From mboxrd@z Thu Jan 1 00:00:00 1970 Message-ID: <499B5136.2030107@domain.hid> Date: Wed, 18 Feb 2009 01:07:18 +0100 From: Gilles Chanteperdrix MIME-Version: 1.0 References: <499B36C5.3060104@domain.hid> <499B3F35.2080604@domain.hid> In-Reply-To: <499B3F35.2080604@domain.hid> Content-Type: text/plain; charset=ISO-8859-2 Content-Transfer-Encoding: 7bit Subject: Re: [Xenomai-help] BUG fs/buffer.c with Linux 2.6.26,27 List-Id: Help regarding installation and common use of Xenomai List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Jan Kiszka Cc: xenomai@xenomai.org Jan Kiszka wrote: > Roman Pisl wrote: >> Hello, >> I'm porting our application to Xenomai and I found a bug with Linux >> 2.6.26 and 2.6.27 (and also .19) and Xenomai 2.4.6.1. This happens when >> the application is running and there is hard disk activity. It happens >> also when latency test is running for a while. Linux 2.6.16 with Xenomai >> 2.4.6.1 works without problem. So this is solution for now. >> >> Tested on x86 - thinkpad laptop and pentium 4 desktop. >> >> I'm not a kernel hacker but surely try to help with fixing. I'd like to >> see this fixed. > > /me, too. But could you first of all check if the problem persists even > with latest adeos-ipipe-2.6.27.13-x86-2.2-05.patch [1]? > > If it does: Go back to 2.6.26 (in order to have the ipipe tracer). > Before the oops can occur, extend /proc/ipipe/trace/back_trace_points > to, say, one thousand ("echo 1000 > /proc/..."). Then post the resulting > trace. It may include some hints about when Linux IRQs were last enabled > and when they became stuck at disabled. Also, please attach your .config. Other traces show that it happens in __ipipe_unstall_iret_root. -- Gilles.