Date: Fri, 30 Aug 2013 13:49:03 +0200
From: Jiri Olsa <jolsa@redhat.com>
To: Jean Pihet <jean.pihet@linaro.org>
Cc: David Ahern <dsahern@gmail.com>, linux-kernel@vger.kernel.org
Subject: Re: Fwd: perf: callchain feature on ARM - question about libunwind
 API
Message-ID: <20130830114903.GC1023@krava.brq.redhat.com>
References: <CAFrcx1n5eA4zz+nUhraE_fi_Usc=PCZTC=RCT=S285K4=jqHSg@mail.gmail.com>
 <CAFrcx1nr4-qyzC=zFRdnMavHy5p_r9Ng+zSVwNPk6snYtBm6gw@mail.gmail.com>
 <521CCFD8.90201@gmail.com>
 <20130829135216.GA1032@krava.brq.redhat.com>
 <CAFrcx1=U8SapYd=Hw8YaAnKNggJMuY4r03zYeevc0Ttq5hFr7Q@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <CAFrcx1=U8SapYd=Hw8YaAnKNggJMuY4r03zYeevc0Ttq5hFr7Q@mail.gmail.com>
User-Agent: Mutt/1.5.21 (2010-09-15)
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1087
Lines: 30

On Thu, Aug 29, 2013 at 07:15:00PM +0200, Jean Pihet wrote:

SNIP

> 
> The problem is that it is slow as hell (couple of minutes to report
> the callchain from a simple test app ;-|). I will look at the sampling
> rate and the caching next.

FYI there's elfutils alternative grewing now via libdw providing
API to do the 'remote unwind' as in libunwind. I'm not completely
sure, but I think it should support both .eh_frames and .debug_frames.

check:
git://git.fedorahosted.org/git/elfutils.git branch jankratochvil/unwindx86

the 'tests/backtrace-data.c' should be an example of what we want in perf

somewhere I should have perf changes to support this in
early version which no longer relevant.. but I recall it
was substantially faster due to some libdw inner caching

it's on my long term TODO list, so any help is welcome! ;-)

jirka
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/