Re: Official ATI Driver : problem build fglrx module

2005-07-08 Thread Alexander Nagel
Hi,
check http://www.rage3d.com/board/forumdisplay.php?f=61
there's a patch somehere for 2.6.12
greets
Alex


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Re: Official ATI Driver : problem build fglrx module

2005-07-08 Thread GOMBAS Gabor
On Fri, Jul 08, 2005 at 10:17:53AM -0400, Lennart Sorensen wrote:

> Well I am sure my boss would love if I could make the ati driver work on
> his Dell X610 which has some quite new ATI chip.  I even told him to
> avoid the ones with an ATI chip in it.

Laptops are a different thing. ATI tries to push driver support to the
laptop makers (which has _some_ justification since ATI has no control
over how its chip is being integrated, and laptop makers are known to do
really weird things sometimes). Unfortunately laptop makers do not want
to publish Linux drivers themselves.

You can say the situation is improving since some years ago the ATI
driver supported only the original "Built by ATI" cards and not the
3rd-party "Powered by ATI" ones (you had to hack the driver to make it
work with a 3rd-party card).

Gabor

-- 
 -
 MTA SZTAKI Computer and Automation Research Institute
Hungarian Academy of Sciences
 -


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Re: Official ATI Driver : problem build fglrx module

2005-07-08 Thread Lennart Sorensen
On Fri, Jul 08, 2005 at 04:02:19PM +0200, GOMBAS Gabor wrote:
> It works for me using Flavio Stanchina's packaging. For 2.6.12 you have
> to edit the kernel module source and replace the dev->slot_name
> references with pci_name(dev) and you also have to fix the
> inter_module_get() thing (I just commented it out as it is not needed on
> my setup).

Well I am sure my boss would love if I could make the ati driver work on
his Dell X610 which has some quite new ATI chip.  I even told him to
avoid the ones with an ATI chip in it.

> NVidia and Vmware also have regular problems with new kernels when the
> internal interfaces change.

That is certainly true, and is true of all binary only drivers (and
actually any driver not natively included with the kernel).

> Actually I heard that the _design_ of the ATI driver is much better than
> NVidia because they try to use some of the existing kernel
> infrastructure instead of reinventing it completely. Of course the
> actual code may be worse.

The nvidia driver is WAY too big. :)  My experience over the years with
ATI is that I love the hardware in general, and I hate their driver
bugs.  Maybe some day I can have a bug free ATI driver, but it's not
looking too promising.  One without any major bugs would be nice as a
start, but I am not holding my breath.  Only major bug I have seen/heard
of so far with nvidia had to do with xvideo on multihead setups, and
well I don't have a multihead setup.

Len Sorensen


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Re: Official ATI Driver : problem build fglrx module

2005-07-08 Thread GOMBAS Gabor
On Thu, Jul 07, 2005 at 05:01:37PM -0400, Lennart Sorensen wrote:

> Also the last I read a about a week or two ago was that the ati driver
> does not work with 2.6.12 at this time, so unless you have some other
> patches to apply you aren't likely to get it to work.

It works for me using Flavio Stanchina's packaging. For 2.6.12 you have
to edit the kernel module source and replace the dev->slot_name
references with pci_name(dev) and you also have to fix the
inter_module_get() thing (I just commented it out as it is not needed on
my setup).

> Of course since you already have the ATI its too late to recomend
> staying away from ati and their lousy drivers in general.

NVidia and Vmware also have regular problems with new kernels when the
internal interfaces change.

Actually I heard that the _design_ of the ATI driver is much better than
NVidia because they try to use some of the existing kernel
infrastructure instead of reinventing it completely. Of course the
actual code may be worse.

Gabor

-- 
 -
 MTA SZTAKI Computer and Automation Research Institute
Hungarian Academy of Sciences
 -


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Re: Official ATI Driver : problem build fglrx module

2005-07-07 Thread Javier Kohen
Hi Sylvain,

El jue, 07-07-2005 a las 22:52 +0200, Sylvain Archenault escribió:
> -BEGIN PGP SIGNED MESSAGE-
> Hash: SHA1
> 
> Hello everybody,
> 
> I would like to know if there is somebody who have succeeded to compile
> the module. I download the driver on ati official website :
> fglrx64_4_3_0-8.14.13-1.x86_64.rpm

Yes, I have. I installed Flavio Stanchina's Debian package from source
(http://xoomer.virgilio.it/flavio.stanchina/debian/fglrx-installer.html), 
followed all the instructions until the point where you build the module, then 
ran "m-a autoinstall fglrx" and stopped it (Ctrl-S) right after it finished 
uncompressing the archive. While this process was on hold I applied some of the 
patches from this thread (1) (http://ati.cchtml.com/show_bug.cgi?id=136, which 
seems to be down momentarily), then resumed the m-a process. The resulting 
module works with the kernel's AGPGART using the same setup that I had for 
kernel 2.6.11.

1) Look at one of my last comments in the ATI thread for a list of what
patches to apply. Don't apply the "integrated" patch, as it's missing
the important part that makes it work with the kernel's AGPGART.

Maybe somebody can recommend a better way to handle the
stopping/resuming of module-assistant. I couldn't find an option to stop
it from unpacking the sources every time and I was too lazy to patch the
tarball directly.

> I read that the problem may be due to gcc, actually i use gcc-4.0, and i
> compile a kernel with this version of gcc, So i think i have to keep
> this version.

Try gcc-4.0 with the previous patches, but if it doesn't work then you
might have to switch back to 3.4 or 3.3. I'm using 3.3 personally
because I was lazy to have make-kpkg use gcc-3.4.

Greetings,
-- 
Javier Kohen <[EMAIL PROTECTED]>
ICQ: blashyrkh #2361802
Jabber: [EMAIL PROTECTED]


signature.asc
Description: This is a digitally signed message part


Re: Official ATI Driver : problem build fglrx module

2005-07-07 Thread Markus Boas
Am Donnerstag 07 Juli 2005 23:00 schrieb Basile STARYNKEVITCH:
> Le Thu, Jul 07, 2005 at 10:52:49PM +0200, Sylvain Archenault écrivait/wrote:
> > I would like to know if there is somebody who have succeeded to compile
> > the module. I download the driver on ati official website :
> > fglrx64_4_3_0-8.14.13-1.x86_64.rpm
> >
> > When i tun make.sh in /lib/modules/fglrx/build_mod I've got this :
> >
> >
> >   CC [M]  /lib/modules/fglrx/build_mod/2.6.x/nvidia-agp.o
> > /lib/modules/fglrx/build_mod/2.6.x/nvidia-agp.c:57: error: static
> > declaration of '__fgl_agp_try_unsupported' follows non-static declaration
> > /lib/modules/fglrx/build_mod/2.6.x/agp_backend.h:92: error: previous
> > declaration of '__fgl_agp_try_unsupported' was here
> > make[2]: *** [/lib/modules/fglrx/build_mod/2.6.x/nvidia-agp.o] Erreur 1
> > make[1]: *** [_module_/lib/modules/fglrx/build_mod/2.6.x] Erreur 2
> > make[1]: quittant le répertoire « /usr/src/linux-2.6.12.2 »
>
> I've also got similar problems, with a 2.6.13-rc2 kernel.
>
> I've also got further compilation errors. In some file[s] I replaced
> pci->slot_name with pci->smart_name (details may be wrong, this is
> from my human memory).
>
> Out of curiosity, what is the chipset Sylvain is compiling for? Mine
> is described in message
> http://lists.debian.org/debian-amd64/2005/06/msg00788.html
>
> And I don't succeed in running X (either Xorg -latest CVS- or Xfree
> 34.2 (from Debian/Sid/AMD64)) in 1280x800 resolution. I can give up 3D
> acceleration...
>
> Regards.
>
Hi
the prob with /2.6.x/nvidia-agp have be done with a patch.
At the end I have a fglx.ko but I can't load it.

Ryven



Re: Official ATI Driver : problem build fglrx module

2005-07-07 Thread Lennart Sorensen
On Thu, Jul 07, 2005 at 10:52:49PM +0200, Sylvain Archenault wrote:
> -BEGIN PGP SIGNED MESSAGE-
> Hash: SHA1
> 
> Hello everybody,
> 
> I would like to know if there is somebody who have succeeded to compile
> the module. I download the driver on ati official website :
> fglrx64_4_3_0-8.14.13-1.x86_64.rpm
> 
> When i tun make.sh in /lib/modules/fglrx/build_mod I've got this :
> 
> 
> ATI module generator V 2.0
> ==
> initializing...
> cleaning...
> patching 'highmem.h'...
> assuming new VMA API since we do have kernel 2.6.x...
> doing Makefile based build for kernel 2.6.x and higher
> make -C /lib/modules/2.6.12.2/build
> SUBDIRS=/lib/modules/fglrx/build_mod/2.6.x modules
> make[1]: entrant dans le r?pertoire ? /usr/src/linux-2.6.12.2 ?
>   CC [M]  /lib/modules/fglrx/build_mod/2.6.x/agp3.o
>   CC [M]  /lib/modules/fglrx/build_mod/2.6.x/nvidia-agp.o
> /lib/modules/fglrx/build_mod/2.6.x/nvidia-agp.c:57: error: static
> declaration of '__fgl_agp_try_unsupported' follows non-static declaration
> /lib/modules/fglrx/build_mod/2.6.x/agp_backend.h:92: error: previous
> declaration of '__fgl_agp_try_unsupported' was here
> make[2]: *** [/lib/modules/fglrx/build_mod/2.6.x/nvidia-agp.o] Erreur 1
> make[1]: *** [_module_/lib/modules/fglrx/build_mod/2.6.x] Erreur 2
> make[1]: quittant le r?pertoire ? /usr/src/linux-2.6.12.2 ?
> make: *** [kmod_build] Erreur 2
> build failed with return value 2
> 
> 
> I read that the problem may be due to gcc, actually i use gcc-4.0, and i
> compile a kernel with this version of gcc, So i think i have to keep
> this version.
> 
> I've just noticed that the translated page of flavio page was out to
> date, so i will have a look to this new page tomorrow, maybe this will
> help me to figure it out.

First of all I would recomend looking through the archive or searching
google and finding where someone made debian packages for building the
ati driver instead.  Trying to work with the rpm is nuts.

Also the last I read a about a week or two ago was that the ati driver
does not work with 2.6.12 at this time, so unless you have some other
patches to apply you aren't likely to get it to work.

Of course since you already have the ATI its too late to recomend
staying away from ati and their lousy drivers in general.

Len Sorensen


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Re: Official ATI Driver : problem build fglrx module

2005-07-07 Thread Basile STARYNKEVITCH
Le Thu, Jul 07, 2005 at 10:52:49PM +0200, Sylvain Archenault écrivait/wrote:
> 
> I would like to know if there is somebody who have succeeded to compile
> the module. I download the driver on ati official website :
> fglrx64_4_3_0-8.14.13-1.x86_64.rpm
> 
> When i tun make.sh in /lib/modules/fglrx/build_mod I've got this :
> 
> 
>   CC [M]  /lib/modules/fglrx/build_mod/2.6.x/nvidia-agp.o
> /lib/modules/fglrx/build_mod/2.6.x/nvidia-agp.c:57: error: static
> declaration of '__fgl_agp_try_unsupported' follows non-static declaration
> /lib/modules/fglrx/build_mod/2.6.x/agp_backend.h:92: error: previous
> declaration of '__fgl_agp_try_unsupported' was here
> make[2]: *** [/lib/modules/fglrx/build_mod/2.6.x/nvidia-agp.o] Erreur 1
> make[1]: *** [_module_/lib/modules/fglrx/build_mod/2.6.x] Erreur 2
> make[1]: quittant le répertoire « /usr/src/linux-2.6.12.2 »

I've also got similar problems, with a 2.6.13-rc2 kernel.

I've also got further compilation errors. In some file[s] I replaced
pci->slot_name with pci->smart_name (details may be wrong, this is
from my human memory).

Out of curiosity, what is the chipset Sylvain is compiling for? Mine
is described in message
http://lists.debian.org/debian-amd64/2005/06/msg00788.html

And I don't succeed in running X (either Xorg -latest CVS- or Xfree
34.2 (from Debian/Sid/AMD64)) in 1280x800 resolution. I can give up 3D
acceleration...

Regards.


-- 
Basile STARYNKEVITCH http://starynkevitch.net/Basile/ 
email: basile(at)starynkevitch(dot)net 
8, rue de la Faïencerie, 92340 Bourg La Reine, France


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Official ATI Driver : problem build fglrx module

2005-07-07 Thread Sylvain Archenault
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1

Hello everybody,

I would like to know if there is somebody who have succeeded to compile
the module. I download the driver on ati official website :
fglrx64_4_3_0-8.14.13-1.x86_64.rpm

When i tun make.sh in /lib/modules/fglrx/build_mod I've got this :


ATI module generator V 2.0
==
initializing...
cleaning...
patching 'highmem.h'...
assuming new VMA API since we do have kernel 2.6.x...
doing Makefile based build for kernel 2.6.x and higher
make -C /lib/modules/2.6.12.2/build
SUBDIRS=/lib/modules/fglrx/build_mod/2.6.x modules
make[1]: entrant dans le répertoire « /usr/src/linux-2.6.12.2 »
  CC [M]  /lib/modules/fglrx/build_mod/2.6.x/agp3.o
  CC [M]  /lib/modules/fglrx/build_mod/2.6.x/nvidia-agp.o
/lib/modules/fglrx/build_mod/2.6.x/nvidia-agp.c:57: error: static
declaration of '__fgl_agp_try_unsupported' follows non-static declaration
/lib/modules/fglrx/build_mod/2.6.x/agp_backend.h:92: error: previous
declaration of '__fgl_agp_try_unsupported' was here
make[2]: *** [/lib/modules/fglrx/build_mod/2.6.x/nvidia-agp.o] Erreur 1
make[1]: *** [_module_/lib/modules/fglrx/build_mod/2.6.x] Erreur 2
make[1]: quittant le répertoire « /usr/src/linux-2.6.12.2 »
make: *** [kmod_build] Erreur 2
build failed with return value 2


I read that the problem may be due to gcc, actually i use gcc-4.0, and i
compile a kernel with this version of gcc, So i think i have to keep
this version.

I've just noticed that the translated page of flavio page was out to
date, so i will have a look to this new page tomorrow, maybe this will
help me to figure it out.


-BEGIN PGP SIGNATURE-
Version: GnuPG v1.4.1 (GNU/Linux)

iD8DBQFCzZYh3M6ekcMWoxURAkNBAKCK2wSv+o7SouTRlueY9bpPX1TrEQCaAz4C
P8/onoKF/QxiiEAcvxcKprY=
=bGDx
-END PGP SIGNATURE-


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]