Re: pkg/35416: agp driver hangs on ALi/ULi M1689 chipset

[Martin Husemann <martin%duskware.de@localhost>]

The following reply was made to PR kern/35416; it has been noted by GNATS.

From: Martin Husemann <martin%duskware.de@localhost>
To: Gary Duzan <gary%duzan.org@localhost>
Cc: gnats-bugs%NetBSD.org@localhost
Subject: Re: pkg/35416: agp driver hangs on ALi/ULi M1689 chipset
Date: Sat, 5 Apr 2008 13:42:54 +0200

 On Sat, Apr 05, 2008 at 07:23:40AM -0400, Gary Duzan wrote:
 >    I believe so. However, the problem isn't the size, it is that
 > the wrong driver attaches, and I assume the aperture size is being
 > pulled from the wrong place.
 
 The check for zero size should be added there, however the right fix is
 to add an agp_match function in agp.c:
 
 #if NAGP_ALI > 0
         { PCI_VENDOR_ALI,       -1,
           NULL,                 agp_ali_attach },
 #endif
 
 change the NULL to agp_ali_match and provide a
 
   int agp_ali_match(const struct pci_attach_args *)
 
 function. Make it return 0 for your chipset, but 1 by default. When that
 works, try if agp_amd works for your chipset, and if it does, create a
 second entry for it with an inverse matching function.
 
 Martin