Skip to contentSkip to conversation

perf_events: Optimize perf_event_task_tick()

tip:perf/core tip-bot for Peter Zijlstra 1 participant 1 message Feb 4, 2010

lore
Share & tools
lore
#0 lore raw
Commit-ID:  9717e6cd3db22eade7dbae0fc9235c66325a7132
Gitweb:     http://git.kernel.org/tip/9717e6cd3db22eade7dbae0fc9235c66325a7132
Author:     Peter Zijlstra <[email protected]>
AuthorDate: Thu, 28 Jan 2010 13:57:44 +0100
Committer:  Ingo Molnar <[email protected]>
CommitDate: Thu, 4 Feb 2010 09:59:49 +0100

perf_events: Optimize perf_event_task_tick()

Pretty much all of the calls do perf_disable/perf_enable cycles, pull that out to cut back on hardware programming.

Cc Mike Galbraith <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Frederic Weisbecker <[email protected]>

LKML-Reference: <new-submission>

---
 kernel/perf_event.c |    8 ++++----
 1 files changed, 4 insertions(+), 4 deletions(-)
kernel/perf_event.c +4 -4
index 40f8b07..087025f 100644
@@ -1573,12 +1573,8 @@ static void rotate_ctx(struct perf_event_context *ctx)
 	raw_spin_lock(&ctx->lock);
 
 	/* Rotate the first entry last of non-pinned groups */
-	perf_disable();
-
 	list_rotate_left(&ctx->flexible_groups);
 
-	perf_enable();
-
 	raw_spin_unlock(&ctx->lock);
 }
 
@@ -1593,6 +1589,8 @@ void perf_event_task_tick(struct task_struct *curr)
 	cpuctx = &__get_cpu_var(perf_cpu_context);
 	ctx = curr->perf_event_ctxp;
 
+	perf_disable();
+
 	perf_ctx_adjust_freq(&cpuctx->ctx);
 	if (ctx)
 		perf_ctx_adjust_freq(ctx);
@@ -1608,6 +1606,8 @@ void perf_event_task_tick(struct task_struct *curr)
 	cpu_ctx_sched_in(cpuctx, EVENT_FLEXIBLE);
 	if (ctx)
 		task_ctx_sched_in(curr, EVENT_FLEXIBLE);
+
+	perf_enable();
 }
 
 static int event_enable_on_exec(struct perf_event *event,