From patchwork Tue Aug 16 10:45:06 2016 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Punit Agrawal X-Patchwork-Id: 9283303 Return-Path: Received: from mail.wl.linuxfoundation.org (pdx-wl-mail.web.codeaurora.org [172.30.200.125]) by pdx-korg-patchwork.web.codeaurora.org (Postfix) with ESMTP id E3E35607FD for ; Tue, 16 Aug 2016 10:46:38 +0000 (UTC) Received: from mail.wl.linuxfoundation.org (localhost [127.0.0.1]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id D38DA285DD for ; Tue, 16 Aug 2016 10:46:38 +0000 (UTC) Received: by mail.wl.linuxfoundation.org (Postfix, from userid 486) id C7B13285DE; Tue, 16 Aug 2016 10:46:38 +0000 (UTC) X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on pdx-wl-mail.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-6.9 required=2.0 tests=BAYES_00,RCVD_IN_DNSWL_HI autolearn=unavailable version=3.3.1 Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id EE80A285DD for ; Tue, 16 Aug 2016 10:46:36 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753467AbcHPKqY (ORCPT ); Tue, 16 Aug 2016 06:46:24 -0400 Received: from fw-tnat.cambridge.arm.com ([217.140.96.140]:21479 "EHLO cam-smtp0.cambridge.arm.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1751861AbcHPKqW (ORCPT ); Tue, 16 Aug 2016 06:46:22 -0400 Received: from e105922-lin.cambridge.arm.com (e105922-lin.cambridge.arm.com [10.1.194.52]) by cam-smtp0.cambridge.arm.com (8.13.8/8.13.8) with SMTP id u7GAjoHf031891; Tue, 16 Aug 2016 11:45:50 +0100 Received: by e105922-lin.cambridge.arm.com (sSMTP sendmail emulation); Tue, 16 Aug 2016 11:45:50 +0100 From: Punit Agrawal To: linux-kernel@vger.kernel.org, kvm@vger.kernel.org, kvmarm@lists.cs.columbia.edu, linux-arm-kernel@lists.infradead.org Cc: Punit Agrawal , Christoffer Dall , Marc Zyngier , Steven Rostedt , Ingo Molnar , Will Deacon Subject: [RFC PATCH 1/7] perf/trace: Add notification for perf trace events Date: Tue, 16 Aug 2016 11:45:06 +0100 Message-Id: <1471344312-26685-2-git-send-email-punit.agrawal@arm.com> X-Mailer: git-send-email 2.8.1 In-Reply-To: <1471344312-26685-1-git-send-email-punit.agrawal@arm.com> References: <1471344312-26685-1-git-send-email-punit.agrawal@arm.com> Sender: kvm-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: kvm@vger.kernel.org X-Virus-Scanned: ClamAV using ClamSMTP Add a mechanism to notify listeners about perf trace event state changes. This enables listeners to take actions requiring the event context (e.g., attached process). The notification mechanism can be used to reduce trace point based profiling overhead by enabling/disabling hardware traps for specific contexts (e.g., virtual machines). Signed-off-by: Punit Agrawal Cc: Steven Rostedt Cc: Ingo Molnar --- include/linux/trace_events.h | 3 +++ kernel/trace/trace_event_perf.c | 24 ++++++++++++++++++++++++ 2 files changed, 27 insertions(+) diff --git a/include/linux/trace_events.h b/include/linux/trace_events.h index be00761..5924032 100644 --- a/include/linux/trace_events.h +++ b/include/linux/trace_events.h @@ -505,6 +505,9 @@ perf_trace_buf_submit(void *raw_data, int size, int rctx, u16 type, { perf_tp_event(type, count, raw_data, size, regs, head, rctx, task); } + +extern int perf_trace_notifier_register(struct notifier_block *nb); +extern int perf_trace_notifier_unregister(struct notifier_block *nb); #endif #endif /* _LINUX_TRACE_EVENT_H */ diff --git a/kernel/trace/trace_event_perf.c b/kernel/trace/trace_event_perf.c index 562fa69..9aaaacf 100644 --- a/kernel/trace/trace_event_perf.c +++ b/kernel/trace/trace_event_perf.c @@ -6,10 +6,12 @@ */ #include +#include #include #include "trace.h" static char __percpu *perf_trace_buf[PERF_NR_CONTEXTS]; +static RAW_NOTIFIER_HEAD(perf_trace_notifier_list); /* * Force it to be aligned to unsigned long to avoid misaligned accesses @@ -86,6 +88,26 @@ static int perf_trace_event_perm(struct trace_event_call *tp_event, return 0; } +int perf_trace_notifier_register(struct notifier_block *nb) +{ + return raw_notifier_chain_register(&perf_trace_notifier_list, nb); +} + +int perf_trace_notifier_unregister(struct notifier_block *nb) +{ + return raw_notifier_chain_unregister(&perf_trace_notifier_list, nb); +} + +static void perf_trace_notify(enum trace_reg event, struct perf_event *p_event) +{ + /* + * We use raw notifiers here as we are called with the + * event_mutex held. + */ + raw_notifier_call_chain(&perf_trace_notifier_list, + event, p_event); +} + static int perf_trace_event_reg(struct trace_event_call *tp_event, struct perf_event *p_event) { @@ -176,6 +198,7 @@ out: static int perf_trace_event_open(struct perf_event *p_event) { struct trace_event_call *tp_event = p_event->tp_event; + perf_trace_notify(TRACE_REG_PERF_OPEN, p_event); return tp_event->class->reg(tp_event, TRACE_REG_PERF_OPEN, p_event); } @@ -183,6 +206,7 @@ static void perf_trace_event_close(struct perf_event *p_event) { struct trace_event_call *tp_event = p_event->tp_event; tp_event->class->reg(tp_event, TRACE_REG_PERF_CLOSE, p_event); + perf_trace_notify(TRACE_REG_PERF_CLOSE, p_event); } static int perf_trace_event_init(struct trace_event_call *tp_event,