Bug#541235: xserver-xorg-video-ati: Random Lockups when DRI enabled
2009/8/17 Stefano pietran...@gmail.com: On Thu, 2009-08-13 at 11:01 +0200, Michel Dänzer wrote: Hmm. One thing I notice is that there's almost no EXA offscreen memory, due to your enormous Virtual directive. Normally this should only affect performance, not correctness, but it might be worth trying a smaller maximum desktop size just in case. Hello, I tried different configurations for my Virtual screen. Here are the logs (hope I've picked the right lines!): Virtual: 2944x1848 (BIG) (II) EXA(0): Offscreen pixmap area of 786432 bytes VirtuaL: NONE (II) EXA(0): Offscreen pixmap area of 11436032 bytes Virtual: 1920x1848 (VERTICAL) (II) EXA(0): Offscreen pixmap area of 12255232 bytes VirtuaL: 2944x1080 (HORIZONTAL) (II) EXA(0): Offscreen pixmap area of 14352384 bytes I thought that the configuration with most free memory would have been the second one (no virtual screen) but surprisingly to me it is not the case. By default xrandr 1.2 grabs a square framebuffer so rotation will work (e.g., 1920x1920). Now I'm using the 2944x1080 (HORIZONTAL) Virtual screen with AGPMode 4 (default) and DRI enabled. The issues I had seem to be gone. However I cannot activate Compositing when I use my external screen since X consumes almost the 80% of the CPU when I move the windows around the screen. If I use my laptop monitor (1024x768) Compositing works quite well, although I experienced some lock-ups when I tried to modify the options of gnome-do in order to activate the Docky. Your desktop is larger than the coordinate limits of the texture engine so it's likely falling back to software. Can we consider this bug closed? Yes. Can you do anything to prevent users to make the same mistakes I made? radeon KMS (kernel modesetting) includes a unified memory manager that will allow you to resize your desktop on the fly without needing to add the virtual stuff. Alex -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#541235: xserver-xorg-video-ati: Random Lockups when DRI enabled
On Thu, 2009-08-13 at 11:01 +0200, Michel Dänzer wrote: Hmm. One thing I notice is that there's almost no EXA offscreen memory, due to your enormous Virtual directive. Normally this should only affect performance, not correctness, but it might be worth trying a smaller maximum desktop size just in case. Hello, I tried different configurations for my Virtual screen. Here are the logs (hope I've picked the right lines!): Virtual: 2944x1848 (BIG) (II) EXA(0): Offscreen pixmap area of 786432 bytes VirtuaL: NONE (II) EXA(0): Offscreen pixmap area of 11436032 bytes Virtual: 1920x1848 (VERTICAL) (II) EXA(0): Offscreen pixmap area of 12255232 bytes VirtuaL: 2944x1080 (HORIZONTAL) (II) EXA(0): Offscreen pixmap area of 14352384 bytes I thought that the configuration with most free memory would have been the second one (no virtual screen) but surprisingly to me it is not the case. Now I'm using the 2944x1080 (HORIZONTAL) Virtual screen with AGPMode 4 (default) and DRI enabled. The issues I had seem to be gone. However I cannot activate Compositing when I use my external screen since X consumes almost the 80% of the CPU when I move the windows around the screen. If I use my laptop monitor (1024x768) Compositing works quite well, although I experienced some lock-ups when I tried to modify the options of gnome-do in order to activate the Docky. Can we consider this bug closed? Can you do anything to prevent users to make the same mistakes I made? Thank you, Stefano -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#541235: xserver-xorg-video-ati: Random Lockups when DRI enabled
On Wed, 2009-08-12 at 18:14 -0700, David Rees wrote: This bug sounds like a duplicate of these two upstream: http://bugs.freedesktop.org/show_bug.cgi?id=20348 http://bugs.freedesktop.org/show_bug.cgi?id=13176 At least for me, enabling XAA and XaaNoSolidFillRect seems to get decent performance without locking up. That would make sense if you have an R420 GPU, this is a known erratum of that core. Stefano has an RV350 though, which isn't affected by this. Unfortunately there's a virtually unlimited number of problems that can cause GPU lockups, and people tend to focus too much on the symptoms (which tend to be similar) rather than the causes / triggers. -- Earthling Michel Dänzer |http://www.vmware.com Libre software enthusiast | Debian, X and DRI developer -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#541235: xserver-xorg-video-ati: Random Lockups when DRI enabled
On Wed, 2009-08-12 at 17:35 -0400, Stefano wrote: AGPMode 4 was the default mode, thus it does not work I tried AGPMode 2 but with no results, at the time I'm writing this email I'm using AGPMode 1 Did that work better? but still I have the display not behaving correctly (as the attached image in my first email) Hmm. One thing I notice is that there's almost no EXA offscreen memory, due to your enormous Virtual directive. Normally this should only affect performance, not correctness, but it might be worth trying a smaller maximum desktop size just in case. -- Earthling Michel Dänzer |http://www.vmware.com Libre software enthusiast | Debian, X and DRI developer -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#541235: xserver-xorg-video-ati: Random Lockups when DRI enabled
Hi Michel, On Thu, 2009-08-13 at 11:01 +0200, Michel Dänzer wrote: On Wed, 2009-08-12 at 17:35 -0400, Stefano wrote: AGPMode 4 was the default mode, thus it does not work I tried AGPMode 2 but with no results, at the time I'm writing this email I'm using AGPMode 1 Did that work better? in fact, I did not have any improvement by changing the AGPMode, AGPMode 1 resulted in a system lockup as well. At this moment I'm using Option BusType PCI and it seems to work. However I've not yet pushed the graphics to see if it really work and I still have some issues with the horizontal lines displaying on screen (although it seems to me that now this problem is less noticeable). but still I have the display not behaving correctly (as the attached image in my first email) Hmm. One thing I notice is that there's almost no EXA offscreen memory, due to your enormous Virtual directive. Normally this should only affect performance, not correctness, but it might be worth trying a smaller maximum desktop size just in case. I'm using the enormous Virtual screen you see in my xorg.conf because I sometimes use a dual screen 1024x768 (laptop) + 1920x1080 (external monitor). I suspected that this could cause problems but I'm not an expert in this field and I wanted to test all the options with my favorite configuration. I can try to reduce the Virtual size and post the relevant part of the X log. At least for me, enabling XAA and XaaNoSolidFillRect seems to get decent performance without locking up. -Dave Dave, I'm using EXA because my X log said that the driver worked better with it instead of XAA, but as I already said, I'm not an expert at all ;-) I can try these options as well and then report to Debian Bugs Install the xserver-xorg-dev package. 'apt-get build-dep xserver-xorg-video-ati' should get you the needed packages. Cheers, Julien Julien, thank you for the suggestion. Now I can run autogen.sh but the (new) problem is when I run make. Here is the output: $ make make all-recursive make[1]: Entering directory `/usr/src/xf86-video-ati-6.12.2' Making all in src make[2]: Entering directory `/usr/src/xf86-video-ati-6.12.2/src' if /bin/sh ../libtool --mode=compile gcc -DHAVE_CONFIG_H -I. -I. -I.. -I./AtomBios/includes -Wall -I/usr/include/xorg -I/usr/include/pixman-1 -I/usr/include/drm -I/usr/include/X11/dri-DDISABLE_EASF -DENABLE_ALL_SERVICE_FUNCTIONS -DATOM_BIOS -DATOM_BIOS_PARSER -DDRIVER_PARSER -g -O2 -MT ati.lo -MD -MP -MF .deps/ati.Tpo \ -c -o ati.lo `test -f 'ati.c' || echo './'`ati.c; \ then mv -f .deps/ati.Tpo .deps/ati.Plo; \ else rm -f .deps/ati.Tpo; exit 1; \ fi libtool: compile: gcc -DHAVE_CONFIG_H -I. -I. -I.. -I./AtomBios/includes -Wall -I/usr/include/xorg -I/usr/include/pixman-1 -I/usr/include/drm -I/usr/include/X11/dri -DDISABLE_EASF -DENABLE_ALL_SERVICE_FUNCTIONS -DATOM_BIOS -DATOM_BIOS_PARSER -DDRIVER_PARSER -g -O2 -MT ati.lo -MD -MP -MF .deps/ati.Tpo -c ati.c -fPIC -DPIC -o .libs/ati.o if /bin/sh ../libtool --mode=compile gcc -DHAVE_CONFIG_H -I. -I. -I.. -I./AtomBios/includes -Wall -I/usr/include/xorg -I/usr/include/pixman-1 -I/usr/include/drm -I/usr/include/X11/dri-DDISABLE_EASF -DENABLE_ALL_SERVICE_FUNCTIONS -DATOM_BIOS -DATOM_BIOS_PARSER -DDRIVER_PARSER -g -O2 -MT atimodule.lo -MD -MP -MF .deps/atimodule.Tpo \ -c -o atimodule.lo `test -f 'atimodule.c' || echo './'`atimodule.c; \ then mv -f .deps/atimodule.Tpo .deps/atimodule.Plo; \ else rm -f .deps/atimodule.Tpo; exit 1; \ fi libtool: compile: gcc -DHAVE_CONFIG_H -I. -I. -I.. -I./AtomBios/includes -Wall -I/usr/include/xorg -I/usr/include/pixman-1 -I/usr/include/drm -I/usr/include/X11/dri -DDISABLE_EASF -DENABLE_ALL_SERVICE_FUNCTIONS -DATOM_BIOS -DATOM_BIOS_PARSER -DDRIVER_PARSER -g -O2 -MT atimodule.lo -MD -MP -MF .deps/atimodule.Tpo -c atimodule.c -fPIC -DPIC -o .libs/atimodule.o atimodule.c:39: error: ‘PACKAGE_VERSION_MAJOR’ undeclared here (not in a function) atimodule.c:39: error: ‘PACKAGE_VERSION_MINOR’ undeclared here (not in a function) atimodule.c:39: error: ‘PACKAGE_VERSION_PATCHLEVEL’ undeclared here (not in a function) make[2]: *** [atimodule.lo] Error 1 make[2]: Leaving directory `/usr/src/xf86-video-ati-6.12.2/src' make[1]: *** [all-recursive] Error 1 make[1]: Leaving directory `/usr/src/xf86-video-ati-6.12.2' make: *** [all] Error 2 What do I miss now? I'll update you about the situation. Thank you guys, Stefano -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#541235: xserver-xorg-video-ati: Random Lockups when DRI enabled
Package: xserver-xorg-video-ati Version: 1:6.12.2-3 Severity: important *** Please type your report below this line *** Hello, I've been hesitating a lot before posting this bug and still am not 100% sure if it's worth doing it, but whatever... Brief description: I use the radeon driver (automatically recognized). When I start X with the option DRI enabled - that is by default - the system randomly locks (sometimes I can only move the mouse pointer, sometimes I can't do anything) and I must shut it down manually using the shutdown button of my laptop. My video card: ATI MOBILITY RADEON 9700/64MB Some explanations: I have had this bug since a long time ago but I read in Brice's blog that this was an issue due to the 7.3 X.org version. Since, I've upgraded my system to version 7.4 hoping that the issue would be fixed. Surprisingly, it seemed that everything worked fine. I used the direct rendering for a whole day without a problem even in a dual screen configuration and with compiz. Unfortunately, the next day the system unexpectedly locked up when gdm was started. I tried to change my xorg.conf file many times without any good result and finally I switched back to Option DRI off (and of course I removed compiz!). From time to time (upon an update of the xserver-xorg packets) I check if the issue is fixed but the answer sadly is: no! When DRI is enables, either the system locks up when gdm starts or it locks up after a random amount of time while I'm logged in. Moreover, when DRI is enabled I notice that the video displays horizontal black lines (see the attached screenshot). I also tried to upgrade the kernel, but I have the very same issue either with the 2.6.26 or 2.6.30 kernels. I have this issue when working with dual monitors in different configurations (only the external monitor on, big desktop, same-as) but also when I use the laptop monitor and the VGA cable is disconnected. I attach the relevant parts of my X log file that I recovered after the last lock up (external monitor on, laptop monitor off). I think that this issue is somehow serious (even if I post the bug as important) because the radeon driver is automatically loaded by X with DRI enabled and can lock up the computer of freshly installed systems. Thank you, Stefano -- Package-specific info: /var/lib/x11/X.roster does not exist. /var/lib/x11/X.md5sum does not exist. X server symlink status: lrwxrwxrwx 1 root root 13 2008-09-06 15:49 /etc/X11/X - /usr/bin/Xorg -rwxr-xr-x 1 root root 1689976 2009-08-06 12:55 /usr/bin/Xorg /var/lib/x11/xorg.conf.roster does not exist. VGA-compatible devices on PCI bus: 01:00.0 VGA compatible controller: ATI Technologies Inc RV350 [Mobility Radeon 9600 M10] /var/lib/x11/xorg.conf.md5sum does not exist. Xorg X server configuration file status: -rw-r--r-- 1 root root 3425 2009-08-12 09:53 /etc/X11/xorg.conf Contents of /etc/X11/xorg.conf: # /etc/X11/xorg.conf (xorg X Window System server configuration file) Section Device Identifier ATI MOBILITY RADEON 9700/64MB Option AccelMethodEXA # Option DRI off EndSection Section Monitor Identifier Laptop LCD Option DPMS on EndSection Section Screen Identifier Laptop Screen Device ATI MOBILITY RADEON 9700/64MB Monitor Laptop LCD SubSection Display Virtual 2944 1848 #for dual monitors EndSubSection EndSection Section ServerFlags Option DontZap off # kill X with ctrl+alt+backspace EndSection Section Extensions OptionComposite Enable EndSection Section ServerLayout Identifier Default Layout EndSection Relevant contents of Xorg X server log file /var/log/Xorg.0.log.old: X.Org X Server 1.6.3 Release Date: 2009-7-31 X Protocol Version 11, Revision 0 Build Operating System: Linux 2.6.30.4-dsa-ia32 i686 Debian Current Operating System: Linux pietranera 2.6.30-1-686 #1 SMP Mon Aug 3 16:18:30 UTC 2009 i686 Build Date: 06 August 2009 04:49:57PM xorg-server 2:1.6.3-1+b1 (bui...@murphy.debian.org) (==) Log file: /var/log/Xorg.0.log (==) Using config file: /etc/X11/xorg.conf (==) ServerLayout Default Layout (**) |--Screen Laptop Screen (0) (**) | |--Monitor Laptop LCD (**) | |--Device ATI MOBILITY RADEON 9700/64MB (**) |--Input Device Synaptics Touchpad (**) Option DontZap off (==) Automatically adding devices (==) Automatically enabling devices (WW) The directory /usr/share/fonts/X11/cyrillic does not exist. Entry deleted from font path. (==) FontPath set to: /usr/share/fonts/X11/misc, /usr/share/fonts/X11/100dpi/:unscaled, /usr/share/fonts/X11/75dpi/:unscaled, /usr/share/fonts/X11/Type1, /usr/share/fonts/X11/100dpi, /usr/share/fonts/X11/75dpi, /var/lib/defoma/x-ttcidfont-conf.d/dirs/TrueType, built-ins (==) ModulePath set to /usr/lib/xorg/modules (**) Extension
Bug#541235: xserver-xorg-video-ati: Random Lockups when DRI enabled
On Wed, Aug 12, 2009 at 1:06 PM, Stefanopietran...@gmail.com wrote: Package: xserver-xorg-video-ati Version: 1:6.12.2-3 Severity: important *** Please type your report below this line *** Hello, I've been hesitating a lot before posting this bug and still am not 100% sure if it's worth doing it, but whatever... Brief description: I use the radeon driver (automatically recognized). When I start X with the option DRI enabled - that is by default - the system randomly locks (sometimes I can only move the mouse pointer, sometimes I can't do anything) and I must shut it down manually using the shutdown button of my laptop. My video card: ATI MOBILITY RADEON 9700/64MB Can you try the xf86-video-ati driver from git master? Also does changing the AGPMode or forcing PCI mode help? Option AGPMode X where X = 1 or 2 or 4 or 8 or Option BusType PCI Please try these options separately. Alex -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#541235: xserver-xorg-video-ati: Random Lockups when DRI enabled
Hi, AGPMode 8 is not supported by my video card AGPMode 4 was the default mode, thus it does not work I tried AGPMode 2 but with no results, at the time I'm writing this email I'm using AGPMode 1 but still I have the display not behaving correctly (as the attached image in my first email) and likely the system will hang up sooner or later. This is odd, because I've been playing a video game that benefits of DRI with no problems and when I quit the system locked up in few minutes. What I can add is that the lockup occurred when I was playing with the windows with Alt+Tab. I also tried to compile the xf86-video-ati-6.12.2 driver but when running autogen.sh I have the following error: ./configure: line 11076: syntax error near unexpected token `XINERAMA,' ./configure: line 11076: `XORG_DRIVER_CHECK_EXT(XINERAMA, xineramaproto)' and the process quits. Any suggestion? I will try the option BusType PCI later, and the xf86-video-ati driver if I'll be able to compile it. Thanks, Stefano On Wed, 2009-08-12 at 13:44 -0400, Alex Deucher wrote: Can you try the xf86-video-ati driver from git master? Also does changing the AGPMode or forcing PCI mode help? Option AGPMode X where X = 1 or 2 or 4 or 8 or Option BusType PCI Please try these options separately. Alex -- Stefano Un geco sul muro in paziente attesa della prossima falena -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#541235: xserver-xorg-video-ati: Random Lockups when DRI enabled
On Wed, Aug 12, 2009 at 17:35:40 -0400, Stefano wrote: I also tried to compile the xf86-video-ati-6.12.2 driver but when running autogen.sh I have the following error: ./configure: line 11076: syntax error near unexpected token `XINERAMA,' ./configure: line 11076: `XORG_DRIVER_CHECK_EXT(XINERAMA, xineramaproto)' and the process quits. Any suggestion? Install the xserver-xorg-dev package. 'apt-get build-dep xserver-xorg-video-ati' should get you the needed packages. Cheers, Julien -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#541235: xserver-xorg-video-ati: Random Lockups when DRI enabled
This bug sounds like a duplicate of these two upstream: http://bugs.freedesktop.org/show_bug.cgi?id=20348 http://bugs.freedesktop.org/show_bug.cgi?id=13176 At least for me, enabling XAA and XaaNoSolidFillRect seems to get decent performance without locking up. -Dave -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
