[systemd-devel] [PATCHSET RE-RESEND] update unified hierarchy support
Tejun Heo
htejun at fb.com
Fri Mar 25 18:34:00 UTC 2016
(sorry, of course forgot to attach the patches)
(bounced for not being subscribed, resending...)
Hello,
Unified hierarchy is available on the 4.5 kernel but there have been
several updates.
1. The __DEVEL__sane_behavior flag is gone. Unified hierarchy is now
available as "cgroup2" filesystem type with its own super magic
number.
2. "cgroup.populated" file is replaced with "populated" field of
"cgroup.events" file.
3. A zombie task remains associated with the cgroup it was associated
with at the time of death instead of being moved immediately to
root. This means that pid to unit lookup may return a slice if the
session or service unit the pid belonged to is already gone.
Three patches are attached addressing each of the above.
Thanks!
--
tejun
-------------- next part --------------
>From 278a39f0a8fa34cd899c6a08e76626c987a4713e Mon Sep 17 00:00:00 2001
From: Tejun Heo <htejun at fb.com>
Date: Fri, 25 Mar 2016 11:38:50 -0400
Subject: [PATCH 1/3] core: update unified hierarchy support
Unified hierarchy is official as of Linux v4.5 and now available through a new
filesystem type, cgroup2, with its own super magic. Update mount logic
accordingly.
Signed-off-by: Tejun Heo <htejun at fb.com>
---
src/basic/cgroup-util.c | 2 +-
src/basic/missing.h | 4 ++++
src/core/mount-setup.c | 2 +-
3 files changed, 6 insertions(+), 2 deletions(-)
diff --git a/src/basic/cgroup-util.c b/src/basic/cgroup-util.c
index 56c1fca..5124b5b 100644
--- a/src/basic/cgroup-util.c
+++ b/src/basic/cgroup-util.c
@@ -2129,7 +2129,7 @@ int cg_unified(void) {
if (statfs("/sys/fs/cgroup/", &fs) < 0)
return -errno;
- if (F_TYPE_EQUAL(fs.f_type, CGROUP_SUPER_MAGIC))
+ if (F_TYPE_EQUAL(fs.f_type, CGROUP2_SUPER_MAGIC))
unified_cache = true;
else if (F_TYPE_EQUAL(fs.f_type, TMPFS_MAGIC))
unified_cache = false;
diff --git a/src/basic/missing.h b/src/basic/missing.h
index 034e334..66cd592 100644
--- a/src/basic/missing.h
+++ b/src/basic/missing.h
@@ -437,6 +437,10 @@ struct btrfs_ioctl_quota_ctl_args {
#define CGROUP_SUPER_MAGIC 0x27e0eb
#endif
+#ifndef CGROUP2_SUPER_MAGIC
+#define CGROUP2_SUPER_MAGIC 0x63677270
+#endif
+
#ifndef TMPFS_MAGIC
#define TMPFS_MAGIC 0x01021994
#endif
diff --git a/src/core/mount-setup.c b/src/core/mount-setup.c
index de1a361..32fe51c 100644
--- a/src/core/mount-setup.c
+++ b/src/core/mount-setup.c
@@ -94,7 +94,7 @@ static const MountPoint mount_table[] = {
#endif
{ "tmpfs", "/run", "tmpfs", "mode=755", MS_NOSUID|MS_NODEV|MS_STRICTATIME,
NULL, MNT_FATAL|MNT_IN_CONTAINER },
- { "cgroup", "/sys/fs/cgroup", "cgroup", "__DEVEL__sane_behavior", MS_NOSUID|MS_NOEXEC|MS_NODEV,
+ { "cgroup", "/sys/fs/cgroup", "cgroup2", NULL, MS_NOSUID|MS_NOEXEC|MS_NODEV,
cg_is_unified_wanted, MNT_FATAL|MNT_IN_CONTAINER },
{ "tmpfs", "/sys/fs/cgroup", "tmpfs", "mode=755", MS_NOSUID|MS_NOEXEC|MS_NODEV|MS_STRICTATIME,
cg_is_legacy_wanted, MNT_FATAL|MNT_IN_CONTAINER },
--
2.5.5
-------------- next part --------------
>From 0fed0c3cdebe72557db528572ed2c531e32e7d5a Mon Sep 17 00:00:00 2001
From: Tejun Heo <htejun at fb.com>
Date: Fri, 25 Mar 2016 11:38:50 -0400
Subject: [PATCH 2/3] core: update populated event handling in unified
hierarchy
Earlier during the development of unified hierarchy, the populated event was
reported through by the dedicated "cgroup.populated" file; however, the
interface was updated so that it's reported through the "populated" field of
"cgroup.events" file. Update populated event handling logic accordingly.
Signed-off-by: Tejun Heo <htejun at fb.com>
---
src/basic/cgroup-util.c | 45 ++++++++++++++++++++++++++++++++++++---------
src/basic/cgroup-util.h | 2 ++
src/core/cgroup.c | 6 +++---
src/nspawn/nspawn-cgroup.c | 3 +--
4 files changed, 42 insertions(+), 14 deletions(-)
diff --git a/src/basic/cgroup-util.c b/src/basic/cgroup-util.c
index 5124b5b..5043180 100644
--- a/src/basic/cgroup-util.c
+++ b/src/basic/cgroup-util.c
@@ -101,6 +101,39 @@ int cg_read_pid(FILE *f, pid_t *_pid) {
return 1;
}
+int cg_read_event(const char *controller, const char *path, const char *event,
+ char **val)
+{
+ _cleanup_free_ char *events = NULL, *content = NULL;
+ char *p, *line;
+ int r;
+
+ r = cg_get_path(controller, path, "cgroup.events", &events);
+ if (r < 0)
+ return r;
+
+ r = read_full_file(events, &content, NULL);
+ if (r < 0)
+ return r;
+
+ p = content;
+ while ((line = strsep(&p, "\n"))) {
+ char *key;
+
+ key = strsep(&line, " ");
+ if (!key || !line)
+ return -EINVAL;
+
+ if (strcmp(key, event))
+ continue;
+
+ *val = strdup(line);
+ return 0;
+ }
+
+ return -ENOENT;
+}
+
int cg_enumerate_subgroups(const char *controller, const char *path, DIR **_d) {
_cleanup_free_ char *fs = NULL;
int r;
@@ -1007,18 +1040,12 @@ int cg_is_empty_recursive(const char *controller, const char *path) {
return unified;
if (unified > 0) {
- _cleanup_free_ char *populated = NULL, *t = NULL;
+ _cleanup_free_ char *t = NULL;
/* On the unified hierarchy we can check empty state
- * via the "cgroup.populated" attribute. */
+ * via the "populated" attribute of "cgroup.events". */
- r = cg_get_path(controller, path, "cgroup.populated", &populated);
- if (r < 0)
- return r;
-
- r = read_one_line_file(populated, &t);
- if (r == -ENOENT)
- return 1;
+ r = cg_read_event(controller, path, "populated", &t);
if (r < 0)
return r;
diff --git a/src/basic/cgroup-util.h b/src/basic/cgroup-util.h
index ad1edd9..4254e51 100644
--- a/src/basic/cgroup-util.h
+++ b/src/basic/cgroup-util.h
@@ -96,6 +96,8 @@ static inline bool CGROUP_BLKIO_WEIGHT_IS_OK(uint64_t x) {
int cg_enumerate_processes(const char *controller, const char *path, FILE **_f);
int cg_read_pid(FILE *f, pid_t *_pid);
+int cg_read_event(const char *controller, const char *path, const char *event,
+ char **val);
int cg_enumerate_subgroups(const char *controller, const char *path, DIR **_d);
int cg_read_subgroup(DIR *d, char **fn);
diff --git a/src/core/cgroup.c b/src/core/cgroup.c
index 39235a9..9c34928 100644
--- a/src/core/cgroup.c
+++ b/src/core/cgroup.c
@@ -765,7 +765,7 @@ int unit_set_cgroup_path(Unit *u, const char *path) {
}
int unit_watch_cgroup(Unit *u) {
- _cleanup_free_ char *populated = NULL;
+ _cleanup_free_ char *events = NULL;
int r;
assert(u);
@@ -791,11 +791,11 @@ int unit_watch_cgroup(Unit *u) {
if (r < 0)
return log_oom();
- r = cg_get_path(SYSTEMD_CGROUP_CONTROLLER, u->cgroup_path, "cgroup.populated", &populated);
+ r = cg_get_path(SYSTEMD_CGROUP_CONTROLLER, u->cgroup_path, "cgroup.events", &events);
if (r < 0)
return log_oom();
- u->cgroup_inotify_wd = inotify_add_watch(u->manager->cgroup_inotify_fd, populated, IN_MODIFY);
+ u->cgroup_inotify_wd = inotify_add_watch(u->manager->cgroup_inotify_fd, events, IN_MODIFY);
if (u->cgroup_inotify_wd < 0) {
if (errno == ENOENT) /* If the directory is already
diff --git a/src/nspawn/nspawn-cgroup.c b/src/nspawn/nspawn-cgroup.c
index 9f9a475..a63c30e 100644
--- a/src/nspawn/nspawn-cgroup.c
+++ b/src/nspawn/nspawn-cgroup.c
@@ -55,8 +55,7 @@ int chown_cgroup(pid_t pid, uid_t uid_shift) {
"cgroup.events",
"cgroup.clone_children",
"cgroup.controllers",
- "cgroup.subtree_control",
- "cgroup.populated")
+ "cgroup.subtree_control")
if (fchownat(fd, fn, uid_shift, uid_shift, 0) < 0)
log_full_errno(errno == ENOENT ? LOG_DEBUG : LOG_WARNING, errno,
"Failed to chown() cgroup file %s, ignoring: %m", fn);
--
2.5.5
-------------- next part --------------
>From 80d7d0ad42ad910acd06517f3f3862eeb1c9dc59 Mon Sep 17 00:00:00 2001
From: Tejun Heo <htejun at fb.com>
Date: Fri, 25 Mar 2016 11:38:50 -0400
Subject: [PATCH 3/3] core: update invoke_sigchld_event() to handle NULL
->sigchld_event()
After receiving SIGCHLD, one of the ways manager_dispatch_sigchld() maps the
now zombie $PID to its unit is through manager_get_unit_by_pid_cgroup() which
reads /proc/$PID/cgroup and looks up the unit associated with the cgroup path.
On non-unified cgroup hierarchies, a process is immediately migrated to the
root cgroup on death and the cgroup lookup would always have returned the unit
associated with it, making it rather pointless but safe. On unified hierarchy,
a zombie remains associated with the cgroup that it was associated with at the
time of death and thus manager_get_unit_by_pid_cgroup() will look up the unit
properly.
However, by the time manager_dispatch_sigchld() is running, the original cgroup
may have become empty and it and its associated unit might already have been
removed. If the cgroup path doesn't yield a match, manager_dispatch_sigchld()
keeps pruning the leaf component. This means that the function may return a
slice unit for a pid and as a slice doesn't have ->sigchld_event() handler,
calling invoke_sigchld_event() on it causes a segfault.
This patch updates invoke_sigchld_event() so that it skips calling if the
handler is not set.
Signed-off-by: Tejun Heo <htejun at fb.com>
---
src/core/manager.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/src/core/manager.c b/src/core/manager.c
index f13e933..78f2a21 100644
--- a/src/core/manager.c
+++ b/src/core/manager.c
@@ -1631,7 +1631,9 @@ static void invoke_sigchld_event(Manager *m, Unit *u, const siginfo_t *si) {
log_unit_debug(u, "Child "PID_FMT" belongs to %s", si->si_pid, u->id);
unit_unwatch_pid(u, si->si_pid);
- UNIT_VTABLE(u)->sigchld_event(u, si->si_pid, si->si_code, si->si_status);
+
+ if (UNIT_VTABLE(u)->sigchld_event)
+ UNIT_VTABLE(u)->sigchld_event(u, si->si_pid, si->si_code, si->si_status);
}
static int manager_dispatch_sigchld(Manager *m) {
--
2.5.5
More information about the systemd-devel
mailing list