[uml-devel] page_mapcount(page) went negative

[uml-devel] page_mapcount(page) went negative

[Anthony Brock]

One of my customer UML instances started crashing last night. It was
originally running under a very old version of the kernel, so I upgraded to
2.6.28.7. However, we are still encountering the crashes. I've verified that
the customer has not exhausted their disk space, but the error seems to
indicate something related to memory:

Eeek! page_mapcount(page) went negative! (-1)
  page pfn = 1261	
  page->flags = 0
  page->count = 2
  page->mapping = 00000000
  vma->vm_ops = 0x0
Kernel panic - not syncing: Kernel mode signal 4

EIP: 0073:[<40388868>] CPU: 0 Not tainted ESP: 007b:bfa7f060 EFLAGS:
00000246
    Not tainted
EAX: ffffffda EBX: 00000007 ECX: 080a3190 EDX: 00000000
ESI: 00000000 EDI: 00000000 EBP: bfa7fff8 DS: 007b ES: 007b
081fdbb4:  [<0806a327>] show_regs+0xc4/0xc9
081fdbe0:  [<08059a16>] panic_exit+0x25/0x3b
081fdbf4:  [<0808411b>] notifier_call_chain+0x27/0x4c
081fdc1c:  [<0808416e>] atomic_notifier_call_chain+0x15/0x17
081fdc2c:  [<081984b1>] panic+0x4c/0xd2
081fdc48:  [<080593cf>] relay_signal+0x31/0x6c
081fdc68:  [<080650a2>] sig_handler_common+0x61/0x70
081fdce0:  [<080651ca>] sig_handler+0x31/0x3d
081fdcec:  [<08065363>] handle_signal+0x4f/0x7d
081fdd0c:  [<08066a0b>] hard_handler+0xf/0x14
081fdd1c:  [<ffffe420>] _etext+0xf7e64328/0x0

It appears to be very repeatable at the moment as the crashes are occurring
within a few minutes of boot. I noticed a similar bug report in the archives
for kernel version 2.6.24. Is this a known issue? If so, is there a
work-around? What additional information should I supply?

I can supply the text from the earlier kernel version if that might help.
Any help is appreciated!

Tony


------------------------------------------------------------------------------
Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA
-OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise
-Strategies to boost innovation and cut costs with open source participation
-Receive a $600 discount off the registration fee with the source code: SFAD
http://p.sf.net/sfu/XcvMzF8H
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel

Re: [uml-devel] page_mapcount(page) went negative

[brocka]

This is definitely repeatable. Unfortunately I don't have a lot of  
visibility into what is actually happening in the instance (or  
triggering the crash), but it does appear to be memory related.

The instance originally had 64M of memory (almost 628M of swap in two  
files) and crashed with 5 minutes of boot (starting last night for  
some reason, after running for almost 2 months without issue). We  
increased the time between crashes to every 30 minutes by increasing  
the physical memory to 96M. It now has 128M of physical memory and  
crashes approximately 90 minutes after boot. I'm waiting for the next  
crash.

The configuration for the guest kernel can be found here:

http://www.coherenthosting.com/prj/uml/miscellaneous/config-uml_2.6.28.7-00-32bit

The cpuinfo of the host machine can be found here:

http://www.coherenthosting.com/prj/uml/miscellaneous/2.6.28.7-host-cpuinfo

For reference I am placing a copy of the crash output from when this  
instance was still running the kernel version 2.6.20.6 here:

http://www.coherenthosting.com/prj/uml/miscellaneous/2.6.20.6-crash-log.txt

Any help is appreciated!

Tony


Quoting "Anthony Brock" <brocka@sterlingcgi.com>:
> One of my customer UML instances started crashing last night. It was
> originally running under a very old version of the kernel, so I upgraded to
> 2.6.28.7. However, we are still encountering the crashes. I've verified that
> the customer has not exhausted their disk space, but the error seems to
> indicate something related to memory:
>
> Eeek! page_mapcount(page) went negative! (-1)
>   page pfn = 1261
>   page->flags = 0
>   page->count = 2
>   page->mapping = 00000000
>   vma->vm_ops = 0x0
> Kernel panic - not syncing: Kernel mode signal 4
>
> EIP: 0073:[<40388868>] CPU: 0 Not tainted ESP: 007b:bfa7f060 EFLAGS:
> 00000246
>     Not tainted
> EAX: ffffffda EBX: 00000007 ECX: 080a3190 EDX: 00000000
> ESI: 00000000 EDI: 00000000 EBP: bfa7fff8 DS: 007b ES: 007b
> 081fdbb4:  [<0806a327>] show_regs+0xc4/0xc9
> 081fdbe0:  [<08059a16>] panic_exit+0x25/0x3b
> 081fdbf4:  [<0808411b>] notifier_call_chain+0x27/0x4c
> 081fdc1c:  [<0808416e>] atomic_notifier_call_chain+0x15/0x17
> 081fdc2c:  [<081984b1>] panic+0x4c/0xd2
> 081fdc48:  [<080593cf>] relay_signal+0x31/0x6c
> 081fdc68:  [<080650a2>] sig_handler_common+0x61/0x70
> 081fdce0:  [<080651ca>] sig_handler+0x31/0x3d
> 081fdcec:  [<08065363>] handle_signal+0x4f/0x7d
> 081fdd0c:  [<08066a0b>] hard_handler+0xf/0x14
> 081fdd1c:  [<ffffe420>] _etext+0xf7e64328/0x0
>
> It appears to be very repeatable at the moment as the crashes are occurring
> within a few minutes of boot. I noticed a similar bug report in the archives
> for kernel version 2.6.24. Is this a known issue? If so, is there a
> work-around? What additional information should I supply?
>
> I can supply the text from the earlier kernel version if that might help.
> Any help is appreciated!
>
> Tony

------------------------------------------------------------------------------
Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA
-OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise
-Strategies to boost innovation and cut costs with open source participation
-Receive a $600 discount off the registration fee with the source code: SFAD
http://p.sf.net/sfu/XcvMzF8H
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel