2.6.17.3 kernel panic

2.6.17.3 kernel panic

[Krzysztof Halasa]

Hi,

Just got a kernel panic, it tried to print something:

BUG: unable to handle kernel paging request at virtual address d47ddc68
eip = c010b247
*pde = 0
(repeated throughout the screen)

It's a laptop - mobile Celeron Coppermine 600 MHz, 128 MB RAM, i440BX
- Fujitsu-Siemens Liteline Plus 419A (LF6).
The kernel is 2.6.17.3 + Alan's IDE patch, standard config.

Restarted the machine and it seems it's do_page_fault:

c010ae3d T do_page_fault
c010b368 t .text.lock.fault

Not sure if the following means anything:

.L138:
	.loc 1 551 0
	movl	8(%esp), %edx
	movl	%ebx, 496(%edx) <<<<<<<<<<<<<<<< faults here
	.loc 1 552 0
	movl	$14, 500(%edx)
	.loc 1 553 0
	movl	(%esp), %eax
	movl	%eax, 504(%edx)
	.loc 1 554 0
	movl	%eax, %ecx
	movl	$.LC12, %eax
	movl	4(%esp), %edx
.LCFI53:
	call	die


arch/i386/mm.c:
/*
 * Oops. The kernel tried to access some bad page. We'll have to
 * terminate things with extreme prejudice.
 */
...
	tsk->thread.cr2 = address; <<<<<<<<<<<<<<<< left side faults here
	tsk->thread.trap_no = 14;
	tsk->thread.error_code = error_code;
	die("Oops", regs, error_code);
	bust_spinlocks(0);
	do_exit(SIGKILL);

This machine has a history of mysterious hard freezes since ca 2.6.0
(black screen of death, I think there were no such problems with 2.4).
The CPU lacks LAPIC. It could be hardware, but I've never spotted any
specific problem, memtest86 doesn't show anything either.

While this time the machine was almost (but not completely) idle,
the hard freezes seems to occur mostly with simultaneous disk and LAN
activity (Ethernet is Tulip 21143 32-bit PC Card, by DLink, the disk
is IBM/Hitachi IC25N020ATMR04-0 Rev: MO1O).

$ cat /proc/iomem 
00000000-0009f7ff : System RAM
0009f800-0009ffff : reserved
000a0000-000bffff : Video RAM area
000c0000-000cbfff : Video ROM
000f0000-000fffff : System ROM
00100000-07feffff : System RAM
  00100000-002a889a : Kernel code
  002a889b-00349233 : Kernel data
07ff0000-07fffbff : ACPI Tables
07fffc00-07ffffff : ACPI Non-volatile Storage
10000000-11ffffff : PCI CardBus #02
  10000000-1003ffff : 0000:02:00.0
12000000-13ffffff : PCI CardBus #02
  12000000-120003ff : 0000:02:00.0
    12000000-120003ff : tulip
14000000-15ffffff : PCI CardBus #06
16000000-17ffffff : PCI CardBus #06
18000000-18000fff : 0000:00:0a.0
  18000000-18000fff : yenta_socket
18001000-18001fff : 0000:00:0a.1
  18001000-18001fff : yenta_socket
f4000000-f7ffffff : 0000:00:00.0
f8000000-fbffffff : PCI Bus #01
  f8000000-fbffffff : 0000:01:00.0
fedfe000-fedfffff : 0000:00:0d.0
  fedfe000-fedfffff : Maestro3
fffeac00-ffffffff : reserved

What could it be?
How could I debug it?
-- 
Krzysztof Halasa

Re: 2.6.17.3 kernel panic

[Michal Piotrowski]

Hi Krzysztof,

On 16/07/06, Krzysztof Halasa <khc@pm.waw.pl> wrote:
> Hi,
>
> Just got a kernel panic, it tried to print something:
>
> BUG: unable to handle kernel paging request at virtual address d47ddc68
> eip = c010b247
> *pde = 0
> (repeated throughout the screen)
>
> It's a laptop - mobile Celeron Coppermine 600 MHz, 128 MB RAM, i440BX
> - Fujitsu-Siemens Liteline Plus 419A (LF6).
> The kernel is 2.6.17.3 + Alan's IDE patch, standard config.
>
> Restarted the machine and it seems it's do_page_fault:
>
> c010ae3d T do_page_fault
> c010b368 t .text.lock.fault
[snip]
>
> What could it be?
> How could I debug it?

please try
gdb vmlinux
list *0xc010b247

> --
> Krzysztof Halasa
> -

Regards,
Michal

-- 
Michal K. K. Piotrowski
LTG - Linux Testers Group
(http://www.stardust.webpages.pl/ltg/wiki/)

Re: 2.6.17.3 kernel panic

[Krzysztof Halasa]

"Michal Piotrowski" <michal.k.k.piotrowski@gmail.com> writes:

>> c010ae3d T do_page_fault
>> c010b368 t .text.lock.fault
> [snip]
>>
>> What could it be?
>> How could I debug it?
>
> please try
> gdb vmlinux
> list *0xc010b247

I actually included listing of the code fragment in question (both
assembly and C).
-- 
Krzysztof Halasa

Re: 2.6.17.3 kernel panic

[Michal Piotrowski]

On 16/07/06, Krzysztof Halasa <khc@pm.waw.pl> wrote:
> "Michal Piotrowski" <michal.k.k.piotrowski@gmail.com> writes:
>
> >> c010ae3d T do_page_fault
> >> c010b368 t .text.lock.fault
> > [snip]
> >>
> >> What could it be?
> >> How could I debug it?
> >
> > please try
> > gdb vmlinux
> > list *0xc010b247
>
> I actually included listing of the code fragment in question (both
> assembly and C).

Ah, yes I see. Sorry for the noise.

Have you tried memtest86+ (http://www.memtest.org/)?

> --
> Krzysztof Halasa
>

Regards,
Michal

-- 
Michal K. K. Piotrowski
LTG - Linux Testers Group
(http://www.stardust.webpages.pl/ltg/wiki/)

Re: 2.6.17.3 kernel panic

[Krzysztof Halasa]

"Michal Piotrowski" <michal.k.k.piotrowski@gmail.com> writes:

> Have you tried memtest86+ (http://www.memtest.org/)?

Both that and memtest86, several times. Will run it again, though.
-- 
Krzysztof Halasa

Re: 2.6.17.3 kernel panic

[Michal Piotrowski]

On 16/07/06, Krzysztof Halasa <khc@pm.waw.pl> wrote:
> "Michal Piotrowski" <michal.k.k.piotrowski@gmail.com> writes:
>
> > Have you tried memtest86+ (http://www.memtest.org/)?
>
> Both that and memtest86, several times. Will run it again, though.

So if you don't have any bad blacks... this it similar to my bug
report http://www.ussg.iu.edu/hypermail/linux/kernel/0606.3/index.html#2558

As Dave Jones has said "Looks like a use after free".

> --
> Krzysztof Halasa
>

Regards,
Michal

-- 
Michal K. K. Piotrowski
LTG - Linux Testers Group
(http://www.stardust.webpages.pl/ltg/wiki/)

Re: 2.6.17.3 kernel panic

[Krzysztof Halasa]

"Michal Piotrowski" <michal.k.k.piotrowski@gmail.com> writes:

>> Both that and memtest86, several times. Will run it again, though.
>
> So if you don't have any bad blacks...

None if you mean HDD. memtest86+ from FC5 boot disk haven't found
anything wrong with RAM, either.

> this it similar to my bug
> report http://www.ussg.iu.edu/hypermail/linux/kernel/0606.3/index.html#2558

Not very, actually. Looks it's time to give this machine some
real exercise. The problem is it's random.
-- 
Krzysztof Halasa

Re: 2.6.17.3 kernel panic

[Grzegorz Kulewski]

On Mon, 17 Jul 2006, Krzysztof Halasa wrote:
> None if you mean HDD. memtest86+ from FC5 boot disk haven't found
> anything wrong with RAM, either.
>
>> this it similar to my bug
>> report http://www.ussg.iu.edu/hypermail/linux/kernel/0606.3/index.html#2558
>
> Not very, actually. Looks it's time to give this machine some
> real exercise. The problem is it's random.

Have you tried prime (from GIMPS) on this machine? It can detect errors on 
systems that have no apparent or visible problems and memtest does not 
show anything bad. It suprised me several times showing problems on 
machines that I could swear were ok.

But maybe it is not hardware fault at all...


Grzegorz Kulewski


PS. Greetings to pm.waw.pl admin! :-)