2010-08-19 15:26:41

by David Howells

Subject: [PATCH] oom: __task_cred() need rcu_read_lock()

From: KOSAKI Motohiro <[email protected]>

dump_tasks() needs to hold the RCU read lock around its access of the target
task's UID. To this end it should use task_uid() as it only needs that one
thing from the creds.

The fact that dump_tasks() holds tasklist_lock is insufficient to prevent the
target process replacing its credentials on another CPU.

Then, this patch change to call rcu_read_lock() explicitly.


===================================================
[ INFO: suspicious rcu_dereference_check() usage. ]
---------------------------------------------------
mm/oom_kill.c:410 invoked rcu_dereference_check() without protection!

other info that might help us debug this:

rcu_scheduler_active = 1, debug_locks = 1
4 locks held by kworker/1:2/651:
#0: (events){+.+.+.}, at: [<ffffffff8106aae7>]
process_one_work+0x137/0x4a0
#1: (moom_work){+.+...}, at: [<ffffffff8106aae7>]
process_one_work+0x137/0x4a0
#2: (tasklist_lock){.+.+..}, at: [<ffffffff810fafd4>]
out_of_memory+0x164/0x3f0
#3: (&(&p->alloc_lock)->rlock){+.+...}, at: [<ffffffff810fa48e>]
find_lock_task_mm+0x2e/0x70

Signed-off-by: KOSAKI Motohiro <[email protected]>
Signed-off-by: David Howells <[email protected]>
---

mm/oom_kill.c | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/mm/oom_kill.c b/mm/oom_kill.c
index 5014e50..7b03102 100644
--- a/mm/oom_kill.c
+++ b/mm/oom_kill.c
@@ -372,7 +372,7 @@ static void dump_tasks(const struct mem_cgroup *mem)
}

pr_info("[%5d] %5d %5d %8lu %8lu %3u %3d %5d %s\n",
- task->pid, __task_cred(task)->uid, task->tgid,
+ task->pid, task_uid(task), task->tgid,
task->mm->total_vm, get_mm_rss(task->mm),
task_cpu(task), task->signal->oom_adj,
task->signal->oom_score_adj, task->comm);

2010-08-19 15:51:12

by Paul E. McKenney

Subject: Re: [PATCH] oom: __task_cred() need rcu_read_lock()

On Thu, Aug 19, 2010 at 04:26:18PM +0100, David Howells wrote:
> From: KOSAKI Motohiro <[email protected]>
>
> dump_tasks() needs to hold the RCU read lock around its access of the target
> task's UID. To this end it should use task_uid() as it only needs that one
> thing from the creds.
>
> The fact that dump_tasks() holds tasklist_lock is insufficient to prevent the
> target process replacing its credentials on another CPU.
>
> Then, this patch change to call rcu_read_lock() explicitly.
>
>
> ===================================================
> [ INFO: suspicious rcu_dereference_check() usage. ]
> ---------------------------------------------------
> mm/oom_kill.c:410 invoked rcu_dereference_check() without protection!
>
> other info that might help us debug this:
>
> rcu_scheduler_active = 1, debug_locks = 1
> 4 locks held by kworker/1:2/651:
> #0: (events){+.+.+.}, at: [<ffffffff8106aae7>]
> process_one_work+0x137/0x4a0
> #1: (moom_work){+.+...}, at: [<ffffffff8106aae7>]
> process_one_work+0x137/0x4a0
> #2: (tasklist_lock){.+.+..}, at: [<ffffffff810fafd4>]
> out_of_memory+0x164/0x3f0
> #3: (&(&p->alloc_lock)->rlock){+.+...}, at: [<ffffffff810fa48e>]
> find_lock_task_mm+0x2e/0x70
>
> Signed-off-by: KOSAKI Motohiro <[email protected]>
> Signed-off-by: David Howells <[email protected]>

Looks good to me!

Acked-by: Paul E. McKenney <[email protected]>

> ---
>
> mm/oom_kill.c | 2 +-
> 1 files changed, 1 insertions(+), 1 deletions(-)
>
> diff --git a/mm/oom_kill.c b/mm/oom_kill.c
> index 5014e50..7b03102 100644
> --- a/mm/oom_kill.c
> +++ b/mm/oom_kill.c
> @@ -372,7 +372,7 @@ static void dump_tasks(const struct mem_cgroup *mem)
> }
>
> pr_info("[%5d] %5d %5d %8lu %8lu %3u %3d %5d %s\n",
> - task->pid, __task_cred(task)->uid, task->tgid,
> + task->pid, task_uid(task), task->tgid,
> task->mm->total_vm, get_mm_rss(task->mm),
> task_cpu(task), task->signal->oom_adj,
> task->signal->oom_score_adj, task->comm);
>

2010-08-19 20:35:10

by David Rientjes

Subject: Re: [PATCH] oom: __task_cred() need rcu_read_lock()

On Thu, 19 Aug 2010, David Howells wrote:

> From: KOSAKI Motohiro <[email protected]>
>
> dump_tasks() needs to hold the RCU read lock around its access of the target
> task's UID. To this end it should use task_uid() as it only needs that one
> thing from the creds.
>
> The fact that dump_tasks() holds tasklist_lock is insufficient to prevent the
> target process replacing its credentials on another CPU.
>
> Then, this patch change to call rcu_read_lock() explicitly.
>
>
> ===================================================
> [ INFO: suspicious rcu_dereference_check() usage. ]
> ---------------------------------------------------
> mm/oom_kill.c:410 invoked rcu_dereference_check() without protection!
>
> other info that might help us debug this:
>
> rcu_scheduler_active = 1, debug_locks = 1
> 4 locks held by kworker/1:2/651:
> #0: (events){+.+.+.}, at: [<ffffffff8106aae7>]
> process_one_work+0x137/0x4a0
> #1: (moom_work){+.+...}, at: [<ffffffff8106aae7>]
> process_one_work+0x137/0x4a0
> #2: (tasklist_lock){.+.+..}, at: [<ffffffff810fafd4>]
> out_of_memory+0x164/0x3f0
> #3: (&(&p->alloc_lock)->rlock){+.+...}, at: [<ffffffff810fa48e>]
> find_lock_task_mm+0x2e/0x70
>
> Signed-off-by: KOSAKI Motohiro <[email protected]>
> Signed-off-by: David Howells <[email protected]>

Acked-by: David Rientjes <[email protected]>

2010-08-20 00:09:24

by KOSAKI Motohiro

Subject: Re: [PATCH] oom: __task_cred() need rcu_read_lock()

> From: KOSAKI Motohiro <[email protected]>
>
> dump_tasks() needs to hold the RCU read lock around its access of the target
> task's UID. To this end it should use task_uid() as it only needs that one
> thing from the creds.
>
> The fact that dump_tasks() holds tasklist_lock is insufficient to prevent the
> target process replacing its credentials on another CPU.
>
> Then, this patch change to call rcu_read_lock() explicitly.
>
>
> ===================================================
> [ INFO: suspicious rcu_dereference_check() usage. ]
> ---------------------------------------------------
> mm/oom_kill.c:410 invoked rcu_dereference_check() without protection!
>
> other info that might help us debug this:
>
> rcu_scheduler_active = 1, debug_locks = 1
> 4 locks held by kworker/1:2/651:
> #0: (events){+.+.+.}, at: [<ffffffff8106aae7>]
> process_one_work+0x137/0x4a0
> #1: (moom_work){+.+...}, at: [<ffffffff8106aae7>]
> process_one_work+0x137/0x4a0
> #2: (tasklist_lock){.+.+..}, at: [<ffffffff810fafd4>]
> out_of_memory+0x164/0x3f0
> #3: (&(&p->alloc_lock)->rlock){+.+...}, at: [<ffffffff810fa48e>]
> find_lock_task_mm+0x2e/0x70
>
> Signed-off-by: KOSAKI Motohiro <[email protected]>
> Signed-off-by: David Howells <[email protected]>
> ---
>
> mm/oom_kill.c | 2 +-
> 1 files changed, 1 insertions(+), 1 deletions(-)
>
> diff --git a/mm/oom_kill.c b/mm/oom_kill.c
> index 5014e50..7b03102 100644
> --- a/mm/oom_kill.c
> +++ b/mm/oom_kill.c
> @@ -372,7 +372,7 @@ static void dump_tasks(const struct mem_cgroup *mem)
> }
>
> pr_info("[%5d] %5d %5d %8lu %8lu %3u %3d %5d %s\n",
> - task->pid, __task_cred(task)->uid, task->tgid,
> + task->pid, task_uid(task), task->tgid,
> task->mm->total_vm, get_mm_rss(task->mm),
> task_cpu(task), task->signal->oom_adj,
> task->signal->oom_score_adj, task->comm);

Thank you!