From mboxrd@z Thu Jan  1 00:00:00 1970
Message-ID: <5108FABD.8090401@sirius-es.it>
Date: Wed, 30 Jan 2013 11:49:33 +0100
From: Matteo Facchinetti <matteo.facchinetti@sirius-es.it>
MIME-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Subject: [Xenomai] FPU problems in Xenomai 2.4.x
List-Id: Discussions about the Xenomai project <xenomai.xenomai.org>
List-Unsubscribe: <http://www.xenomai.org/mailman/options/xenomai>,
	<mailto:xenomai-request@xenomai.org?subject=unsubscribe>
List-Archive: <http://www.xenomai.org/pipermail/xenomai>
List-Post: <mailto:xenomai@xenomai.org>
List-Help: <mailto:xenomai-request@xenomai.org?subject=help>
List-Subscribe: <http://www.xenomai.org/mailman/listinfo/xenomai>,
	<mailto:xenomai-request@xenomai.org?subject=subscribe>
To: xenomai@xenomai.org

Hi,

in a PPC mpc5125 system we're using xenomai 2.4.x (from git repository) 
with kernel 2.6.29
and have applied 
ksrc_arch_powerpc_patches_adeos-ipipe-2.6.29.4-powerpc-DENX-2.6-02.patch

We have encountered a problem in a userspace application that makes 
heavy use of the FPU.
In specific, our application consists about two concurrent processes:
- the first use FPU in a xenomai_task
- the second use FPU in a normal linux thread

After a random time of test, Kernel crash with the following oops:

Unrecoverable FP Unavailable Exception 801 at c0003888
Oops: Unrecoverable FP Unavailable Exception, sig: 6 [#1]
PREEMPT MPC5125 ADS
Modules linked in: xeno_can_mscan xeno_can mchpar1xxx ipv6
NIP: c0003888 LR: c0006c14 CTR: 1008879c
REGS: ced71e40 TRAP: 0801   Not tainted  (2.6.29.1)
MSR: 00009032 <EE,ME,IR,DR>  CR: 84000444  XER: 00000000
TASK = cec164e0[4069] 'gui_powerpc.elf' THREAD: ced70000
GPR00: c0006c04 ced71ef0 cec164e0 ced71efc ced71ef8 cec166e8 102b24fe 
000000cc
GPR08: 102b2500 c0520000 00000000 00000000 0ec166e8
NIP [c0003888] cvt_fd+0x0/0x1c
LR [c0006c14] fix_alignment+0x6cc/0x84c
Call Trace:
[ced71ef0] [c0006c04] fix_alignment+0x6bc/0x84c (unreliable)
[ced71f30] [c0010cec] alignment_exception+0x8c/0x128
[ced71f40] [c0013820] ret_from_except_full+0x0/0x4c
--- Exception: 600 at 0x100887ac
     LR = 0x100887a4
Instruction dump:
dbe30110 fc00048e d8030118 41820014 80850084 38602900 7c841878 90850084
38a00000 3c80c052 90a46044 4e800020 <c8050118> fdfe058e c0030000 d8040000
---[ end trace 0cdaae6afca79e76 ]---
note: gui_powerpc.elf[4069] exited with preempt_count 1

In relation about mail:
http://www.xenomai.org/pipermail/xenomai-core/2010-10/msg00007.html
We commented out local_irq_save_hw_cond(flags).
Kernel doesn't crash but application die with fpu register corrupted.

Does anyone know if xenomai 2.4.x with 2.6.29 kernel series have 
problems with heavy FPU usage?

Matteo Facchinetti
Sirius Electronic Systems s.r.l.