OOPS: dcache_dir_lseek, 2.4.21 and 2.4.20

I see this from time to time when I leave a UML instance running for a long period of time (I get this oops on the host, not the UML, of course). It is not immediately reproducible, but has happened several times under similar circumstances. Previously I had seen it on 2.4.20 as well.

 - mdz


Unable to handle kernel NULL pointer dereference at virtual address 00000000
c014c4d2
*pde = 00000000
Oops: 0002
CPU:    0
EIP:    0010:[<c014c4d2>]    Not tainted
Using defaults from ksymoops -t elf32-i386 -a i386
EFLAGS: 00010282
eax: 00000000   ebx: c22762e0   ecx: cbb159e8   edx: c22762e0
esi: 00000002   edi: 00000000   ebp: c22762c0   esp: c663df84
ds: 0018   es: 0018   ss: 0018
Process linux (pid: 936, stackpage=c663d000)
Stack: fffffff2 c22762c0 00000000 c014c3c0 c63e97c0 ffffffea c013d7b5 c63e97c0 
       00000002 00000000 00000000 c663c000 00000002 a691bafc a691ba8c c0108eff 
       00000037 00000002 00000000 00000002 a691bafc a691ba8c 00000013 0000002b 
Call Trace:    [<c014c3c0>] [<c013d7b5>] [<c0108eff>]
Code: 89 10 89 42 04 8b 54 24 1c 8b 42 08 8b 40 08 8d 48 68 ff 40 

Trace; c014c3c0 <dcache_dir_lseek+0/140>
Trace; c013d7b5 <sys_lseek+65/a0>
Trace; c0108eff <system_call+33/38>

Code; c014c4d2 <dcache_dir_lseek+112/140>
00000000 <_EIP>:
Code; c014c4d2 <dcache_dir_lseek+112/140> <=====
0: 89 10 mov %edx,(%eax) <=====
Code; c014c4d4 <dcache_dir_lseek+114/140>
2: 89 42 04 mov %eax,0x4(%edx)
Code; c014c4d7 <dcache_dir_lseek+117/140>
5: 8b 54 24 1c mov 0x1c(%esp,1),%edx
Code; c014c4db <dcache_dir_lseek+11b/140>
9: 8b 42 08 mov 0x8(%edx),%eax
Code; c014c4de <dcache_dir_lseek+11e/140>
c: 8b 40 08 mov 0x8(%eax),%eax
Code; c014c4e1 <dcache_dir_lseek+121/140>
f: 8d 48 68 lea 0x68(%eax),%ecx
Code; c014c4e4 <dcache_dir_lseek+124/140>
12: ff 40 00 incl 0x0(%eax)