+ LIST_REMOVE(device_bandwidths, c->blockio_device_bandwidths, b);
+ free(b->path);
+ free(b);
+}
+
+void cgroup_context_done(CGroupContext *c) {
+ assert(c);
+
+ while (c->blockio_device_weights)
+ cgroup_context_free_blockio_device_weight(c, c->blockio_device_weights);
+
+ while (c->blockio_device_bandwidths)
+ cgroup_context_free_blockio_device_bandwidth(c, c->blockio_device_bandwidths);
+
+ while (c->device_allow)
+ cgroup_context_free_device_allow(c, c->device_allow);
+}
+
+void cgroup_context_dump(CGroupContext *c, FILE* f, const char *prefix) {
+ CGroupBlockIODeviceBandwidth *b;
+ CGroupBlockIODeviceWeight *w;
+ CGroupDeviceAllow *a;
+ char u[FORMAT_TIMESPAN_MAX];
+
+ assert(c);
+ assert(f);
+
+ prefix = strempty(prefix);
+
+ fprintf(f,
+ "%sCPUAccounting=%s\n"
+ "%sBlockIOAccounting=%s\n"
+ "%sMemoryAccounting=%s\n"
+ "%sTasksAccounting=%s\n"
+ "%sCPUShares=%" PRIu64 "\n"
+ "%sStartupCPUShares=%" PRIu64 "\n"
+ "%sCPUQuotaPerSecSec=%s\n"
+ "%sBlockIOWeight=%" PRIu64 "\n"
+ "%sStartupBlockIOWeight=%" PRIu64 "\n"
+ "%sMemoryLimit=%" PRIu64 "\n"
+ "%sTasksMax=%" PRIu64 "\n"
+ "%sDevicePolicy=%s\n"
+ "%sDelegate=%s\n",
+ prefix, yes_no(c->cpu_accounting),
+ prefix, yes_no(c->blockio_accounting),
+ prefix, yes_no(c->memory_accounting),
+ prefix, yes_no(c->tasks_accounting),
+ prefix, c->cpu_shares,
+ prefix, c->startup_cpu_shares,
+ prefix, format_timespan(u, sizeof(u), c->cpu_quota_per_sec_usec, 1),
+ prefix, c->blockio_weight,
+ prefix, c->startup_blockio_weight,
+ prefix, c->memory_limit,
+ prefix, c->tasks_max,
+ prefix, cgroup_device_policy_to_string(c->device_policy),
+ prefix, yes_no(c->delegate));
+
+ LIST_FOREACH(device_allow, a, c->device_allow)
+ fprintf(f,
+ "%sDeviceAllow=%s %s%s%s\n",
+ prefix,
+ a->path,
+ a->r ? "r" : "", a->w ? "w" : "", a->m ? "m" : "");
+
+ LIST_FOREACH(device_weights, w, c->blockio_device_weights)
+ fprintf(f,
+ "%sBlockIODeviceWeight=%s %" PRIu64,
+ prefix,
+ w->path,
+ w->weight);
+
+ LIST_FOREACH(device_bandwidths, b, c->blockio_device_bandwidths) {
+ char buf[FORMAT_BYTES_MAX];
+
+ fprintf(f,
+ "%s%s=%s %s\n",
+ prefix,
+ b->read ? "BlockIOReadBandwidth" : "BlockIOWriteBandwidth",
+ b->path,
+ format_bytes(buf, sizeof(buf), b->bandwidth));
+ }
+}
+
+static int lookup_blkio_device(const char *p, dev_t *dev) {
+ struct stat st;
+ int r;
+
+ assert(p);
+ assert(dev);
+
+ r = stat(p, &st);
+ if (r < 0)
+ return log_warning_errno(errno, "Couldn't stat device %s: %m", p);
+
+ if (S_ISBLK(st.st_mode))
+ *dev = st.st_rdev;
+ else if (major(st.st_dev) != 0) {
+ /* If this is not a device node then find the block
+ * device this file is stored on */
+ *dev = st.st_dev;
+
+ /* If this is a partition, try to get the originating
+ * block device */
+ block_get_whole_disk(*dev, dev);
+ } else {
+ log_warning("%s is not a block device and file system block device cannot be determined or is not local.", p);
+ return -ENODEV;
+ }
+
+ return 0;
+}
+
+static int whitelist_device(const char *path, const char *node, const char *acc) {
+ char buf[2+DECIMAL_STR_MAX(dev_t)*2+2+4];
+ struct stat st;
+ int r;
+
+ assert(path);
+ assert(acc);
+
+ if (stat(node, &st) < 0) {
+ log_warning("Couldn't stat device %s", node);
+ return -errno;
+ }
+
+ if (!S_ISCHR(st.st_mode) && !S_ISBLK(st.st_mode)) {
+ log_warning("%s is not a device.", node);
+ return -ENODEV;
+ }
+
+ sprintf(buf,
+ "%c %u:%u %s",
+ S_ISCHR(st.st_mode) ? 'c' : 'b',
+ major(st.st_rdev), minor(st.st_rdev),
+ acc);
+
+ r = cg_set_attribute("devices", path, "devices.allow", buf);
+ if (r < 0)
+ log_full_errno(IN_SET(r, -ENOENT, -EROFS, -EINVAL, -EACCES) ? LOG_DEBUG : LOG_WARNING, r,
+ "Failed to set devices.allow on %s: %m", path);
+
+ return r;
+}
+
+static int whitelist_major(const char *path, const char *name, char type, const char *acc) {
+ _cleanup_fclose_ FILE *f = NULL;
+ char line[LINE_MAX];
+ bool good = false;
+ int r;
+
+ assert(path);
+ assert(acc);
+ assert(type == 'b' || type == 'c');
+
+ f = fopen("/proc/devices", "re");
+ if (!f)
+ return log_warning_errno(errno, "Cannot open /proc/devices to resolve %s (%c): %m", name, type);
+
+ FOREACH_LINE(line, f, goto fail) {
+ char buf[2+DECIMAL_STR_MAX(unsigned)+3+4], *p, *w;
+ unsigned maj;
+
+ truncate_nl(line);
+
+ if (type == 'c' && streq(line, "Character devices:")) {
+ good = true;
+ continue;
+ }
+
+ if (type == 'b' && streq(line, "Block devices:")) {
+ good = true;
+ continue;
+ }
+
+ if (isempty(line)) {
+ good = false;
+ continue;
+ }
+
+ if (!good)
+ continue;
+
+ p = strstrip(line);
+
+ w = strpbrk(p, WHITESPACE);
+ if (!w)
+ continue;
+ *w = 0;
+
+ r = safe_atou(p, &maj);
+ if (r < 0)
+ continue;
+ if (maj <= 0)
+ continue;
+
+ w++;
+ w += strspn(w, WHITESPACE);
+
+ if (fnmatch(name, w, 0) != 0)
+ continue;
+
+ sprintf(buf,
+ "%c %u:* %s",
+ type,
+ maj,
+ acc);
+
+ r = cg_set_attribute("devices", path, "devices.allow", buf);
+ if (r < 0)
+ log_full_errno(IN_SET(r, -ENOENT, -EROFS, -EINVAL, -EACCES) ? LOG_DEBUG : LOG_WARNING, r,
+ "Failed to set devices.allow on %s: %m", path);
+ }
+
+ return 0;
+
+fail:
+ log_warning_errno(errno, "Failed to read /proc/devices: %m");
+ return -errno;
+}
+
+void cgroup_context_apply(CGroupContext *c, CGroupMask mask, const char *path, uint32_t netclass, ManagerState state) {
+ bool is_root;
+ int r;
+
+ assert(c);
+ assert(path);
+
+ if (mask == 0)
+ return;
+
+ /* Some cgroup attributes are not supported on the root cgroup,
+ * hence silently ignore */
+ is_root = isempty(path) || path_equal(path, "/");
+ if (is_root)
+ /* Make sure we don't try to display messages with an empty path. */
+ path = "/";
+
+ /* We generally ignore errors caused by read-only mounted
+ * cgroup trees (assuming we are running in a container then),
+ * and missing cgroups, i.e. EROFS and ENOENT. */
+
+ if ((mask & CGROUP_MASK_CPU) && !is_root) {
+ char buf[MAX(DECIMAL_STR_MAX(uint64_t), DECIMAL_STR_MAX(usec_t)) + 1];
+
+ sprintf(buf, "%" PRIu64 "\n",
+ IN_SET(state, MANAGER_STARTING, MANAGER_INITIALIZING) && c->startup_cpu_shares != CGROUP_CPU_SHARES_INVALID ? c->startup_cpu_shares :
+ c->cpu_shares != CGROUP_CPU_SHARES_INVALID ? c->cpu_shares : CGROUP_CPU_SHARES_DEFAULT);
+ r = cg_set_attribute("cpu", path, "cpu.shares", buf);
+ if (r < 0)
+ log_full_errno(IN_SET(r, -ENOENT, -EROFS, -EACCES) ? LOG_DEBUG : LOG_WARNING, r,
+ "Failed to set cpu.shares on %s: %m", path);
+
+ sprintf(buf, USEC_FMT "\n", CGROUP_CPU_QUOTA_PERIOD_USEC);
+ r = cg_set_attribute("cpu", path, "cpu.cfs_period_us", buf);
+ if (r < 0)
+ log_full_errno(IN_SET(r, -ENOENT, -EROFS, -EACCES) ? LOG_DEBUG : LOG_WARNING, r,
+ "Failed to set cpu.cfs_period_us on %s: %m", path);
+
+ if (c->cpu_quota_per_sec_usec != USEC_INFINITY) {
+ sprintf(buf, USEC_FMT "\n", c->cpu_quota_per_sec_usec * CGROUP_CPU_QUOTA_PERIOD_USEC / USEC_PER_SEC);
+ r = cg_set_attribute("cpu", path, "cpu.cfs_quota_us", buf);
+ } else
+ r = cg_set_attribute("cpu", path, "cpu.cfs_quota_us", "-1");
+ if (r < 0)
+ log_full_errno(IN_SET(r, -ENOENT, -EROFS, -EACCES) ? LOG_DEBUG : LOG_WARNING, r,
+ "Failed to set cpu.cfs_quota_us on %s: %m", path);
+ }
+
+ if (mask & CGROUP_MASK_BLKIO) {
+ char buf[MAX(DECIMAL_STR_MAX(uint64_t)+1,
+ DECIMAL_STR_MAX(dev_t)*2+2+DECIMAL_STR_MAX(uint64_t)+1)];
+ CGroupBlockIODeviceWeight *w;
+ CGroupBlockIODeviceBandwidth *b;
+
+ if (!is_root) {
+ sprintf(buf, "%" PRIu64 "\n",
+ IN_SET(state, MANAGER_STARTING, MANAGER_INITIALIZING) && c->startup_blockio_weight != CGROUP_BLKIO_WEIGHT_INVALID ? c->startup_blockio_weight :
+ c->blockio_weight != CGROUP_BLKIO_WEIGHT_INVALID ? c->blockio_weight : CGROUP_BLKIO_WEIGHT_DEFAULT);
+ r = cg_set_attribute("blkio", path, "blkio.weight", buf);
+ if (r < 0)
+ log_full_errno(IN_SET(r, -ENOENT, -EROFS, -EACCES) ? LOG_DEBUG : LOG_WARNING, r,
+ "Failed to set blkio.weight on %s: %m", path);
+
+ /* FIXME: no way to reset this list */
+ LIST_FOREACH(device_weights, w, c->blockio_device_weights) {
+ dev_t dev;
+
+ r = lookup_blkio_device(w->path, &dev);
+ if (r < 0)
+ continue;
+
+ sprintf(buf, "%u:%u %" PRIu64 "\n", major(dev), minor(dev), w->weight);
+ r = cg_set_attribute("blkio", path, "blkio.weight_device", buf);
+ if (r < 0)
+ log_full_errno(IN_SET(r, -ENOENT, -EROFS, -EACCES) ? LOG_DEBUG : LOG_WARNING, r,
+ "Failed to set blkio.weight_device on %s: %m", path);
+ }
+ }
+
+ /* FIXME: no way to reset this list */
+ LIST_FOREACH(device_bandwidths, b, c->blockio_device_bandwidths) {
+ const char *a;
+ dev_t dev;
+
+ r = lookup_blkio_device(b->path, &dev);
+ if (r < 0)
+ continue;
+
+ a = b->read ? "blkio.throttle.read_bps_device" : "blkio.throttle.write_bps_device";
+
+ sprintf(buf, "%u:%u %" PRIu64 "\n", major(dev), minor(dev), b->bandwidth);
+ r = cg_set_attribute("blkio", path, a, buf);
+ if (r < 0)
+ log_full_errno(IN_SET(r, -ENOENT, -EROFS, -EACCES) ? LOG_DEBUG : LOG_WARNING, r,
+ "Failed to set %s on %s: %m", a, path);
+ }
+ }
+
+ if ((mask & CGROUP_MASK_MEMORY) && !is_root) {
+ if (c->memory_limit != (uint64_t) -1) {
+ char buf[DECIMAL_STR_MAX(uint64_t) + 1];
+
+ sprintf(buf, "%" PRIu64 "\n", c->memory_limit);
+
+ if (cg_unified() <= 0)
+ r = cg_set_attribute("memory", path, "memory.limit_in_bytes", buf);
+ else
+ r = cg_set_attribute("memory", path, "memory.max", buf);
+
+ } else {
+ if (cg_unified() <= 0)
+ r = cg_set_attribute("memory", path, "memory.limit_in_bytes", "-1");
+ else
+ r = cg_set_attribute("memory", path, "memory.max", "max");
+ }
+
+ if (r < 0)
+ log_full_errno(IN_SET(r, -ENOENT, -EROFS, -EACCES) ? LOG_DEBUG : LOG_WARNING, r,
+ "Failed to set memory.limit_in_bytes/memory.max on %s: %m", path);
+ }
+
+ if ((mask & CGROUP_MASK_DEVICES) && !is_root) {
+ CGroupDeviceAllow *a;
+
+ /* Changing the devices list of a populated cgroup
+ * might result in EINVAL, hence ignore EINVAL
+ * here. */
+
+ if (c->device_allow || c->device_policy != CGROUP_AUTO)
+ r = cg_set_attribute("devices", path, "devices.deny", "a");
+ else
+ r = cg_set_attribute("devices", path, "devices.allow", "a");
+ if (r < 0)
+ log_full_errno(IN_SET(r, -ENOENT, -EROFS, -EINVAL, -EACCES) ? LOG_DEBUG : LOG_WARNING, r,
+ "Failed to reset devices.list on %s: %m", path);
+
+ if (c->device_policy == CGROUP_CLOSED ||
+ (c->device_policy == CGROUP_AUTO && c->device_allow)) {
+ static const char auto_devices[] =
+ "/dev/null\0" "rwm\0"
+ "/dev/zero\0" "rwm\0"
+ "/dev/full\0" "rwm\0"
+ "/dev/random\0" "rwm\0"
+ "/dev/urandom\0" "rwm\0"
+ "/dev/tty\0" "rwm\0"
+ "/dev/pts/ptmx\0" "rw\0"; /* /dev/pts/ptmx may not be duplicated, but accessed */
+
+ const char *x, *y;
+
+ NULSTR_FOREACH_PAIR(x, y, auto_devices)
+ whitelist_device(path, x, y);
+
+ whitelist_major(path, "pts", 'c', "rw");
+ whitelist_major(path, "kdbus", 'c', "rw");
+ whitelist_major(path, "kdbus/*", 'c', "rw");
+ }
+
+ LIST_FOREACH(device_allow, a, c->device_allow) {
+ char acc[4];
+ unsigned k = 0;
+
+ if (a->r)
+ acc[k++] = 'r';
+ if (a->w)
+ acc[k++] = 'w';
+ if (a->m)
+ acc[k++] = 'm';
+
+ if (k == 0)
+ continue;
+
+ acc[k++] = 0;
+
+ if (startswith(a->path, "/dev/"))
+ whitelist_device(path, a->path, acc);
+ else if (startswith(a->path, "block-"))
+ whitelist_major(path, a->path + 6, 'b', acc);
+ else if (startswith(a->path, "char-"))
+ whitelist_major(path, a->path + 5, 'c', acc);
+ else
+ log_debug("Ignoring device %s while writing cgroup attribute.", a->path);
+ }
+ }
+
+ if ((mask & CGROUP_MASK_PIDS) && !is_root) {
+
+ if (c->tasks_max != (uint64_t) -1) {
+ char buf[DECIMAL_STR_MAX(uint64_t) + 2];
+
+ sprintf(buf, "%" PRIu64 "\n", c->tasks_max);
+ r = cg_set_attribute("pids", path, "pids.max", buf);
+ } else
+ r = cg_set_attribute("pids", path, "pids.max", "max");
+
+ if (r < 0)
+ log_full_errno(IN_SET(r, -ENOENT, -EROFS, -EACCES) ? LOG_DEBUG : LOG_WARNING, r,
+ "Failed to set pids.max on %s: %m", path);
+ }
+
+ if (mask & CGROUP_MASK_NET_CLS) {
+ char buf[DECIMAL_STR_MAX(uint32_t)];
+
+ sprintf(buf, "%" PRIu32, netclass);
+
+ r = cg_set_attribute("net_cls", path, "net_cls.classid", buf);
+ if (r < 0)
+ log_full_errno(IN_SET(r, -ENOENT, -EROFS, -EACCES) ? LOG_DEBUG : LOG_WARNING, r,
+ "Failed to set net_cls.classid on %s: %m", path);
+ }
+}
+
+CGroupMask cgroup_context_get_mask(CGroupContext *c) {
+ CGroupMask mask = 0;
+
+ /* Figure out which controllers we need */
+
+ if (c->cpu_accounting ||
+ c->cpu_shares != CGROUP_CPU_SHARES_INVALID ||
+ c->startup_cpu_shares != CGROUP_CPU_SHARES_INVALID ||
+ c->cpu_quota_per_sec_usec != USEC_INFINITY)
+ mask |= CGROUP_MASK_CPUACCT | CGROUP_MASK_CPU;
+
+ if (c->blockio_accounting ||
+ c->blockio_weight != CGROUP_BLKIO_WEIGHT_INVALID ||
+ c->startup_blockio_weight != CGROUP_BLKIO_WEIGHT_INVALID ||
+ c->blockio_device_weights ||
+ c->blockio_device_bandwidths)
+ mask |= CGROUP_MASK_BLKIO;
+
+ if (c->memory_accounting ||
+ c->memory_limit != (uint64_t) -1)
+ mask |= CGROUP_MASK_MEMORY;
+
+ if (c->device_allow ||
+ c->device_policy != CGROUP_AUTO)
+ mask |= CGROUP_MASK_DEVICES;
+
+ if (c->tasks_accounting ||
+ c->tasks_max != (uint64_t) -1)
+ mask |= CGROUP_MASK_PIDS;
+
+ if (c->netclass_type != CGROUP_NETCLASS_TYPE_NONE)
+ mask |= CGROUP_MASK_NET_CLS;
+
+ return mask;
+}
+
+CGroupMask unit_get_own_mask(Unit *u) {
+ CGroupContext *c;
+
+ /* Returns the mask of controllers the unit needs for itself */
+
+ c = unit_get_cgroup_context(u);
+ if (!c)
+ return 0;
+
+ /* If delegation is turned on, then turn on all cgroups,
+ * unless we are on the legacy hierarchy and the process we
+ * fork into it is known to drop privileges, and hence
+ * shouldn't get access to the controllers.
+ *
+ * Note that on the unified hierarchy it is safe to delegate
+ * controllers to unprivileged services. */
+
+ if (c->delegate) {
+ ExecContext *e;
+
+ e = unit_get_exec_context(u);
+ if (!e ||
+ exec_context_maintains_privileges(e) ||
+ cg_unified() > 0)
+ return _CGROUP_MASK_ALL;
+ }
+
+ return cgroup_context_get_mask(c);
+}
+
+CGroupMask unit_get_members_mask(Unit *u) {
+ assert(u);
+
+ /* Returns the mask of controllers all of the unit's children
+ * require, merged */
+
+ if (u->cgroup_members_mask_valid)
+ return u->cgroup_members_mask;
+
+ u->cgroup_members_mask = 0;
+
+ if (u->type == UNIT_SLICE) {
+ Unit *member;
+ Iterator i;
+
+ SET_FOREACH(member, u->dependencies[UNIT_BEFORE], i) {
+
+ if (member == u)
+ continue;
+
+ if (UNIT_DEREF(member->slice) != u)
+ continue;
+
+ u->cgroup_members_mask |=
+ unit_get_own_mask(member) |
+ unit_get_members_mask(member);
+ }
+ }
+
+ u->cgroup_members_mask_valid = true;
+ return u->cgroup_members_mask;
+}
+
+CGroupMask unit_get_siblings_mask(Unit *u) {
+ assert(u);
+
+ /* Returns the mask of controllers all of the unit's siblings
+ * require, i.e. the members mask of the unit's parent slice
+ * if there is one. */
+
+ if (UNIT_ISSET(u->slice))
+ return unit_get_members_mask(UNIT_DEREF(u->slice));
+
+ return unit_get_own_mask(u) | unit_get_members_mask(u);
+}
+
+CGroupMask unit_get_subtree_mask(Unit *u) {
+
+ /* Returns the mask of this subtree, meaning of the group
+ * itself and its children. */
+
+ return unit_get_own_mask(u) | unit_get_members_mask(u);
+}
+
+CGroupMask unit_get_target_mask(Unit *u) {
+ CGroupMask mask;
+
+ /* This returns the cgroup mask of all controllers to enable
+ * for a specific cgroup, i.e. everything it needs itself,
+ * plus all that its children need, plus all that its siblings
+ * need. This is primarily useful on the legacy cgroup
+ * hierarchy, where we need to duplicate each cgroup in each
+ * hierarchy that shall be enabled for it. */
+
+ mask = unit_get_own_mask(u) | unit_get_members_mask(u) | unit_get_siblings_mask(u);
+ mask &= u->manager->cgroup_supported;
+
+ return mask;
+}
+
+CGroupMask unit_get_enable_mask(Unit *u) {
+ CGroupMask mask;
+
+ /* This returns the cgroup mask of all controllers to enable
+ * for the children of a specific cgroup. This is primarily
+ * useful for the unified cgroup hierarchy, where each cgroup
+ * controls which controllers are enabled for its children. */
+
+ mask = unit_get_members_mask(u);
+ mask &= u->manager->cgroup_supported;
+
+ return mask;
+}
+
+/* Recurse from a unit up through its containing slices, propagating
+ * mask bits upward. A unit is also member of itself. */
+void unit_update_cgroup_members_masks(Unit *u) {
+ CGroupMask m;
+ bool more;
+
+ assert(u);
+
+ /* Calculate subtree mask */
+ m = unit_get_subtree_mask(u);
+
+ /* See if anything changed from the previous invocation. If
+ * not, we're done. */
+ if (u->cgroup_subtree_mask_valid && m == u->cgroup_subtree_mask)
+ return;
+
+ more =
+ u->cgroup_subtree_mask_valid &&
+ ((m & ~u->cgroup_subtree_mask) != 0) &&
+ ((~m & u->cgroup_subtree_mask) == 0);
+
+ u->cgroup_subtree_mask = m;
+ u->cgroup_subtree_mask_valid = true;
+
+ if (UNIT_ISSET(u->slice)) {
+ Unit *s = UNIT_DEREF(u->slice);
+
+ if (more)
+ /* There's more set now than before. We
+ * propagate the new mask to the parent's mask
+ * (not caring if it actually was valid or
+ * not). */
+
+ s->cgroup_members_mask |= m;
+
+ else
+ /* There's less set now than before (or we
+ * don't know), we need to recalculate
+ * everything, so let's invalidate the
+ * parent's members mask */
+
+ s->cgroup_members_mask_valid = false;
+
+ /* And now make sure that this change also hits our
+ * grandparents */
+ unit_update_cgroup_members_masks(s);
+ }
+}
+
+static const char *migrate_callback(CGroupMask mask, void *userdata) {
+ Unit *u = userdata;
+
+ assert(mask != 0);
+ assert(u);
+
+ while (u) {
+ if (u->cgroup_path &&
+ u->cgroup_realized &&
+ (u->cgroup_realized_mask & mask) == mask)
+ return u->cgroup_path;
+
+ u = UNIT_DEREF(u->slice);
+ }
+
+ return NULL;
+}
+
+char *unit_default_cgroup_path(Unit *u) {
+ _cleanup_free_ char *escaped = NULL, *slice = NULL;
+ int r;
+
+ assert(u);
+
+ if (unit_has_name(u, SPECIAL_ROOT_SLICE))
+ return strdup(u->manager->cgroup_root);
+
+ if (UNIT_ISSET(u->slice) && !unit_has_name(UNIT_DEREF(u->slice), SPECIAL_ROOT_SLICE)) {
+ r = cg_slice_to_path(UNIT_DEREF(u->slice)->id, &slice);
+ if (r < 0)
+ return NULL;
+ }
+
+ escaped = cg_escape(u->id);
+ if (!escaped)
+ return NULL;
+
+ if (slice)
+ return strjoin(u->manager->cgroup_root, "/", slice, "/", escaped, NULL);
+ else
+ return strjoin(u->manager->cgroup_root, "/", escaped, NULL);
+}
+
+int unit_set_cgroup_path(Unit *u, const char *path) {
+ _cleanup_free_ char *p = NULL;
+ int r;
+
+ assert(u);
+
+ if (path) {
+ p = strdup(path);
+ if (!p)
+ return -ENOMEM;
+ } else
+ p = NULL;
+
+ if (streq_ptr(u->cgroup_path, p))
+ return 0;
+
+ if (p) {
+ r = hashmap_put(u->manager->cgroup_unit, p, u);
+ if (r < 0)
+ return r;
+ }
+
+ unit_release_cgroup(u);
+
+ u->cgroup_path = p;
+ p = NULL;
+
+ return 1;
+}
+
+int unit_watch_cgroup(Unit *u) {
+ _cleanup_free_ char *populated = NULL;
+ int r;
+
+ assert(u);
+
+ if (!u->cgroup_path)
+ return 0;
+
+ if (u->cgroup_inotify_wd >= 0)
+ return 0;
+
+ /* Only applies to the unified hierarchy */
+ r = cg_unified();
+ if (r < 0)
+ return log_unit_error_errno(u, r, "Failed detect wether the unified hierarchy is used: %m");
+ if (r == 0)
+ return 0;
+
+ /* Don't watch the root slice, it's pointless. */
+ if (unit_has_name(u, SPECIAL_ROOT_SLICE))
+ return 0;
+
+ r = hashmap_ensure_allocated(&u->manager->cgroup_inotify_wd_unit, &trivial_hash_ops);
+ if (r < 0)
+ return log_oom();
+
+ r = cg_get_path(SYSTEMD_CGROUP_CONTROLLER, u->cgroup_path, "cgroup.populated", &populated);
+ if (r < 0)
+ return log_oom();
+
+ u->cgroup_inotify_wd = inotify_add_watch(u->manager->cgroup_inotify_fd, populated, IN_MODIFY);
+ if (u->cgroup_inotify_wd < 0) {
+
+ if (errno == ENOENT) /* If the directory is already
+ * gone we don't need to track
+ * it, so this is not an error */
+ return 0;
+
+ return log_unit_error_errno(u, errno, "Failed to add inotify watch descriptor for control group %s: %m", u->cgroup_path);
+ }
+
+ r = hashmap_put(u->manager->cgroup_inotify_wd_unit, INT_TO_PTR(u->cgroup_inotify_wd), u);
+ if (r < 0)
+ return log_unit_error_errno(u, r, "Failed to add inotify watch descriptor to hash map: %m");
+
+ return 0;
+}
+
+static int unit_create_cgroup(
+ Unit *u,
+ CGroupMask target_mask,
+ CGroupMask enable_mask) {
+
+ CGroupContext *c;
+ int r;
+
+ assert(u);
+
+ c = unit_get_cgroup_context(u);
+ if (!c)
+ return 0;
+
+ if (!u->cgroup_path) {
+ _cleanup_free_ char *path = NULL;
+
+ path = unit_default_cgroup_path(u);
+ if (!path)
+ return log_oom();
+
+ r = unit_set_cgroup_path(u, path);
+ if (r == -EEXIST)
+ return log_unit_error_errno(u, r, "Control group %s exists already.", path);
+ if (r < 0)
+ return log_unit_error_errno(u, r, "Failed to set unit's control group path to %s: %m", path);
+ }
+
+ /* First, create our own group */
+ r = cg_create_everywhere(u->manager->cgroup_supported, target_mask, u->cgroup_path);
+ if (r < 0)
+ return log_unit_error_errno(u, r, "Failed to create cgroup %s: %m", u->cgroup_path);
+
+ /* Start watching it */
+ (void) unit_watch_cgroup(u);
+
+ /* Enable all controllers we need */
+ r = cg_enable_everywhere(u->manager->cgroup_supported, enable_mask, u->cgroup_path);
+ if (r < 0)
+ log_unit_warning_errno(u, r, "Failed to enable controllers on cgroup %s, ignoring: %m", u->cgroup_path);
+
+ /* Keep track that this is now realized */
+ u->cgroup_realized = true;
+ u->cgroup_realized_mask = target_mask;
+
+ if (u->type != UNIT_SLICE && !c->delegate) {
+
+ /* Then, possibly move things over, but not if
+ * subgroups may contain processes, which is the case
+ * for slice and delegation units. */
+ r = cg_migrate_everywhere(u->manager->cgroup_supported, u->cgroup_path, u->cgroup_path, migrate_callback, u);
+ if (r < 0)
+ log_unit_warning_errno(u, r, "Failed to migrate cgroup from to %s, ignoring: %m", u->cgroup_path);
+ }
+
+ return 0;
+}
+
+int unit_attach_pids_to_cgroup(Unit *u) {
+ int r;
+ assert(u);
+
+ r = unit_realize_cgroup(u);
+ if (r < 0)
+ return r;
+
+ r = cg_attach_many_everywhere(u->manager->cgroup_supported, u->cgroup_path, u->pids, migrate_callback, u);
+ if (r < 0)
+ return r;
+
+ return 0;
+}
+
+static bool unit_has_mask_realized(Unit *u, CGroupMask target_mask) {
+ assert(u);
+
+ return u->cgroup_realized && u->cgroup_realized_mask == target_mask;
+}
+
+static int unit_find_free_netclass_cgroup(Unit *u, uint32_t *ret) {
+
+ uint32_t start, i;
+ Manager *m;
+
+ assert(u);
+
+ m = u->manager;
+
+ i = start = m->cgroup_netclass_registry_last;
+
+ do {
+ i++;
+
+ if (!hashmap_get(m->cgroup_netclass_registry, UINT_TO_PTR(i))) {
+ m->cgroup_netclass_registry_last = i;
+ *ret = i;
+ return 0;
+ }
+
+ if (i == UINT32_MAX)
+ i = CGROUP_NETCLASS_FIXED_MAX;
+
+ } while (i != start);
+
+ return -ENOBUFS;
+}
+
+int unit_add_to_netclass_cgroup(Unit *u) {
+
+ CGroupContext *cc;
+ Unit *first;
+ void *key;
+ int r;
+
+ assert(u);
+
+ cc = unit_get_cgroup_context(u);
+ if (!cc)
+ return 0;
+
+ switch (cc->netclass_type) {
+ case CGROUP_NETCLASS_TYPE_NONE:
+ return 0;
+
+ case CGROUP_NETCLASS_TYPE_FIXED:
+ u->cgroup_netclass_id = cc->netclass_id;
+ break;
+
+ case CGROUP_NETCLASS_TYPE_AUTO:
+ /* Allocate a new ID in case it was requested and not done yet */
+ if (u->cgroup_netclass_id == 0) {
+ r = unit_find_free_netclass_cgroup(u, &u->cgroup_netclass_id);
+ if (r < 0)
+ return r;
+
+ log_debug("Dynamically assigned netclass cgroup id %" PRIu32 " to %s", u->cgroup_netclass_id, u->id);
+ }
+
+ break;
+ }
+
+ r = hashmap_ensure_allocated(&u->manager->cgroup_netclass_registry, &trivial_hash_ops);
+ if (r < 0)
+ return r;
+
+ key = UINT32_TO_PTR(u->cgroup_netclass_id);
+ first = hashmap_get(u->manager->cgroup_netclass_registry, key);
+
+ if (first) {
+ LIST_PREPEND(cgroup_netclass, first, u);
+ return hashmap_replace(u->manager->cgroup_netclass_registry, key, u);
+ }
+
+ return hashmap_put(u->manager->cgroup_netclass_registry, key, u);
+}
+
+int unit_remove_from_netclass_cgroup(Unit *u) {
+
+ Unit *head;
+ void *key;
+
+ assert(u);
+
+ key = UINT32_TO_PTR(u->cgroup_netclass_id);
+
+ LIST_FIND_HEAD(cgroup_netclass, u, head);
+ LIST_REMOVE(cgroup_netclass, head, u);
+
+ if (head)
+ return hashmap_replace(u->manager->cgroup_netclass_registry, key, head);
+
+ hashmap_remove(u->manager->cgroup_netclass_registry, key);
+
+ return 0;
+}
+
+/* Check if necessary controllers and attributes for a unit are in place.
+ *
+ * If so, do nothing.
+ * If not, create paths, move processes over, and set attributes.
+ *
+ * Returns 0 on success and < 0 on failure. */
+static int unit_realize_cgroup_now(Unit *u, ManagerState state) {
+ CGroupMask target_mask, enable_mask;
+ int r;
+
+ assert(u);
+
+ if (u->in_cgroup_queue) {
+ LIST_REMOVE(cgroup_queue, u->manager->cgroup_queue, u);
+ u->in_cgroup_queue = false;
+ }
+
+ target_mask = unit_get_target_mask(u);
+ if (unit_has_mask_realized(u, target_mask))
+ return 0;
+
+ /* First, realize parents */
+ if (UNIT_ISSET(u->slice)) {
+ r = unit_realize_cgroup_now(UNIT_DEREF(u->slice), state);
+ if (r < 0)
+ return r;
+ }
+
+ /* And then do the real work */
+ enable_mask = unit_get_enable_mask(u);
+ r = unit_create_cgroup(u, target_mask, enable_mask);
+ if (r < 0)