Re: slaballocation panic on boot

[Galen Sampson <galen_sampson@xxxxxxxxx>]

I would like to bring good news but unfortunately that is not the case.  I have
updated to sources with todays (9/25) fixes.  The machine running 4.8-p10 is
rebooted into the dragonfly kernel so make installworld can be performed.  
'make installworld 2>&1 | tee /devel/dragonfly/buildlog/installworld.log' does
not get written, but on the console only the first shell script 'for _x in [
cp....' is printed.  The system always fails here.  Sometimes with ddb,
sometimes without.  The panics I have seen have been double faults.  The core
dumps I have collected yield useless information (see attached).

Notes:
kernel config is attacted (no slaballoc, no memap)
useless crashdump attached

regards,
Galen

--- Matthew Dillon <dillon@xxxxxxxxxxxxxxxxxxxx> wrote:
>     Ok, it's definitely DargonFly, then.  I have committed a number of fixes
>     to the slab allocator, could you retest?
> 
> 						-Matt
> 
> :This is a follow up.  I have wiped the machine in question clean and started
> :from freebsd 4.8.  The machine works extremely well in 4.8.  I have updated
> the
> :machine to freebsd 4.8-p10, and the machine continues to work fine.  If I
> :install a dragonfly kernel and reboot to do 'make installworld' the machine
> :will hard reboot immediately and the world will not be installed (no panic
> :either, even with ddb).  This is both with and without the slab allocator. 
> The
> :dragonfly sources are from 9/24.
> :
> :Has anyone tried running dragonfly on a pentium that does not have mmx?
> :
> :I'm waiting for matt's patches to the slab allocator at which point I will
> test
> :again.
> :
> :regards,
> :Galen

__________________________________
Do you Yahoo!?
The New Yahoo! Shopping - with improved product search
http://shopping.yahoo.com

Attachment: FUBAR1
Description: FUBAR1

Attachment: debug.script
Description: debug.script