Re: Reboot while booting with new per-CPU allocator

From: Stephane E. Potvin <sepotvin_at_videotron.ca>
Date: Thu, 16 Jun 2005 10:55:14 -0400
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Robert Watson wrote:
> 
> On Thu, 16 Jun 2005, Stephane E. Potvin wrote:
> 
>>  Kernel malloc layers malloc_type allocation over one of two underlying
>>  allocators: a set of power-of-two UMA zones for small allocations,
>> and the
>> ...
>>  modifications to vmstat in order to restore "vmstat -m" on core dumps
>> will
>>  follow shortly.
>>
>> I get spontaneous reboots while the kernel is loading after the kernel
>> finds the APICs while booting verbose:
> 
> 
> It sounds like you've one a binary search to track this down and it's
> definitely that commit that did it?  I.e., if you specifically back out
> the associated changes to kern_malloc.c and malloc.h locally with a
> curret kernel, all is well?
> 
> Have you tried running with kern_malloc.c:1.141 (Jun 10) from Joseph
> Koshy, which corrects a bug in the deregistering of malloc types?
> 
> It looks like you're running with a serial console, but if not, could
> you do so and make sure there are no last second printfs that get eaten
> by the reboot clearing the video console?
> 

I did a binary search to find the offending commit. Reverting the changes
to kern_malloc.c and malloc.h locally did fix the problem. That's what I'm
currently using to type this message. I also had to revert the changes to
vmstat to make the world compile.

Using head sources from yesterday (Jun 15) exhibit the same problem
unfortunately.

The computer that has this problem is a Dell XPS laptop which, being quite
recent, doesn't have any serial port. The boot output I gave is a mix of
interpolating the beginning from a working image (with only the
kern_malloc.c and malloc.h differing) and getting the last part with a
digital camera. I guessed it wouldn't cause a problem as the kern_malloc.c
diff didn't seems to print anything during the boot. There might be some
last second printfs going to the console before the reboot but in that
case they are much too fast to be seen even to the naked eyes.

Steph
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (FreeBSD)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFCsZLSmdOXtTCX/nsRAtG8AKDc858m6/64Iv4aKs90mZysXulc+ACfQuQo
Tp0ZxB6AECs15V8Fz87lR5c=
=pLjb
-----END PGP SIGNATURE-----
Received on Thu Jun 16 2005 - 12:56:20 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:36 UTC