|
Frequent crashes on Mountain Lion with example patch!I am getting desperate while experiencing frequent crashes and even kernel panics on Mountain Lion (10.8.0 and 10.8.2) with an (quasi-official) qtz example patch! If simply starting the patch AsteroidsQC.qtz by Gary C Martin (from http://quartzcomposer.com/compositions), it takes less than 3 minutes... and than the crash pattern is always the same -- no matter, if executed by the Finder, QuickTimePlayer or Quartz Composer. Crashes however only occur on my iMac (3.4GHz, i7, AMD Radeon HD 6970M 2048 MB) but not on my old MacBook Pro (2.53GHz, Core 2 Duo, NVIDIA GeForce 9600M GT 512 MB) while both running 10.8.2) What can i do? Might this be due to a hardware defect of the graphic card? Would already be helpful, if people on ML would like to just run the above mentioned example patch and note/confirm the results. Crashes look like these: Thread 9 Crashed:: com.apple.CoreAnimation.CAOpenGLLayer 0 libGPUSupportMercury.dylib 0x000000010e3211c6 gldObjectUnpurgeable + 189 1 GLEngine 0x0000000111bffa1d gleObjectUnpurgeable + 241 2 GLEngine 0x0000000111b21103 glObjectUnpurgeableAPPLE_Exec + 266 3 com.apple.QuartzComposer 0x00007fff998caa2e -[QCImageTextureBuffer setPurgeable:] + 612 or Thread 7 Crashed:: com.apple.CoreAnimation.CAOpenGLLayer 0 libGPUSupportMercury.dylib 0x0000000113fcee87 gldObjectPurgeable + 339 1 GLEngine 0x0000000113aab6c7 gleObjectPurgeable + 393 2 GLEngine 0x00000001139ccfa9 glObjectPurgeableAPPLE_Exec + 266 3 com.apple.QuartzComposer 0x00007fff998ca91b -[QCImageTextureBuffer setPurgeable:] + 337 etc.. Kernel panic locks report that: Kernel Extensions in backtrace: com.apple.iokit.IOAcceleratorFamily(19.0.26)[7DC411F1-A646-3A4A-B160-A1492F5EC0ED]@0xffffff7f90635000->0xffffff7f9067bfff dependency: com.apple.iokit.IOPCIFamily(2.7.2)[B1B77B26-7984-302F-BA8E-544DD3D75E73]@0xffffff7f8efae000 dependency: com.apple.iokit.IONDRVSupport(2.3.5)[86DDB71C-A73A-3EBE-AC44-0BC9A38B9A44]@0xffffff7f8f4d0000 dependency: com.apple.iokit.IOGraphicsFamily(2.3.5)[803496D0-ADAD-3ADB-B071-8A0A197DA53D]@0xffffff7f8f48d000 com.apple.AMDRadeonAccelerator(1.0)[D89E4100-9C89-3C99-B34B-F7790E1A0B8B]@0xffffff7f90687000->0xffffff7f90a54fff dependency: com.apple.iokit.IOAcceleratorFamily(19.0.26)[7DC411F1-A646-3A4A-B160-A1492F5EC0ED]@0xffffff7f90635000 dependency: com.apple.iokit.IOPCIFamily(2.7.2)[B1B77B26-7984-302F-BA8E-544DD3D75E73]@0xffffff7f8efae000 dependency: com.apple.iokit.IONDRVSupport(2.3.5)[86DDB71C-A73A-3EBE-AC44-0BC9A38B9A44]@0xffffff7f8f4d0000 dependency: com.apple.iokit.IOGraphicsFamily(2.3.5)[803496D0-ADAD-3ADB-B071-8A0A197DA53D]@0xffffff7f8f48d000 BSD process name corresponding to current thread: Quartz Composer or Kernel Extensions in backtrace: com.apple.iokit.IOAcceleratorFamily(19.0.26)[7DC411F1-A646-3A4A-B160-A1492F5EC0ED]@0xffffff7fa6835000->0xffffff7fa687bfff dependency: com.apple.iokit.IOPCIFamily(2.7.2)[B1B77B26-7984-302F-BA8E-544DD3D75E73]@0xffffff7fa51ae000 dependency: com.apple.iokit.IONDRVSupport(2.3.5)[86DDB71C-A73A-3EBE-AC44-0BC9A38B9A44]@0xffffff7fa56d0000 dependency: com.apple.iokit.IOGraphicsFamily(2.3.5)[803496D0-ADAD-3ADB-B071-8A0A197DA53D]@0xffffff7fa568d000 com.apple.AMDRadeonAccelerator(1.0)[D89E4100-9C89-3C99-B34B-F7790E1A0B8B]@0xffffff7fa6887000->0xffffff7fa6c54fff dependency: com.apple.iokit.IOAcceleratorFamily(19.0.26)[7DC411F1-A646-3A4A-B160-A1492F5EC0ED]@0xffffff7fa6835000 dependency: com.apple.iokit.IOPCIFamily(2.7.2)[B1B77B26-7984-302F-BA8E-544DD3D75E73]@0xffffff7fa51ae000 dependency: com.apple.iokit.IONDRVSupport(2.3.5)[86DDB71C-A73A-3EBE-AC44-0BC9A38B9A44]@0xffffff7fa56d0000 dependency: com.apple.iokit.IOGraphicsFamily(2.3.5)[803496D0-ADAD-3ADB-B071-8A0A197DA53D]@0xffffff7fa568d000 BSD process name corresponding to current thread: Quartz Composer For further details see attachment. |
I think you are ran into a known issue with QC on 10.8 : https://openradar.appspot.com/12105667
Indeed, that clarifies a lot -- thank you for pointing me there and linking the issue right back to this thread.
Wondered why I wasn't able to reproduce this particular problem - was running on an nVidia 9400, not an ATI card form factor.
It just appears that smokris filed a bug based on this thread.
Asteroids.qtz has an old Tiger era js patch, iirc. Other users have reported similar issues with that version of the JS patch. It's weird to see the crash at that place, but I've had similar erroneous backtraces with JS patch crashes.
I think the bug may be that they didn't handle deprecating the patch gracefully - there's not a great way to do it with QC either, besides the way Kineme handled it with the older GL Tools patches, where users get a prompt that there's a more recent version. With the js patch, this would be impossible, because the script has to be changed between versions.