From mboxrd@z Thu Jan  1 00:00:00 1970
Message-ID: <4ACB696F.3080907@domain.hid>
Date: Tue, 06 Oct 2009 17:59:43 +0200
From: Jan Kiszka <jan.kiszka@domain.hid>
MIME-Version: 1.0
References: <4AC5E895.8010107@domain.hid>
	<4AC5F49E.3070608@domain.hid>	<4AC5FEAD.7000608@domain.hid>
	<4AC77576.9070700@domain.hid>	<4AC9E655.5090006@domain.hid>
	<4AC9ED6B.4050004@domain.hid>	<4AC9F4FA.5070409@domain.hid>
	<4ACA013F.8040602@domain.hid> <4ACB4389.10109@domain.hid>
	<4ACB4C70.80406@domain.hid> <4ACB6626.6020707@domain.hid>
In-Reply-To: <4ACB6626.6020707@domain.hid>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Subject: Re: [Xenomai-help] Freeze on SMP x86-64
List-Id: Help regarding installation and common use of Xenomai
	<xenomai.xenomai.org>
List-Unsubscribe: <https://mail.gna.org/listinfo/xenomai-help>,
	<mailto:xenomai-help-request@domain.hid>
List-Archive: </public/xenomai-help>
List-Post: <mailto:xenomai@xenomai.org>
List-Help: <mailto:xenomai-help-request@domain.hid>
List-Subscribe: <https://mail.gna.org/listinfo/xenomai-help>,
	<mailto:xenomai-help-request@domain.hid>
To: Roman Pisl <rpisl@domain.hid>
Cc: "xenomai@xenomai.org" <xenomai@xenomai.org>

Roman Pisl wrote:
> On 6.10.2009 15:56, Jan Kiszka wrote:
>> Roman Pisl wrote:
>>> ...
>>>> Is there a chance to grab the panic log via serial line? The first of
>>>> your two screenshots indicate that there might have happened something
>>>> interesting before the NMI triggered, but it's scrolled out of sight.
>>>>
>>>> Jan
>>>>
>>> Hello,
>>> here I send kernel panic (triggered by rt_task_sleep and latency test)
>>> grabbed via serial line.
>> Hmm, still only NMI alarms...
>>
>>> I'm ready to help with resolving this issue.
>> Could you try enabling the I-pipe tracer with function tracing? It
>> should dump a trace of the functions that were executed before the
>> watchdog triggered. You may need to extend the number of tracepoints a
>> bit, try "echo 1000>  /proc/ipipe/trace/back_trace_points".
>>
>> Jan
>>
> 
> Then I had I/O activity on the background and got this.
> 

Try without CONFIG_IPIPE_DEBUG_INTERNAL (it triggers when we printk from
NMI context). Please also raise the NMI threshold
(CONFIG_XENO_HW_NMI_DEBUG_LATENCY_MAX) to 1 ms or so. It may otherwise
trigger too early if there are hardware related delays + the tracing
overhead.

Jan

-- 
Siemens AG, Corporate Technology, CT SE 2
Corporate Competence Center Embedded Linux