Oops in 2.6.x maybe r8169

Oops in 2.6.x maybe r8169

[Matthias Bernges]

Hello,

since I use the Realtek 8169 Network card I get a kernel Oops
after which the kernel hangs completly.
I tried Kernel 2.6.6, 2.6.7 and 2.6.8.1. It appears randomly but
only if the machine has high load and high network traffic.
I've attached the output of the Oops with Kernel 2.6.8.1 (using
ksymoops) with kernel 2.6.7 I don't get a oops the PC just hangs.
2.6.6 gives me a veery long output which I can't catch (completly)
because I don't have a null modem cable.



Oops: 0000 [#1]
CPU:    0
EIP:    0060:[<c0271498>]    Not tainted
EFLAGS: 00010206    (2.6.8.1)
eax: 20000100   ebx: 0002bc49   ecx: 00000090   edx: 0000003c
esi: 00000000   edi: c15eb220   ebp: 00000009   esp: c0497f80
ds: 007b   es: 007b  ss: 0068
Stack: dff2c800 c15eb3d0 decde080 00000202 decde000 fffd4440 00000004
e0807e00       00000014 c0497000 c0271901 c15eb000 c15eb220 e0807e00
00000001 c15eb220       deceb460 04000001 00000000 df3af8d4 c0105a3a
00000005 c15eb000 df3af8d4 Call Trace:
Code: 8b 46 60 ba 3c 00 00 00 83 f8 3b 0f 46 c2 ff 47 0c 01 47 14


>>EIP; c0271498 <SELECT_DRIVE+18/50>   <=====

>>edi; c15eb220 <pg0+113d220/3fb50000>
>>esp; c0497f80 <per_cpu__tvec_bases+ec0/1008>

Code;  c0271498 <SELECT_DRIVE+18/50>
00000000 <_EIP>:
Code;  c0271498 <SELECT_DRIVE+18/50>   <=====
   0:   8b 46 60                  mov    0x60(%esi),%eax   <=====
Code;  c027149b <SELECT_DRIVE+1b/50>
   3:   ba 3c 00 00 00            mov    $0x3c,%edx
Code;  c02714a0 <SELECT_DRIVE+20/50>
   8:   83 f8 3b                  cmp    $0x3b,%eax
Code;  c02714a3 <SELECT_DRIVE+23/50>
   b:   0f 46 c2                  cmovbe %edx,%eax
Code;  c02714a6 <SELECT_DRIVE+26/50>
   e:   ff 47 0c                  incl   0xc(%edi)
Code;  c02714a9 <SELECT_DRIVE+29/50>
  11:   01 47 14                  add    %eax,0x14(%edi)

<0>Kernel panic: Fatal exception in interrupt

Re: Oops in 2.6.x maybe r8169 (maybe disk related as well)

[Francois Romieu]

Matthias Bernges <mbernges@rumms.uni-mannheim.de> :
[...]
> since I use the Realtek 8169 Network card I get a kernel Oops
> after which the kernel hangs completly.
> I tried Kernel 2.6.6, 2.6.7 and 2.6.8.1. It appears randomly but
> only if the machine has high load and high network traffic.

Can you give 2.6.9-rc3 a try ?

[...]
> >>EIP; c0271498 <SELECT_DRIVE+18/50>   <=====
> 
> >>edi; c15eb220 <pg0+113d220/3fb50000>
> >>esp; c0497f80 <per_cpu__tvec_bases+ec0/1008>
> 
> Code;  c0271498 <SELECT_DRIVE+18/50>
> 00000000 <_EIP>:
> Code;  c0271498 <SELECT_DRIVE+18/50>   <=====
>    0:   8b 46 60                  mov    0x60(%esi),%eax   <=====
> Code;  c027149b <SELECT_DRIVE+1b/50>
>    3:   ba 3c 00 00 00            mov    $0x3c,%edx

Your ata subsytem does not seem happy.

Can you provide:
- a short description of the system;
- the revision of your compiler;
- lspci -vx output;
- /sbin/lsmod output;
- complete dmesg after boot;
- vmstat 1 for a few seconds during network load;
- the content of /proc/interrupts adter a few seconds of network load.

--
Ueimor

Re: Oops in 2.6.x maybe r8169 (maybe disk related as well)

[Matthias Bernges]

Hello,

sorry that the answer took so long but I had some exams.

Am Di 05.10.04 um 12:59 CEST schrieb Francois Romieu
<romieu@fr.zoreil.com>:

> Matthias Bernges <mbernges@rumms.uni-mannheim.de> :
> [...]
> > since I use the Realtek 8169 Network card I get a kernel Oops
> > after which the kernel hangs completly.
> > I tried Kernel 2.6.6, 2.6.7 and 2.6.8.1. It appears randomly but
> > only if the machine has high load and high network traffic.
> 
> Can you give 2.6.9-rc3 a try ?

I tried 2.6.9-rc4 and had the same problem.

> 
> [...]
> > >>EIP; c0271498 <SELECT_DRIVE+18/50>   <=====
> > 
> > >>edi; c15eb220 <pg0+113d220/3fb50000>
> > >>esp; c0497f80 <per_cpu__tvec_bases+ec0/1008>
> > 
> > Code;  c0271498 <SELECT_DRIVE+18/50>
> > 00000000 <_EIP>:
> > Code;  c0271498 <SELECT_DRIVE+18/50>   <=====
> >    0:   8b 46 60                  mov    0x60(%esi),%eax   <=====
> > Code;  c027149b <SELECT_DRIVE+1b/50>
> >    3:   ba 3c 00 00 00            mov    $0x3c,%edx
> 
> Your ata subsytem does not seem happy.
> 
> Can you provide:
> - a short description of the system;

Gentoo Linux, Duron 900Mhz, 3 Nics (3com, Via-Rhine On Board and r8169),
Via Chipset, 2x2GB HardDisk, DVD, Something more?

> - the revision of your compiler;

lise root # gcc --version
gcc (GCC) 3.3.4 20040623 (Gentoo Linux 3.3.4-r1, ssp-3.3.2-2, pie-8.7.6)

> - lspci -vx output;

http://www.matthiasbernges.de/linux/lspci

> - /sbin/lsmod output;

Module                  Size  Used by
rtc                    10296  0 

> - complete dmesg after boot;

http://www.matthiasbernges.de/linux/dmesg

> - vmstat 1 for a few seconds during network load;
http://www.matthiasbernges.de/linux/vmstat

> - the content of /proc/interrupts adter a few seconds of network load.

           CPU0       
  0:     795867          XT-PIC  timer
  1:          8          XT-PIC  i8042
  2:          0          XT-PIC  cascade
  3:          0          XT-PIC  ehci_hcd, CMI8738-MC6
  5:     866353          XT-PIC  uhci_hcd, eth2
  8:          2          XT-PIC  rtc
  9:          0          XT-PIC  acpi
 10:      10964          XT-PIC  uhci_hcd, eth0
 11:     215952          XT-PIC  uhci_hcd, eth1
 12:         58          XT-PIC  i8042
 14:      11944          XT-PIC  ide0
 15:         64          XT-PIC  ide1
NMI:          0 
ERR:          0



matthias