P. Berrange 已提交
3789 3790 3791
}


3792
static char *qemudDomainDumpXML(virDomainPtr dom,
3793
                                int flags) {
3794 3795 3796
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    char *ret = NULL;
3797 3798
    unsigned long balloon;
    int err;
3799

3800
    qemuDriverLock(driver);
3801
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
3802 3803
    qemuDriverUnlock(driver);

D
Daniel P. Berrange 已提交
3804
    if (!vm) {
3805 3806 3807 3808
        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);
3809
        goto cleanup;
D
Daniel P. Berrange 已提交
3810 3811
    }

3812 3813 3814 3815 3816 3817 3818
    /* 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;

3819 3820 3821 3822 3823 3824
    ret = virDomainDefFormat(dom->conn,
                             (flags & VIR_DOMAIN_XML_INACTIVE) && vm->newDef ?
                             vm->newDef : vm->def,
                             flags);

cleanup:
3825 3826
    if (vm)
        virDomainObjUnlock(vm);
3827
    return ret;
D
Daniel P. Berrange 已提交
3828 3829 3830
}


3831 3832 3833 3834
static char *qemuDomainXMLFromNative(virConnectPtr conn,
                                     const char *format,
                                     const char *config,
                                     unsigned int flags ATTRIBUTE_UNUSED) {
3835
    struct qemud_driver *driver = conn->privateData;
3836 3837 3838 3839 3840 3841 3842 3843 3844
    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;
    }

3845
    def = qemuParseCommandLineString(conn, driver->caps, config);
3846 3847 3848 3849 3850 3851 3852 3853 3854 3855
    if (!def)
        goto cleanup;

    xml = virDomainDefFormat(conn, def, VIR_DOMAIN_XML_INACTIVE);

cleanup:
    virDomainDefFree(def);
    return xml;
}

3856 3857 3858 3859 3860 3861
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;
3862
    virDomainChrDef monitor_chr;
3863 3864 3865 3866 3867 3868 3869 3870 3871 3872
    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;

3873 3874
    qemuDriverLock(driver);

3875 3876 3877 3878 3879 3880 3881 3882 3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940
    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;
    }

3941 3942
    if (qemuPrepareMonitorChr(conn, driver, &monitor_chr, def->name) < 0)
        goto cleanup;
3943 3944

    if (qemudBuildCommandLine(conn, driver, def,
3945
                              &monitor_chr, qemuCmdFlags,
3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 3969 3970
                              &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:
3971
    qemuDriverUnlock(driver);
3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983 3984
    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;
}


3985
static int qemudListDefinedDomains(virConnectPtr conn,
3986
                            char **const names, int nnames) {
3987
    struct qemud_driver *driver = conn->privateData;
3988
    int got = 0, i;
3989

3990
    qemuDriverLock(driver);
3991
    for (i = 0 ; i < driver->domains.count && got < nnames ; i++) {
3992
        virDomainObjLock(driver->domains.objs[i]);
3993 3994
        if (!virDomainIsActive(driver->domains.objs[i])) {
            if (!(names[got++] = strdup(driver->domains.objs[i]->def->name))) {
3995
                virReportOOMError(conn);
3996
                virDomainObjUnlock(driver->domains.objs[i]);
3997 3998
                goto cleanup;
            }
3999
        }
4000
        virDomainObjUnlock(driver->domains.objs[i]);
D
Daniel P. Berrange 已提交
4001
    }
4002

4003
    qemuDriverUnlock(driver);
D
Daniel P. Berrange 已提交
4004
    return got;
4005 4006 4007

 cleanup:
    for (i = 0 ; i < got ; i++)
4008
        VIR_FREE(names[i]);
4009
    qemuDriverUnlock(driver);
4010
    return -1;
D
Daniel P. Berrange 已提交
4011 4012
}

4013
static int qemudNumDefinedDomains(virConnectPtr conn) {
4014
    struct qemud_driver *driver = conn->privateData;
4015 4016
    int n = 0, i;

4017
    qemuDriverLock(driver);
4018 4019
    for (i = 0 ; i < driver->domains.count ; i++)
        if (!virDomainIsActive(driver->domains.objs[i]))
4020
            n++;
4021
    qemuDriverUnlock(driver);
4022

4023
    return n;
D
Daniel P. Berrange 已提交
4024 4025 4026
}


4027
static int qemudDomainStart(virDomainPtr dom) {
4028 4029 4030
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4031
    virDomainEventPtr event = NULL;
4032

4033
    qemuDriverLock(driver);
4034
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4035

4036
    if (!vm) {
4037 4038 4039 4040
        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);
4041
        goto cleanup;
4042 4043
    }

4044
    ret = qemudStartVMDaemon(dom->conn, driver, vm, NULL, -1);
4045
    if (ret != -1)
4046 4047 4048
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STARTED,
                                         VIR_DOMAIN_EVENT_STARTED_BOOTED);
4049 4050

cleanup:
4051 4052
    if (vm)
        virDomainObjUnlock(vm);
4053
    if (event)
4054
        qemuDomainEventQueue(driver, event);
4055
    qemuDriverUnlock(driver);
4056
    return ret;
D
Daniel P. Berrange 已提交
4057 4058 4059
}


4060
static virDomainPtr qemudDomainDefine(virConnectPtr conn, const char *xml) {
4061
    struct qemud_driver *driver = conn->privateData;
4062
    virDomainDefPtr def;
4063
    virDomainObjPtr vm = NULL;
4064
    virDomainPtr dom = NULL;
4065
    virDomainEventPtr event = NULL;
4066
    int newVM = 1;
4067

4068
    qemuDriverLock(driver);
4069 4070
    if (!(def = virDomainDefParseString(conn, driver->caps, xml,
                                        VIR_DOMAIN_XML_INACTIVE)))
4071
        goto cleanup;
4072

4073 4074 4075
    if (virSecurityDriverVerify(conn, def) < 0)
        goto cleanup;

4076 4077
    /* See if a VM with matching UUID already exists */
    vm = virDomainFindByUUID(&driver->domains, def->uuid);
4078
    if (vm) {
4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089
        /* 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 */
4090
        virDomainObjUnlock(vm);
4091
        newVM = 0;
4092 4093 4094 4095 4096 4097 4098 4099 4100 4101 4102
    } 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;
        }
4103
    }
4104

4105 4106 4107
    if (!(vm = virDomainAssignDef(conn,
                                  &driver->domains,
                                  def))) {
4108
        goto cleanup;
4109
    }
4110
    def = NULL;
4111
    vm->persistent = 1;
4112

4113 4114
    if (virDomainSaveConfig(conn,
                            driver->configDir,
4115
                            vm->newDef ? vm->newDef : vm->def) < 0) {
4116 4117
        virDomainRemoveInactive(&driver->domains,
                                vm);
4118
        vm = NULL;
4119
        goto cleanup;
4120 4121
    }

4122 4123 4124 4125 4126
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_DEFINED,
                                     newVM ?
                                     VIR_DOMAIN_EVENT_DEFINED_ADDED :
                                     VIR_DOMAIN_EVENT_DEFINED_UPDATED);
4127

4128
    dom = virGetDomain(conn, vm->def->name, vm->def->uuid);
4129
    if (dom) dom->id = vm->def->id;
4130 4131

cleanup:
4132
    virDomainDefFree(def);
4133 4134
    if (vm)
        virDomainObjUnlock(vm);
4135 4136
    if (event)
        qemuDomainEventQueue(driver, event);
4137
    qemuDriverUnlock(driver);
4138
    return dom;
D
Daniel P. Berrange 已提交
4139 4140
}

4141
static int qemudDomainUndefine(virDomainPtr dom) {
4142 4143
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
4144
    virDomainEventPtr event = NULL;
4145
    int ret = -1;
D
Daniel P. Berrange 已提交
4146

4147
    qemuDriverLock(driver);
4148
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4149

D
Daniel P. Berrange 已提交
4150
    if (!vm) {
4151 4152 4153 4154
        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);
4155
        goto cleanup;
D
Daniel P. Berrange 已提交
4156 4157
    }

4158
    if (virDomainIsActive(vm)) {
4159
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
4160
                         "%s", _("cannot delete active domain"));
4161
        goto cleanup;
D
Daniel P. Berrange 已提交
4162 4163
    }

4164 4165 4166
    if (!vm->persistent) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
                         "%s", _("cannot undefine transient domain"));
4167
        goto cleanup;
4168 4169 4170
    }

    if (virDomainDeleteConfig(dom->conn, driver->configDir, driver->autostartDir, vm) < 0)
4171
        goto cleanup;
D
Daniel P. Berrange 已提交
4172

4173 4174 4175
    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_UNDEFINED,
                                     VIR_DOMAIN_EVENT_UNDEFINED_REMOVED);
4176

4177 4178
    virDomainRemoveInactive(&driver->domains,
                            vm);
4179
    vm = NULL;
4180
    ret = 0;
D
Daniel P. Berrange 已提交
4181

4182
cleanup:
4183 4184
    if (vm)
        virDomainObjUnlock(vm);
4185 4186
    if (event)
        qemuDomainEventQueue(driver, event);
4187
    qemuDriverUnlock(driver);
4188
    return ret;
D
Daniel P. Berrange 已提交
4189 4190
}

4191
/* Return the disks name for use in monitor commands */
4192
static char *qemudDiskDeviceName(const virConnectPtr conn,
4193
                                 const virDomainDiskDefPtr disk) {
4194 4195 4196 4197 4198 4199

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

    if (virDiskNameToBusDeviceIndex(disk, &busid, &devid) < 0) {
4200
        qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
4201 4202 4203 4204 4205 4206 4207
                         _("cannot convert disk '%s' to bus/device index"),
                         disk->dst);
        return NULL;
    }

    switch (disk->bus) {
        case VIR_DOMAIN_DISK_BUS_IDE:
4208
            if (disk->device== VIR_DOMAIN_DISK_DEVICE_DISK)
4209
                ret = virAsprintf(&devname, "ide%d-hd%d", busid, devid);
4210
            else
4211
                ret = virAsprintf(&devname, "ide%d-cd%d", busid, devid);
4212 4213
            break;
        case VIR_DOMAIN_DISK_BUS_SCSI:
4214
            if (disk->device == VIR_DOMAIN_DISK_DEVICE_DISK)
4215
                ret = virAsprintf(&devname, "scsi%d-hd%d", busid, devid);
4216
            else
4217
                ret = virAsprintf(&devname, "scsi%d-cd%d", busid, devid);
4218 4219
            break;
        case VIR_DOMAIN_DISK_BUS_FDC:
4220
            ret = virAsprintf(&devname, "floppy%d", devid);
4221 4222
            break;
        case VIR_DOMAIN_DISK_BUS_VIRTIO:
4223
            ret = virAsprintf(&devname, "virtio%d", devid);
4224 4225
            break;
        default:
4226
            qemudReportError(conn, NULL, NULL, VIR_ERR_NO_SUPPORT,
4227 4228 4229 4230 4231 4232
                             _("Unsupported disk name mapping for bus '%s'"),
                             virDomainDiskBusTypeToString(disk->bus));
            return NULL;
    }

    if (ret == -1) {
4233
        virReportOOMError(conn);
4234 4235 4236 4237 4238 4239
        return NULL;
    }

    return devname;
}

4240 4241
static int qemudDomainChangeEjectableMedia(virConnectPtr conn,
                                           virDomainObjPtr vm,
4242 4243
                                           virDomainDeviceDefPtr dev)
{
4244
    virDomainDiskDefPtr origdisk = NULL, newdisk;
4245
    char *cmd, *reply, *safe_path;
4246
    char *devname = NULL;
4247
    unsigned int qemuCmdFlags;
4248
    int i;
4249

4250
    origdisk = NULL;
4251
    newdisk = dev->data.disk;
4252 4253 4254 4255
    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];
4256
            break;
4257
        }
4258 4259 4260
    }

    if (!origdisk) {
4261
        qemudReportError(conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
4262 4263 4264 4265 4266 4267 4268 4269
                         _("No device with bus '%s' and target '%s'"),
                         virDomainDiskBusTypeToString(newdisk->bus),
                         newdisk->dst);
        return -1;
    }

    if (qemudExtractVersionInfo(vm->def->emulator,
                                NULL,
4270
                                &qemuCmdFlags) < 0)
4271 4272 4273
        return -1;

    if (qemuCmdFlags & QEMUD_CMD_FLAG_DRIVE) {
4274
        if (!(devname = qemudDiskDeviceName(conn, newdisk)))
4275 4276 4277 4278 4279 4280 4281 4282 4283
            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 {
4284
            qemudReportError(conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
4285 4286 4287 4288 4289 4290 4291 4292
                             _("Emulator version does not support removable "
                               "media for device '%s' and target '%s'"),
                               virDomainDiskDeviceTypeToString(newdisk->device),
                               newdisk->dst);
            return -1;
        }

        if (!devname) {
4293
            virReportOOMError(conn);
4294 4295 4296
            return -1;
        }
    }
4297

4298
    if (newdisk->src) {
4299 4300
        safe_path = qemudEscapeMonitorArg(newdisk->src);
        if (!safe_path) {
4301
            virReportOOMError(conn);
4302
            VIR_FREE(devname);
4303 4304
            return -1;
        }
4305
        if (virAsprintf(&cmd, "change %s \"%s\"", devname, safe_path) == -1) {
4306
            virReportOOMError(conn);
4307
            VIR_FREE(safe_path);
4308
            VIR_FREE(devname);
4309 4310
            return -1;
        }
4311
        VIR_FREE(safe_path);
4312

4313
    } else if (virAsprintf(&cmd, "eject %s", devname) == -1) {
4314
        virReportOOMError(conn);
4315
        VIR_FREE(devname);
4316 4317
        return -1;
    }
4318
    VIR_FREE(devname);
4319

4320
    if (qemudMonitorCommand(vm, cmd, &reply) < 0) {
4321
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
C
Cole Robinson 已提交
4322
                         "%s", _("could not change cdrom media"));
4323
        VIR_FREE(cmd);
4324 4325
        return -1;
    }
4326 4327 4328 4329

    /* If the command failed qemu prints:
     * device not found, device is locked ...
     * No message is printed on success it seems */
4330
    DEBUG ("%s: ejectable media change reply: %s", vm->def->name, reply);
4331
    if (strstr(reply, "\ndevice ")) {
4332
        qemudReportError (conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
C
Cole Robinson 已提交
4333
                          _("changing cdrom media failed: %s"), reply);
4334 4335 4336 4337
        VIR_FREE(reply);
        VIR_FREE(cmd);
        return -1;
    }
4338 4339
    VIR_FREE(reply);
    VIR_FREE(cmd);
4340

4341 4342
    VIR_FREE(origdisk->src);
    origdisk->src = newdisk->src;
4343
    newdisk->src = NULL;
4344
    origdisk->type = newdisk->type;
4345 4346 4347
    return 0;
}

4348 4349 4350
static int qemudDomainAttachPciDiskDevice(virConnectPtr conn,
                                          virDomainObjPtr vm,
                                          virDomainDeviceDefPtr dev)
4351 4352
{
    int ret, i;
4353
    char *cmd, *reply, *s;
4354 4355
    char *safe_path;
    const char* type = virDomainDiskBusTypeToString(dev->data.disk->bus);
4356
    int tryOldSyntax = 0;
4357 4358 4359

    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (STREQ(vm->def->disks[i]->dst, dev->data.disk->dst)) {
4360
            qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
4361 4362 4363 4364 4365 4366
                           _("target %s already exists"), dev->data.disk->dst);
            return -1;
        }
    }

    if (VIR_REALLOC_N(vm->def->disks, vm->def->ndisks+1) < 0) {
4367
        virReportOOMError(conn);
4368 4369 4370
        return -1;
    }

4371
try_command:
4372 4373
    safe_path = qemudEscapeMonitorArg(dev->data.disk->src);
    if (!safe_path) {
4374
        virReportOOMError(conn);
4375 4376 4377
        return -1;
    }

4378 4379
    ret = virAsprintf(&cmd, "pci_add %s storage file=%s,if=%s",
                      (tryOldSyntax ? "0": "pci_addr=auto"), safe_path, type);
4380 4381
    VIR_FREE(safe_path);
    if (ret == -1) {
4382
        virReportOOMError(conn);
4383 4384 4385
        return ret;
    }

4386
    if (qemudMonitorCommand(vm, cmd, &reply) < 0) {
4387
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
4388 4389 4390 4391 4392
                         _("cannot attach %s disk"), type);
        VIR_FREE(cmd);
        return -1;
    }

4393
    DEBUG ("%s: pci_add reply: %s", vm->def->name, reply);
4394
    /* If the command succeeds qemu prints:
4395 4396 4397 4398 4399 4400 4401 4402
     * OK bus 0, slot XXX...
     * or
     * OK domain 0, bus 0, slot XXX
     */
    if ((s = strstr(reply, "OK ")) &&
        (s = strstr(s, "slot "))) {
        char *dummy = s;
        s += strlen("slot ");
4403 4404

        if (virStrToLong_i ((const char*)s, &dummy, 10, &dev->data.disk->slotnum) == -1)
4405
            VIR_WARN("%s", _("Unable to parse slot number\n"));
4406 4407 4408 4409 4410 4411 4412
        /* XXX not neccessarily always going to end up in domain 0 / bus 0 :-( */
        /* XXX this slotnum is not persistant across restarts :-( */
    } else if (!tryOldSyntax && strstr(reply, "invalid char in expression")) {
        VIR_FREE(reply);
        VIR_FREE(cmd);
        tryOldSyntax = 1;
        goto try_command;
4413
    } else {
4414
        qemudReportError (conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
4415
                          _("adding %s disk failed: %s"), type, reply);
4416 4417 4418 4419 4420 4421 4422 4423 4424 4425 4426 4427 4428
        VIR_FREE(reply);
        VIR_FREE(cmd);
        return -1;
    }

    vm->def->disks[vm->def->ndisks++] = dev->data.disk;
    qsort(vm->def->disks, vm->def->ndisks, sizeof(*vm->def->disks),
          virDomainDiskQSort);

    VIR_FREE(reply);
    VIR_FREE(cmd);
    return 0;
}
4429

4430 4431 4432
static int qemudDomainAttachUsbMassstorageDevice(virConnectPtr conn,
                                                 virDomainObjPtr vm,
                                                 virDomainDeviceDefPtr dev)
4433
{
4434 4435
    int ret, i;
    char *safe_path;
4436 4437
    char *cmd, *reply;

4438 4439
    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (STREQ(vm->def->disks[i]->dst, dev->data.disk->dst)) {
4440
            qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
4441 4442 4443 4444 4445
                           _("target %s already exists"), dev->data.disk->dst);
            return -1;
        }
    }

4446
    if (VIR_REALLOC_N(vm->def->disks, vm->def->ndisks+1) < 0) {
4447
        virReportOOMError(conn);
4448
        return -1;
4449 4450
    }

4451 4452
    safe_path = qemudEscapeMonitorArg(dev->data.disk->src);
    if (!safe_path) {
4453
        virReportOOMError(conn);
4454 4455 4456
        return -1;
    }

4457
    ret = virAsprintf(&cmd, "usb_add disk:%s", safe_path);
4458
    VIR_FREE(safe_path);
4459
    if (ret == -1) {
4460
        virReportOOMError(conn);
4461 4462 4463
        return ret;
    }

4464
    if (qemudMonitorCommand(vm, cmd, &reply) < 0) {
4465
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
4466
                         "%s", _("cannot attach usb disk"));
4467 4468 4469 4470
        VIR_FREE(cmd);
        return -1;
    }

4471
    DEBUG ("%s: attach_usb reply: %s",vm->def->name,  reply);
4472 4473 4474
    /* If the command failed qemu prints:
     * Could not add ... */
    if (strstr(reply, "Could not add ")) {
4475
        qemudReportError (conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
4476
                          "%s",
4477
                          _("adding usb disk failed"));
4478 4479 4480 4481
        VIR_FREE(reply);
        VIR_FREE(cmd);
        return -1;
    }
4482

4483 4484 4485
    vm->def->disks[vm->def->ndisks++] = dev->data.disk;
    qsort(vm->def->disks, vm->def->ndisks, sizeof(*vm->def->disks),
          virDomainDiskQSort);
4486

4487 4488 4489 4490 4491
    VIR_FREE(reply);
    VIR_FREE(cmd);
    return 0;
}

4492 4493 4494
static int qemudDomainAttachHostDevice(virConnectPtr conn,
                                       virDomainObjPtr vm,
                                       virDomainDeviceDefPtr dev)
4495 4496 4497 4498
{
    int ret;
    char *cmd, *reply;

4499
    if (VIR_REALLOC_N(vm->def->hostdevs, vm->def->nhostdevs+1) < 0) {
4500
        virReportOOMError(conn);
4501 4502
        return -1;
    }
4503

4504
    if (dev->data.hostdev->source.subsys.u.usb.vendor) {
4505 4506 4507
        ret = virAsprintf(&cmd, "usb_add host:%.4x:%.4x",
                          dev->data.hostdev->source.subsys.u.usb.vendor,
                          dev->data.hostdev->source.subsys.u.usb.product);
4508
    } else {
4509 4510 4511
        ret = virAsprintf(&cmd, "usb_add host:%.3d.%.3d",
                          dev->data.hostdev->source.subsys.u.usb.bus,
                          dev->data.hostdev->source.subsys.u.usb.device);
4512 4513
    }
    if (ret == -1) {
4514
        virReportOOMError(conn);
4515 4516 4517
        return -1;
    }

4518
    if (qemudMonitorCommand(vm, cmd, &reply) < 0) {
4519
        qemudReportError(conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
4520 4521 4522 4523 4524
                         "%s", _("cannot attach usb device"));
        VIR_FREE(cmd);
        return -1;
    }

4525
    DEBUG ("%s: attach_usb reply: %s", vm->def->name, reply);
4526 4527 4528
    /* If the command failed qemu prints:
     * Could not add ... */
    if (strstr(reply, "Could not add ")) {
4529
        qemudReportError (conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
4530 4531 4532 4533 4534 4535
                          "%s",
                          _("adding usb device failed"));
        VIR_FREE(reply);
        VIR_FREE(cmd);
        return -1;
    }
4536

4537
    vm->def->hostdevs[vm->def->nhostdevs++] = dev->data.hostdev;
4538

4539 4540
    VIR_FREE(reply);
    VIR_FREE(cmd);
4541 4542 4543
    return 0;
}

4544 4545
static int qemudDomainAttachDevice(virDomainPtr dom,
                                   const char *xml) {
4546 4547 4548 4549
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    virDomainDeviceDefPtr dev = NULL;
    int ret = -1;
4550

4551
    qemuDriverLock(driver);
4552
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4553
    if (!vm) {
4554 4555 4556 4557
        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);
4558
        goto cleanup;
4559 4560 4561
    }

    if (!virDomainIsActive(vm)) {
4562
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
4563
                         "%s", _("cannot attach device on inactive domain"));
4564
        goto cleanup;
4565 4566
    }

4567 4568
    dev = virDomainDeviceDefParse(dom->conn, driver->caps, vm->def, xml,
                                  VIR_DOMAIN_XML_INACTIVE);
4569 4570 4571
    if (dev == NULL)
        goto cleanup;

4572

4573 4574
    if (dev->type == VIR_DOMAIN_DEVICE_DISK) {
        switch (dev->data.disk->device) {
4575 4576
        case VIR_DOMAIN_DISK_DEVICE_CDROM:
        case VIR_DOMAIN_DISK_DEVICE_FLOPPY:
4577 4578
            if (driver->securityDriver)
                driver->securityDriver->domainSetSecurityImageLabel(dom->conn, vm, dev->data.disk);
4579 4580 4581 4582

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

4583
            ret = qemudDomainChangeEjectableMedia(dom->conn, vm, dev);
4584
            break;
4585

4586
        case VIR_DOMAIN_DISK_DEVICE_DISK:
4587 4588
            if (driver->securityDriver)
                driver->securityDriver->domainSetSecurityImageLabel(dom->conn, vm, dev->data.disk);
4589 4590 4591 4592

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

4593
            if (dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_USB) {
4594
                ret = qemudDomainAttachUsbMassstorageDevice(dom->conn, vm, dev);
4595 4596
            } else if (dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_SCSI ||
                       dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_VIRTIO) {
4597
                ret = qemudDomainAttachPciDiskDevice(dom->conn, vm, dev);
4598 4599 4600 4601 4602
            } else {
                qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                                 _("disk bus '%s' cannot be hotplugged."),
                                 virDomainDiskBusTypeToString(dev->data.disk->bus));
                goto cleanup;
4603 4604
            }
            break;
4605

4606 4607
        default:
            qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
4608 4609
                             _("disk device type '%s' cannot be hotplugged"),
                             virDomainDiskDeviceTypeToString(dev->data.disk->device));
4610
            goto cleanup;
4611
        }
4612
    } else if (dev->type == VIR_DOMAIN_DEVICE_HOSTDEV &&
4613 4614
               dev->data.hostdev->mode == VIR_DOMAIN_HOSTDEV_MODE_SUBSYS &&
               dev->data.hostdev->source.subsys.type == VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB) {
4615 4616 4617
        if (qemuDomainSetDeviceOwnership(dom->conn, driver, dev, 0) < 0)
            goto cleanup;

4618
        ret = qemudDomainAttachHostDevice(dom->conn, vm, dev);
4619
    } else {
4620
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
4621 4622
                         _("device type '%s' cannot be attached"),
                         virDomainDeviceTypeToString(dev->type));
4623
        goto cleanup;
4624 4625
    }

4626
    if (!ret && virDomainSaveStatus(dom->conn, driver->stateDir, vm) < 0)
4627 4628
        ret = -1;

4629
cleanup:
4630 4631 4632
    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
4633
        virDomainDeviceDefFree(dev);
4634
    }
4635 4636
    if (vm)
        virDomainObjUnlock(vm);
4637
    qemuDriverUnlock(driver);
4638 4639 4640
    return ret;
}

4641 4642
static int qemudDomainDetachPciDiskDevice(virConnectPtr conn,
                                          virDomainObjPtr vm, virDomainDeviceDefPtr dev)
4643 4644
{
    int i, ret = -1;
4645 4646
    char *cmd = NULL;
    char *reply = NULL;
4647
    virDomainDiskDefPtr detach = NULL;
4648
    int tryOldSyntax = 0;
4649 4650 4651 4652 4653 4654 4655 4656 4657

    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) {
4658
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
4659
                         _("disk %s not found"), dev->data.disk->dst);
4660
        goto cleanup;
4661 4662 4663
    }

    if (detach->slotnum < 1) {
4664
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
4665
                         _("disk %s cannot be detached - invalid slot number %d"),
4666
                           detach->dst, detach->slotnum);
4667
        goto cleanup;
4668 4669
    }

4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680
try_command:
    if (tryOldSyntax) {
        if (virAsprintf(&cmd, "pci_del 0 %d", detach->slotnum) < 0) {
            virReportOOMError(conn);
            goto cleanup;
        }
    } else {
        if (virAsprintf(&cmd, "pci_del pci_addr=0:0:%d", detach->slotnum) < 0) {
            virReportOOMError(conn);
            goto cleanup;
        }
4681 4682
    }

4683
    if (qemudMonitorCommand(vm, cmd, &reply) < 0) {
4684
        qemudReportError(conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
4685
                          _("failed to execute detach disk %s command"), detach->dst);
4686
        goto cleanup;
4687 4688
    }

4689
    DEBUG ("%s: pci_del reply: %s",vm->def->name,  reply);
4690 4691 4692 4693 4694 4695

    if (!tryOldSyntax &&
        strstr(reply, "extraneous characters")) {
        tryOldSyntax = 1;
        goto try_command;
    }
4696 4697
    /* If the command fails due to a wrong slot qemu prints: invalid slot,
     * nothing is printed on success */
4698 4699
    if (strstr(reply, "invalid slot") ||
        strstr(reply, "Invalid pci address")) {
4700
        qemudReportError (conn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
4701 4702
                          _("failed to detach disk %s: invalid slot %d: %s"),
                          detach->dst, detach->slotnum, reply);
4703
        goto cleanup;
4704 4705 4706 4707 4708
    }

    if (vm->def->ndisks > 1) {
        vm->def->disks[i] = vm->def->disks[--vm->def->ndisks];
        if (VIR_REALLOC_N(vm->def->disks, vm->def->ndisks) < 0) {
4709
            virReportOOMError(conn);
4710
            goto cleanup;
4711 4712 4713 4714 4715 4716 4717 4718
        }
        qsort(vm->def->disks, vm->def->ndisks, sizeof(*vm->def->disks),
              virDomainDiskQSort);
    } else {
        VIR_FREE(vm->def->disks[0]);
        vm->def->ndisks = 0;
    }
    ret = 0;
4719 4720

cleanup:
4721 4722 4723 4724 4725 4726 4727
    VIR_FREE(reply);
    VIR_FREE(cmd);
    return ret;
}

static int qemudDomainDetachDevice(virDomainPtr dom,
                                   const char *xml) {
4728 4729 4730 4731
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    virDomainDeviceDefPtr dev = NULL;
    int ret = -1;
4732

4733
    qemuDriverLock(driver);
4734
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4735
    if (!vm) {
4736 4737 4738 4739
        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);
4740
        goto cleanup;
4741 4742 4743
    }

    if (!virDomainIsActive(vm)) {
4744
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
G
Guido Günther 已提交
4745
                         "%s", _("cannot detach device on inactive domain"));
4746
        goto cleanup;
4747 4748
    }

4749 4750
    dev = virDomainDeviceDefParse(dom->conn, driver->caps, vm->def, xml,
                                  VIR_DOMAIN_XML_INACTIVE);
4751 4752 4753
    if (dev == NULL)
        goto cleanup;

4754 4755 4756 4757

    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 ||
4758
         dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_VIRTIO)) {
4759
        ret = qemudDomainDetachPciDiskDevice(dom->conn, vm, dev);
4760
        if (driver->securityDriver)
4761
            driver->securityDriver->domainRestoreSecurityImageLabel(dom->conn, dev->data.disk);
4762 4763
        if (qemuDomainSetDeviceOwnership(dom->conn, driver, dev, 1) < 0)
            VIR_WARN0("Fail to restore disk device ownership");
4764
    }
4765
    else
4766 4767 4768
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                         "%s", _("only SCSI or virtio disk device can be detached dynamically"));

4769
    if (!ret && virDomainSaveStatus(dom->conn, driver->stateDir, vm) < 0)
4770 4771
        ret = -1;

4772 4773
cleanup:
    virDomainDeviceDefFree(dev);
4774 4775
    if (vm)
        virDomainObjUnlock(vm);
4776
    qemuDriverUnlock(driver);
4777 4778 4779
    return ret;
}

4780
static int qemudDomainGetAutostart(virDomainPtr dom,
4781
                                   int *autostart) {
4782 4783 4784
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int ret = -1;
4785

4786
    qemuDriverLock(driver);
4787
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4788 4789
    qemuDriverUnlock(driver);

4790
    if (!vm) {
4791 4792 4793 4794
        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);
4795
        goto cleanup;
4796 4797 4798
    }

    *autostart = vm->autostart;
4799
    ret = 0;
4800

4801
cleanup:
4802 4803
    if (vm)
        virDomainObjUnlock(vm);
4804
    return ret;
4805 4806
}

4807
static int qemudDomainSetAutostart(virDomainPtr dom,
4808
                                   int autostart) {
4809 4810
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
4811 4812
    char *configFile = NULL, *autostartLink = NULL;
    int ret = -1;
4813

4814
    qemuDriverLock(driver);
4815
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4816

4817
    if (!vm) {
4818 4819 4820 4821
        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);
4822
        goto cleanup;
4823 4824
    }

4825 4826 4827
    if (!vm->persistent) {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INTERNAL_ERROR,
                         "%s", _("cannot set autostart for transient domain"));
4828
        goto cleanup;
4829 4830
    }

4831 4832
    autostart = (autostart != 0);

4833 4834 4835 4836 4837
    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;
4838

4839 4840
        if (autostart) {
            int err;
4841

4842
            if ((err = virFileMakePath(driver->autostartDir))) {
4843 4844 4845
                virReportSystemError(dom->conn, err,
                                     _("cannot create autostart directory %s"),
                                     driver->autostartDir);
4846 4847
                goto cleanup;
            }
4848

4849
            if (symlink(configFile, autostartLink) < 0) {
4850 4851 4852
                virReportSystemError(dom->conn, errno,
                                     _("Failed to create symlink '%s to '%s'"),
                                     autostartLink, configFile);
4853 4854 4855 4856
                goto cleanup;
            }
        } else {
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
4857 4858 4859
                virReportSystemError(dom->conn, errno,
                                     _("Failed to delete symlink '%s'"),
                                     autostartLink);
4860 4861
                goto cleanup;
            }
4862 4863
        }

4864
        vm->autostart = autostart;
4865
    }
4866
    ret = 0;
4867

4868 4869 4870
cleanup:
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
4871 4872
    if (vm)
        virDomainObjUnlock(vm);
4873
    qemuDriverUnlock(driver);
4874
    return ret;
4875 4876
}

4877 4878 4879 4880 4881 4882 4883 4884 4885
/* 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)
{
4886
    struct qemud_driver *driver = dom->conn->privateData;
4887
    char *dummy, *info = NULL;
4888
    const char *p, *eol;
4889
    const char *qemu_dev_name = NULL;
4890
    size_t len;
4891
    int i, ret = -1;
4892
    virDomainObjPtr vm;
4893
    virDomainDiskDefPtr disk = NULL;
4894

4895
    qemuDriverLock(driver);
4896
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
4897
    qemuDriverUnlock(driver);
4898
    if (!vm) {
4899 4900 4901 4902
        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);
4903
        goto cleanup;
4904
    }
4905
    if (!virDomainIsActive (vm)) {
4906
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
4907
                          "%s", _("domain is not running"));
4908
        goto cleanup;
4909 4910
    }

4911 4912 4913 4914 4915 4916 4917 4918
    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (STREQ(path, vm->def->disks[i]->dst)) {
            disk = vm->def->disks[i];
            break;
        }
    }

    if (!disk) {
4919 4920
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
                          _("invalid path: %s"), path);
4921
        goto cleanup;
4922 4923
    }

4924
    qemu_dev_name = qemudDiskDeviceName(dom->conn, disk);
4925
    if (!qemu_dev_name)
4926
        goto cleanup;
4927 4928
    len = strlen (qemu_dev_name);

4929
    if (qemudMonitorCommand (vm, "info blockstats", &info) < 0) {
4930 4931
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                          "%s", _("'info blockstats' command failed"));
4932
        goto cleanup;
4933
    }
4934
    DEBUG ("%s: info blockstats reply: %s", vm->def->name, info);
4935 4936 4937 4938 4939 4940

    /* 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
4941
    if (strstr(info, "\ninfo ")) {
4942 4943 4944
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                          "%s",
                          _("'info blockstats' not supported by this qemu"));
4945
        goto cleanup;
4946 4947 4948 4949 4950 4951 4952 4953 4954 4955 4956 4957 4958 4959 4960 4961 4962 4963 4964 4965 4966 4967 4968 4969 4970 4971
    }

    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) {
4972
                if (STRPREFIX (p, "rd_bytes=")) {
4973 4974
                    p += 9;
                    if (virStrToLong_ll (p, &dummy, 10, &stats->rd_bytes) == -1)
4975 4976
                        DEBUG ("%s: error reading rd_bytes: %s",
                               vm->def->name, p);
4977
                } else if (STRPREFIX (p, "wr_bytes=")) {
4978 4979
                    p += 9;
                    if (virStrToLong_ll (p, &dummy, 10, &stats->wr_bytes) == -1)
4980 4981
                        DEBUG ("%s: error reading wr_bytes: %s",
                               vm->def->name, p);
4982
                } else if (STRPREFIX (p, "rd_operations=")) {
4983 4984
                    p += 14;
                    if (virStrToLong_ll (p, &dummy, 10, &stats->rd_req) == -1)
4985 4986
                        DEBUG ("%s: error reading rd_req: %s",
                               vm->def->name, p);
4987
                } else if (STRPREFIX (p, "wr_operations=")) {
4988 4989
                    p += 14;
                    if (virStrToLong_ll (p, &dummy, 10, &stats->wr_req) == -1)
4990 4991
                        DEBUG ("%s: error reading wr_req: %s",
                               vm->def->name, p);
4992
                } else
4993
                    DEBUG ("%s: unknown block stat near %s", vm->def->name, p);
4994 4995 4996 4997 4998 4999

                /* Skip to next label. */
                p = strchr (p, ' ');
                if (!p || p >= eol) break;
                p++;
            }
5000
            ret = 0;
5001
            goto cleanup;
5002 5003 5004 5005 5006 5007 5008 5009 5010 5011 5012
        }

        /* 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);
5013
 cleanup:
5014 5015
    VIR_FREE(qemu_dev_name);
    VIR_FREE(info);
5016 5017
    if (vm)
        virDomainObjUnlock(vm);
5018
    return ret;
5019 5020
}

5021
#ifdef __linux__
5022 5023 5024 5025 5026
static int
qemudDomainInterfaceStats (virDomainPtr dom,
                           const char *path,
                           struct _virDomainInterfaceStats *stats)
{
5027 5028
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
5029
    int i;
5030
    int ret = -1;
5031

5032
    qemuDriverLock(driver);
5033
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5034 5035
    qemuDriverUnlock(driver);

5036
    if (!vm) {
5037 5038 5039 5040
        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);
5041
        goto cleanup;
5042 5043
    }

5044
    if (!virDomainIsActive(vm)) {
5045
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
5046
                         "%s", _("domain is not running"));
5047
        goto cleanup;
5048 5049 5050 5051
    }

    if (!path || path[0] == '\0') {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
5052
                         "%s", _("NULL or empty path"));
5053
        goto cleanup;
5054 5055 5056
    }

    /* Check the path is one of the domain's network interfaces. */
5057 5058
    for (i = 0 ; i < vm->def->nnets ; i++) {
        if (vm->def->nets[i]->ifname &&
5059 5060 5061 5062
            STREQ (vm->def->nets[i]->ifname, path)) {
            ret = 0;
            break;
        }
5063 5064
    }

5065 5066 5067 5068 5069
    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);
5070

5071
cleanup:
5072 5073
    if (vm)
        virDomainObjUnlock(vm);
5074 5075
    return ret;
}
5076
#else
5077 5078 5079 5080
static int
qemudDomainInterfaceStats (virDomainPtr dom,
                           const char *path ATTRIBUTE_UNUSED,
                           struct _virDomainInterfaceStats *stats ATTRIBUTE_UNUSED)
5081 5082 5083 5084
    qemudReportError (dom->conn, dom, NULL, VIR_ERR_NO_SUPPORT,
                      "%s", __FUNCTION__);
    return -1;
}
5085
#endif
5086

5087 5088 5089 5090 5091 5092 5093
static int
qemudDomainBlockPeek (virDomainPtr dom,
                      const char *path,
                      unsigned long long offset, size_t size,
                      void *buffer,
                      unsigned int flags ATTRIBUTE_UNUSED)
{
5094 5095 5096
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    int fd = -1, ret = -1, i;
5097

5098
    qemuDriverLock(driver);
5099
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5100 5101
    qemuDriverUnlock(driver);

5102
    if (!vm) {
5103 5104 5105 5106
        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);
5107
        goto cleanup;
5108 5109 5110 5111
    }

    if (!path || path[0] == '\0') {
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
J
Jim Meyering 已提交
5112
                         "%s", _("NULL or empty path"));
5113
        goto cleanup;
5114 5115 5116
    }

    /* Check the path belongs to this domain. */
5117 5118
    for (i = 0 ; i < vm->def->ndisks ; i++) {
        if (vm->def->disks[i]->src != NULL &&
5119 5120 5121 5122
            STREQ (vm->def->disks[i]->src, path)) {
            ret = 0;
            break;
        }
5123 5124
    }

5125 5126 5127 5128 5129
    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) {
5130 5131
            virReportSystemError (dom->conn, errno,
                                  _("%s: failed to open"), path);
5132 5133
            goto cleanup;
        }
5134

5135 5136 5137 5138 5139 5140
        /* 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) {
5141 5142
            virReportSystemError (dom->conn, errno,
                                  _("%s: failed to seek or read"), path);
5143 5144 5145 5146 5147 5148 5149
            goto cleanup;
        }

        ret = 0;
    } else {
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
                          "%s", _("invalid path"));
5150 5151
    }

5152 5153 5154
cleanup:
    if (fd >= 0)
        close (fd);
5155 5156
    if (vm)
        virDomainObjUnlock(vm);
5157 5158 5159
    return ret;
}

R
Richard W.M. Jones 已提交
5160 5161 5162 5163 5164 5165
static int
qemudDomainMemoryPeek (virDomainPtr dom,
                       unsigned long long offset, size_t size,
                       void *buffer,
                       unsigned int flags)
{
5166 5167 5168
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
    char cmd[256], *info = NULL;
R
Richard W.M. Jones 已提交
5169 5170 5171
    char tmp[] = TEMPDIR "/qemu.mem.XXXXXX";
    int fd = -1, ret = -1;

5172
    qemuDriverLock(driver);
5173
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
5174
    qemuDriverUnlock(driver);
R
Richard W.M. Jones 已提交
5175 5176

    if (!vm) {
5177 5178 5179 5180
        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);
5181 5182 5183 5184 5185 5186 5187
        goto cleanup;
    }

    if (flags != VIR_MEMORY_VIRTUAL) {
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_INVALID_ARG,
                          "%s", _("QEMU driver only supports virtual memory addrs"));
        goto cleanup;
R
Richard W.M. Jones 已提交
5188 5189
    }

5190
    if (!virDomainIsActive(vm)) {
5191
        qemudReportError(dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
R
Richard W.M. Jones 已提交
5192
                         "%s", _("domain is not running"));
5193
        goto cleanup;
R
Richard W.M. Jones 已提交
5194 5195 5196 5197
    }

    /* Create a temporary filename. */
    if ((fd = mkstemp (tmp)) == -1) {
5198 5199
        virReportSystemError (dom->conn, errno,
                              _("mkstemp(\"%s\") failed"), tmp);
5200
        goto cleanup;
R
Richard W.M. Jones 已提交
5201 5202 5203 5204
    }

    /* Issue the memsave command. */
    snprintf (cmd, sizeof cmd, "memsave %llu %zi \"%s\"", offset, size, tmp);
5205
    if (qemudMonitorCommand (vm, cmd, &info) < 0) {
R
Richard W.M. Jones 已提交
5206
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
5207
                          "%s", _("'memsave' command failed"));
5208
        goto cleanup;
R
Richard W.M. Jones 已提交
5209 5210
    }

5211
    DEBUG ("%s: memsave reply: %s", vm->def->name, info);
R
Richard W.M. Jones 已提交
5212 5213 5214

    /* Read the memory file into buffer. */
    if (saferead (fd, buffer, size) == (ssize_t) -1) {
5215 5216 5217
        virReportSystemError (dom->conn, errno,
                              _("failed to read temporary file "
                                "created with template %s"), tmp);
5218
        goto cleanup;
R
Richard W.M. Jones 已提交
5219 5220 5221
    }

    ret = 0;
5222 5223 5224

cleanup:
    VIR_FREE(info);
R
Richard W.M. Jones 已提交
5225 5226
    if (fd >= 0) close (fd);
    unlink (tmp);
5227 5228
    if (vm)
        virDomainObjUnlock(vm);
R
Richard W.M. Jones 已提交
5229 5230 5231
    return ret;
}

5232

5233 5234
static int
qemudDomainEventRegister (virConnectPtr conn,
5235
                          virConnectDomainEventCallback callback,
5236 5237
                          void *opaque,
                          virFreeCallback freecb)
5238
{
5239 5240 5241
    struct qemud_driver *driver = conn->privateData;
    int ret;

5242
    qemuDriverLock(driver);
5243 5244
    ret = virDomainEventCallbackListAdd(conn, driver->domainEventCallbacks,
                                        callback, opaque, freecb);
5245
    qemuDriverUnlock(driver);
5246

5247
    return ret;
5248 5249 5250 5251
}

static int
qemudDomainEventDeregister (virConnectPtr conn,
5252
                            virConnectDomainEventCallback callback)
5253
{
5254 5255 5256
    struct qemud_driver *driver = conn->privateData;
    int ret;

5257
    qemuDriverLock(driver);
5258 5259 5260 5261 5262 5263
    if (driver->domainEventDispatching)
        ret = virDomainEventCallbackListMarkDelete(conn, driver->domainEventCallbacks,
                                                   callback);
    else
        ret = virDomainEventCallbackListRemove(conn, driver->domainEventCallbacks,
                                               callback);
5264
    qemuDriverUnlock(driver);
5265

5266
    return ret;
5267 5268
}

5269 5270 5271 5272 5273
static void qemuDomainEventDispatchFunc(virConnectPtr conn,
                                        virDomainEventPtr event,
                                        virConnectDomainEventCallback cb,
                                        void *cbopaque,
                                        void *opaque)
5274
{
5275
    struct qemud_driver *driver = opaque;
5276

5277 5278 5279 5280 5281 5282 5283 5284 5285 5286 5287 5288 5289 5290 5291 5292 5293 5294 5295 5296 5297 5298 5299 5300 5301 5302 5303 5304 5305 5306 5307 5308 5309 5310 5311 5312 5313 5314 5315 5316 5317 5318 5319 5320
    /* 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);
5321 5322
}

D
Daniel Veillard 已提交
5323 5324 5325 5326 5327 5328 5329 5330 5331 5332 5333 5334 5335 5336 5337 5338 5339 5340
/* 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;
5341 5342
    struct qemud_driver *driver = dconn->privateData;
    virDomainDefPtr def = NULL;
D
Daniel Veillard 已提交
5343 5344 5345 5346 5347
    virDomainObjPtr vm = NULL;
    int this_port;
    char hostname [HOST_NAME_MAX+1];
    char migrateFrom [64];
    const char *p;
5348
    virDomainEventPtr event = NULL;
5349 5350 5351
    int ret = -1;;

    *uri_out = NULL;
D
Daniel Veillard 已提交
5352

5353
    qemuDriverLock(driver);
D
Daniel Veillard 已提交
5354 5355 5356
    if (!dom_xml) {
        qemudReportError (dconn, NULL, NULL, VIR_ERR_INTERNAL_ERROR,
                          "%s", _("no domain XML passed"));
5357
        goto cleanup;
D
Daniel Veillard 已提交
5358 5359 5360 5361 5362 5363 5364 5365 5366 5367 5368 5369 5370 5371 5372 5373 5374 5375
    }

    /* 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 */
        if (gethostname (hostname, HOST_NAME_MAX+1) == -1) {
5376 5377
            virReportSystemError (dconn, errno,
                                  "%s", _("failed to determine host name"));
5378
            goto cleanup;
D
Daniel Veillard 已提交
5379 5380 5381
        }

        /* Caller frees */
5382
        if (virAsprintf(uri_out, "tcp:%s:%d", hostname, this_port) < 0) {
5383
            virReportOOMError (dconn);
5384
            goto cleanup;
D
Daniel Veillard 已提交
5385 5386 5387 5388 5389 5390
        }
    } 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.
         */
5391
        if (!STRPREFIX (uri_in, "tcp:")) {
D
Daniel Veillard 已提交
5392 5393
            qemudReportError (dconn, NULL, NULL, VIR_ERR_INVALID_ARG,
                  "%s", _("only tcp URIs are supported for KVM migrations"));
5394
            goto cleanup;
D
Daniel Veillard 已提交
5395 5396 5397 5398 5399 5400 5401 5402 5403
        }

        /* 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"));
5404
            goto cleanup;
D
Daniel Veillard 已提交
5405 5406 5407 5408
        }
    }

    /* Parse the domain XML. */
5409 5410
    if (!(def = virDomainDefParseString(dconn, driver->caps, dom_xml,
                                        VIR_DOMAIN_XML_INACTIVE))) {
D
Daniel Veillard 已提交
5411 5412
        qemudReportError (dconn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                          "%s", _("failed to parse XML"));
5413
        goto cleanup;
D
Daniel Veillard 已提交
5414 5415 5416 5417 5418 5419 5420 5421 5422 5423 5424 5425 5426 5427 5428 5429
    }

    /* 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"));
5430
        goto cleanup;
D
Daniel Veillard 已提交
5431 5432 5433 5434 5435 5436 5437 5438 5439
    }
#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);
5440
            goto cleanup;
D
Daniel Veillard 已提交
5441
        }
5442
        virDomainObjUnlock(vm);
D
Daniel Veillard 已提交
5443 5444 5445 5446 5447 5448 5449
    }

    if (!(vm = virDomainAssignDef(dconn,
                                  &driver->domains,
                                  def))) {
        qemudReportError (dconn, NULL, NULL, VIR_ERR_OPERATION_FAILED,
                          "%s", _("failed to assign new VM"));
5450
        goto cleanup;
D
Daniel Veillard 已提交
5451
    }
5452
    def = NULL;
D
Daniel Veillard 已提交
5453 5454 5455 5456 5457 5458 5459 5460

    /* 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);
5461
    if (qemudStartVMDaemon (dconn, driver, vm, migrateFrom, -1) < 0) {
5462
        if (!vm->persistent) {
D
Daniel Veillard 已提交
5463
            virDomainRemoveInactive(&driver->domains, vm);
5464 5465
            vm = NULL;
        }
5466
        goto cleanup;
D
Daniel Veillard 已提交
5467
    }
5468 5469 5470 5471

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STARTED,
                                     VIR_DOMAIN_EVENT_STARTED_MIGRATED);
5472
    ret = 0;
D
Daniel Veillard 已提交
5473

5474 5475 5476 5477 5478
cleanup:
    virDomainDefFree(def);
    if (ret != 0) {
        VIR_FREE(*uri_out);
    }
5479 5480
    if (vm)
        virDomainObjUnlock(vm);
5481 5482
    if (event)
        qemuDomainEventQueue(driver, event);
5483
    qemuDriverUnlock(driver);
5484
    return ret;
D
Daniel Veillard 已提交
5485 5486 5487 5488 5489 5490 5491 5492 5493 5494 5495 5496
}

/* 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,
                           unsigned long flags ATTRIBUTE_UNUSED,
                           const char *dname ATTRIBUTE_UNUSED,
                           unsigned long resource)
{
5497 5498
    struct qemud_driver *driver = dom->conn->privateData;
    virDomainObjPtr vm;
5499
    virDomainEventPtr event = NULL;
D
Daniel Veillard 已提交
5500 5501
    char *safe_uri;
    char cmd[HOST_NAME_MAX+50];
5502 5503
    char *info = NULL;
    int ret = -1;
5504
    int paused = 0;
D
Daniel Veillard 已提交
5505

5506
    qemuDriverLock(driver);
5507
    vm = virDomainFindByUUID(&driver->domains, dom->uuid);
D
Daniel Veillard 已提交
5508
    if (!vm) {
5509 5510 5511 5512
        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);
5513
        goto cleanup;
D
Daniel Veillard 已提交
5514 5515 5516
    }

    if (!virDomainIsActive(vm)) {
5517
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_OPERATION_INVALID,
D
Daniel Veillard 已提交
5518
                          "%s", _("domain is not running"));
5519
        goto cleanup;
D
Daniel Veillard 已提交
5520 5521
    }

5522 5523
    if (!(flags & VIR_MIGRATE_LIVE)) {
        /* Pause domain for non-live migration */
5524 5525 5526 5527 5528
        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;
        }
5529
        DEBUG ("%s: stop reply: %s", vm->def->name, info);
5530
        VIR_FREE(info);
5531
        paused = 1;
5532

5533 5534 5535 5536 5537 5538
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_SUSPENDED,
                                         VIR_DOMAIN_EVENT_SUSPENDED_MIGRATED);
        if (event)
            qemuDomainEventQueue(driver, event);
        event = NULL;
5539 5540
    }

D
Daniel Veillard 已提交
5541 5542 5543
    if (resource > 0) {
        /* Issue migrate_set_speed command.  Don't worry if it fails. */
        snprintf (cmd, sizeof cmd, "migrate_set_speed %lum", resource);
5544
        qemudMonitorCommand (vm, cmd, &info);
D
Daniel Veillard 已提交
5545

5546
        DEBUG ("%s: migrate_set_speed reply: %s", vm->def->name, info);
D
Daniel Veillard 已提交
5547 5548 5549 5550 5551 5552
        VIR_FREE (info);
    }

    /* Issue the migrate command. */
    safe_uri = qemudEscapeMonitorArg (uri);
    if (!safe_uri) {
5553
        virReportOOMError (dom->conn);
5554
        goto cleanup;
D
Daniel Veillard 已提交
5555 5556 5557 5558
    }
    snprintf (cmd, sizeof cmd, "migrate \"%s\"", safe_uri);
    VIR_FREE (safe_uri);

5559
    if (qemudMonitorCommand (vm, cmd, &info) < 0) {
D
Daniel Veillard 已提交
5560 5561
        qemudReportError (dom->conn, dom, NULL, VIR_ERR_OPERATION_FAILED,
                          "%s", _("migrate operation failed"));
5562
        goto cleanup;
D
Daniel Veillard 已提交
5563 5564
    }

5565
    DEBUG ("%s: migrate reply: %s", vm->def->name, info);
D
Daniel Veillard 已提交
5566 5567 5568 5569 5570

    /* 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);
5571
        goto cleanup;
D
Daniel Veillard 已提交
5572 5573 5574 5575
    }

    /* Clean up the source domain. */
    qemudShutdownVMDaemon (dom->conn, driver, vm);
5576
    paused = 0;
5577 5578 5579 5580

    event = virDomainEventNewFromObj(vm,
                                     VIR_DOMAIN_EVENT_STOPPED,
                                     VIR_DOMAIN_EVENT_STOPPED_MIGRATED);
5581
    if (!vm->persistent) {
D
Daniel Veillard 已提交
5582
        virDomainRemoveInactive(&driver->domains, vm);
5583 5584
        vm = NULL;
    }
5585
    ret = 0;
D
Daniel Veillard 已提交
5586

5587
cleanup:
5588 5589 5590
    /* Note that we have to free info *first*, since we are re-using the
     * variable below (and otherwise might cause a memory leak)
     */
5591
    VIR_FREE(info);
5592 5593 5594 5595 5596 5597 5598 5599

    if (paused) {
        /* we got here through some sort of failure; start the domain again */
        if (qemudMonitorCommand (vm, "cont", &info) < 0) {
            /* 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
             */
5600 5601
            VIR_ERROR(_("Failed to resume guest %s after failure\n"),
                      vm->def->name);
5602 5603
        }
        else {
5604
            DEBUG ("%s: cont reply: %s", vm->def->name, info);
5605 5606 5607 5608 5609 5610 5611 5612
            VIR_FREE(info);
        }

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

5613 5614
    if (vm)
        virDomainObjUnlock(vm);
5615 5616
    if (event)
        qemuDomainEventQueue(driver, event);
5617
    qemuDriverUnlock(driver);
5618
    return ret;
D
Daniel Veillard 已提交
5619 5620 5621 5622 5623 5624 5625 5626 5627 5628 5629 5630
}

/* 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)
{
5631 5632 5633
    struct qemud_driver *driver = dconn->privateData;
    virDomainObjPtr vm;
    virDomainPtr dom = NULL;
5634
    virDomainEventPtr event = NULL;
D
Daniel Veillard 已提交
5635 5636
    char *info = NULL;

5637
    qemuDriverLock(driver);
5638
    vm = virDomainFindByName(&driver->domains, dname);
D
Daniel Veillard 已提交
5639
    if (!vm) {
5640 5641
        qemudReportError (dconn, NULL, NULL, VIR_ERR_NO_DOMAIN,
                          _("no domain with matching name '%s'"), dname);
5642
        goto cleanup;
D
Daniel Veillard 已提交
5643 5644 5645 5646 5647 5648 5649 5650 5651
    }

    /* 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);
        VIR_FREE(info);
        vm->state = VIR_DOMAIN_RUNNING;
5652 5653 5654
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_RESUMED,
                                         VIR_DOMAIN_EVENT_RESUMED_MIGRATED);
D
Daniel Veillard 已提交
5655
        virDomainSaveStatus(dconn, driver->stateDir, vm);
D
Daniel Veillard 已提交
5656 5657
    } else {
        qemudShutdownVMDaemon (dconn, driver, vm);
5658 5659 5660
        event = virDomainEventNewFromObj(vm,
                                         VIR_DOMAIN_EVENT_STOPPED,
                                         VIR_DOMAIN_EVENT_STOPPED_FAILED);
5661
        if (!vm->persistent) {
D
Daniel Veillard 已提交
5662
            virDomainRemoveInactive(&driver->domains, vm);
5663 5664
            vm = NULL;
        }
D
Daniel Veillard 已提交
5665
    }
5666 5667

cleanup:
5668 5669
    if (vm)
        virDomainObjUnlock(vm);
5670 5671
    if (event)
        qemuDomainEventQueue(driver, event);
5672
    qemuDriverUnlock(driver);
5673
    return dom;
D
Daniel Veillard 已提交
5674 5675
}

5676 5677 5678 5679 5680 5681 5682 5683 5684 5685 5686 5687 5688 5689 5690 5691
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;

5692
    def = virNodeDeviceDefParseString(dev->conn, xml, EXISTING_DEVICE);
5693 5694 5695 5696 5697 5698 5699 5700 5701 5702 5703 5704 5705 5706 5707 5708 5709 5710 5711 5712 5713 5714 5715 5716 5717 5718 5719 5720 5721 5722 5723 5724 5725 5726 5727 5728 5729 5730 5731 5732 5733 5734 5735 5736 5737 5738 5739 5740 5741 5742 5743 5744 5745 5746 5747 5748 5749 5750 5751 5752 5753 5754 5755 5756 5757 5758 5759 5760 5761 5762 5763 5764 5765 5766 5767 5768 5769 5770 5771 5772 5773 5774 5775 5776 5777 5778 5779 5780 5781 5782 5783 5784 5785 5786 5787 5788 5789 5790
    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)
{
    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 (pciResetDevice(dev->conn, pci) < 0)
        goto out;

    ret = 0;
out:
    pciFreeDevice(dev->conn, pci);
    return ret;
}

5791 5792 5793 5794 5795
static virDriver qemuDriver = {
    VIR_DRV_QEMU,
    "QEMU",
    qemudOpen, /* open */
    qemudClose, /* close */
D
Daniel Veillard 已提交
5796
    qemudSupportsFeature, /* supports_feature */
5797 5798
    qemudGetType, /* type */
    qemudGetVersion, /* version */
5799
    qemudGetHostname, /* getHostname */
5800
    qemudGetMaxVCPUs, /* getMaxVcpus */
5801
    nodeGetInfo, /* nodeGetInfo */
5802 5803 5804
    qemudGetCapabilities, /* getCapabilities */
    qemudListDomains, /* listDomains */
    qemudNumDomains, /* numOfDomains */
5805
    qemudDomainCreate, /* domainCreateXML */
5806 5807 5808 5809 5810
    qemudDomainLookupByID, /* domainLookupByID */
    qemudDomainLookupByUUID, /* domainLookupByUUID */
    qemudDomainLookupByName, /* domainLookupByName */
    qemudDomainSuspend, /* domainSuspend */
    qemudDomainResume, /* domainResume */
5811
    qemudDomainShutdown, /* domainShutdown */
5812 5813 5814
    NULL, /* domainReboot */
    qemudDomainDestroy, /* domainDestroy */
    qemudDomainGetOSType, /* domainGetOSType */
5815 5816 5817
    qemudDomainGetMaxMemory, /* domainGetMaxMemory */
    qemudDomainSetMaxMemory, /* domainSetMaxMemory */
    qemudDomainSetMemory, /* domainSetMemory */
5818 5819 5820
    qemudDomainGetInfo, /* domainGetInfo */
    qemudDomainSave, /* domainSave */
    qemudDomainRestore, /* domainRestore */
P
Paolo Bonzini 已提交
5821
    qemudDomainCoreDump, /* domainCoreDump */
5822
    qemudDomainSetVcpus, /* domainSetVcpus */
5823 5824 5825 5826
#if HAVE_SCHED_GETAFFINITY
    qemudDomainPinVcpu, /* domainPinVcpu */
    qemudDomainGetVcpus, /* domainGetVcpus */
#else
5827 5828
    NULL, /* domainPinVcpu */
    NULL, /* domainGetVcpus */
5829
#endif
5830
    qemudDomainGetMaxVcpus, /* domainGetMaxVcpus */
5831 5832
    qemudDomainGetSecurityLabel, /* domainGetSecurityLabel */
    qemudNodeGetSecurityModel, /* nodeGetSecurityModel */
5833
    qemudDomainDumpXML, /* domainDumpXML */
5834
    qemuDomainXMLFromNative, /* domainXmlFromNative */
5835
    qemuDomainXMLToNative, /* domainXMLToNative */
5836 5837
    qemudListDefinedDomains, /* listDefinedDomains */
    qemudNumDefinedDomains, /* numOfDefinedDomains */
5838 5839 5840
    qemudDomainStart, /* domainCreate */
    qemudDomainDefine, /* domainDefineXML */
    qemudDomainUndefine, /* domainUndefine */
5841
    qemudDomainAttachDevice, /* domainAttachDevice */
5842
    qemudDomainDetachDevice, /* domainDetachDevice */
5843 5844 5845 5846 5847
    qemudDomainGetAutostart, /* domainGetAutostart */
    qemudDomainSetAutostart, /* domainSetAutostart */
    NULL, /* domainGetSchedulerType */
    NULL, /* domainGetSchedulerParameters */
    NULL, /* domainSetSchedulerParameters */
D
Daniel Veillard 已提交
5848 5849
    NULL, /* domainMigratePrepare (v1) */
    qemudDomainMigratePerform, /* domainMigratePerform */
5850
    NULL, /* domainMigrateFinish */
5851
    qemudDomainBlockStats, /* domainBlockStats */
5852
    qemudDomainInterfaceStats, /* domainInterfaceStats */
5853
    qemudDomainBlockPeek, /* domainBlockPeek */
R
Richard W.M. Jones 已提交
5854
    qemudDomainMemoryPeek, /* domainMemoryPeek */
5855 5856
    nodeGetCellsFreeMemory, /* nodeGetCellsFreeMemory */
    nodeGetFreeMemory,  /* getFreeMemory */
5857 5858
    qemudDomainEventRegister, /* domainEventRegister */
    qemudDomainEventDeregister, /* domainEventDeregister */
D
Daniel Veillard 已提交
5859 5860
    qemudDomainMigratePrepare2, /* domainMigratePrepare2 */
    qemudDomainMigrateFinish2, /* domainMigrateFinish2 */
5861 5862 5863
    qemudNodeDeviceDettach, /* nodeDeviceDettach */
    qemudNodeDeviceReAttach, /* nodeDeviceReAttach */
    qemudNodeDeviceReset, /* nodeDeviceReset */
5864 5865 5866
};


5867
static virStateDriver qemuStateDriver = {
5868 5869 5870 5871
    .initialize = qemudStartup,
    .cleanup = qemudShutdown,
    .reload = qemudReload,
    .active = qemudActive,
5872
};
5873

5874
int qemuRegister(void) {
5875 5876 5877 5878
    virRegisterDriver(&qemuDriver);
    virRegisterStateDriver(&qemuStateDriver);
    return 0;
}