core: unified cgroup hierarchy support

This patch set adds full support the new unified cgroup hierarchy logic
of modern kernels.

A new kernel command line option "systemd.unified_cgroup_hierarchy=1" is
added. If specified the unified hierarchy is mounted to /sys/fs/cgroup
instead of a tmpfs. No further hierarchies are mounted. The kernel
command line option defaults to off. We can turn it on by default as
soon as the kernel's APIs regarding this are stabilized (but even then
downstream distros might want to turn this off, as this will break any
tools that access cgroupfs directly).

It is possibly to choose for each boot individually whether the unified
or the legacy hierarchy is used. nspawn will by default provide the
legacy hierarchy to containers if the host is using it, and the unified
otherwise. However it is possible to run containers with the unified
hierarchy on a legacy host and vice versa, by setting the
$UNIFIED_CGROUP_HIERARCHY environment variable for nspawn to 1 or 0,
respectively.

The unified hierarchy provides reliable cgroup empty notifications for
the first time, via inotify. To make use of this we maintain one
manager-wide inotify fd, and each cgroup to it.

This patch also removes cg_delete() which is unused now.

On kernel 4.2 only the "memory" controller is compatible with the
unified hierarchy, hence that's the only controller systemd exposes when
booted in unified heirarchy mode.

This introduces a new enum for enumerating supported controllers, plus a
related enum for the mask bits mapping to it. The core is changed to
make use of this everywhere.

This moves PID 1 into a new "init.scope" implicit scope unit in the root
slice. This is necessary since on the unified hierarchy cgroups may
either contain subgroups or processes but not both. PID 1 hence has to
move out of the root cgroup (strictly speaking the root cgroup is the
only one where processes and subgroups are still allowed, but in order
to support containers nicey, we move PID 1 into the new scope in all
cases.) This new unit is also used on legacy hierarchy setups. It's
actually pretty useful on all systems, as it can then be used to filter
journal messages coming from PID 1, and so on.

The root slice ("-.slice") is now implicitly created and started (and
does not require a unit file on disk anymore), since
that's where "init.scope" is located and the slice needs to be started
before the scope can.

To check whether we are in unified or legacy hierarchy mode we use
statfs() on /sys/fs/cgroup. If the .f_type field reports tmpfs we are in
legacy mode, if it reports cgroupfs we are in unified mode.

This patch set carefuly makes sure that cgls and cgtop continue to work
as desired.

When invoking nspawn as a service it will implicitly create two
subcgroups in the cgroup it is using, one to move the nspawn process
into, the other to move the actual container processes into. This is
done because of the requirement that cgroups may either contain
processes or other subgroups.

1 files changed, 27 insertions, 141 deletions

diff --git a/src/core/unit.c b/src/core/unit.c
index 34d3adcd3b..8c07c6140d 100644
--- a/src/core/unit.c
+++ b/src/core/unit.c
@@ -91,6 +91,7 @@ Unit *unit_new(Manager *m, size_t size) {
         u->unit_file_state = _UNIT_FILE_STATE_INVALID;
         u->unit_file_preset = -1;
         u->on_failure_job_mode = JOB_REPLACE;
+        u->cgroup_inotify_wd = -1;
 
         RATELIMIT_INIT(u->auto_stop_ratelimit, 10 * USEC_PER_SEC, 16);
 
@@ -525,10 +526,7 @@ void unit_free(Unit *u) {
         if (u->in_cgroup_queue)
                 LIST_REMOVE(cgroup_queue, u->manager->cgroup_queue, u);
 
-        if (u->cgroup_path) {
-                hashmap_remove(u->manager->cgroup_unit, u->cgroup_path);
-                free(u->cgroup_path);
-        }
+        unit_release_cgroup(u);
 
         manager_update_failed_units(u->manager, u, false);
         set_remove(u->manager->startup_units, u);
@@ -1801,7 +1799,7 @@ void unit_notify(Unit *u, UnitActiveState os, UnitActiveState ns, bool reload_su
 
         /* Make sure the cgroup is always removed when we become inactive */
         if (UNIT_IS_INACTIVE_OR_FAILED(ns))
-                unit_destroy_cgroup_if_empty(u);
+                unit_prune_cgroup(u);
 
         /* Note that this doesn't apply to RemainAfterExit services exiting
          * successfully, since there's no change of state in that case. Which is
@@ -2028,70 +2026,7 @@ void unit_unwatch_all_pids(Unit *u) {
         while (!set_isempty(u->pids))
                 unit_unwatch_pid(u, PTR_TO_LONG(set_first(u->pids)));
 
-        set_free(u->pids);
-        u->pids = NULL;
-}
-
-static int unit_watch_pids_in_path(Unit *u, const char *path) {
-        _cleanup_closedir_ DIR *d = NULL;
-        _cleanup_fclose_ FILE *f = NULL;
-        int ret = 0, r;
-
-        assert(u);
-        assert(path);
-
-        /* Adds all PIDs from a specific cgroup path to the set of PIDs we watch. */
-
-        r = cg_enumerate_processes(SYSTEMD_CGROUP_CONTROLLER, path, &f);
-        if (r >= 0) {
-                pid_t pid;
-
-                while ((r = cg_read_pid(f, &pid)) > 0) {
-                        r = unit_watch_pid(u, pid);
-                        if (r < 0 && ret >= 0)
-                                ret = r;
-                }
-                if (r < 0 && ret >= 0)
-                        ret = r;
-
-        } else if (ret >= 0)
-                ret = r;
-
-        r = cg_enumerate_subgroups(SYSTEMD_CGROUP_CONTROLLER, path, &d);
-        if (r >= 0) {
-                char *fn;
-
-                while ((r = cg_read_subgroup(d, &fn)) > 0) {
-                        _cleanup_free_ char *p = NULL;
-
-                        p = strjoin(path, "/", fn, NULL);
-                        free(fn);
-
-                        if (!p)
-                                return -ENOMEM;
-
-                        r = unit_watch_pids_in_path(u, p);
-                        if (r < 0 && ret >= 0)
-                                ret = r;
-                }
-                if (r < 0 && ret >= 0)
-                        ret = r;
-
-        } else if (ret >= 0)
-                ret = r;
-
-        return ret;
-}
-
-int unit_watch_all_pids(Unit *u) {
-        assert(u);
-
-        /* Adds all PIDs from our cgroup to the set of PIDs we watch */
-
-        if (!u->cgroup_path)
-                return -ENOENT;
-
-        return unit_watch_pids_in_path(u, u->cgroup_path);
+        u->pids = set_free(u->pids);
 }
 
 void unit_tidy_watch_pids(Unit *u, pid_t except1, pid_t except2) {
@@ -2400,31 +2335,6 @@ char *unit_dbus_path(Unit *u) {
         return unit_dbus_path_from_name(u->id);
 }
 
-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_slice(Unit *u, Unit *slice) {
         assert(u);
         assert(slice);
@@ -2447,6 +2357,10 @@ int unit_set_slice(Unit *u, Unit *slice) {
         if (slice->type != UNIT_SLICE)
                 return -EINVAL;
 
+        if (unit_has_name(u, SPECIAL_INIT_SCOPE) &&
+            !unit_has_name(slice, SPECIAL_ROOT_SLICE))
+                return -EPERM;
+
         if (UNIT_DEREF(u->slice) == slice)
                 return 0;
 
@@ -2495,7 +2409,7 @@ int unit_set_default_slice(Unit *u) {
                 slice_name = b;
         } else
                 slice_name =
-                        u->manager->running_as == MANAGER_SYSTEM
+                        u->manager->running_as == MANAGER_SYSTEM && !unit_has_name(u, SPECIAL_INIT_SCOPE)
                         ? SPECIAL_SYSTEM_SLICE
                         : SPECIAL_ROOT_SLICE;
 
@@ -2704,40 +2618,6 @@ void unit_serialize_item(Unit *u, FILE *f, const char *key, const char *value) {
         fprintf(f, "%s=%s\n", key, value);
 }
 
-static 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;
-        }
-
-        if (u->cgroup_path) {
-                log_unit_debug(u, "Changing cgroup path from %s to %s.", u->cgroup_path, strna(p));
-                hashmap_remove(u->manager->cgroup_unit, u->cgroup_path);
-                free(u->cgroup_path);
-        }
-
-        u->cgroup_path = p;
-        p = NULL;
-
-        return 0;
-}
-
 int unit_deserialize(Unit *u, FILE *f, FDSet *fds) {
         ExecRuntime **rt = NULL;
         size_t offset;
@@ -2868,6 +2748,8 @@ int unit_deserialize(Unit *u, FILE *f, FDSet *fds) {
                         if (r < 0)
                                 log_unit_debug_errno(u, r, "Failed to set cgroup path %s, ignoring: %m", v);
 
+                        (void) unit_watch_cgroup(u);
+
                         continue;
                 } else if (streq(l, "cgroup-realized")) {
                         int b;
@@ -3600,18 +3482,22 @@ int unit_kill_context(
 
                 } else if (r > 0) {
 
-                        /* FIXME: For now, we will not wait for the
-                         * cgroup members to die if we are running in
-                         * a container or if this is a delegation
-                         * unit, simply because cgroup notification is
-                         * unreliable in these cases. It doesn't work
-                         * at all in containers, and outside of
-                         * containers it can be confused easily by
-                         * left-over directories in the cgroup --
-                         * which however should not exist in
-                         * non-delegated units. */
-
-                        if  (detect_container(NULL) == 0 && !unit_cgroup_delegate(u))
+                        /* FIXME: For now, on the legacy hierarchy, we
+                         * will not wait for the cgroup members to die
+                         * if we are running in a container or if this
+                         * is a delegation unit, simply because cgroup
+                         * notification is unreliable in these
+                         * cases. It doesn't work at all in
+                         * containers, and outside of containers it
+                         * can be confused easily by left-over
+                         * directories in the cgroup -- which however
+                         * should not exist in non-delegated units. On
+                         * the unified hierarchy that's different,
+                         * there we get proper events. Hence rely on
+                         * them.*/
+
+                        if  (cg_unified() > 0 ||
+                             (detect_container(NULL) == 0 && !unit_cgroup_delegate(u)))
                                 wait_for_exit = true;
 
                         if (c->send_sighup && k != KILL_KILL) {