Re: [Xenomai-help] Kernel crash during queue create/destroy

[Jan Kiszka <jan.kiszka@domain.hid>]

[-- Attachment #1: Type: text/plain, Size: 2164 bytes --]

Philippe Gerum wrote:
> On Wed, 2006-11-15 at 23:36 +0100, Jan Kiszka wrote:
>> Philippe Gerum wrote:
>>> On Wed, 2006-11-15 at 22:01 +0100, Jan Kiszka wrote:
>>>
>>>> /me unfortunately failed to reproduce your problem here. Instead, I
>>>> found a regression in SVN head - different story for a different thread.
>>>>
>>> It's reproducible here. This bug triggers if you leave enough time
>>> between queue creation and deletion for Linux to deal with its usual
>>> business, like running workqueues... Additionally, this bug would not
>>> trigger with different queue names passed to the creation routines. It
>>> seems to be caused by out-of-sequence create/delete requests of /proc
>>> entries relayed to the proc_fs subsystem, which does not perform any
>>> sanity checks on the data it is submitted.
>>>
>> Yes, that delay makes the difference. I just added a huge one (100
>> ticks), and now I get tones of this:
>>
>>> xenomai 2.2.4 timer-test
>>> task shadow:0
>>> timer set mode:0
>>> task sleep:0
>>> testing XENOMAI q-functions
>>> Bad page state in process 'maintask'
>>> page:c10e8b20 flags:0x80000400 mapping:00000000 mapcount:0 count:0
>>> Trying to fix it up, but a reboot is needed
>>> Backtrace:
>>>  <c0103055> show_trace+0x12/0x14  <c01035e0> dump_stack+0x1c/0x1e
>>>  <c0157104> bad_page+0x49/0x73  <c01577d8> free_hot_cold_page+0x68/0x13d
>>>  <c01578fb> free_hot_page+0xf/0x11  <c0157a17> __free_pages+0x2e/0x39
>>>  <c01646f2> __vunmap+0x90/0xbc  <c01647c4> vfree+0x2e/0x30
>>>  <c01388ca> xnheap_destroy_mapped+0xe2/0x11b  <c8887d93> rt_queue_delete+0xb5/0xf2 [xeno_native]
>>>  <c88836d9> __rt_queue_delete+0x4e/0x72 [xeno_native]  <c0141b20> losyscall_event+0xa3/0x145
>>>  <c013594f> __ipipe_dispatch_event+0xac/0x16c  <c0109fd2> __ipipe_syscall_root+0x78/0x100
>>>  <c02a5520> system_call+0x20/0x38 
>> Is this also what you see? Looks quite different on first sight. [Note:
>> it's on a qemu box.]
>>
> 
> Uh, no. Try removing the second part of your heap patch, regarding the
> memory unreservation routine.
> 

Ouch, that part of my patch was nonsense, indeed. Fine again now.

Jan


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 250 bytes --]