> I'm hitting a bug in one of my composition and i'm having a hard time > figuring out where the problem lies. > There seems to be a memory leak somewhere, and it crashes after more than 4 > hours. > I saved the crashlog this morning, the crash log refers to the composition > running as a QB application but I've had crashes inside the editor too.
It would be handy to have additional crash logs - leaks don't often crash identically each time. > Can anyone help me decypher the crashlog? > Process: Cobalt_132_NightWorks_28_2 [6955] > Path: > /Users/USER/Desktop/Cobalt_132_NightWorks_28_2.app/Contents/MacOS/Cobalt_132_NightWorks_28_2 > Identifier: com.mycompanyname.Cobalt_132_NightWorks_28_2 > Version: ??? (1.0) > Code Type: X86-64 (Native) You're running 64 bit (good) - that means a system memory leak isn't likely to be fatal. Further down, the total VM size is something like 3.3GB, which is relatively large, but not indicative of a leak in a 64bit process. > Parent Process: launchd [118] > > Date/Time: 2012-02-29 07:38:22.657 +0100 > OS Version: Mac OS X 10.7.2 (11C74) > Report Version: 9 > > Interval Since Last Report: 136852 sec > Crashes Since Last Report: 6 > Per-App Interval Since Last Report: 18303 sec > Per-App Crashes Since Last Report: 1 > Anonymous UUID: AF017B63-6D34-4966-9BD2-814149224020 > > Crashed Thread: 4 CVDisplayLink Thread 4 crashed, and you're using display link (good). > Exception Type: EXC_BAD_INSTRUCTION (SIGILL) > Exception Codes: 0x0000000000000001, 0x0000000000000000 looks like an intentional crash (SIGILL instead of SIGABRT, SIGTERM, or SIGINT). > Thread 4 Crashed:: CVDisplayLink > 0 com.apple.ATIRadeonX3000GLDriver 0x0000000105ef0fb2 0x105eeb000 + 24498 > 1 com.apple.ATIRadeonX3000GLDriver 0x0000000105f2ce3c 0x105eeb000 + 269884 The crash was in the drivers (ATI drivers, to be specific). It's unclear what's going on (because there are no symbols). Using some internal tools to symbolicate the driver, it's unclear what happened; the driver triggered an intentional crash because of an unknown return value when submitting a flush (a glFlush doesn't take any arguments, so it's not invalid input). It's possible an allocation failed internal to the driver, or it could be some other bug that caused an unexpected error to happen. > 2 GLEngine 0x0000000105c267ff glFlushRender_Exec + > 68 There's the glFlush (glFlushRenderAPPLE most likely) ... > 3 com.apple.CoreImage 0x00007fff8641f203 > fe_accel_context_flush(fe_accel_context_struct*) + 39 > 4 com.apple.CoreImage 0x00007fff8641ec01 > fe_accel_context_release(fe_accel_context_struct*) + 37 > 5 com.apple.CoreImage 0x00007fff8642b19c -[FEOpenGLContext > updateBuffer] + 242 > 6 com.apple.CoreImage 0x00007fff8642b62b -[FEOpenGLContext > _begin] + 99 > 7 com.apple.CoreImage 0x00007fff86463bb0 > FEApplyTreeNode::render1(FETreeContext*, FEShape const&, > fe_kernel_target_struct*, int, float*, FETreeTexture*) + 1576 > 8 com.apple.CoreImage 0x00007fff864645a7 > FEApplyTreeNode::render2(FETreeContext*, FEShape const*, void (*)(FEContext*, > void*, FEFormat), void (*)(FEContext*, void*), void*) + 1597 > 9 com.apple.CoreImage 0x00007fff86461b19 > FETreeNode::render_(FETreeContext*, FEShape const*, void (*)(FEContext*, > void*, FEFormat), void (*)(FEContext*, void*), void*) + 149 > 10 com.apple.CoreImage 0x00007fff8646539e > FETreeNode::render(FETreeContext*, FEShape const&, void (*)(FEContext*, > void*, FEFormat), void (*)(FEContext*, void*), void*) + 108 > 11 com.apple.CoreImage 0x00007fff8646913a > FETreeContext::renderTree(FETreeNode*, FEShape const&, void (*)(FEContext*, > void*, FEFormat), void (*)(FEContext*, void*), void*) + 276 > 12 com.apple.CoreImage 0x00007fff8646944e > FETreeContext::renderImage_(FEImage*, CGRect, FEShape const*, > CGAffineTransform, CGColorSpace*, FEFormat, bool, void (*)(FEContext*, void*, > FEFormat), void (*)(FEContext*, void*), void*) + 688 > 13 com.apple.CoreImage 0x00007fff864696d7 > FETreeContext::renderImage(FEImage*, CGRect, FEShape const*, > CGAffineTransform, CGColorSpace*, FEFormat, bool, void (*)(FEContext*, void*, > FEFormat), void (*)(FEContext*, void*), void*) + 375 > 14 com.apple.CoreImage 0x00007fff86440377 -[FEImage(Internal) > _renderWithContext:bounds:transform:colorSpace:format:premultiplied:setupCallback:finishCallback:callbackData:] > + 247 > 15 com.apple.CoreImage 0x00007fff8643fdfa -[FEImage > renderToContext:bounds:transform:destinationFormat:colorSpace:] + 218 > 16 com.apple.CoreImage 0x00007fff863f653e > renderImage(FEContext*, CGRect, CGAffineTransform, unsigned int, void*) + 137 > 17 com.apple.CoreImage 0x00007fff8640fb66 > -[CIOpenGLContextImpl renderWithBounds:matrix:function:info:] + 258 > 18 com.apple.CoreImage 0x00007fff863f64a7 -[CIContextImpl > render:] + 168 > 19 com.apple.CoreImage 0x00007fff863f4728 -[CIContext > drawImage:inRect:fromRect:] + 1098 > 20 com.apple.CoreImage 0x00007fff863f4a51 -[CIContext > drawImage:atPoint:fromRect:] + 140 ... and it was called by CoreImage ... > 21 com.apple.QuartzComposer 0x00007fff871ebd90 > -[QCProvider_CoreImage > createTextureBufferForManager:withFormat:target:transformation:bounds:colorSpace:options:] > + 1368 ... who in turn was called by QC to create a texture from a CIImage. This is a normal operation, but it's possible that textures are leaking. Use OpenGL Driver Monitor to look at the number of textures, and see if it grows without bound. > VM Region Summary: > ReadOnly portion of Libraries: Total=191.5M resident=130.4M(68%) > swapped_out_or_unallocated=61.1M(32%) > Writable regions: Total=3.0G written=292.3M(9%) resident=865.4M(28%) > swapped_out=0K(0%) unallocated=2.2G(72%) > > REGION TYPE VIRTUAL > =========== ======= > CG backing stores 5260K > CG image 120K > CG shared images 1192K > CoreGraphics 16K > CoreServices 924K > IOKit 1.1G > IOKit (reserved) 3072K reserved VM address space > (unallocated) > JS JIT generated code 128.0M > JS JIT generated code (reserved) 896.0M reserved VM address space > (unallocated) > JS VM register file 4096K > JS garbage collector 208K > MALLOC 903.7M > MALLOC guard page 32K > Memory tag=240 4K > Memory tag=242 12K > Memory tag=251 36K > STACK GUARD 56.0M > Stack 11.0M > VM_ALLOCATE 16.2M > __CI_BITMAP 80K > __DATA 16.9M > __IMAGE 1256K > __LINKEDIT 49.8M > __TEXT 141.7M > __UNICODE 544K > mapped file 27.7M > shared memory 4304K > =========== ======= > TOTAL 3.3G > TOTAL, minus reserved VM space 2.4G As for system memory (not vram), this looks mostly ok - however, IOKit being 1.1GB is a little alarming (OpenGL textures are reported there I think), and it's right about at your GPU's limit... > Model: MacBookPro8,2, BootROM MBP81.0047.B24, 4 processors, Intel Core i7, > 2.4 GHz, 4 GB, SMC 1.69f3 > Graphics: AMD Radeon HD 6770M, AMD Radeon HD 6770M, PCIe, 1024 MB ... here the GPU reports 1GB of vram (and 4GB of system memory). You could try running this on a different GPU to see if the problem persists (preferably an intel or nvidia GPU, so you'll definitely get a different driver with different allocation/paging behavior). If it does, it could be a texture leak. If it doesn't persist, it might be a driver bug. -- Christopher Wright [email protected]
_______________________________________________ Do not post admin requests to the list. They will be ignored. Quartzcomposer-dev mailing list ([email protected]) Help/Unsubscribe/Update your Subscription: https://lists.apple.com/mailman/options/quartzcomposer-dev/archive%40mail-archive.com This email sent to [email protected]
