I've uploaded a simple test case here:
http://dl.getdropbox.com/u/550950/gc-crash.zip
On Sat, Oct 24, 2009 at 11:18 PM, Ken Ferry <kenfe...@gmail.com> wrote:
> If you have extracted this to a test project, does that mean you now
> have a reproducing case whose source you can share? If so, why don't
> you put it up somewhere.
>
> You could also try doing a Build and Analyze in Xcode. It is possible
> to make memory management mistakes in gc, and the analyzer will catch
> a large class of them.
>
> On Saturday, October 24, 2009, slasktrattena...@gmail.com
> <slasktrattena...@gmail.com> wrote:
>> Weird... It seems that sourceImage (the generic one only) is being
>> garbage collected in the middle of the drawing. If I disable GC just
>> before doing the drawing, and enable it again afterwards, the process
>> stops crashing.
>>
>> I verified this by making a new project, where I placed the
>> screensaver view in a normal window. With GC disabled, there were no
>> crashes. As soon as I enabled GC, it started crashing. Backtrace says:
>>
>> #0 0x91c008c5 in ripc_RemoveEntry ()
>> #1 0x91bdd7b0 in ripc_AcquireImage ()
>> #2 0x91bdb3be in ripc_DrawImage ()
>> #3 0x937eab60 in CGContextDrawImage ()
>> #4 0x92c4d3c7 in __-[NSImageRep
>> drawInRect:fromRect:operation:fraction:respectFlipped:hints:]_block_invoke_1
>> ()
>> #5 0x92c4ca4d in -[NSImageRep
>> drawInRect:fromRect:operation:fraction:respectFlipped:hints:] ()
>> #6 0x92c4b586 in __-[NSImage
>> drawInRect:fromRect:operation:fraction:respectFlipped:hints:]_block_invoke_1
>> ()
>> #7 0x92c48e1b in -[NSImage
>> _usingBestRepresentationForRect:context:hints:body:] ()
>> #8 0x92c48950 in -[NSImage
>> drawInRect:fromRect:operation:fraction:respectFlipped:hints:] ()
>> #9 0x92d06d9b in -[NSImage drawInRect:fromRect:operation:fraction:] ()
>> #10 0x92d335de in -[NSImage _composite:delta:fromRect:toPoint:] ()
>> #11 0x92d32af5 in -[NSImage compositeToPoint:fromRect:operation:] ()
>> #12 0x92d32a8d in -[NSImage compositeToPoint:operation:] ()
>> #13 0x00002acd in +[NSImage(Reflection)
>> appendReflectionToImage:fraction:] (self=0xa03195c0, _cmd=0x9324,
>> sourceImage=0x1251440, fraction=0.5) at
>> /Users/fabian/Lounge/Lounge/NSImage+Reflection.m:29
>> #14 0x000047b0 in -[MainController updateTrackInfo:] (self=0x1026cc0,
>> _cmd=0x908a, dict=0x123a260) at
>> /Users/fabian/Lounge/Lounge/MainController.m:379
>> #15 0x94b41b65 in __NSFireDelayedPerform ()
>> #16 0x93fcbeee in __CFRunLoopRun ()
>> #17 0x93fc9d34 in CFRunLoopRunSpecific ()
>> #18 0x93fc9b61 in CFRunLoopRunInMode ()
>> #19 0x94dbafec in RunCurrentEventLoopInMode ()
>> #20 0x94dbada3 in ReceiveNextEventCommon ()
>> #21 0x94dbac28 in BlockUntilNextEventMatchingListInMode ()
>> #22 0x92c10b99 in _DPSNextEvent ()
>> #23 0x92c1040e in -[NSApplication
>> nextEventMatchingMask:untilDate:inMode:dequeue:] ()
>> #24 0x92bd25fb in -[NSApplication run] ()
>> #25 0x92bca695 in NSApplicationMain ()
>> #26 0x00002108 in main (argc=0x1, argv=0xbffff7d4) at
>> /Users/fabian/Lounge/Lounge/main.m:13
>>
>> On Sat, Oct 24, 2009 at 7:36 PM, slasktrattena...@gmail.com
>> <slasktrattena...@gmail.com> wrote:
>>> Thanks for your reply.
>>>
>>> On Sat, Oct 24, 2009 at 6:03 PM, Jens Alfke <j...@mooseyard.com> wrote:
>>>>
>>>> It would be really good to get a backtrace. Are you launching the process
>>>> with gdb, or at least attaching before the crash?
>>>
>>> Yes. I'm launching with gdb and pass the "debug" argument to Screen
>>> Saver Engine. This way the screen saver is running on the desktop
>>> window level.
>>>
>>>> And after the crash, gdb
>>>> is confused and can't get info about the process?
>>>
>>> Sort of. I think the problem is no app can have keyboard focus as long
>>> as the screensaver is running/frontmost. So Xcode is not responding to
>>> keyboard events until I terminate the process.
>>>
>>>> If you run without gdb, do you get a regular "unexpectedly quit" alert with
>>>> the option to send a crash report (which includes a backtrace)?
>>>
>>> No. The process just hangs and I have to force a reboot.
>>>
>>>> Have you tried running with malloc in scribbling/heap-checking mode, to see
>>>> if anything funky's going on with the heap? ('man malloc' for details on
>>>> the
>>>> environment variables to set for this.)
>>>
>>> I'm not sure what to expect, but I did try just now to enable most of
>>> these settings and they didn't return anything that looked helpful in
>>> gdb.
>>>
>> _______________________________________________
>>
>> Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
>>
>> Please do not post admin requests or moderator comments to the list.
>> Contact the moderators at cocoa-dev-admins(at)lists.apple.com
>>
>> Help/Unsubscribe/Update your Subscription:
>> http://lists.apple.com/mailman/options/cocoa-dev/kenferry%40gmail.com
>>
>> This email sent to kenfe...@gmail.com
>>
>
_______________________________________________
Cocoa-dev mailing list (Cocoa-dev@lists.apple.com)
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
