-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hello,

On Jul 13, 2007, at 10:33, Masafumi Hotta wrote:

>> Hmm, weird.
>> Please run 'ofctl -p' and mail me the output
>
> I tried to run 'ofctl -p',but no 'ofctl' on 3.1.
> 4.99.22 kernel stops 'db>' ,so I can't run 'ofctl'.

ofctl works on 3.1, the ioctl()s it uses have been present in NetBSD  
for ages. Just build it from a 4.0 source tree. The version in -current  
uses proplib soo building it on 3.1 might be a little more of an  
effort.

> But I found 'dmesg' is available on 'db>',and noticed one thing.
> No 'ofb0' is found, 'macfb0' is detected.

Eh, macfb is a mac68k thing ;)

> Here is several lines of dmesg.
> machfb0 at pci0 dev 16 function 0: ATI Technologies Rage Mobility M3  
> (AGP) (rev.
>  0x02)
> machfb0: 64 MB aperture at 0x94000000, 16 KB registers at 0x90000000
> machfb0: 16384 KB (unknown type) 14.318 MHz, maximum RAMDAC clock 230  
> MHz
> machfb0: chip ID mismatch, 0xff7f != 0x4c46
>
> ----------------------------------------------------------------------- 
> -----------
> I am wondering that kernel without 'machfb0' works ?

It should.
Looks like what you have is based on a Rage XL which is known broken  
for now :/
I thought I had machfb explicitly not match that kind of hardware but  
apparently I'm wrong - I'll fix that. Unfortunately I don't have the  
hardware to debug this problem :(
So, just disable machfb in your kernel config and ofb should take over  
again.

have fun
Michael
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (Darwin)

iQEVAwUBRpec2MpnzkX8Yg2nAQI/2wf/TG0INEWCIuW8RAhQIRApHwOfPHLA6FOQ
Mtn4UvO93D/2LHDFLDgslIh9som5i+W99ZHaC0s5EXu4tipiusB4ZZsuJF03vk4m
tngtjW41iV3qSY3vNQjHBK/zc9y8OF24oGhUD3TR/eObJ9mUDI2YD20WDX5DgFjo
i9jAuf8BDUlW/q1ywQKGAr7STN+W28SbexpVHEKQNSO9v7UWS3gugDTm1oRWA48L
mg4YMkZdM8QRjelupcafY4/9gvJnAn7yHlFthiMSHvA4kZxLJidsnmPD5fzLC4Zv
K4PmJGMPQdBBOIOOgJY97VU1/jxYq7F0vuzJnOtdkJ9BjQiYiXoqeg==
=J5/3
-----END PGP SIGNATURE-----