Re: [Xenomai-help] Testsuite fails on 2.5-rc3

[Philippe Gerum <rpm@xenomai.org>]

On Thu, 2009-09-10 at 14:46 -0600, Jonathan Haws wrote:
> > > > > Thu Jan  1 00:06:52 UTC 1970
> > > > > running: ./run -- -T 120 # switchtest
> > > > > *
> > > > > *
> > > > > * Type ^C to stop this application.
> > > > > *
> > > > > *
> > > > > == Testing FPU check routines...
> > > > > r0: 1 != 2
> > > > > r1: 1 != 2
> > > > > r2: 1 != 2
> > > > > r3: 1 != 2
> > > > > r4: 1 != 2
> > > > > r5: 1 != 2
> > > > > r6: 1 != 2
> > > > > r7: 1 != 2
> > > > > r8: 1 != 2
> > > > > r9: 1 != 2
> > > > > r10: 1 != 2
> > > > > r11: 1 != 2
> > > > > r12: 1 != 2
> > > > > r13: 1 != 2
> > > > > r14: 1 != 2
> > > > > r15: 1 != 2
> > > > > r16: 1 != 2
> > > > > r17: 1 != 2
> > > > > r18: 1 != 2
> > > > > r19: 1 != 2
> > > > > r20: 1 != 2
> > > > > r21: 1 != 2
> > > > > r22: 1 != 2
> > > > > r23: 1 != 2
> > > > > r24: 1 != 2
> > > > > r25: 1 != 2
> > > > > r26: 1 != 2
> > > > > r27: 1 != 2
> > > > > r28: 1 != 2
> > > > > r29: 1 != 2
> > > > > r30: 1 != 2
> > > > > r31: 1 != 2
> > > > > == FPU check routines: OK.
> > > > > == Threads: sleeper_ufps-0 rtk-1Oops: Exception in kernel
> > mode,
> > > > sig: 11 [#1]
> > > > > PREEMPT Kilauea
> > > > > Modules linked in:
> > > > > NIP: c01e876c LR: c01e86e8 CTR: c0070fb0
> > > > > REGS: cf1e5440 TRAP: 0700   Not tainted  (2.6.30.3-wolverine-
> > > > dirty)
> > > > > MSR: 00029030 <EE,ME,CE,IR,DR>  CR: 40000002  XER: 0000005f
> > > > > TASK = cd412000[819] 'switchtest' THREAD: cd448000
> > > > > GPR00: 00000005 cf1e54f0 cd412000 00000000 00000005 00000001
> > > > a4500f69 00000000
> > > > > GPR08: 000011af 00000000 00000017 c0390000 20000002 00000000
> > > > 00021030 c0070d8c
> > > > > GPR16: cd471208 00000000 00000000 00000000 c0394108 c0381248
> > > > 003d0900 00000000
> > > > > GPR24: c02f5580 cf1e5500 cd4711d0 cd4560b8 00000006 cd471208
> > > > 00000000 00000000
> > > > > NIP [c01e876c] rtswitch_ktask+0xb0/0x3e4
> > > > > LR [c01e86e8] rtswitch_ktask+0x2c/0x3e4
> > > > > Call Trace:
> > > > > Instruction dump:
> > > > > 7c0003b8 7f9c0038 7f892278 21290000 7d3c0194 7d3c4b78 4182009c
> > > > 1c1703e8
> > > > > 39200000 7c002214 9001000c 91210008 <c8010008> fc200090
> > fc400090
> > > > fc600090
> > > > > Kernel panic - not syncing: Fatal exception in interrupt
> > > > > Call Trace:
> > > > > Rebooting in 180 seconds..
> > > >
> > > > One thing I can tell for sure reading this trace is that the
> > error
> > > > is
> > > > not caused by the FPU. The error is a "program check exception".
> > >
> > > So, what could be causing the program check in switchtest?  Could
> > it be a misconfiguration of Xenomai in my kernel, or the fact that I
> > had re-patched my kernel with a newer I-pipe patch and then I tried
> > to re-patch again with an older patch?
> > >
> > 
> > This one is an Heisenbug from switchtest. It was known to happen
> > with
> > FPU-disabled threads as well. I thought I had pinned it a month ago,
> > but
> > I was wrong.
> 
> Just FYI, I got xeno-test to run correctly under 2.5-rc3 (once I had /dev/rtheap and /dev/rtpXX in place) and got this exact same error with switchtest.  So, it happens in both the stable release and the development branch.
> 

2.4.9.1 and 2.5-rc3 are mostly in sync wrt to switchtest.

> Is this just an artifact of switchtest, or is it something that will come up in my own code?

Switchtest definitely has a bug, but the fact that it tests the FPU is
not relevant to this issue (IOW, the test engine has a problem, not what
is being tested).

> 
> Thanks,
> 
> Jonathan
-- 
Philippe.