Subject: Re: problems (freezes, ide timeout) with -current/i386 _and_ MP
To: Markus W Kilbinger <kilbi@rad.rwth-aachen.de>
From: Frank van der Linden <fvdl@wasabisystems.com>
List: port-i386
Date: 11/30/2002 21:50:22
On Sat, Nov 30, 2002 at 09:02:29PM +0100, Markus W Kilbinger wrote:
> On two i386 dual boards (asus p2b-d and a7m266-d) I see sporadic
> system freezes if running a MP kernel with -current later than Nov.,
> 26.

Be sure to have an up to date i386/intr.c and include/intr.h, there
was a lockup problem related to serial interrupts that was fixed
on the 27th.

> Another problem I noticed is: cdrecord fails exactly and
> reproduceably during passing the 21st mb of the image file with:
> 
>   pciide0:0:0: lost interrupt
>         type: ata tc_bcount: 65536 tc_skip: 0
>   pciide0:0:0: bus-master DMA error: missing interrupt, status=0x61
>   pciide0:0:0: device timeout, c_bcount=65536, c_skip0

That's kind of strange. If this was a random interrupt loss problem,
I'd suspect the interrupt code, but being able to reproduce it
exactly at that point doesn't seem to confirm that at all.

Here are some things to do:

	* Have a completely up to date kernel
	* If you get lockups, try LOCKDEBUG (and have a serial console
	  attached, otherwise you can't see DDB)
	* Try a kernel without MULTIPROCESSOR, but with cpu* and ioapic*
	  at mainbus. This will use the ioapic for interrupts, but
	  only one CPU.

- Frank

-- 
Frank van der Linden                                    fvdl@wasabisystems.com
==============================================================================
Quality NetBSD Development, Support & Service.   http://www.wasabisystems.com/