Core Data: solve a strange EXC_BAD_ACCESS error

I am facing a really strange problem with Core Data. Let's describe it:

Definitions

Let's say I have two models, ModelA and ModelB. In the data model ModelA has a reference to ModelB as a one-to-many association, and consequently ModelB has a one-to-one association with ModelA.

Update

When the application launches (especially at first launch), or when the user asks, I have to create or update all the ModelB instances for each ModelA instance. ModelA instances are predetermined. For each ModelA instance I have about 200 instances of ModelB.

I use a code like this:

ModelB *model = [NSEntityDescription insertNewObjectForEntityForName:@"ModelB"
                                              inManagedObjectContext:context];
model.value = [NSNumber numberWithDouble:myValue];
model.modelA = modelA; // I pass modelA as a parameter to the function
[modelA addModelBObject:model];

I do not persist data immediately (since I have lots of data to save), but I do it at the end of the process.

The error

Sometimes, and only sometimes, I get an EXC_BAD_ACCESS error at this line:

model.value = [NSNumber numberWithDouble:myValue];

Enabling the zombies, I cannot see nothing more than that stupid EXC_BAD_ACCESS, no more info.

The really strange error

I tried to set that value in different ways, but nothing changed. Then, I tried to retain the NSNumber, but nothing changed. Then, finally, I tried to retain the model once created and I got the usual EXC_BAD_ACCESS, but correspondingly to the creation of the model, that is to say here:

ModelB *model = [[NSEntityDescription insertNewObjectForEntityForName:@"ModelB"
                                               inManagedObjectContext:context] retain];

Ideas?

Have you got any ideas on how to solve this issue? Am I doing something wrong? By the way, this problem apparently does not occur if I slow things down a lot (eg. saving the context every time I create a new model), but this really slow the whole application ...

Edit

In some rare cases I get this stack trace:

Serious application error.  Exception was caught during Core Data change processing.  This is usually a bug within an observer of NSManagedObjectContextObjectsDidChangeNotification.  -[__NSCFSet addObject:]: attempt to insert nil with userInfo (null)
2011-06-15 11:36:59.864 myApp[457:607] *** Terminating app due to uncaught exception 'NSInvalidArgumentException', reason: '-[__NSCFSet addObject:]: attempt to insert nil'
*** Call stack at first throw:
(
0   CoreFoundation                      0x313dc64f __exceptionPreprocess + 114
1   libobjc.A.dylib                     0x34b3dc5d objc_exception_throw + 24
2   CoreFoundation                      0x313dc491 +[NSException raise:format:arguments:] + 68
3   CoreFoundation                      0x313dc4cb +[NSException raise:format:] + 34
4   CoreFoundation                      0x31351089 -[__NSCFSet addObject:] + 152
5   CoreData                            0x35136dd9 -[NSManagedObjectContext(_NSInternalChangeProcessing) _processPendingUpdates:] + 524
6   CoreData                            0x350f4b3d -[NSManagedObjectContext(_NSInternalChangeProcessing) _processRecentChanges:] + 724
7   CoreData                            0x351363a5 -[NSManagedObjectContext processPendingChanges] + 16
8   CoreData                            0x350d027f _performRunLoopAction + 126
9   CoreFoundation                      0x313b3a35 __CFRUNLOOP_IS_CALLING_OUT_TO_AN_OBSERVER_CALLBACK_FUNCTION__ + 16
10  CoreFoundation                      0x313b5465 __CFRunLoopDoObservers + 412
11  CoreFoundation                      0x313b675b __CFRunLoopRun + 854
12  CoreFoundation                      0x31346ec3 CFRunLoopRunSpecific + 230
13  CoreFoundation                      0x31346dcb CFRunLoopRunInMode + 58
14  GraphicsServices                    0x3658841f GSEventRunModal + 114
15  GraphicsServices                    0x365884cb GSEventRun + 62
16  UIKit                               0x368ded69 -[UIApplication _run] + 404
17  UIKit                               0x368dc807 UIApplicationMain + 670
18  myApp                               0x000028cf main + 82
19  myApp                               0x00002878 start + 40
)
terminate called after throwing an instance of 'NSException'

Edit 2

This is the stack trace of the exception:

#0  0x313f1460 in __CFBasicHashAddValue ()
#1  0x3133fff8 in CFBasicHashAddValue ()
#2  0x31344162 in CFSetAddValue ()
#3  0x31351012 in -[__NSCFSet addObject:] ()
#4  0x3514211a in _PFFastMOCObjectWillChange ()
#5  0x3512ed46 in _PF_ManagedObject_WillChangeValueForKeyIndex ()
#6  0x35132e7e in _sharedIMPL_setvfk_core ()
#7  0x3513316a in _svfk_2 ()
#8  0x0003b750 in -[_TassoStorico setValoreValue:] (self=0x6d97bf0, _cmd=0x49064, value_=1.02600002) at _TassoStorico.m:87
#9  0x0001b62e in -[EuriborParser(hidden) readStoricoForzato] (self=0x74200d0, _cmd=0x48ff7) at EuriborParser.m:236
#10 0x31349f02 in -[NSObject(NSObject) performSelector:withObject:] ()
#11 0x000441c4 in -[MBProgressHUD launchExecution] (self=0x90a6ff0, _cmd=0x4b83f) at MBProgressHUD.m:482
#12 0x352b3388 in -[NSThread main] ()
#13 0x353255cc in __NSThread__main__ ()
#14 0x34e20310 in _pthread_start ()
#15 0x34e21bbc in thread_start ()

Vincent Guerci

CoreData is not thread-safe; looks like that's your issue try looking at this SO question

I had a similar issue (EA_BAD_ACCESS on [managedContext save])which was caused by a KeyValueObserver that had not been removed when it should have been. It was quite hard to track down. This meant that I was getting this sort of behaviour even with ARC.

Reset Your Managed Object Context After a Persistent Store Migration

I had similar issues. I also work with a couple of threads, and I use Core Data’s migratePersistentStore:toURL:options:withType:error: function.

In my case the error happened right after a migration in another thread. I thought that my issues were thread-related, so I began to double-down on thread-safety in my code, but in reality a single call to managedObjectContext.reset() before doing any subsequent operations on objects in the managed object context solved all issues.

Core Data Behavior

Strangely enough existingObjectWithID:error: was always giving me back a reference to a non-existing object, so that I was accessing non-existent objects and receiving EXC_BAD_ACCESS. After resetting the context the function exposed the correct behavior.

These two lines are redundant and possibly dangerous:

model.modelA = modelA; // I pass modelA as a parameter to the function
[modelA addModelBObject:model];

The managed object context will automatically sent the reciprocal relationship for you so there is no need to do it manually. In fact, I think it might cause a dangerous loop. You can use just one line or the other and both sides of the relationship will be automatically set. I recommend just:

model.modelA = modelA;

... because it is simpler.

来源：https://stackoverflow.com/questions/6355126/core-data-solve-a-strange-exc-bad-access-error

标签

objective-c

core-data

exception-handling

exc-bad-access