Re: kernel stack overflow profiler.. two very long oopses

Michael L. Galbraith (mikeg@mikeg.weiden.de)
Sun, 16 Feb 1997 06:04:38 +0100 (MET)


On Sat, 15 Feb 1997, Ingo Molnar wrote:

> > kernel stack overflow. Forcing Oops.
>
> i'm not an SCSI expert, but i've snipped out some imho interesting
> functions:
>
> > Trace: c02192b1 <do_sd_request+139/190>
> > Trace: c02192b1 <do_sd_request+139/190>
> > Trace: c02192b1 <do_sd_request+139/190>
> > Trace: c02192b1 <do_sd_request+139/190>
>
> damn. This means we are recursing 4 ways deep in the Adaptec driver? Isnt
> this a no-no?
>

Possibly because of tagged queueing enabled (wa guess)? This seems to work
fine.. except for vfat/dos file systems. I have copied >1.6 Gig ext2 fs
the same way with no illness (several times).

Also, ran into same stack corruption with tagged queueing disabled.

>
> i will try to trade my Adaptec for a BusLogic as soon as possible ...
>

I figured that this driver would _eventually_ get fixed since sooo many
people have 2940s (tried/ka-booom)... even whimpered politely to author.

but if I do...

Will I then run into that nasty sector mapping problem?

It _would_ be sorta nice to be able to backup again. God's probably tired of
hearing the same fervent prayers after every little crash. :))

-Mike