Re: [PATCH] perf top: Fix a race in callchain handling

From: Namhyung Kim <namhyung@gmail.com>
To: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Cc: Namhyung Kim <namhyung.kim@lge.com>,
	Peter Zijlstra <a.p.zijlstra@chello.nl>,
	Paul Mackerras <paulus@samba.org>, Ingo Molnar <mingo@redhat.com>,
	LKML <linux-kernel@vger.kernel.org>,
	Frederic Weisbecker <fweisbec@gmail.com>
Subject: Re: [PATCH] perf top: Fix a race in callchain handling
Date: Sat, 12 May 2012 01:25:48 +0900	[thread overview]
Message-ID: <87boluwupf.fsf@gmail.com> (raw)
In-Reply-To: <20120511150111.GA2485@infradead.org> (Arnaldo Carvalho de Melo's message of "Fri, 11 May 2012 12:01:11 -0300")

Hi,

On Fri, 11 May 2012 12:01:11 -0300, Arnaldo Carvalho de Melo wrote:
> Em Fri, May 11, 2012 at 11:10:59AM +0900, Namhyung Kim escreveu:
>> On Mon, 07 May 2012 10:32:22 +0900, Namhyung Kim wrote:
>> > I thought about it before, but it still cannot protect it from accessing
>> > a hists by multiple concurrent threads. IOW if two threads call the
>> > function to a same hists at the same time, ->callchain_collapse_cursor
>> > would still get the race problem - so crashed.
>> >
>> > I guess callchain_cursor should be thread-local, eventually. No need to
>> > make it hist-local IMHO.
>> 
>> So, any thoughts?
>
> Can you try writing a patch using TLS?
>

Well, I can do it, but I still think the current patch is just enough
for now though. My concern is the TLS might make perf less portable
especially for embedded systems which are not using glibc.

Anyway I'll give it a spin as you ask.

Thanks,
Namhyung