I'm experiencing a weird issue on iOS16.
There is no code change, it works fine on lower OS versions. The stack traces are:
The first case:
Crashed: com.apple.main-thread
EXC_BAD_ACCESS KERN_INVALID_ADDRESS 0x00000000e5bde410
0
libobjc.A.dylib
objc_release_x21 + 16
1
CoreData
-[_PFManagedObjectReferenceQueue _processReferenceQueue:] + 1020
2
CoreData
-[NSManagedObjectContext _processRecentChanges:] + 112
3
CoreData
_performRunLoopAction + 412
4
CoreFoundation
__CFRUNLOOP_IS_CALLING_OUT_TO_AN_OBSERVER_CALLBACK_FUNCTION__ + 36
5
CoreFoundation
__CFRunLoopDoObservers + 532
6
CoreFoundation
__CFRunLoopRun + 1048
7
CoreFoundation
CFRunLoopRunSpecific + 612
8
GraphicsServices
GSEventRunModal + 164
9
UIKitCore
-[UIApplication _run] + 888
10
UIKitCore
UIApplicationMain + 340
The second case, I believe this stack trace has the same root cause as the previous one (same trend, same UI page, only happens on iOS 16).
Crashed: com.apple.main-thread
EXC_BAD_ACCESS KERN_INVALID_ADDRESS 0x0033003200390070
0
libobjc.A.dylib
objc_msgSend + 32
1
CoreFoundation
-[__NSArrayM dealloc] + 188
2
MyApplication
MyClass.m - Line 361
-[MyClass loadMessages:] + 361
3
MyApplication
MyClass.m - Line 125
__74-[MyClass requestRecentMessagesAndDiscardExistingMessagesCompletion:]_block_invoke + 125
4
libdispatch.dylib
_dispatch_call_block_and_release + 32
5
libdispatch.dylib
_dispatch_client_callout + 20
6
libdispatch.dylib
_dispatch_main_queue_drain + 928
7
libdispatch.dylib
_dispatch_main_queue_callback_4CF + 44
8
CoreFoundation
__CFRUNLOOP_IS_SERVICING_THE_MAIN_DISPATCH_QUEUE__ + 16
9
CoreFoundation
__CFRunLoopRun + 2036
10
CoreFoundation
CFRunLoopRunSpecific + 612
11
GraphicsServices
GSEventRunModal + 164
12
UIKitCore
-[UIApplication _run] + 888
13
UIKitCore
UIApplicationMain + 340
MyClass.m is an Objective-C class, it has a property:
@property (nonatomic, strong) NSArray<Message *> *messages; // Message is NSManagedObject
In the second stacktrace, frame -[OurClass loadMessages:] + 361, the messages array is deallocated:
self.messages = [[NSArray alloc] init...];
So my guess is, somehow the messages are over-released. If the messages are released in MyClass before, then the crash happens as the first stack trace, otherwise, it happens as the second stack trace.
I've turned on -com.apple.CoreData.ConcurrencyDebug 1 to try to find out the concurrency issues and fixed them, but no luck.
Any help would be much appreciated. Thanks in advance!