Hello,

I was testing 2.6.9-ac6 on one of our backup routers (usually running 
rock solid with a 2.4 kernel), but it consistently crashed within 10-20 
minutes after boot. Crash messages follow:

Unable to handle kernel paging request at virtual address 10c6483f
  printing eip:
c01428f2
*pde = 00000000
Oops: 0002 [#1]
Modules linked in: w83781d i2c_sensor i2c_viapro i2c_core dummy e100 mii 
bridge rtc
CPU:    0
EIP:    0060:[<c01428f2>]    Not tainted VLI
EFLAGS: 00010296   (2.6.9-ac6)
EIP is at fput+0x2/0x20
eax: 10c6482b   ebx: c6336078   ecx: cd8f239c   edx: 10c6482b
esi: c6336000   edi: c6336008   ebp: 00000001   esp: cf8e7e84
ds: 007b   es: 007b   ss: 0068
Process NeTraMet (pid: 1023, threadinfo=cf8e6000 task=cfd54090)
Stack: c01518c5 00000000 00000000 00000080 c0151d11 cf8e7ee8 00000000 
00000000
        00000080 00000008 000000f8 00000000 00000000 000000f8 cf8e6000 
ce42b8cc
        ce42b8c8 ce42b8c4 ce42b8d8 ce42b8d4 ce42b8d0 00000000 00000008 
00000000
Call Trace:
  [<c01518c5>] poll_freewait+0x35/0x50
  [<c0151d11>] do_select+0x2b1/0x2d0
  [<c01518e0>] __pollwait+0x0/0xa0
  [<c0151d47>] select_bits_alloc+0x17/0x20
  [<c01520d8>] sys_select+0x378/0x550
  [<c018a172>] copy_to_user+0x32/0x40
  [<c0108f8f>] timer_interrupt+0x2f/0xd0
  [<c01f54d4>] sock_get_timestamp+0x54/0x70
  [<c01f2a6d>] sock_ioctl+0x21d/0x230
  [<c0103c93>] syscall_call+0x7/0xb
Code: c0 e8 43 0d fd ff a1 28 77 28 c0 5a a3 e0 09 32 c0 59 eb 10 90 8d 
74 26 00 68 a0 7a 25 c0 e8 26 0d fd ff 58 31 c0 5b 5f c3 89 c2 <ff> 4a 
14 0f 94 c0 84 c0 74 07 89 d0 e9 0d 0000 00 c3 8d b6 00

Following this the machine was unreachable through the network, and 
after 60 seconds it was rebooted by the software watchdog.

The running process (NeTraMet) is a network traffic meter doing network 
flow accounting, and every 10 minutes the flows are collected by a 
central accounting server (by NeMaC). The router consistently crashed 
the first or second time this "flow collecting" ran after boot.

The kernel configuration and boot messages, as well as three sets of 
crash messages are attached. Please note, that after the first two 
crashes, CONFIG_PREEMPT was disabled and the kernel was recompiled to 
rule this out as the cause of the crashes. The kernel .config and boot 
messages are from the kernel which had preemption disabled, and the 
third crash in the attached file is the one shown above.

Any suggestions would be much appreciated - I'm willing to test patches 
that may fix the problem.

Regards,
Anders K. Pedersen