X-Git-Url: https://www.chiark.greenend.org.uk/ucgi/~ianmdlvl/git?a=blobdiff_plain;f=src%2Fcore%2Fcgroup.c;h=70fc925b4dfedde9bcfde694acf600f545af1b76;hb=a2c0e528b8b5ba370527db279605e4e4135689c1;hp=36e9f53db8f1f4834fe56361f19fab4dbe6da1cb;hpb=637f421e5c6ae76f5c59c3badd809bf5f04fca80;p=elogind.git diff --git a/src/core/cgroup.c b/src/core/cgroup.c index 36e9f53db..70fc925b4 100644 --- a/src/core/cgroup.c +++ b/src/core/cgroup.c @@ -27,6 +27,8 @@ #include "cgroup-util.h" #include "cgroup.h" +#define CGROUP_CPU_QUOTA_PERIOD_USEC ((usec_t) 100 * USEC_PER_MSEC) + void cgroup_context_init(CGroupContext *c) { assert(c); @@ -39,9 +41,7 @@ void cgroup_context_init(CGroupContext *c) { c->blockio_weight = (unsigned long) -1; c->startup_blockio_weight = (unsigned long) -1; - c->cpu_quota_per_sec_usec = (usec_t) -1; - c->cpu_quota_usec = (usec_t) -1; - c->cpu_quota_period_usec = 100*USEC_PER_MSEC; + c->cpu_quota_per_sec_usec = USEC_INFINITY; } void cgroup_context_free_device_allow(CGroupContext *c, CGroupDeviceAllow *a) { @@ -84,37 +84,11 @@ void cgroup_context_done(CGroupContext *c) { cgroup_context_free_device_allow(c, c->device_allow); } -usec_t cgroup_context_get_cpu_quota_usec(CGroupContext *c) { - assert(c); - - /* Returns the absolute CPU quota */ - - if (c->cpu_quota_usec != (usec_t) -1) - return c->cpu_quota_usec; - else if (c->cpu_quota_per_sec_usec != (usec_t) -1) - return c->cpu_quota_per_sec_usec*c->cpu_quota_period_usec/USEC_PER_SEC; - else - return (usec_t) -1; -} - -usec_t cgroup_context_get_cpu_quota_per_sec_usec(CGroupContext *c) { - assert(c); - - /* Returns the CPU quota relative to 1s */ - - if (c->cpu_quota_usec != (usec_t) -1) - return c->cpu_quota_usec*USEC_PER_SEC/c->cpu_quota_period_usec; - else if (c->cpu_quota_per_sec_usec != (usec_t) -1) - return c->cpu_quota_per_sec_usec; - else - return (usec_t) -1; -} - void cgroup_context_dump(CGroupContext *c, FILE* f, const char *prefix) { CGroupBlockIODeviceBandwidth *b; CGroupBlockIODeviceWeight *w; CGroupDeviceAllow *a; - char t[FORMAT_TIMESPAN_MAX], s[FORMAT_TIMESPAN_MAX], u[FORMAT_TIMESPAN_MAX]; + char u[FORMAT_TIMESPAN_MAX]; assert(c); assert(f); @@ -127,25 +101,23 @@ void cgroup_context_dump(CGroupContext *c, FILE* f, const char *prefix) { "%sMemoryAccounting=%s\n" "%sCPUShares=%lu\n" "%sStartupCPUShares=%lu\n" - "%sCPUQuota=%s\n" "%sCPUQuotaPerSecSec=%s\n" - "%sCPUQuotaPeriodSec=%s\n" "%sBlockIOWeight=%lu\n" "%sStartupBlockIOWeight=%lu\n" "%sMemoryLimit=%" PRIu64 "\n" - "%sDevicePolicy=%s\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, c->cpu_shares, prefix, c->startup_cpu_shares, - prefix, strna(format_timespan(u, sizeof(u), cgroup_context_get_cpu_quota_usec(c), 1)), - prefix, strna(format_timespan(t, sizeof(t), cgroup_context_get_cpu_quota_per_sec_usec(c), 1)), - prefix, strna(format_timespan(s, sizeof(s), c->cpu_quota_period_usec, 1)), + 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, cgroup_device_policy_to_string(c->device_policy)); + prefix, cgroup_device_policy_to_string(c->device_policy), + prefix, yes_no(c->delegate)); LIST_FOREACH(device_allow, a, c->device_allow) fprintf(f, @@ -181,10 +153,8 @@ static int lookup_blkio_device(const char *p, dev_t *dev) { assert(dev); r = stat(p, &st); - if (r < 0) { - log_warning("Couldn't stat device %s: %m", p); - return -errno; - } + 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; @@ -230,7 +200,7 @@ static int whitelist_device(const char *path, const char *node, const char *acc) r = cg_set_attribute("devices", path, "devices.allow", buf); if (r < 0) - log_warning("Failed to set devices.allow on %s: %s", path, strerror(-r)); + log_full(r == -ENOENT ? LOG_DEBUG : LOG_WARNING, "Failed to set devices.allow on %s: %s", path, strerror(-r)); return r; } @@ -246,10 +216,8 @@ static int whitelist_major(const char *path, const char *name, char type, const assert(type == 'b' || type == 'c'); f = fopen("/proc/devices", "re"); - if (!f) { - log_warning("Cannot open /proc/devices to resolve %s (%c): %m", name, type); - return -errno; - } + 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; @@ -302,13 +270,13 @@ static int whitelist_major(const char *path, const char *name, char type, const r = cg_set_attribute("devices", path, "devices.allow", buf); if (r < 0) - log_warning("Failed to set devices.allow on %s: %s", path, strerror(-r)); + log_full(r == -ENOENT ? LOG_DEBUG : LOG_WARNING, "Failed to set devices.allow on %s: %s", path, strerror(-r)); } return 0; fail: - log_warning("Failed to read /proc/devices: %m"); + log_warning_errno(errno, "Failed to read /proc/devices: %m"); return -errno; } @@ -328,28 +296,26 @@ void cgroup_context_apply(CGroupContext *c, CGroupControllerMask mask, const cha if ((mask & CGROUP_CPU) && !is_root) { char buf[MAX(DECIMAL_STR_MAX(unsigned long), DECIMAL_STR_MAX(usec_t)) + 1]; - usec_t q; sprintf(buf, "%lu\n", - state == MANAGER_STARTING && c->startup_cpu_shares != (unsigned long) -1 ? c->startup_cpu_shares : + IN_SET(state, MANAGER_STARTING, MANAGER_INITIALIZING) && c->startup_cpu_shares != (unsigned long) -1 ? c->startup_cpu_shares : c->cpu_shares != (unsigned long) -1 ? c->cpu_shares : 1024); r = cg_set_attribute("cpu", path, "cpu.shares", buf); if (r < 0) - log_warning("Failed to set cpu.shares on %s: %s", path, strerror(-r)); + log_full(r == -ENOENT ? LOG_DEBUG : LOG_WARNING, "Failed to set cpu.shares on %s: %s", path, strerror(-r)); - sprintf(buf, USEC_FMT "\n", c->cpu_quota_period_usec); + 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_warning("Failed to set cpu.cfs_period_us on %s: %s", path, strerror(-r)); + log_full(r == -ENOENT ? LOG_DEBUG : LOG_WARNING, "Failed to set cpu.cfs_period_us on %s: %s", path, strerror(-r)); - q = cgroup_context_get_cpu_quota_usec(c); - if (q != (usec_t) -1) { - sprintf(buf, USEC_FMT "\n", q); + 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_warning("Failed to set cpu.cfs_quota_us on %s: %s", path, strerror(-r)); + log_full(r == -ENOENT ? LOG_DEBUG : LOG_WARNING, "Failed to set cpu.cfs_quota_us on %s: %s", path, strerror(-r)); } if (mask & CGROUP_BLKIO) { @@ -360,11 +326,11 @@ void cgroup_context_apply(CGroupContext *c, CGroupControllerMask mask, const cha CGroupBlockIODeviceBandwidth *b; if (!is_root) { - sprintf(buf, "%lu\n", state == MANAGER_STARTING && c->startup_blockio_weight != (unsigned long) -1 ? c->startup_blockio_weight : + sprintf(buf, "%lu\n", IN_SET(state, MANAGER_STARTING, MANAGER_INITIALIZING) && c->startup_blockio_weight != (unsigned long) -1 ? c->startup_blockio_weight : c->blockio_weight != (unsigned long) -1 ? c->blockio_weight : 1000); r = cg_set_attribute("blkio", path, "blkio.weight", buf); if (r < 0) - log_warning("Failed to set blkio.weight on %s: %s", path, strerror(-r)); + log_full(r == -ENOENT ? LOG_DEBUG : LOG_WARNING, "Failed to set blkio.weight on %s: %s", path, strerror(-r)); /* FIXME: no way to reset this list */ LIST_FOREACH(device_weights, w, c->blockio_device_weights) { @@ -377,7 +343,7 @@ void cgroup_context_apply(CGroupContext *c, CGroupControllerMask mask, const cha sprintf(buf, "%u:%u %lu", major(dev), minor(dev), w->weight); r = cg_set_attribute("blkio", path, "blkio.weight_device", buf); if (r < 0) - log_error("Failed to set blkio.weight_device on %s: %s", path, strerror(-r)); + log_full(r == -ENOENT ? LOG_DEBUG : LOG_WARNING, "Failed to set blkio.weight_device on %s: %s", path, strerror(-r)); } } @@ -395,7 +361,7 @@ void cgroup_context_apply(CGroupContext *c, CGroupControllerMask mask, const cha sprintf(buf, "%u:%u %" PRIu64 "\n", major(dev), minor(dev), b->bandwidth); r = cg_set_attribute("blkio", path, a, buf); if (r < 0) - log_error("Failed to set %s on %s: %s", a, path, strerror(-r)); + log_full(r == -ENOENT ? LOG_DEBUG : LOG_WARNING, "Failed to set %s on %s: %s", a, path, strerror(-r)); } } @@ -409,7 +375,7 @@ void cgroup_context_apply(CGroupContext *c, CGroupControllerMask mask, const cha r = cg_set_attribute("memory", path, "memory.limit_in_bytes", "-1"); if (r < 0) - log_error("Failed to set memory.limit_in_bytes on %s: %s", path, strerror(-r)); + log_full(r == -ENOENT ? LOG_DEBUG : LOG_WARNING, "Failed to set memory.limit_in_bytes on %s: %s", path, strerror(-r)); } if ((mask & CGROUP_DEVICE) && !is_root) { @@ -420,7 +386,7 @@ void cgroup_context_apply(CGroupContext *c, CGroupControllerMask mask, const cha else r = cg_set_attribute("devices", path, "devices.allow", "a"); if (r < 0) - log_warning("Failed to reset devices.list on %s: %s", path, strerror(-r)); + log_full(r == -ENOENT ? LOG_DEBUG : LOG_WARNING, "Failed to reset devices.list on %s: %s", path, strerror(-r)); if (c->device_policy == CGROUP_CLOSED || (c->device_policy == CGROUP_AUTO && c->device_allow)) { @@ -479,8 +445,7 @@ CGroupControllerMask cgroup_context_get_mask(CGroupContext *c) { if (c->cpu_accounting || c->cpu_shares != (unsigned long) -1 || c->startup_cpu_shares != (unsigned long) -1 || - c->cpu_quota_usec != (usec_t) -1 || - c->cpu_quota_per_sec_usec != (usec_t) -1) + c->cpu_quota_per_sec_usec != USEC_INFINITY) mask |= CGROUP_CPUACCT | CGROUP_CPU; if (c->blockio_accounting || @@ -494,7 +459,8 @@ CGroupControllerMask cgroup_context_get_mask(CGroupContext *c) { c->memory_limit != (uint64_t) -1) mask |= CGROUP_MEMORY; - if (c->device_allow || c->device_policy != CGROUP_AUTO) + if (c->device_allow || + c->device_policy != CGROUP_AUTO) mask |= CGROUP_DEVICE; return mask; @@ -507,6 +473,19 @@ CGroupControllerMask unit_get_cgroup_mask(Unit *u) { if (!c) return 0; + /* If delegation is turned on, then turn on all cgroups, + * unless the process we fork into it is known to drop + * privileges anyway, and shouldn't get access to the + * controllers anyway. */ + + if (c->delegate) { + ExecContext *e; + + e = unit_get_exec_context(u); + if (!e || exec_context_maintains_privileges(e)) + return _CGROUP_CONTROLLER_MASK_ALL; + } + return cgroup_context_get_mask(c); } @@ -647,10 +626,8 @@ static int unit_create_cgroups(Unit *u, CGroupControllerMask mask) { /* First, create our own group */ r = cg_create_everywhere(u->manager->cgroup_supported, mask, u->cgroup_path); - if (r < 0) { - log_error("Failed to create cgroup %s: %s", u->cgroup_path, strerror(-r)); - return r; - } + if (r < 0) + return log_error_errno(r, "Failed to create cgroup %s: %m", u->cgroup_path); /* Keep track that this is now realized */ u->cgroup_realized = true; @@ -659,7 +636,7 @@ static int unit_create_cgroups(Unit *u, CGroupControllerMask mask) { /* Then, possibly move things over */ r = cg_migrate_everywhere(u->manager->cgroup_supported, u->cgroup_path, u->cgroup_path, migrate_callback, u); if (r < 0) - log_warning("Failed to migrate cgroup from to %s: %s", u->cgroup_path, strerror(-r)); + log_warning_errno(r, "Failed to migrate cgroup from to %s: %m", u->cgroup_path); return 0; } @@ -732,7 +709,7 @@ unsigned manager_dispatch_cgroup_queue(Manager *m) { r = unit_realize_cgroup_now(i, state); if (r < 0) - log_warning("Failed to realize cgroups for queued unit %s: %s", i->id, strerror(-r)); + log_warning_errno(r, "Failed to realize cgroups for queued unit %s: %m", i->id); n++; } @@ -815,7 +792,7 @@ void unit_destroy_cgroup(Unit *u) { r = cg_trim_everywhere(u->manager->cgroup_supported, u->cgroup_path, !unit_has_name(u, SPECIAL_ROOT_SLICE)); if (r < 0) - log_debug("Failed to destroy cgroup %s: %s", u->cgroup_path, strerror(-r)); + log_debug_errno(r, "Failed to destroy cgroup %s: %m", u->cgroup_path); hashmap_remove(u->manager->cgroup_unit, u->cgroup_path); @@ -865,7 +842,6 @@ pid_t unit_search_main_pid(Unit *u) { int manager_setup_cgroup(Manager *m) { _cleanup_free_ char *path = NULL; - char *e; int r; assert(m); @@ -875,15 +851,15 @@ int manager_setup_cgroup(Manager *m) { m->cgroup_root = NULL; r = cg_pid_get_path(SYSTEMD_CGROUP_CONTROLLER, 0, &m->cgroup_root); - if (r < 0) { - log_error("Cannot determine cgroup we are running in: %s", strerror(-r)); - return r; - } + if (r < 0) + return log_error_errno(r, "Cannot determine cgroup we are running in: %m"); /* LEGACY: Already in /system.slice? If so, let's cut this * off. This is to support live upgrades from older systemd * versions where PID 1 was moved there. */ if (m->running_as == SYSTEMD_SYSTEM) { + char *e; + e = endswith(m->cgroup_root, "/" SPECIAL_SYSTEM_SLICE); if (!e) e = endswith(m->cgroup_root, "/system"); @@ -899,46 +875,42 @@ int manager_setup_cgroup(Manager *m) { /* 2. Show data */ r = cg_get_path(SYSTEMD_CGROUP_CONTROLLER, m->cgroup_root, NULL, &path); - if (r < 0) { - log_error("Cannot find cgroup mount point: %s", strerror(-r)); - return r; - } + if (r < 0) + return log_error_errno(r, "Cannot find cgroup mount point: %m"); log_debug("Using cgroup controller " SYSTEMD_CGROUP_CONTROLLER ". File system hierarchy is at %s.", path); + if (!m->test_run) { - /* 3. Install agent */ - if (m->running_as == SYSTEMD_SYSTEM) { - r = cg_install_release_agent(SYSTEMD_CGROUP_CONTROLLER, SYSTEMD_CGROUP_AGENT_PATH); + /* 3. Install agent */ + if (m->running_as == SYSTEMD_SYSTEM) { + r = cg_install_release_agent(SYSTEMD_CGROUP_CONTROLLER, SYSTEMD_CGROUP_AGENT_PATH); + if (r < 0) + log_warning_errno(r, "Failed to install release agent, ignoring: %m"); + else if (r > 0) + log_debug("Installed release agent."); + else + log_debug("Release agent already installed."); + } + + /* 4. Make sure we are in the root cgroup */ + r = cg_create_and_attach(SYSTEMD_CGROUP_CONTROLLER, m->cgroup_root, 0); if (r < 0) - log_warning("Failed to install release agent, ignoring: %s", strerror(-r)); - else if (r > 0) - log_debug("Installed release agent."); - else - log_debug("Release agent already installed."); - } + return log_error_errno(r, "Failed to create root cgroup hierarchy: %m"); - /* 4. Make sure we are in the root cgroup */ - r = cg_create_and_attach(SYSTEMD_CGROUP_CONTROLLER, m->cgroup_root, 0); - if (r < 0) { - log_error("Failed to create root cgroup hierarchy: %s", strerror(-r)); - return r; - } + /* 5. And pin it, so that it cannot be unmounted */ + safe_close(m->pin_cgroupfs_fd); - /* 5. And pin it, so that it cannot be unmounted */ - safe_close(m->pin_cgroupfs_fd); + m->pin_cgroupfs_fd = open(path, O_RDONLY|O_CLOEXEC|O_DIRECTORY|O_NOCTTY|O_NONBLOCK); + if (m->pin_cgroupfs_fd < 0) + return log_error_errno(errno, "Failed to open pin file: %m"); - m->pin_cgroupfs_fd = open(path, O_RDONLY|O_CLOEXEC|O_DIRECTORY|O_NOCTTY|O_NONBLOCK); - if (m->pin_cgroupfs_fd < 0) { - log_error("Failed to open pin file: %m"); - return -errno; + /* 6. Always enable hierarchial support if it exists... */ + cg_set_attribute("memory", "/", "memory.use_hierarchy", "1"); } - /* 6. Figure out which controllers are supported */ + /* 7. Figure out which controllers are supported */ m->cgroup_supported = cg_mask_supported(); - /* 7. Always enable hierarchial support if it exists... */ - cg_set_attribute("memory", "/", "memory.use_hierarchy", "1"); - return 0; }