| Summary: | [GM965] X segfaults in intel_renderbuffer when closing googleearth | ||
|---|---|---|---|
| Product: | Mesa | Reporter: | Albert Damen <albrt> |
| Component: | Drivers/DRI/i965 | Assignee: | Ian Romanick <idr> |
| Status: | RESOLVED FIXED | QA Contact: | |
| Severity: | critical | ||
| Priority: | high | Keywords: | patch |
| Version: | git | ||
| Hardware: | x86-64 (AMD64) | ||
| OS: | Linux (All) | ||
| Whiteboard: | |||
| i915 platform: | i915 features: | ||
| Attachments: |
Xorg.log
xorg.conf Stacktrace Possible fix |
||
Created attachment 28001 [details]
xorg.conf
Created attachment 28002 [details]
Stacktrace
7.5 is fine. Mass version move, cvs -> git The same crash still happens with latest versions from git (29 Aug). New versions are: -- xf86-video-intel: 7c48c21b22bf5862c5a35bda1635753cc5a7197c (master) -- xserver: 24716b9254fa7d609792596723a192bb044a7d3f (1.6.99.1) -- mesa: da1248bee5471f8da2277118a23b53d308721fca (7.6-devel) -- libdrm: 73b59c894380995a2889b98e79acadd2da0bb237 (2.4.13) -- kernel: 2.6.31-8-generic (based on 2.6.31-rc7) Created attachment 29046 [details] [review] Possible fix I have done some further debugging, by setting a watch on the intel_context's driDrawable->refcount in gdb: - While googleearth is running, refcount is 2 and does not change - When I close googleearth, the driContext unbinds from the drawable (driUnbindContext). Refcount is now 1. - Then driDestroyDrawable reduces refcount to 0. The drawable will now be destroyed and driContext->driDrawablePriv is set to NULL. driDrawable of the intel_context is not changed. - Finally the intel_context is destroyed (intelDestroyContext). As driDrawable still points to the drawable, this wants to destroy the drawable again, which fails (sigsegv). Assuming the intel_context can only have a valid drawable if the driContext has a valid drawable, I added a NULL check on driContextPriv->driDrawablePriv. This solves the X crash for me. See attached patch. (In reply to comment #6) > Created an attachment (id=29046) [details] > Possible fix > > I have done some further debugging, by setting a watch on the intel_context's > driDrawable->refcount in gdb: > > - While googleearth is running, refcount is 2 and does not change > - When I close googleearth, the driContext unbinds from the drawable > (driUnbindContext). Refcount is now 1. > - Then driDestroyDrawable reduces refcount to 0. The drawable will now be > destroyed and driContext->driDrawablePriv is set to NULL. driDrawable of the > intel_context is not changed. > - Finally the intel_context is destroyed (intelDestroyContext). As driDrawable > still points to the drawable, this wants to destroy the drawable again, which > fails (sigsegv). > > Assuming the intel_context can only have a valid drawable if the driContext has > a valid drawable, I added a NULL check on driContextPriv->driDrawablePriv. This > solves the X crash for me. See attached patch. > Based on this description, I think this is a duplicate of bug #23418. I wish I had read this before working on that bug. It would have gone much quicker. :) Anyway, could you verify that the following commit fixes this problem in your environment: commit 2921a2555d0a76fa649b23c31e3264bbc78b2ff5 Author: Ian Romanick <ian.d.romanick@intel.com> Date: Wed Sep 16 07:39:58 2009 -0700 intel: Deassociated drawables from private context struct in intelUnbindContext The generic DRI infrastructure makes sure that __DRIcontextRec::driDrawablePriv and __DRIcontextRec::driReadablePriv are set to NULL after unbinding a context. However, the intel_context structure keeps cached copies of these pointers. If these cached pointers are not NULLed and the drawable is actually destroyed after unbinding the context (typically by way of glXDestroyWindow), freed memory will be dereferenced in intelDestroyContext. This should fix bug #23418. I updated to mesa 7.7-devel, commit f911d196cf, which includes your fix. This indeed solved the crash. Thanks! |
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.
Created attachment 28000 [details] Xorg.log Bug description: When closing googleearth, X crashes and is restarted. Program received signal SIGSEGV, Segmentation fault. 0x00007f167e64d65c in intel_renderbuffer (rb=0x404) at ../intel/intel_fbo.h:89 System environment: -- chipset: GM965 (8086:2a02) -- system architecture: 64-bit -- xf86-video-intel: bb3007384298cb57625ec0b3868dff9b23568f3e (2.7.99.902) -- xserver: 48a9d65b88f56d1f8ab3bf824a4fe48c2f68725f (1.6.99.1) -- mesa: 8c30292a6e48448318d84582df876f35c490f968 (7.6-devel) -- libdrm: eea95ed8af24300e5a5d2489dfe0d73c24300651 (2.4.12) -- kernel: 2.6.31-4-generic (based on 2.6.31-rc4) -- Linux distribution: Ubuntu Karmic (development version) -- Machine or mobo model: Compal IFL91 laptop (T7500 CPU, 4GB Ram) -- Display connector: LVDS -- googleearth: 5.0.11337.1968 -- KMS enabled Reproducing steps: Run googleearth, zoom in (lower altitude) and move around a bit. Then close googleearth and see X crash.