Subject: Re: pciide lost interrupt - losing access to the file system
To: Johan Ihren <>
From: Manuel Bouyer <>
List: current-users
Date: 10/16/1999 19:26:56
On Fri, Oct 15, 1999 at 09:31:38AM +0200, Johan Ihren wrote:
> I'm seeing the same thing, except in my case it is while trying to
> install 1.4 onto an old 486-machine (no PCI). I.e. what I see is
> basically:
> wdc0:0:0: device timeout
>         type: ata
>         c_bcount: 0
>         c_skip: 0
> over and over when booting from the install floppy. First I thought it
> must be broken h/w, but since other OSes (not to mention by name)
> install nicely it seems like a NetBSD problem.

It's not the same problem: it's not surprising to have some lost interrupt (
or other problems) with some combination of IDE hardware, and the code should
be able to deal with this (the problem Rick reported let me believe there is
a bug in the recovery code).
In your case it seem that *no* interrupts are never generated, maybe for
one specific command (because c_bcount & c_skip == 0, this is not a data
transfer which trigger the problem). I suspect here that maybe your drive is
not fully conformant, or that you have one of these "intelligent" IDE
controllers which does fabcy things ...

Manuel Bouyer, LIP6, Universite Paris VI.