Possible bug: nasty interaction between framebuffer and DRI/DRM

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

Hi,

I seem to have discovered a nasty interaction between framebuffer and DRI/DRM 
in OpenGL applications. I can reproduce this problem in at least Return to 
Castle Wolfenstein (version 1.1b) and Tux Racer.

I am using a 32 MB ATI Rage Pro 128 on Debian woody (XFree86 4.1). If I run 
Tuxracer or RTCW, and wait for the game to load, everything is alright. 
However if I switch to a spare virtual terminal, and switch back to X, the 
fonts become illegible -- they just appear as coloured boxes with no 
outlines, making things unplayable.

The problem does not occur when using software based OpenGL rendering which is 
why I believe DRI/DRM is involved. This problem only occurs when framebuffer 
is enabled in the kernel, if framebuffer is disabled the problem disappears, 
so I believe framebuffer is also at fault.

Greatful for any help...

Chris Howells
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQE892SKF8Iu1zN5WiwRAgMbAJ0V5bWa85Lfsgu4FFtKeIj4c/YX8gCcCrn6
YU1lq0HuQpefLVGsDvFzN4E=
=qzHZ
-----END PGP SIGNATURE-----


_______________________________________________________________

Don't miss the 2002 Sprint PCS Application Developer's Conference
August 25-28 in Las Vegas -- http://devcon.sprintpcs.com/adp/index.cfm

Hi,

On Friday 31 May 2002 12:54 pm, Chris Howells wrote:

I seem to have discovered a nasty interaction between framebuffer and
DRI/DRM in OpenGL applications. I can reproduce this problem in at least
Return to Castle Wolfenstein (version 1.1b) and Tux Racer.

Meant to say this before...

I can reproduce this under both kernel 2.4.17 and 2.4.18.


_______________________________________________________________

Don't miss the 2002 Sprint PCS Application Developer's Conference
August 25-28 in Las Vegas -- http://devcon.sprintpcs.com/adp/index.cfm

On Fri, 31 May 2002, Chris Howells wrote:

I seem to have discovered a nasty interaction between framebuffer and DRI/DRM 
in OpenGL applications. I can reproduce this problem in at least Return to 
Castle Wolfenstein (version 1.1b) and Tux Racer.

I am using a 32 MB ATI Rage Pro 128 on Debian woody (XFree86 4.1). If I run 
Tuxracer or RTCW, and wait for the game to load, everything is alright. 
However if I switch to a spare virtual terminal, and switch back to X, the 
fonts become illegible -- they just appear as coloured boxes with no 
outlines, making things unplayable.

The problem does not occur when using software based OpenGL rendering which is 
why I believe DRI/DRM is involved. This problem only occurs when framebuffer 
is enabled in the kernel, if framebuffer is disabled the problem disappears, 
so I believe framebuffer is also at fault.

Do you use `option UseFBDev' in your XF86Config?

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds


_______________________________________________________________

Don't miss the 2002 Sprint PCS Application Developer's Conference
August 25-28 in Las Vegas -- http://devcon.sprintpcs.com/adp/index.cfm

Chris Howells wrote:

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

Hi,

I seem to have discovered a nasty interaction between framebuffer and DRI/DRM in OpenGL applications. I can reproduce this problem in at least Return to Castle Wolfenstein (version 1.1b) and Tux Racer.

I am using a 32 MB ATI Rage Pro 128 on Debian woody (XFree86 4.1). If I run Tuxracer or RTCW, and wait for the game to load, everything is alright. However if I switch to a spare virtual terminal, and switch back to X, the fonts become illegible -- they just appear as coloured boxes with no outlines, making things unplayable.

It sounds like the texture cache is getting wiped out, but the DRI drivers aren't realizing that it's happened.  This shouldn't be too hard to fix if someone has time.

Keith



_______________________________________________________________

Don't miss the 2002 Sprint PCS Application Developer's Conference
August 25-28 in Las Vegas -- http://devcon.sprintpcs.com/adp/index.cfm

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

Hi,

On Friday 31 May 2002 1:08 pm, Geert Uytterhoeven wrote:

Do you use `option UseFBDev' in your XF86Config?

No I'm not. Should I be?

Cheers

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQE898NRF8Iu1zN5WiwRAiU5AJ95WKTct4sFOXe+ruIMvlICGxyhFACggQvB
0RaUngCVTJQdFTS3H3ceb7c=
=/n6Q
-----END PGP SIGNATURE-----


_______________________________________________________________

Don't miss the 2002 Sprint PCS Application Developer's Conference
August 25-28 in Las Vegas -- http://devcon.sprintpcs.com/adp/index.cfm

On Fri, 31 May 2002, Chris Howells wrote:

On Friday 31 May 2002 1:08 pm, Geert Uytterhoeven wrote:

quoted

Do you use `option UseFBDev' in your XF86Config?

No I'm not. Should I be?

Yes. Without that option, the X server will change the graphics chip's
registers behind the back of aty128fb.

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds


_______________________________________________________________

Don't miss the 2002 Sprint PCS Application Developer's Conference
August 25-28 in Las Vegas -- http://devcon.sprintpcs.com/adp/index.cfm