Objective-c – SIGABRT AND SIGTERM THREAD in all the programs

The last two are identical; "atomic" is the default behavior (note that it is not actually a keyword; it is specified only by the absence of nonatomic -- atomic was added as a keyword in recent versions of llvm/clang).

Assuming that you are @synthesizing the method implementations, atomic vs. non-atomic changes the generated code. If you are writing your own setter/getters, atomic/nonatomic/retain/assign/copy are merely advisory. (Note: @synthesize is now the default behavior in recent versions of LLVM. There is also no need to declare instance variables; they will be synthesized automatically, too, and will have an _ prepended to their name to prevent accidental direct access).

With "atomic", the synthesized setter/getter will ensure that a whole value is always returned from the getter or set by the setter, regardless of setter activity on any other thread. That is, if thread A is in the middle of the getter while thread B calls the setter, an actual viable value -- an autoreleased object, most likely -- will be returned to the caller in A.

In nonatomic, no such guarantees are made. Thus, nonatomic is considerably faster than "atomic".

What "atomic" does not do is make any guarantees about thread safety. If thread A is calling the getter simultaneously with thread B and C calling the setter with different values, thread A may get any one of the three values returned -- the one prior to any setters being called or either of the values passed into the setters in B and C. Likewise, the object may end up with the value from B or C, no way to tell.

Ensuring data integrity -- one of the primary challenges of multi-threaded programming -- is achieved by other means.

Adding to this:

atomicity of a single property also cannot guarantee thread safety when multiple dependent properties are in play.

Consider:

 @property(atomic, copy) NSString *firstName;
 @property(atomic, copy) NSString *lastName;
 @property(readonly, atomic, copy) NSString *fullName;

In this case, thread A could be renaming the object by calling setFirstName: and then calling setLastName:. In the meantime, thread B may call fullName in between thread A's two calls and will receive the new first name coupled with the old last name.

To address this, you need a transactional model. I.e. some other kind of synchronization and/or exclusion that allows one to exclude access to fullName while the dependent properties are being updated.

status: this has been seen as recently as Mac OS 10.8 and Xcode 4.4.

tl;dr: This can occur in two contexts: when running on the device and when running on the simulator. When running on the device, disconnecting and reconnecting the device seems to fix things.

Mike Ash suggested

launchctl list|grep UIKitApplication|awk '{print $3}'|xargs launchctl remove

This doesn't work all the time. In fact, it's never worked for me but it clearly works in some cases. Just don't know which cases. So it's worth trying.

Otherwise, the only known way to fix this is to restart the user launchd. Rebooting will do that but there is a less drastic/faster way. You'll need to create another admin user, but you only have to do that once. When things wedge, log out as yourself, log in as that user, and kill the launchd that belongs to your main user, e.g.,

sudo kill -9 `ps aux | egrep 'user_id .*[0-9] /sbin/launchd' | awk '{print $2}'`

substituting your main user name for user_id. Logging in again as your normal user gets you back to a sane state. Kinda painful, but less so than a full reboot.

details:

This has started happening more often with Lion/Xcode 4.2. (Personally, I never saw it before that combination.)

The bug seems to be in launchd, which inherits the app process as a child when the debugger stops debugging it without killing it. This is usually signaled by the app becoming a zombie, having a process status of Z in ps.

The core issue appears to be in the bootstrap name server which is implemented in launchd. This (to the extent I understand it) maps app ids to mach ports. When the bug is triggered, the app dies but doesn't get cleaned out of the bootstrap server's name server map and as result, the bootstrap server refuses to allow another instance of the app to be registered under the same name.

It was hoped (see the comments) that forcing launchd to wait() for the zombie would fix things but it doesn't. It's not the zombie status that's the core problem (which is why some zombies are benign) but the bootstrap name server and there's no known way to clear this short of killing launchd.

It looks like the bug is triggered by something bad between Xcode, gdb, and the user launchd. I just repeated the wedge by running an app in the iphone simulator, having it stopped within gdb, and then doing a build and run to the ipad simulator. It seems to be sensitive to switching simulators (iOS 4.3/iOS 5, iPad/iPhone). It doesn't happen all the time but fairly frequently when I'm switching simulators a lot.

Killing launchd while you're logged in will screw up your session. Logging out and logging back in doesn't kill the user launchd; OS X keeps the existing process around. A reboot will fix things, but that's painful. The instructions above are faster.

I've submitted a bug to Apple, FWIW. rdar://10330930