strange kernel hang in 2.2.{13-17}+RAID+NFS

[Neil Brown]

Hi,
 I have a problem that I hope someone can shed some light on.

 I have an NFS server with a RAID5 filesystem which has been running
 2.2.13+raid+nfsd  for several months.
 About a week ago it froze with the message

     Unable to handle kernel paging request

 (out of arch/i386/mm/fault.c)
 what is particularly strange is that it didn't go on to print
    " at virtual address ......

 as one would expect.
 This problem, which first happened on a Thursday, happened again on
 Friday and Saturday :-)

 I upgraded to 2.2.17pre14+RAID+NFS and it stayed up for longer, but
 died again the next Saturday (this time with 
    Unable to handle kernel NULL pointer dereference
 )

 When it was hung, nothing, not even SysRq would work.

 So my question is, what could possibly be causing it to hang here and
 not print out all the useful information that should follow?
 If I can get that info out, I could atleast make a reasonable attempt
 at finding out what is wrong.

 The machine is a Dual-PII with an aha2940 scsi controller and
 ee-pro100b ethernet card, 512Meg RAM.

NeilBrown