perf: Don't disable preemption in perf_pending_task().

perf_pending_task() is invoked in task context and disables preemption
because perf_swevent_get_recursion_context() used to access per-CPU
variables. The other reason is to create a RCU read section while
accessing the perf_event.

The recursion counter is no longer a per-CPU accounter so disabling
preemption is no longer required. The RCU section is needed and must be
created explicit.

Replace the preemption-disable section with a explicit RCU-read section.

Signed-off-by: Sebastian Andrzej Siewior <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Tested-by: Marco Elver <[email protected]>
Link: https://lore.kernel.org/r/[email protected]

Author: Sebastian Andrzej Siewior

kernel/events/core.c

@@ -5208,10 +5208,9 @@ static void perf_pending_task_sync(struct perf_event *event)
 	}
 
 	/*
-	 * All accesses related to the event are within the same
-	 * non-preemptible section in perf_pending_task(). The RCU
-	 * grace period before the event is freed will make sure all
-	 * those accesses are complete by then.
+	 * All accesses related to the event are within the same RCU section in
+	 * perf_pending_task(). The RCU grace period before the event is freed
+	 * will make sure all those accesses are complete by then.
 	 */
 	rcuwait_wait_event(&event->pending_work_wait, !event->pending_work, TASK_UNINTERRUPTIBLE);
 }
@@ -6831,7 +6830,7 @@ static void perf_pending_task(struct callback_head *head)
 	 * critical section as the ->pending_work reset. See comment in
 	 * perf_pending_task_sync().
 	 */
-	preempt_disable_notrace();
+	rcu_read_lock();
 	/*
 	 * If we 'fail' here, that's OK, it means recursion is already disabled
 	 * and we won't recurse 'further'.
@@ -6844,10 +6843,10 @@ static void perf_pending_task(struct callback_head *head)
 		local_dec(&event->ctx->nr_pending);
 		rcuwait_wake_up(&event->pending_work_wait);
 	}
+	rcu_read_unlock();
 
 	if (rctx >= 0)
 		perf_swevent_put_recursion_context(rctx);
-	preempt_enable_notrace();
 }
 
 #ifdef CONFIG_GUEST_PERF_EVENTS