Help in debugging Memory corruption issue on Hugemem kernel

Help in debugging Memory corruption issue on Hugemem kernel

[Swamy_Gautham]

Please CC me when replying, thanks.

The issue is with a Multipathing driver on RH-AS-3.0
(2.4.21-9.0.1.ELhugemem) kernel.
The driver is corrupting memory which leads to a panic at a later point in
time when a "tar" or "cp" command is run. This happens only on hugemem
kernels and not on SMP kernels.

Happens only with machines that have at least 6GB of physical memory.

Here is a panic stack trace.

****************************************************************************
********************************************************
invalid operand: 0000

CPU: 0
EIP: 0060:[<02155c9d>] Tainted: P 
EFLAGS: 00010006

EIP is at rmqueue [kernel] 0x1fd (2.4.21-9.0.1.ELhugemem/i686)
eax: 00000000 ebx: 00000020 ecx: 000f6000 edx: 001797e0
esi: 023a2514 edi: 023a1300 ebp: 088798ac esp: ea2b9e74
ds: 0068 es: 0068 ss: 0068
Process tar (pid: 24913, stackpage=ea2b9000)
Stack: e8a7a740 06d280a8 000f6000 0213e921 00000005 000837e0 00083400
00000202 
00000000 023a1300 023a2624 00000000 000001d2 00000000 02155ff7 00000000 
00000fff ea86b200 00000000 f5300fff ea2b9f24 00000001 0215f03b f5301000 
Call Trace: [<0213e921>] follow_page [kernel] 0x1b1 (0xea2b9e80)
[<02155ff7>] __alloc_pages [kernel] 0x97 (0xea2b9eac)
[<0215f03b>] get_user_size [kernel] 0x4b (0xea2b9ecc)
[<021492e8>] do_generic_file_write [kernel] 0x168 (0xea2b9ef0)
[<021498ef>] generic_file_write [kernel] 0x13f (0xea2b9f48)
[<0213402b>] update_process_times_statistical [kernel] 0x7b (0xea2b9f5c)
[<f88bce89>] ext3_file_write [ext3] 0x39 (0xea2b9f74)
[<02161523>] sys_write [kernel] 0xa3 (0xea2b9f94)

Code: Bad EIP value.

CPU#3 is frozen.
CPU#1 is frozen.
****************************************************************************
*************************************************************

Any ideas as to how to debug this memory corruption issue ?

I have turned on "Debug High Memory Support" and "Debug Memory allocations"
via kdb, I do not know how to use this feature. Has anyone used this
debugging support.

Thanks 
Gautham

Re: Help in debugging Memory corruption issue on Hugemem kernel

[Christoph Hellwig]

On Sun, Apr 25, 2004 at 03:17:26PM +0100, Swamy_Gautham@emc.com wrote:
> Please CC me when replying, thanks.
> 
> The issue is with a Multipathing driver on RH-AS-3.0
> (2.4.21-9.0.1.ELhugemem) kernel.
> The driver is corrupting memory which leads to a panic at a later point in
> time when a "tar" or "cp" command is run. This happens only on hugemem
> kernels and not on SMP kernels.
> 
> Happens only with machines that have at least 6GB of physical memory.

Could you please post the url to the mentioned driver's sorce?