I'm experiencing a reproducible lockup when I launch googleearth. The program hangs during the initialization and the system become unresponsive: I can't switch to a console and ctrl+alt+del can't reboot the machine. However I can log in from another machine in my network, and top indicates that googleearth-bin CPU usage is 99-100%; if I kill the process, X starts using 99-100% CPU and become unkillable. I recompiled xorg-server, mesa and xf86-video-ati and I attached gdb to the X process; I obtained a meaningless backtrace, but it pointed me to libdrm.so.2; recompiling libdrm, I obtained the backtrace attached to this bug. I'm currently using xorg-server 1.2.0, mesa 6.5.2, xf86-video-ati 6.6.3 and libdrm 2.3.0 on a ATI Radeon X550 (RV370) 5B63 (PCIE), Mapped VideoRAM: 262144 kByte (128 bit DDR SDRAM).
Created attachment 8588 [details] gdb log
Created attachment 8613 [details] drm log I enabled the drm debug with "echo 1 > /sys/module/drm/parameters/debug" and obtained a log of some 60 mb. Most of the log was the repetition of several hundred of thousands of [drm:radeon_freelist_get] done_age = 3134137 with some blocks like [drm:radeon_freelist_get] returning NULL! [drm:radeon_cp_dispatch_texture] radeon_cp_dispatch_texture: EAGAIN [drm:drm_ioctl] ret = fffffff5 [drm:drm_ioctl] pid=28444, cmd=0xc018644e, nr=0x4e, dev 0xe200, auth=1 = 3134137 every ~1000 lines. I removed most of them from the attached log.
Hello. Should I provide more information?
You are facing a lockup, which is somethings hard to debug. Unfortunetly there isn't much you can do if you can't investigate this by yourself (tracking done which command is triggering the lockup). Moreover r300 got a bunch of debugging reg but we don't know any (ATI/AMD do not provide any informations). Last change is that some dev can reproduce the lookup and investigate this (which might take time as with the limited amount of information we have, moreover such things is painful and time consuming). I plane my self to look at some r300 lockup in near future but i can't give any time frame.
Does your recent comment to bug 9446 imply that this is working for you now? If not, please attach the full xorg.conf and Xorg.0.log.
Oh, well. When I realized that compiz wasn't locking anymore in September I did try again with googleearth (precisely on September 15) and it still locked up immediately after displaying the globe. I was using xorg-server 1.4, mesa 7.0.1 xf86-video-ati-6.7.192 and drm from kernel 2.6.22. Now I'm using xorg-server 1.4, mesa 7.0.1 plus commit http://gitweb.freedesktop.org/?p=mesa/mesa.git;a=commit;h=32699696e31234c8d4e4b08f255ba2134ec12db5 xf86-video-ati 6.7.195 and drm from kernel 2.6.23; I reinstalled googleearth 10 minutes ago (version 4.2.198.2451 according to my package manager) and I couldn't make it crash under compiz while with metacity it doesn't starts and complains that: do_wait: drmWaitVBlank returned -1, IRQs don't seem to be working correctly. Try running with LIBGL_THROTTLE_REFRESH and LIBL_SYNC_REFRESH unset. When it works, with compiz, on console it displays lines after lines of this: wx=0.312690, wy=0.328990, rx=0.639990, ry=0.330000 gx=0.210000, gy=0.710000, bx=0.149990, by=0.059990 wx=0.312690, wy=0.328990, rx=0.639990, ry=0.330000 gx=0.210000, gy=0.710000, bx=0.149990, by=0.059990 wx=0.312690, wy=0.328990, rx=0.639990, ry=0.330000 gx=0.210000, gy=0.710000, bx=0.149990, by=0.059990 When I opened the bug report occasionally I couldn't reproduce the crash (maybe it did work 1 out of 10 times), so I'm not sure that the problem is really fixed, I'm going to try again after a reboot, or after some hours of use of the computer.
Created attachment 12218 [details] Xorg.0.log
Created attachment 12219 [details] xorg.conf
Giacomo, could you check if it hangs for you using method described here: http://lists.freedesktop.org/archives/xorg/2007-November/030139.html ? I'm seeing lockups of my ati mobile x300 and I'm wondering if it's the same problem.
(In reply to comment #9) > Giacomo, could you check if it hangs for you using method described here: > http://lists.freedesktop.org/archives/xorg/2007-November/030139.html ? That's bug 12815 FWIW.
(In reply to comment #10) > (In reply to comment #9) > > Giacomo, could you check if it hangs for you using method described here: > > http://lists.freedesktop.org/archives/xorg/2007-November/030139.html ? > > That's bug 12815 FWIW. > Sorry, I missed Arkadiusz's request. On topic, now, using xorg-server 1.4, mesa 7.0.2 and xf86-video-ati 6.7.196 (minus a commit, see bug 13231) and drm from kernel 2.6.23, googleearth works reliably with compiz and fails with do_wait: drmWaitVBlank returned -1, IRQs don't seem to be working correctly. Try running with LIBGL_THROTTLE_REFRESH and LIBL_SYNC_REFRESH unset. on metacity 2.20.1. Anyway it's not a lock anymore, so should I close the bug?
(In reply to comment #11) > On topic, now, using xorg-server 1.4, mesa 7.0.2 and xf86-video-ati 6.7.196 > (minus a commit, see bug 13231) and drm from kernel 2.6.23, googleearth works > reliably with compiz and fails with > > do_wait: drmWaitVBlank returned -1, IRQs don't seem to be working correctly. > Try running with LIBGL_THROTTLE_REFRESH and LIBL_SYNC_REFRESH unset. This is probably (related to) bug 12902. It should only hang for about 3 seconds and then continue working though, doesn't it? > on metacity 2.20.1. Anyway it's not a lock anymore, so should I close the bug? Yes, let's let this one rest in peace finally. :)
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.