vircgroup.c 94.3 KB
Newer Older
1
/*
2
 * vircgroup.c: methods for managing control cgroups
3
 *
4
 * Copyright (C) 2010-2015 Red Hat, Inc.
5 6
 * Copyright IBM Corp. 2008
 *
O
Osier Yang 已提交
7 8 9 10 11 12 13 14 15 16 17
 * 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
18
 * License along with this library.  If not, see
O
Osier Yang 已提交
19
 * <http://www.gnu.org/licenses/>.
20 21 22
 */
#include <config.h>

23
#ifdef __linux__
24
# include <mntent.h>
25
# include <sys/mount.h>
26 27
# include <fcntl.h>
# include <sys/stat.h>
28

29 30 31 32 33
# ifdef MAJOR_IN_MKDEV
#  include <sys/mkdev.h>
# elif MAJOR_IN_SYSMACROS
#  include <sys/sysmacros.h>
# endif
34

35 36 37 38 39
# include <sys/types.h>
# include <signal.h>
# include <dirent.h>
# include <unistd.h>
#endif /* __linux__ */
40

41
#define LIBVIRT_VIRCGROUPPRIV_H_ALLOW
42 43
#include "vircgrouppriv.h"

44
#include "virutil.h"
45
#include "viralloc.h"
46
#include "vircgroupbackend.h"
47
#include "virerror.h"
48
#include "virlog.h"
E
Eric Blake 已提交
49
#include "virfile.h"
50
#include "virhash.h"
51
#include "virhashcode.h"
52
#include "virstring.h"
53
#include "virsystemd.h"
54
#include "virtypedparam.h"
55
#include "virhostcpu.h"
56
#include "virthread.h"
57

58 59
VIR_LOG_INIT("util.cgroup");

60 61
#define VIR_FROM_THIS VIR_FROM_CGROUP

62
#define CGROUP_NB_TOTAL_CPU_STAT_PARAM 3
63
#define CGROUP_NB_PER_CPU_STAT_PARAM   1
64

65 66
VIR_ENUM_IMPL(virCgroupController,
              VIR_CGROUP_CONTROLLER_LAST,
R
Ryota Ozaki 已提交
67
              "cpu", "cpuacct", "cpuset", "memory", "devices",
68
              "freezer", "blkio", "net_cls", "perf_event",
69 70
              "name=systemd",
);
71

E
Eric Blake 已提交
72

73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110
/**
 * virCgroupGetDevicePermsString:
 *
 * @perms: Bitwise or of VIR_CGROUP_DEVICE permission bits
 *
 * Returns string corresponding to the appropriate bits set.
 */
const char *
virCgroupGetDevicePermsString(int perms)
{
    if (perms & VIR_CGROUP_DEVICE_READ) {
        if (perms & VIR_CGROUP_DEVICE_WRITE) {
            if (perms & VIR_CGROUP_DEVICE_MKNOD)
                return "rwm";
            else
                return "rw";
        } else {
            if (perms & VIR_CGROUP_DEVICE_MKNOD)
                return "rm";
            else
                return "r";
        }
    } else {
        if (perms & VIR_CGROUP_DEVICE_WRITE) {
            if (perms & VIR_CGROUP_DEVICE_MKNOD)
                return "wm";
            else
                return "w";
        } else {
            if (perms & VIR_CGROUP_DEVICE_MKNOD)
                return "m";
            else
                return "";
        }
    }
}


111
#ifdef __linux__
E
Eric Blake 已提交
112 113
bool
virCgroupAvailable(void)
114
{
115 116
    size_t i;
    virCgroupBackendPtr *backends = virCgroupBackendGetAll();
117

118
    if (!backends)
119 120
        return false;

121 122 123
    for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) {
        if (backends[i] && backends[i]->available())
            return true;
124 125
    }

126
    return false;
127 128
}

E
Eric Blake 已提交
129 130 131 132 133 134

static int
virCgroupPartitionNeedsEscaping(const char *path)
{
    FILE *fp = NULL;
    int ret = 0;
135
    VIR_AUTOFREE(char *) line = NULL;
E
Eric Blake 已提交
136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193
    size_t buflen;

    /* If it starts with 'cgroup.' or a '_' of any
     * of the controller names from /proc/cgroups,
     * then we must prefix a '_'
     */
    if (STRPREFIX(path, "cgroup."))
        return 1;

    if (path[0] == '_' ||
        path[0] == '.')
        return 1;

    if (!(fp = fopen("/proc/cgroups", "r"))) {
        /* The API contract is that we return ENXIO
         * if cgroups are not available on a host */
        if (errno == ENOENT)
            errno = ENXIO;
        virReportSystemError(errno, "%s",
                             _("Cannot open /proc/cgroups"));
        return -1;
    }

    /*
     * Data looks like this:
     * #subsys_name hierarchy num_cgroups enabled
     * cpuset  2 4  1
     * cpu     3 48 1
     * cpuacct 3 48 1
     * memory  4 4  1
     * devices 5 4  1
     * freezer 6 4  1
     * net_cls 7 1  1
     */
    while (getline(&line, &buflen, fp) > 0) {
        char *tmp;
        size_t len;

        if (STRPREFIX(line, "#subsys_name"))
            continue;

        tmp = strchrnul(line, ' ');
        *tmp = '\0';
        len = tmp - line;

        if (STRPREFIX(path, line) &&
            path[len] == '.') {
            ret = 1;
            goto cleanup;
        }
    }

    if (ferror(fp)) {
        virReportSystemError(errno, "%s",
                             _("Error while reading /proc/cgroups"));
        goto cleanup;
    }

194
 cleanup:
E
Eric Blake 已提交
195 196 197 198 199
    VIR_FORCE_FCLOSE(fp);
    return ret;
}


200
int
E
Eric Blake 已提交
201 202 203
virCgroupPartitionEscape(char **path)
{
    int rc;
204
    char *newstr = NULL;
E
Eric Blake 已提交
205 206 207 208

    if ((rc = virCgroupPartitionNeedsEscaping(*path)) <= 0)
        return rc;

209
    if (virAsprintf(&newstr, "_%s", *path) < 0)
E
Eric Blake 已提交
210 211
        return -1;

212 213 214
    VIR_FREE(*path);
    *path = newstr;

E
Eric Blake 已提交
215 216
    return 0;
}
E
Eric Blake 已提交
217 218


219 220 221 222
/*
 * Process /proc/mounts figuring out what controllers are
 * mounted and where
 */
223
static int
224
virCgroupDetectMounts(virCgroupPtr group)
225
{
226
    FILE *mounts = NULL;
227 228
    struct mntent entry;
    char buf[CGROUP_MAX_VAL];
229
    int ret = -1;
230
    size_t i;
231

232
    mounts = fopen("/proc/mounts", "r");
233
    if (mounts == NULL) {
234
        virReportSystemError(errno, "%s", _("Unable to open /proc/mounts"));
235
        return -1;
236 237 238
    }

    while (getmntent_r(mounts, &entry, buf, sizeof(buf)) != NULL) {
239 240 241 242 243 244 245 246
        for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) {
            if (group->backends[i] &&
                group->backends[i]->detectMounts(group,
                                                 entry.mnt_type,
                                                 entry.mnt_opts,
                                                 entry.mnt_dir) < 0) {
                goto cleanup;
            }
247
        }
248 249
    }

250 251
    ret = 0;
 cleanup:
252
    VIR_FORCE_FCLOSE(mounts);
253
    return ret;
254 255
}

256 257

/*
258 259 260 261
 * virCgroupDetectPlacement:
 * @group: the group to process
 * @path: the relative path to append, not starting with '/'
 *
262 263
 * Process /proc/self/cgroup figuring out what cgroup
 * sub-path the current process is assigned to. ie not
264 265 266 267 268 269 270 271 272 273 274 275 276 277
 * necessarily in the root. The contents of this file
 * looks like
 *
 * 9:perf_event:/
 * 8:blkio:/
 * 7:net_cls:/
 * 6:freezer:/
 * 5:devices:/
 * 4:memory:/
 * 3:cpuacct,cpu:/
 * 2:cpuset:/
 * 1:name=systemd:/user/berrange/2
 *
 * It then appends @path to each detected path.
278
 */
E
Eric Blake 已提交
279 280 281 282
static int
virCgroupDetectPlacement(virCgroupPtr group,
                         pid_t pid,
                         const char *path)
283
{
284 285
    FILE *mapping  = NULL;
    char line[1024];
286
    int ret = -1;
287
    VIR_AUTOFREE(char *) procfile = NULL;
288

289
    VIR_DEBUG("Detecting placement for pid %lld path %s",
M
Michal Privoznik 已提交
290
              (long long) pid, path);
291 292 293 294
    if (pid == -1) {
        if (VIR_STRDUP(procfile, "/proc/self/cgroup") < 0)
            goto cleanup;
    } else {
M
Michal Privoznik 已提交
295 296
        if (virAsprintf(&procfile, "/proc/%lld/cgroup",
                        (long long) pid) < 0)
297 298 299 300
            goto cleanup;
    }

    mapping = fopen(procfile, "r");
301
    if (mapping == NULL) {
302 303 304 305
        virReportSystemError(errno,
                             _("Unable to open '%s'"),
                             procfile);
        goto cleanup;
306 307
    }

308
    while (fgets(line, sizeof(line), mapping) != NULL) {
309
        size_t i;
310
        char *controllers = strchr(line, ':');
311 312
        char *selfpath = controllers ? strchr(controllers + 1, ':') : NULL;
        char *nl = selfpath ? strchr(selfpath, '\n') : NULL;
313

314
        if (!controllers || !selfpath)
315 316 317 318 319
            continue;

        if (nl)
            *nl = '\0';

320
        *selfpath = '\0';
321
        controllers++;
322
        selfpath++;
323

324 325 326 327 328 329
        for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) {
            if (group->backends[i] &&
                group->backends[i]->detectPlacement(group, path, controllers,
                                                    selfpath) < 0) {
                goto cleanup;
            }
330 331 332
        }
    }

333
    ret = 0;
334

335
 cleanup:
336
    VIR_FORCE_FCLOSE(mapping);
337
    return ret;
338 339
}

E
Eric Blake 已提交
340

341 342 343 344 345 346 347
static int
virCgroupDetect(virCgroupPtr group,
                pid_t pid,
                int controllers,
                const char *path,
                virCgroupPtr parent)
{
348
    size_t i;
349 350
    bool backendAvailable = false;
    int controllersAvailable = 0;
351
    virCgroupBackendPtr *backends = virCgroupBackendGetAll();
352 353 354 355

    VIR_DEBUG("group=%p controllers=%d path=%s parent=%p",
              group, controllers, path, parent);

356 357 358 359 360
    if (!backends)
        return -1;

    for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) {
        if (backends[i] && backends[i]->available()) {
361 362
            group->backends[i] = backends[i];
            backendAvailable = true;
363 364 365
        }
    }

366
    if (!backendAvailable) {
367 368 369 370 371
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("no cgroup backend available"));
        return -1;
    }

372
    if (parent) {
373 374 375 376 377 378
        for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) {
            if (group->backends[i] &&
                group->backends[i]->copyMounts(group, parent) < 0) {
                return -1;
            }
        }
379 380 381 382 383
    } else {
        if (virCgroupDetectMounts(group) < 0)
            return -1;
    }

384 385 386 387 388 389 390 391
    for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) {
        if (group->backends[i]) {
            int rc = group->backends[i]->detectControllers(group, controllers);
            if (rc < 0)
                return -1;
            controllersAvailable |= rc;
        }
    }
392

393
    /* Check that at least 1 controller is available */
394
    if (controllersAvailable == 0) {
395 396 397
        virReportSystemError(ENXIO, "%s",
                             _("At least one cgroup controller is required"));
        return -1;
398
    }
399

400 401 402
    /* In some cases we can copy part of the placement info
     * based on the parent cgroup...
     */
403 404 405 406 407 408 409 410
    if (parent || path[0] == '/') {
        for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) {
            if (group->backends[i] &&
                group->backends[i]->copyPlacement(group, path, parent) < 0) {
                return -1;
            }
        }
    }
411 412 413 414

    /* ... but use /proc/cgroups to fill in the rest */
    if (virCgroupDetectPlacement(group, pid, path) < 0)
        return -1;
415

416
    /* Check that for every mounted controller, we found our placement */
417 418 419 420 421 422
    for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) {
        if (group->backends[i] &&
            group->backends[i]->validatePlacement(group, pid) < 0) {
            return -1;
        }
    }
423

424
    return 0;
425 426
}

427

428
char *
429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456
virCgroupGetBlockDevString(const char *path)
{
    char *ret = NULL;
    struct stat sb;

    if (stat(path, &sb) < 0) {
        virReportSystemError(errno,
                             _("Path '%s' is not accessible"),
                             path);
        return NULL;
    }

    if (!S_ISBLK(sb.st_mode)) {
        virReportSystemError(EINVAL,
                             _("Path '%s' must be a block device"),
                             path);
        return NULL;
    }

    /* Automatically append space after the string since all callers
     * use it anyway */
    if (virAsprintf(&ret, "%d:%d ", major(sb.st_rdev), minor(sb.st_rdev)) < 0)
        return NULL;

    return ret;
}


457
int
E
Eric Blake 已提交
458 459 460 461
virCgroupSetValueStr(virCgroupPtr group,
                     int controller,
                     const char *key,
                     const char *value)
462
{
463
    VIR_AUTOFREE(char *) keypath = NULL;
464
    char *tmp = NULL;
465

466 467
    if (virCgroupPathOfController(group, controller, key, &keypath) < 0)
        return -1;
468

469
    VIR_DEBUG("Set value '%s' to '%s'", keypath, value);
470
    if (virFileWriteStr(keypath, value, 0) < 0) {
471 472 473 474 475
        if (errno == EINVAL &&
            (tmp = strrchr(keypath, '/'))) {
            virReportSystemError(errno,
                                 _("Invalid value '%s' for '%s'"),
                                 value, tmp + 1);
476
            return -1;
477
        }
478 479
        virReportSystemError(errno,
                             _("Unable to write to '%s'"), keypath);
480
        return -1;
481 482
    }

483
    return 0;
484 485
}

E
Eric Blake 已提交
486

487
int
E
Eric Blake 已提交
488 489 490 491
virCgroupGetValueStr(virCgroupPtr group,
                     int controller,
                     const char *key,
                     char **value)
492
{
493 494
    VIR_AUTOFREE(char *) keypath = NULL;
    int rc;
495

496
    *value = NULL;
497

498 499
    if (virCgroupPathOfController(group, controller, key, &keypath) < 0)
        return -1;
500

501
    VIR_DEBUG("Get value %s", keypath);
502

503 504 505
    if ((rc = virFileReadAll(keypath, 1024*1024, value)) < 0) {
        virReportSystemError(errno,
                             _("Unable to read from '%s'"), keypath);
506
        return -1;
507 508
    }

509 510 511
    /* Terminated with '\n' has sometimes harmful effects to the caller */
    if (rc > 0 && (*value)[rc - 1] == '\n')
        (*value)[rc - 1] = '\0';
512

513
    return 0;
514 515
}

E
Eric Blake 已提交
516

517
int
518 519 520 521 522 523
virCgroupGetValueForBlkDev(virCgroupPtr group,
                           int controller,
                           const char *key,
                           const char *path,
                           char **value)
{
524 525
    VIR_AUTOFREE(char *) prefix = NULL;
    VIR_AUTOFREE(char *) str = NULL;
526 527
    char **lines = NULL;
    int ret = -1;
528 529

    if (virCgroupGetValueStr(group, controller, key, &str) < 0)
530
        goto error;
531 532

    if (!(prefix = virCgroupGetBlockDevString(path)))
533
        goto error;
534 535

    if (!(lines = virStringSplit(str, "\n", -1)))
536
        goto error;
537

538
    if (VIR_STRDUP(*value, virStringListGetFirstWithPrefix(lines, prefix)) < 0)
539
        goto error;
540

541 542 543 544
    ret = 0;
 error:
    virStringListFree(lines);
    return ret;
545 546 547
}


548
int
E
Eric Blake 已提交
549 550 551 552
virCgroupSetValueU64(virCgroupPtr group,
                     int controller,
                     const char *key,
                     unsigned long long int value)
553
{
554
    VIR_AUTOFREE(char *) strval = NULL;
555

556 557
    if (virAsprintf(&strval, "%llu", value) < 0)
        return -1;
558

559
    return virCgroupSetValueStr(group, controller, key, strval);
560 561 562
}


563
int
E
Eric Blake 已提交
564 565 566 567
virCgroupSetValueI64(virCgroupPtr group,
                     int controller,
                     const char *key,
                     long long int value)
568
{
569
    VIR_AUTOFREE(char *) strval = NULL;
570

571 572
    if (virAsprintf(&strval, "%lld", value) < 0)
        return -1;
573

574
    return virCgroupSetValueStr(group, controller, key, strval);
575 576
}

E
Eric Blake 已提交
577

578
int
E
Eric Blake 已提交
579 580 581 582
virCgroupGetValueI64(virCgroupPtr group,
                     int controller,
                     const char *key,
                     long long int *value)
583
{
584
    VIR_AUTOFREE(char *) strval = NULL;
585

586
    if (virCgroupGetValueStr(group, controller, key, &strval) < 0)
587
        return -1;
588

589 590 591 592
    if (virStrToLong_ll(strval, NULL, 10, value) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Unable to parse '%s' as an integer"),
                       strval);
593
        return -1;
594
    }
595

596
    return 0;
597 598
}

E
Eric Blake 已提交
599

600
int
E
Eric Blake 已提交
601 602 603 604
virCgroupGetValueU64(virCgroupPtr group,
                     int controller,
                     const char *key,
                     unsigned long long int *value)
605
{
606
    VIR_AUTOFREE(char *) strval = NULL;
607

608
    if (virCgroupGetValueStr(group, controller, key, &strval) < 0)
609
        return -1;
610

611 612 613 614
    if (virStrToLong_ull(strval, NULL, 10, value) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Unable to parse '%s' as an integer"),
                       strval);
615
        return -1;
616
    }
617

618
    return 0;
619 620 621
}


E
Eric Blake 已提交
622 623 624 625 626
static int
virCgroupMakeGroup(virCgroupPtr parent,
                   virCgroupPtr group,
                   bool create,
                   unsigned int flags)
627
{
628 629 630 631 632 633 634 635
    size_t i;

    for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) {
        if (group->backends[i] &&
            group->backends[i]->makeGroup(parent, group, create, flags) < 0) {
            virCgroupRemove(group);
            return -1;
        }
636 637
    }

638
    return 0;
639 640
}

641

642 643 644 645 646 647 648 649 650 651 652 653 654 655
/**
 * virCgroupNew:
 * @path: path for the new group
 * @parent: parent group, or NULL
 * @controllers: bitmask of controllers to activate
 *
 * Create a new cgroup storing it in @group.
 *
 * If @path starts with a '/' it is treated as an
 * absolute path, and @parent is ignored. Otherwise
 * it is treated as being relative to @parent. If
 * @parent is NULL, then the placement of the current
 * process is used.
 *
656
 * Returns 0 on success, -1 on error
657
 */
658
int
E
Eric Blake 已提交
659 660 661 662 663
virCgroupNew(pid_t pid,
             const char *path,
             virCgroupPtr parent,
             int controllers,
             virCgroupPtr *group)
664
{
665 666
    VIR_DEBUG("pid=%lld path=%s parent=%p controllers=%d group=%p",
              (long long) pid, path, parent, controllers, group);
667
    *group = NULL;
668

669 670
    if (VIR_ALLOC((*group)) < 0)
        goto error;
671

672
    if (path[0] == '/' || !parent) {
673 674
        if (VIR_STRDUP((*group)->path, path) < 0)
            goto error;
675 676 677 678
    } else {
        if (virAsprintf(&(*group)->path, "%s%s%s",
                        parent->path,
                        STREQ(parent->path, "") ? "" : "/",
679 680
                        path) < 0)
            goto error;
681 682
    }

683
    if (virCgroupDetect(*group, pid, controllers, path, parent) < 0)
684
        goto error;
685

686 687
    return 0;

688
 error:
689
    virCgroupFree(group);
690
    *group = NULL;
691

692
    return -1;
693
}
694

695

696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713
static int
virCgroupAddTaskInternal(virCgroupPtr group,
                         pid_t pid,
                         unsigned int flags)
{
    size_t i;

    for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) {
        if (group->backends[i] &&
            group->backends[i]->addTask(group, pid, flags) < 0) {
            return -1;
        }
    }

    return 0;
}


714
/**
715
 * virCgroupAddProcess:
716
 *
717 718
 * @group: The cgroup to add a process to
 * @pid: The pid of the process to add
719
 *
720
 * Will add the process to all controllers, except the
721 722 723 724 725
 * systemd unit controller.
 *
 * Returns: 0 on success, -1 on error
 */
int
726
virCgroupAddProcess(virCgroupPtr group, pid_t pid)
727
{
728
    return virCgroupAddTaskInternal(group, pid, VIR_CGROUP_TASK_PROCESS);
729 730 731
}

/**
732
 * virCgroupAddMachineProcess:
733
 *
734 735
 * @group: The cgroup to add a process to
 * @pid: The pid of the process to add
736
 *
737
 * Will add the process to all controllers, including the
738 739 740 741 742
 * systemd unit controller.
 *
 * Returns: 0 on success, -1 on error
 */
int
743
virCgroupAddMachineProcess(virCgroupPtr group, pid_t pid)
744
{
745 746 747
    return virCgroupAddTaskInternal(group, pid,
                                    VIR_CGROUP_TASK_PROCESS |
                                    VIR_CGROUP_TASK_SYSTEMD);
748 749
}

750 751 752 753 754 755 756 757 758 759 760 761 762 763 764
/**
 * virCgroupAddThread:
 *
 * @group: The cgroup to add a thread to
 * @pid: The pid of the thread to add
 *
 * Will add the thread to all controllers, except the
 * systemd unit controller.
 *
 * Returns: 0 on success, -1 on error
 */
int
virCgroupAddThread(virCgroupPtr group,
                   pid_t pid)
{
765
    return virCgroupAddTaskInternal(group, pid, VIR_CGROUP_TASK_THREAD);
766 767
}

E
Eric Blake 已提交
768 769 770

static int
virCgroupSetPartitionSuffix(const char *path, char **res)
771
{
772
    char **tokens;
773
    size_t i;
774
    int ret = -1;
775

776
    if (!(tokens = virStringSplit(path, "/", 0)))
777
        return ret;
778

779
    for (i = 0; tokens[i] != NULL; i++) {
780 781 782 783 784 785 786 787 788 789 790 791 792 793 794
        /* Whitelist the 3 top level fixed dirs
         * NB i == 0 is "", since we have leading '/'
         */
        if (i == 1 &&
            (STREQ(tokens[i], "machine") ||
             STREQ(tokens[i], "system") ||
             STREQ(tokens[i], "user"))) {
            continue;
        }
        /* If there is no suffix set already, then
         * add ".partition"
         */
        if (STRNEQ(tokens[i], "") &&
            !strchr(tokens[i], '.')) {
            if (VIR_REALLOC_N(tokens[i],
795
                              strlen(tokens[i]) + strlen(".partition") + 1) < 0)
796
                goto cleanup;
797 798
            strcat(tokens[i], ".partition");
        }
799

800
        if (virCgroupPartitionEscape(&(tokens[i])) < 0)
801
            goto cleanup;
802 803
    }

804
    if (!(*res = virStringListJoin((const char **)tokens, "/")))
805
        goto cleanup;
806

807 808 809 810 811
    ret = 0;

 cleanup:
    virStringListFree(tokens);
    return ret;
812 813
}

E
Eric Blake 已提交
814

815 816 817 818 819 820 821
/**
 * virCgroupNewPartition:
 * @path: path for the partition
 * @create: true to create the cgroup tree
 * @controllers: mask of controllers to create
 *
 * Creates a new cgroup to represent the resource
822
 * partition path identified by @path.
823
 *
824
 * Returns 0 on success, -1 on failure
825
 */
E
Eric Blake 已提交
826 827 828 829 830
int
virCgroupNewPartition(const char *path,
                      bool create,
                      int controllers,
                      virCgroupPtr *group)
831
{
832
    int ret = -1;
833 834
    VIR_AUTOFREE(char *) parentPath = NULL;
    VIR_AUTOFREE(char *) newPath = NULL;
835
    virCgroupPtr parent = NULL;
836 837 838
    VIR_DEBUG("path=%s create=%d controllers=%x",
              path, create, controllers);

839 840 841 842 843 844
    if (path[0] != '/') {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Partition path '%s' must start with '/'"),
                       path);
        return -1;
    }
845

846
    if (virCgroupSetPartitionSuffix(path, &newPath) < 0)
847
        goto cleanup;
848

849 850
    if (virCgroupNew(-1, newPath, NULL, controllers, group) < 0)
        goto cleanup;
851

852
    if (STRNEQ(newPath, "/")) {
853
        char *tmp;
854
        if (VIR_STRDUP(parentPath, newPath) < 0)
855
            goto cleanup;
856 857 858 859 860

        tmp = strrchr(parentPath, '/');
        tmp++;
        *tmp = '\0';

861
        if (virCgroupNew(-1, parentPath, NULL, controllers, &parent) < 0)
862
            goto cleanup;
863

864
        if (virCgroupMakeGroup(parent, *group, create, VIR_CGROUP_NONE) < 0)
865
            goto cleanup;
866 867
    }

868 869 870
    ret = 0;
 cleanup:
    if (ret != 0)
871 872
        virCgroupFree(group);
    virCgroupFree(&parent);
873
    return ret;
874 875
}

876

G
Gao feng 已提交
877
/**
878
* virCgroupNewSelf:
G
Gao feng 已提交
879 880 881
*
* @group: Pointer to returned virCgroupPtr
*
882 883 884
* Obtain a cgroup representing the config of the
* current process
*
885
* Returns 0 on success, or -1 on error
G
Gao feng 已提交
886
*/
E
Eric Blake 已提交
887 888
int
virCgroupNewSelf(virCgroupPtr *group)
G
Gao feng 已提交
889
{
890
    return virCgroupNewDetect(-1, -1, group);
G
Gao feng 已提交
891
}
892

893

894 895 896 897 898 899 900 901
/**
 * virCgroupNewDomainPartition:
 *
 * @partition: partition holding the domain
 * @driver: name of the driver
 * @name: name of the domain
 * @group: Pointer to returned virCgroupPtr
 *
902
 * Returns 0 on success, or -1 on error
903
 */
E
Eric Blake 已提交
904 905 906 907 908 909
int
virCgroupNewDomainPartition(virCgroupPtr partition,
                            const char *driver,
                            const char *name,
                            bool create,
                            virCgroupPtr *group)
910
{
911
    VIR_AUTOFREE(char *)grpname = NULL;
912

913
    if (virAsprintf(&grpname, "%s.libvirt-%s",
914
                    name, driver) < 0)
915
        return -1;
916

917
    if (virCgroupPartitionEscape(&grpname) < 0)
918
        return -1;
919

920
    if (virCgroupNew(-1, grpname, partition, -1, group) < 0)
921
        return -1;
922 923 924 925 926 927 928 929 930 931 932

    /*
     * Create a cgroup with memory.use_hierarchy enabled to
     * surely account memory usage of lxc with ns subsystem
     * enabled. (To be exact, memory and ns subsystems are
     * enabled at the same time.)
     *
     * The reason why doing it here, not a upper group, say
     * a group for driver, is to avoid overhead to track
     * cumulative usage that we don't need.
     */
E
Eric Blake 已提交
933 934
    if (virCgroupMakeGroup(partition, *group, create,
                           VIR_CGROUP_MEM_HIERACHY) < 0) {
935
        virCgroupFree(group);
936
        return -1;
937 938
    }

939
    return 0;
940
}
941

E
Eric Blake 已提交
942

943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960
/**
 * virCgroupNewThread:
 *
 * @domain: group for the domain
 * @name: enum to generate the name for the new thread
 * @id: id of the vcpu or iothread
 * @create: true to create if not already existing
 * @group: Pointer to returned virCgroupPtr
 *
 * Returns 0 on success, or -1 on error
 */
int
virCgroupNewThread(virCgroupPtr domain,
                   virCgroupThreadName nameval,
                   int id,
                   bool create,
                   virCgroupPtr *group)
{
961
    VIR_AUTOFREE(char *) name = NULL;
962 963 964 965 966
    int controllers;

    switch (nameval) {
    case VIR_CGROUP_THREAD_VCPU:
        if (virAsprintf(&name, "vcpu%d", id) < 0)
967
            return -1;
968 969 970
        break;
    case VIR_CGROUP_THREAD_EMULATOR:
        if (VIR_STRDUP(name, "emulator") < 0)
971
            return -1;
972 973 974
        break;
    case VIR_CGROUP_THREAD_IOTHREAD:
        if (virAsprintf(&name, "iothread%d", id) < 0)
975
            return -1;
976 977 978 979
        break;
    case VIR_CGROUP_THREAD_LAST:
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("unexpected name value %d"), nameval);
980
        return -1;
981 982 983 984 985 986 987
    }

    controllers = ((1 << VIR_CGROUP_CONTROLLER_CPU) |
                   (1 << VIR_CGROUP_CONTROLLER_CPUACCT) |
                   (1 << VIR_CGROUP_CONTROLLER_CPUSET));

    if (virCgroupNew(-1, name, domain, controllers, group) < 0)
988
        return -1;
989

990
    if (virCgroupMakeGroup(domain, *group, create, VIR_CGROUP_THREAD) < 0) {
991
        virCgroupFree(group);
992
        return -1;
993 994
    }

995
    return 0;
996 997 998
}


E
Eric Blake 已提交
999 1000 1001 1002
int
virCgroupNewDetect(pid_t pid,
                   int controllers,
                   virCgroupPtr *group)
1003
{
1004
    return virCgroupNew(pid, "", NULL, controllers, group);
1005 1006
}

E
Eric Blake 已提交
1007

1008 1009 1010
/*
 * Returns 0 on success (but @group may be NULL), -1 on fatal error
 */
E
Eric Blake 已提交
1011 1012 1013 1014 1015
int
virCgroupNewDetectMachine(const char *name,
                          const char *drivername,
                          pid_t pid,
                          int controllers,
1016
                          char *machinename,
E
Eric Blake 已提交
1017
                          virCgroupPtr *group)
1018
{
1019 1020
    size_t i;

1021
    if (virCgroupNewDetect(pid, controllers, group) < 0) {
1022 1023 1024 1025 1026
        if (virCgroupNewIgnoreError())
            return 0;
        return -1;
    }

1027 1028 1029 1030 1031 1032 1033 1034 1035 1036
    for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) {
        if ((*group)->backends[i] &&
            !(*group)->backends[i]->validateMachineGroup(*group, name,
                                                         drivername,
                                                         machinename)) {
            VIR_DEBUG("Failed to validate machine name for '%s' driver '%s'",
                      name, drivername);
            virCgroupFree(group);
            return 0;
        }
1037 1038 1039 1040 1041
    }

    return 0;
}

E
Eric Blake 已提交
1042

1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053
static int
virCgroupEnableMissingControllers(char *path,
                                  pid_t pidleader,
                                  int controllers,
                                  virCgroupPtr *group)
{
    virCgroupPtr parent = NULL;
    char *offset = path;
    int ret = -1;

    if (virCgroupNew(pidleader,
1054
                     "/",
1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094
                     NULL,
                     controllers,
                     &parent) < 0)
        return ret;

    for (;;) {
        virCgroupPtr tmp;
        char *t = strchr(offset + 1, '/');
        if (t)
            *t = '\0';

        if (virCgroupNew(pidleader,
                         path,
                         parent,
                         controllers,
                         &tmp) < 0)
            goto cleanup;

        if (virCgroupMakeGroup(parent, tmp, true, VIR_CGROUP_NONE) < 0) {
            virCgroupFree(&tmp);
            goto cleanup;
        }
        if (t) {
            *t = '/';
            offset = t;
            virCgroupFree(&parent);
            parent = tmp;
        } else {
            *group = tmp;
            break;
        }
    }

    ret = 0;
 cleanup:
    virCgroupFree(&parent);
    return ret;
}


1095 1096 1097 1098 1099 1100 1101 1102 1103 1104
/*
 * Returns 0 on success, -1 on fatal error, -2 on systemd not available
 */
static int
virCgroupNewMachineSystemd(const char *name,
                           const char *drivername,
                           const unsigned char *uuid,
                           const char *rootdir,
                           pid_t pidleader,
                           bool isContainer,
1105 1106
                           size_t nnicindexes,
                           int *nicindexes,
1107 1108 1109
                           const char *partition,
                           int controllers,
                           virCgroupPtr *group)
1110
{
1111
    int rv;
1112
    virCgroupPtr init;
1113
    VIR_AUTOFREE(char *) path = NULL;
1114
    size_t i;
1115 1116 1117 1118 1119 1120 1121 1122

    VIR_DEBUG("Trying to setup machine '%s' via systemd", name);
    if ((rv = virSystemdCreateMachine(name,
                                      drivername,
                                      uuid,
                                      rootdir,
                                      pidleader,
                                      isContainer,
1123 1124
                                      nnicindexes,
                                      nicindexes,
1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135
                                      partition)) < 0)
        return rv;

    if (controllers != -1)
        controllers |= (1 << VIR_CGROUP_CONTROLLER_SYSTEMD);

    VIR_DEBUG("Detecting systemd placement");
    if (virCgroupNewDetect(pidleader,
                           controllers,
                           &init) < 0)
        return -1;
1136

1137 1138 1139 1140 1141 1142
    for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) {
        if (init->backends[i] &&
            (path = init->backends[i]->stealPlacement(init))) {
            break;
        }
    }
1143
    virCgroupFree(&init);
1144 1145 1146

    if (!path || STREQ(path, "/") || path[0] != '/') {
        VIR_DEBUG("Systemd didn't setup its controller");
1147
        return -2;
1148 1149
    }

1150 1151
    if (virCgroupEnableMissingControllers(path, pidleader,
                                          controllers, group) < 0) {
1152
        return -1;
1153 1154
    }

1155 1156 1157 1158 1159 1160 1161 1162
    if (virCgroupAddProcess(*group, pidleader) < 0) {
        virErrorPtr saved = virSaveLastError();
        virCgroupRemove(*group);
        virCgroupFree(group);
        if (saved) {
            virSetError(saved);
            virFreeError(saved);
        }
1163 1164
    }

1165
    return 0;
1166
}
1167

E
Eric Blake 已提交
1168

1169 1170 1171
/*
 * Returns 0 on success, -1 on fatal error
 */
1172
int virCgroupTerminateMachine(const char *name)
1173
{
1174
    return virSystemdTerminateMachine(name);
1175 1176 1177
}


1178 1179 1180
static int
virCgroupNewMachineManual(const char *name,
                          const char *drivername,
1181
                          pid_t pidleader,
1182 1183 1184 1185
                          const char *partition,
                          int controllers,
                          virCgroupPtr *group)
{
1186 1187
    virCgroupPtr parent = NULL;
    int ret = -1;
1188 1189

    VIR_DEBUG("Fallback to non-systemd setup");
1190 1191 1192 1193 1194
    if (virCgroupNewPartition(partition,
                              STREQ(partition, "/machine"),
                              controllers,
                              &parent) < 0) {
        if (virCgroupNewIgnoreError())
1195
            goto done;
1196

1197
        goto cleanup;
1198 1199 1200 1201 1202 1203 1204
    }

    if (virCgroupNewDomainPartition(parent,
                                    drivername,
                                    name,
                                    true,
                                    group) < 0)
1205
        goto cleanup;
1206

1207
    if (virCgroupAddProcess(*group, pidleader) < 0) {
1208 1209
        virErrorPtr saved = virSaveLastError();
        virCgroupRemove(*group);
1210
        virCgroupFree(group);
1211 1212 1213 1214 1215 1216
        if (saved) {
            virSetError(saved);
            virFreeError(saved);
        }
    }

1217 1218 1219 1220
 done:
    ret = 0;

 cleanup:
1221
    virCgroupFree(&parent);
1222
    return ret;
1223 1224
}

E
Eric Blake 已提交
1225 1226 1227 1228 1229 1230 1231 1232

int
virCgroupNewMachine(const char *name,
                    const char *drivername,
                    const unsigned char *uuid,
                    const char *rootdir,
                    pid_t pidleader,
                    bool isContainer,
1233 1234
                    size_t nnicindexes,
                    int *nicindexes,
E
Eric Blake 已提交
1235 1236 1237
                    const char *partition,
                    int controllers,
                    virCgroupPtr *group)
1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248
{
    int rv;

    *group = NULL;

    if ((rv = virCgroupNewMachineSystemd(name,
                                         drivername,
                                         uuid,
                                         rootdir,
                                         pidleader,
                                         isContainer,
1249 1250
                                         nnicindexes,
                                         nicindexes,
1251 1252 1253 1254 1255 1256 1257 1258 1259 1260
                                         partition,
                                         controllers,
                                         group)) == 0)
        return 0;

    if (rv == -1)
        return -1;

    return virCgroupNewMachineManual(name,
                                     drivername,
1261
                                     pidleader,
1262 1263 1264 1265 1266
                                     partition,
                                     controllers,
                                     group);
}

E
Eric Blake 已提交
1267 1268 1269

bool
virCgroupNewIgnoreError(void)
1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280
{
    if (virLastErrorIsSystemErrno(ENXIO) ||
        virLastErrorIsSystemErrno(EPERM) ||
        virLastErrorIsSystemErrno(EACCES)) {
        virResetLastError();
        VIR_DEBUG("No cgroups present/configured/accessible, ignoring error");
        return true;
    }
    return false;
}

E
Eric Blake 已提交
1281

E
Eric Blake 已提交
1282 1283 1284 1285 1286 1287
/**
 * virCgroupFree:
 *
 * @group: The group structure to free
 */
void
1288
virCgroupFree(virCgroupPtr *group)
E
Eric Blake 已提交
1289 1290 1291
{
    size_t i;

1292
    if (*group == NULL)
E
Eric Blake 已提交
1293 1294 1295
        return;

    for (i = 0; i < VIR_CGROUP_CONTROLLER_LAST; i++) {
1296 1297 1298
        VIR_FREE((*group)->legacy[i].mountPoint);
        VIR_FREE((*group)->legacy[i].linkPoint);
        VIR_FREE((*group)->legacy[i].placement);
E
Eric Blake 已提交
1299 1300
    }

P
Pavel Hrdina 已提交
1301 1302 1303
    VIR_FREE((*group)->unified.mountPoint);
    VIR_FREE((*group)->unified.placement);

1304 1305
    VIR_FREE((*group)->path);
    VIR_FREE(*group);
E
Eric Blake 已提交
1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320
}


/**
 * virCgroupHasController: query whether a cgroup controller is present
 *
 * @cgroup: The group structure to be queried, or NULL
 * @controller: cgroup subsystem id
 *
 * Returns true if a cgroup controller is mounted and is associated
 * with this cgroup object.
 */
bool
virCgroupHasController(virCgroupPtr cgroup, int controller)
{
1321 1322
    size_t i;

E
Eric Blake 已提交
1323 1324 1325 1326
    if (!cgroup)
        return false;
    if (controller < 0 || controller >= VIR_CGROUP_CONTROLLER_LAST)
        return false;
1327

1328 1329 1330 1331 1332 1333 1334 1335
    for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) {
        if (cgroup->backends[i] &&
            cgroup->backends[i]->hasController(cgroup, controller)) {
            return true;
        }
    }

    return false;
E
Eric Blake 已提交
1336 1337 1338 1339 1340
}


int
virCgroupPathOfController(virCgroupPtr group,
1341
                          unsigned int controller,
E
Eric Blake 已提交
1342 1343 1344
                          const char *key,
                          char **path)
{
1345 1346 1347
    if (controller >= VIR_CGROUP_CONTROLLER_LAST) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Invalid controller id '%d'"), controller);
E
Eric Blake 已提交
1348 1349 1350
        return -1;
    }

1351 1352
    VIR_CGROUP_BACKEND_CALL(group, controller, pathOfController, -1,
                            controller, key, path);
E
Eric Blake 已提交
1353 1354 1355
}


1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373
/**
 * virCgroupGetBlkioIoServiced:
 *
 * @group: The cgroup to get throughput for
 * @bytes_read: Pointer to returned bytes read
 * @bytes_write: Pointer to returned bytes written
 * @requests_read: Pointer to returned read io ops
 * @requests_write: Pointer to returned write io ops
 *
 * Returns: 0 on success, -1 on error
 */
int
virCgroupGetBlkioIoServiced(virCgroupPtr group,
                            long long *bytes_read,
                            long long *bytes_write,
                            long long *requests_read,
                            long long *requests_write)
{
1374 1375
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO,
                            getBlkioIoServiced, -1,
1376 1377
                            bytes_read, bytes_write,
                            requests_read, requests_write);
1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400
}


/**
 * virCgroupGetBlkioIoDeviceServiced:
 *
 * @group: The cgroup to get throughput for
 * @path: The device to get throughput for
 * @bytes_read: Pointer to returned bytes read
 * @bytes_write: Pointer to returned bytes written
 * @requests_read: Pointer to returned read io ops
 * @requests_write: Pointer to returned write io ops
 *
 * Returns: 0 on success, -1 on error
 */
int
virCgroupGetBlkioIoDeviceServiced(virCgroupPtr group,
                                  const char *path,
                                  long long *bytes_read,
                                  long long *bytes_write,
                                  long long *requests_read,
                                  long long *requests_write)
{
1401 1402
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO,
                            getBlkioIoDeviceServiced, -1,
1403 1404
                            path, bytes_read, bytes_write,
                            requests_read, requests_write);
1405 1406 1407
}


1408 1409 1410 1411 1412 1413
/**
 * virCgroupSetBlkioWeight:
 *
 * @group: The cgroup to change io weight for
 * @weight: The Weight for this cgroup
 *
1414
 * Returns: 0 on success, -1 on error
1415
 */
E
Eric Blake 已提交
1416 1417
int
virCgroupSetBlkioWeight(virCgroupPtr group, unsigned int weight)
1418
{
1419 1420
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO,
                            setBlkioWeight, -1, weight);
1421 1422
}

E
Eric Blake 已提交
1423

1424 1425 1426 1427 1428 1429
/**
 * virCgroupGetBlkioWeight:
 *
 * @group: The cgroup to get weight for
 * @Weight: Pointer to returned weight
 *
1430
 * Returns: 0 on success, -1 on error
1431
 */
E
Eric Blake 已提交
1432 1433
int
virCgroupGetBlkioWeight(virCgroupPtr group, unsigned int *weight)
1434
{
1435 1436
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO,
                            getBlkioWeight, -1, weight);
1437 1438
}

1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451
/**
 * virCgroupSetBlkioDeviceReadIops:
 * @group: The cgroup to change block io setting for
 * @path: The path of device
 * @riops: The new device read iops throttle, or 0 to clear
 *
 * Returns: 0 on success, -1 on error
 */
int
virCgroupSetBlkioDeviceReadIops(virCgroupPtr group,
                                const char *path,
                                unsigned int riops)
{
1452 1453
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO,
                            setBlkioDeviceReadIops, -1, path, riops);
1454 1455
}

E
Eric Blake 已提交
1456

1457
/**
1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469
 * virCgroupSetBlkioDeviceWriteIops:
 * @group: The cgroup to change block io setting for
 * @path: The path of device
 * @wiops: The new device write iops throttle, or 0 to clear
 *
 * Returns: 0 on success, -1 on error
 */
int
virCgroupSetBlkioDeviceWriteIops(virCgroupPtr group,
                                 const char *path,
                                 unsigned int wiops)
{
1470 1471
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO,
                            setBlkioDeviceWriteIops, -1, path, wiops);
1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487
}


/**
 * virCgroupSetBlkioDeviceReadBps:
 * @group: The cgroup to change block io setting for
 * @path: The path of device
 * @rbps: The new device read bps throttle, or 0 to clear
 *
 * Returns: 0 on success, -1 on error
 */
int
virCgroupSetBlkioDeviceReadBps(virCgroupPtr group,
                               const char *path,
                               unsigned long long rbps)
{
1488 1489
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO,
                            setBlkioDeviceReadBps, -1, path, rbps);
1490 1491 1492 1493 1494 1495 1496
}

/**
 * virCgroupSetBlkioDeviceWriteBps:
 * @group: The cgroup to change block io setting for
 * @path: The path of device
 * @wbps: The new device write bps throttle, or 0 to clear
1497
 *
1498 1499 1500 1501 1502 1503 1504
 * Returns: 0 on success, -1 on error
 */
int
virCgroupSetBlkioDeviceWriteBps(virCgroupPtr group,
                                const char *path,
                                unsigned long long wbps)
{
1505 1506
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO,
                            setBlkioDeviceWriteBps, -1, path, wbps);
1507 1508 1509 1510 1511 1512 1513
}


/**
 * virCgroupSetBlkioDeviceWeight:
 * @group: The cgroup to change block io setting for
 * @path: The path of device
1514 1515
 * @weight: The new device weight (100-1000),
 * (10-1000) after kernel 2.6.39, or 0 to clear
1516
 *
1517
 * Returns: 0 on success, -1 on error
1518
 */
E
Eric Blake 已提交
1519 1520 1521 1522
int
virCgroupSetBlkioDeviceWeight(virCgroupPtr group,
                              const char *path,
                              unsigned int weight)
1523
{
1524 1525
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO,
                            setBlkioDeviceWeight, -1, path, weight);
1526
}
1527

1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540
/**
 * virCgroupGetBlkioDeviceReadIops:
 * @group: The cgroup to gather block io setting for
 * @path: The path of device
 * @riops: Returned device read iops throttle, 0 if there is none
 *
 * Returns: 0 on success, -1 on error
 */
int
virCgroupGetBlkioDeviceReadIops(virCgroupPtr group,
                                const char *path,
                                unsigned int *riops)
{
1541 1542
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO,
                            getBlkioDeviceReadIops, -1, path, riops);
1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557
}

/**
 * virCgroupGetBlkioDeviceWriteIops:
 * @group: The cgroup to gather block io setting for
 * @path: The path of device
 * @wiops: Returned device write iops throttle, 0 if there is none
 *
 * Returns: 0 on success, -1 on error
 */
int
virCgroupGetBlkioDeviceWriteIops(virCgroupPtr group,
                                 const char *path,
                                 unsigned int *wiops)
{
1558 1559
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO,
                            getBlkioDeviceWriteIops, -1, path, wiops);
1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574
}

/**
 * virCgroupGetBlkioDeviceReadBps:
 * @group: The cgroup to gather block io setting for
 * @path: The path of device
 * @rbps: Returned device read bps throttle, 0 if there is none
 *
 * Returns: 0 on success, -1 on error
 */
int
virCgroupGetBlkioDeviceReadBps(virCgroupPtr group,
                               const char *path,
                               unsigned long long *rbps)
{
1575 1576
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO,
                            getBlkioDeviceReadBps, -1, path, rbps);
1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591
}

/**
 * virCgroupGetBlkioDeviceWriteBps:
 * @group: The cgroup to gather block io setting for
 * @path: The path of device
 * @wbps: Returned device write bps throttle, 0 if there is none
 *
 * Returns: 0 on success, -1 on error
 */
int
virCgroupGetBlkioDeviceWriteBps(virCgroupPtr group,
                                const char *path,
                                unsigned long long *wbps)
{
1592 1593
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO,
                            getBlkioDeviceWriteBps, -1, path, wbps);
1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608
}

/**
 * virCgroupGetBlkioDeviceWeight:
 * @group: The cgroup to gather block io setting for
 * @path: The path of device
 * @weight: Returned device weight, 0 if there is none
 *
 * Returns: 0 on success, -1 on error
 */
int
virCgroupGetBlkioDeviceWeight(virCgroupPtr group,
                              const char *path,
                              unsigned int *weight)
{
1609 1610
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_BLKIO,
                            getBlkioDeviceWeight, -1, path, weight);
1611 1612
}

1613

1614 1615 1616 1617 1618 1619 1620 1621
/**
 * virCgroupSetMemory:
 *
 * @group: The cgroup to change memory for
 * @kb: The memory amount in kilobytes
 *
 * Returns: 0 on success
 */
E
Eric Blake 已提交
1622 1623
int
virCgroupSetMemory(virCgroupPtr group, unsigned long long kb)
1624
{
1625 1626
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_MEMORY,
                            setMemory, -1, kb);
1627 1628
}

E
Eric Blake 已提交
1629

1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651
/**
 * virCgroupGetMemoryStat:
 *
 * @group: The cgroup to change memory for
 * @cache: page cache memory in KiB
 * @activeAnon: anonymous and swap cache memory in KiB
 * @inactiveAnon: anonymous and swap cache memory in KiB
 * @activeFile: file-backed memory in KiB
 * @inactiveFile: file-backed memory in KiB
 * @unevictable: memory that cannot be reclaimed KiB
 *
 * Returns: 0 on success, -1 on error
 */
int
virCgroupGetMemoryStat(virCgroupPtr group,
                       unsigned long long *cache,
                       unsigned long long *activeAnon,
                       unsigned long long *inactiveAnon,
                       unsigned long long *activeFile,
                       unsigned long long *inactiveFile,
                       unsigned long long *unevictable)
{
1652 1653
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_MEMORY,
                            getMemoryStat, -1, cache,
1654 1655 1656
                            activeAnon, inactiveAnon,
                            activeFile, inactiveFile,
                            unevictable);
1657 1658 1659
}


R
Ryota Ozaki 已提交
1660 1661 1662 1663 1664 1665 1666 1667
/**
 * virCgroupGetMemoryUsage:
 *
 * @group: The cgroup to change memory for
 * @kb: Pointer to returned used memory in kilobytes
 *
 * Returns: 0 on success
 */
E
Eric Blake 已提交
1668 1669
int
virCgroupGetMemoryUsage(virCgroupPtr group, unsigned long *kb)
R
Ryota Ozaki 已提交
1670
{
1671 1672
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_MEMORY,
                            getMemoryUsage, -1, kb);
R
Ryota Ozaki 已提交
1673 1674
}

E
Eric Blake 已提交
1675

1676 1677 1678 1679 1680 1681 1682 1683
/**
 * virCgroupSetMemoryHardLimit:
 *
 * @group: The cgroup to change memory hard limit for
 * @kb: The memory amount in kilobytes
 *
 * Returns: 0 on success
 */
E
Eric Blake 已提交
1684 1685
int
virCgroupSetMemoryHardLimit(virCgroupPtr group, unsigned long long kb)
1686
{
1687 1688
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_MEMORY,
                            setMemoryHardLimit, -1, kb);
1689 1690
}

E
Eric Blake 已提交
1691

1692 1693 1694 1695 1696 1697 1698 1699
/**
 * virCgroupGetMemoryHardLimit:
 *
 * @group: The cgroup to get the memory hard limit for
 * @kb: The memory amount in kilobytes
 *
 * Returns: 0 on success
 */
E
Eric Blake 已提交
1700 1701
int
virCgroupGetMemoryHardLimit(virCgroupPtr group, unsigned long long *kb)
1702
{
1703 1704
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_MEMORY,
                            getMemoryHardLimit, -1, kb);
1705 1706
}

E
Eric Blake 已提交
1707

1708 1709 1710 1711 1712 1713 1714 1715
/**
 * virCgroupSetMemorySoftLimit:
 *
 * @group: The cgroup to change memory soft limit for
 * @kb: The memory amount in kilobytes
 *
 * Returns: 0 on success
 */
E
Eric Blake 已提交
1716 1717
int
virCgroupSetMemorySoftLimit(virCgroupPtr group, unsigned long long kb)
1718
{
1719 1720
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_MEMORY,
                            setMemorySoftLimit, -1, kb);
1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731
}


/**
 * virCgroupGetMemorySoftLimit:
 *
 * @group: The cgroup to get the memory soft limit for
 * @kb: The memory amount in kilobytes
 *
 * Returns: 0 on success
 */
E
Eric Blake 已提交
1732 1733
int
virCgroupGetMemorySoftLimit(virCgroupPtr group, unsigned long long *kb)
1734
{
1735 1736
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_MEMORY,
                            getMemorySoftLimit, -1, kb);
1737 1738
}

E
Eric Blake 已提交
1739

1740
/**
1741
 * virCgroupSetMemSwapHardLimit:
1742
 *
1743 1744
 * @group: The cgroup to change mem+swap hard limit for
 * @kb: The mem+swap amount in kilobytes
1745 1746 1747
 *
 * Returns: 0 on success
 */
E
Eric Blake 已提交
1748 1749
int
virCgroupSetMemSwapHardLimit(virCgroupPtr group, unsigned long long kb)
1750
{
1751 1752
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_MEMORY,
                            setMemSwapHardLimit, -1, kb);
1753 1754
}

E
Eric Blake 已提交
1755

1756
/**
1757
 * virCgroupGetMemSwapHardLimit:
1758
 *
1759 1760
 * @group: The cgroup to get mem+swap hard limit for
 * @kb: The mem+swap amount in kilobytes
1761 1762 1763
 *
 * Returns: 0 on success
 */
E
Eric Blake 已提交
1764 1765
int
virCgroupGetMemSwapHardLimit(virCgroupPtr group, unsigned long long *kb)
1766
{
1767 1768
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_MEMORY,
                            getMemSwapHardLimit, -1, kb);
1769 1770
}

E
Eric Blake 已提交
1771

G
Gao feng 已提交
1772 1773 1774 1775 1776 1777 1778 1779
/**
 * virCgroupGetMemSwapUsage:
 *
 * @group: The cgroup to get mem+swap usage for
 * @kb: The mem+swap amount in kilobytes
 *
 * Returns: 0 on success
 */
E
Eric Blake 已提交
1780 1781
int
virCgroupGetMemSwapUsage(virCgroupPtr group, unsigned long long *kb)
G
Gao feng 已提交
1782
{
1783 1784
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_MEMORY,
                            getMemSwapUsage, -1, kb);
G
Gao feng 已提交
1785 1786
}

E
Eric Blake 已提交
1787

1788 1789 1790 1791 1792 1793 1794 1795
/**
 * virCgroupSetCpusetMems:
 *
 * @group: The cgroup to set cpuset.mems for
 * @mems: the numa nodes to set
 *
 * Returns: 0 on success
 */
E
Eric Blake 已提交
1796 1797
int
virCgroupSetCpusetMems(virCgroupPtr group, const char *mems)
1798
{
1799 1800
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPUSET,
                            setCpusetMems, -1, mems);
1801 1802
}

E
Eric Blake 已提交
1803

1804 1805 1806 1807 1808 1809 1810 1811
/**
 * virCgroupGetCpusetMems:
 *
 * @group: The cgroup to get cpuset.mems for
 * @mems: the numa nodes to get
 *
 * Returns: 0 on success
 */
E
Eric Blake 已提交
1812 1813
int
virCgroupGetCpusetMems(virCgroupPtr group, char **mems)
1814
{
1815 1816
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPUSET,
                            getCpusetMems, -1, mems);
1817 1818
}

E
Eric Blake 已提交
1819

1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830
/**
 * virCgroupSetCpusetMemoryMigrate:
 *
 * @group: The cgroup to set cpuset.memory_migrate for
 * @migrate: Whether to migrate the memory on change or not
 *
 * Returns: 0 on success
 */
int
virCgroupSetCpusetMemoryMigrate(virCgroupPtr group, bool migrate)
{
1831 1832
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPUSET,
                            setCpusetMemoryMigrate, -1, migrate);
1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846
}


/**
 * virCgroupGetCpusetMemoryMigrate:
 *
 * @group: The cgroup to get cpuset.memory_migrate for
 * @migrate: Migration setting
 *
 * Returns: 0 on success
 */
int
virCgroupGetCpusetMemoryMigrate(virCgroupPtr group, bool *migrate)
{
1847 1848
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPUSET,
                            getCpusetMemoryMigrate, -1, migrate);
1849 1850 1851
}


1852 1853 1854 1855 1856 1857
/**
 * virCgroupSetCpusetCpus:
 *
 * @group: The cgroup to set cpuset.cpus for
 * @cpus: the cpus to set
 *
N
Nitesh Konkar 已提交
1858
 * Returns: 0 on success
1859
 */
E
Eric Blake 已提交
1860 1861
int
virCgroupSetCpusetCpus(virCgroupPtr group, const char *cpus)
1862
{
1863 1864
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPUSET,
                            setCpusetCpus, -1, cpus);
1865 1866
}

E
Eric Blake 已提交
1867

1868 1869 1870 1871 1872 1873
/**
 * virCgroupGetCpusetCpus:
 *
 * @group: The cgroup to get cpuset.cpus for
 * @cpus: the cpus to get
 *
N
Nitesh Konkar 已提交
1874
 * Returns: 0 on success
1875
 */
E
Eric Blake 已提交
1876 1877
int
virCgroupGetCpusetCpus(virCgroupPtr group, char **cpus)
1878
{
1879 1880
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPUSET,
                            getCpusetCpus, -1, cpus);
1881 1882
}

E
Eric Blake 已提交
1883

1884 1885 1886
/**
 * virCgroupDenyAllDevices:
 *
1887
 * @group: The cgroup to deny all permissions, for all devices
1888 1889 1890
 *
 * Returns: 0 on success
 */
E
Eric Blake 已提交
1891 1892
int
virCgroupDenyAllDevices(virCgroupPtr group)
1893
{
1894 1895
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_DEVICES,
                            denyAllDevices, -1);
1896 1897
}

1898 1899 1900
/**
 * virCgroupAllowAllDevices:
 *
1901
 * Allows the permission for all devices by setting lines similar
1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914
 * to these ones (obviously the 'm' permission is an example):
 *
 * 'b *:* m'
 * 'c *:* m'
 *
 * @group: The cgroup to allow devices for
 * @perms: Bitwise or of VIR_CGROUP_DEVICE permission bits to allow
 *
 * Returns: 0 on success
 */
int
virCgroupAllowAllDevices(virCgroupPtr group, int perms)
{
1915 1916
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_DEVICES,
                            allowAllDevices, -1, perms);
1917 1918
}

E
Eric Blake 已提交
1919

1920 1921 1922 1923 1924
/**
 * virCgroupAllowDevice:
 *
 * @group: The cgroup to allow a device for
 * @type: The device type (i.e., 'c' or 'b')
1925 1926
 * @major: The major number of the device, a negative value means '*'
 * @minor: The minor number of the device, a negative value means '*'
1927
 * @perms: Bitwise or of VIR_CGROUP_DEVICE permission bits to allow
1928 1929 1930
 *
 * Returns: 0 on success
 */
E
Eric Blake 已提交
1931 1932 1933
int
virCgroupAllowDevice(virCgroupPtr group, char type, int major, int minor,
                     int perms)
1934
{
1935 1936
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_DEVICES,
                            allowDevice, -1, type, major, minor, perms);
1937
}
1938

E
Eric Blake 已提交
1939

1940 1941 1942 1943 1944
/**
 * virCgroupAllowDevicePath:
 *
 * @group: The cgroup to allow the device for
 * @path: the device to allow
1945
 * @perms: Bitwise or of VIR_CGROUP_DEVICE permission bits to allow
1946
 * @ignoreEacces: Ignore lack of permission (mostly for NFS mounts)
1947 1948 1949 1950
 *
 * Queries the type of device and its major/minor number, and
 * adds that to the cgroup ACL
 *
1951
 * Returns: 0 on success, 1 if path exists but is not a device or is not
1952
 * accessible, or * -1 on error
1953
 */
E
Eric Blake 已提交
1954
int
1955 1956 1957 1958
virCgroupAllowDevicePath(virCgroupPtr group,
                         const char *path,
                         int perms,
                         bool ignoreEacces)
1959 1960 1961
{
    struct stat sb;

1962
    if (stat(path, &sb) < 0) {
1963 1964 1965
        if (errno == EACCES && ignoreEacces)
            return 1;

1966 1967 1968 1969 1970
        virReportSystemError(errno,
                             _("Path '%s' is not accessible"),
                             path);
        return -1;
    }
1971 1972

    if (!S_ISCHR(sb.st_mode) && !S_ISBLK(sb.st_mode))
1973
        return 1;
1974

1975 1976
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_DEVICES,
                            allowDevice, -1,
1977 1978 1979 1980
                            S_ISCHR(sb.st_mode) ? 'c' : 'b',
                            major(sb.st_rdev),
                            minor(sb.st_rdev),
                            perms);
1981
}
D
Daniel P. Berrange 已提交
1982

1983 1984 1985 1986 1987 1988

/**
 * virCgroupDenyDevice:
 *
 * @group: The cgroup to deny a device for
 * @type: The device type (i.e., 'c' or 'b')
1989 1990
 * @major: The major number of the device, a negative value means '*'
 * @minor: The minor number of the device, a negative value means '*'
1991
 * @perms: Bitwise or of VIR_CGROUP_DEVICE permission bits to deny
1992 1993 1994
 *
 * Returns: 0 on success
 */
E
Eric Blake 已提交
1995 1996 1997
int
virCgroupDenyDevice(virCgroupPtr group, char type, int major, int minor,
                    int perms)
1998
{
1999 2000
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_DEVICES,
                            denyDevice, -1, type, major, minor, perms);
2001 2002
}

E
Eric Blake 已提交
2003

2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
/**
 * virCgroupDenyDevicePath:
 *
 * @group: The cgroup to deny the device for
 * @path: the device to deny
 * @perms: Bitwise or of VIR_CGROUP_DEVICE permission bits to allow
 * @ignoreEacces: Ignore lack of permission (mostly for NFS mounts)
 *
 * Queries the type of device and its major/minor number, and
 * removes it from the cgroup ACL
 *
 * Returns: 0 on success, 1 if path exists but is not a device or is not
 * accessible, or -1 on error.
 */
E
Eric Blake 已提交
2018
int
2019 2020 2021 2022
virCgroupDenyDevicePath(virCgroupPtr group,
                        const char *path,
                        int perms,
                        bool ignoreEacces)
2023 2024 2025
{
    struct stat sb;

2026
    if (stat(path, &sb) < 0) {
2027 2028 2029
        if (errno == EACCES && ignoreEacces)
            return 1;

2030 2031 2032 2033 2034
        virReportSystemError(errno,
                             _("Path '%s' is not accessible"),
                             path);
        return -1;
    }
2035 2036

    if (!S_ISCHR(sb.st_mode) && !S_ISBLK(sb.st_mode))
2037
        return 1;
2038

2039 2040
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_DEVICES,
                            denyDevice, -1,
2041 2042 2043 2044
                            S_ISCHR(sb.st_mode) ? 'c' : 'b',
                            major(sb.st_rdev),
                            minor(sb.st_rdev),
                            perms);
2045 2046
}

E
Eric Blake 已提交
2047

2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063
/* This function gets the sums of cpu time consumed by all vcpus.
 * For example, if there are 4 physical cpus, and 2 vcpus in a domain,
 * then for each vcpu, the cpuacct.usage_percpu looks like this:
 *   t0 t1 t2 t3
 * and we have 2 groups of such data:
 *   v\p   0   1   2   3
 *   0   t00 t01 t02 t03
 *   1   t10 t11 t12 t13
 * for each pcpu, the sum is cpu time consumed by all vcpus.
 *   s0 = t00 + t10
 *   s1 = t01 + t11
 *   s2 = t02 + t12
 *   s3 = t03 + t13
 */
static int
virCgroupGetPercpuVcpuSum(virCgroupPtr group,
2064
                          virBitmapPtr guestvcpus,
2065
                          unsigned long long *sum_cpu_time,
2066 2067
                          size_t nsum,
                          virBitmapPtr cpumap)
2068
{
2069
    int ret = -1;
2070
    ssize_t i = -1;
2071
    virCgroupPtr group_vcpu = NULL;
2072

2073
    while ((i = virBitmapNextSetBit(guestvcpus, i)) >= 0) {
2074
        VIR_AUTOFREE(char *) buf = NULL;
2075 2076
        char *pos;
        unsigned long long tmp;
2077
        ssize_t j;
2078

J
John Ferlan 已提交
2079 2080
        if (virCgroupNewThread(group, VIR_CGROUP_THREAD_VCPU, i,
                               false, &group_vcpu) < 0)
2081
            goto cleanup;
2082 2083

        if (virCgroupGetCpuacctPercpuUsage(group_vcpu, &buf) < 0)
2084
            goto cleanup;
2085 2086

        pos = buf;
2087 2088 2089
        for (j = virBitmapNextSetBit(cpumap, -1);
             j >= 0 && j < nsum;
             j = virBitmapNextSetBit(cpumap, j)) {
2090 2091 2092
            if (virStrToLong_ull(pos, &pos, 10, &tmp) < 0) {
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("cpuacct parse error"));
2093
                goto cleanup;
2094 2095 2096
            }
            sum_cpu_time[j] += tmp;
        }
2097

2098
        virCgroupFree(&group_vcpu);
2099 2100
    }

2101 2102
    ret = 0;
 cleanup:
2103
    virCgroupFree(&group_vcpu);
2104
    return ret;
2105 2106 2107
}


2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127
/**
 * virCgroupGetPercpuStats:
 * @cgroup: cgroup data structure
 * @params: typed parameter array where data is returned
 * @nparams: cardinality of @params
 * @start_cpu: offset of physical CPU to get data for
 * @ncpus: number of physical CPUs to get data for
 * @nvcpupids: number of vCPU threads for a domain (actual number of vcpus)
 *
 * This function is the worker that retrieves data in the appropriate format
 * for the terribly designed 'virDomainGetCPUStats' API. Sharing semantics with
 * the API, this function has two modes of operation depending on magic settings
 * of the input arguments. Please refer to docs of 'virDomainGetCPUStats' for
 * the usage patterns of the similarly named arguments.
 *
 * @nvcpupids determines the count of active vcpu threads for the vm. If the
 * threads could not be detected the percpu data is skipped.
 *
 * Please DON'T use this function anywhere else.
 */
2128 2129 2130 2131 2132
int
virCgroupGetPercpuStats(virCgroupPtr group,
                        virTypedParameterPtr params,
                        unsigned int nparams,
                        int start_cpu,
2133
                        unsigned int ncpus,
2134
                        virBitmapPtr guestvcpus)
2135
{
2136
    int ret = -1;
2137
    size_t i;
2138
    int need_cpus, total_cpus;
2139
    char *pos;
2140 2141
    VIR_AUTOFREE(char *) buf = NULL;
    VIR_AUTOFREE(unsigned long long *) sum_cpu_time = NULL;
2142 2143 2144
    virTypedParameterPtr ent;
    int param_idx;
    unsigned long long cpu_time;
2145
    virBitmapPtr cpumap = NULL;
2146 2147

    /* return the number of supported params */
2148
    if (nparams == 0 && ncpus != 0) {
2149
        if (!guestvcpus)
2150 2151 2152 2153
            return CGROUP_NB_PER_CPU_STAT_PARAM;
        else
            return CGROUP_NB_PER_CPU_STAT_PARAM + 1;
    }
2154 2155

    /* To parse account file, we need to know how many cpus are present.  */
2156
    if (!(cpumap = virHostCPUGetPresentBitmap()))
2157
        return -1;
2158

2159 2160
    total_cpus = virBitmapSize(cpumap);

2161
    /* return total number of cpus */
2162 2163 2164 2165
    if (ncpus == 0) {
        ret = total_cpus;
        goto cleanup;
    }
2166

2167
    if (start_cpu >= total_cpus) {
2168 2169
        virReportError(VIR_ERR_INVALID_ARG,
                       _("start_cpu %d larger than maximum of %d"),
2170
                       start_cpu, total_cpus - 1);
2171
        goto cleanup;
2172 2173 2174 2175
    }

    /* we get percpu cputime accounting info. */
    if (virCgroupGetCpuacctPercpuUsage(group, &buf))
2176
        goto cleanup;
2177 2178 2179 2180 2181 2182
    pos = buf;

    /* return percpu cputime in index 0 */
    param_idx = 0;

    /* number of cpus to compute */
J
Ján Tomko 已提交
2183
    need_cpus = MIN(total_cpus, start_cpu + ncpus);
2184

J
Ján Tomko 已提交
2185
    for (i = 0; i < need_cpus; i++) {
J
Ján Tomko 已提交
2186
        if (!virBitmapIsBitSet(cpumap, i)) {
2187 2188
            cpu_time = 0;
        } else if (virStrToLong_ull(pos, &pos, 10, &cpu_time) < 0) {
2189 2190
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("cpuacct parse error"));
2191
            goto cleanup;
2192 2193 2194 2195 2196 2197
        }
        if (i < start_cpu)
            continue;
        ent = &params[(i - start_cpu) * nparams + param_idx];
        if (virTypedParameterAssign(ent, VIR_DOMAIN_CPU_STATS_CPUTIME,
                                    VIR_TYPED_PARAM_ULLONG, cpu_time) < 0)
2198
            goto cleanup;
2199 2200
    }

2201
    /* return percpu vcputime in index 1 */
2202
    param_idx = 1;
2203

2204
    if (guestvcpus && param_idx < nparams) {
2205
        if (VIR_ALLOC_N(sum_cpu_time, need_cpus) < 0)
2206
            goto cleanup;
2207 2208
        if (virCgroupGetPercpuVcpuSum(group, guestvcpus, sum_cpu_time,
                                      need_cpus, cpumap) < 0)
2209
            goto cleanup;
2210 2211

        for (i = start_cpu; i < need_cpus; i++) {
2212 2213
            int idx = (i - start_cpu) * nparams + param_idx;
            if (virTypedParameterAssign(&params[idx],
2214 2215 2216
                                        VIR_DOMAIN_CPU_STATS_VCPUTIME,
                                        VIR_TYPED_PARAM_ULLONG,
                                        sum_cpu_time[i]) < 0)
2217
                goto cleanup;
2218 2219 2220
        }

        param_idx++;
2221 2222
    }

2223 2224 2225 2226 2227
    ret = param_idx;

 cleanup:
    virBitmapFree(cpumap);
    return ret;
2228 2229
}

2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279

int
virCgroupGetDomainTotalCpuStats(virCgroupPtr group,
                                virTypedParameterPtr params,
                                int nparams)
{
    unsigned long long cpu_time;
    int ret;

    if (nparams == 0) /* return supported number of params */
        return CGROUP_NB_TOTAL_CPU_STAT_PARAM;
    /* entry 0 is cputime */
    ret = virCgroupGetCpuacctUsage(group, &cpu_time);
    if (ret < 0) {
        virReportSystemError(-ret, "%s", _("unable to get cpu account"));
        return -1;
    }

    if (virTypedParameterAssign(&params[0], VIR_DOMAIN_CPU_STATS_CPUTIME,
                                VIR_TYPED_PARAM_ULLONG, cpu_time) < 0)
        return -1;

    if (nparams > 1) {
        unsigned long long user;
        unsigned long long sys;

        ret = virCgroupGetCpuacctStat(group, &user, &sys);
        if (ret < 0) {
            virReportSystemError(-ret, "%s", _("unable to get cpu account"));
            return -1;
        }

        if (virTypedParameterAssign(&params[1],
                                    VIR_DOMAIN_CPU_STATS_USERTIME,
                                    VIR_TYPED_PARAM_ULLONG, user) < 0)
            return -1;
        if (nparams > 2 &&
            virTypedParameterAssign(&params[2],
                                    VIR_DOMAIN_CPU_STATS_SYSTEMTIME,
                                    VIR_TYPED_PARAM_ULLONG, sys) < 0)
            return -1;

        if (nparams > CGROUP_NB_TOTAL_CPU_STAT_PARAM)
            nparams = CGROUP_NB_TOTAL_CPU_STAT_PARAM;
    }

    return nparams;
}


E
Eric Blake 已提交
2280 2281
int
virCgroupSetCpuShares(virCgroupPtr group, unsigned long long shares)
2282
{
2283 2284
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPU,
                            setCpuShares, -1, shares);
2285 2286
}

E
Eric Blake 已提交
2287 2288 2289

int
virCgroupGetCpuShares(virCgroupPtr group, unsigned long long *shares)
2290
{
2291 2292
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPU,
                            getCpuShares, -1, shares);
2293
}
2294

E
Eric Blake 已提交
2295

2296 2297 2298 2299 2300 2301 2302 2303
/**
 * virCgroupSetCpuCfsPeriod:
 *
 * @group: The cgroup to change cpu.cfs_period_us for
 * @cfs_period: The bandwidth period in usecs
 *
 * Returns: 0 on success
 */
E
Eric Blake 已提交
2304 2305
int
virCgroupSetCpuCfsPeriod(virCgroupPtr group, unsigned long long cfs_period)
2306
{
2307 2308
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPU,
                            setCpuCfsPeriod, -1, cfs_period);
2309 2310
}

E
Eric Blake 已提交
2311

2312 2313 2314 2315 2316 2317 2318 2319
/**
 * virCgroupGetCpuCfsPeriod:
 *
 * @group: The cgroup to get cpu.cfs_period_us for
 * @cfs_period: Pointer to the returned bandwidth period in usecs
 *
 * Returns: 0 on success
 */
E
Eric Blake 已提交
2320 2321
int
virCgroupGetCpuCfsPeriod(virCgroupPtr group, unsigned long long *cfs_period)
2322
{
2323 2324
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPU,
                            getCpuCfsPeriod, -1, cfs_period);
2325 2326
}

E
Eric Blake 已提交
2327

2328 2329 2330 2331 2332 2333 2334 2335 2336
/**
 * virCgroupSetCpuCfsQuota:
 *
 * @group: The cgroup to change cpu.cfs_quota_us for
 * @cfs_quota: the cpu bandwidth (in usecs) that this tg will be allowed to
 *             consume over period
 *
 * Returns: 0 on success
 */
E
Eric Blake 已提交
2337 2338
int
virCgroupSetCpuCfsQuota(virCgroupPtr group, long long cfs_quota)
2339
{
2340 2341
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPU,
                            setCpuCfsQuota, -1, cfs_quota);
2342 2343
}

E
Eric Blake 已提交
2344 2345 2346

int
virCgroupGetCpuacctPercpuUsage(virCgroupPtr group, char **usage)
2347
{
2348 2349
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPUACCT,
                            getCpuacctPercpuUsage, -1, usage);
2350 2351
}

E
Eric Blake 已提交
2352

2353
int
E
Eric Blake 已提交
2354 2355 2356 2357 2358
virCgroupRemoveRecursively(char *grppath)
{
    DIR *grpdir;
    struct dirent *ent;
    int rc = 0;
E
Eric Blake 已提交
2359
    int direrr;
E
Eric Blake 已提交
2360

J
Ján Tomko 已提交
2361
    if (virDirOpenQuiet(&grpdir, grppath) < 0) {
E
Eric Blake 已提交
2362 2363 2364 2365 2366 2367 2368
        if (errno == ENOENT)
            return 0;
        rc = -errno;
        VIR_ERROR(_("Unable to open %s (%d)"), grppath, errno);
        return rc;
    }

E
Eric Blake 已提交
2369 2370 2371
    /* This is best-effort cleanup: we want to log failures with just
     * VIR_ERROR instead of normal virReportError */
    while ((direrr = virDirRead(grpdir, &ent, NULL)) > 0) {
2372
        VIR_AUTOFREE(char *) path = NULL;
E
Eric Blake 已提交
2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383

        if (ent->d_type != DT_DIR) continue;

        if (virAsprintf(&path, "%s/%s", grppath, ent->d_name) == -1) {
            rc = -ENOMEM;
            break;
        }
        rc = virCgroupRemoveRecursively(path);
        if (rc != 0)
            break;
    }
E
Eric Blake 已提交
2384 2385 2386 2387 2388
    if (direrr < 0) {
        rc = -errno;
        VIR_ERROR(_("Failed to readdir for %s (%d)"), grppath, errno);
    }

J
Ján Tomko 已提交
2389
    VIR_DIR_CLOSE(grpdir);
E
Eric Blake 已提交
2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415

    VIR_DEBUG("Removing cgroup %s", grppath);
    if (rmdir(grppath) != 0 && errno != ENOENT) {
        rc = -errno;
        VIR_ERROR(_("Unable to remove %s (%d)"), grppath, errno);
    }

    return rc;
}


/**
 * virCgroupRemove:
 *
 * @group: The group to be removed
 *
 * It first removes all child groups recursively
 * in depth first order and then removes @group
 * because the presence of the child groups
 * prevents removing @group.
 *
 * Returns: 0 on success
 */
int
virCgroupRemove(virCgroupPtr group)
{
2416 2417 2418 2419 2420 2421 2422 2423 2424 2425
    size_t i;

    for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) {
        if (group->backends[i] &&
            group->backends[i]->remove(group) < 0) {
            return -1;
        }
    }

    return 0;
E
Eric Blake 已提交
2426 2427 2428
}


2429 2430 2431
/*
 * Returns 1 if some PIDs are killed, 0 if none are killed, or -1 on error
 */
E
Eric Blake 已提交
2432
static int
2433 2434 2435 2436 2437
virCgroupKillInternal(virCgroupPtr group,
                      int signum,
                      virHashTablePtr pids,
                      int controller,
                      const char *taskFile)
2438
{
2439
    int ret = -1;
2440
    bool killedAny = false;
2441
    VIR_AUTOFREE(char *) keypath = NULL;
2442
    bool done = false;
E
Eric Blake 已提交
2443 2444 2445
    FILE *fp = NULL;
    VIR_DEBUG("group=%p path=%s signum=%d pids=%p",
              group, group->path, signum, pids);
2446

2447
    if (virCgroupPathOfController(group, controller, taskFile, &keypath) < 0)
2448
        return -1;
2449 2450 2451 2452 2453 2454 2455

    /* PIDs may be forking as we kill them, so loop
     * until there are no new PIDs found
     */
    while (!done) {
        done = true;
        if (!(fp = fopen(keypath, "r"))) {
2456 2457 2458 2459 2460 2461
            if (errno == ENOENT) {
                VIR_DEBUG("No file %s, assuming done", keypath);
                killedAny = false;
                goto done;
            }

2462 2463 2464
            virReportSystemError(errno,
                                 _("Failed to read %s"),
                                 keypath);
2465 2466 2467
            goto cleanup;
        } else {
            while (!feof(fp)) {
M
Michal Privoznik 已提交
2468 2469
                long pid_value;
                if (fscanf(fp, "%ld", &pid_value) != 1) {
2470 2471
                    if (feof(fp))
                        break;
2472 2473 2474
                    virReportSystemError(errno,
                                         _("Failed to read %s"),
                                         keypath);
E
Eric Blake 已提交
2475
                    goto cleanup;
2476
                }
2477
                if (virHashLookup(pids, (void*)pid_value))
2478 2479
                    continue;

M
Michal Privoznik 已提交
2480
                VIR_DEBUG("pid=%ld", pid_value);
2481 2482
                /* Cgroups is a Linux concept, so this cast is safe.  */
                if (kill((pid_t)pid_value, signum) < 0) {
2483
                    if (errno != ESRCH) {
2484
                        virReportSystemError(errno,
M
Michal Privoznik 已提交
2485
                                             _("Failed to kill process %ld"),
2486
                                             pid_value);
2487 2488 2489 2490
                        goto cleanup;
                    }
                    /* Leave RC == 0 since we didn't kill one */
                } else {
2491
                    killedAny = true;
2492 2493 2494
                    done = false;
                }

2495
                ignore_value(virHashAddEntry(pids, (void*)pid_value, (void*)1));
2496 2497 2498 2499 2500
            }
            VIR_FORCE_FCLOSE(fp);
        }
    }

2501
 done:
2502
    ret = killedAny ? 1 : 0;
2503

2504
 cleanup:
E
Eric Blake 已提交
2505
    VIR_FORCE_FCLOSE(fp);
2506

2507
    return ret;
2508 2509 2510
}


E
Eric Blake 已提交
2511 2512
static uint32_t
virCgroupPidCode(const void *name, uint32_t seed)
2513
{
M
Michal Privoznik 已提交
2514
    long pid_value = (long)(intptr_t)name;
2515
    return virHashCodeGen(&pid_value, sizeof(pid_value), seed);
2516
}
E
Eric Blake 已提交
2517 2518 2519 2520


static bool
virCgroupPidEqual(const void *namea, const void *nameb)
2521 2522 2523
{
    return namea == nameb;
}
E
Eric Blake 已提交
2524 2525 2526 2527


static void *
virCgroupPidCopy(const void *name)
2528 2529 2530 2531
{
    return (void*)name;
}

E
Eric Blake 已提交
2532

2533
int
E
Eric Blake 已提交
2534 2535 2536
virCgroupKillRecursiveInternal(virCgroupPtr group,
                               int signum,
                               virHashTablePtr pids,
2537 2538
                               int controller,
                               const char *taskFile,
E
Eric Blake 已提交
2539
                               bool dormdir)
2540
{
2541
    int ret = -1;
2542
    int rc;
2543
    bool killedAny = false;
2544
    VIR_AUTOFREE(char *) keypath = NULL;
2545
    DIR *dp = NULL;
2546
    virCgroupPtr subgroup = NULL;
2547
    struct dirent *ent;
E
Eric Blake 已提交
2548
    int direrr;
E
Eric Blake 已提交
2549 2550
    VIR_DEBUG("group=%p path=%s signum=%d pids=%p",
              group, group->path, signum, pids);
2551

2552
    if (virCgroupPathOfController(group, controller, "", &keypath) < 0)
2553
        return -1;
2554

2555 2556
    if ((rc = virCgroupKillInternal(group, signum, pids,
                                    controller, taskFile)) < 0) {
2557
        goto cleanup;
2558
    }
2559 2560
    if (rc == 1)
        killedAny = true;
2561

2562
    VIR_DEBUG("Iterate over children of %s (killedAny=%d)", keypath, killedAny);
J
Ján Tomko 已提交
2563
    if ((rc = virDirOpenIfExists(&dp, keypath)) < 0)
2564
        goto cleanup;
J
Ján Tomko 已提交
2565 2566 2567 2568 2569

    if (rc == 0) {
        VIR_DEBUG("Path %s does not exist, assuming done", keypath);
        killedAny = false;
        goto done;
2570 2571
    }

E
Eric Blake 已提交
2572
    while ((direrr = virDirRead(dp, &ent, keypath)) > 0) {
2573 2574 2575 2576 2577
        if (ent->d_type != DT_DIR)
            continue;

        VIR_DEBUG("Process subdir %s", ent->d_name);

2578
        if (virCgroupNew(-1, ent->d_name, group, -1, &subgroup) < 0)
2579 2580
            goto cleanup;

E
Eric Blake 已提交
2581
        if ((rc = virCgroupKillRecursiveInternal(subgroup, signum, pids,
2582
                                                 controller, taskFile, true)) < 0)
2583 2584
            goto cleanup;
        if (rc == 1)
2585
            killedAny = true;
2586 2587 2588

        if (dormdir)
            virCgroupRemove(subgroup);
2589

2590
        virCgroupFree(&subgroup);
2591
    }
E
Eric Blake 已提交
2592 2593
    if (direrr < 0)
        goto cleanup;
2594

2595
 done:
2596
    ret = killedAny ? 1 : 0;
2597

2598
 cleanup:
2599
    virCgroupFree(&subgroup);
J
Ján Tomko 已提交
2600
    VIR_DIR_CLOSE(dp);
2601
    return ret;
2602 2603
}

E
Eric Blake 已提交
2604 2605 2606

int
virCgroupKillRecursive(virCgroupPtr group, int signum)
2607
{
2608 2609 2610
    int ret = 0;
    int rc;
    size_t i;
2611
    bool backendAvailable = false;
2612
    virCgroupBackendPtr *backends = virCgroupBackendGetAll();
2613 2614 2615 2616 2617 2618
    virHashTablePtr pids = virHashCreateFull(100,
                                             NULL,
                                             virCgroupPidCode,
                                             virCgroupPidEqual,
                                             virCgroupPidCopy,
                                             NULL);
2619

2620
    VIR_DEBUG("group=%p path=%s signum=%d", group, group->path, signum);
2621

2622
    for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) {
2623 2624
        if (backends && backends[i] && backends[i]->available()) {
            backendAvailable = true;
2625 2626 2627 2628 2629 2630 2631 2632 2633 2634
            rc = backends[i]->killRecursive(group, signum, pids);
            if (rc < 0) {
                ret = -1;
                goto cleanup;
            }
            if (rc > 0)
                ret = rc;
        }
    }

2635 2636 2637 2638 2639 2640
    if (!backends || !backendAvailable) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("no cgroup backend available"));
        goto cleanup;
    }

2641 2642
 cleanup:
    virHashFree(pids);
2643
    return ret;
2644 2645 2646
}


E
Eric Blake 已提交
2647 2648
int
virCgroupKillPainfully(virCgroupPtr group)
2649
{
2650
    size_t i;
2651
    int ret;
2652
    VIR_DEBUG("cgroup=%p path=%s", group, group->path);
2653
    for (i = 0; i < 15; i++) {
2654 2655 2656 2657 2658 2659
        int signum;
        if (i == 0)
            signum = SIGTERM;
        else if (i == 8)
            signum = SIGKILL;
        else
J
Ján Tomko 已提交
2660
            signum = 0; /* Just check for existence */
2661

2662 2663 2664 2665
        ret = virCgroupKillRecursive(group, signum);
        VIR_DEBUG("Iteration %zu rc=%d", i, ret);
        /* If ret == -1 we hit error, if 0 we ran out of PIDs */
        if (ret <= 0)
2666 2667 2668 2669
            break;

        usleep(200 * 1000);
    }
2670 2671
    VIR_DEBUG("Complete %d", ret);
    return ret;
2672
}
2673

E
Eric Blake 已提交
2674

2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686
/**
 * virCgroupGetCpuCfsQuota:
 *
 * @group: The cgroup to get cpu.cfs_quota_us for
 * @cfs_quota: Pointer to the returned cpu bandwidth (in usecs) that this tg
 *             will be allowed to consume over period
 *
 * Returns: 0 on success
 */
int
virCgroupGetCpuCfsQuota(virCgroupPtr group, long long *cfs_quota)
{
2687 2688
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPU,
                            getCpuCfsQuota, -1, cfs_quota);
2689 2690 2691 2692 2693 2694
}


int
virCgroupGetCpuacctUsage(virCgroupPtr group, unsigned long long *usage)
{
2695 2696
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPUACCT,
                            getCpuacctUsage, -1, usage);
2697 2698 2699 2700 2701 2702 2703
}


int
virCgroupGetCpuacctStat(virCgroupPtr group, unsigned long long *user,
                        unsigned long long *sys)
{
2704 2705
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_CPUACCT,
                            getCpuacctStat, -1, user, sys);
2706 2707 2708 2709 2710 2711
}


int
virCgroupSetFreezerState(virCgroupPtr group, const char *state)
{
2712 2713
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_FREEZER,
                            setFreezerState, -1, state);
2714 2715 2716 2717 2718 2719
}


int
virCgroupGetFreezerState(virCgroupPtr group, char **state)
{
2720 2721
    VIR_CGROUP_BACKEND_CALL(group, VIR_CGROUP_CONTROLLER_FREEZER,
                            getFreezerState, -1, state);
2722 2723 2724
}


E
Eric Blake 已提交
2725
int
2726 2727
virCgroupBindMount(virCgroupPtr group, const char *oldroot,
                   const char *mountopts)
2728
{
2729 2730 2731 2732 2733 2734 2735 2736 2737 2738
    size_t i;

    for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) {
        if (group->backends[i] &&
            group->backends[i]->bindMount(group, oldroot, mountopts) < 0) {
            return -1;
        }
    }

    return 0;
2739
}
2740 2741


2742 2743 2744 2745 2746
int virCgroupSetOwner(virCgroupPtr cgroup,
                      uid_t uid,
                      gid_t gid,
                      int controllers)
{
2747 2748 2749 2750 2751 2752 2753 2754 2755 2756
    size_t i;

    for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) {
        if (cgroup->backends[i] &&
            cgroup->backends[i]->setOwner(cgroup, uid, gid, controllers) < 0) {
            return -1;
        }
    }

    return 0;
2757 2758 2759
}


2760 2761 2762 2763 2764 2765 2766 2767 2768 2769
/**
 * virCgroupSupportsCpuBW():
 * Check whether the host supports CFS bandwidth.
 *
 * Return true when CFS bandwidth is supported,
 * false when CFS bandwidth is not supported.
 */
bool
virCgroupSupportsCpuBW(virCgroupPtr cgroup)
{
2770 2771
    VIR_CGROUP_BACKEND_CALL(cgroup, VIR_CGROUP_CONTROLLER_CPU,
                            supportsCpuBW, false);
2772 2773
}

2774 2775 2776
int
virCgroupHasEmptyTasks(virCgroupPtr cgroup, int controller)
{
2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787
    size_t i;

    for (i = 0; i < VIR_CGROUP_BACKEND_TYPE_LAST; i++) {
        if (cgroup->backends[i]) {
            int rc = cgroup->backends[i]->hasEmptyTasks(cgroup, controller);
            if (rc <= 0)
                return rc;
        }
    }

    return 1;
2788
}
2789

2790 2791 2792
bool
virCgroupControllerAvailable(int controller)
{
2793 2794
    virCgroupPtr cgroup;
    bool ret = false;
2795 2796

    if (virCgroupNewSelf(&cgroup) < 0)
2797
        return ret;
2798

2799
    ret = virCgroupHasController(cgroup, controller);
2800
    virCgroupFree(&cgroup);
2801
    return ret;
2802 2803
}

2804
#else /* !__linux__ */
2805

2806 2807 2808 2809 2810 2811 2812
bool
virCgroupAvailable(void)
{
    return false;
}


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
int
virCgroupNewPartition(const char *path ATTRIBUTE_UNUSED,
                      bool create ATTRIBUTE_UNUSED,
                      int controllers ATTRIBUTE_UNUSED,
                      virCgroupPtr *group ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENXIO, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupNewSelf(virCgroupPtr *group ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENXIO, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupNewDomainPartition(virCgroupPtr partition ATTRIBUTE_UNUSED,
                            const char *driver ATTRIBUTE_UNUSED,
                            const char *name ATTRIBUTE_UNUSED,
                            bool create ATTRIBUTE_UNUSED,
                            virCgroupPtr *group ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENXIO, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859
int
virCgroupNewThread(virCgroupPtr domain ATTRIBUTE_UNUSED,
                   virCgroupThreadName nameval ATTRIBUTE_UNUSED,
                   int id ATTRIBUTE_UNUSED,
                   bool create ATTRIBUTE_UNUSED,
                   virCgroupPtr *group ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENXIO, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870
int
virCgroupNewDetect(pid_t pid ATTRIBUTE_UNUSED,
                   int controllers ATTRIBUTE_UNUSED,
                   virCgroupPtr *group ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENXIO, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


2871 2872 2873 2874 2875
int
virCgroupNewDetectMachine(const char *name ATTRIBUTE_UNUSED,
                          const char *drivername ATTRIBUTE_UNUSED,
                          pid_t pid ATTRIBUTE_UNUSED,
                          int controllers ATTRIBUTE_UNUSED,
2876
                          char *machinename ATTRIBUTE_UNUSED,
2877 2878 2879 2880 2881 2882 2883
                          virCgroupPtr *group ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENXIO, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}

J
Ján Tomko 已提交
2884

2885
int virCgroupTerminateMachine(const char *name ATTRIBUTE_UNUSED)
J
Ján Tomko 已提交
2886 2887 2888 2889 2890 2891 2892
{
    virReportSystemError(ENXIO, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


2893 2894 2895 2896 2897 2898 2899
int
virCgroupNewMachine(const char *name ATTRIBUTE_UNUSED,
                    const char *drivername ATTRIBUTE_UNUSED,
                    const unsigned char *uuid ATTRIBUTE_UNUSED,
                    const char *rootdir ATTRIBUTE_UNUSED,
                    pid_t pidleader ATTRIBUTE_UNUSED,
                    bool isContainer ATTRIBUTE_UNUSED,
2900 2901
                    size_t nnicindexes ATTRIBUTE_UNUSED,
                    int *nicindexes ATTRIBUTE_UNUSED,
2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918
                    const char *partition ATTRIBUTE_UNUSED,
                    int controllers ATTRIBUTE_UNUSED,
                    virCgroupPtr *group ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENXIO, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


bool
virCgroupNewIgnoreError(void)
{
    VIR_DEBUG("No cgroups present/configured/accessible, ignoring error");
    return true;
}

2919 2920

void
2921
virCgroupFree(virCgroupPtr *group ATTRIBUTE_UNUSED)
2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935
{
    virReportSystemError(ENXIO, "%s",
                         _("Control groups not supported on this platform"));
}


bool
virCgroupHasController(virCgroupPtr cgroup ATTRIBUTE_UNUSED,
                       int controller ATTRIBUTE_UNUSED)
{
    return false;
}


2936 2937
int
virCgroupPathOfController(virCgroupPtr group ATTRIBUTE_UNUSED,
2938
                          unsigned int controller ATTRIBUTE_UNUSED,
2939 2940 2941 2942 2943 2944 2945 2946 2947
                          const char *key ATTRIBUTE_UNUSED,
                          char **path ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENXIO, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


2948
int
2949 2950
virCgroupAddProcess(virCgroupPtr group ATTRIBUTE_UNUSED,
                    pid_t pid ATTRIBUTE_UNUSED)
2951 2952 2953 2954 2955 2956 2957
{
    virReportSystemError(ENXIO, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


2958
int
2959 2960
virCgroupAddMachineProcess(virCgroupPtr group ATTRIBUTE_UNUSED,
                           pid_t pid ATTRIBUTE_UNUSED)
2961 2962 2963 2964 2965 2966 2967
{
    virReportSystemError(ENXIO, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


2968 2969 2970 2971 2972 2973 2974 2975 2976 2977
int
virCgroupAddThread(virCgroupPtr group ATTRIBUTE_UNUSED,
                   pid_t pid ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENXIO, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004
int
virCgroupGetBlkioIoServiced(virCgroupPtr group ATTRIBUTE_UNUSED,
                            long long *bytes_read ATTRIBUTE_UNUSED,
                            long long *bytes_write ATTRIBUTE_UNUSED,
                            long long *requests_read ATTRIBUTE_UNUSED,
                            long long *requests_write ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENXIO, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupGetBlkioIoDeviceServiced(virCgroupPtr group ATTRIBUTE_UNUSED,
                                  const char *path ATTRIBUTE_UNUSED,
                                  long long *bytes_read ATTRIBUTE_UNUSED,
                                  long long *bytes_write ATTRIBUTE_UNUSED,
                                  long long *requests_read ATTRIBUTE_UNUSED,
                                  long long *requests_write ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENXIO, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024
int
virCgroupSetBlkioWeight(virCgroupPtr group ATTRIBUTE_UNUSED,
                        unsigned int weight ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENXIO, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupGetBlkioWeight(virCgroupPtr group ATTRIBUTE_UNUSED,
                        unsigned int *weight ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENXIO, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


3025 3026 3027 3028 3029 3030 3031 3032 3033 3034
int
virCgroupSetBlkioDeviceWeight(virCgroupPtr group ATTRIBUTE_UNUSED,
                              const char *path ATTRIBUTE_UNUSED,
                              unsigned int weight ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}

3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073 3074
int
virCgroupSetBlkioDeviceReadIops(virCgroupPtr group ATTRIBUTE_UNUSED,
                                const char *path ATTRIBUTE_UNUSED,
                                unsigned int riops ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}

int
virCgroupSetBlkioDeviceWriteIops(virCgroupPtr group ATTRIBUTE_UNUSED,
                                 const char *path ATTRIBUTE_UNUSED,
                                 unsigned int wiops ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}

int
virCgroupSetBlkioDeviceReadBps(virCgroupPtr group ATTRIBUTE_UNUSED,
                               const char *path ATTRIBUTE_UNUSED,
                               unsigned long long rbps ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}

int
virCgroupSetBlkioDeviceWriteBps(virCgroupPtr group ATTRIBUTE_UNUSED,
                                const char *path ATTRIBUTE_UNUSED,
                                unsigned long long wbps ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}

3075 3076 3077
int
virCgroupGetBlkioDeviceWeight(virCgroupPtr group ATTRIBUTE_UNUSED,
                              const char *path ATTRIBUTE_UNUSED,
3078
                              unsigned int *weight ATTRIBUTE_UNUSED)
3079 3080 3081 3082 3083 3084 3085 3086 3087
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}

int
virCgroupGetBlkioDeviceReadIops(virCgroupPtr group ATTRIBUTE_UNUSED,
                                const char *path ATTRIBUTE_UNUSED,
3088
                                unsigned int *riops ATTRIBUTE_UNUSED)
3089 3090 3091 3092 3093 3094 3095 3096 3097
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}

int
virCgroupGetBlkioDeviceWriteIops(virCgroupPtr group ATTRIBUTE_UNUSED,
                                 const char *path ATTRIBUTE_UNUSED,
3098
                                 unsigned int *wiops ATTRIBUTE_UNUSED)
3099 3100 3101 3102 3103 3104 3105 3106 3107
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}

int
virCgroupGetBlkioDeviceReadBps(virCgroupPtr group ATTRIBUTE_UNUSED,
                               const char *path ATTRIBUTE_UNUSED,
3108
                               unsigned long long *rbps ATTRIBUTE_UNUSED)
3109 3110 3111 3112 3113 3114 3115 3116 3117
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}

int
virCgroupGetBlkioDeviceWriteBps(virCgroupPtr group ATTRIBUTE_UNUSED,
                                const char *path ATTRIBUTE_UNUSED,
3118
                                unsigned long long *wbps ATTRIBUTE_UNUSED)
3119 3120 3121 3122 3123
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}
3124

3125 3126 3127 3128 3129 3130 3131 3132 3133 3134
int
virCgroupSetMemory(virCgroupPtr group ATTRIBUTE_UNUSED,
                   unsigned long long kb ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


P
Pavel Hrdina 已提交
3135 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149
int
virCgroupGetMemoryStat(virCgroupPtr group ATTRIBUTE_UNUSED,
                       unsigned long long *cache ATTRIBUTE_UNUSED,
                       unsigned long long *activeAnon ATTRIBUTE_UNUSED,
                       unsigned long long *inactiveAnon ATTRIBUTE_UNUSED,
                       unsigned long long *activeFile ATTRIBUTE_UNUSED,
                       unsigned long long *inactiveFile ATTRIBUTE_UNUSED,
                       unsigned long long *unevictable ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 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
int
virCgroupGetMemoryUsage(virCgroupPtr group ATTRIBUTE_UNUSED,
                        unsigned long *kb ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupSetMemoryHardLimit(virCgroupPtr group ATTRIBUTE_UNUSED,
                            unsigned long long kb ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupGetMemoryHardLimit(virCgroupPtr group ATTRIBUTE_UNUSED,
                            unsigned long long *kb ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupSetMemorySoftLimit(virCgroupPtr group ATTRIBUTE_UNUSED,
                            unsigned long long kb ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupGetMemorySoftLimit(virCgroupPtr group ATTRIBUTE_UNUSED,
                            unsigned long long *kb ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupSetMemSwapHardLimit(virCgroupPtr group ATTRIBUTE_UNUSED,
                             unsigned long long kb ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupGetMemSwapHardLimit(virCgroupPtr group ATTRIBUTE_UNUSED,
                             unsigned long long *kb ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupGetMemSwapUsage(virCgroupPtr group ATTRIBUTE_UNUSED,
                         unsigned long long *kb ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupSetCpusetMems(virCgroupPtr group ATTRIBUTE_UNUSED,
                       const char *mems ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupGetCpusetMems(virCgroupPtr group ATTRIBUTE_UNUSED,
                       char **mems ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}

3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268

int
virCgroupSetCpusetMemoryMigrate(virCgroupPtr group ATTRIBUTE_UNUSED,
                                bool migrate ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupGetCpusetMemoryMigrate(virCgroupPtr group ATTRIBUTE_UNUSED,
                                bool *migrate ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}

3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288

int
virCgroupSetCpusetCpus(virCgroupPtr group ATTRIBUTE_UNUSED,
                       const char *cpus ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupGetCpusetCpus(virCgroupPtr group ATTRIBUTE_UNUSED,
                       char **cpus ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}

3289 3290 3291 3292 3293 3294 3295 3296
int
virCgroupAllowAllDevices(virCgroupPtr group ATTRIBUTE_UNUSED,
                         int perms ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}
3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319

int
virCgroupDenyAllDevices(virCgroupPtr group ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupAllowDevice(virCgroupPtr group ATTRIBUTE_UNUSED,
                     char type ATTRIBUTE_UNUSED,
                     int major ATTRIBUTE_UNUSED,
                     int minor ATTRIBUTE_UNUSED,
                     int perms ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


3320 3321 3322
int
virCgroupAllowDevicePath(virCgroupPtr group ATTRIBUTE_UNUSED,
                         const char *path ATTRIBUTE_UNUSED,
3323 3324
                         int perms ATTRIBUTE_UNUSED,
                         bool ignoreEaccess ATTRIBUTE_UNUSED)
3325 3326 3327 3328 3329 3330 3331
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344
int
virCgroupDenyDevice(virCgroupPtr group ATTRIBUTE_UNUSED,
                    char type ATTRIBUTE_UNUSED,
                    int major ATTRIBUTE_UNUSED,
                    int minor ATTRIBUTE_UNUSED,
                    int perms ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


3345 3346 3347
int
virCgroupDenyDevicePath(virCgroupPtr group ATTRIBUTE_UNUSED,
                        const char *path ATTRIBUTE_UNUSED,
3348 3349
                        int perms ATTRIBUTE_UNUSED,
                        bool ignoreEacces ATTRIBUTE_UNUSED)
3350 3351 3352 3353 3354 3355 3356
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415
int
virCgroupSetCpuShares(virCgroupPtr group ATTRIBUTE_UNUSED,
                      unsigned long long shares ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupGetCpuShares(virCgroupPtr group ATTRIBUTE_UNUSED,
                      unsigned long long *shares ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupSetCpuCfsPeriod(virCgroupPtr group ATTRIBUTE_UNUSED,
                         unsigned long long cfs_period ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupGetCpuCfsPeriod(virCgroupPtr group ATTRIBUTE_UNUSED,
                         unsigned long long *cfs_period ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupSetCpuCfsQuota(virCgroupPtr group ATTRIBUTE_UNUSED,
                        long long cfs_quota ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupRemove(virCgroupPtr group ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENXIO, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


3416 3417 3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433 3434
int
virCgroupKillRecursive(virCgroupPtr group ATTRIBUTE_UNUSED,
                       int signum ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupKillPainfully(virCgroupPtr group ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475
int
virCgroupGetCpuCfsQuota(virCgroupPtr group ATTRIBUTE_UNUSED,
                        long long *cfs_quota ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupGetCpuacctUsage(virCgroupPtr group ATTRIBUTE_UNUSED,
                         unsigned long long *usage ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupGetCpuacctPercpuUsage(virCgroupPtr group ATTRIBUTE_UNUSED,
                               char **usage ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupGetCpuacctStat(virCgroupPtr group ATTRIBUTE_UNUSED,
                        unsigned long long *user ATTRIBUTE_UNUSED,
                        unsigned long long *sys ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486
int
virCgroupGetDomainTotalCpuStats(virCgroupPtr group ATTRIBUTE_UNUSED,
                                virTypedParameterPtr params ATTRIBUTE_UNUSED,
                                int nparams ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506
int
virCgroupSetFreezerState(virCgroupPtr group ATTRIBUTE_UNUSED,
                         const char *state ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupGetFreezerState(virCgroupPtr group ATTRIBUTE_UNUSED,
                         char **state ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


E
Eric Blake 已提交
3507
int
3508 3509 3510
virCgroupBindMount(virCgroupPtr group ATTRIBUTE_UNUSED,
                   const char *oldroot ATTRIBUTE_UNUSED,
                   const char *mountopts ATTRIBUTE_UNUSED)
3511
{
3512 3513 3514
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
3515
}
3516

3517 3518 3519 3520 3521 3522 3523 3524

bool
virCgroupSupportsCpuBW(virCgroupPtr cgroup ATTRIBUTE_UNUSED)
{
    VIR_DEBUG("Control groups not supported on this platform");
    return false;
}

E
Eric Blake 已提交
3525 3526 3527 3528 3529 3530

int
virCgroupGetPercpuStats(virCgroupPtr group ATTRIBUTE_UNUSED,
                        virTypedParameterPtr params ATTRIBUTE_UNUSED,
                        unsigned int nparams ATTRIBUTE_UNUSED,
                        int start_cpu ATTRIBUTE_UNUSED,
J
Ján Tomko 已提交
3531
                        unsigned int ncpus ATTRIBUTE_UNUSED,
3532
                        virBitmapPtr guestvcpus ATTRIBUTE_UNUSED)
E
Eric Blake 已提交
3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


int
virCgroupSetOwner(virCgroupPtr cgroup ATTRIBUTE_UNUSED,
                  uid_t uid ATTRIBUTE_UNUSED,
                  gid_t gid ATTRIBUTE_UNUSED,
                  int controllers ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}

3551 3552 3553 3554 3555 3556 3557 3558 3559
int
virCgroupHasEmptyTasks(virCgroupPtr cgroup ATTRIBUTE_UNUSED,
                       int controller ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}

3560 3561 3562 3563 3564
bool
virCgroupControllerAvailable(int controller ATTRIBUTE_UNUSED)
{
    return false;
}
3565
#endif /* !__linux__ */
3566 3567 3568 3569 3570 3571 3572


int
virCgroupDelThread(virCgroupPtr cgroup,
                   virCgroupThreadName nameval,
                   int idx)
{
3573
    virCgroupPtr new_cgroup = NULL;
3574 3575 3576 3577 3578 3579 3580

    if (cgroup) {
        if (virCgroupNewThread(cgroup, nameval, idx, false, &new_cgroup) < 0)
            return -1;

        /* Remove the offlined cgroup */
        virCgroupRemove(new_cgroup);
3581
        virCgroupFree(&new_cgroup);
3582 3583 3584 3585
    }

    return 0;
}