From mboxrd@z Thu Jan  1 00:00:00 1970
From: Andi Kleen <ak@linux.intel.com>
Subject: Re: PEBS level 2/3 breaks dwarf unwinding! [WAS: Re: Broken dwarf
 unwinding - wrong stack pointer register value?]
Date: Tue, 6 Nov 2018 09:26:07 -0800
Message-ID: <20181106172607.GT6218@tassilo.jf.intel.com>
References: <2335309.gnWok9HYb4@agathebauer>
 <3227038.olIWmsCzzY@agathebauer>
 <20181105205119.GC25674@krava>
 <3799078.YBnU1OB0PF@agathebauer>
 <20181106001037.GQ6218@tassilo.jf.intel.com>
 <20181106083957.GB8081@krava>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Return-path: <linux-kernel-owner@vger.kernel.org>
Content-Disposition: inline
In-Reply-To: <20181106083957.GB8081@krava>
Sender: linux-kernel-owner@vger.kernel.org
To: Jiri Olsa <jolsa@redhat.com>
Cc: Milian Wolff <milian.wolff@kdab.com>, linux-kernel@vger.kernel.org, Jiri Olsa <jolsa@kernel.org>, namhyung@kernel.org, linux-perf-users@vger.kernel.org, Arnaldo Carvalho <acme@kernel.org>
List-Id: linux-perf-users.vger.kernel.org

> hum, is this about having 'large pebs' or there's this window
> if there's also only single pebs record allowed? which should
> be case for dwarf unwind

With large PEBS today there is never any stack unwind because
stack unwinding can be only done from a PMI.

The window happens even with single PEBS. It's related
to the CPU being pipelined, so complex events may not
happen fully atomically.

-Andi