Next-in-Thread Next Message

Question Kernel oops when ejecting CBEM56G 

Forum: Xircom PCMCIA Ethernet Adapter Issues
Date: 2000, Feb 09
From: Eyvind Bernhardsen eyvind

I just upgraded to pcmcia-cs 3.1.9 (precompiled Debian "woody" version), which has fixed earlier problems I had with pcmcia, but which Oopses whenever the card is ejected, cardctl eject is called, or the system shuts down/reboots.

I'm not sure if I had this problem before, since I was pretty much happy to have the card _not_ lock the system solid (which it frequently did with pcmcia-cs 3.1.8), and didn't bother to check the kernel logs much once things were up and running :)

The system is a Dell Latitude CPtV466GT running Debian "potato" (aka "frozen") with kernel 2.2.14 plus the ext3 patch, the international kernel patch, and the 2.3.39 USB backport.

The output of ksymoops is as follows:

ksymoops 2.3.3 on i686 2.2.14ext3.  Options used
     -V (default)
     -k /proc/ksyms (default)
     -l /proc/modules (default)
     -o /lib/modules/2.2.14ext3/ (default)
     -m /boot/ (specified)

Unable to handle kernel paging request at virtual address 5a5a5ab6
current->tss.cr3 = 06f90000, %cr3 = 06f90000
*pde = 00000000
Oops: 0000
CPU:    0
EIP:    0010:[<c80575f2>]
Using defaults from ksymoops -t elf32-i386 -a i386
EFLAGS: 00010206
eax: 5a5a5a5a   ebx: c0233800   ecx: 5a5a5a5a   edx: c80599fc
esi: c80599fc   edi: c8052000   ebp: c6f8ff3c   esp: c6f8ff2c
ds: 0018   es: 0018   ss: 0018
Process rmmod (pid: 584, process nr: 36, stackpage=c6f8f000)
Stack: 00004000 c7e11ee0 0000000c c011ba98 c6f8ff6c c805790e c8059be8 00000000
       c6fbb150 c0116d73 c8052000 fffffff0 c7144180 08054006 00000000 c6f8e000
       bfffecac c0117130 c8052000 fffffff0 c6f8ffb0 bfffecac c4b77000 fffffffe
Call Trace: [<c011ba98>] [<c805790e>] [<c8059be8>] [<c0116d73>] [<c8052000>] [<c0117130>] [<c8052000>]
       [<c0116775>] [<c8052000>] [<c0116737>] [<c010a041>] [<c0109f3c>]
Code: 8b 59 5c 8d 53 0c 8a 83 72 05 00 00 24 03 3c 02 75 29 83 c4

>>EIP; c80575f2 <[tulip_cb]tulip_reap+2e/78>   <=====
Trace; c011ba98 <handle_mm_fault+c8/13c>
Trace; c805790e <[tulip_cb]cleanup_module+1a/84>
Trace; c8059be8 <[tulip_cb]tulip_ops+0/13>
Trace; c0116d73 <qm_info+9f/b8>
Trace; c8052000 <[tulip_cb]__module_kernel_version+0/1a>
Trace; c0117130 <free_module+20/a8>
Trace; c8052000 <[tulip_cb]__module_kernel_version+0/1a>
Trace; c0116775 <sys_delete_module+13d/1ec>
Trace; c8052000 <[tulip_cb]__module_kernel_version+0/1a>
Trace; c0116737 <sys_delete_module+ff/1ec>
Trace; c010a041 <error_code+2d/34>
Trace; c0109f3c <system_call+34/38>
Code;  c80575f2 <[tulip_cb]tulip_reap+2e/78>
00000000 <_EIP>:
Code;  c80575f2 <[tulip_cb]tulip_reap+2e/78>   <=====
   0:   8b 59 5c                  mov    0x5c(%ecx),%ebx   <=====
Code;  c80575f5 <[tulip_cb]tulip_reap+31/78>
   3:   8d 53 0c                  lea    0xc(%ebx),%edx
Code;  c80575f8 <[tulip_cb]tulip_reap+34/78>
   6:   8a 83 72 05 00 00         mov    0x572(%ebx),%al
Code;  c80575fe <[tulip_cb]tulip_reap+3a/78>
   c:   24 03                     and    $0x3,%al
Code;  c8057600 <[tulip_cb]tulip_reap+3c/78>
   e:   3c 02                     cmp    $0x2,%al
Code;  c8057602 <[tulip_cb]tulip_reap+3e/78>
  10:   75 29                     jne    3b <_EIP+0x3b> c805762d <[tulip_cb]tulip_reap+69/78>
Code;  c8057604 <[tulip_cb]tulip_reap+40/78>
  12:   83 c4 00                  add    $0x0,%esp

I've managed to parse the disassembled output enough to figure out that the kernel oopses in tulip_reap() while trying to dereference (*devp)->priv on line 3459 of tulip_cb.c, which might mean that root_tulip_dev is invalid, but I don't know how to debug the kernel so I can't find out :) What I _do_ know is that tulip_reap is called twice; once while loading the module, and once while cleaning up after it (I get the "tulip_reap()" message twice in my kernel log). The first invocation works fine, obviously.

Since I don't usually eject the network card once I've inserted it this isn't really a big deal for me, but I'd be happy to provide my configuration, log messages etc. if anyone wants to take a harder look at it.

Next-in-Thread Next Message

Messages Inline: 1 All Outline: 1 2 All

1. None Ok, I think I found the problem by David Hinds, 2000, Feb 09
(_ Sad Possibly, but... by Eyvind Bernhardsen, 2000, Feb 10
(_ Ok Yes! by Eyvind Bernhardsen, 2000, Feb 11

Add Message to: "Kernel oops when ejecting CBEM56G"

Members Subscribe Admin Mode Show Frames Help for HyperNews at 1.10