Mailing List Archive: Re: Oops with latest CVS

Re: Oops with latest CVS

Aug 13, 2001, 9:53 AM

Post #1 of 3 (650 views)

Quick update on my problem:

I just tried with a 2.4.8 kernel (only added Dell patches for PERC
RAID controller), and I get the same behavior: second configured
machine crashes as soon as it tries to talk to first ...

Any hint ?

TIA,
-- pw

--
______________________________________________________________________
Pierre Wendling Email: pwendling@example.com
Sopra Group, DMI/SC, Annecy, France. Tel: (33) 450 33 3350
#include <std/disclaimer.h> Fax: (33) 450 33 3078

Re: Re: Oops with latest CVS [ In reply to ]

philipp.reisner at example

Aug 14, 2001, 12:09 PM

Post #2 of 3 (637 views)

Permalink

Yes. I know that it crashes. -- At least the user mode linux boxes
of my "test cluster" are also crashing. I am working on it.

-philipp

* Pierre Wendling <pwendling@example.com> [010813 18:53]:
>
> Quick update on my problem:
>
> I just tried with a 2.4.8 kernel (only added Dell patches for PERC
> RAID controller), and I get the same behavior: second configured
> machine crashes as soon as it tries to talk to first ...
>
> Any hint ?
>
> TIA,
> -- pw
>
> --
> ______________________________________________________________________
> Pierre Wendling Email: pwendling@example.com
> Sopra Group, DMI/SC, Annecy, France. Tel: (33) 450 33 3350
> #include <std/disclaimer.h> Fax: (33) 450 33 3078
>
>
>
> _______________________________________________
> DRBD-devel mailing list
> DRBD-devel@example.com
> http://lists.sourceforge.net/lists/listinfo/drbd-devel
>

Re: Re: Oops with latest CVS [ In reply to ]

philipp.reisner at example

Aug 14, 2001, 1:36 PM

Post #3 of 3 (651 views)

Permalink

Today I had the patience to track the bug in drbd_req-2.4.c down.

Shame on me it turned out to be:
nbh = (struct buffer_head*)((char*)req)+sizeof(drbd_request_t);
vs.
nbh = (struct buffer_head*)(((char*)req)+sizeof(drbd_request_t));

BTW, I still have the vmalloc/vfree in my code basis. I do not have
problems with it.

-Philipp

* Pierre Wendling <pwendling@example.com> [010814 12:41]:
> >>>>> "JD" == dolze <dolze@example.com> writes:
> JD> change in drbd_main.c the occurence of vmalloc(xx) into
> JD> kmalloc(xx, GFP_DRBD) and vfree into kfree. After recompile it
> JD> should work - but don't expect wonders.
>
> I tried this on a 2.4.8 kernel with no luck. The second machine still
> crashes.
>
> I have been trying to get a crash dump using lkcd (SGI opensource
> project), but unfortunately, the kernel hit another problem when taking
> the dump (Scheduling in interrupt, kernel BUG in sched.c:706). I don't
> have too much time to track down this one.
>
> Does anyone know a working crash dump patch/tool that is available on
> the recent 2.4.x kernels ?
>
> TIA
> --pw
>
> ______________________________________________________________________
> Pierre Wendling Email: pwendling@example.com
> Sopra Group, DMI/SC, Annecy, France. Tel: (33) 450 33 3350
> #include <std/disclaimer.h> Fax: (33) 450 33 3078
>
>
> _______________________________________________
> DRBD-devel mailing list
> DRBD-devel@example.com
> http://lists.sourceforge.net/lists/listinfo/drbd-devel
>