diff mbox series

[for-4.9.y] cgroup: Fix deadlock in cpu hotplug path

Message ID 1539195816-16015-1-git-send-email-amit.pundir@linaro.org
State New
Headers show
Series [for-4.9.y] cgroup: Fix deadlock in cpu hotplug path | expand

Commit Message

Amit Pundir Oct. 10, 2018, 6:23 p.m. UTC
From: Prateek Sood <prsood@codeaurora.org>


commit 116d2f7496c51b2e02e8e4ecdd2bdf5fb9d5a641 upstream.

Deadlock during cgroup migration from cpu hotplug path when a task T is
being moved from source to destination cgroup.

kworker/0:0
cpuset_hotplug_workfn()
   cpuset_hotplug_update_tasks()
      hotplug_update_tasks_legacy()
        remove_tasks_in_empty_cpuset()
          cgroup_transfer_tasks() // stuck in iterator loop
            cgroup_migrate()
              cgroup_migrate_add_task()

In cgroup_migrate_add_task() it checks for PF_EXITING flag of task T.
Task T will not migrate to destination cgroup. css_task_iter_start()
will keep pointing to task T in loop waiting for task T cg_list node
to be removed.

Task T
do_exit()
  exit_signals() // sets PF_EXITING
  exit_task_namespaces()
    switch_task_namespaces()
      free_nsproxy()
        put_mnt_ns()
          drop_collected_mounts()
            namespace_unlock()
              synchronize_rcu()
                _synchronize_rcu_expedited()
                  schedule_work() // on cpu0 low priority worker pool
                  wait_event() // waiting for work item to execute

Task T inserted a work item in the worklist of cpu0 low priority
worker pool. It is waiting for expedited grace period work item
to execute. This work item will only be executed once kworker/0:0
complete execution of cpuset_hotplug_workfn().

kworker/0:0 ==> Task T ==>kworker/0:0

In case of PF_EXITING task being migrated from source to destination
cgroup, migrate next available task in source cgroup.

Signed-off-by: Prateek Sood <prsood@codeaurora.org>

Signed-off-by: Tejun Heo <tj@kernel.org>

[AmitP: Upstream commit cherry-pick failed, so I picked the
        backported changes from CAF/msm-4.9 tree instead:
        https://source.codeaurora.org/quic/la/kernel/msm-4.9/commit/?id=49b74f1696417b270c89cd893ca9f37088928078]
Signed-off-by: Amit Pundir <amit.pundir@linaro.org>

---
This patch can be cleanly applied and build tested on 4.4.y and 3.18.y
as well but I couldn't find it in msm-4.4 and msm-3.18 trees. So this
patch is really untested on those stable trees.
Build tested on 4.9.131, 4.4.159 and 3.18.123 for ARCH=arm/arm64 allmodconfig.

 kernel/cgroup.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

-- 
2.7.4

Comments

Greg KH Oct. 11, 2018, 9:23 a.m. UTC | #1
On Wed, Oct 10, 2018 at 11:53:36PM +0530, Amit Pundir wrote:
> From: Prateek Sood <prsood@codeaurora.org>

> 

> commit 116d2f7496c51b2e02e8e4ecdd2bdf5fb9d5a641 upstream.

> 

> Deadlock during cgroup migration from cpu hotplug path when a task T is

> being moved from source to destination cgroup.

> 

> kworker/0:0

> cpuset_hotplug_workfn()

>    cpuset_hotplug_update_tasks()

>       hotplug_update_tasks_legacy()

>         remove_tasks_in_empty_cpuset()

>           cgroup_transfer_tasks() // stuck in iterator loop

>             cgroup_migrate()

>               cgroup_migrate_add_task()

> 

> In cgroup_migrate_add_task() it checks for PF_EXITING flag of task T.

> Task T will not migrate to destination cgroup. css_task_iter_start()

> will keep pointing to task T in loop waiting for task T cg_list node

> to be removed.

> 

> Task T

> do_exit()

>   exit_signals() // sets PF_EXITING

>   exit_task_namespaces()

>     switch_task_namespaces()

>       free_nsproxy()

>         put_mnt_ns()

>           drop_collected_mounts()

>             namespace_unlock()

>               synchronize_rcu()

>                 _synchronize_rcu_expedited()

>                   schedule_work() // on cpu0 low priority worker pool

>                   wait_event() // waiting for work item to execute

> 

> Task T inserted a work item in the worklist of cpu0 low priority

> worker pool. It is waiting for expedited grace period work item

> to execute. This work item will only be executed once kworker/0:0

> complete execution of cpuset_hotplug_workfn().

> 

> kworker/0:0 ==> Task T ==>kworker/0:0

> 

> In case of PF_EXITING task being migrated from source to destination

> cgroup, migrate next available task in source cgroup.

> 

> Signed-off-by: Prateek Sood <prsood@codeaurora.org>

> Signed-off-by: Tejun Heo <tj@kernel.org>

> [AmitP: Upstream commit cherry-pick failed, so I picked the

>         backported changes from CAF/msm-4.9 tree instead:

>         https://source.codeaurora.org/quic/la/kernel/msm-4.9/commit/?id=49b74f1696417b270c89cd893ca9f37088928078]

> Signed-off-by: Amit Pundir <amit.pundir@linaro.org>

> ---

> This patch can be cleanly applied and build tested on 4.4.y and 3.18.y

> as well but I couldn't find it in msm-4.4 and msm-3.18 trees. So this

> patch is really untested on those stable trees.

> Build tested on 4.9.131, 4.4.159 and 3.18.123 for ARCH=arm/arm64 allmodconfig.


Now applied, thanks.

greg k-h
diff mbox series

Patch

diff --git a/kernel/cgroup.c b/kernel/cgroup.c
index 4c233437ee1a..bb0cf1caf1cd 100644
--- a/kernel/cgroup.c
+++ b/kernel/cgroup.c
@@ -4386,7 +4386,11 @@  int cgroup_transfer_tasks(struct cgroup *to, struct cgroup *from)
 	 */
 	do {
 		css_task_iter_start(&from->self, &it);
-		task = css_task_iter_next(&it);
+
+		do {
+			task = css_task_iter_next(&it);
+		} while (task && (task->flags & PF_EXITING));
+
 		if (task)
 			get_task_struct(task);
 		css_task_iter_end(&it);