Subject: Re: hardening pcmcia_cis.c against mapping failures
To: None <martin@rumolt.teuto.de>
From: John Kohl <jtk@kolvir.arlington.ma.us>
List: tech-kern
Date: 06/27/2000 17:03:43
>>>>> "Martin" == Martin Husemann <martin@rumolt.teuto.de> writes:

Martin> Hey!
>> On my ThinkPad 600E, using the default RBUS address for cardbus blows chunks
>> and leads to bogus data reads for the CIS data on a card.  That causes a
>> page fault when debugging is not enabled.

Martin> I'm seeing the same on a different ThinkPad, but only for some cards.
Martin> I've a PR open for this, with a "fix": it inserts a few delays, which
Martin> makes the data read OK.

Martin> Making it foolproof is another (good) thing. Does the card work with your
Martin> patch?

The card doesn't work, but neither does it cause a page fault.
I can get it working fine if I also adjust the rbus base offset (in
arch/i386/i386/rbus_machdep.c, I think it is) to 1GB instead of 2GB.

That's probably another option that should be listed in some config
file/hints for laptops....

-- 
==John Kohl <jtk@kolvir.arlington.ma.us>, <john_kohl@alum.mit.edu>
Home page: <http://people.ne.mediaone.net/jtk/>
Bicycling and Skiing to keep fit.