May 2008 - Radeon HD - openSUSE Mailing Lists

Re: [radeonhd] System-Crash with radeonhd or ATI-gflrx-driver and Sapphire RADEON HD 3870
by Matthias Hopf 29 May '08

29 May '08

Please keep this on the list. Others might have experience with some of your symptoms. On May 28, 08 20:32:58 +0200, Marcus Ziegelmeier wrote: > > -------- Original-Nachricht -------- > > Datum: Wed, 28 May 2008 14:24:50 +0200 > > Von: Matthias Hopf <mhopf(a)suse.de> > > An: Marcus Ziegelmeier <StMarcus(a)gmx.net> > > Betreff: Re: [radeonhd] System-Crash with radeonhd or ATI-gflrx-driver and Sapphire RADEON HD 3870 > > > On May 28, 08 13:49:00 +0200, Marcus Ziegelmeier wrote: > > > So, i'll try to check on the versions and i guess i should get the same > > versions for recompiling the drivers ones more? > > > > Yes, and don't forget to configure or autogen again. > > So, > i fixed the different Version an now have 7.3.xxx for xserver. > The GUI starts without the workaround, BUT > * still cryptographic symols instead of normal whitespaces on every console screen :-( Haven't heard of that. Maybe you should try Option "AccelMethod" "ShadowFB" and check whether that's an acceleration framework issue. > * still not showing a resolution of 1920x1200 even if it automatical configured in xorg.conf, so i think the systems recognizes that a higher resolution should be possible Can't debug this without a log (-logverbose 7) > * still kdetv says it could not work with the video-plugin and i could not find any solution - but i'm not sure this problem belongs to the radeonhd-driver There is no XVideo support yet. Matthias -- Matthias Hopf <mhopf(a)suse.de> __ __ __ Maxfeldstr. 5 / 90409 Nuernberg (_ | | (_ |__ mat(a)mshopf.de Phone +49-911-74053-715 __) |_| __) |__ R & D www.mshopf.de -- To unsubscribe, e-mail: radeonhd+unsubscribe(a)opensuse.org For additional commands, e-mail: radeonhd+help(a)opensuse.org

1 0

xf86-video-radeonhd:master: 1 commit(s)
by mhopf＠kemper.freedesktop.org 29 May '08

29 May '08

Reporting from xf86-video-radeonhd: Radeon HD video driver Please visit: http://gitweb.freedesktop.org/?p=xorg/driver/xf86-video-radeonhd Or get your own copy by using: git-clone git://anongit.freedesktop.org/xorg/driver/xf86-video-radeonhd Commit against master at 5d9a8cb8...: commit 9ce56cd38799a33590fbe46e554efc2519177fe7 Author: Matthias Hopf <mhopf(a)suse.de> Date: Thu May 29 11:27:41 2008 +0200 dri: move ansi-c header inclusion to correct location. diffstat: src/rhd_dri.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) gitweb url: http://gitweb.freedesktop.org/?p=xorg/driver/xf86-video-radeonhd;a=commit;h… -- To unsubscribe, e-mail: radeonhd+unsubscribe(a)opensuse.org For additional commands, e-mail: radeonhd+help(a)opensuse.org

1 0

RE: [radeonhd] State of DRI on r5xx
by Egbert Eich 29 May '08

29 May '08

Bridgman, John writes: > Before "2D acceleration" was introduced the radeonhd driver still had > shadowfb support which made a big difference in 2D performance. I'm > pretty sure Matthias mentioned that the 3D support did not co-exist with > shadowfb either but I'm not 100% sure of that. > > Without shadowfb support, performance would be on a par with fglrx > running without DRI, ie really slow. > Right, however shadowfb should only matter for R5xx. IGP (like RS690) should be OK. Cheers, Egbert. -- To unsubscribe, e-mail: radeonhd+unsubscribe(a)opensuse.org For additional commands, e-mail: radeonhd+help(a)opensuse.org

1 0

xf86-video-radeonhd:master: 1 commit(s)
by libv＠kemper.freedesktop.org 28 May '08

28 May '08

Reporting from xf86-video-radeonhd: Radeon HD video driver Please visit: http://gitweb.freedesktop.org/?p=xorg/driver/xf86-video-radeonhd Or get your own copy by using: git-clone git://anongit.freedesktop.org/xorg/driver/xf86-video-radeonhd Commit against master at 5568da86...: commit 5d9a8cb863f5935ea442310864615dfcaf44a971 Author: Luc Verhaegen <libv(a)skynet.be> Date: Thu May 29 01:16:37 2008 +0200 MC: never reread rhdPtr->FBIntAddress. The drm never really changes the FB location, just goes around in a circle. Actually changing the MC view of the FB would be insanity, as everything we do depends on this internal view. We nominated a value in PreInit, and then use this consistently. So: * remove the code to repetitively get this nominated value. * stop saving the offsets twice by saving them again in DRIFinishScreenInit. * tighten up the debug logging surrounding addressing. diffstat: src/rhd_dri.c | 7 ------- src/rhd_driver.c | 23 ++++++++++++----------- src/rhd_mc.c | 47 ++++++++++++++++------------------------------- src/rhd_mc.h | 1 - src/rhd_vga.c | 2 +- 5 files changed, 29 insertions(+), 51 deletions(-) gitweb url: http://gitweb.freedesktop.org/?p=xorg/driver/xf86-video-radeonhd;a=commit;h… -- To unsubscribe, e-mail: radeonhd+unsubscribe(a)opensuse.org For additional commands, e-mail: radeonhd+help(a)opensuse.org

1 0

Re: [radeonhd] DRI support hit master
by Egbert Eich 28 May '08

28 May '08

Adam Williamson writes: > On Wed, 2008-05-28 at 09:54 -0700, Adam Williamson wrote: > > > Okay, got this fixed thanks to our X guy, Paulo Andrade. He reminded me > > that it's the exact same issue we ran into (and was fixed upstream) a > > while back. As he puts it: > > Er, I should clarify: we ran into it, reported it, it was fixed, and now > it's been re-introduced. :) In another file, yes. The same solution should however apply to this file. Cheers, Egbert. -- To unsubscribe, e-mail: radeonhd+unsubscribe(a)opensuse.org For additional commands, e-mail: radeonhd+help(a)opensuse.org

1 0

Re: [radeonhd] DRI support hit master
by Egbert Eich 28 May '08

28 May '08

Adam Williamson writes: > On Tue, 2008-05-27 at 15:31 -0700, Adam Williamson wrote: > > On Tue, 2008-05-27 at 18:01 +0200, Matthias Hopf wrote: > > > I just pushed DRI support to master for radeonhd. Yippie! > > > > Having trouble building this on current Mandriva Cooker (which is not > > bleeding edge X.org, in case that's important). Well, I figure it's this > > that's breaking, but hey, it could be the scaler or RS780 stuff. > > Anyway :) Here's the error: > > > > In file included from /usr/include/xorg/opaque.h:48, > > from /usr/include/xorg/windowstr.h:60, > > from /usr/include/xorg/exa.h:36, > > from rhd_dri.c:60: > > /usr/include/setjmp.h:49: error: conflicting types for 'xf86jmp_buf' > > /usr/include/xorg/xf86_libc.h:99: error: previous declaration of 'xf86jmp_buf' was here > > /usr/include/setjmp.h:53: error: expected ')' before '==' token > > /usr/include/setjmp.h:82: error: conflicting types for 'xf86longjmp' > > /usr/include/xorg/xf86_ansic.h:292: error: previous declaration of 'xf86longjmp' was here > > > > any advice on the best fix? Thanks. > > Okay, got this fixed thanks to our X guy, Paulo Andrade. He reminded me > that it's the exact same issue we ran into (and was fixed upstream) a > while back. As he puts it: > > "We just are not linking by default with the builtin X Server > libc_wrapper, and it was already removed from the X Server long ago... > only still exists on X Server 1.4. I think SuSE may have some > dependency on libc_wrapper. I left the problem "on purpose" in Mandriva > to ensure new compilations will not link with libc_wrapper, and directly > call libc functions." > > The patch to fix this in MDV was basically to stop it ever including > xf86_ansic.h : > > *** > > @@ -131,12 +131,6 @@ AM_CONDITIONAL(USE_DRI, test "x$USE_DRI" > CFLAGS="$SAVED_CFLAGS" > CPPFLAGS="$SAVED_CPPFLAGS" > > -AC_CHECK_HEADER([${sdkdir}/xf86_ansic.h], > - [AC_DEFINE([HAVE_XF86_ANSIC_H], [1], > - [Define to 1 if you have <${sdkdir}/xf86_ansic.h>.]) > - have_xf86_ansic_h=yes]) > -AM_CONDITIONAL([HAVE_XF86_ANSIC_H], [test "x$have_xf86_ansic_h" = "xyes"]) > - > SAVED_CFLAGS="$CFLAGS" > CFLAGS="$CFLAGS $XORG_CFLAGS" > This problem has been fixed by me a while ago for other files. The commit ID for this was: 3293c60d0e8b026f3fedcc182ac86bb05d6a1ced I can't test this right now but maybe a similar fix could be persued there. Cheers, Egbert. -- To unsubscribe, e-mail: radeonhd+unsubscribe(a)opensuse.org For additional commands, e-mail: radeonhd+help(a)opensuse.org

1 0

Re: [radeonhd] DRI support hit master
by Matthias Hopf 28 May '08

28 May '08

On May 28, 08 14:15:00 +0200, Klaas Tammling wrote: > Sorry if it sounds a bit silly but I never heard about DRI and just read Direct Rendering Interface / Infrastructure, depending on context ;) > about what it actually is. So if DRI is now in your master does that mean > with your next stable release there will be 3d acceleration? Yes, for r5xx. But you need the latest DRM (kernel module), Xserver, and Mesa (actually, the r500-support branch). Eventually, everything will work with stable releases, but that takes some time. > Btw: Is that normal that if I want to reply to your mailinglist I only have > the private e-mail address so send to? The mailing list is configured that way. This is the source for constant discussion, but many people consider it the right (TM) way to do. All decent mailers have a special "reply-to-mailinglist" function. Matthias -- Matthias Hopf <mhopf(a)suse.de> __ __ __ Maxfeldstr. 5 / 90409 Nuernberg (_ | | (_ |__ mat(a)mshopf.de Phone +49-911-74053-715 __) |_| __) |__ R & D www.mshopf.de -- To unsubscribe, e-mail: radeonhd+unsubscribe(a)opensuse.org For additional commands, e-mail: radeonhd+help(a)opensuse.org

2 1

Re: [radeonhd] 0x95C5:0x174B:0xE670: Sapphire Radeon HD 3450
by Graeme Hamilton 28 May '08

28 May '08

On Tue, 2008-05-27 at 12:33 +0200, Matthias Hopf wrote: > Nope. Crtc 0 shouldn't change, and that's why I need a --logverbose 7 > log, to verify this. > > CU > > Matthias > Hi Matthias, Please find the verbose log attached as requested. Graeme -- Helpdesk Support Analyst UHI Millennium Institute Synergie Building Fairways Business Park Inverness IV2 6AA Creating the University of the Highlands and Islands / A' Cruthachadh Oilthigh na Gaidhealtachd's nan Eilean A Millennium Commission lottery project. A limited company registered in Scotland No 148203. Recognised Scottish Charity No. SC022228. Registered office: 12B Ness Walk, Inverness, IV3 5SQ.

3 4

Re: [radeonhd] State of DRI on r5xx
by Marvin 28 May '08

28 May '08

Am Donnerstag 22 Mai 2008 21:54:30 schrieb Sebastian Brocks: > Marvin schrieb: > > as I get the same error with a totaly different chipset RS690, it is > > possible that this might be a 64bit issue, as you are also using x86_64. > > I tried the "ati" driver and dri worked well with it (but is very > > unstable). Can you also confirm this, Sebastian? > > I'm also using x86_64, but have not yet tested the "ati" driver. doh! I was to fast... The ati driver also shows the problem. So maybe it is because they share common code, or it is somewhere else (mesa/Xorg). Greetings Marvin -- To unsubscribe, e-mail: radeonhd+unsubscribe(a)opensuse.org For additional commands, e-mail: radeonhd+help(a)opensuse.org

5 9

System-Crash with radeonhd or ATI-gflrx-driver and Sapphire RADEON HD 3870
by Marcus Ziegelmeier 28 May '08

28 May '08

Hallo guys, i'm always new in the world of Suse and Linux, but i'm willing to learn and getting more and more to know the good sides on OSS. I bought a new Computer 2 weeks ago and tried to get everything fixed an running, but i'm still stuck with one main problem: First some details on my system, if they may be important? MB: Asus P5K Premium RAM: 8GB CPU: Intel Quadcore Q9450 GPU: Sapphire Radeon HD 3870 Dual DVI with 512MB Passivcooled BS: (Windows Vista x64, Windows XP x64) and OpenSUSE 10.3 64 Monitor: Acer PW243 24" & Samsung 17" So, when installing the system, Suse installs a rudimental VESA driver and that is working quite well, but only with 1 Monitor and on a max resolution of 1200x1024. Also my old TV-card (WinTV PCI - BT878) does not work with that VESA driver. So i need to get any ATI driver working, but everything i tried failed at last. As a newby i tried the 1-Click install for ATI Cards from the Suse-ATI Page, but after running "SaX2 -r" or "SaX2 -r -m 0=fglrx" my system crashes down when "startx" is called automatically. It results in a black screen and the whole computer doesn't response anymore. The same happend with 8.41, 8.47 and 8.49 Version. I also tried the new radeonhd opensource driver with the same results - a full system crash. Refering to the Manual the HD 3870 should be supported and i think i'm doing everything right. Has anyone noticed a similar problem or knows any workaround for the HD 3870? A friend of mine, very familiar with opensuse tried the same and said, that the driver might cause the kernel to crash. The files (logs & hardware infos) are already sent to suse, but i thought it could be helpful to ask here on the list as well. So, any ideas or advise for help are very welcome. Thanks and best regards, Marcus Ziegelmeier -- Psssst! Schon vom neuen GMX MultiMessenger gehört? Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger -- To unsubscribe, e-mail: radeonhd+unsubscribe(a)opensuse.org For additional commands, e-mail: radeonhd+help(a)opensuse.org

2 3