Subject: Re: pciide lost interrupt
To: Rick Byers <firstname.lastname@example.org>
From: Manuel Bouyer <email@example.com>
Date: 11/28/1999 22:56:38
On Sun, Nov 28, 1999 at 02:42:25PM -0500, Rick Byers wrote:
> I'm still having problems with "lost interrupt" messages. Maybe I'm just
> unlucky but it seems to be hapenning more frequently now that I've
> upgrated to 1.4P. With sources from Nov 23, I get errors like this
> frequently (when doing constant disk access):
> Nov 28 13:49:52 Rick /netbsd: pciide0:0:1: lost interrupt
> Nov 28 13:49:53 Rick /netbsd: type: ata
> Nov 28 13:49:53 Rick /netbsd: c_bcount: 65536
> Nov 28 13:49:53 Rick /netbsd: c_skip: 0
> Nov 28 13:49:53 Rick /netbsd: pciide0:0:1: Bus-Master DMA error: missing
> interrupt, status=0x61
> Nov 28 13:49:53 Rick /netbsd: wd1b: DMA error writing fsbn 6544 of
> 6544-6671 (wd1 bn 54928; cn 54 tn 7 sn 55), retrying
> Nov 28 13:49:53 Rick /netbsd: wd1: soft error (corrected)
> Nov 28 13:49:55 Rick /netbsd: usl_detachtimeout
> When I get errors like this, the system freezes up for about 5-10 seconds,
> but I no longer loose access to the disk completely. Any ideas what might
> be causing this? Is there anything I can do to help track it down?
Status 0x61 means that the bus-master engine is still active. This means
the transfert is not done.
I would suspect some electrical problem on the IDE bus which makes the
data transfer between HD and controller hang (some kind of deadlock).
Did you try another cable ? Also make sure there is something at both end
of the cable, and that the controller is really at one end.
If you're using DMA mode 2 you can try to force the kernel to DMA mode 1
with flags in the kernel config file ('man 4 wd' for details, it's also in
the GENERIC config file). The same is true for Ultra-DMA (i.e. try UDMA mode 1
if you're using UDMA mode 2).
Manuel Bouyer, LIP6, Universite Paris VI. Manuel.Bouyer@lip6.fr