[BUGREPORT] kernel BUG in page_alloc.c:141!

[BUGREPORT] kernel BUG in page_alloc.c:141!

[Erik McKee]

Hello!

Booted 2.5.24, and it ran fine for sometime, before it dead(live) locked,
causing a reboot.  Attempts to reboot were met with the following bug
immediatly after calibrating delay loop, which equates out to an
if(bad_range(buddy1,zone)) BUG; in __free_pages_ok:

ksymoops 2.3.7 on i586 2.4.19-pre10-ac2.  Options used
     -v /dosc/linux-2.5/vmlinux (specified)
     -K (specified)
     -L (specified)
     -O (specified)
     -m /boot/Sys-2.5.24 (specified)

kernel BUG at page_alloc.c:141!
invalid operand: 0000
CPU:    0
EIP:    010:[<c012bddb>]   Not tainted
Using defaults from ksymoops -t elf32-i386 -a i386
EFLAGS: 00010082
eax: 00001000   ebx: f1c728d6     ecx: 00000000       edx: 00001000
esi: 00000000   edi: c101ce20     ebp: 00000cd7       esp: c0277f58
ds: 0018   es: 0018   ss: 0018
Stack: c101ce44 00800000 00000cd7 00007ff0 00001000 c1001008 c101ce44 c0264000
       c1000008 c0264018 00000213 ffffffff 0000066b c012c5b3 c027dba2 00000000
       c0276000 c0105000 0008e000 ffffffff c02e5000 00000000 00000a8d c02b9800
Call Trace: [<c012c5b3>] [<c0105000>] [<c0105000>] [<c0105000>]
Code: 0f 0b 8d 00 81 9e 22 c0 8b 44 24 18 2b 44 24 14 8b 5c 21 10

>>EIP; c012bddb <__free_pages_ok+18b/2b0>   <=====
Trace; c012c5b3 <__free_pages+1b/1c>
Trace; c0105000 <_stext+0/0>
Trace; c0105000 <_stext+0/0>
Trace; c0105000 <_stext+0/0>
Code;  c012bddb <__free_pages_ok+18b/2b0>
00000000 <_EIP>:
Code;  c012bddb <__free_pages_ok+18b/2b0>   <=====
   0:   0f 0b                     ud2a      <=====
Code;  c012bddd <__free_pages_ok+18d/2b0>
   2:   8d 00                     lea    (%eax),%eax
Code;  c012bddf <__free_pages_ok+18f/2b0>
   4:   81 9e 22 c0 8b 44 24      sbbl   $0x442b1824,0x448bc022(%esi)
Code;  c012bde6 <__free_pages_ok+196/2b0>
   b:   18 2b 44 
Code;  c012bde9 <__free_pages_ok+199/2b0>
   e:   24 14                     and    $0x14,%al
Code;  c012bdeb <__free_pages_ok+19b/2b0>
  10:   8b 5c 21 10               mov    0x10(%ecx,1),%ebx

 <0>Kernel panic: Attempted to kill the idle task!



__________________________________________________
Do You Yahoo!?
Yahoo! - Official partner of 2002 FIFA World Cup
http://fifaworldcup.yahoo.com

Re: [BUGREPORT] kernel BUG in page_alloc.c:141!

[William Lee Irwin III]

On Fri, Jun 21, 2002 at 12:15:28PM -0700, Erik McKee wrote:
> Booted 2.5.24, and it ran fine for sometime, before it dead(live) locked,
> causing a reboot.  Attempts to reboot were met with the following bug
> immediatly after calibrating delay loop, which equates out to an
> if(bad_range(buddy1,zone)) BUG; in __free_pages_ok:

This looks odd. Can you by any chance disassemble the parts before this?
Or better yet, reproduce it with a kernel compiled with -g and objdump
--source --disassemble vmlinux to get the disassembly of __free_pages_ok()?


Thanks,
Bill

Re: [BUGREPORT] kernel BUG in page_alloc.c:141!

[William Lee Irwin III]

On Fri, Jun 21, 2002 at 12:15:28PM -0700, Erik McKee wrote:
>> Booted 2.5.24, and it ran fine for sometime, before it dead(live) locked,
>> causing a reboot.  Attempts to reboot were met with the following bug
>> immediatly after calibrating delay loop, which equates out to an
>> if(bad_range(buddy1,zone)) BUG; in __free_pages_ok:

On Fri, Jun 21, 2002 at 01:06:13PM -0700, William Lee Irwin III wrote:
> This looks odd. Can you by any chance disassemble the parts before this?
> Or better yet, reproduce it with a kernel compiled with -g and objdump
> --source --disassemble vmlinux to get the disassembly of __free_pages_ok()?

Oh, also include the updated oops if you can. What I'll be trying to do
is figure out what variables are in which registers and stack locations
to get their values and then fish their values out of the oops.


Thanks,
Bill

Re: [BUGREPORT] kernel BUG in page_alloc.c:141!

[Kai Germaschewski]

On Fri, 21 Jun 2002, William Lee Irwin III wrote:

> On Fri, Jun 21, 2002 at 12:15:28PM -0700, Erik McKee wrote:
> > Booted 2.5.24, and it ran fine for sometime, before it dead(live) locked,
> > causing a reboot.  Attempts to reboot were met with the following bug
> > immediatly after calibrating delay loop, which equates out to an
> > if(bad_range(buddy1,zone)) BUG; in __free_pages_ok:
> 
> This looks odd. Can you by any chance disassemble the parts before this?
> Or better yet, reproduce it with a kernel compiled with -g and objdump
> --source --disassemble vmlinux to get the disassembly of __free_pages_ok()?

"make mm/page_alloc.lst" may simplify this task. However, the usage of the
various macros seems to confuse gcc -g / objdump somewhat, so the output
isn't as clear as it could be.

--Kai

Re: [BUGREPORT] kernel BUG in page_alloc.c:141!

[William Lee Irwin III]

On Fri, Jun 21, 2002 at 12:15:28PM -0700, Erik McKee wrote:
>>> Booted 2.5.24, and it ran fine for sometime, before it dead(live) locked,
>>> causing a reboot.  Attempts to reboot were met with the following bug
>>> immediatly after calibrating delay loop, which equates out to an
>>> if(bad_range(buddy1,zone)) BUG; in __free_pages_ok:

On Fri, 21 Jun 2002, William Lee Irwin III wrote:
>> This looks odd. Can you by any chance disassemble the parts before this?
>> Or better yet, reproduce it with a kernel compiled with -g and objdump
>> --source --disassemble vmlinux to get the disassembly of __free_pages_ok()?

On Fri, Jun 21, 2002 at 03:23:13PM -0500, Kai Germaschewski wrote:
> "make mm/page_alloc.lst" may simplify this task. However, the usage of the
> various macros seems to confuse gcc -g / objdump somewhat, so the output
> isn't as clear as it could be.
> --Kai

I've had to work around that before, I'll probably still be able to
recognize it. It's still a royal PITA.

Cheers,
Bill