Summary: | x11-base/xorg-server-1.8.0 server crashes randomly or if I use xrandr | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Zoltan Puskas <zoltan> |
Component: | [OLD] Unspecified | Assignee: | Gentoo X packagers <x11> |
Status: | RESOLVED NEEDINFO | ||
Severity: | critical | CC: | asturm, darkbasic, Hugo.Mildenberger, igor.poboiko, johannes.hirte, meyerm, zhalassy |
Priority: | High | ||
Version: | unspecified | ||
Hardware: | AMD64 | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Attachments: |
kdm.log of the few last crashes
the pervious xorg.0.log with crash details at the end New log with errors Xorg 1.8.2, kde 4.4, ati rhd 4350, moving a window to the other screen, crash logs |
Description
Zoltan Puskas
2010-05-06 12:18:02 UTC
Created attachment 230589 [details]
kdm.log of the few last crashes
Created attachment 230591 [details]
the pervious xorg.0.log with crash details at the end
(In reply to comment #2) > Created an attachment (id=230591) [details] > the pervious xorg.0.log with crash details at the end > Could be fixed with soon-to-be-released xorg-server-1.8.1: http://cgit.freedesktop.org/xorg/xserver/commit/?id=41bdb6c003cca3ef0ff88d9c7de318115bab1ba2 Created attachment 232309 [details]
New log with errors
Both X restart when using xrandr, and random freezes continue. I have not been using xrandr (or any external display for a time now), but I did a try for every new xorg release.
The attached kdm.log contains info about the freezes. Apparently the Inel drivers are to blame. I get (see also log):
X: intel_bufmgr_gem.c:900: drm_intel_gem_bo_unreference_locked_timed: Assertion `((&bo_gem->refcount)->atomic) > 0' failed.
[mi] EQ overflowing. The server is probably stuck in an infinite loop.
So the problem persists in both xorg-1.8.1 and xorg-1.8.1-r1. And it still happens randomly.
(In reply to comment #4) > Created an attachment (id=232309) [details] > New log with errors > > Both X restart when using xrandr, and random freezes continue. I have not been > using xrandr (or any external display for a time now), but I did a try for > every new xorg release. > > The attached kdm.log contains info about the freezes. Apparently the Inel > drivers are to blame. I get (see also log): > > X: intel_bufmgr_gem.c:900: drm_intel_gem_bo_unreference_locked_timed: Assertion > `((&bo_gem->refcount)->atomic) > 0' failed. > [mi] EQ overflowing. The server is probably stuck in an infinite loop. > > So the problem persists in both xorg-1.8.1 and xorg-1.8.1-r1. And it still > happens randomly. > Please post this info at bugzilla.kernel.org, this is the first time I see any log output concerning the X freezes and it seems to be an issue with the kernel drivers. Strangely enough, I couldn't reproduce any freezes anymore on my system, even though going back and forth in kernel history as well as xf86-video-intel versions. I can't imagine what 'solved' it (besides, I'm using zen-sources), however I've still got the xrandr-X-reset issue. Freezing frequency also seems to be vary over users' systems and kernel .configs as someone reported less freezes when adapting someone elses .config. There are lots of peoples with this problem (including me). Please see this thread: http://forums.gentoo.org/viewtopic-t-820551-start-200.html The freezes are not XServer-1.8 dependend. I've had this with xorg-server-1.7.7 too. (In reply to comment #7) > The freezes are not XServer-1.8 dependend. I've had this with xorg-server-1.7.7 > too. > Yes, the freezes also occur with xorg-server-1.8.2, latest live builds of xf86-video-intel, libdrm, dri2proto, as well as mesa-9999 and 2.6.35_rc5-git5-drm system, as soon as hardware rendering is turned on. With software rendering, the system is absolutely stable. Last time my system froze, I was able to switch to back to tty1 and catch these errors: intel_bufmgr_gem.c:1010: Error setting to CPU domain 10666: Input/output error intel_bufmgr_gem.c:1010: Error setting to CPU domain 10474: Input/output error intel_bufmgr_gem.c:1088: Error setting domain xxxx: Input/output error I am firing up X through startx which gives me more verbose info. 2010/7/21 <bugzilla-daemon@gentoo.org>: > Yes, the freezes also occur with xorg-server-1.8.2, latest live builds of > xf86-video-intel, libdrm, dri2proto, as well as mesa-9999 and > 2.6.35_rc5-git5-drm system, as soon as hardware rendering is turned on. With > software rendering, the system is absolutely stable. Fucking intel drivers, I just emerged xorg-server 1.8.2, xf86-video-intel 2.12, libdrm git and 2.6.25-rc5... :-( Couldn't someone compile the relevant parts with -g, put "splitdebug installsources" in FEATURES and enable core dumps to go after the reason? libglx, which is sometimes directly involved here has at least one hidden memory bug (for example bug #320055). I think I have the same problem, but I don't think it's Intel driver related, more like the GLX engine is faulty. I have an ATI Radeon HD 4350, driven with the xorg's opensource "ati" driver. I have two screens. I use a single virtual screen with xrandr, kde 4.4 on it. If I grab any window, waiting it to fade, then grabbing to the other monitor, as soon as the mouse pointer touches the edge of the source monitor, it crashes X. If i use only one monitor, it never crashes. If i use IceWM, it never crashes. Will attach my logs. Created attachment 246681 [details]
Xorg 1.8.2, kde 4.4, ati rhd 4350, moving a window to the other screen, crash logs
kwin is putting a lot of pressure on xorg and X drivers. Please each file a *separate* bug if you can reproduce with the latest ~arch versions of all the Xorg stack. Thanks |