> The error may well be due to a bogus kernel pointer being used for bi=
tmap=20
> operations too, of course. That's the more likely explanation (bogus=20
> pointers _usually_ result in byte or word corruption rather than just=
a=20
> single bit error, but this can be unlucky).
If it might be of interest : I also encountered a memory corruption wit=
h
1.3.95 :
ksymoops output...) :
divide error: 0000
CPU: 0
EIP: 0010:[do_truncate+35/128]
EFLAGS: 00010202
eax: 0046f4e4 ebx: 0046f4b0 ecx: 0046f4b0
edx: 0046f4b0
esi: 0000202a edi: 00000000 ebp: 005b8f94
esp: 005b8f1c
ds: 0018 es: 0018 fs: 002b gs: 002b ss: 0018
Process bash (pid: 4989, process nr: 4, stackpage=3D005b8000)
Stack: 00000000 00000202 000081a4 000081a4 00000002 0012c730 0046f4b0 0=
0000002
00000000 00000202 0012d098 0046f4b0 00000000 006fe540 000001b6 0=
032c000
bffffd28 0046f4b0 004b7000 0032c017 0000000d 00122570 0032c000 0=
0000202
Call Trace: [permission+32/192] [open_namei+1048/1152] [do_open+128/384=
] [sys_open+50/96] [system_call+89/160]
>>EIP: 121863 <do_truncate+23/80>
Trace: 12c730 <permission+20/c0>
Trace: 12d098 <open_namei+418/480>
Trace: 122570 <do_open+80/180>
Trace: 1226a2 <sys_open+32/60>
Trace: 10a5d9 <system_call+59/a0>
Code: 121863 <do_truncate+23/80> idivl 0x18(%esp,1),%eax
Code: 121867 <do_truncate+27/80> movl $0x48,0xc(%esp,1)
Code: 12186f <do_truncate+2f/80> leal 0xc(%esp,1),%eax
Code: 121873 <do_truncate+33/80> pushl %eax
Code: 121874 <do_truncate+34/80> pushl %ebx
Code: 121875 <do_truncate+35/80> call 9090009d <_EIP+9090009d>
(idivl is bogus -- at this place we have "newattrs.ia_size =3D length;"=
from
open.c:68 (in do_truncate)). This caused processes to die on SIGFPE eac=
h
time they tried to truncate a file. No other symptom was visible, shutd=
own
went OK.
Never had RAM problems before. It happened in the middle of a session.
Everything is OK so far with 1.3.96.
Hope this helps...
--=20
Thomas.Quinot@Cuivre.FdN.FR <URL:http://Web.FdN.FR/~tquinot=
/>