[Developers] Fwd: svn head .dmg

[Bart Matthaei]

Tom's build (the one on the website) already has your patch applied.

I've also fixed this in the trunk.

Tom: you said you applied 'a version' of Brion's patch. Can you send me
your version of Controller.m so I can check what you did and if it improves
anything over Brion's patch? If so, I can also change that and commit.

Cheers,

Bart

On Fri, Jan 19, 2007 at 12:35:04AM -0800, Brion Vibber wrote:
> Bart Matthaei wrote:
> > What do you guys think. Should I place it as a beta on the website so
> > people can use it?
> 
> If it's current SVN head, then it has a crashing bug and may crash
> frequently, at least for some people.
> 
> My first home build crashed a couple times a day on my MacBook (crashes
> strongly associated with putting the machine to sleep) until I tracked
> it down and fixed the bug.
> 
> The bug is an uninitialized pointer being dereferenced due to incorrect
> handling of the case where no unlocked keys are available on the
> keychain during non-interactive checks.
> 
> 
> I posted my one-line patch against trunk and a link to a binary I built
> from it in December:
> 
> http://www.sshkeychain.org/pipermail/developers/2006-December/000040.html
> 
> I've been comfortably using my patched build on my MacBook for the last
> three weeks with nary a problem. It would be super if the patch could be
> applied...
> 
> 
> 
> Here's a stack trace of the crash, if you don't believe me... ;)
> 
> Exception:  EXC_BAD_ACCESS (0x0001)
> Codes:      KERN_INVALID_ADDRESS (0x0001) at 0x24c95d01
> 
> Thread 1 Crashed:
> 0   SSHKeychain              	0x0000885a -[Controller
> askPassphrase:withInteraction:] + 1402 (Controller.m:453)
> 1   libobjc.A.dylib          	0x90a59d76 objc_msgSendv + 54
> 2   com.apple.Foundation     	0x925ff5d7 -[NSInvocation invoke] + 1341
> 3   com.apple.Foundation     	0x9264c74f -[NSConnection
> dispatchInvocation:] + 487
> 4   com.apple.Foundation     	0x9264ab51 -[NSConnection
> handleRequest:sequence:] + 1273
> 5   com.apple.Foundation     	0x9264a3ae -[NSConnection
> handlePortCoder:] + 922
> 6   com.apple.Foundation     	0x92649fef -[NSConcretePortCoder dispatch]
> + 188
> 7   com.apple.Foundation     	0x92649a4c __NSFireMachPort + 307
> 8   com.apple.CoreFoundation 	0x9083b3c5 __CFMachPortPerform + 136
> 9   com.apple.CoreFoundation 	0x9082b66d CFRunLoopRunSpecific + 2904
> 10  com.apple.CoreFoundation 	0x9082ab0e CFRunLoopRunInMode + 61
> 11  com.apple.Foundation     	0x926b8e14 -[NSConnection run] + 214
> 12  com.apple.Foundation     	0x925f836c forkThreadForFunction + 123
> 13  libSystem.B.dylib        	0x90023d87 _pthread_body + 84
> 
> 
> and my scribbled notes about the function during debugging...
> 
> if interaction is false
> and [searchList count] is 0
> then SecKeychainFindGenericPassword doesn't get run
> 
> caught it!
> interaction is NO
> 
> 
> -- brion vibber (brion @ pobox.com / brion @ wikimedia.org)
> _______________________________________________
> developers mailing list
> developers at sshkeychain.org
> http://www.sshkeychain.org/cgi-bin/mailman/listinfo/developers
> 

-- 
Bart Matthaei                      bart at ambrero.nl

Ambrero Software
http://www.ambrero.nl/