From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <linux-kernel-owner@vger.kernel.org>
Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand
	id S1753201Ab3LMSHA (ORCPT <rfc822;w@1wt.eu>);
	Fri, 13 Dec 2013 13:07:00 -0500
Received: from merlin.infradead.org ([205.233.59.134]:49742 "EHLO
	merlin.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
	with ESMTP id S1752504Ab3LMSG7 (ORCPT
	<rfc822;linux-kernel@vger.kernel.org>);
	Fri, 13 Dec 2013 13:06:59 -0500
Date: Fri, 13 Dec 2013 19:06:38 +0100
From: Peter Zijlstra <peterz@infradead.org>
To: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>,
        Ingo Molnar <mingo@redhat.com>, linux-kernel@vger.kernel.org,
        David Ahern <dsahern@gmail.com>,
        Frederic Weisbecker <fweisbec@gmail.com>, Jiri Olsa <jolsa@redhat.com>,
        Mike Galbraith <efault@gmx.de>, Namhyung Kim <namhyung@gmail.com>,
        Paul Mackerras <paulus@samba.org>,
        Stephane Eranian <eranian@google.com>, Andi Kleen <ak@linux.intel.com>
Subject: Re: [PATCH v0 01/71] perf: Disable all pmus on unthrottling and
 rescheduling
Message-ID: <20131213180638.GR21999@twins.programming.kicks-ass.net>
References: <1386765443-26966-1-git-send-email-alexander.shishkin@linux.intel.com>
 <1386765443-26966-2-git-send-email-alexander.shishkin@linux.intel.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <1386765443-26966-2-git-send-email-alexander.shishkin@linux.intel.com>
User-Agent: Mutt/1.5.21 (2012-12-30)
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

On Wed, Dec 11, 2013 at 02:36:13PM +0200, Alexander Shishkin wrote:
> Currently, only one pmu in a context gets disabled during unthrottling
> and event_sched_{out,in}, however, events in one context may belong to
> different pmus, which results in pmus being reprogrammed while they are
> still enabled. This patch temporarily disables pmus that correspond to
> each event in the context while these events are being modified.
> 
> Signed-off-by: Alexander Shishkin <alexander.shishkin@linux.intel.com>
> ---
>  kernel/events/core.c | 27 ++++++++++++++++++++++++---
>  1 file changed, 24 insertions(+), 3 deletions(-)
> 
> diff --git a/kernel/events/core.c b/kernel/events/core.c
> index 403b781..d656cd6 100644
> --- a/kernel/events/core.c
> +++ b/kernel/events/core.c
> @@ -1396,6 +1396,9 @@ event_sched_out(struct perf_event *event,
>  	if (event->state != PERF_EVENT_STATE_ACTIVE)
>  		return;
>  
> +	if (event->pmu != ctx->pmu)
> +		perf_pmu_disable(event->pmu);
> +
>  	event->state = PERF_EVENT_STATE_INACTIVE;
>  	if (event->pending_disable) {
>  		event->pending_disable = 0;
> @@ -1412,6 +1415,9 @@ event_sched_out(struct perf_event *event,
>  		ctx->nr_freq--;
>  	if (event->attr.exclusive || !cpuctx->active_oncpu)
>  		cpuctx->exclusive = 0;
> +
> +	if (event->pmu != ctx->pmu)
> +		perf_pmu_enable(event->pmu);
>  }
>  
>  static void

Hmm, indeed. Does it make sense to drop the conditional?
perf_pmu_{en,dis}able() is recursive and the thinking is that if its the
same PMU the cacheline is hot because we touched it already recently
anyway, so the unconditional inc/dec might actually be faster.. dunno.