From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752361AbZFWKEg (ORCPT ); Tue, 23 Jun 2009 06:04:36 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1754544AbZFWKEU (ORCPT ); Tue, 23 Jun 2009 06:04:20 -0400 Received: from hera.kernel.org ([140.211.167.34]:33661 "EHLO hera.kernel.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755492AbZFWKES (ORCPT ); Tue, 23 Jun 2009 06:04:18 -0400 Date: Tue, 23 Jun 2009 10:03:47 GMT From: tip-bot for Peter Zijlstra To: linux-tip-commits@vger.kernel.org Cc: acme@redhat.com, paulus@samba.org, linux-kernel@vger.kernel.org, hpa@zytor.com, mingo@redhat.com, a.p.zijlstra@chello.nl, efault@gmx.de, tglx@linutronix.de, mingo@elte.hu Reply-To: mingo@redhat.com, hpa@zytor.com, linux-kernel@vger.kernel.org, paulus@samba.org, acme@redhat.com, a.p.zijlstra@chello.nl, efault@gmx.de, tglx@linutronix.de, mingo@elte.hu In-Reply-To: References: Subject: [tip:perfcounters/urgent] perf_counter: Optimize perf_swcounter_event() Message-ID: Git-Commit-ID: f29ac756a40d0f1bb07d682ea521e7b666ff06d5 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]); Tue, 23 Jun 2009 10:03:48 +0000 (UTC) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Commit-ID: f29ac756a40d0f1bb07d682ea521e7b666ff06d5 Gitweb: http://git.kernel.org/tip/f29ac756a40d0f1bb07d682ea521e7b666ff06d5 Author: Peter Zijlstra AuthorDate: Fri, 19 Jun 2009 18:27:26 +0200 Committer: Ingo Molnar CommitDate: Tue, 23 Jun 2009 11:42:44 +0200 perf_counter: Optimize perf_swcounter_event() Similar to tracepoints, use an enable variable to reduce overhead when unused. Only look for a counter of a particular event type when we know there is at least one in the system. Signed-off-by: Peter Zijlstra LKML-Reference: Cc: Mike Galbraith Cc: Paul Mackerras Cc: Arnaldo Carvalho de Melo Signed-off-by: Ingo Molnar --- include/linux/perf_counter.h | 11 ++++++++++- kernel/perf_counter.c | 18 +++++++++++++++--- 2 files changed, 25 insertions(+), 4 deletions(-) diff --git a/include/linux/perf_counter.h b/include/linux/perf_counter.h index 89698d8..e7213e4 100644 --- a/include/linux/perf_counter.h +++ b/include/linux/perf_counter.h @@ -669,7 +669,16 @@ static inline int is_software_counter(struct perf_counter *counter) (counter->attr.type != PERF_TYPE_HW_CACHE); } -extern void perf_swcounter_event(u32, u64, int, struct pt_regs *, u64); +extern atomic_t perf_swcounter_enabled[PERF_COUNT_SW_MAX]; + +extern void __perf_swcounter_event(u32, u64, int, struct pt_regs *, u64); + +static inline void +perf_swcounter_event(u32 event, u64 nr, int nmi, struct pt_regs *regs, u64 addr) +{ + if (atomic_read(&perf_swcounter_enabled[event])) + __perf_swcounter_event(event, nr, nmi, regs, addr); +} extern void __perf_counter_mmap(struct vm_area_struct *vma); diff --git a/kernel/perf_counter.c b/kernel/perf_counter.c index 1a933a2..7515c76 100644 --- a/kernel/perf_counter.c +++ b/kernel/perf_counter.c @@ -3317,8 +3317,8 @@ out: put_cpu_var(perf_cpu_context); } -void -perf_swcounter_event(u32 event, u64 nr, int nmi, struct pt_regs *regs, u64 addr) +void __perf_swcounter_event(u32 event, u64 nr, int nmi, + struct pt_regs *regs, u64 addr) { struct perf_sample_data data = { .regs = regs, @@ -3509,9 +3509,19 @@ static const struct pmu *tp_perf_counter_init(struct perf_counter *counter) } #endif +atomic_t perf_swcounter_enabled[PERF_COUNT_SW_MAX]; + +static void sw_perf_counter_destroy(struct perf_counter *counter) +{ + u64 event = counter->attr.config; + + atomic_dec(&perf_swcounter_enabled[event]); +} + static const struct pmu *sw_perf_counter_init(struct perf_counter *counter) { const struct pmu *pmu = NULL; + u64 event = counter->attr.config; /* * Software counters (currently) can't in general distinguish @@ -3520,7 +3530,7 @@ static const struct pmu *sw_perf_counter_init(struct perf_counter *counter) * to be kernel events, and page faults are never hypervisor * events. */ - switch (counter->attr.config) { + switch (event) { case PERF_COUNT_SW_CPU_CLOCK: pmu = &perf_ops_cpu_clock; @@ -3541,6 +3551,8 @@ static const struct pmu *sw_perf_counter_init(struct perf_counter *counter) case PERF_COUNT_SW_PAGE_FAULTS_MAJ: case PERF_COUNT_SW_CONTEXT_SWITCHES: case PERF_COUNT_SW_CPU_MIGRATIONS: + atomic_inc(&perf_swcounter_enabled[event]); + counter->destroy = sw_perf_counter_destroy; pmu = &perf_ops_generic; break; }