> Date: Tue, 19 Nov 1996 19:20:25 -0500 (EST)
> From: Ion Badulescu <ionut@moisil.wal.rhno.columbia.edu>
> To: Dan Merillat <Dan@merillat.org>
> Cc: Linux Kernel List <linux-kernel@vger.rutgers.edu>
> Subject: Re: Funky PCI bug.
>
> On Tue, 19 Nov 1996, Dan Merillat wrote:
>
> [snip]
>
> > Oh, and before anyone tells me to read include/linux/pci.h...
> > the 3940 works fine in most boards... and has been working with linux for
> > quite some time. It's just this one old MB that tickles the bug.
>
> There is no way to use a PCI board if the BIOS does not do its job. You
> can't even leave it in, as it will most likely conflict (I/O portwise)
> with the other PCI boards; the BIOS won't know about it and won't
> reconfigure it. As Linux relies entirely on the BIOS for PCI management...
> it can't really do anything to prevent such problems.
As I said in the message, there was NO conflict, UNTIL I read /proc/pci
So linux got the board working, but it had some bad pointers floating
around. Linux is tolerant of screwups with other hardware (cmd 640...)
and it should not have died so badly when I cat'd a file in /proc.
It worked FINE until then.,
--Dan