braindead radeon.h

[Michael Lorenz]

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

Hello,

On Jul 5, 2006, at 19:13, Dave Airlie wrote:

>> I just found that the radeon driver in xorg 7.0 ignores the MMIO BAR's
>> actual size and always tries to mmap 0x80000 bytes ( a value #defined
>> in radeon.h ) - even with chips that have only 0x10000 there like my
>> M9+.
>> This fails on operating systems that actually pay attention what users
>> can mmap - like NetBSD when using ttyE* and probably anything using
>> /dev/xf86 - which allows to mmap only resources that actually belong  
>> to
>> the graphics chip.
>> Has this been fixed in a later release?
>
> Yes I think I fixed it in
> http://gitweb.freedesktop.org/?p=xorg-driver-xf86-video-ati;a=commit; 
> h=85e2ee6de6fd347aad66785ec7119f92013cd994
>
> which is in 6.6.1 for X.org 7.1.

Thanks, I'll have a look.

> I've just pushed a bunch of fixes to the X.org 7.0 compatible branch
> of ATI (ati-1-0-branch) in git. with that fix in there. I'm not sure
> I'll make a release from that branch but I could be persuaded if
> enough 7.0 users ask.

It's broken on more or less any OS / hardware combination that doesn't  
map PCI resources through /dev/mem so fixing it in the 7.0 branch would  
probably be a good idea.

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

iQEVAwUBRKxLt8pnzkX8Yg2nAQK2Qwf/VeSj0gTBYFK2V7lKl5jOhv+dkMvkNtGG
6v/6322TedVBSiqQ/Y//jZJR3pfFLoU/7VtGi1JzNNh8YVLiiZF6QAj/akq1DoIV
/jp+aKh6CfdecKZdjcErHdTAzTmXUmWMD9qyqyrulx45755dL2Aw6TFtrnBHq8hU
5qELmpWkaf5tBhR0goUO3wOm4XqGi4w3rt0Qg3fhUvFJ7TaA9tYfxJCQ4x0UWpqU
Zd3T2r0IDAnO33NQduUs14A+jqBe9tGIwu9wHuVVqbkwHqvYoSz9ZgZ5vYkmKHld
PH+3vilcEnlr8Q8YyBRgRdYT5tiO6tLpnYNG8GY+2e7QvPdY/b5xXA==
=M7kr
-----END PGP SIGNATURE-----