[Dri-devel] Patch: Shared IRQ's for i810, plus less kernel oops

2001-06-14 Thread Sottek, Matthew J


 This patch adds in shared IRQ's. I added it in a manner that
goes along with the template code so it should only alter
the i810 driver.

  Plus, the i810 drm isn't working in XFree 4.1.0. I get
regular oopsen when shutting down X leaving me with a blank
console (if I'm lucky). With shared interrupts I usually get
a wedged machine. This patch removes the NULL deref's that
were ruining my day, but I doubt this is the real answer.
I probably just hid the problem better.
  Seems a lot of things get called with
dev->dev_private == NULL which gets dereferenced a lot.

Can someone who knows the template code better look into
this and perhaps come up with a real fix?

Thanks,
-Matt


 drm.diff


Re: [Dri-devel] X4.1.0 radeon driver instead of 2.4.x driver

2001-06-14 Thread Gareth Hughes

Ed Schernau wrote:
> 
> David Miller suggested I use the X-supplied driver.  How is this change made?
>  I tried compiling X4.1.0 from source, which failed with syntax errors.
> 
> Any points appreciated - is this something obvious?

Can you give an example of the syntax errors?  Something has gone very
wrong if it's failing like this...

-- Gareth

___
Dri-devel mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/dri-devel



[Dri-devel] X4.1.0 radeon driver instead of 2.4.x driver

2001-06-14 Thread Ed Schernau

David Miller suggested I use the X-supplied driver.  How is this change made? 
 I tried compiling X4.1.0 from source, which failed with syntax errors.

Any points appreciated - is this something obvious?

Ed

___
Dri-devel mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/dri-devel



[Dri-devel] Re: CVS Update: xc (branch: trunk) 4.1.0

2001-06-14 Thread Dieter Nützel

Hello,

UT show the same texture corruption running XFree86 DRI 4.1.0 (after David's 
update) with the tdfx driver (V5 5500, 1280x1024x24) as I reported some days 
ago for the trunk-branch.

Every other thing I've tested looks OK.
Even Xv (good work Alan).

But I have some Linux 2.4.5-ac14 shm problems, now.
Read my post here about it or at LKM.

Regards,
Dieter


___
Dri-devel mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/dri-devel



[Dri-devel] Interrupt sharing gone?

2001-06-14 Thread Sottek, Matthew J

The i810 driver used to have interrupt sharing before moving to the template
code, it appears to have lost this functionality since the change. This is a
real problem since a lot of the i810's and i815's have integrated lans that
are using the same interrupt as the graphics controller.

  As I recall Keith or Jeff added that for us after the issue was discovered
during
development. Is there a reason the templated code shouldn't allow interrupt
sharing?

  I think that this hasn't been merged into the kernel tree yet, when that
happens
there will probably be a lot of broken systems. I just noticed it myself
since I
haven't been tracking the drm in the DRI tree.

 -Matt



___
Dri-devel mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/dri-devel



[Dri-devel] I can't believe it. TDFX drivers nearly as fast as under Win2000

2001-06-14 Thread Catonga

Hello,

I just read this great review of XFree 4.1.0
on "the duke of url" Website:

http://www.thedukeofurl.org/reviews/misc/xfree86410/index.shtml


And what i saw was more than impressive!

The Voodoo 3 card is now nearly as fast as under Windows 2000
in Quake 3.


I never imagined that this could get ever be possible.

Thanks a lot to all that made this possible.


Best Regards
Oliver C.



___
Dri-devel mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/dri-devel



[Dri-devel] Possible other platform status page?

2001-06-14 Thread Jeffrey M Einhorn

I was wondering if the dri website had a status page for non-i386 linux
dri support like freebsd or alpha/linux.
thanks,
   jeff  


___
Dri-devel mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/dri-devel



Re: [Dri-devel] GL_DECAL and texture corruption fixes for MGA driver

2001-06-14 Thread Karl Lessard


Andy Isaacson wrote:

>On Wed, Jun 13, 2001 at 12:17:34PM -0400, Karl Lessard wrote:
>
>>Hi everybody,
>>
>>this patch fixes some problems in texturing in the mga driver.
>>
>>1) It fixes a texture corruption problem, as the mga driver did not use 
>>the right texture format in some case
>>  ( I've seen that when an application wants to store a texture in 
>>RGB5_A1, but it is actually stored in ARGB8 by Mesa).
>>
>>2) I've enabled again the hardware acceleration for GL_DECAL texture 
>>function, for both single and multi-texturing. Blending in single 
>>texturing needs to be emulated through multi-texturing, and problems 
>>with multi-texturing were probably relied to the bug above (as in Tribes2).
>>
>>That way, anybody who tries to play Tribes2 with a mga board will be 
>>impressed by the improvment made by this patch. Please, try it!
>>
>
>Wow.  This patch takes Tribes2 from "utterly unplayable at 0.5 FPS" to
>"at least I can look around and shoot at stuff" at something like 10 or
>20 FPS.  (G400 OEM 32MB, dual PIII 666.)  I'm still not sure it's up to
>snuff with the Windows version, but this is a huge improvement.
>

If you observe that some scenes a slower that others, make sure that 
you've disable the CVA extension before running the game by setting to 
"1" the option "disableExtCompiledVertexArray" in you ClientPrefs file. 
It seems to be a bottle-neck of the driver.

>
>
>Thanks, Karl!
>
No problem Andy! :)

Karl


___
Dri-devel mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/dri-devel



[Dri-devel] Re: DRI on ATi Xpert2000 working!!!

2001-06-14 Thread Michel Dänzer

David Dawes wrote:
> 
> On Thu, Jun 14, 2001 at 05:40:53PM +0200, Michel Dänzer wrote:
> >[EMAIL PROTECTED] wrote:
> >>
> >>  Well I have not tested Xpert2000, but r128 and radeons seems to work
> >> fine. The key to getting dri support in 4.1.0 is to recompile modules
> >> from os-support. The ones that come with linux kernel would not work.
> >
> >This means that the r128 DRM version in XFree86 should be bumped such that
> >it doesn't try to work with the one in the current 2.4 kernels.
> 
> It looks like the patchlevel was increased, but the minor (or
> major?) version should have been if they're not compatible.

The major version would have to be bumped if the 4.1.0 DRM doesn't work with
older XFree86 versions, otherwise only the minor one, right?

Someone will have to verify that.


-- 
Earthling Michel Dänzer (MrCooper)\   Debian GNU/Linux (powerpc) developer
CS student, Free Software enthusiast   \XFree86 and DRI project member

___
Dri-devel mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/dri-devel



[Dri-devel] Re: [Xpert]DRI on ATi Xpert2000 working!!!

2001-06-14 Thread David Dawes

On Thu, Jun 14, 2001 at 05:40:53PM +0200, Michel Dänzer wrote:
>[EMAIL PROTECTED] wrote:
>> 
>>  Well I have not tested Xpert2000, but r128 and radeons seems to work
>> fine. The key to getting dri support in 4.1.0 is to recompile modules from
>> os-support. The ones that come with linux kernel would not work.
>
>This means that the r128 DRM version in XFree86 should be bumped such that it
>doesn't try to work with the one in the current 2.4 kernels.

It looks like the patchlevel was increased, but the minor (or
major?) version should have been if they're not compatible.

BTW, here are the versions:

  Driver  XFree86 4.1.0 Linux 2.4.4 kernelXFree86 4.1.0 checks for
---
  gamma  1.0.0   1.0.0   1.0+.x
  i810   1.1.0   1.1.0   1.1+.x
  mga3.0.2   2.0.1   3.0+.x
  r128   2.1.6   2.1.2   2.1+.x
  radeon 1.1.1   1.0.0   1.1+.x
  tdfx   1.0.0   1.0.0   1.0+.x

Note: "N+" means values >= N.  The checks imply that drm modules with newer
minor version numbers will work with older XFree86 driver modules
(if the major version number hasn't changed), and that patchlevel changes
indicate changes that don't affect the user<->kernel interface.

David
--
David Dawes  Email: [EMAIL PROTECTED]
Founder/President, The XFree86 Project, Inc  Phone: +1 510 687 6857
http://www.xfree86.org/

___
Dri-devel mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/dri-devel



Re: [Dri-devel] problem w/ gl programs on rage128

2001-06-14 Thread Michel Dänzer

Daniel Rose wrote:

> I'm having some troubles with GL programs under DRI and XF 4.1.0 on my
> ATI Rage 128..
> What happens is that in certain GL apps, objects are missing (ie.
> transparent) and in other apps, the entire app turns a lovely shade of
> magenta or blue.. in all cases, you can see the actual intended program
> and/or objects for a VERY split second if you defocus them refocus the
> window. I've had this problem a couple of times, on different
> distributions, but with the same video card. It is running in
> accelerated mode, i can tell by looking at glxinfo, which outputs that
> it is indeed using the Mesa Rage128 driver..

Please try building the drm from the XFree86 4.1.0 source instead of using the
one from the Linux kernel. If you don't know how to do that, look at
http://www.xfree86.org/~alanh/ .


-- 
Earthling Michel Dänzer (MrCooper)\   Debian GNU/Linux (powerpc) developer
CS student, Free Software enthusiast   \XFree86 and DRI project member

___
Dri-devel mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/dri-devel



[Dri-devel] Re: [Xpert]DRI on ATi Xpert2000 working!!!

2001-06-14 Thread Mike Heald

On Thursday 14 June 2001 16:40, Michel Dänzer wrote:
> [EMAIL PROTECTED] wrote:
> >  Well I have not tested Xpert2000, but r128 and radeons seems to work
> > fine. The key to getting dri support in 4.1.0 is to recompile modules
> > from os-support. The ones that come with linux kernel would not work.
>
> This means that the r128 DRM version in XFree86 should be bumped such that
> it doesn't try to work with the one in the current 2.4 kernels.


Cool, now everything seems great! There's still a few things
that aren't 100%, there's some walls in Unreal whose edges
appear jagged when they should be striaight, and the scrolling
text in XMMS is flickery instead of smooth. I've tried
Option ``shadowfb'' but it doesnt help. Any suggestions?

Mike
p.s. thanks for all the help getting it running in the first place! :)

___
Dri-devel mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/dri-devel



[Dri-devel] Re: [Xpert]DRI on ATi Xpert2000 working!!!

2001-06-14 Thread Michel Dänzer

[EMAIL PROTECTED] wrote:
> 
>  Well I have not tested Xpert2000, but r128 and radeons seems to work
> fine. The key to getting dri support in 4.1.0 is to recompile modules from
> os-support. The ones that come with linux kernel would not work.

This means that the r128 DRM version in XFree86 should be bumped such that it
doesn't try to work with the one in the current 2.4 kernels.


-- 
Earthling Michel Dänzer (MrCooper)\   Debian GNU/Linux (powerpc) developer
CS student, Free Software enthusiast   \XFree86 and DRI project member

___
Dri-devel mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/dri-devel



Re: [Dri-devel] GL_DECAL and texture corruption fixes for MGA driver

2001-06-14 Thread Andy Isaacson

On Wed, Jun 13, 2001 at 12:17:34PM -0400, Karl Lessard wrote:
> Hi everybody,
> 
> this patch fixes some problems in texturing in the mga driver.
> 
> 1) It fixes a texture corruption problem, as the mga driver did not use 
> the right texture format in some case
>   ( I've seen that when an application wants to store a texture in 
> RGB5_A1, but it is actually stored in ARGB8 by Mesa).
> 
> 2) I've enabled again the hardware acceleration for GL_DECAL texture 
> function, for both single and multi-texturing. Blending in single 
> texturing needs to be emulated through multi-texturing, and problems 
> with multi-texturing were probably relied to the bug above (as in Tribes2).
> 
> That way, anybody who tries to play Tribes2 with a mga board will be 
> impressed by the improvment made by this patch. Please, try it!

Wow.  This patch takes Tribes2 from "utterly unplayable at 0.5 FPS" to
"at least I can look around and shoot at stuff" at something like 10 or
20 FPS.  (G400 OEM 32MB, dual PIII 666.)  I'm still not sure it's up to
snuff with the Windows version, but this is a huge improvement.

Thanks, Karl!

-andy

___
Dri-devel mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/dri-devel



Re: [Dri-devel] radeon.o kernel module version is 1.0.0 but version 1.1.x is needed

2001-06-14 Thread David S. Miller


root writes:
 > Questions:
 > What does the DRI version depend on?  I assumed a new X and a new kernel
 > would update this.
 > Is there a way to get a DRI version 1.1.x ?
 > 
 > Currently, I am unable to play any 3d games, this card is angering me.

You need to use the Radeon kernel module from XFree86-4.1.x not from
the current 2.4.x kernel as these two trees are not synchronized.

Later,
David S. Miller
[EMAIL PROTECTED]

___
Dri-devel mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/dri-devel



[Dri-devel] radeon.o kernel module version is 1.0.0 but version 1.1.x is needed

2001-06-14 Thread root

Hello all,  I hope this is the right forum for this.

I'm running X4.1.0 (from RPM) and kernel 2.4.5 (from kernel.org), my
Linux 2.4.5 system has agpgart and radeon compiled as modules.

When I start X, I get the message in the subject line.

Questions:
What does the DRI version depend on?  I assumed a new X and a new kernel
would update this.
Is there a way to get a DRI version 1.1.x ?

Currently, I am unable to play any 3d games, this card is angering me.

Any help appreciated,
Ed Schernau

___
Dri-devel mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/dri-devel