Re: [RFC PATCH] tracing: Don't call wakeup() when committing the event

From: Vaibhav Nagarnaik
Date: Tue May 03 2011 - 19:28:01 EST


On Tue, May 3, 2011 at 4:08 PM, Frederic Weisbecker <fweisbec@xxxxxxxxx> wrote:
> On Tue, May 03, 2011 at 06:47:32PM -0400, Steven Rostedt wrote:
>> On Wed, 2011-05-04 at 00:09 +0200, Frederic Weisbecker wrote:
>>
>> > I wonder if we should have a lite version of wake_up() that checks
>> > if the list of waiters is empty before locking the queue.
>> > After all we don't care much about tight races for tracing.
>>
>> Since tracing is a special case here, we probably could get away to just
>> do it ourselves. That is, have our own flag that determines if we should
>> wakeup or not.
>
> Yep, agreed.
>
How about this? Check if the waitqueue is active and if true, operate on the
trace_wait queue.

diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
index 06f4458..e03ee24 100644
--- a/kernel/trace/trace.c
+++ b/kernel/trace/trace.c
@@ -353,7 +353,7 @@ void trace_wake_up(void)
{
int cpu;

- if (trace_flags & TRACE_ITER_BLOCK)
+ if (trace_flags & TRACE_ITER_BLOCK || !waitqueue_active(&trace_wait))
return;
/*
* The runqueue_is_locked() can fail, but this is the best we

Vaibhav Nagarnaik
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/