Re: [PATCH v5 00/10] latched RB-trees and __module_address()

[Rusty Russell <rusty@rustcorp.com.au>]

Peter Zijlstra <peterz@infradead.org> writes:
> On Tue, Apr 14, 2015 at 12:27:05PM +0930, Rusty Russell wrote:

Hmm, this blows up:

> +static void module_assert_mutex_or_preempt(void)
> +{
> +#ifdef CONFIG_LOCKDEP
> +	if (!unlikely(debug_locks))
> +		return;
> +
> +	WARN_ON(!rcu_held_lock_sched_held() &&
> +		!lockdep_is_held(&module_mutex));

That's rcu_read_lock_sched_held, not rcu_held_lock_sched_held().

Also, your unlikely is weird and backwards.

I changed it as below, and folded.  It's in modules-next.

I'm now going to do some *actual* testing, and I'll go all Torvalds on
you if this spews warnings...

Cheers,
Rusty.

diff --git a/kernel/module.c b/kernel/module.c
index 4a89b88b4428..a15899e00ca9 100644
--- a/kernel/module.c
+++ b/kernel/module.c
@@ -113,10 +113,10 @@ static void module_assert_mutex(void)
 static void module_assert_mutex_or_preempt(void)
 {
 #ifdef CONFIG_LOCKDEP
-	if (!unlikely(debug_locks))
+	if (unlikely(!debug_locks))
 		return;
 
-	WARN_ON(!rcu_held_lock_sched_held() &&
+	WARN_ON(!rcu_read_lock_sched_held() &&
 		!lockdep_is_held(&module_mutex));
 #endif
 }