vircgroup.c 121.8 KB
Newer Older
1
/*
2
 * vircgroup.c: methods for managing control cgroups
3
 *
E
Eric Blake 已提交
4
 * Copyright (C) 2010-2014 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 23 24 25 26
 *
 * Authors:
 *  Dan Smith <danms@us.ibm.com>
 */
#include <config.h>

#include <stdio.h>
27
#if defined HAVE_MNTENT_H && defined HAVE_GETMNTENT_R
28
# include <mntent.h>
D
Daniel P. Berrange 已提交
29
#endif
30 31 32
#if defined HAVE_SYS_MOUNT_H
# include <sys/mount.h>
#endif
33 34 35 36 37 38
#include <fcntl.h>
#include <string.h>
#include <errno.h>
#include <stdlib.h>
#include <sys/stat.h>
#include <sys/types.h>
39
#include <signal.h>
40
#include <dirent.h>
41

42 43 44
#define __VIR_CGROUP_ALLOW_INCLUDE_PRIV_H__
#include "vircgrouppriv.h"

45
#include "virutil.h"
46
#include "viralloc.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

56 57
#include "nodeinfo.h"

58 59
VIR_LOG_INIT("util.cgroup");

60 61
#define CGROUP_MAX_VAL 512

62 63
#define VIR_FROM_THIS VIR_FROM_CGROUP

64
#define CGROUP_NB_TOTAL_CPU_STAT_PARAM 3
65
#define CGROUP_NB_PER_CPU_STAT_PARAM   1
66

67
#if defined(__linux__) && defined(HAVE_GETMNTENT_R) && \
68
    defined(_DIRENT_HAVE_D_TYPE) && defined(_SC_CLK_TCK)
69 70 71
# define VIR_CGROUP_SUPPORTED
#endif

72
VIR_ENUM_IMPL(virCgroupController, VIR_CGROUP_CONTROLLER_LAST,
R
Ryota Ozaki 已提交
73
              "cpu", "cpuacct", "cpuset", "memory", "devices",
74 75
              "freezer", "blkio", "net_cls", "perf_event",
              "name=systemd");
76

77 78 79 80 81 82 83 84
typedef enum {
    VIR_CGROUP_NONE = 0, /* create subdir under each cgroup if possible. */
    VIR_CGROUP_MEM_HIERACHY = 1 << 0, /* call virCgroupSetMemoryUseHierarchy
                                       * before creating subcgroups and
                                       * attaching tasks
                                       */
} virCgroupFlags;

E
Eric Blake 已提交
85

86
#ifdef VIR_CGROUP_SUPPORTED
E
Eric Blake 已提交
87 88
bool
virCgroupAvailable(void)
89
{
90
    bool ret = false;
91 92 93 94 95 96 97 98 99 100 101
    FILE *mounts = NULL;
    struct mntent entry;
    char buf[CGROUP_MAX_VAL];

    if (!virFileExists("/proc/cgroups"))
        return false;

    if (!(mounts = fopen("/proc/mounts", "r")))
        return false;

    while (getmntent_r(mounts, &entry, buf, sizeof(buf)) != NULL) {
102 103 104 105
        /* We're looking for at least one 'cgroup' fs mount,
         * which is *not* a named mount. */
        if (STREQ(entry.mnt_type, "cgroup") &&
            !strstr(entry.mnt_opts, "name=")) {
106 107 108 109 110 111 112 113 114
            ret = true;
            break;
        }
    }

    VIR_FORCE_FCLOSE(mounts);
    return ret;
}

E
Eric Blake 已提交
115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179

static int
virCgroupPartitionNeedsEscaping(const char *path)
{
    FILE *fp = NULL;
    int ret = 0;
    char *line = NULL;
    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;
    }

180
 cleanup:
E
Eric Blake 已提交
181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201
    VIR_FREE(line);
    VIR_FORCE_FCLOSE(fp);
    return ret;
}


static int
virCgroupPartitionEscape(char **path)
{
    size_t len = strlen(*path) + 1;
    int rc;
    char escape = '_';

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

    if (VIR_INSERT_ELEMENT(*path, 0, len, escape) < 0)
        return -1;

    return 0;
}
E
Eric Blake 已提交
202 203


204
static bool
205 206 207
virCgroupValidateMachineGroup(virCgroupPtr group,
                              const char *name,
                              const char *drivername,
208
                              const char *partition,
209
                              bool stripEmulatorSuffix)
210 211 212 213
{
    size_t i;
    bool valid = false;
    char *partname;
214
    char *scopename;
215 216 217 218 219 220 221 222

    if (virAsprintf(&partname, "%s.libvirt-%s",
                    name, drivername) < 0)
        goto cleanup;

    if (virCgroupPartitionEscape(&partname) < 0)
        goto cleanup;

223 224 225 226 227 228 229 230 231
    if (!partition)
        partition = "/machine";

    if (!(scopename = virSystemdMakeScopeName(name, drivername, partition)))
        goto cleanup;

    if (virCgroupPartitionEscape(&scopename) < 0)
        goto cleanup;

232 233 234
    for (i = 0; i < VIR_CGROUP_CONTROLLER_LAST; i++) {
        char *tmp;

235 236 237
        if (i == VIR_CGROUP_CONTROLLER_SYSTEMD)
            continue;

238 239 240 241 242 243
        if (!group->controllers[i].placement)
            continue;

        tmp = strrchr(group->controllers[i].placement, '/');
        if (!tmp)
            goto cleanup;
244 245 246 247 248 249 250 251 252 253 254 255

        if (stripEmulatorSuffix &&
            (i == VIR_CGROUP_CONTROLLER_CPU ||
             i == VIR_CGROUP_CONTROLLER_CPUACCT ||
             i == VIR_CGROUP_CONTROLLER_CPUSET)) {
            if (STREQ(tmp, "/emulator"))
                *tmp = '\0';
            tmp = strrchr(group->controllers[i].placement, '/');
            if (!tmp)
                goto cleanup;
        }

256 257 258
        tmp++;

        if (STRNEQ(tmp, name) &&
259 260
            STRNEQ(tmp, partname) &&
            STRNEQ(tmp, scopename)) {
E
Eric Blake 已提交
261 262 263 264
            VIR_DEBUG("Name '%s' for controller '%s' does not match "
                      "'%s', '%s' or '%s'",
                      tmp, virCgroupControllerTypeToString(i),
                      name, partname, scopename);
265
            goto cleanup;
266
        }
267 268 269 270 271 272
    }

    valid = true;

 cleanup:
    VIR_FREE(partname);
273
    VIR_FREE(scopename);
274 275
    return valid;
}
E
Eric Blake 已提交
276

L
Lai Jiangshan 已提交
277

E
Eric Blake 已提交
278 279 280
static int
virCgroupCopyMounts(virCgroupPtr group,
                    virCgroupPtr parent)
281
{
282
    size_t i;
283
    for (i = 0; i < VIR_CGROUP_CONTROLLER_LAST; i++) {
284 285 286
        if (!parent->controllers[i].mountPoint)
            continue;

287 288 289
        if (VIR_STRDUP(group->controllers[i].mountPoint,
                       parent->controllers[i].mountPoint) < 0)
            return -1;
290

291 292 293
        if (VIR_STRDUP(group->controllers[i].linkPoint,
                       parent->controllers[i].linkPoint) < 0)
            return -1;
294 295 296 297
    }
    return 0;
}

E
Eric Blake 已提交
298

299 300 301 302
/*
 * Process /proc/mounts figuring out what controllers are
 * mounted and where
 */
E
Eric Blake 已提交
303 304
static int
virCgroupDetectMounts(virCgroupPtr group)
305
{
306
    size_t i;
307
    FILE *mounts = NULL;
308 309 310 311 312
    struct mntent entry;
    char buf[CGROUP_MAX_VAL];

    mounts = fopen("/proc/mounts", "r");
    if (mounts == NULL) {
313 314 315
        virReportSystemError(errno, "%s",
                             _("Unable to open /proc/mounts"));
        return -1;
316 317 318
    }

    while (getmntent_r(mounts, &entry, buf, sizeof(buf)) != NULL) {
319 320
        if (STRNEQ(entry.mnt_type, "cgroup"))
            continue;
321

322
        for (i = 0; i < VIR_CGROUP_CONTROLLER_LAST; i++) {
323 324 325 326 327 328 329 330 331 332 333 334
            const char *typestr = virCgroupControllerTypeToString(i);
            int typelen = strlen(typestr);
            char *tmp = entry.mnt_opts;
            while (tmp) {
                char *next = strchr(tmp, ',');
                int len;
                if (next) {
                    len = next-tmp;
                    next++;
                } else {
                    len = strlen(tmp);
                }
335 336 337 338
                /* NB, the same controller can appear >1 time in mount list
                 * due to bind mounts from one location to another. Pick the
                 * first entry only
                 */
339
                if (typelen == len && STREQLEN(typestr, tmp, len) &&
340 341 342 343 344
                    !group->controllers[i].mountPoint) {
                    char *linksrc;
                    struct stat sb;
                    char *tmp2;

345 346
                    if (VIR_STRDUP(group->controllers[i].mountPoint,
                                   entry.mnt_dir) < 0)
347
                        goto error;
348 349 350

                    tmp2 = strrchr(entry.mnt_dir, '/');
                    if (!tmp2) {
351 352 353
                        virReportError(VIR_ERR_INTERNAL_ERROR,
                                       _("Missing '/' separator in cgroup mount '%s'"),
                                       entry.mnt_dir);
354 355
                        goto error;
                    }
356

357 358 359
                    /* If it is a co-mount it has a filename like "cpu,cpuacct"
                     * and we must identify the symlink path */
                    if (strchr(tmp2 + 1, ',')) {
360
                        *tmp2 = '\0';
361 362
                        if (virAsprintf(&linksrc, "%s/%s",
                                        entry.mnt_dir, typestr) < 0)
363
                            goto error;
364 365 366 367 368 369 370 371
                        *tmp2 = '/';

                        if (lstat(linksrc, &sb) < 0) {
                            if (errno == ENOENT) {
                                VIR_WARN("Controller %s co-mounted at %s is missing symlink at %s",
                                         typestr, entry.mnt_dir, linksrc);
                                VIR_FREE(linksrc);
                            } else {
372
                                virReportSystemError(errno,
E
Eric Blake 已提交
373 374
                                                     _("Cannot stat %s"),
                                                     linksrc);
375 376 377 378 379 380 381 382 383 384 385 386
                                goto error;
                            }
                        } else {
                            if (!S_ISLNK(sb.st_mode)) {
                                VIR_WARN("Expecting a symlink at %s for controller %s",
                                         linksrc, typestr);
                            } else {
                                group->controllers[i].linkPoint = linksrc;
                            }
                        }
                    }
                }
387 388 389
                tmp = next;
            }
        }
390 391
    }

392
    VIR_FORCE_FCLOSE(mounts);
393

394
    return 0;
395

396
 error:
397
    VIR_FORCE_FCLOSE(mounts);
398
    return -1;
399 400
}

401

E
Eric Blake 已提交
402 403 404 405
static int
virCgroupCopyPlacement(virCgroupPtr group,
                       const char *path,
                       virCgroupPtr parent)
406
{
407
    size_t i;
408
    for (i = 0; i < VIR_CGROUP_CONTROLLER_LAST; i++) {
409 410 411
        if (!group->controllers[i].mountPoint)
            continue;

412 413 414
        if (i == VIR_CGROUP_CONTROLLER_SYSTEMD)
            continue;

415
        if (path[0] == '/') {
416 417
            if (VIR_STRDUP(group->controllers[i].placement, path) < 0)
                return -1;
418 419
        } else {
            /*
420 421 422
             * parent == "/" + path="" => "/"
             * parent == "/libvirt.service" + path == "" => "/libvirt.service"
             * parent == "/libvirt.service" + path == "foo" => "/libvirt.service/foo"
423 424 425 426 427 428 429
             */
            if (virAsprintf(&group->controllers[i].placement,
                            "%s%s%s",
                            parent->controllers[i].placement,
                            (STREQ(parent->controllers[i].placement, "/") ||
                             STREQ(path, "") ? "" : "/"),
                            path) < 0)
430
                return -1;
431 432 433 434 435 436 437
        }
    }

    return 0;
}


438
/*
439 440 441 442
 * virCgroupDetectPlacement:
 * @group: the group to process
 * @path: the relative path to append, not starting with '/'
 *
443 444
 * Process /proc/self/cgroup figuring out what cgroup
 * sub-path the current process is assigned to. ie not
445 446 447 448 449 450 451 452 453 454 455 456 457 458
 * 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.
459
 */
E
Eric Blake 已提交
460 461 462 463
static int
virCgroupDetectPlacement(virCgroupPtr group,
                         pid_t pid,
                         const char *path)
464
{
465
    size_t i;
466 467
    FILE *mapping  = NULL;
    char line[1024];
468
    int ret = -1;
469
    char *procfile;
470

471 472
    VIR_DEBUG("Detecting placement for pid %lld path %s",
              (unsigned long long)pid, path);
473 474 475 476 477 478 479 480 481 482
    if (pid == -1) {
        if (VIR_STRDUP(procfile, "/proc/self/cgroup") < 0)
            goto cleanup;
    } else {
        if (virAsprintf(&procfile, "/proc/%llu/cgroup",
                        (unsigned long long)pid) < 0)
            goto cleanup;
    }

    mapping = fopen(procfile, "r");
483
    if (mapping == NULL) {
484 485 486 487
        virReportSystemError(errno,
                             _("Unable to open '%s'"),
                             procfile);
        goto cleanup;
488 489
    }

490 491
    while (fgets(line, sizeof(line), mapping) != NULL) {
        char *controllers = strchr(line, ':');
492 493
        char *selfpath = controllers ? strchr(controllers + 1, ':') : NULL;
        char *nl = selfpath ? strchr(selfpath, '\n') : NULL;
494

495
        if (!controllers || !selfpath)
496 497 498 499 500
            continue;

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

501
        *selfpath = '\0';
502
        controllers++;
503
        selfpath++;
504

505
        for (i = 0; i < VIR_CGROUP_CONTROLLER_LAST; i++) {
506 507 508
            const char *typestr = virCgroupControllerTypeToString(i);
            int typelen = strlen(typestr);
            char *tmp = controllers;
509

510 511 512 513
            while (tmp) {
                char *next = strchr(tmp, ',');
                int len;
                if (next) {
514
                    len = next - tmp;
515 516 517 518
                    next++;
                } else {
                    len = strlen(tmp);
                }
519 520

                /*
521 522 523
                 * selfpath == "/" + path="" -> "/"
                 * selfpath == "/libvirt.service" + path == "" -> "/libvirt.service"
                 * selfpath == "/libvirt.service" + path == "foo" -> "/libvirt.service/foo"
524
                 */
525
                if (typelen == len && STREQLEN(typestr, tmp, len) &&
526 527 528 529 530 531 532 533 534 535 536 537 538 539
                    group->controllers[i].mountPoint != NULL &&
                    group->controllers[i].placement == NULL) {
                    if (i == VIR_CGROUP_CONTROLLER_SYSTEMD) {
                        if (VIR_STRDUP(group->controllers[i].placement,
                                       selfpath) < 0)
                            goto cleanup;
                    } else {
                        if (virAsprintf(&group->controllers[i].placement,
                                        "%s%s%s", selfpath,
                                        (STREQ(selfpath, "/") ||
                                         STREQ(path, "") ? "" : "/"),
                                        path) < 0)
                            goto cleanup;
                    }
540
                }
541 542 543 544 545 546

                tmp = next;
            }
        }
    }

547
    ret = 0;
548

549
 cleanup:
550
    VIR_FREE(procfile);
551
    VIR_FORCE_FCLOSE(mapping);
552

553
    return ret;
554 555
}

E
Eric Blake 已提交
556 557 558 559 560 561 562

static int
virCgroupDetect(virCgroupPtr group,
                pid_t pid,
                int controllers,
                const char *path,
                virCgroupPtr parent)
563
{
564 565
    size_t i;
    size_t j;
566 567
    VIR_DEBUG("group=%p controllers=%d path=%s parent=%p",
              group, controllers, path, parent);
568

569 570 571 572 573 574
    if (parent) {
        if (virCgroupCopyMounts(group, parent) < 0)
            return -1;
    } else {
        if (virCgroupDetectMounts(group) < 0)
            return -1;
575 576
    }

577
    if (controllers >= 0) {
578
        VIR_DEBUG("Filtering controllers %d", controllers);
579
        for (i = 0; i < VIR_CGROUP_CONTROLLER_LAST; i++) {
580
            VIR_DEBUG("Controller '%s' wanted=%s, mount='%s'",
581
                      virCgroupControllerTypeToString(i),
582 583
                      (1 << i) & controllers ? "yes" : "no",
                      NULLSTR(group->controllers[i].mountPoint));
584
            if (((1 << i) & controllers)) {
585
                /* Remove non-existent controllers  */
586
                if (!group->controllers[i].mountPoint) {
587
                    VIR_DEBUG("Requested controller '%s' not mounted, ignoring",
588
                              virCgroupControllerTypeToString(i));
589
                    controllers &= ~(1 << i);
590 591 592 593
                }
            } else {
                /* Check whether a request to disable a controller
                 * clashes with co-mounting of controllers */
594
                for (j = 0; j < VIR_CGROUP_CONTROLLER_LAST; j++) {
595 596 597 598 599 600 601
                    if (j == i)
                        continue;
                    if (!((1 << j) & controllers))
                        continue;

                    if (STREQ_NULLABLE(group->controllers[i].mountPoint,
                                       group->controllers[j].mountPoint)) {
602 603 604 605 606
                        virReportSystemError(EINVAL,
                                             _("Controller '%s' is not wanted, but '%s' is co-mounted"),
                                             virCgroupControllerTypeToString(i),
                                             virCgroupControllerTypeToString(j));
                        return -1;
607 608 609 610 611 612 613 614
                    }
                }
                VIR_FREE(group->controllers[i].mountPoint);
            }
        }
    } else {
        VIR_DEBUG("Auto-detecting controllers");
        controllers = 0;
615
        for (i = 0; i < VIR_CGROUP_CONTROLLER_LAST; i++) {
616 617 618 619 620 621 622
            VIR_DEBUG("Controller '%s' present=%s",
                      virCgroupControllerTypeToString(i),
                      group->controllers[i].mountPoint ? "yes" : "no");
            if (group->controllers[i].mountPoint == NULL)
                continue;
            controllers |= (1 << i);
        }
623
    }
624

625
    /* Check that at least 1 controller is available */
626
    if (!controllers) {
627 628 629
        virReportSystemError(ENXIO, "%s",
                             _("At least one cgroup controller is required"));
        return -1;
630
    }
631

632 633 634 635 636 637 638 639 640 641
    /* In some cases we can copy part of the placement info
     * based on the parent cgroup...
     */
    if ((parent || path[0] == '/') &&
        virCgroupCopyPlacement(group, path, parent) < 0)
        return -1;

    /* ... but use /proc/cgroups to fill in the rest */
    if (virCgroupDetectPlacement(group, pid, path) < 0)
        return -1;
642

643 644 645 646
    /* Check that for every mounted controller, we found our placement */
    for (i = 0; i < VIR_CGROUP_CONTROLLER_LAST; i++) {
        if (!group->controllers[i].mountPoint)
            continue;
647

648 649 650 651 652 653
        if (!group->controllers[i].placement) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Could not find placement for controller %s at %s"),
                           virCgroupControllerTypeToString(i),
                           group->controllers[i].placement);
            return -1;
654
        }
655

656
        VIR_DEBUG("Detected mount/mapping %zu:%s at %s in %s for pid %llu", i,
657 658
                  virCgroupControllerTypeToString(i),
                  group->controllers[i].mountPoint,
659 660
                  group->controllers[i].placement,
                  (unsigned long long)pid);
661 662
    }

663
    return 0;
664 665
}

666

E
Eric Blake 已提交
667 668 669 670 671
static int
virCgroupSetValueStr(virCgroupPtr group,
                     int controller,
                     const char *key,
                     const char *value)
672
{
673
    int ret = -1;
674
    char *keypath = NULL;
675
    char *tmp = NULL;
676

677 678
    if (virCgroupPathOfController(group, controller, key, &keypath) < 0)
        return -1;
679

680
    VIR_DEBUG("Set value '%s' to '%s'", keypath, value);
681
    if (virFileWriteStr(keypath, value, 0) < 0) {
682 683 684 685 686 687 688
        if (errno == EINVAL &&
            (tmp = strrchr(keypath, '/'))) {
            virReportSystemError(errno,
                                 _("Invalid value '%s' for '%s'"),
                                 value, tmp + 1);
            goto cleanup;
        }
689 690 691
        virReportSystemError(errno,
                             _("Unable to write to '%s'"), keypath);
        goto cleanup;
692 693
    }

694
    ret = 0;
695

696
 cleanup:
697 698
    VIR_FREE(keypath);
    return ret;
699 700
}

E
Eric Blake 已提交
701 702 703 704 705 706

static int
virCgroupGetValueStr(virCgroupPtr group,
                     int controller,
                     const char *key,
                     char **value)
707 708
{
    char *keypath = NULL;
709
    int ret = -1, rc;
710

711
    *value = NULL;
712

713 714
    if (virCgroupPathOfController(group, controller, key, &keypath) < 0)
        return -1;
715

716
    VIR_DEBUG("Get value %s", keypath);
717

718 719 720 721
    if ((rc = virFileReadAll(keypath, 1024*1024, value)) < 0) {
        virReportSystemError(errno,
                             _("Unable to read from '%s'"), keypath);
        goto cleanup;
722 723
    }

724 725 726
    /* Terminated with '\n' has sometimes harmful effects to the caller */
    if (rc > 0 && (*value)[rc - 1] == '\n')
        (*value)[rc - 1] = '\0';
727

728 729
    ret = 0;

730
 cleanup:
731 732
    VIR_FREE(keypath);
    return ret;
733 734
}

E
Eric Blake 已提交
735 736 737 738 739 740

static int
virCgroupSetValueU64(virCgroupPtr group,
                     int controller,
                     const char *key,
                     unsigned long long int value)
741 742
{
    char *strval = NULL;
743
    int ret;
744

745 746
    if (virAsprintf(&strval, "%llu", value) < 0)
        return -1;
747

748
    ret = virCgroupSetValueStr(group, controller, key, strval);
749 750 751

    VIR_FREE(strval);

752
    return ret;
753 754 755
}


E
Eric Blake 已提交
756 757 758 759 760
static int
virCgroupSetValueI64(virCgroupPtr group,
                     int controller,
                     const char *key,
                     long long int value)
761 762
{
    char *strval = NULL;
763
    int ret;
764

765 766
    if (virAsprintf(&strval, "%lld", value) < 0)
        return -1;
767

768
    ret = virCgroupSetValueStr(group, controller, key, strval);
769 770 771

    VIR_FREE(strval);

772
    return ret;
773 774
}

E
Eric Blake 已提交
775 776 777 778 779 780

static int
virCgroupGetValueI64(virCgroupPtr group,
                     int controller,
                     const char *key,
                     long long int *value)
781 782
{
    char *strval = NULL;
783
    int ret = -1;
784

785 786
    if (virCgroupGetValueStr(group, controller, key, &strval) < 0)
        goto cleanup;
787

788 789 790 791 792 793
    if (virStrToLong_ll(strval, NULL, 10, value) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Unable to parse '%s' as an integer"),
                       strval);
        goto cleanup;
    }
794

795 796
    ret = 0;

797
 cleanup:
798 799
    VIR_FREE(strval);
    return ret;
800 801
}

E
Eric Blake 已提交
802 803 804 805 806 807

static int
virCgroupGetValueU64(virCgroupPtr group,
                     int controller,
                     const char *key,
                     unsigned long long int *value)
808 809
{
    char *strval = NULL;
810
    int ret = -1;
811

812 813
    if (virCgroupGetValueStr(group, controller, key, &strval) < 0)
        goto cleanup;
814

815 816 817 818 819 820
    if (virStrToLong_ull(strval, NULL, 10, value) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Unable to parse '%s' as an integer"),
                       strval);
        goto cleanup;
    }
821

822 823
    ret = 0;

824
 cleanup:
825 826
    VIR_FREE(strval);
    return ret;
827 828 829
}


E
Eric Blake 已提交
830 831
static int
virCgroupCpuSetInherit(virCgroupPtr parent, virCgroupPtr group)
832
{
833
    size_t i;
834 835 836 837 838
    const char *inherit_values[] = {
        "cpuset.cpus",
        "cpuset.mems",
    };

839
    VIR_DEBUG("Setting up inheritance %s -> %s", parent->path, group->path);
840
    for (i = 0; i < ARRAY_CARDINALITY(inherit_values); i++) {
841
        char *value;
842

843 844 845 846
        if (virCgroupGetValueStr(parent,
                                 VIR_CGROUP_CONTROLLER_CPUSET,
                                 inherit_values[i],
                                 &value) < 0)
847
            return -1;
848 849 850

        VIR_DEBUG("Inherit %s = %s", inherit_values[i], value);

851 852 853 854 855
        if (virCgroupSetValueStr(group,
                                 VIR_CGROUP_CONTROLLER_CPUSET,
                                 inherit_values[i],
                                 value) < 0) {
            VIR_FREE(value);
856
            return -1;
857
        }
858
        VIR_FREE(value);
859 860
    }

861
    return 0;
862 863
}

E
Eric Blake 已提交
864 865 866

static int
virCgroupSetMemoryUseHierarchy(virCgroupPtr group)
867 868 869 870
{
    unsigned long long value;
    const char *filename = "memory.use_hierarchy";

871 872 873
    if (virCgroupGetValueU64(group,
                             VIR_CGROUP_CONTROLLER_MEMORY,
                             filename, &value) < 0)
874
        return -1;
875 876 877 878 879 880

    /* Setting twice causes error, so if already enabled, skip setting */
    if (value == 1)
        return 0;

    VIR_DEBUG("Setting up %s/%s", group->path, filename);
881 882 883
    if (virCgroupSetValueU64(group,
                             VIR_CGROUP_CONTROLLER_MEMORY,
                             filename, 1) < 0)
884
        return -1;
885

886
    return 0;
887 888
}

E
Eric Blake 已提交
889 890 891 892 893 894

static int
virCgroupMakeGroup(virCgroupPtr parent,
                   virCgroupPtr group,
                   bool create,
                   unsigned int flags)
895
{
896
    size_t i;
897
    int ret = -1;
898

899
    VIR_DEBUG("Make group %s", group->path);
900
    for (i = 0; i < VIR_CGROUP_CONTROLLER_LAST; i++) {
901 902
        char *path = NULL;

903 904 905 906 907 908
        /* We must never mkdir() in systemd's hierarchy */
        if (i == VIR_CGROUP_CONTROLLER_SYSTEMD) {
            VIR_DEBUG("Not creating systemd controller group");
            continue;
        }

909
        /* Skip over controllers that aren't mounted */
910 911 912
        if (!group->controllers[i].mountPoint) {
            VIR_DEBUG("Skipping unmounted controller %s",
                      virCgroupControllerTypeToString(i));
913
            continue;
914
        }
915

916
        if (virCgroupPathOfController(group, i, "", &path) < 0)
917
            return -1;
918

919 920 921
        /* As of Feb 2011, clang can't see that the above function
         * call did not modify group. */
        sa_assert(group->controllers[i].mountPoint);
922

923
        VIR_DEBUG("Make controller %s", path);
924
        if (!virFileExists(path)) {
925 926
            if (!create ||
                mkdir(path, 0755) < 0) {
927 928 929 930
                if (errno == EEXIST) {
                    VIR_FREE(path);
                    continue;
                }
931 932 933 934 935
                /* With a kernel that doesn't support multi-level directory
                 * for blkio controller, libvirt will fail and disable all
                 * other controllers even though they are available. So
                 * treat blkio as unmounted if mkdir fails. */
                if (i == VIR_CGROUP_CONTROLLER_BLKIO) {
936
                    VIR_DEBUG("Ignoring mkdir failure with blkio controller. Kernel probably too old");
937 938 939 940
                    VIR_FREE(group->controllers[i].mountPoint);
                    VIR_FREE(path);
                    continue;
                } else {
941 942 943
                    virReportSystemError(errno,
                                         _("Failed to create controller %s for group"),
                                         virCgroupControllerTypeToString(i));
944
                    VIR_FREE(path);
945
                    goto cleanup;
946
                }
947
            }
948 949
            if (group->controllers[VIR_CGROUP_CONTROLLER_CPUSET].mountPoint != NULL &&
                (i == VIR_CGROUP_CONTROLLER_CPUSET ||
E
Eric Blake 已提交
950 951
                 STREQ(group->controllers[i].mountPoint,
                       group->controllers[VIR_CGROUP_CONTROLLER_CPUSET].mountPoint))) {
952
                if (virCgroupCpuSetInherit(parent, group) < 0) {
953
                    VIR_FREE(path);
954
                    goto cleanup;
955
                }
956
            }
957 958 959 960
            /*
             * Note that virCgroupSetMemoryUseHierarchy should always be
             * called prior to creating subcgroups and attaching tasks.
             */
961 962
            if ((flags & VIR_CGROUP_MEM_HIERACHY) &&
                (group->controllers[VIR_CGROUP_CONTROLLER_MEMORY].mountPoint != NULL) &&
963
                (i == VIR_CGROUP_CONTROLLER_MEMORY ||
E
Eric Blake 已提交
964 965
                 STREQ(group->controllers[i].mountPoint,
                       group->controllers[VIR_CGROUP_CONTROLLER_MEMORY].mountPoint))) {
966
                if (virCgroupSetMemoryUseHierarchy(group) < 0) {
967
                    VIR_FREE(path);
968
                    goto cleanup;
969 970
                }
            }
971 972 973 974 975
        }

        VIR_FREE(path);
    }

976
    VIR_DEBUG("Done making controllers for group");
977 978
    ret = 0;

979
 cleanup:
980
    return ret;
981 982
}

983

984 985 986 987 988 989 990 991 992 993 994 995 996 997
/**
 * 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.
 *
998
 * Returns 0 on success, -1 on error
999
 */
E
Eric Blake 已提交
1000 1001 1002 1003 1004 1005
static int
virCgroupNew(pid_t pid,
             const char *path,
             virCgroupPtr parent,
             int controllers,
             virCgroupPtr *group)
1006
{
1007 1008
    VIR_DEBUG("parent=%p path=%s controllers=%d",
              parent, path, controllers);
1009
    *group = NULL;
1010

1011 1012
    if (VIR_ALLOC((*group)) < 0)
        goto error;
1013

1014
    if (path[0] == '/' || !parent) {
1015 1016
        if (VIR_STRDUP((*group)->path, path) < 0)
            goto error;
1017 1018 1019 1020
    } else {
        if (virAsprintf(&(*group)->path, "%s%s%s",
                        parent->path,
                        STREQ(parent->path, "") ? "" : "/",
1021 1022
                        path) < 0)
            goto error;
1023 1024
    }

1025
    if (virCgroupDetect(*group, pid, controllers, path, parent) < 0)
1026
        goto error;
1027

1028 1029
    return 0;

1030
 error:
1031 1032
    virCgroupFree(group);
    *group = NULL;
1033

1034
    return -1;
1035
}
1036

1037

1038 1039 1040 1041 1042 1043
/**
 * virCgroupAddTask:
 *
 * @group: The cgroup to add a task to
 * @pid: The pid of the task to add
 *
1044
 * Returns: 0 on success, -1 on error
1045
 */
E
Eric Blake 已提交
1046 1047
int
virCgroupAddTask(virCgroupPtr group, pid_t pid)
1048
{
1049
    int ret = -1;
1050
    size_t i;
1051

1052
    for (i = 0; i < VIR_CGROUP_CONTROLLER_LAST; i++) {
1053 1054 1055
        /* Skip over controllers not mounted */
        if (!group->controllers[i].mountPoint)
            continue;
1056

1057 1058 1059 1060
        /* We must never add tasks in systemd's hierarchy */
        if (i == VIR_CGROUP_CONTROLLER_SYSTEMD)
            continue;

E
Eric Blake 已提交
1061
        if (virCgroupSetValueU64(group, i, "tasks", pid) < 0)
1062
            goto cleanup;
1063 1064
    }

1065
    ret = 0;
1066
 cleanup:
1067
    return ret;
1068 1069
}

E
Eric Blake 已提交
1070

1071 1072 1073 1074 1075 1076 1077
/**
 * virCgroupAddTaskController:
 *
 * @group: The cgroup to add a task to
 * @pid: The pid of the task to add
 * @controller: The cgroup controller to be operated on
 *
1078
 * Returns: 0 on success or -1 on error
1079
 */
E
Eric Blake 已提交
1080 1081
int
virCgroupAddTaskController(virCgroupPtr group, pid_t pid, int controller)
1082
{
1083 1084 1085 1086 1087
    if (controller < 0 || controller >= VIR_CGROUP_CONTROLLER_LAST) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Controller %d out of range"), controller);
        return -1;
    }
1088

1089 1090 1091 1092 1093 1094
    if (!group->controllers[controller].mountPoint) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Controller '%s' not mounted"),
                       virCgroupControllerTypeToString(controller));
        return -1;
    }
1095 1096 1097 1098 1099 1100

    return virCgroupSetValueU64(group, controller, "tasks",
                                (unsigned long long)pid);
}


E
Eric Blake 已提交
1101 1102 1103 1104
static int
virCgroupAddTaskStrController(virCgroupPtr group,
                              const char *pidstr,
                              int controller)
1105 1106 1107 1108 1109 1110
{
    char *str = NULL, *cur = NULL, *next = NULL;
    unsigned long long p = 0;
    int rc = 0;
    char *endp;

1111 1112
    if (VIR_STRDUP(str, pidstr) < 0)
        return -1;
1113 1114 1115

    cur = str;
    while (*cur != '\0') {
1116 1117 1118
        if (virStrToLong_ull(cur, &endp, 10, &p) < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Cannot parse '%s' as an integer"), cur);
1119
            goto cleanup;
1120
        }
1121

1122 1123 1124 1125 1126 1127 1128 1129
        if (virCgroupAddTaskController(group, p, controller) < 0) {
            /* A thread that exits between when we first read the source
             * tasks and now is not fatal.  */
            if (virLastErrorIsSystemErrno(ESRCH))
                virResetLastError();
            else
                goto cleanup;
        }
1130 1131 1132 1133 1134 1135 1136 1137 1138 1139

        next = strchr(cur, '\n');
        if (next) {
            cur = next + 1;
            *next = '\0';
        } else {
            break;
        }
    }

1140
 cleanup:
1141 1142 1143 1144
    VIR_FREE(str);
    return rc;
}

E
Eric Blake 已提交
1145

1146 1147 1148 1149 1150 1151
/**
 * virCgroupMoveTask:
 *
 * @src_group: The source cgroup where all tasks are removed from
 * @dest_group: The destination where all tasks are added to
 *
1152
 * Returns: 0 on success or -1 on failure
1153
 */
E
Eric Blake 已提交
1154 1155
int
virCgroupMoveTask(virCgroupPtr src_group, virCgroupPtr dest_group)
1156
{
1157
    int ret = -1;
1158
    char *content = NULL;
1159
    size_t i;
1160

1161
    for (i = 0; i < VIR_CGROUP_CONTROLLER_LAST; i++) {
1162 1163 1164
        if (!src_group->controllers[i].mountPoint ||
            !dest_group->controllers[i].mountPoint)
            continue;
1165

1166 1167 1168 1169
        /* We must never move tasks in systemd's hierarchy */
        if (i == VIR_CGROUP_CONTROLLER_SYSTEMD)
            continue;

1170 1171 1172 1173 1174
        /* New threads are created in the same group as their parent;
         * but if a thread is created after we first read we aren't
         * aware that it needs to move.  Therefore, we must iterate
         * until content is empty.  */
        while (1) {
J
Ján Tomko 已提交
1175
            VIR_FREE(content);
1176 1177 1178
            if (virCgroupGetValueStr(src_group, i, "tasks", &content) < 0)
                return -1;

1179 1180
            if (!*content)
                break;
1181

1182
            if (virCgroupAddTaskStrController(dest_group, content, i) < 0)
1183 1184
                goto cleanup;
        }
1185
    }
1186

1187
    ret = 0;
1188
 cleanup:
1189
    VIR_FREE(content);
1190
    return ret;
1191
}
1192

1193

E
Eric Blake 已提交
1194 1195
static int
virCgroupSetPartitionSuffix(const char *path, char **res)
1196
{
1197
    char **tokens;
1198
    size_t i;
1199
    int ret = -1;
1200

1201
    if (!(tokens = virStringSplit(path, "/", 0)))
1202
        return ret;
1203

1204
    for (i = 0; tokens[i] != NULL; i++) {
1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219
        /* 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],
1220
                              strlen(tokens[i]) + strlen(".partition") + 1) < 0)
1221 1222 1223
                goto cleanup;
            strcat(tokens[i], ".partition");
        }
1224

1225
        if (virCgroupPartitionEscape(&(tokens[i])) < 0)
1226
            goto cleanup;
1227 1228
    }

1229
    if (!(*res = virStringJoin((const char **)tokens, "/")))
1230
        goto cleanup;
1231 1232

    ret = 0;
1233

1234
 cleanup:
1235 1236 1237 1238
    virStringFreeList(tokens);
    return ret;
}

E
Eric Blake 已提交
1239

1240 1241 1242 1243 1244 1245 1246 1247 1248
/**
 * 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
 * partition path identified by @name.
 *
1249
 * Returns 0 on success, -1 on failure
1250
 */
E
Eric Blake 已提交
1251 1252 1253 1254 1255
int
virCgroupNewPartition(const char *path,
                      bool create,
                      int controllers,
                      virCgroupPtr *group)
1256
{
1257
    int ret = -1;
1258 1259
    char *parentPath = NULL;
    virCgroupPtr parent = NULL;
1260
    char *newPath = NULL;
1261 1262 1263
    VIR_DEBUG("path=%s create=%d controllers=%x",
              path, create, controllers);

1264 1265 1266 1267 1268 1269
    if (path[0] != '/') {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Partition path '%s' must start with '/'"),
                       path);
        return -1;
    }
1270

1271
    if (virCgroupSetPartitionSuffix(path, &newPath) < 0)
1272 1273
        goto cleanup;

1274
    if (virCgroupNew(-1, newPath, NULL, controllers, group) < 0)
1275 1276
        goto cleanup;

1277
    if (STRNEQ(newPath, "/")) {
1278
        char *tmp;
1279
        if (VIR_STRDUP(parentPath, newPath) < 0)
1280 1281 1282 1283 1284 1285
            goto cleanup;

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

1286
        if (virCgroupNew(-1, parentPath, NULL, controllers, &parent) < 0)
1287 1288
            goto cleanup;

1289
        if (virCgroupMakeGroup(parent, *group, create, VIR_CGROUP_NONE) < 0) {
1290 1291 1292 1293 1294
            virCgroupRemove(*group);
            goto cleanup;
        }
    }

1295
    ret = 0;
1296
 cleanup:
1297
    if (ret != 0)
1298 1299 1300
        virCgroupFree(group);
    virCgroupFree(&parent);
    VIR_FREE(parentPath);
1301
    VIR_FREE(newPath);
1302
    return ret;
1303 1304
}

1305

G
Gao feng 已提交
1306
/**
1307
* virCgroupNewSelf:
G
Gao feng 已提交
1308 1309 1310
*
* @group: Pointer to returned virCgroupPtr
*
1311 1312 1313
* Obtain a cgroup representing the config of the
* current process
*
1314
* Returns 0 on success, or -1 on error
G
Gao feng 已提交
1315
*/
E
Eric Blake 已提交
1316 1317
int
virCgroupNewSelf(virCgroupPtr *group)
G
Gao feng 已提交
1318
{
1319
    return virCgroupNewDetect(-1, -1, group);
G
Gao feng 已提交
1320
}
1321

1322

1323 1324 1325 1326 1327 1328 1329 1330
/**
 * virCgroupNewDomainPartition:
 *
 * @partition: partition holding the domain
 * @driver: name of the driver
 * @name: name of the domain
 * @group: Pointer to returned virCgroupPtr
 *
1331
 * Returns 0 on success, or -1 on error
1332
 */
E
Eric Blake 已提交
1333 1334 1335 1336 1337 1338
int
virCgroupNewDomainPartition(virCgroupPtr partition,
                            const char *driver,
                            const char *name,
                            bool create,
                            virCgroupPtr *group)
1339
{
1340
    int ret = -1;
1341
    char *grpname = NULL;
1342

1343
    if (virAsprintf(&grpname, "%s.libvirt-%s",
1344
                    name, driver) < 0)
1345
        goto cleanup;
1346

1347 1348
    if (virCgroupPartitionEscape(&grpname) < 0)
        goto cleanup;
1349

1350
    if (virCgroupNew(-1, grpname, partition, -1, group) < 0)
1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362
        goto cleanup;

    /*
     * 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
1363 1364
    if (virCgroupMakeGroup(partition, *group, create,
                           VIR_CGROUP_MEM_HIERACHY) < 0) {
1365 1366 1367
        virCgroupRemove(*group);
        virCgroupFree(group);
        goto cleanup;
1368 1369
    }

1370 1371
    ret = 0;

1372
 cleanup:
1373
    VIR_FREE(grpname);
1374
    return ret;
1375
}
1376

E
Eric Blake 已提交
1377

1378
/**
1379
 * virCgroupNewVcpu:
1380
 *
1381
 * @domain: group for the domain
1382
 * @vcpuid: id of the vcpu
1383
 * @create: true to create if not already existing
1384 1385
 * @group: Pointer to returned virCgroupPtr
 *
1386
 * Returns 0 on success, or -1 on error
1387
 */
E
Eric Blake 已提交
1388 1389 1390 1391 1392
int
virCgroupNewVcpu(virCgroupPtr domain,
                 int vcpuid,
                 bool create,
                 virCgroupPtr *group)
1393
{
1394 1395
    int ret = -1;
    char *name = NULL;
1396
    int controllers;
1397

1398
    if (virAsprintf(&name, "vcpu%d", vcpuid) < 0)
1399
        goto cleanup;
1400

1401 1402 1403 1404
    controllers = ((1 << VIR_CGROUP_CONTROLLER_CPU) |
                   (1 << VIR_CGROUP_CONTROLLER_CPUACCT) |
                   (1 << VIR_CGROUP_CONTROLLER_CPUSET));

1405
    if (virCgroupNew(-1, name, domain, controllers, group) < 0)
1406
        goto cleanup;
1407

1408 1409 1410 1411
    if (virCgroupMakeGroup(domain, *group, create, VIR_CGROUP_NONE) < 0) {
        virCgroupRemove(*group);
        virCgroupFree(group);
        goto cleanup;
1412 1413
    }

1414
    ret = 0;
1415
 cleanup:
1416 1417
    VIR_FREE(name);
    return ret;
1418 1419
}

E
Eric Blake 已提交
1420

1421
/**
1422
 * virCgroupNewEmulator:
1423
 *
1424 1425
 * @domain: group for the domain
 * @create: true to create if not already existing
1426 1427
 * @group: Pointer to returned virCgroupPtr
 *
1428
 * Returns: 0 on success or -1 on error
1429
 */
E
Eric Blake 已提交
1430 1431 1432 1433
int
virCgroupNewEmulator(virCgroupPtr domain,
                     bool create,
                     virCgroupPtr *group)
1434
{
1435
    int ret = -1;
1436
    int controllers;
1437

1438 1439 1440 1441
    controllers = ((1 << VIR_CGROUP_CONTROLLER_CPU) |
                   (1 << VIR_CGROUP_CONTROLLER_CPUACCT) |
                   (1 << VIR_CGROUP_CONTROLLER_CPUSET));

1442
    if (virCgroupNew(-1, "emulator", domain, controllers, group) < 0)
1443
        goto cleanup;
1444

1445 1446 1447 1448
    if (virCgroupMakeGroup(domain, *group, create, VIR_CGROUP_NONE) < 0) {
        virCgroupRemove(*group);
        virCgroupFree(group);
        goto cleanup;
1449 1450
    }

1451
    ret = 0;
1452
 cleanup:
1453
    return ret;
1454
}
1455

1456

E
Eric Blake 已提交
1457 1458 1459 1460
int
virCgroupNewDetect(pid_t pid,
                   int controllers,
                   virCgroupPtr *group)
1461
{
1462
    return virCgroupNew(pid, "", NULL, controllers, group);
1463 1464
}

E
Eric Blake 已提交
1465

1466 1467 1468
/*
 * Returns 0 on success (but @group may be NULL), -1 on fatal error
 */
E
Eric Blake 已提交
1469 1470 1471 1472 1473 1474 1475
int
virCgroupNewDetectMachine(const char *name,
                          const char *drivername,
                          pid_t pid,
                          const char *partition,
                          int controllers,
                          virCgroupPtr *group)
1476
{
1477
    if (virCgroupNewDetect(pid, controllers, group) < 0) {
1478 1479 1480 1481 1482
        if (virCgroupNewIgnoreError())
            return 0;
        return -1;
    }

E
Eric Blake 已提交
1483 1484
    if (!virCgroupValidateMachineGroup(*group, name, drivername, partition,
                                       true)) {
1485 1486
        VIR_DEBUG("Failed to validate machine name for '%s' driver '%s'",
                  name, drivername);
1487 1488 1489 1490 1491 1492 1493
        virCgroupFree(group);
        return 0;
    }

    return 0;
}

E
Eric Blake 已提交
1494

1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508
/*
 * Returns 0 on success, -1 on fatal error, -2 on systemd not available
 */
static int
virCgroupNewMachineSystemd(const char *name,
                           const char *drivername,
                           bool privileged,
                           const unsigned char *uuid,
                           const char *rootdir,
                           pid_t pidleader,
                           bool isContainer,
                           const char *partition,
                           int controllers,
                           virCgroupPtr *group)
1509 1510
{
    int ret = -1;
1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534
    int rv;
    virCgroupPtr init, parent = NULL;
    char *path = NULL;
    char *offset;

    VIR_DEBUG("Trying to setup machine '%s' via systemd", name);
    if ((rv = virSystemdCreateMachine(name,
                                      drivername,
                                      privileged,
                                      uuid,
                                      rootdir,
                                      pidleader,
                                      isContainer,
                                      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;
1535

1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599
    path = init->controllers[VIR_CGROUP_CONTROLLER_SYSTEMD].placement;
    init->controllers[VIR_CGROUP_CONTROLLER_SYSTEMD].placement = NULL;
    virCgroupFree(&init);

    if (!path || STREQ(path, "/") || path[0] != '/') {
        VIR_DEBUG("Systemd didn't setup its controller");
        ret = -2;
        goto cleanup;
    }

    offset = path;

    if (virCgroupNew(pidleader,
                     "",
                     NULL,
                     controllers,
                     &parent) < 0)
        goto cleanup;


    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;
        }
    }

    if (virCgroupAddTask(*group, pidleader) < 0) {
        virErrorPtr saved = virSaveLastError();
        virCgroupRemove(*group);
        virCgroupFree(group);
        if (saved) {
            virSetError(saved);
            virFreeError(saved);
        }
    }

    ret = 0;
 cleanup:
    virCgroupFree(&parent);
    VIR_FREE(path);
    return ret;
}
1600

E
Eric Blake 已提交
1601

1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613
static int
virCgroupNewMachineManual(const char *name,
                          const char *drivername,
                          pid_t pidleader,
                          const char *partition,
                          int controllers,
                          virCgroupPtr *group)
{
    virCgroupPtr parent = NULL;
    int ret = -1;

    VIR_DEBUG("Fallback to non-systemd setup");
1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640
    if (virCgroupNewPartition(partition,
                              STREQ(partition, "/machine"),
                              controllers,
                              &parent) < 0) {
        if (virCgroupNewIgnoreError())
            goto done;

        goto cleanup;
    }

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

    if (virCgroupAddTask(*group, pidleader) < 0) {
        virErrorPtr saved = virSaveLastError();
        virCgroupRemove(*group);
        virCgroupFree(group);
        if (saved) {
            virSetError(saved);
            virFreeError(saved);
        }
    }

1641
 done:
1642 1643
    ret = 0;

1644
 cleanup:
1645 1646 1647 1648
    virCgroupFree(&parent);
    return ret;
}

E
Eric Blake 已提交
1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660

int
virCgroupNewMachine(const char *name,
                    const char *drivername,
                    bool privileged,
                    const unsigned char *uuid,
                    const char *rootdir,
                    pid_t pidleader,
                    bool isContainer,
                    const char *partition,
                    int controllers,
                    virCgroupPtr *group)
1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688
{
    int rv;

    *group = NULL;

    if ((rv = virCgroupNewMachineSystemd(name,
                                         drivername,
                                         privileged,
                                         uuid,
                                         rootdir,
                                         pidleader,
                                         isContainer,
                                         partition,
                                         controllers,
                                         group)) == 0)
        return 0;

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

    return virCgroupNewMachineManual(name,
                                     drivername,
                                     pidleader,
                                     partition,
                                     controllers,
                                     group);
}

E
Eric Blake 已提交
1689 1690 1691

bool
virCgroupNewIgnoreError(void)
1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702
{
    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
1703

E
Eric Blake 已提交
1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798
/**
 * virCgroupFree:
 *
 * @group: The group structure to free
 */
void
virCgroupFree(virCgroupPtr *group)
{
    size_t i;

    if (*group == NULL)
        return;

    for (i = 0; i < VIR_CGROUP_CONTROLLER_LAST; i++) {
        VIR_FREE((*group)->controllers[i].mountPoint);
        VIR_FREE((*group)->controllers[i].linkPoint);
        VIR_FREE((*group)->controllers[i].placement);
    }

    VIR_FREE((*group)->path);
    VIR_FREE(*group);
}


/**
 * 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)
{
    if (!cgroup)
        return false;
    if (controller < 0 || controller >= VIR_CGROUP_CONTROLLER_LAST)
        return false;
    return cgroup->controllers[controller].mountPoint != NULL;
}


int
virCgroupPathOfController(virCgroupPtr group,
                          int controller,
                          const char *key,
                          char **path)
{
    if (controller == -1) {
        size_t i;
        for (i = 0; i < VIR_CGROUP_CONTROLLER_LAST; i++) {
            /* Reject any controller with a placement
             * of '/' to avoid doing bad stuff to the root
             * cgroup
             */
            if (group->controllers[i].mountPoint &&
                group->controllers[i].placement &&
                STRNEQ(group->controllers[i].placement, "/")) {
                controller = i;
                break;
            }
        }
    }
    if (controller == -1) {
        virReportSystemError(ENOSYS, "%s",
                             _("No controllers are mounted"));
        return -1;
    }

    if (group->controllers[controller].mountPoint == NULL) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Controller '%s' is not mounted"),
                       virCgroupControllerTypeToString(controller));
        return -1;
    }

    if (group->controllers[controller].placement == NULL) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Controller '%s' is not enabled for group"),
                       virCgroupControllerTypeToString(controller));
        return -1;
    }

    if (virAsprintf(path, "%s%s/%s",
                    group->controllers[controller].mountPoint,
                    group->controllers[controller].placement,
                    key ? key : "") < 0)
        return -1;

    return 0;
}


1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899
/**
 * 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)
{
    long long stats_val;
    char *str1 = NULL, *str2 = NULL, *p1, *p2;
    size_t i;
    int ret = -1;

    const char *value_names[] = {
        "Read ",
        "Write "
    };
    long long *bytes_ptrs[] = {
        bytes_read,
        bytes_write
    };
    long long *requests_ptrs[] = {
        requests_read,
        requests_write
    };

    *bytes_read = 0;
    *bytes_write = 0;
    *requests_read = 0;
    *requests_write = 0;

    if (virCgroupGetValueStr(group,
                             VIR_CGROUP_CONTROLLER_BLKIO,
                             "blkio.throttle.io_service_bytes", &str1) < 0)
        goto cleanup;

    if (virCgroupGetValueStr(group,
                             VIR_CGROUP_CONTROLLER_BLKIO,
                             "blkio.throttle.io_serviced", &str2) < 0)
        goto cleanup;

    /* sum up all entries of the same kind, from all devices */
    for (i = 0; i < ARRAY_CARDINALITY(value_names); i++) {
        p1 = str1;
        p2 = str2;

        while ((p1 = strstr(p1, value_names[i]))) {
            p1 += strlen(value_names[i]);
            if (virStrToLong_ll(p1, &p1, 10, &stats_val) < 0) {
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("Cannot parse byte %sstat '%s'"),
                               value_names[i],
                               p1);
                goto cleanup;
            }

            if (stats_val < 0 ||
                (stats_val > 0 && *bytes_ptrs[i] > (LLONG_MAX - stats_val)))
            {
                virReportError(VIR_ERR_OVERFLOW,
                               _("Sum of byte %sstat overflows"),
                               value_names[i]);
                goto cleanup;
            }
            *bytes_ptrs[i] += stats_val;
        }

        while ((p2 = strstr(p2, value_names[i]))) {
            p2 += strlen(value_names[i]);
            if (virStrToLong_ll(p2, &p2, 10, &stats_val) < 0) {
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("Cannot parse %srequest stat '%s'"),
                               value_names[i],
                               p2);
                goto cleanup;
            }

            if (stats_val < 0 ||
                (stats_val > 0 && *requests_ptrs[i] > (LLONG_MAX - stats_val)))
            {
                virReportError(VIR_ERR_OVERFLOW,
                               _("Sum of %srequest stat overflows"),
                               value_names[i]);
                goto cleanup;
            }
            *requests_ptrs[i] += stats_val;
        }
    }

    ret = 0;

1900
 cleanup:
1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
    VIR_FREE(str2);
    VIR_FREE(str1);
    return ret;
}


/**
 * 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)
{
    char *str1 = NULL, *str2 = NULL, *str3 = NULL, *p1, *p2;
    struct stat sb;
    size_t i;
    int ret = -1;

    const char *value_names[] = {
        "Read ",
        "Write "
    };
    long long *bytes_ptrs[] = {
        bytes_read,
        bytes_write
    };
    long long *requests_ptrs[] = {
        requests_read,
        requests_write
    };

    if (stat(path, &sb) < 0) {
        virReportSystemError(errno,
                             _("Path '%s' is not accessible"),
                             path);
        return -1;
    }

    if (!S_ISBLK(sb.st_mode)) {
        virReportSystemError(EINVAL,
                             _("Path '%s' must be a block device"),
                             path);
        return -1;
    }

    if (virCgroupGetValueStr(group,
                             VIR_CGROUP_CONTROLLER_BLKIO,
                             "blkio.throttle.io_service_bytes", &str1) < 0)
        goto cleanup;

    if (virCgroupGetValueStr(group,
                             VIR_CGROUP_CONTROLLER_BLKIO,
                             "blkio.throttle.io_serviced", &str2) < 0)
        goto cleanup;

    if (virAsprintf(&str3, "%d:%d ", major(sb.st_rdev), minor(sb.st_rdev)) < 0)
        goto cleanup;

    if (!(p1 = strstr(str1, str3))) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Cannot find byte stats for block device '%s'"),
                       str3);
        goto cleanup;
    }

    if (!(p2 = strstr(str2, str3))) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Cannot find request stats for block device '%s'"),
                       str3);
        goto cleanup;
    }

    for (i = 0; i < ARRAY_CARDINALITY(value_names); i++) {
        if (!(p1 = strstr(p1, value_names[i]))) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Cannot find byte %sstats for block device '%s'"),
                           value_names[i], str3);
            goto cleanup;
        }

        if (virStrToLong_ll(p1 + strlen(value_names[i]), &p1, 10, bytes_ptrs[i]) < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Cannot parse %sstat '%s'"),
                           value_names[i], p1 + strlen(value_names[i]));
            goto cleanup;
        }

        if (!(p2 = strstr(p2, value_names[i]))) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Cannot find request %sstats for block device '%s'"),
                           value_names[i], str3);
            goto cleanup;
        }

        if (virStrToLong_ll(p2 + strlen(value_names[i]), &p2, 10, requests_ptrs[i]) < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Cannot parse %sstat '%s'"),
                           value_names[i], p2 + strlen(value_names[i]));
            goto cleanup;
        }
    }

    ret = 0;

2018
 cleanup:
2019 2020 2021 2022 2023 2024 2025
    VIR_FREE(str3);
    VIR_FREE(str2);
    VIR_FREE(str1);
    return ret;
}


2026 2027 2028 2029 2030 2031
/**
 * virCgroupSetBlkioWeight:
 *
 * @group: The cgroup to change io weight for
 * @weight: The Weight for this cgroup
 *
2032
 * Returns: 0 on success, -1 on error
2033
 */
E
Eric Blake 已提交
2034 2035
int
virCgroupSetBlkioWeight(virCgroupPtr group, unsigned int weight)
2036 2037 2038 2039 2040 2041 2042
{
    return virCgroupSetValueU64(group,
                                VIR_CGROUP_CONTROLLER_BLKIO,
                                "blkio.weight",
                                weight);
}

E
Eric Blake 已提交
2043

2044 2045 2046 2047 2048 2049
/**
 * virCgroupGetBlkioWeight:
 *
 * @group: The cgroup to get weight for
 * @Weight: Pointer to returned weight
 *
2050
 * Returns: 0 on success, -1 on error
2051
 */
E
Eric Blake 已提交
2052 2053
int
virCgroupGetBlkioWeight(virCgroupPtr group, unsigned int *weight)
2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064
{
    unsigned long long tmp;
    int ret;
    ret = virCgroupGetValueU64(group,
                               VIR_CGROUP_CONTROLLER_BLKIO,
                               "blkio.weight", &tmp);
    if (ret == 0)
        *weight = tmp;
    return ret;
}

2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108
/**
 * 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)
{
    char *str;
    struct stat sb;
    int ret;

    if (stat(path, &sb) < 0) {
        virReportSystemError(errno,
                             _("Path '%s' is not accessible"),
                             path);
        return -1;
    }

    if (!S_ISBLK(sb.st_mode)) {
        virReportSystemError(EINVAL,
                             _("Path '%s' must be a block device"),
                             path);
        return -1;
    }

    if (virAsprintf(&str, "%d:%d %u", major(sb.st_rdev),
                    minor(sb.st_rdev), riops) < 0)
        return -1;

    ret = virCgroupSetValueStr(group,
                               VIR_CGROUP_CONTROLLER_BLKIO,
                               "blkio.throttle.read_iops_device",
                               str);

    VIR_FREE(str);
    return ret;
}

E
Eric Blake 已提交
2109

2110
/**
2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203
 * 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)
{
    char *str;
    struct stat sb;
    int ret;

    if (stat(path, &sb) < 0) {
        virReportSystemError(errno,
                             _("Path '%s' is not accessible"),
                             path);
        return -1;
    }

    if (!S_ISBLK(sb.st_mode)) {
        virReportSystemError(EINVAL,
                             _("Path '%s' must be a block device"),
                             path);
        return -1;
    }

    if (virAsprintf(&str, "%d:%d %u", major(sb.st_rdev),
                    minor(sb.st_rdev), wiops) < 0)
        return -1;

    ret = virCgroupSetValueStr(group,
                               VIR_CGROUP_CONTROLLER_BLKIO,
                               "blkio.throttle.write_iops_device",
                               str);

    VIR_FREE(str);
    return ret;
}


/**
 * 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)
{
    char *str;
    struct stat sb;
    int ret;

    if (stat(path, &sb) < 0) {
        virReportSystemError(errno,
                             _("Path '%s' is not accessible"),
                             path);
        return -1;
    }

    if (!S_ISBLK(sb.st_mode)) {
        virReportSystemError(EINVAL,
                             _("Path '%s' must be a block device"),
                             path);
        return -1;
    }

    if (virAsprintf(&str, "%d:%d %llu", major(sb.st_rdev),
                    minor(sb.st_rdev), rbps) < 0)
        return -1;

    ret = virCgroupSetValueStr(group,
                               VIR_CGROUP_CONTROLLER_BLKIO,
                               "blkio.throttle.read_bps_device",
                               str);

    VIR_FREE(str);
    return ret;
}

/**
 * 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
2204
 *
2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247
 * Returns: 0 on success, -1 on error
 */
int
virCgroupSetBlkioDeviceWriteBps(virCgroupPtr group,
                                const char *path,
                                unsigned long long wbps)
{
    char *str;
    struct stat sb;
    int ret;

    if (stat(path, &sb) < 0) {
        virReportSystemError(errno,
                             _("Path '%s' is not accessible"),
                             path);
        return -1;
    }

    if (!S_ISBLK(sb.st_mode)) {
        virReportSystemError(EINVAL,
                             _("Path '%s' must be a block device"),
                             path);
        return -1;
    }

    if (virAsprintf(&str, "%d:%d %llu", major(sb.st_rdev),
                    minor(sb.st_rdev), wbps) < 0)
        return -1;

    ret = virCgroupSetValueStr(group,
                               VIR_CGROUP_CONTROLLER_BLKIO,
                               "blkio.throttle.write_bps_device",
                               str);

    VIR_FREE(str);
    return ret;
}


/**
 * virCgroupSetBlkioDeviceWeight:
 * @group: The cgroup to change block io setting for
 * @path: The path of device
2248 2249
 * @weight: The new device weight (100-1000),
 * (10-1000) after kernel 2.6.39, or 0 to clear
2250 2251 2252 2253
 *
 * device_weight is treated as a write-only parameter, so
 * there isn't a getter counterpart.
 *
2254
 * Returns: 0 on success, -1 on error
2255
 */
E
Eric Blake 已提交
2256 2257 2258 2259
int
virCgroupSetBlkioDeviceWeight(virCgroupPtr group,
                              const char *path,
                              unsigned int weight)
2260 2261 2262 2263 2264
{
    char *str;
    struct stat sb;
    int ret;

2265 2266 2267 2268 2269 2270
    if (stat(path, &sb) < 0) {
        virReportSystemError(errno,
                             _("Path '%s' is not accessible"),
                             path);
        return -1;
    }
2271

2272 2273 2274 2275 2276 2277
    if (!S_ISBLK(sb.st_mode)) {
        virReportSystemError(EINVAL,
                             _("Path '%s' must be a block device"),
                             path);
        return -1;
    }
2278 2279 2280

    if (virAsprintf(&str, "%d:%d %d", major(sb.st_rdev), minor(sb.st_rdev),
                    weight) < 0)
2281
        return -1;
2282 2283 2284 2285 2286 2287 2288 2289

    ret = virCgroupSetValueStr(group,
                               VIR_CGROUP_CONTROLLER_BLKIO,
                               "blkio.weight_device",
                               str);
    VIR_FREE(str);
    return ret;
}
2290

2291

2292 2293 2294 2295 2296 2297 2298 2299
/**
 * virCgroupSetMemory:
 *
 * @group: The cgroup to change memory for
 * @kb: The memory amount in kilobytes
 *
 * Returns: 0 on success
 */
E
Eric Blake 已提交
2300 2301
int
virCgroupSetMemory(virCgroupPtr group, unsigned long long kb)
2302
{
2303 2304
    unsigned long long maxkb = VIR_DOMAIN_MEMORY_PARAM_UNLIMITED;

2305 2306 2307 2308 2309 2310 2311 2312
    if (kb > maxkb) {
        virReportError(VIR_ERR_INVALID_ARG,
                       _("Memory '%llu' must be less than %llu"),
                       kb, maxkb);
        return -1;
    }

    if (kb == maxkb)
2313 2314 2315 2316 2317 2318 2319 2320 2321
        return virCgroupSetValueI64(group,
                                    VIR_CGROUP_CONTROLLER_MEMORY,
                                    "memory.limit_in_bytes",
                                    -1);
    else
        return virCgroupSetValueU64(group,
                                    VIR_CGROUP_CONTROLLER_MEMORY,
                                    "memory.limit_in_bytes",
                                    kb << 10);
2322 2323
}

E
Eric Blake 已提交
2324

R
Ryota Ozaki 已提交
2325 2326 2327 2328 2329 2330 2331 2332
/**
 * 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
2333 2334
int
virCgroupGetMemoryUsage(virCgroupPtr group, unsigned long *kb)
R
Ryota Ozaki 已提交
2335
{
C
Cole Robinson 已提交
2336
    long long unsigned int usage_in_bytes;
R
Ryota Ozaki 已提交
2337 2338 2339 2340 2341 2342 2343 2344 2345
    int ret;
    ret = virCgroupGetValueU64(group,
                               VIR_CGROUP_CONTROLLER_MEMORY,
                               "memory.usage_in_bytes", &usage_in_bytes);
    if (ret == 0)
        *kb = (unsigned long) usage_in_bytes >> 10;
    return ret;
}

E
Eric Blake 已提交
2346

2347 2348 2349 2350 2351 2352 2353 2354
/**
 * 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
2355 2356
int
virCgroupSetMemoryHardLimit(virCgroupPtr group, unsigned long long kb)
2357 2358 2359 2360
{
    return virCgroupSetMemory(group, kb);
}

E
Eric Blake 已提交
2361

2362 2363 2364 2365 2366 2367 2368 2369
/**
 * 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
2370 2371
int
virCgroupGetMemoryHardLimit(virCgroupPtr group, unsigned long long *kb)
2372 2373
{
    long long unsigned int limit_in_bytes;
2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386
    int ret = -1;

    if (virCgroupGetValueU64(group,
                             VIR_CGROUP_CONTROLLER_MEMORY,
                             "memory.limit_in_bytes", &limit_in_bytes) < 0)
        goto cleanup;

    *kb = limit_in_bytes >> 10;
    if (*kb > VIR_DOMAIN_MEMORY_PARAM_UNLIMITED)
        *kb = VIR_DOMAIN_MEMORY_PARAM_UNLIMITED;

    ret = 0;
 cleanup:
2387 2388 2389
    return ret;
}

E
Eric Blake 已提交
2390

2391 2392 2393 2394 2395 2396 2397 2398
/**
 * 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
2399 2400
int
virCgroupSetMemorySoftLimit(virCgroupPtr group, unsigned long long kb)
2401
{
2402 2403
    unsigned long long maxkb = VIR_DOMAIN_MEMORY_PARAM_UNLIMITED;

2404 2405 2406 2407 2408 2409 2410 2411
    if (kb > maxkb) {
        virReportError(VIR_ERR_INVALID_ARG,
                       _("Memory '%llu' must be less than %llu"),
                       kb, maxkb);
        return -1;
    }

    if (kb == maxkb)
2412 2413 2414 2415 2416 2417 2418 2419 2420
        return virCgroupSetValueI64(group,
                                    VIR_CGROUP_CONTROLLER_MEMORY,
                                    "memory.soft_limit_in_bytes",
                                    -1);
    else
        return virCgroupSetValueU64(group,
                                    VIR_CGROUP_CONTROLLER_MEMORY,
                                    "memory.soft_limit_in_bytes",
                                    kb << 10);
2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431
}


/**
 * 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
2432 2433
int
virCgroupGetMemorySoftLimit(virCgroupPtr group, unsigned long long *kb)
2434 2435
{
    long long unsigned int limit_in_bytes;
2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448
    int ret = -1;

    if (virCgroupGetValueU64(group,
                             VIR_CGROUP_CONTROLLER_MEMORY,
                             "memory.soft_limit_in_bytes", &limit_in_bytes) < 0)
        goto cleanup;

    *kb = limit_in_bytes >> 10;
    if (*kb > VIR_DOMAIN_MEMORY_PARAM_UNLIMITED)
        *kb = VIR_DOMAIN_MEMORY_PARAM_UNLIMITED;

    ret = 0;
 cleanup:
2449 2450 2451
    return ret;
}

E
Eric Blake 已提交
2452

2453
/**
2454
 * virCgroupSetMemSwapHardLimit:
2455
 *
2456 2457
 * @group: The cgroup to change mem+swap hard limit for
 * @kb: The mem+swap amount in kilobytes
2458 2459 2460
 *
 * Returns: 0 on success
 */
E
Eric Blake 已提交
2461 2462
int
virCgroupSetMemSwapHardLimit(virCgroupPtr group, unsigned long long kb)
2463
{
2464 2465
    unsigned long long maxkb = VIR_DOMAIN_MEMORY_PARAM_UNLIMITED;

2466 2467 2468 2469 2470 2471 2472 2473
    if (kb > maxkb) {
        virReportError(VIR_ERR_INVALID_ARG,
                       _("Memory '%llu' must be less than %llu"),
                       kb, maxkb);
        return -1;
    }

    if (kb == maxkb)
2474 2475 2476 2477 2478 2479 2480 2481 2482
        return virCgroupSetValueI64(group,
                                    VIR_CGROUP_CONTROLLER_MEMORY,
                                    "memory.memsw.limit_in_bytes",
                                    -1);
    else
        return virCgroupSetValueU64(group,
                                    VIR_CGROUP_CONTROLLER_MEMORY,
                                    "memory.memsw.limit_in_bytes",
                                    kb << 10);
2483 2484
}

E
Eric Blake 已提交
2485

2486
/**
2487
 * virCgroupGetMemSwapHardLimit:
2488
 *
2489 2490
 * @group: The cgroup to get mem+swap hard limit for
 * @kb: The mem+swap amount in kilobytes
2491 2492 2493
 *
 * Returns: 0 on success
 */
E
Eric Blake 已提交
2494 2495
int
virCgroupGetMemSwapHardLimit(virCgroupPtr group, unsigned long long *kb)
2496 2497
{
    long long unsigned int limit_in_bytes;
2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510
    int ret = -1;

    if (virCgroupGetValueU64(group,
                             VIR_CGROUP_CONTROLLER_MEMORY,
                             "memory.memsw.limit_in_bytes", &limit_in_bytes) < 0)
        goto cleanup;

    *kb = limit_in_bytes >> 10;
    if (*kb > VIR_DOMAIN_MEMORY_PARAM_UNLIMITED)
        *kb = VIR_DOMAIN_MEMORY_PARAM_UNLIMITED;

    ret = 0;
 cleanup:
2511 2512 2513
    return ret;
}

E
Eric Blake 已提交
2514

G
Gao feng 已提交
2515 2516 2517 2518 2519 2520 2521 2522
/**
 * 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
2523 2524
int
virCgroupGetMemSwapUsage(virCgroupPtr group, unsigned long long *kb)
G
Gao feng 已提交
2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535
{
    long long unsigned int usage_in_bytes;
    int ret;
    ret = virCgroupGetValueU64(group,
                               VIR_CGROUP_CONTROLLER_MEMORY,
                               "memory.memsw.usage_in_bytes", &usage_in_bytes);
    if (ret == 0)
        *kb = usage_in_bytes >> 10;
    return ret;
}

E
Eric Blake 已提交
2536

2537 2538 2539 2540 2541 2542 2543 2544
/**
 * 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
2545 2546
int
virCgroupSetCpusetMems(virCgroupPtr group, const char *mems)
2547 2548 2549 2550 2551 2552 2553
{
    return virCgroupSetValueStr(group,
                                VIR_CGROUP_CONTROLLER_CPUSET,
                                "cpuset.mems",
                                mems);
}

E
Eric Blake 已提交
2554

2555 2556 2557 2558 2559 2560 2561 2562
/**
 * 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
2563 2564
int
virCgroupGetCpusetMems(virCgroupPtr group, char **mems)
2565 2566 2567 2568 2569 2570 2571
{
    return virCgroupGetValueStr(group,
                                VIR_CGROUP_CONTROLLER_CPUSET,
                                "cpuset.mems",
                                mems);
}

E
Eric Blake 已提交
2572

2573 2574 2575 2576 2577 2578 2579 2580
/**
 * virCgroupSetCpusetCpus:
 *
 * @group: The cgroup to set cpuset.cpus for
 * @cpus: the cpus to set
 *
 * Retuens: 0 on success
 */
E
Eric Blake 已提交
2581 2582
int
virCgroupSetCpusetCpus(virCgroupPtr group, const char *cpus)
2583 2584 2585 2586 2587 2588 2589
{
    return virCgroupSetValueStr(group,
                                VIR_CGROUP_CONTROLLER_CPUSET,
                                "cpuset.cpus",
                                cpus);
}

E
Eric Blake 已提交
2590

2591 2592 2593 2594 2595 2596 2597 2598
/**
 * virCgroupGetCpusetCpus:
 *
 * @group: The cgroup to get cpuset.cpus for
 * @cpus: the cpus to get
 *
 * Retuens: 0 on success
 */
E
Eric Blake 已提交
2599 2600
int
virCgroupGetCpusetCpus(virCgroupPtr group, char **cpus)
2601 2602 2603 2604 2605 2606 2607
{
    return virCgroupGetValueStr(group,
                                VIR_CGROUP_CONTROLLER_CPUSET,
                                "cpuset.cpus",
                                cpus);
}

E
Eric Blake 已提交
2608

2609 2610 2611
/**
 * virCgroupDenyAllDevices:
 *
2612
 * @group: The cgroup to deny all permissions, for all devices
2613 2614 2615
 *
 * Returns: 0 on success
 */
E
Eric Blake 已提交
2616 2617
int
virCgroupDenyAllDevices(virCgroupPtr group)
2618 2619
{
    return virCgroupSetValueStr(group,
2620 2621 2622
                                VIR_CGROUP_CONTROLLER_DEVICES,
                                "devices.deny",
                                "a");
2623 2624
}

E
Eric Blake 已提交
2625

2626 2627 2628 2629 2630 2631 2632
/**
 * virCgroupAllowDevice:
 *
 * @group: The cgroup to allow a device for
 * @type: The device type (i.e., 'c' or 'b')
 * @major: The major number of the device
 * @minor: The minor number of the device
2633
 * @perms: Bitwise or of VIR_CGROUP_DEVICE permission bits to allow
2634 2635 2636
 *
 * Returns: 0 on success
 */
E
Eric Blake 已提交
2637 2638 2639
int
virCgroupAllowDevice(virCgroupPtr group, char type, int major, int minor,
                     int perms)
2640
{
2641
    int ret = -1;
2642 2643
    char *devstr = NULL;

2644 2645 2646
    if (virAsprintf(&devstr, "%c %i:%i %s%s%s", type, major, minor,
                    perms & VIR_CGROUP_DEVICE_READ ? "r" : "",
                    perms & VIR_CGROUP_DEVICE_WRITE ? "w" : "",
2647 2648
                    perms & VIR_CGROUP_DEVICE_MKNOD ? "m" : "") < 0)
        goto cleanup;
2649

2650 2651 2652 2653 2654
    if (virCgroupSetValueStr(group,
                             VIR_CGROUP_CONTROLLER_DEVICES,
                             "devices.allow",
                             devstr) < 0)
        goto cleanup;
2655

2656 2657
    ret = 0;

2658
 cleanup:
2659 2660
    VIR_FREE(devstr);
    return ret;
2661
}
2662

E
Eric Blake 已提交
2663

2664 2665 2666 2667 2668 2669
/**
 * virCgroupAllowDeviceMajor:
 *
 * @group: The cgroup to allow an entire device major type for
 * @type: The device type (i.e., 'c' or 'b')
 * @major: The major number of the device type
2670
 * @perms: Bitwise or of VIR_CGROUP_DEVICE permission bits to allow
2671 2672 2673
 *
 * Returns: 0 on success
 */
E
Eric Blake 已提交
2674 2675 2676
int
virCgroupAllowDeviceMajor(virCgroupPtr group, char type, int major,
                          int perms)
2677
{
2678
    int ret = -1;
2679 2680
    char *devstr = NULL;

2681 2682 2683
    if (virAsprintf(&devstr, "%c %i:* %s%s%s", type, major,
                    perms & VIR_CGROUP_DEVICE_READ ? "r" : "",
                    perms & VIR_CGROUP_DEVICE_WRITE ? "w" : "",
2684 2685
                    perms & VIR_CGROUP_DEVICE_MKNOD ? "m" : "") < 0)
        goto cleanup;
2686

2687 2688 2689 2690 2691
    if (virCgroupSetValueStr(group,
                             VIR_CGROUP_CONTROLLER_DEVICES,
                             "devices.allow",
                             devstr) < 0)
        goto cleanup;
2692

2693 2694
    ret = 0;

2695
 cleanup:
2696 2697
    VIR_FREE(devstr);
    return ret;
2698 2699
}

E
Eric Blake 已提交
2700

2701 2702 2703 2704 2705
/**
 * virCgroupAllowDevicePath:
 *
 * @group: The cgroup to allow the device for
 * @path: the device to allow
2706
 * @perms: Bitwise or of VIR_CGROUP_DEVICE permission bits to allow
2707 2708 2709 2710
 *
 * Queries the type of device and its major/minor number, and
 * adds that to the cgroup ACL
 *
2711
 * Returns: 0 on success, 1 if path exists but is not a device, or
2712
 * -1 on error
2713
 */
E
Eric Blake 已提交
2714 2715
int
virCgroupAllowDevicePath(virCgroupPtr group, const char *path, int perms)
2716 2717 2718
{
    struct stat sb;

2719 2720 2721 2722 2723 2724
    if (stat(path, &sb) < 0) {
        virReportSystemError(errno,
                             _("Path '%s' is not accessible"),
                             path);
        return -1;
    }
2725 2726

    if (!S_ISCHR(sb.st_mode) && !S_ISBLK(sb.st_mode))
2727
        return 1;
2728 2729 2730 2731

    return virCgroupAllowDevice(group,
                                S_ISCHR(sb.st_mode) ? 'c' : 'b',
                                major(sb.st_rdev),
2732 2733
                                minor(sb.st_rdev),
                                perms);
2734
}
D
Daniel P. Berrange 已提交
2735

2736 2737 2738 2739 2740 2741 2742 2743

/**
 * virCgroupDenyDevice:
 *
 * @group: The cgroup to deny a device for
 * @type: The device type (i.e., 'c' or 'b')
 * @major: The major number of the device
 * @minor: The minor number of the device
2744
 * @perms: Bitwise or of VIR_CGROUP_DEVICE permission bits to deny
2745 2746 2747
 *
 * Returns: 0 on success
 */
E
Eric Blake 已提交
2748 2749 2750
int
virCgroupDenyDevice(virCgroupPtr group, char type, int major, int minor,
                    int perms)
2751
{
2752
    int ret = -1;
2753 2754
    char *devstr = NULL;

2755 2756 2757
    if (virAsprintf(&devstr, "%c %i:%i %s%s%s", type, major, minor,
                    perms & VIR_CGROUP_DEVICE_READ ? "r" : "",
                    perms & VIR_CGROUP_DEVICE_WRITE ? "w" : "",
2758 2759
                    perms & VIR_CGROUP_DEVICE_MKNOD ? "m" : "") < 0)
        goto cleanup;
2760

2761 2762 2763 2764 2765
    if (virCgroupSetValueStr(group,
                             VIR_CGROUP_CONTROLLER_DEVICES,
                             "devices.deny",
                             devstr) < 0)
        goto cleanup;
2766

2767 2768
    ret = 0;

2769
 cleanup:
2770 2771
    VIR_FREE(devstr);
    return ret;
2772 2773
}

E
Eric Blake 已提交
2774

2775 2776 2777 2778 2779 2780
/**
 * virCgroupDenyDeviceMajor:
 *
 * @group: The cgroup to deny an entire device major type for
 * @type: The device type (i.e., 'c' or 'b')
 * @major: The major number of the device type
2781
 * @perms: Bitwise or of VIR_CGROUP_DEVICE permission bits to deny
2782 2783 2784
 *
 * Returns: 0 on success
 */
E
Eric Blake 已提交
2785 2786 2787
int
virCgroupDenyDeviceMajor(virCgroupPtr group, char type, int major,
                         int perms)
2788
{
2789
    int ret = -1;
2790 2791
    char *devstr = NULL;

2792 2793 2794
    if (virAsprintf(&devstr, "%c %i:* %s%s%s", type, major,
                    perms & VIR_CGROUP_DEVICE_READ ? "r" : "",
                    perms & VIR_CGROUP_DEVICE_WRITE ? "w" : "",
2795 2796
                    perms & VIR_CGROUP_DEVICE_MKNOD ? "m" : "") < 0)
        goto cleanup;
2797

2798 2799 2800 2801 2802
    if (virCgroupSetValueStr(group,
                             VIR_CGROUP_CONTROLLER_DEVICES,
                             "devices.deny",
                             devstr) < 0)
        goto cleanup;
2803

2804 2805
    ret = 0;

2806
 cleanup:
2807 2808
    VIR_FREE(devstr);
    return ret;
2809 2810
}

E
Eric Blake 已提交
2811 2812 2813

int
virCgroupDenyDevicePath(virCgroupPtr group, const char *path, int perms)
2814 2815 2816
{
    struct stat sb;

2817 2818 2819 2820 2821 2822
    if (stat(path, &sb) < 0) {
        virReportSystemError(errno,
                             _("Path '%s' is not accessible"),
                             path);
        return -1;
    }
2823 2824

    if (!S_ISCHR(sb.st_mode) && !S_ISBLK(sb.st_mode))
2825
        return 1;
2826 2827 2828 2829

    return virCgroupDenyDevice(group,
                               S_ISCHR(sb.st_mode) ? 'c' : 'b',
                               major(sb.st_rdev),
2830 2831
                               minor(sb.st_rdev),
                               perms);
2832 2833
}

E
Eric Blake 已提交
2834

2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892
/* 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,
                          unsigned int nvcpupids,
                          unsigned long long *sum_cpu_time,
                          unsigned int num)
{
    int ret = -1;
    size_t i;
    char *buf = NULL;
    virCgroupPtr group_vcpu = NULL;

    for (i = 0; i < nvcpupids; i++) {
        char *pos;
        unsigned long long tmp;
        size_t j;

        if (virCgroupNewVcpu(group, i, false, &group_vcpu) < 0)
            goto cleanup;

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

        pos = buf;
        for (j = 0; j < num; j++) {
            if (virStrToLong_ull(pos, &pos, 10, &tmp) < 0) {
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("cpuacct parse error"));
                goto cleanup;
            }
            sum_cpu_time[j] += tmp;
        }

        virCgroupFree(&group_vcpu);
        VIR_FREE(buf);
    }

    ret = 0;
 cleanup:
    virCgroupFree(&group_vcpu);
    VIR_FREE(buf);
    return ret;
}


2893 2894 2895 2896 2897
int
virCgroupGetPercpuStats(virCgroupPtr group,
                        virTypedParameterPtr params,
                        unsigned int nparams,
                        int start_cpu,
2898 2899
                        unsigned int ncpus,
                        unsigned int nvcpupids)
2900 2901 2902
{
    int rv = -1;
    size_t i;
2903
    int need_cpus, total_cpus;
2904 2905
    char *pos;
    char *buf = NULL;
2906
    unsigned long long *sum_cpu_time = NULL;
2907 2908 2909 2910 2911
    virTypedParameterPtr ent;
    int param_idx;
    unsigned long long cpu_time;

    /* return the number of supported params */
2912 2913 2914 2915 2916 2917
    if (nparams == 0 && ncpus != 0) {
        if (nvcpupids == 0)
            return CGROUP_NB_PER_CPU_STAT_PARAM;
        else
            return CGROUP_NB_PER_CPU_STAT_PARAM + 1;
    }
2918 2919

    /* To parse account file, we need to know how many cpus are present.  */
J
Ján Tomko 已提交
2920
    if ((total_cpus = nodeGetCPUCount()) < 0)
2921 2922
        return rv;

J
Ján Tomko 已提交
2923 2924
    if (ncpus == 0)
        return total_cpus;
2925

2926
    if (start_cpu >= total_cpus) {
2927 2928
        virReportError(VIR_ERR_INVALID_ARG,
                       _("start_cpu %d larger than maximum of %d"),
2929
                       start_cpu, total_cpus - 1);
2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941
        goto cleanup;
    }

    /* we get percpu cputime accounting info. */
    if (virCgroupGetCpuacctPercpuUsage(group, &buf))
        goto cleanup;
    pos = buf;

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

    /* number of cpus to compute */
J
Ján Tomko 已提交
2942
    need_cpus = MIN(total_cpus, start_cpu + ncpus);
2943

J
Ján Tomko 已提交
2944
    for (i = 0; i < need_cpus; i++) {
2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957
        if (virStrToLong_ull(pos, &pos, 10, &cpu_time) < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("cpuacct parse error"));
            goto cleanup;
        }
        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)
            goto cleanup;
    }

2958 2959 2960 2961 2962
    if (nvcpupids == 0 || param_idx + 1 >= nparams)
        goto success;
    /* return percpu vcputime in index 1 */
    param_idx++;

J
Ján Tomko 已提交
2963
    if (VIR_ALLOC_N(sum_cpu_time, need_cpus) < 0)
2964
        goto cleanup;
J
Ján Tomko 已提交
2965
    if (virCgroupGetPercpuVcpuSum(group, nvcpupids, sum_cpu_time, need_cpus) < 0)
2966 2967
        goto cleanup;

J
Ján Tomko 已提交
2968
    for (i = start_cpu; i < need_cpus; i++) {
2969 2970 2971 2972
        if (virTypedParameterAssign(&params[(i - start_cpu) * nparams +
                                            param_idx],
                                    VIR_DOMAIN_CPU_STATS_VCPUTIME,
                                    VIR_TYPED_PARAM_ULLONG,
J
Ján Tomko 已提交
2973
                                    sum_cpu_time[i]) < 0)
2974 2975 2976 2977
            goto cleanup;
    }

 success:
2978
    rv = param_idx + 1;
2979

2980
 cleanup:
2981
    VIR_FREE(sum_cpu_time);
2982 2983 2984 2985
    VIR_FREE(buf);
    return rv;
}

2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035

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
3036 3037
int
virCgroupSetCpuShares(virCgroupPtr group, unsigned long long shares)
3038
{
3039 3040
    return virCgroupSetValueU64(group,
                                VIR_CGROUP_CONTROLLER_CPU,
D
Daniel P. Berrange 已提交
3041
                                "cpu.shares", shares);
3042 3043
}

E
Eric Blake 已提交
3044 3045 3046

int
virCgroupGetCpuShares(virCgroupPtr group, unsigned long long *shares)
3047
{
3048 3049
    return virCgroupGetValueU64(group,
                                VIR_CGROUP_CONTROLLER_CPU,
D
Daniel P. Berrange 已提交
3050
                                "cpu.shares", shares);
3051
}
3052

E
Eric Blake 已提交
3053

3054 3055 3056 3057 3058 3059 3060 3061
/**
 * 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
3062 3063
int
virCgroupSetCpuCfsPeriod(virCgroupPtr group, unsigned long long cfs_period)
3064
{
3065
    /* The cfs_period should be greater or equal than 1ms, and less or equal
3066 3067
     * than 1s.
     */
3068 3069 3070 3071 3072 3073
    if (cfs_period < 1000 || cfs_period > 1000000) {
        virReportError(VIR_ERR_INVALID_ARG,
                       _("cfs_period '%llu' must be in range (1000, 1000000)"),
                       cfs_period);
        return -1;
    }
3074 3075 3076 3077 3078 3079

    return virCgroupSetValueU64(group,
                                VIR_CGROUP_CONTROLLER_CPU,
                                "cpu.cfs_period_us", cfs_period);
}

E
Eric Blake 已提交
3080

3081 3082 3083 3084 3085 3086 3087 3088
/**
 * 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
3089 3090
int
virCgroupGetCpuCfsPeriod(virCgroupPtr group, unsigned long long *cfs_period)
3091 3092 3093 3094 3095 3096
{
    return virCgroupGetValueU64(group,
                                VIR_CGROUP_CONTROLLER_CPU,
                                "cpu.cfs_period_us", cfs_period);
}

E
Eric Blake 已提交
3097

3098 3099 3100 3101 3102 3103 3104 3105 3106
/**
 * 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
3107 3108
int
virCgroupSetCpuCfsQuota(virCgroupPtr group, long long cfs_quota)
3109
{
3110 3111 3112 3113 3114 3115 3116 3117
    /* The cfs_quota should be greater or equal than 1ms */
    if (cfs_quota >= 0 &&
        (cfs_quota < 1000 ||
         cfs_quota > ULLONG_MAX / 1000)) {
        virReportError(VIR_ERR_INVALID_ARG,
                       _("cfs_quota '%lld' must be in range (1000, %llu)"),
                       cfs_quota, ULLONG_MAX / 1000);
        return -1;
3118 3119 3120 3121 3122 3123 3124
    }

    return virCgroupSetValueI64(group,
                                VIR_CGROUP_CONTROLLER_CPU,
                                "cpu.cfs_quota_us", cfs_quota);
}

E
Eric Blake 已提交
3125 3126 3127

int
virCgroupGetCpuacctPercpuUsage(virCgroupPtr group, char **usage)
3128 3129 3130 3131 3132
{
    return virCgroupGetValueStr(group, VIR_CGROUP_CONTROLLER_CPUACCT,
                                "cpuacct.usage_percpu", usage);
}

E
Eric Blake 已提交
3133

E
Eric Blake 已提交
3134 3135 3136 3137 3138 3139
int
virCgroupRemoveRecursively(char *grppath)
{
    DIR *grpdir;
    struct dirent *ent;
    int rc = 0;
E
Eric Blake 已提交
3140
    int direrr;
E
Eric Blake 已提交
3141 3142 3143 3144 3145 3146 3147 3148 3149 3150

    grpdir = opendir(grppath);
    if (grpdir == NULL) {
        if (errno == ENOENT)
            return 0;
        rc = -errno;
        VIR_ERROR(_("Unable to open %s (%d)"), grppath, errno);
        return rc;
    }

E
Eric Blake 已提交
3151 3152 3153
    /* 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) {
E
Eric Blake 已提交
3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167
        char *path;

        if (ent->d_name[0] == '.') continue;
        if (ent->d_type != DT_DIR) continue;

        if (virAsprintf(&path, "%s/%s", grppath, ent->d_name) == -1) {
            rc = -ENOMEM;
            break;
        }
        rc = virCgroupRemoveRecursively(path);
        VIR_FREE(path);
        if (rc != 0)
            break;
    }
E
Eric Blake 已提交
3168 3169 3170 3171 3172
    if (direrr < 0) {
        rc = -errno;
        VIR_ERROR(_("Failed to readdir for %s (%d)"), grppath, errno);
    }

E
Eric Blake 已提交
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
    closedir(grpdir);

    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)
{
    int rc = 0;
    size_t i;
    char *grppath = NULL;

    VIR_DEBUG("Removing cgroup %s", group->path);
    for (i = 0; i < VIR_CGROUP_CONTROLLER_LAST; i++) {
        /* Skip over controllers not mounted */
        if (!group->controllers[i].mountPoint)
            continue;

        /* We must never rmdir() in systemd's hierarchy */
        if (i == VIR_CGROUP_CONTROLLER_SYSTEMD)
            continue;

        /* Don't delete the root group, if we accidentally
           ended up in it for some reason */
        if (STREQ(group->controllers[i].placement, "/"))
            continue;

        if (virCgroupPathOfController(group,
                                      i,
                                      NULL,
                                      &grppath) != 0)
            continue;

        VIR_DEBUG("Removing cgroup %s and all child cgroups", grppath);
        rc = virCgroupRemoveRecursively(grppath);
        VIR_FREE(grppath);
    }
    VIR_DEBUG("Done removing cgroup %s", group->path);

    return rc;
}


3235 3236 3237
/*
 * Returns 1 if some PIDs are killed, 0 if none are killed, or -1 on error
 */
E
Eric Blake 已提交
3238 3239
static int
virCgroupKillInternal(virCgroupPtr group, int signum, virHashTablePtr pids)
3240
{
3241
    int ret = -1;
3242
    bool killedAny = false;
3243 3244
    char *keypath = NULL;
    bool done = false;
E
Eric Blake 已提交
3245 3246 3247
    FILE *fp = NULL;
    VIR_DEBUG("group=%p path=%s signum=%d pids=%p",
              group, group->path, signum, pids);
3248

3249
    if (virCgroupPathOfController(group, -1, "tasks", &keypath) < 0)
3250
        return -1;
3251 3252 3253 3254 3255 3256 3257

    /* 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"))) {
3258 3259 3260 3261 3262 3263
            if (errno == ENOENT) {
                VIR_DEBUG("No file %s, assuming done", keypath);
                killedAny = false;
                goto done;
            }

3264 3265 3266
            virReportSystemError(errno,
                                 _("Failed to read %s"),
                                 keypath);
3267 3268 3269
            goto cleanup;
        } else {
            while (!feof(fp)) {
3270 3271
                unsigned long pid_value;
                if (fscanf(fp, "%lu", &pid_value) != 1) {
3272 3273
                    if (feof(fp))
                        break;
3274 3275 3276
                    virReportSystemError(errno,
                                         _("Failed to read %s"),
                                         keypath);
E
Eric Blake 已提交
3277
                    goto cleanup;
3278
                }
3279
                if (virHashLookup(pids, (void*)pid_value))
3280 3281
                    continue;

3282 3283 3284
                VIR_DEBUG("pid=%lu", pid_value);
                /* Cgroups is a Linux concept, so this cast is safe.  */
                if (kill((pid_t)pid_value, signum) < 0) {
3285
                    if (errno != ESRCH) {
3286 3287 3288
                        virReportSystemError(errno,
                                             _("Failed to kill process %lu"),
                                             pid_value);
3289 3290 3291 3292
                        goto cleanup;
                    }
                    /* Leave RC == 0 since we didn't kill one */
                } else {
3293
                    killedAny = true;
3294 3295 3296
                    done = false;
                }

3297
                ignore_value(virHashAddEntry(pids, (void*)pid_value, (void*)1));
3298 3299 3300 3301 3302
            }
            VIR_FORCE_FCLOSE(fp);
        }
    }

3303
 done:
3304
    ret = killedAny ? 1 : 0;
3305

3306
 cleanup:
3307
    VIR_FREE(keypath);
E
Eric Blake 已提交
3308
    VIR_FORCE_FCLOSE(fp);
3309

3310
    return ret;
3311 3312 3313
}


E
Eric Blake 已提交
3314 3315
static uint32_t
virCgroupPidCode(const void *name, uint32_t seed)
3316
{
3317 3318
    unsigned long pid_value = (unsigned long)(intptr_t)name;
    return virHashCodeGen(&pid_value, sizeof(pid_value), seed);
3319
}
E
Eric Blake 已提交
3320 3321 3322 3323


static bool
virCgroupPidEqual(const void *namea, const void *nameb)
3324 3325 3326
{
    return namea == nameb;
}
E
Eric Blake 已提交
3327 3328 3329 3330


static void *
virCgroupPidCopy(const void *name)
3331 3332 3333 3334
{
    return (void*)name;
}

E
Eric Blake 已提交
3335

3336
/*
3337
 * Returns 1 if some PIDs are killed, 0 if none are killed, or -1 on error
3338
 */
E
Eric Blake 已提交
3339 3340
int
virCgroupKill(virCgroupPtr group, int signum)
3341 3342
{
    VIR_DEBUG("group=%p path=%s signum=%d", group, group->path, signum);
3343
    int ret;
3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354
    /* The 'tasks' file in cgroups can contain duplicated
     * pids, so we use a hash to track which we've already
     * killed.
     */
    virHashTablePtr pids = virHashCreateFull(100,
                                             NULL,
                                             virCgroupPidCode,
                                             virCgroupPidEqual,
                                             virCgroupPidCopy,
                                             NULL);

3355
    ret = virCgroupKillInternal(group, signum, pids);
3356 3357 3358

    virHashFree(pids);

3359
    return ret;
3360 3361 3362
}


E
Eric Blake 已提交
3363 3364 3365 3366 3367
static int
virCgroupKillRecursiveInternal(virCgroupPtr group,
                               int signum,
                               virHashTablePtr pids,
                               bool dormdir)
3368
{
3369
    int ret = -1;
3370
    int rc;
3371
    bool killedAny = false;
3372
    char *keypath = NULL;
3373
    DIR *dp = NULL;
3374 3375
    virCgroupPtr subgroup = NULL;
    struct dirent *ent;
E
Eric Blake 已提交
3376
    int direrr;
E
Eric Blake 已提交
3377 3378
    VIR_DEBUG("group=%p path=%s signum=%d pids=%p",
              group, group->path, signum, pids);
3379

3380
    if (virCgroupPathOfController(group, -1, "", &keypath) < 0)
3381
        return -1;
3382

3383
    if ((rc = virCgroupKillInternal(group, signum, pids)) < 0)
3384
        goto cleanup;
3385 3386
    if (rc == 1)
        killedAny = true;
3387

3388
    VIR_DEBUG("Iterate over children of %s (killedAny=%d)", keypath, killedAny);
3389
    if (!(dp = opendir(keypath))) {
3390 3391 3392 3393 3394
        if (errno == ENOENT) {
            VIR_DEBUG("Path %s does not exist, assuming done", keypath);
            killedAny = false;
            goto done;
        }
3395 3396
        virReportSystemError(errno,
                             _("Cannot open %s"), keypath);
3397
        goto cleanup;
3398 3399
    }

E
Eric Blake 已提交
3400
    while ((direrr = virDirRead(dp, &ent, keypath)) > 0) {
3401 3402 3403 3404 3405 3406 3407 3408 3409
        if (STREQ(ent->d_name, "."))
            continue;
        if (STREQ(ent->d_name, ".."))
            continue;
        if (ent->d_type != DT_DIR)
            continue;

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

3410
        if (virCgroupNew(-1, ent->d_name, group, -1, &subgroup) < 0)
3411 3412
            goto cleanup;

E
Eric Blake 已提交
3413 3414
        if ((rc = virCgroupKillRecursiveInternal(subgroup, signum, pids,
                                                 true)) < 0)
3415 3416
            goto cleanup;
        if (rc == 1)
3417
            killedAny = true;
3418 3419 3420 3421 3422 3423

        if (dormdir)
            virCgroupRemove(subgroup);

        virCgroupFree(&subgroup);
    }
E
Eric Blake 已提交
3424 3425
    if (direrr < 0)
        goto cleanup;
3426

3427
 done:
3428
    ret = killedAny ? 1 : 0;
3429

3430
 cleanup:
3431
    virCgroupFree(&subgroup);
3432 3433 3434
    VIR_FREE(keypath);
    if (dp)
        closedir(dp);
3435

3436
    return ret;
3437 3438
}

E
Eric Blake 已提交
3439 3440 3441

int
virCgroupKillRecursive(virCgroupPtr group, int signum)
3442
{
3443
    int ret;
3444 3445 3446 3447 3448 3449 3450 3451
    VIR_DEBUG("group=%p path=%s signum=%d", group, group->path, signum);
    virHashTablePtr pids = virHashCreateFull(100,
                                             NULL,
                                             virCgroupPidCode,
                                             virCgroupPidEqual,
                                             virCgroupPidCopy,
                                             NULL);

3452
    ret = virCgroupKillRecursiveInternal(group, signum, pids, false);
3453 3454 3455

    virHashFree(pids);

3456
    return ret;
3457 3458 3459
}


E
Eric Blake 已提交
3460 3461
int
virCgroupKillPainfully(virCgroupPtr group)
3462
{
3463
    size_t i;
3464
    int ret;
3465
    VIR_DEBUG("cgroup=%p path=%s", group, group->path);
3466
    for (i = 0; i < 15; i++) {
3467 3468 3469 3470 3471 3472
        int signum;
        if (i == 0)
            signum = SIGTERM;
        else if (i == 8)
            signum = SIGKILL;
        else
J
Ján Tomko 已提交
3473
            signum = 0; /* Just check for existence */
3474

3475 3476 3477 3478
        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)
3479 3480 3481 3482
            break;

        usleep(200 * 1000);
    }
3483 3484
    VIR_DEBUG("Complete %d", ret);
    return ret;
3485
}
3486

E
Eric Blake 已提交
3487 3488 3489

static char *
virCgroupIdentifyRoot(virCgroupPtr group)
3490 3491 3492 3493
{
    char *ret = NULL;
    size_t i;

3494
    for (i = 0; i < VIR_CGROUP_CONTROLLER_LAST; i++) {
3495 3496 3497 3498 3499 3500 3501 3502 3503 3504
        char *tmp;
        if (!group->controllers[i].mountPoint)
            continue;
        if (!(tmp = strrchr(group->controllers[i].mountPoint, '/'))) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Could not find directory separator in %s"),
                           group->controllers[i].mountPoint);
            return NULL;
        }

3505 3506 3507
        if (VIR_STRNDUP(ret, group->controllers[i].mountPoint,
                        tmp - group->controllers[i].mountPoint) < 0)
            return NULL;
3508 3509 3510 3511 3512 3513 3514 3515 3516
        return ret;
    }

    virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                   _("Could not find any mounted controllers"));
    return NULL;
}


3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586
/**
 * 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)
{
    return virCgroupGetValueI64(group,
                                VIR_CGROUP_CONTROLLER_CPU,
                                "cpu.cfs_quota_us", cfs_quota);
}


int
virCgroupGetCpuacctUsage(virCgroupPtr group, unsigned long long *usage)
{
    return virCgroupGetValueU64(group,
                                VIR_CGROUP_CONTROLLER_CPUACCT,
                                "cpuacct.usage", usage);
}


int
virCgroupGetCpuacctStat(virCgroupPtr group, unsigned long long *user,
                        unsigned long long *sys)
{
    char *str;
    char *p;
    int ret = -1;
    static double scale = -1.0;

    if (virCgroupGetValueStr(group, VIR_CGROUP_CONTROLLER_CPUACCT,
                             "cpuacct.stat", &str) < 0)
        return -1;

    if (!(p = STRSKIP(str, "user ")) ||
        virStrToLong_ull(p, &p, 10, user) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Cannot parse user stat '%s'"),
                       p);
        goto cleanup;
    }
    if (!(p = STRSKIP(p, "\nsystem ")) ||
        virStrToLong_ull(p, NULL, 10, sys) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Cannot parse sys stat '%s'"),
                       p);
        goto cleanup;
    }
    /* times reported are in system ticks (generally 100 Hz), but that
     * rate can theoretically vary between machines.  Scale things
     * into approximate nanoseconds.  */
    if (scale < 0) {
        long ticks_per_sec = sysconf(_SC_CLK_TCK);
        if (ticks_per_sec == -1) {
            virReportSystemError(errno, "%s",
                                 _("Cannot determine system clock HZ"));
            goto cleanup;
        }
        scale = 1000000000.0 / ticks_per_sec;
    }
    *user *= scale;
    *sys *= scale;

    ret = 0;
3587
 cleanup:
3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610
    VIR_FREE(str);
    return ret;
}


int
virCgroupSetFreezerState(virCgroupPtr group, const char *state)
{
    return virCgroupSetValueStr(group,
                                VIR_CGROUP_CONTROLLER_FREEZER,
                                "freezer.state", state);
}


int
virCgroupGetFreezerState(virCgroupPtr group, char **state)
{
    return virCgroupGetValueStr(group,
                                VIR_CGROUP_CONTROLLER_FREEZER,
                                "freezer.state", state);
}


E
Eric Blake 已提交
3611 3612 3613
int
virCgroupIsolateMount(virCgroupPtr group, const char *oldroot,
                      const char *mountopts)
3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632
{
    int ret = -1;
    size_t i;
    char *opts = NULL;
    char *root = NULL;

    if (!(root = virCgroupIdentifyRoot(group)))
        return -1;

    VIR_DEBUG("Mounting cgroups at '%s'", root);

    if (virFileMakePath(root) < 0) {
        virReportSystemError(errno,
                             _("Unable to create directory %s"),
                             root);
        goto cleanup;
    }

    if (virAsprintf(&opts,
3633
                    "mode=755,size=65536%s", mountopts) < 0)
3634 3635 3636 3637 3638 3639 3640 3641 3642
        goto cleanup;

    if (mount("tmpfs", root, "tmpfs", MS_NOSUID|MS_NODEV|MS_NOEXEC, opts) < 0) {
        virReportSystemError(errno,
                             _("Failed to mount %s on %s type %s"),
                             "tmpfs", root, "tmpfs");
        goto cleanup;
    }

3643
    for (i = 0; i < VIR_CGROUP_CONTROLLER_LAST; i++) {
3644 3645 3646 3647 3648 3649 3650 3651
        if (!group->controllers[i].mountPoint)
            continue;

        if (!virFileExists(group->controllers[i].mountPoint)) {
            char *src;
            if (virAsprintf(&src, "%s%s%s",
                            oldroot,
                            group->controllers[i].mountPoint,
3652
                            group->controllers[i].placement) < 0)
3653 3654
                goto cleanup;

E
Eric Blake 已提交
3655 3656
            VIR_DEBUG("Create mount point '%s'",
                      group->controllers[i].mountPoint);
3657 3658 3659 3660 3661 3662 3663 3664
            if (virFileMakePath(group->controllers[i].mountPoint) < 0) {
                virReportSystemError(errno,
                                     _("Unable to create directory %s"),
                                     group->controllers[i].mountPoint);
                VIR_FREE(src);
                goto cleanup;
            }

E
Eric Blake 已提交
3665 3666
            if (mount(src, group->controllers[i].mountPoint, NULL, MS_BIND,
                      NULL) < 0) {
3667 3668 3669
                virReportSystemError(errno,
                                     _("Failed to bind cgroup '%s' on '%s'"),
                                     src, group->controllers[i].mountPoint);
3670
                VIR_FREE(src);
3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692
                goto cleanup;
            }

            VIR_FREE(src);
        }

        if (group->controllers[i].linkPoint) {
            VIR_DEBUG("Link mount point '%s' to '%s'",
                      group->controllers[i].mountPoint,
                      group->controllers[i].linkPoint);
            if (symlink(group->controllers[i].mountPoint,
                        group->controllers[i].linkPoint) < 0) {
                virReportSystemError(errno,
                                     _("Unable to symlink directory %s to %s"),
                                     group->controllers[i].mountPoint,
                                     group->controllers[i].linkPoint);
                return -1;
            }
        }
    }
    ret = 0;

3693
 cleanup:
3694 3695 3696 3697
    VIR_FREE(root);
    VIR_FREE(opts);
    return ret;
}
3698 3699


3700 3701 3702 3703 3704 3705 3706 3707 3708
int virCgroupSetOwner(virCgroupPtr cgroup,
                      uid_t uid,
                      gid_t gid,
                      int controllers)
{
    int ret = -1;
    size_t i;
    char *base = NULL, *entry = NULL;
    DIR *dh = NULL;
E
Eric Blake 已提交
3709
    int direrr;
3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729

    for (i = 0; i < VIR_CGROUP_CONTROLLER_LAST; i++) {
        struct dirent *de;

        if (!((1 << i) & controllers))
            continue;

        if (!cgroup->controllers[i].mountPoint)
            continue;

        if (virAsprintf(&base, "%s%s", cgroup->controllers[i].mountPoint,
                        cgroup->controllers[i].placement) < 0)
            goto cleanup;

        if (!(dh = opendir(base))) {
            virReportSystemError(errno,
                                 _("Unable to open dir '%s'"), base);
            goto cleanup;
        }

E
Eric Blake 已提交
3730
        while ((direrr = virDirRead(dh, &de, base)) > 0) {
3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746
            if (STREQ(de->d_name, ".") ||
                STREQ(de->d_name, ".."))
                continue;

            if (virAsprintf(&entry, "%s/%s", base, de->d_name) < 0)
                goto cleanup;

            if (chown(entry, uid, gid) < 0) {
                virReportSystemError(errno,
                                     _("cannot chown '%s' to (%u, %u)"),
                                     entry, uid, gid);
                goto cleanup;
            }

            VIR_FREE(entry);
        }
E
Eric Blake 已提交
3747 3748
        if (direrr < 0)
            goto cleanup;
3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772

        if (chown(base, uid, gid) < 0) {
            virReportSystemError(errno,
                                 _("cannot chown '%s' to (%u, %u)"),
                                 base, uid, gid);
            goto cleanup;
        }

        VIR_FREE(base);
        closedir(dh);
        dh = NULL;
    }

    ret = 0;

 cleanup:
    if (dh)
        closedir(dh);
    VIR_FREE(entry);
    VIR_FREE(base);
    return ret;
}


3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796
/**
 * 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)
{
    char *path = NULL;
    int ret = false;

    if (!cgroup)
        return false;

    if (virCgroupPathOfController(cgroup, VIR_CGROUP_CONTROLLER_CPU,
                                  "cpu.cfs_period_us", &path) < 0) {
        virResetLastError();
        goto cleanup;
    }

    ret = virFileExists(path);

3797
 cleanup:
3798 3799 3800 3801 3802
    VIR_FREE(path);
    return ret;
}


3803 3804
#else /* !VIR_CGROUP_SUPPORTED */

3805 3806 3807 3808 3809 3810 3811
bool
virCgroupAvailable(void)
{
    return false;
}


3812 3813 3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870 3871 3872 3873 3874 3875 3876 3877 3878 3879
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;
}


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


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


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;
}


3880 3881 3882 3883 3884 3885 3886 3887 3888 3889 3890 3891 3892
int
virCgroupNewDetectMachine(const char *name ATTRIBUTE_UNUSED,
                          const char *drivername ATTRIBUTE_UNUSED,
                          pid_t pid ATTRIBUTE_UNUSED,
                          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;
}

3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917
int
virCgroupNewMachine(const char *name ATTRIBUTE_UNUSED,
                    const char *drivername ATTRIBUTE_UNUSED,
                    bool privileged ATTRIBUTE_UNUSED,
                    const unsigned char *uuid ATTRIBUTE_UNUSED,
                    const char *rootdir ATTRIBUTE_UNUSED,
                    pid_t pidleader ATTRIBUTE_UNUSED,
                    bool isContainer ATTRIBUTE_UNUSED,
                    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;
}

3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934

void
virCgroupFree(virCgroupPtr *group ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENXIO, "%s",
                         _("Control groups not supported on this platform"));
}


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


3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946
int
virCgroupPathOfController(virCgroupPtr group ATTRIBUTE_UNUSED,
                          int controller ATTRIBUTE_UNUSED,
                          const char *key ATTRIBUTE_UNUSED,
                          char **path ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENXIO, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977
int
virCgroupAddTask(virCgroupPtr group ATTRIBUTE_UNUSED,
                 pid_t pid ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENXIO, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


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


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


3978 3979 3980 3981 3982 3983 3984 3985 3986 3987 3988 3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002 4003 4004
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;
}


4005 4006 4007 4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024
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;
}


4025 4026 4027 4028 4029 4030 4031 4032 4033 4034
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;
}

4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064 4065 4066 4067 4068 4069 4070 4071 4072 4073 4074
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;
}

4075

4076 4077 4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 4094 4095 4096 4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108 4109 4110 4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 4127 4128 4129 4130 4131 4132 4133 4134 4135 4136 4137 4138 4139 4140 4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160 4161 4162 4163 4164 4165 4166 4167 4168 4169 4170 4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 4181 4182 4183 4184 4185 4186 4187 4188 4189 4190 4191 4192 4193 4194 4195 4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218 4219 4220 4221 4222 4223 4224 4225 4226 4227 4228 4229 4230 4231 4232 4233 4234 4235 4236 4237 4238 4239
int
virCgroupSetMemory(virCgroupPtr group ATTRIBUTE_UNUSED,
                   unsigned long long kb ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


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;
}


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;
}


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;
}


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


4240 4241 4242 4243 4244 4245 4246 4247 4248 4249 4250
int
virCgroupAllowDevicePath(virCgroupPtr group ATTRIBUTE_UNUSED,
                         const char *path ATTRIBUTE_UNUSED,
                         int perms ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


4251 4252 4253 4254 4255 4256 4257 4258 4259 4260 4261 4262 4263 4264 4265 4266 4267 4268 4269 4270 4271 4272 4273 4274 4275
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;
}


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


4276 4277 4278 4279 4280 4281 4282 4283 4284 4285 4286
int
virCgroupDenyDevicePath(virCgroupPtr group ATTRIBUTE_UNUSED,
                        const char *path ATTRIBUTE_UNUSED,
                        int perms ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302 4303 4304 4305 4306 4307 4308 4309 4310 4311 4312 4313 4314 4315 4316 4317 4318 4319 4320 4321 4322 4323 4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334 4335 4336
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;
}


4337 4338 4339 4340 4341 4342 4343 4344 4345
int
virCgroupRemoveRecursively(char *grppath ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENXIO, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


4346 4347 4348 4349 4350 4351 4352 4353 4354
int
virCgroupRemove(virCgroupPtr group ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENXIO, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


4355 4356 4357 4358 4359 4360 4361 4362 4363 4364 4365 4366 4367 4368 4369 4370 4371 4372 4373 4374 4375 4376 4377 4378 4379 4380 4381 4382 4383
int
virCgroupKill(virCgroupPtr group ATTRIBUTE_UNUSED,
              int signum ATTRIBUTE_UNUSED)
{
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
}


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;
}


4384 4385 4386 4387 4388 4389 4390 4391 4392 4393 4394 4395 4396 4397 4398 4399 4400 4401 4402 4403 4404 4405 4406 4407 4408 4409 4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420 4421 4422 4423 4424
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;
}


4425 4426 4427 4428 4429 4430 4431 4432 4433 4434 4435
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;
}


4436 4437 4438 4439 4440 4441 4442 4443 4444 4445 4446 4447 4448 4449 4450 4451 4452 4453 4454 4455
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
4456 4457 4458 4459
int
virCgroupIsolateMount(virCgroupPtr group ATTRIBUTE_UNUSED,
                      const char *oldroot ATTRIBUTE_UNUSED,
                      const char *mountopts ATTRIBUTE_UNUSED)
4460
{
4461 4462 4463
    virReportSystemError(ENOSYS, "%s",
                         _("Control groups not supported on this platform"));
    return -1;
4464
}
4465

4466 4467 4468 4469 4470 4471 4472 4473

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

E
Eric Blake 已提交
4474 4475 4476 4477 4478 4479

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
4480 4481
                        unsigned int ncpus ATTRIBUTE_UNUSED,
                        unsigned int nvcpupids ATTRIBUTE_UNUSED)
E
Eric Blake 已提交
4482 4483 4484 4485 4486 4487 4488 4489 4490 4491 4492 4493 4494 4495 4496 4497 4498 4499
{
    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;
}

4500
#endif /* !VIR_CGROUP_SUPPORTED */