Re: [suse-xfree86] ATI 3.9.0 drivers and Suse 9.1
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?
I have the same messages with driver 3.76, kernel-default-2.6.4-54.5 (the last available from YOU). I'm starting to think this kernel and the patch or the drivers does not like each other. Sadly I don't have time to try a newer kernel in the next days, but maybe one of the new ones from ftp://ftp.suse.com/pub/projects/kernel/kotd/i386 will work... Ciao, Roberto
On Mon, May 31, 2004 at 10:09:44PM +0200, r.maurizzi@digitalpha.it wrote:
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?
I have the same messages with driver 3.76, kernel-default-2.6.4-54.5 (the last available from YOU). I'm starting to think this kernel and the patch or the drivers does not like each other.
Did you try the 3.7.6 fglrx/km_fglrx RPMs built from SuSE? ftp://ftp.suse.com/pub/suse/i386/supplementary/X/XFree86/ATI/suse91/fglrx/3.7.6 Required regparm patches for 9.1 are already applied for them. 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 ----------------------------------------------------
The patch I mentioned was firstly made for the 3.7.6 drivers when they first came out 2+ months ago. The thing is that ATI's 3.9.0 drivers are supposed to have the regparm patch (it is said so in the release notes) but Suse 9.1 is the only distribution that doesn't work with these drivers. So I must assume that Suse's kernels have some patch that breaks something. That is why I'm asking for help and to go back to an older driver (which besides anything else caused me reboots) is not an option. Stefan Dirsch wrote:
On Mon, May 31, 2004 at 10:09:44PM +0200, r.maurizzi@digitalpha.it wrote:
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?
I have the same messages with driver 3.76, kernel-default-2.6.4-54.5 (the last available from YOU). I'm starting to think this kernel and the patch or the drivers does not like each other.
Did you try the 3.7.6 fglrx/km_fglrx RPMs built from SuSE?
ftp://ftp.suse.com/pub/suse/i386/supplementary/X/XFree86/ATI/suse91/fglrx/3.7.6
Required regparm patches for 9.1 are already applied for them.
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 ----------------------------------------------------
On Mon, May 31, 2004 at 10:34:53PM +0100, George Billios wrote:
The patch I mentioned was firstly made for the 3.7.6 drivers when they first came out 2+ months ago.
Which patch did you mention?
The thing is that ATI's 3.9.0 drivers are supposed to have the regparm patch (it is said so in the release notes) but Suse 9.1 is the only distribution that doesn't work with these drivers.
Which other distribution with regparm enabled kernels did you try?
So I must assume that Suse's kernels have some patch that breaks something.
I don't think so. I think the regparm adjustments of ATI are simply not complete.
That is why I'm asking for help and to go back to an older driver (which besides anything else caused me reboots) is not an option.
So you did try the 3.7.6 RPMs built by SuSE and these didn't work? Stefan
Stefan Dirsch wrote:
On Mon, May 31, 2004 at 10:09:44PM +0200, r.maurizzi@digitalpha.it wrote:
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?
I have the same messages with driver 3.76, kernel-default-2.6.4-54.5 (the last available from YOU). I'm starting to think this kernel and the patch or the drivers does not like each other.
Did you try the 3.7.6 fglrx/km_fglrx RPMs built from SuSE?
ftp://ftp.suse.com/pub/suse/i386/supplementary/X/XFree86/ATI/suse91/fglrx/3.7.6
Required regparm patches for 9.1 are already applied for them.
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 ----------------------------------------------------
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 have the issue too. From /var/log/messages and /var/log/warn kernel: Badness in copy_page_range at mm/memory.c:331 kernel: Badness in do_no_page at mm/memory.c:1455 I have a couple of other anomalies. From /var/log/XFree86.0.log (WW) fglrx: No matching Device section for instance (BusID PCI:3:0:1) found I think that's because my card - ATI 9600 - is dual head of which I'm using only one of the heads. The other anomaly (II) fglrx(0): doing DRIScreenInit drmOpenDevice: minor is 0 drmOpenDevice: node name is /dev/dri/card0 drmOpenDevice: open result is -1, (Unknown error 999) drmOpenDevice: open result is -1, (Unknown error 999) drmOpenDevice: Open failed drmOpenDevice: minor is 0 drmOpenDevice: node name is /dev/dri/card0 drmOpenDevice: open result is -1, (Unknown error 999) drmOpenDevice: open result is -1, (Unknown error 999) drmOpenDevice: Open failed drmOpenDevice: minor is 0 drmOpenDevice: node name is /dev/dri/card0 drmOpenDevice: open result is 4, (OK) I don't know what the above means. /dev/dri/card0 does exist. But besides from these everything appears ok. Later in the same log I get (II) fglrx(0): DRI initialization successfull! But /usr/X11/bin/fgl_glxgears still won't run but exits immediately with Error: couldn't get fbconfig Anyone know what that means? A problem with that particular program? Anyone know of another way to test if 3D is working properly? I'm using SuSE 9.1 stock with all YOU updates and SuSE's version of the ATI driver - not ATI's version. E.g. I'm not using 3.9.0 but 3.7.6. Maybe the following from my experience is helpful to someone starting out on this nightmare. For the longest time DRI initialization failed on me - also with ATI's driver instead of SuSE's version. I ended up removing the fglrx driver and Mesa completely and then reinstalled SuSE's version of the driver ftp://ftp.suse.com/pub/suse/i386/supplementary/X/XFree86/ATI/suse91/fglrx acccording to the README and Mesa in that order and all with --force, all at run level 3 to ensure that the fglrx driver was not installed and with reboots in between. I'm using the /etc/X11/XF86Config generated by /usr/X11/bin/fglrxconfig with default replies except I changed Option "UseInternalAGPGART" "yes" to Option "UseInternalAGPGART" "no" The build of the kernel module was clean except building of /usr/src/kernel-modules/fglrx/firegl_public.c showed some warnings /usr/src/kernel-modules/fglrx/firegl_public.c: In function `__ke_block_all_signals': /usr/src/kernel-modules/fglrx/firegl_public.c:794: warning: passing arg 1 of `block_all_signals' from incompatible pointer type /usr/src/kernel-modules/fglrx/firegl_public.c: In function `__ke_request_irq': /usr/src/kernel-modules/fglrx/firegl_public.c:1826: warning: assignment from incompatible pointer type /usr/src/kernel-modules/fglrx/firegl_public.c: At top level: /usr/src/kernel-modules/fglrx/firegl_public.c:2260: warning: initialization from incompatible pointer type /usr/src/kernel-modules/fglrx/firegl_public.c:2278: warning: initialization from incompatible pointer type /usr/src/kernel-modules/fglrx/firegl_public.c:2285: warning: initialization from incompatible pointer type /usr/src/kernel-modules/fglrx/firegl_public.c:2292: warning: initialization from incompatible pointer type /usr/src/kernel-modules/fglrx/firegl_public.c:2301: warning: initialization from incompatible pointer type But I don't know if they are important. In the end, running /usr/X11/bin/fglrxinfo should produce this (from what I read) display: :0.0 screen: 0 OpenGL vendor string: ATI Technologies Inc. OpenGL renderer string: RADEON 9600 Generic OpenGL version string: 1.3 (X4.3.0-3.7.6) If it says Mesa instead of ATI then there is a problem.
Stefan Dirsch wrote:
On Mon, May 31, 2004 at 10:34:53PM +0100, George Billios wrote:
Which patch did you mention?
In the first mail the patch that someone created for the 3.9.0 drivers to work with Suse, comes from the patch for the 3.7.6 (and older) drivers. If you can visit the rage3d.com forums there are many informations about the drivers.
Which other distribution with regparm enabled kernels did you try? Not me, but others are using Mandrake, Fedora Core and Gentoo regparm kernels that work without any patch with the 3.9.0 drivers.
I don't think so. I think the regparm adjustments of ATI are simply not complete.
As I said, people with other distributions can install the drivers without errors and without the need for a patch. Probably ATI has also something to do with it, but under these circumstances SuSE must have done something too.
So you did try the 3.7.6 RPMs built by SuSE and these didn't work?
I've tried the 3.7.6 drivers with Suse 9.0 (upgraded kernel to 2.6) when they were released and they would hang the system. I'm now using the 3.9.0 drivers (with the patch to make it work under Suse) and they are stable but I get the errors I mentioned in the first mail. I can't really understand why the 3.9.0 drivers can't be directly installed under SuSE 9.1 . If you are sure this is not Suse's problem then you should contact ATI in order to solve it.
On Mon, May 31, 2004 at 03:13:36PM -0700, Claus Olesen wrote:
I have the issue too. From /var/log/messages and /var/log/warn kernel: Badness in copy_page_range at mm/memory.c:331 kernel: Badness in do_no_page at mm/memory.c:1455
I get the same warnings/errors with using the 3.7.6 RPM built by SuSE with Kernel 2.6.4-54.5-default (update kernel) and kernel-default-2.6.4-52 (original 9.1 kernel), but the driver works fine for me (Radeon 9000 Pro) - for whatever reasons. 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 ----------------------------------------------------
participants (4)
-
Claus Olesen -
George Billios -
r.maurizzi＠digitalpha.it -
Stefan Dirsch