You are not logged in.
So I decided to upgrade my system so that I could play Starcraft 2. I compiled the 2.6.35-rc6 kernel from aur, and did a full pacman -Syu to get the latest Xorg. (xorg-server 1.8.1.902-1)
On reboot, Xorg (as expected) failed to load fglrx so I installed Catalyst 10.7 from aur, did an rmmod and modprobe on it. But when I ran startx, I got a black screen and a complete system lockup. Only a hard restart would do anything.
I saw this problem on the wiki, and tried both things there - wiping the amdpcsdb, and disabling acpi - as well as a healthy amount of rebooting, reinstalling catalyst, updating catalyst-lib32tools... but it freezes on me every time.
Interestingly, the console is a higher resolution than normal ever since the kernel upgrade.
I've got a Radeon 5770 on x86_64. Help?
EDIT: Oh, one other thing I should mention, when I'm booted into console (runlevel 3) the Radeon's fan is working full blast. As soon as I startx (where it freezes), it quiets down to a normal pace, which seems to suggest to me that at least some of fglrx is working...
Last edited by mDuo13 (2010-07-31 16:25:34)
Offline
Interesting things. Discovered that I needed to move fglrx from daemons (where it used to work) to modules. Now it boots into X successfully... almost. My mouse and keyboard (both usb) stop responding the moment X appears. (actually I haven't actually used the mouse since the kernel upgrade, but I know the keyboard works until X starts). Any ideas? Something wrong with the xf86-input* packages or something?
Offline
By the way, I don't post here expecting answers, I post here as a good-luck charm for finding the answers myself.
I needed to restore the evdev and quirks conf files in my /etc/X11/xorg.conf.d/ directory (I had renamed the to .bak as part of earlier troubleshooting).
Looks like now I just have to deal with 2d acceleration artifacts, and that's already well-documented...
Offline
Vi0L0 wrote:@jck: ive got solution well at least its working for me.
Ive updated libva (its not updated in extra repo because its not working for intel users) and xvba-video packages to newest versions, and rebuilded mplayer-vaapi and osd is always working fine now.
But please note that after doing this ie. VLC player from [extra] wont work, simply because its compiled for older version of libva, so if you want to use vlc with those newest packages you will have to rebuild vlc manually.It's not only vlc:
~ % ffmpeg ffmpeg: error while loading shared libraries: libva-0.31.0.6.so.1: cannot open shared object file: No such file or directory
Please include ffmpeg and/or vlc. :-/
Thanks for that info. I've just included ffmpeg.
Although I wont include vlc. To build vlc i need to install 114 packages on any of my i686 and x86_64 virtual systems, and it wont be easy to keep them updated on my slow connection.
Last edited by Vi0L0 (2010-08-01 14:27:06)
Offline
new_xvba solved the problem, thanks violo
Offline
If anyone doesn't want to build:
http://omploader.org/vNTR1eg/vlc-1.1.2- … pkg.tar.xz
Works for me.
฿ 18PRsqbZCrwPUrVnJe1BZvza7bwSDbpxZz
Offline
It just me or XBMC-svn + new xvba freezes and will not play anything? with xvba 6.11 works well, less for 1080p videos are "cut"
Offline
Hi. I'm trying to use vaapi but I'm having problems with x264 videos. When I run "mplayer -vo vaapi:gl -va vaapi [file.mkv]", this is what I get:
and here the output from mplayer:
Warning unknown option device at line 5
MPlayer SVN-r31722-4.5.0 (C) 2000-2010 MPlayer Team
158 audio & 335 video codecs
mplayer: could not connect to socket
mplayer: No such file or directory
Failed to open LIRC support. You will not be able to use your remote control.
Playing Kung.Fu.Panda.2008.720P.BLURAY.X264-OUTDATED.mkv.
libavformat file format detected.
[matroska @ 0x3446ef0] Estimating duration from bitrate, this may be inaccurate
[lavf] stream 0: video (h264), -vid 0
[lavf] stream 1: audio (ac3), -aid 0
VIDEO: [H264] 1280x544 0bpp 24.000 fps 0.0 kbps ( 0.0 kbyte/s)
SUB: Added subtitle file (1): ./Kung.Fu.Panda.2008.720P.BLURAY.X264-OUTDATED.srt
[vo_vaapi] Using OpenGL rendering
libva: libva version 0.31.1-sds1
Xlib: extension "XFree86-DRI" missing on display ":0.0".
libva: va_getDriverName() returns 0
libva: Trying to open /usr/lib/va/drivers/fglrx_drv_video.so
libva: va_openDriver() returns 0
==========================================================================
Opening video decoder: [ffmpeg] FFmpeg's libavcodec codec family
[VD_FFMPEG] VA API accelerated codec.
Selected video codec: [ffh264] vfm: ffmpeg (FFmpeg H.264)
==========================================================================
==========================================================================
Opening audio decoder: [ffmpeg] FFmpeg/libavcodec audio decoders
AUDIO: 48000 Hz, 2 ch, s16le, 640.0 kbit/41.67% (ratio: 80000->192000)
Selected audio codec: [ffac3] afm: ffmpeg (FFmpeg AC-3)
==========================================================================
AO: [oss] 48000Hz 2ch s16le (2 bytes per sample)
Starting playback...
Unsupported PixelFormat 61
[VD_FFMPEG] Trying pixfmt=1.
Movie-Aspect is 2.35:1 - prescaling to correct movie aspect.
VO: [vaapi] 1280x544 => 1280x544 H.264 VA-API Acceleration
[VD_FFMPEG] XVMC-accelerated MPEG-2.
A: 28.2 V: 28.2 A-V: -0.030 ct: 0.061 0/ 0 2% 25% 0.7% 22 0
Any idea of what I'm doing wrong?
Edit: here some extra info
Arch: x86_64
Video: ATI HD5850 - Catalyst drivers
mplayer-vaapi 20100713-3
libva 0.31.1_p4-1
ffmpeg 24460-2
xvba-video 0.7.2-1
Thanks
Last edited by javonera (2010-08-06 21:54:22)
Offline
-= Some news =-
[catalyst] repo update:
* xvba-video is now in version 7.3;
* mplayer-vaapi package rebuilded;
It should work fine with new libva package from [extra] which has some strange versioning now - 1.0.4 (vainfo is showing 0.31.1) umm... what?
* you can probably remove [new_xvba] from pacman.conf - subtitles are fine with new extra's libva, plus do
# pacman -S ffmpeg
if it wasn't updated from [extra] after pacman -Syu
* and vlc is working
btw: i had to change PKGBUILD of xvba-video (drivers are now in /usr/lib/dri/) to make it work with extra's packages - so if there's a problem - as always write it down
Any idea of what I'm doing wrong?
Actually it looks like you aren't doing anything wrong. It's more like your gpu isn't well supported... or maybe it has problems with this particular mkv file.
Remove new_xvba repo, and try newer packages from catalyst and extra repo - maybe this will help...
Last edited by Vi0L0 (2010-08-07 03:50:40)
Offline
Thank you! It works fine!
[henry@HENRY-NB ~]$ vainfo
libva: libva version 0.31.1
Xlib: extension "XFree86-DRI" missing on display ":0.0".
libva: va_getDriverName() returns 0
libva: Trying to open /usr/lib/dri/fglrx_drv_video.so
libva: va_openDriver() returns 0
vainfo: VA API version: 0.31
vainfo: Driver version: Splitted-Desktop Systems XvBA backend for VA API - 0.7.3
vainfo: Supported profile and entrypoints
VAProfileMPEG2Simple : VAEntrypointIDCT
VAProfileMPEG2Main : VAEntrypointIDCT
VAProfileH264High : VAEntrypointVLD
VAProfileVC1Advanced : VAEntrypointVLD
Offline
-= Some news =-
[catalyst] repo update:
* xvba-video is now in version 7.3;
* mplayer-vaapi package rebuilded;
It should work fine with new libva package from [extra] which has some strange versioning now - 1.0.4 (vainfo is showing 0.31.1) umm... what?* you can probably remove [new_xvba] from pacman.conf - subtitles are fine with new extra's libva, plus do
# pacman -S ffmpeg
if it wasn't updated from [extra] after pacman -Syu
* and vlc is workingbtw: i had to change PKGBUILD of xvba-video (drivers are now in /usr/lib/dri/) to make it work with extra's packages - so if there's a problem - as always write it down
javonera wrote:Any idea of what I'm doing wrong?
Actually it looks like you aren't doing anything wrong. It's more like your gpu isn't well supported... or maybe it has problems with this particular mkv file.
Remove new_xvba repo, and try newer packages from catalyst and extra repo - maybe this will help...
No changes with updates. It happens with all x264 files, so you must be right about my gpu isn't well supported. Guess I'll have to wait newer drivers.
Thanks Vi0L0.
Offline
Offline
"New" driver:
Thanks for info!
New Updated Driver serves OpenCL 1.1 support.
I've updated AUR's catalyst-test with this new stuff.
+ NOTE: before building please download ati_catalyst_10.7b_opencl1.1_linux.zip using this site:
http://support.amd.com/us/kbarticles/Pa … river.aspx
and move it next to PKGBUILD
Hopefully without the black 2D artifacts ...
It looks like artifacts are still appearing.
Last edited by Vi0L0 (2010-08-09 02:38:24)
Offline
X.Org Server 1.9 Is Hitting Ubuntu 10.10 Soon
Yeeeeeeeeaaah
Offline
X.Org Server 1.9 Is Hitting Ubuntu 10.10 Soon
Yeeeeeeeeaaah
Good for us, bad for AMD/Ati (now they have to support it).
Offline
"No risk = no fun"
It's been some time since catalyst package got hooked and overall it looks pretty good - only little (already fixed) cosmetic bugs, lots of happy people .
But there's one problem - our auto-recompilation of fglrx module isn't working as good as it should...
Problem: when kernel26 is hitting some main update (ie. 2.6.34-ARCH -> 2.6.35-ARCH) fglrx module won't be auto-recompiled because kernel26-headers is installed after kernel26 package, so our hook doesn't have headers and module cannot be built.
Idea: What if we force kernel26-headers to be updated at first?
Solution: add kernel26-headers to SyncFirst array of /etc/pacman.conf
I believe it's a good idea, although it could be problematic ie. when mirror you are using didn't synced completely and you cannot reach this package, but i believe that we are all using more than one mirror, with official ftp.archlinux.org at the bottom of the list ;-).
So i'm pretty sure that it will work fine.
We have already said A and B, so it's probably a time to say C and finally have automatic catalyst ;-).
What are you thinking about that idea and solution? For every sceptic i got a candy .
Ofcourse i already tested it and i've got working solution that will add/remove headers automatically to/from pacman.conf .
I also think it's good idea to add to repo new package called catalyst-module-only which will contain only fglrx module compiled for current kernel26 for people who have problems with module compilation (didn't heard about anything like that, but still it could be useful).
Offtopic: as an off-topic i would like to show you brilliant band from France called 'Justice' and their mix. I'm not fan of techno/electric/house but this set is imo great, complicated and not boring at all (c'mon, at least wait for woman's vocal) ;P.
Last edited by Vi0L0 (2010-08-10 14:23:16)
Offline
About that kernel26 update... When the new kernel will hit core, is it enough to use this catalyst_module_build (or how it is called)?
Also: I'm happy with the repo. (besides the bugs from the driver itself)
Offline
Also: I'm happy with the repo. (besides the bugs from the driver itself)
+1. Vi0l0 made a fantastic job with the catalyst repo.
Offline
AUR's catalyst has been updated. Changelog :
- automatically adding (or removing when pacman -R catalyst) kernel26-headers to SyncFirst array of pacman.conf file, so now fglrx should be auto-recompiled successfully even when kernel26 is hitting main update. Now if pacman will ask you about updating kernel26-headers in first place - tell yes;
- generating log with debugging informations about every fglrx building, log is placed at /var/log/catalyst-install.log, every entry is only ~30 lines long, newest entries are stored at the bottom of the log;
- output of catalyst_build_module is now clean - it's giving only basic info:
'Ok.' - when module was compiled successfully;
'Failed!!! Check out log: /var/log/catalyst-install.log' - when failed.
I'm now thinking about updating [catalyst] repo with new build...
Edit: repo is updated now.
About that kernel26 update... When the new kernel will hit core, is it enough to use this catalyst_module_build (or how it is called)?
yes, after update of kernel26 and kernel26-headers type catalyst_build_module kernel_version:
# catalyst_build_module 2.6.35-ARCH
Or just add kernel26-headers to SyncFirst array of pacman.conf file, and it should be auto-recompiled
Last edited by Vi0L0 (2010-08-10 14:23:40)
Offline
I've posted this issue a few places around the net, but haven't really got a good answer yet.
It appears that something about fglrx is causing my system to hang when shutting down. Its all the way to "Power off" and then freezes. I ran some tests.
If I modprobe fglrx and try to shutdown without ever running an X server, it shuts down okay.
If I modprobe fglrx, run an X server, kill the X server and rmmod fglrx, the hang still occurs when I try to power down.
If I only run an X server on the vesa driver, everything works fine on shutdown.
Restarts of the computer always work fine -- it is only shutdowns that cause the hang.
I've looked through the logs and nothing seems to be different in them between a restart and shutdown. I believe the the hang may be occurring after logging has already stopped.
Running on Mobility Radeon 5830 (HP Envy 15 laptop Core i5) with Catalyst 10.7 -- Arch Linux x64, kernel 2.6.34. I don't know whether this is a weird bug with Evergreen, with Arch (I haven't tried another distro on this laptop yet), or just with my laptop. Any ideas?
Offline
Hi when I add fglrx to rc.conf when booting a see
/etc/rc.d/functions: line 192: /etc/rc.d/fglrx: no such file
And in catalyst 10.7 colors in some web pages are different. Like cyaneous is pink, black is dark blue.
Offline
Hi when I add fglrx to rc.conf when booting a see
/etc/rc.d/functions: line 192: /etc/rc.d/fglrx: no such file
And in catalyst 10.7 colors in some web pages are different. Like cyaneous is pink, black is dark blue.
Heja,
place fglrx in MODULES, not DAEMONS array
Offline
Ow.. thanks... But still problem with colors.
Offline
The xorg-server-1.8-catalyst-maximize-fix package in AUR has been updated to version 1.8.2-1. I usually wait for [extra] to catch up with releases, but 1.8.2 has been out for almost 2 months now and there is simply no reason not to update.
Get it here: http://aur.archlinux.org/packages.php?ID=36328
@Vi0L0: If I recall correctly, you have this package in [catalyst]. If so, can you update it as well?
It is better to keep your mouth shut and be thought a fool than to open it and remove all doubt. (Mark Twain)
Offline
The xorg-server-1.8-catalyst-maximize-fix package in AUR has been updated to version 1.8.2-1. I usually wait for [extra] to catch up with releases, but 1.8.2 has been out for almost 2 months now and there is simply no reason not to update.
Get it here: http://aur.archlinux.org/packages.php?ID=36328
@Vi0L0: If I recall correctly, you have this package in [catalyst]. If so, can you update it as well?
There's no need of updating from 1.8.2-rc2 to 1.8.2 because there is no difference between them.
Quoting xserver 1.8.2 announce:
"no new commits over RC2"
Last edited by Vi0L0 (2010-08-12 14:46:15)
Offline