ATI 3.9.0 drivers and Suse 9.1
Hi, Me and others have problems with the new 3.9.0 driver. If you can check the threads at rage3d forums you might be able to help us. First even though the drivers compile correcly when you modprobe them you get the following error: fglrx: no version for "struct_module" found: kernel tainted. fglrx: module license 'Proprietary. (C) 2002 - ATI Technologies, Starnberg, GERMANY' taints kernel. [fglrx] Maximum main memory to use for locked dma buffers: 681 MBytes. Unable to handle kernel NULL pointer dereference at virtual address 00000000 printing eip: f2986d58 *pde = 00000000 Oops: 0002 [#1] It must be something with the regparm, which the drivers clame to support and in order to make the work you must apply a custom patch from here: http://www.rage3d.com/board/showthread.php?s=&threadid=33761546 Even though the drivers load with this patch, every time you run a GL app you get errors like these: kernel: Badness in do_no_page at mm/memory.c:1455 or kernel: Badness in copy_page_range at mm/memory.c:331 This seems to happen mostly is Suse 9.1 kernels and not to other distributions. What do you think is the problem? Thanks
Am Montag, 31. Mai 2004 20:06 schrieb George Billios:
Hi, Me and others have problems with the new 3.9.0 driver. If you can check the threads at rage3d forums you might be able to help us. First even though the drivers compile correcly when you modprobe them you get the following error:
My request to SUSE inst support for my 9.1-INst. related to RAdeon drivers was answered with a link to the 3.7.6-version of ATI-Driver. So I think, they know about problems with the 3.9er. Hartmut
Hi I was able to reproduce this problem with the 3.9.0 drivers from ATI. After reverting the regparm changes added by ATI in firegl_public.h and adding the regparm patches I used for 3.7.6 the driver works fine for me. The warning messages kernel: Badness in do_no_page at mm/memory.c:1455 kernel: Badness in copy_page_range at mm/memory.c:331 are still active, but seem not to be fatal for a working driver. I've uploaded new RPMs to ftp://ftp.suse.com/pub/suse/i386/supplementary/X/XFree86/ATI/suse91/fglrx/3.9.0 Hope this helps. Stefan On Mon, May 31, 2004 at 07:06:44PM +0100, George Billios wrote:
Hi, Me and others have problems with the new 3.9.0 driver. If you can check the threads at rage3d forums you might be able to help us. First even though the drivers compile correcly when you modprobe them you get the following error:
fglrx: no version for "struct_module" found: kernel tainted. fglrx: module license 'Proprietary. (C) 2002 - ATI Technologies, Starnberg, GERMANY' taints kernel. [fglrx] Maximum main memory to use for locked dma buffers: 681 MBytes. Unable to handle kernel NULL pointer dereference at virtual address 00000000 printing eip: f2986d58 *pde = 00000000 Oops: 0002 [#1]
It must be something with the regparm, which the drivers clame to support and in order to make the work you must apply a custom patch from here: http://www.rage3d.com/board/showthread.php?s=&threadid=33761546
Even though the drivers load with this patch, every time you run a GL app you get errors like these: kernel: Badness in do_no_page at mm/memory.c:1455 or kernel: Badness in copy_page_range at mm/memory.c:331
This seems to happen mostly is Suse 9.1 kernels and not to other distributions. What do you think is the problem?
Public Key available ---------------------------------------------------- Stefan Dirsch (Res. & Dev.) SUSE LINUX AG Tel: 0911-740 53 0 Maxfeldstrasse 5 FAX: 0911-740 53 479 D-90409 Nürnberg http://www.suse.de Germany ----------------------------------------------------
Nice :) I wish I could try them but yesterday my pc locked again because of the fglrx driver and I'm back to radeon driver again. It is very funny that the screen just locks, you can only move the mouse but nothing happens and there is nothing written in the kernel messages. The same thing that happened with the 3.7.6 drivers only this time it happened not immediately but after 2 days. Another thing. With the radeon drivers, xv works fine but with the fglrx drivers xv doesn't work. This seems to be a problem with the Xorg version used in all distributions. A guy from the rage3d forum who downloaded the latest Xorg CVS yesterday, said that this is solved. When do you expect to update the Xorg version, if you intend to do so? Thanks again for the quick answers. Stefan Dirsch wrote:
Hi
I was able to reproduce this problem with the 3.9.0 drivers from ATI. After reverting the regparm changes added by ATI in firegl_public.h and adding the regparm patches I used for 3.7.6 the driver works fine for me. The warning messages
kernel: Badness in do_no_page at mm/memory.c:1455 kernel: Badness in copy_page_range at mm/memory.c:331
are still active, but seem not to be fatal for a working driver. I've uploaded new RPMs to
ftp://ftp.suse.com/pub/suse/i386/supplementary/X/XFree86/ATI/suse91/fglrx/3.9.0
Hope this helps.
Stefan
On Mon, May 31, 2004 at 07:06:44PM +0100, George Billios wrote:
Hi, Me and others have problems with the new 3.9.0 driver. If you can check the threads at rage3d forums you might be able to help us. First even though the drivers compile correcly when you modprobe them you get the following error:
fglrx: no version for "struct_module" found: kernel tainted. fglrx: module license 'Proprietary. (C) 2002 - ATI Technologies, Starnberg, GERMANY' taints kernel. [fglrx] Maximum main memory to use for locked dma buffers: 681 MBytes. Unable to handle kernel NULL pointer dereference at virtual address 00000000 printing eip: f2986d58 *pde = 00000000 Oops: 0002 [#1]
It must be something with the regparm, which the drivers clame to support and in order to make the work you must apply a custom patch from here: http://www.rage3d.com/board/showthread.php?s=&threadid=33761546
Even though the drivers load with this patch, every time you run a GL app you get errors like these: kernel: Badness in do_no_page at mm/memory.c:1455 or kernel: Badness in copy_page_range at mm/memory.c:331
This seems to happen mostly is Suse 9.1 kernels and not to other distributions. What do you think is the problem?
Public Key available ---------------------------------------------------- Stefan Dirsch (Res. & Dev.) SUSE LINUX AG Tel: 0911-740 53 0 Maxfeldstrasse 5 FAX: 0911-740 53 479 D-90409 Nürnberg http://www.suse.de Germany ----------------------------------------------------
Besides the regparm, what can I expect from the 3.9.0 fglrx drivers? I
tried upgrading to X11R6.7.0 yesterday with no luck, and I'm back on the
fglrx 3.7.6 and XFree86 4.3.99 again. With Suse 9.1 I've only been able
to play Neverwinter Nights and get 3D support, UT2004 segfaults
immediately, and Quake 3 won't even load. I'm blaming this on
XFree/fglrx since my brother and my roommate both can run these fine on
Nvidia drivers. Can anybody give input on where the best route to
getting the performance back would be?
Thanks,
Scott
-----Original Message-----
From: George Billios
Nice :) I wish I could try them but yesterday my pc locked again because of the fglrx driver and I'm back to radeon driver again. It is very funny that the screen just locks, you can only move the mouse but nothing happens and there is nothing written in the kernel messages. The same thing that happened with the 3.7.6 drivers only this time it happened not immediately but after 2 days.
Another thing. With the radeon drivers, xv works fine but with the fglrx drivers xv doesn't work. This seems to be a problem with the Xorg version used in all distributions. A guy from the rage3d forum who downloaded the latest Xorg CVS yesterday, said that this is solved. When do you expect to update the Xorg version, if you intend to do so?
Thanks again for the quick answers.
Stefan Dirsch wrote:
Hi
I was able to reproduce this problem with the 3.9.0 drivers from ATI. After reverting the regparm changes added by ATI in firegl_public.h and adding the regparm patches I used for 3.7.6 the driver works fine for me. The warning messages
kernel: Badness in do_no_page at mm/memory.c:1455 kernel: Badness in copy_page_range at mm/memory.c:331
are still active, but seem not to be fatal for a working driver. I've uploaded new RPMs to
ftp://ftp.suse.com/pub/suse/i386/supplementary/X/XFree86/ATI/suse91/fgl rx/3.9.0
Hope this helps.
Stefan
On Mon, May 31, 2004 at 07:06:44PM +0100, George Billios wrote:
Hi, Me and others have problems with the new 3.9.0 driver. If you can
check
the threads at rage3d forums you might be able to help us. First even though the drivers compile correcly when you modprobe them you get the following error:
fglrx: no version for "struct_module" found: kernel tainted. fglrx: module license 'Proprietary. (C) 2002 - ATI Technologies, Starnberg, GERMANY' taints kernel. [fglrx] Maximum main memory to use for locked dma buffers: 681 MBytes. Unable to handle kernel NULL pointer dereference at virtual address 00000000 printing eip: f2986d58 *pde = 00000000 Oops: 0002 [#1]
It must be something with the regparm, which the drivers clame to support and in order to make the work you must apply a custom patch from here: http://www.rage3d.com/board/showthread.php?s=&threadid=33761546
Even though the drivers load with this patch, every time you run a GL app you get errors like these: kernel: Badness in do_no_page at mm/memory.c:1455 or kernel: Badness in copy_page_range at mm/memory.c:331
This seems to happen mostly is Suse 9.1 kernels and not to other distributions. What do you think is the problem?
Public Key available ---------------------------------------------------- Stefan Dirsch (Res. & Dev.) SUSE LINUX AG Tel: 0911-740 53 0 Maxfeldstrasse 5 FAX: 0911-740 53 479 D-90409 Nürnberg http://www.suse.de Germany ----------------------------------------------------
-- To unsubscribe check the List-Unsubscribe header For additional commands, email: suse-xfree86-help@suse.com
On Tue, Jun 01, 2004 at 07:17:29PM +0100, George Billios wrote:
Another thing. With the radeon drivers, xv works fine but with the fglrx drivers xv doesn't work. This seems to be a problem with the Xorg version used in all distributions. A guy from the rage3d forum who downloaded the latest Xorg CVS yesterday, said that this is solved.
I don't think, that this is related to any X.Org release. But you need Option "VideoOverlay" "on" in the Device section of /etc/X11/XF86Config, I just have learned.
When do you expect to update the Xorg version, if you intend to do so?
When the next X.Org release will be released. Hopefully still this summer. Stefan Public Key available ---------------------------------------------------- Stefan Dirsch (Res. & Dev.) SUSE LINUX AG Tel: 0911-740 53 0 Maxfeldstrasse 5 FAX: 0911-740 53 479 D-90409 Nürnberg http://www.suse.de Germany ----------------------------------------------------
Unfortunately the option doesn't change anything. This is a known bug.
When the next X.Org release will be released. Hopefully still this summer.
That's good news. Stefan Dirsch wrote:
On Tue, Jun 01, 2004 at 07:17:29PM +0100, George Billios wrote:
Another thing. With the radeon drivers, xv works fine but with the fglrx drivers xv doesn't work. This seems to be a problem with the Xorg version used in all distributions. A guy from the rage3d forum who downloaded the latest Xorg CVS yesterday, said that this is solved.
I don't think, that this is related to any X.Org release. But you need
Option "VideoOverlay" "on"
in the Device section of /etc/X11/XF86Config, I just have learned.
When do you expect to update the Xorg version, if you intend to do so?
When the next X.Org release will be released. Hopefully still this summer.
Stefan
Public Key available ---------------------------------------------------- Stefan Dirsch (Res. & Dev.) SUSE LINUX AG Tel: 0911-740 53 0 Maxfeldstrasse 5 FAX: 0911-740 53 479 D-90409 Nürnberg http://www.suse.de Germany ----------------------------------------------------
I am having the same problems that I had with 9.0. I have a Radeon 8500 installed in SuSE 9.1 kernel 2.6.4-54.5-default with all updates. I have installed ATI 3.9.0 and 3.7.6 SuSE rpm's as per the readme and all have concluded the same: negativity@phantom:~> fglrxinfo display: :0.0 screen: 0 OpenGL vendor string: Mesa project: www.mesa3d.org OpenGL renderer string: Mesa GLX Indirect OpenGL version string: 1.2 (1.4 Mesa 5.0.2) I have used both the sax2 and fglrxconfig config methods to enable 3d. I have noticed that sax2 writes to XF86Config and fglrxconfig writes to XF86Config-4. I've renamed the Config-4 to XF86Config. I am missing something in the process, can someone lend me a brain cell? Can it be something with my VIA KT400 chipset? I've "heard" they don't play well. Any comments will be much appreciated! BTW, also have tried the generic ATI drivers but still render OpenGL with MESA. ~Kenny :::!~!!!!!:. .xUHWH!! !!?M88WHX:. .X*#M@$!! !X!M$$$$$$WWx:. :!!!!!!?H! :!$!$$$$$$$$$$8X: !!~ ~:~!! :~!$!#$$$$$$$$$$8X: :!~::!H!< ~.U$X!?R$$$$$$$$MM! ~!~!!!!~~ .:XW$$$U!!?$$$$$$RMM! !:~~~ .:!M"T#$$$$WX??#MRRMMM! ~?WuxiW*` `"#$$$$8!!!!??!!! :X- M$$$$ `"T#$T~!8$WUXU~ :%` ~#$$$m: ~!~ ?$$$$$$ :!`.- ~T$$$$8xx. .xWW- ~""##*" ..... -~~:<` ! ~?T#$$@@W@*?$$ /` W$@@M!!! .!~~ !! .:XUW$W!~ `"~: : #"~~`.:x%`!! !H: !WM$$$$Ti.: .!WUn+!` :::~:!!`:X~ .: ?H.!u "$$$B$$$!W:U!T$$M~ .~~ :X@!.-~ ?@WTWo("*$$$W$TH$! ` Wi.~!X$?!-~ : ?$$$B$Wu("**$RM! $R@i.~~ ! : ~$$$$$B$$en:`` ?MXT@Wx.~ : ~"##*$$$$M~ insanity: doing the same thing over and over again and expecting different results. All your oil are belong to U.S.
Kenny, This is common problem according to Google. I don't know what the correct answer is but you may want to take a look at what I posted yesterday. That worked for me and you may want to try out something similarly hazardous. The end result should be that your /var/log/XFree86.0.log should say DRI initialization successful and not failed. -claus On Tuesday 01 June 2004 18:35, kenny wrote:
I am having the same problems that I had with 9.0. I have a Radeon 8500 installed in SuSE 9.1 kernel 2.6.4-54.5-default with all updates. I have installed ATI 3.9.0 and 3.7.6 SuSE rpm's as per the readme and all have concluded the same:
negativity@phantom:~> fglrxinfo display: :0.0 screen: 0 OpenGL vendor string: Mesa project: www.mesa3d.org OpenGL renderer string: Mesa GLX Indirect OpenGL version string: 1.2 (1.4 Mesa 5.0.2)
I have used both the sax2 and fglrxconfig config methods to enable 3d. I have noticed that sax2 writes to XF86Config and fglrxconfig writes to XF86Config-4. I've renamed the Config-4 to XF86Config.
I am missing something in the process, can someone lend me a brain cell? Can it be something with my VIA KT400 chipset? I've "heard" they don't play well. Any comments will be much appreciated!
BTW, also have tried the generic ATI drivers but still render OpenGL with MESA.
~Kenny
:::!~!!!!!:.
.xUHWH!! !!?M88WHX:. .X*#M@$!! !X!M$$$$$$WWx:.
:!!!!!!?H! :!$!$$$$$$$$$$8X:
!!~ ~:~!! :~!$!#$$$$$$$$$$8X: :!~::!H!< ~.U$X!?R$$$$$$$$MM!
~!~!!!!~~ .:XW$$$U!!?$$$$$$RMM! !:~~~ .:!M"T#$$$$WX??#MRRMMM! ~?WuxiW*` `"#$$$$8!!!!??!!!
:X- M$$$$ `"T#$T~!8$WUXU~ : :%` ~#$$$m: ~!~ ?$$$$$$ : :!`.- ~T$$$$8xx. .xWW- ~""##*"
..... -~~:<` ! ~?T#$$@@W@*?$$ /` W$@@M!!! .!~~ !! .:XUW$W!~ `"~: : #"~~`.:x%`!! !H: !WM$$$$Ti.: .!WUn+!`
:::~:!!`:X~ .: ?H.!u "$$$B$$$!W:U!T$$M~
.~~ :X@!.-~ ?@WTWo("*$$$W$TH$! ` Wi.~!X$?!-~ : ?$$$B$Wu("**$RM! $R@i.~~ ! : ~$$$$$B$$en:`` ?MXT@Wx.~ : ~"##*$$$$M~
insanity: doing the same thing over and over again and expecting different results. All your oil are belong to U.S.
List, I first used VNC Viewer under Windows and was impressed by its effectiveness for remote assistance, and its free. I like the fact that the user at the PC and myself can both view the same desktop - me remotley and the user at the computer. I have recentley used VNC Server and Viewer under Linux. Works fine but how do I view the remote desktop currentley used by the user at the PC? When I VNC into the Linux box I get to see the desktop - however it is a different (new?) desktop - not the one currenly used by the user at the Linux PC. Is it possible to replicate the functionality provided by default under WIndows where both users see the same desktop? Does the X Windows server support this? Thanks in advance.. Jonathan Hughes Goodyear South Africa
On Friday 11 June 2004 08:39, jonathan_hughes@goodyear.co.za wrote:
Is it possible to replicate the functionality provided by default under WIndows where both users see the same desktop? Does the X Windows server support this?
If you are using KDE this is very easy to do with Desktop Sharing. The user whose desktop needs to be shared invites the other user for instance via e-mail. I don't use VNC often so I do not know about other implementations, but I guess there are more of them available. -Teemu
On Friday 11 June 2004 01:39 am, jonathan_hughes@goodyear.co.za wrote:
Is it possible to replicate the functionality provided by default under WIndows where both users see the same desktop? Does the X Windows server support this? Chris Freyer
Jonathan-- I'm running SuSE 9.0 and I don't have the problem you have. I'm running Desktop Sharing 1.0 instead of VNC. It has some nice features, such as the ability to send email invitations, manage the duration of the login, and allow/disallow the viewer to control the mouse/keyboard.
participants (9)
-
Chris Freyer -
Claus Olesen -
Ferguson, Scott W -
George Billios -
Hartmut Berghoff -
jonathan_hughes＠goodyear.co.za -
kenny -
Stefan Dirsch -
Teemu Nikkilä