Oops while accessing /proc/stat (2.4.19)

Oops while accessing /proc/stat (2.4.19)

[Bjoern Krombholz]

Hello,

i'm currently have a problem that every program that tries to read from
/proc/stat like `uptime', `free', `cat /proc/stat' etc. segfaults.


strace shows:
--->
open("/proc/stat", O_RDONLY|O_LARGEFILE) = 3
fstat64(3, {st_mode=S_IFREG|0444, st_size=0, ...}) = 0
brk(0x8050000)                          = 0x8050000
read(3,  <unfinished ...>
<---


First oops:
--->
ksymoops 2.4.6 on i686 2.4.19.  Options used
     -V (default)
     -k /proc/ksyms (default)
     -l /proc/modules (default)
     -o /lib/modules/2.4.19/ (default)
     -m /boot/System.map-2.4.19 (specified)

kernel: c01f2ad9
kernel: Oops: 0002
kernel: CPU:    0
kernel: EIP:    0010:[number+1049/1088]    Tainted: PF
kernel: EFLAGS: 00010203
kernel: eax: 00001000   ebx: 00000000   ecx: 003db837   edx: cdb65f6c
kernel: esi: cdb65f6c   edi: c2e370e0   ebp: 003db837   esp: cdb65ee4
kernel: ds: 0018   es: 0018   ss: 0018
kernel: Process gkrellm (pid: 815, stackpage=cdb65000)
kernel: Stack: c0151923 cbbb6000 00001000 c01ff613 cdb65f18 001f2b86 0397e6cc 00000001
kernel:        00000000 cbbb6000 cdb65f6c cdb65f5c c01ff613 003db838 0397e6cc 001f2b86
kernel:        011f4d1f 00000000 00000c00 cbbb6000 00001000 001f2b86 c0224b2c cdb65f6c
kernel: Call Trace:    [proc_sprintf+51/80] [proc_file_read+248/416] [sys_read+150/240]
kernel: Code: c6 00 20 ff 84 24 c4 00 00 00 89 e8 4d 85 c0 7f d8 8b 84 24
Using defaults from ksymoops -t elf32-i386 -a i386


>>edx; cdb65f6c <_end+d8c6294/20597328>
>>esi; cdb65f6c <_end+d8c6294/20597328>
>>edi; c2e370e0 <_end+2b97408/20597328>
>>esp; cdb65ee4 <_end+d8c620c/20597328>

Code;  00000000 Before first symbol
00000000 <_EIP>:
Code;  00000000 Before first symbol
   0:   c6 00 20                  movb   $0x20,(%eax)
Code;  00000003 Before first symbol
   3:   ff 84 24 c4 00 00 00      incl   0xc4(%esp,1)
Code;  0000000a Before first symbol
   a:   89 e8                     mov    %ebp,%eax
Code;  0000000c Before first symbol
   c:   4d                        dec    %ebp
Code;  0000000d Before first symbol
   d:   85 c0                     test   %eax,%eax
Code;  0000000f Before first symbol
   f:   7f d8                     jg     ffffffe9 <_EIP+0xffffffe9> ffffffe9 <END_OF_CODE+1ef57b06/????>
Code;  00000011 Before first symbol
  11:   8b 84 24 00 00 00 00      mov    0x0(%esp,1),%eax
<---


System:
CPU:    AMD Athlon XP
MB:     DFI AD70-SC (Via KT266 + 8233)
Kernel: 2.4.19 (extra modules: alsa-0.9.0rc1, lm_sensors-2.6.4)



Thx,
Bjoern Krombholz

Re: Oops while accessing /proc/stat (2.4.19)

[Adrian Bunk]

On Mon, 26 Aug 2002, Bjoern Krombholz wrote:

> Hello,

Hi Bjoern,

> i'm currently have a problem that every program that tries to read from
> /proc/stat like `uptime', `free', `cat /proc/stat' etc. segfaults.
>...
> kernel: c01f2ad9
> kernel: Oops: 0002
> kernel: CPU:    0
> kernel: EIP:    0010:[number+1049/1088]    Tainted: PF
>...

which binary-only modules (e.g. NVidia) are loaded on your computer? Is
the problem reproducible without them ever loaded since the last reboot?

> Thx,
> Bjoern Krombholz

cu
Adrian

-- 

You only think this is a free country. Like the US the UK spends a lot of
time explaining its a free country because its a police state.
								Alan Cox

Re: Oops while accessing /proc/stat (2.4.19)

[Bjoern Krombholz]

On Wed, Aug 28, 2002 at 11:08:24PM +0200, Adrian Bunk wrote:
> On Mon, 26 Aug 2002, Bjoern Krombholz wrote:

Hi Adrian,

> > i'm currently have a problem that every program that tries to read from
> > /proc/stat like `uptime', `free', `cat /proc/stat' etc. segfaults.
> >...
> > kernel: c01f2ad9
> > kernel: Oops: 0002
> > kernel: CPU:    0
> > kernel: EIP:    0010:[number+1049/1088]    Tainted: PF
> >...
> 
> which binary-only modules (e.g. NVidia) are loaded on your computer? Is
> the problem reproducible without them ever loaded since the last reboot?
> 

Actually it happened after the decission of not using NVidia drivers any
longer. I didn't insmod them in this session and at the time of the crash
I was happy to reach 10 days uptime (the system crashed quite often
before, every 3 to 7 days, and completely locked, so I had no chance to
get to know what these crashes were caused by).

So, NVidia wasn't the problem, but maybe VMWare. I shut VMware down a few
hours before the first Oops occured.


Bjoern


PS:
One of the `crash|lock types' that happened most often is when the system/pci
bus is on high load, e.g. copying some big files from one partition to
another one while watching TV with xawtv (standard bttv driver); or while
using "transcode" (converting video files) and whatching TV or whatching
movies. It's always the same that the system locks while reading/writing
(might be only one of those operations) to the hard disk; the HD LED keeps
on.

Maybe this is related to VIAs hardware in some way, I don't know but'd
like to. :)

Re: Oops while accessing /proc/stat (2.4.19)

[Adrian Bunk]

On Thu, 29 Aug 2002, Bjoern Krombholz wrote:

>...
> > which binary-only modules (e.g. NVidia) are loaded on your computer? Is
> > the problem reproducible without them ever loaded since the last reboot?
> >
>
> Actually it happened after the decission of not using NVidia drivers any
> longer. I didn't insmod them in this session and at the time of the crash
> I was happy to reach 10 days uptime (the system crashed quite often
> before, every 3 to 7 days, and completely locked, so I had no chance to
> get to know what these crashes were caused by).
>
> So, NVidia wasn't the problem, but maybe VMWare. I shut VMware down a few
> hours before the first Oops occured.

It would be interesting to see whether this appears again without the
VMware module loaded before.

> Bjoern
>
>
> PS:
> One of the `crash|lock types' that happened most often is when the system/pci
> bus is on high load, e.g. copying some big files from one partition to
> another one while watching TV with xawtv (standard bttv driver); or while
> using "transcode" (converting video files) and whatching TV or whatching
> movies. It's always the same that the system locks while reading/writing
> (might be only one of those operations) to the hard disk; the HD LED keeps
> on.
>
> Maybe this is related to VIAs hardware in some way, I don't know but'd
> like to. :)


It might be:
1. a hardware problem
2. a bug in a non-free module
3. a bug in the kernel a non-free module suffers from
4. a bug in the kernel that is not related to non-free modules

If your problem is 1. or 4. there are people who can help you, in the case
of 2. or 3. it's impossible for everyone except the vendor of these moduls
to help you if these are binary-only modules (even in the case 3. where
it's a kernel bug because it's impossible to debug it).


cu
Adrian

-- 

You only think this is a free country. Like the US the UK spends a lot of
time explaining its a free country because its a police state.
								Alan Cox