From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1759072AbZEKKPy (ORCPT ); Mon, 11 May 2009 06:15:54 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1757940AbZEKKNq (ORCPT ); Mon, 11 May 2009 06:13:46 -0400 Received: from hera.kernel.org ([140.211.167.34]:50845 "EHLO hera.kernel.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1758730AbZEKKNp (ORCPT ); Mon, 11 May 2009 06:13:45 -0400 Date: Mon, 11 May 2009 10:12:41 GMT From: tip-bot for Paul Mackerras To: linux-tip-commits@vger.kernel.org Cc: linux-kernel@vger.kernel.org, paulus@samba.org, hpa@zytor.com, mingo@redhat.com, a.p.zijlstra@chello.nl, tglx@linutronix.de, cjashfor@linux.vnet.ibm.com, mingo@elte.hu Reply-To: mingo@redhat.com, hpa@zytor.com, paulus@samba.org, linux-kernel@vger.kernel.org, a.p.zijlstra@chello.nl, tglx@linutronix.de, cjashfor@linux.vnet.ibm.com, mingo@elte.hu In-Reply-To: <18951.48034.485580.498953@drongo.ozlabs.ibm.com> References: <18951.48034.485580.498953@drongo.ozlabs.ibm.com> Subject: [tip:perfcounters/core] perf_counter: don't count scheduler ticks as context switches Message-ID: Git-Commit-ID: a08b159fc243dbfe415250466d24cfc5010deee5 X-Mailer: tip-git-log-daemon MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Disposition: inline X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.0 (hera.kernel.org [127.0.0.1]); Mon, 11 May 2009 10:12:43 +0000 (UTC) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Commit-ID: a08b159fc243dbfe415250466d24cfc5010deee5 Gitweb: http://git.kernel.org/tip/a08b159fc243dbfe415250466d24cfc5010deee5 Author: Paul Mackerras AuthorDate: Mon, 11 May 2009 15:46:10 +1000 Committer: Ingo Molnar CommitDate: Mon, 11 May 2009 12:10:53 +0200 perf_counter: don't count scheduler ticks as context switches The context-switch software counter gives inflated values at present because each scheduler tick and each process-wide counter enable/disable prctl gets counted as a context switch. This happens because perf_counter_task_tick, perf_counter_task_disable and perf_counter_task_enable all call perf_counter_task_sched_out, which calls perf_swcounter_event to record a context switch event. This fixes it by introducing a variant of perf_counter_task_sched_out with two underscores in front for internal use within the perf_counter code, and makes perf_counter_task_{tick,disable,enable} call it. This variant doesn't record a context switch event, and takes a struct perf_counter_context *. This adds the new variant rather than changing the behaviour or interface of perf_counter_task_sched_out because that is called from other code. [ Impact: fix inflated context-switch event counts ] Signed-off-by: Paul Mackerras Cc: Peter Zijlstra Cc: Corey Ashford LKML-Reference: <18951.48034.485580.498953@drongo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar --- kernel/perf_counter.c | 16 +++++++++++----- 1 files changed, 11 insertions(+), 5 deletions(-) diff --git a/kernel/perf_counter.c b/kernel/perf_counter.c index a5bdc93..7373b96 100644 --- a/kernel/perf_counter.c +++ b/kernel/perf_counter.c @@ -837,6 +837,14 @@ void perf_counter_task_sched_out(struct task_struct *task, int cpu) cpuctx->task_ctx = NULL; } +static void __perf_counter_task_sched_out(struct perf_counter_context *ctx) +{ + struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); + + __perf_counter_sched_out(ctx, cpuctx); + cpuctx->task_ctx = NULL; +} + static void perf_counter_cpu_sched_out(struct perf_cpu_context *cpuctx) { __perf_counter_sched_out(&cpuctx->ctx, cpuctx); @@ -943,15 +951,13 @@ int perf_counter_task_disable(void) struct perf_counter *counter; unsigned long flags; u64 perf_flags; - int cpu; if (likely(!ctx->nr_counters)) return 0; local_irq_save(flags); - cpu = smp_processor_id(); - perf_counter_task_sched_out(curr, cpu); + __perf_counter_task_sched_out(ctx); spin_lock(&ctx->lock); @@ -989,7 +995,7 @@ int perf_counter_task_enable(void) local_irq_save(flags); cpu = smp_processor_id(); - perf_counter_task_sched_out(curr, cpu); + __perf_counter_task_sched_out(ctx); spin_lock(&ctx->lock); @@ -1054,7 +1060,7 @@ void perf_counter_task_tick(struct task_struct *curr, int cpu) ctx = &curr->perf_counter_ctx; perf_counter_cpu_sched_out(cpuctx); - perf_counter_task_sched_out(curr, cpu); + __perf_counter_task_sched_out(ctx); rotate_ctx(&cpuctx->ctx); rotate_ctx(ctx);