OSDN Git Service

completion: Use lockdep_assert_RT_in_threaded_ctx() in complete_all()
authorSebastian Siewior <bigeasy@linutronix.de>
Mon, 23 Mar 2020 15:20:19 +0000 (16:20 +0100)
committerPeter Zijlstra <peterz@infradead.org>
Mon, 23 Mar 2020 17:40:25 +0000 (18:40 +0100)
The warning was intended to spot complete_all() users from hardirq
context on PREEMPT_RT. The warning as-is will also trigger in interrupt
handlers, which are threaded on PREEMPT_RT, which was not intended.

Use lockdep_assert_RT_in_threaded_ctx() which triggers in non-preemptive
context on PREEMPT_RT.

Fixes: a5c6234e1028 ("completion: Use simple wait queues")
Reported-by: kernel test robot <rong.a.chen@intel.com>
Suggested-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20200323152019.4qjwluldohuh3by5@linutronix.de
include/linux/lockdep.h
kernel/sched/completion.c

index 425b4ce..206774a 100644 (file)
@@ -711,6 +711,21 @@ do {                                                                       \
 # define lockdep_assert_in_irq() do { } while (0)
 #endif
 
+#ifdef CONFIG_PROVE_RAW_LOCK_NESTING
+
+# define lockdep_assert_RT_in_threaded_ctx() do {                      \
+               WARN_ONCE(debug_locks && !current->lockdep_recursion && \
+                         current->hardirq_context &&                   \
+                         !(current->hardirq_threaded || current->irq_config),  \
+                         "Not in threaded context on PREEMPT_RT as expected\n");       \
+} while (0)
+
+#else
+
+# define lockdep_assert_RT_in_threaded_ctx() do { } while (0)
+
+#endif
+
 #ifdef CONFIG_LOCKDEP
 void lockdep_rcu_suspicious(const char *file, const int line, const char *s);
 #else
index f15e961..a778554 100644 (file)
@@ -58,7 +58,7 @@ void complete_all(struct completion *x)
 {
        unsigned long flags;
 
-       WARN_ON(irqs_disabled());
+       lockdep_assert_RT_in_threaded_ctx();
 
        raw_spin_lock_irqsave(&x->wait.lock, flags);
        x->done = UINT_MAX;