Hello Marcelo, hello world,
I'm having massive problems with Linux 2.4.20 and modul loading. In fact,
it seems to have something to do with devfs. Everytime i try to load a
module which is supposed to create something new in /dev, i get an Oops.
I noticed that behaviour with vmware-module and now also with ALSA
0.9.0rc6. Is there a fix available for this yet?
Call trace (ksymoops):
EIP: 0010:[<c01185d5>] Not tainted
Using defaults from ksymoops -t elf32-i386 -a i386
EFLAGS: 00010046
eax: ddb67b34 ebx: 00000000 ecx: d3a13ef8 edx: d3a13ef0
esi: 00000246 edi: 00000000 ebp: ddb67a00 esp: d3a13eb8
ds: 0018 es: 0018 ss: 0018
Process xmms (pid: 4406, stackpage=d3a13000)
Stack: d3a13ef0 ddb67b34 e292e6ea d70acb40 dba8c640 df6fb040 dba8c648 d309eec0
d309eed8 00000000 00000000 00000000 00000003 d309eec0 00000000 d3a12000
d309eec0 d369a940 00000000 00000000 00000000 00000000 c017e500 e292c000
Call Trace: [<e292e6ea>] [<c017e500>] [<c017f19f>] [<c014314e>] [<c01386a3>]
[<c01385c7>] [<c0138963>] [<c010737f>]
Code: 89 4b 04 89 5a 08 89 41 04 89 08 56 9d 8b 1c 24 8b 74 24 04
>>EIP; c01185d5 <add_wait_queue+15/30> <=====
>>eax; ddb67b34 <_end+1d7d60d0/224e35fc>
>>ecx; d3a13ef8 <_end+13682494/224e35fc>
>>edx; d3a13ef0 <_end+1368248c/224e35fc>
>>ebp; ddb67a00 <_end+1d7d5f9c/224e35fc>
>>esp; d3a13eb8 <_end+13682454/224e35fc>
Trace; e292e6ea <[snd-pcm-oss]snd_pcm_oss_open+17a/250>
Trace; c017e500 <devfs_get_ops+60/90>
Trace; c017f19f <devfs_open+13f/1a0>
Trace; c014314e <vfs_permission+7e/140>
Trace; c01386a3 <dentry_open+d3/1d0>
Trace; c01385c7 <filp_open+67/70>
Trace; c0138963 <sys_open+53/a0>
Trace; c010737f <system_call+33/38>
Code; c01185d5 <add_wait_queue+15/30>
00000000 <_EIP>:
Code; c01185d5 <add_wait_queue+15/30> <=====
0: 89 4b 04 mov %ecx,0x4(%ebx) <=====
Code; c01185d8 <add_wait_queue+18/30>
3: 89 5a 08 mov %ebx,0x8(%edx)
Code; c01185db <add_wait_queue+1b/30>
6: 89 41 04 mov %eax,0x4(%ecx)
Code; c01185de <add_wait_queue+1e/30>
9: 89 08 mov %ecx,(%eax)
Code; c01185e0 <add_wait_queue+20/30>
b: 56 push %esi
Code; c01185e1 <add_wait_queue+21/30>
c: 9d popf
Code; c01185e2 <add_wait_queue+22/30>
d: 8b 1c 24 mov (%esp,1),%ebx
Code; c01185e5 <add_wait_queue+25/30>
10: 8b 74 24 04 mov 0x4(%esp,1),%esi
--
.''`. Name: Martin Loschwitz
: :' : E-Mail: [email protected]
`. `'` www: http://www.madkiss.org/
`- Use Debian GNU/Linux - http://www.debian.org
On Sat, 2002-11-30 at 08:49, Martin Loschwitz wrote:
> Hello Marcelo, hello world,
>
> I'm having massive problems with Linux 2.4.20 and modul loading. In fact,
> it seems to have something to do with devfs. Everytime i try to load a
> module which is supposed to create something new in /dev, i get an Oops.
> I noticed that behaviour with vmware-module and now also with ALSA
> 0.9.0rc6. Is there a fix available for this yet?
Linux 2,4.20 doesnt include vmware or ALSA. The fact you list those
modules alone suggests that the problem is that you haven't rebuilt them
for the new kernel
On Sat, 2002-11-30 at 08:49, Martin Loschwitz wrote:
> I'm having massive problems with Linux 2.4.20 and modul loading.
> In fact, it seems to have something to do with devfs.
On Sat Nov 30 2002 - 09:36:57 EST, Alan Cox wrote:
> Linux 2,4.20 doesnt include vmware or ALSA. The fact you list those
> modules alone suggests that the problem is that you haven't rebuilt
> them for the new kernel.
Well maybe that's his problem and maybe it isn't, but he's not the only
person thinking that there's something strange about devfs in 2.4.20. I
have a 2.4.20-SMP box that is deadlocking when loading modules via
devfs. The NMI watchdog has produced two oopsen for me, which I have
published to this list, and while one oops is also from the ALSA
modules, the other is from mga.o. Both oopsen occur in the
__write_lock_failed() function and have devfs_open() in the module
stack. I have most definitely NOT forgotten to recompile everything for
2.4.20, and I am using symbol versioning anyway.
Chris
> Well maybe that's his problem and maybe it isn't, but he's not the only
> person thinking that there's something strange about devfs in 2.4.20. I
> have a 2.4.20-SMP box that is deadlocking when loading modules via
> devfs. The NMI watchdog has produced two oopsen for me, which I have
Could be. I don't test devfs at all