Philipp Rumpf wrote:
> On Sun, Nov 29, 1998 at 03:40:14PM +0100, Christian Loth wrote:
> > > > Yep, that reproduced it. Well, not exactly like that. Instead of a hardlock
> > > > I had a totally unstable system afterwards: X crashing, not being able
> > > > to start it up again, shells not starting up when trying to log in at
> > > > a console, after a while no console switching possible.
> > >
> > > What is most important is if you got the log messages again.
> >
> > Yes, it was the same log message again. But with one more interesting thing
> > I found...the thing with not being able to log in:
>
> > Nov 29 15:13:38 loth kernel: Out of memory for bash.
> >
> > Memory leak?
>
> Maybe.
>
> > > > However, I have no idea how to solve this problem. Is it a fault of the
> > > > kernel? Or a fault of my screwed system? If the second, any suggestions
> > > > how I could fix it?
> > >
> > > If I am right, it is a problem of your kernel when having a broken fat fs
> > > mounted. Here is a small patch that should fix the problem you reported
> > > first. Please report if you can still lock your system with this patch
> > > applied. Thanks,
> > >
> >
> > Ok, compiling...will tell you after reboot and test.
>
> Oops, just read the code again ... no, my first patch should do exactly nothing.
> Sorry for wasting your time. I can give you another patch, but I do not really
> understand what it would change ...
Well it *did* something. I didn't get the log messages anymore and the
find process ended in a segmentation fault, and the following oops
was produced (and run through ksymoops):
Unable to handle kernel paging request at virtual address 8110289c
current->tss.cr3 = 0347f000, %cr3 = 0347f000
*pde = 00000000
Oops: 0000
CPU: 0
EIP: 0010:[<c0127893>]
EFLAGS: 00010287
eax: 00000019 ebx: c2ee6120 ecx: 8110289c edx: c0f42c94
esi: 00000200 edi: 303d0b25 ebp: 00000301 esp: c33cdd64
ds: 0018 es: 0018 ss: 0018
Process find (pid: 286, process nr: 33, stackpage=c33cd000)
Stack: 00000301 c02d7200 c4a1a999 c0127f96 00000301 0029d728 00000200 c34e9074
c4a1a0a3 00000301 0029d728 00000200 0029d728 003e9400 c02d7200 c34e9074
00000301 c4a1e233 c02d7200 0029d728 c2ee73e0 c33cde3c c34e9074 00000000
Call Trace: [<c4a1a999>] [<c0127f96>] [<c4a1a0a3>] [<c4a1e233>] [<c4a1b376>] [<c4a287a0>] [<c4a10000>]
[<c4a1a735>] [<c0127ecf>] [<c4a1d820>] [<c010e032>] [<c010e0ac>] [<c0108e7d>] [<c011ba6b>] [<c011bac0>]
[<c011bc66>] [<c4a1b413>] [<c4a1b3c0>] [<c012f594>] [<c012f6d0>] [<c012f594>] [<c0107ac4>]
Code: 3b 9a 08 fc 1b c0 75 11 8b 43 1c 89 82 08 fc 1b c0 e9 a5 00
Using `/System.map' to map addresses to symbols.
>>EIP: c0127893 <getblk+53/340>
Trace: c4a1a999
Trace: c0127f96 <bread+1a/88>
Trace: c4a1a0a3
Trace: c4a1e233
Trace: c4a1b376
Trace: c4a287a0
Trace: c4a10000
Trace: c4a1a735
Trace: c0127ecf <__brelse+23/44>
Trace: c4a1d820
Trace: c010e032 <smp_local_timer_interrupt+ca/134>
Trace: c010e0ac <smp_apic_timer_interrupt+10/18>
Trace: c0108e7d <apic_timer_interrupt+1d/28>
Trace: c011ba6b <do_anonymous_page+7b/8c>
Trace: c011bac0 <do_no_page+44/e8>
Trace: c011bc66 <handle_mm_fault+102/1cc>
Trace: c4a1b413
Trace: c4a1b3c0
Trace: c012f594 <filldir>
Trace: c012f6d0 <sys_getdents+b8/144>
Trace: c012f594 <filldir>
Trace: c0107ac4 <system_call+34/40>
Code: c0127893 <getblk+53/340>
Code: c0127893 <getblk+53/340> 3b 9a 08 fc 1b cmpl 0xc01bfc08(%edx),%ebx
Code: c0127899 <getblk+59/340> 75 11 jne c01278ac <getblk+6c/340>
Code: c012789b <getblk+5b/340> 8b 43 1c movl 0x1c(%ebx),%eax
Code: c012789e <getblk+5e/340> 89 82 08 fc 1b movl %eax,0xc01bfc08(%edx)
Code: c01278a4 <getblk+64/340> e9 a5 00 00 90 jmp 900000bb <_EIP+0x900000bb>
Code: c01278a9 <getblk+69/340> 90 nop
Code: c01278aa <getblk+6a/340> 90 nop
Cannot read eip address from EIP: line. Is this a valid oops file?
- Chris
-- /----------------------------------------------------------------------------\ | Christian Loth, chris@rom.mud.de | REALM OF MAGIC | | http://rom.mud.de/~chris | telnet rom.mud.de 4000 | | | http://rom.mud.de/ | \----------------------------------------------------------------------------/- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/