Re: Advice on debugging a GC related crash
Re: Advice on debugging a GC related crash
- Subject: Re: Advice on debugging a GC related crash
- From: Harry Jordan <email@hidden>
- Date: Sat, 26 Sep 2009 18:39:47 +0100
Wow. Thank you Bill. That's ore than I could have hoped for. You'r
right of course, there was some unexpected code getting loaded:
0x1d3000 - 0x1dbffc +com.sweetpproductions.SafariCookies 0.8 (0.8)
<9739F20D-A5C1-015A-FB1F-B0F66CD64AB9> /Library/InputManagers/Safari
Cookies/Safari Cookies.bundle/Contents/MacOS/Safari Cookies
Everything else except for this input manager seems quite standard, so
I suspect this could be the culprit. For now I'm going to wait and see
if I get any more crash reports in this vein before I investigate any
further.
Thank you again
Harry
On 25 Sep 2009, at 16:01, Bill Bumgarner wrote:
On Sep 25, 2009, at 4:32 AM, Harry Jordan wrote:
Thread 3 Crashed: Dispatch queue: Garbage Collection Work Queue
0 com.apple.CoreFoundation 0x95c6c393
__CFTypeCollectionRelease + 83
1 com.apple.CoreFoundation 0x95c68eb8
__CFBasicHashStandardCallback + 296
2 com.apple.CoreFoundation 0x95c68b78 __CFBasicHashDrain +
568
3 com.apple.Foundation 0x92a20f64 -[NSCFDictionary
finalize] + 45
4 libobjc.A.dylib 0x9724618d finalizeOneObject + 59
5 libauto.dylib 0x9769d612 Auto::foreach_block_do
(auto_zone_cursor*, void (*)(void*, void*), void*) + 66
Check the crash log for libraries that you don't recognize that
aren't from Apple. It might be that the user has some random bit
of driverware that is buggy. That has been the typical source of a
crash like this where only 1 of many users see this. In particular,
there were a couple of random web cam type things that seemed to
have been the source of this kind of failure.
The other source has been bundle unloading. Unloading of bundles
in Objective-C is incredibly hard to get right. Hard to the point
of considering it truly unsupported. If a bundle was unloaded and
there was a reference to an NSString constant in that bundle and the
string had been put into a dictionary that was later collected after
unload, you'll see a crash like this.
The best evidence in this case will be a hole in the addresses of
the dylibs. If you are really lucky, register êx (on i386,
anyway) will contain an address that points to the hole.
In any case, ask the user what kind of software/hardware they have
installed that is of the non-app persuasion.
b.bum
_______________________________________________
Cocoa-dev mailing list (email@hidden)
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:
This email sent to email@hidden