Skip to content

Commit

Permalink
perf: Fix perf_lock_task_context() vs RCU
Browse files Browse the repository at this point in the history
Jiri managed to trigger this warning:

 [] ======================================================
 [] [ INFO: possible circular locking dependency detected ]
 [] 3.10.0+ #228 Tainted: G        W
 [] -------------------------------------------------------
 [] p/6613 is trying to acquire lock:
 []  (rcu_node_0){..-...}, at: [<ffffffff810ca797>] rcu_read_unlock_special+0xa7/0x250
 []
 [] but task is already holding lock:
 []  (&ctx->lock){-.-...}, at: [<ffffffff810f2879>] perf_lock_task_context+0xd9/0x2c0
 []
 [] which lock already depends on the new lock.
 []
 [] the existing dependency chain (in reverse order) is:
 []
 [] -> CyanogenMod#4 (&ctx->lock){-.-...}:
 [] -> CyanogenMod#3 (&rq->lock){-.-.-.}:
 [] -> CyanogenMod#2 (&p->pi_lock){-.-.-.}:
 [] -> #1 (&rnp->nocb_gp_wq[1]){......}:
 [] -> #0 (rcu_node_0){..-...}:

Paul was quick to explain that due to preemptible RCU we cannot call
rcu_read_unlock() while holding scheduler (or nested) locks when part
of the read side critical section was preemptible.

Therefore solve it by making the entire RCU read side non-preemptible.

Also pull out the retry from under the non-preempt to play nice with RT.

Reported-by: Jiri Olsa <[email protected]>
Helped-out-by: Paul E. McKenney <[email protected]>
Cc: <[email protected]>
Signed-off-by: Peter Zijlstra <[email protected]>
Signed-off-by: Ingo Molnar <[email protected]>
  • Loading branch information
Peter Zijlstra authored and Ingo Molnar committed Jul 12, 2013
1 parent 06f4179 commit 058ebd0
Showing 1 changed file with 14 additions and 1 deletion.
15 changes: 14 additions & 1 deletion kernel/events/core.c
Original file line number Diff line number Diff line change
Expand Up @@ -947,8 +947,18 @@ perf_lock_task_context(struct task_struct *task, int ctxn, unsigned long *flags)
{
struct perf_event_context *ctx;

rcu_read_lock();
retry:
/*
* One of the few rules of preemptible RCU is that one cannot do
* rcu_read_unlock() while holding a scheduler (or nested) lock when
* part of the read side critical section was preemptible -- see
* rcu_read_unlock_special().
*
* Since ctx->lock nests under rq->lock we must ensure the entire read
* side critical section is non-preemptible.
*/
preempt_disable();
rcu_read_lock();
ctx = rcu_dereference(task->perf_event_ctxp[ctxn]);
if (ctx) {
/*
Expand All @@ -964,6 +974,8 @@ perf_lock_task_context(struct task_struct *task, int ctxn, unsigned long *flags)
raw_spin_lock_irqsave(&ctx->lock, *flags);
if (ctx != rcu_dereference(task->perf_event_ctxp[ctxn])) {
raw_spin_unlock_irqrestore(&ctx->lock, *flags);
rcu_read_unlock();
preempt_enable();
goto retry;
}

Expand All @@ -973,6 +985,7 @@ perf_lock_task_context(struct task_struct *task, int ctxn, unsigned long *flags)
}
}
rcu_read_unlock();
preempt_enable();
return ctx;
}

Expand Down

0 comments on commit 058ebd0

Please sign in to comment.