Hi, while trying to play video with mplayer (video-output: Xv), xserver crashed. It also kept starting&crashing in loop (kdm managed).
Created attachment 6641 [details] xserver log Interesting points also might be the failed allocs of GART memory. Also not really sure what does following mean: (WW) I810(0): PRB0_HEAD (0x00000000) and PRB0_TAIL (0x00000110) indicate ring buffer not flushed
Oh, i should also note that i have xorg-server-1.1.1 and xf86-video-i810-1.6.5.
Was it an immediate crash when trying to play, or did you resize/move the window or do something else when it crashed ?? Also, the log you've uploaded is useless as it's not the initial crash message. You need to disable kdm and start X manually to get a good log.
Oh, bummer, will try to make it crash w/o kdm. Yeah, actually thinking of it, i resized the window, then everything froze and the server crashed.
any news ?
Created attachment 6674 [details] xserver log Hi, here's a a new crash log. I've started the X server with startx and was switching to/from fullscreen to make it crash. It didn't want to crash, then i tried mplayer's -vo sdl output (using Xv backend) and pressing 'n' key, that cycles the size of the video window. After 3-4 switches the X server locked-up, the screen stayed gray. I shutdown the notebook and copied the X server log. The output looks vaguely similar to the previous one to me tho'.
Can you download a test driver from http://www.fairlite.demon.co.uk/intel.html and install that, and get another crash log. It'll help debug the exact cause. Thanks.
Is your driver the same version as mine? Because I cannot make it crash. I will try to make it crash tomorrow.
Yes, it's 1.6.5 which extra debug enabled.
(In reply to comment #9) > Yes, it's 1.6.5 which extra debug enabled. Sorry for the delay - here are two crash logs, they don't differ that much. The testscase is still the same - switch fullscreen/aspec-ration in mplayer and bang!. Could it perhaps be caused by the failed allocs?
Created attachment 7173 [details] crash log (20060928) first crash log
Created attachment 7174 [details] second crash log
I suggest you get the current git repository for the xf86-video-intel driver and in i830_video.c at the very first line you'll see a #define statement. Change that value at the end of the line to a 1 and recompile the driver and install it. Then get the system to crash again which will produce much more detailed logs.
Created attachment 7178 [details] xorg log (git version) Here it is. While reading through the log, i've noticed that the gart allocs no longer fail. Anyway, hope it helps.
Created attachment 7184 [details] [review] Attempt to fix xv lockup Can you try this patch against the source code you downloaded to see if it helps.
Created attachment 7186 [details] Xorg log (git + patch) Well, crashed again - this time with gray screen :) BTW, I had to tweak the patch a little (.x2 to ->x2).
Created attachment 7188 [details] [review] Another attempt to fix xv lockup O.k. reverse that last patch and apply this, which will also output a little more debug if the problem persists.
Created attachment 7189 [details] [review] Yet another if xv2.patch fails try this, but make sure you revert xv2.patch fully before applying.
Created attachment 7190 [details] Xorg log (git + fix2) Unfortunately another crash.
Created attachment 7191 [details] Xorg log (git + fix3) Nice! No crash this time, except no video - just blue screen :)
Created attachment 7192 [details] [review] Another Whoops, made a bit of a typo with xv3. Try this.
Created attachment 7193 [details] Xorg log (git + fix4) Video now works, but unfortunately I managed to crash it :(
Created attachment 7194 [details] [review] Another5 Another to try, but..... Can you tell me a little bit more about what you are doing. From the logs it looks as though you are resizing to fullscreen then back to a smaller windows, then back up to fullscreen and to a window again. You do this multiple times until it crashes - is that right ? Is there anything different you are doing when the crash happens ?
Created attachment 7196 [details] Xorg log (git + fix5) Crashed again. Well I have fairly simple testcase. 1) run mplayer in fullscreen 2) go back to window-mode 3) resize the window wildly or change the aspect (n-key) ratio (changes the window size) 4) bang! Apart from these steps I'm not doing anything else. I usually takes one or two changes to crash.
Created attachment 7197 [details] [review] Another6 Another....
Created attachment 7200 [details] Xorg log (git + fix6) Again crash :( I was thinking, would access to my machine help you in any way? I could provide you with ssh access if you'd want.
Created attachment 7201 [details] [review] ugh ssh access wouldn't really help as once the box is hosed, I'd need to reboot etc. Anyway, it's easier for me to keep pushing test patches. It's actually guess work at the moment, as there's not an awful lot the logs are giving me, so I'm hoping to trigger some other event with the current modifications. Try this one though.
Created attachment 7202 [details] [review] keep you busy :-) Just to keep you busy, if that doesn't work - try this.
Created attachment 7204 [details] fix8 Both (fix7 and fix8) crashed. Btw i tried to ssh into the machine and everything is working, except X :) So you could do some sort of post-mortem analysis if such thing is possible :) Anyway, there's another thing I didn't mention - when resizing there's this "black flash" on the screen. I'll make a video with my camera - maybe it has some connection with it, maybe not :)
Created attachment 7205 [details] [review] last one for today try this.
Created attachment 7206 [details] Xorg log (git + fix9) Bingo! I've been torturing it for a long while and it didn't crash - usually it crahsed after 2nd switch! :) What did you change in this patch?
Excellent. Thanks for testing.
(In reply to comment #32) > Excellent. Thanks for testing. I should thank you! If you ever go to Prague, drop a message - I owe you a beer or cookie :-)
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.