Over-released object iOS app crash - ios

My app is crashing with the last message in the device console:
objc[5105]: Cannot form weak reference to instance (0x10801ec00) of class UIPageViewController. It is possible that this object was over-released, or is in the process of deallocation.
Randomly app also crashed with the same error message but UINavigationController instead of UIPageViewController. A bug is reproducible on the simulator and physical devices (iOS 11.2.5).
Profiling with "Zombie" template in Instruments did not give any valuable information. I found that outdated guide useful:
http://www.corbinstreehouse.com/blog/2007/10/instruments-on-leopard-how-to-debug-those-random-crashes-in-your-cocoa-app/
My answer might not fix your issue but I hope it will help to track it. Additionally, I'll appreciate if someone explains why ARC fired on a different thread.

By subclassing UIPageViewController and setting breakpoint in it's dealloc I found that it's not executed on the main thread.
don't do this in production - only play around when debugging
Synchronously removing its view from view hierarchy on the main thread did fix the issue. I have no view controller container responsible for that:
- (void)dealloc {
dispatch_sync(dispatch_get_main_queue(), ^{
[self.view removeFromSuperview];
});
}
However it was not even about the view, while just delaying execution was enough to fix the crash, and the view will be cleaned together with its superview.
- (void)dealloc {
dispatch_sync(dispatch_get_main_queue(), ^{
NSLog(#"%#", self);
});
}
View controllers were stored in the array, and I knew that when this array was destroyed, view controllers are as well. I checked on which thread object holding the only one reference to this array is destroyed, and it is main thread. ARC should release objects on same thread where the last reference to the object was removed, but that is not the case in my app. https://stackoverflow.com/a/32800112
How did I fix my issue:
I don't why it was so. I did not search for a reason anymore when I found that issue will also disappear when I manually remove reference to the array, right before removing the last reference to its owner.

Related

Detecting if view still exists or active in completionHandler block

In my app I'm polling a web service for status updates, using a completionHandler block and making changes to the current view based on returned results when the callback executes.
- (void) tickTimer
{
[MyWebService myWebMethod:param1 completionHandler:^(NSString *result) {
// does view still exist?
[self myUpdateMethod];
// does property still exist?
self.theResult = result;
// does child view still exist?
_txtUpdate.text = result;
}];
}
But in the interim, it's possible that the view may have been unloaded as the user navigates elsewhere.
So a couple of questions:
What happens to a view when a new one is loaded and it gets pushed to the background? I imagine it gets garbage collected at some point, but how do I tell if it's still safe to access by any of the references above, and what would happen if it's not?
If the view does still exist, how do I tell if it is also still the foreground view?
So, blocks create strong references to all objects pointers that are referred to in their closure. Due to this, your block is going to force [self] to stay in memory until the block is destroyed. If this isn't the behavior you want you should create a weak pointer to self and refer to it inside of the block:
__weak typeof(self) weakSelf = self;
So a couple of questions:
What happens to a view when a new one is loaded and it gets pushed to
the background? I imagine it gets garbage collected at some point, but
how do I tell if it's still safe to access by any of the references
above, and what would happen if it's not?
If your view stays in the view hierarchy, it will stay in memory. Once there are no more references to the view it will be dealloced.
If you use a weak pointer like outlined above, then [weakSelf] will be nil if the view has been dealloced
If the view does still exist, how do I tell if it is also still the
foreground view?
I'm not sure what you mean by foreground view, but if you want to see if it's still in the view hierarchy then you can check the property -(UIView *)superview. If superview is nil, then it's not on the screen
If you use ARC right, it will not let you use deallocated viewcontroller.
You can use viewDidAppear and viewDidDisappear methods to know visible yours viewcontroller or not.

iOS UIViewController,why "self" become to "wild pointer" in viewDidUnload?

my code snippet:
- (void)viewDidUnload{
[super viewDidUnload];
self.statusView = nil;
self.tableView = nil;
self.noDataView = nil;
}
In a rare situation, my app crashed in line self.noDataView = nil;. When I debug by po self, it seemed that it's pointing something other than current controller. What is possible reason?
PS:self.tableView's delegate and dataSource is set to self in init method. Does that have any relation to this?
First, [super viewDidUnload] should be used as the last statement. However, that won't fix your error, probably.
The reason for your problem is quite simple. Your controller is overreleased somewhere. Do you have zombie detection enabled? The code where the application crashes is usually irrelevant because the problem happened earlier.
viewWillUnload is deprecated now, you can't count on it anymore, any question about it will lead you to the below references.
From Apple:
In iOS 6, the viewWillUnload and viewDidUnload methods of
UIViewController are now deprecated. If you were using these methods
to release data, use the didReceiveMemoryWarning method instead. You
can also use this method to release references to the view
controller’s view if it is not being used. You would need to test that
the view is not in a window before doing this.
And Quote WWDC 2012:
The method viewWillUnload and viewDidUnload. We're not going to call
them anymore. I mean, there's kind of a cost-benifit equation and
analysis that we went through. In the early days, there was a real
performance need for us to ensure that on memory warnings we unloaded
views. There was all kinds of graphics and backing stores and so forth
that would also get unloaded. We now unload those independently of the
view, so it isn't that big of a deal for us for those to be unloaded,
and there were so many bugs where there would be pointers into.
Edit:
For your problem in iOS 5.1, viewDidUnload is used to release anything you have made when the view was created, so unless you are creating or retaining objects it in viewDidLoad or your nib, you may not release them in viewDidUnload.

Strange behaviour while deallocation of an object

I have a view, which is subclass of UIWebView. It has a property called Contact which is a managed object. The view uses templating engine to create a html with the object and then load into UIWebView. I thought it would be a better idea to monitor the object in the view itself, such that whenever something changes in the object, the view refreshes automatically. So, observed for certain attributes of the managed object in the view itself. And then to avoid the notification coalesce, I have made it such that the reload is done with
[self performSelector:#selector(refresh) afterDelay:0 ].
It refresh the webview automatically whenever it finds the change but also gives some strange crash. The crash says [MyWebView retain] message sent to deallocated object. I know I have properly removed observing values in dealloc method. But, it seems like dealloc gets triggered after a while. I have a strange issue related to releasing the view. The view stays for a while, although the view controller is already released and then releases after may 2/3 seconds. It is really strange. I think the crash is because of this.
Please do suggest me any idea. I will be glad to hear your suggestion. There are something wrong certainly, if anybody could point me I would really be grateful.
Using the delegate design pattern can cause EXC_BAD_ACESS KERN_INVALID_ADDRESS crashes if not used properly. If you have processing that is running in background threads that use the delegate design pattern, where in the object you set SELF as the delegate then you must remove SELF as the delegate in the dealloc method (even under ARC) by setting the delegate reference to nil, or there is a possibility that the object will try to call back into your deallocated object using the delegate design pattern. So if you have something like this in your object.
[_xmlParser setDelegate:self];
you should always have a dealloc method even under ARC to prevent the possibility of a crash in the case where your object gets destroyed while still doing work. It is very common to have your object destroyed while doing work. imagine a UIViewController that shows images from the internet. If you had a FetchImage class that used the delegate design pattern to lookup images that then calls a routine on the object when the lookup finishes, it is easily for the user to pop into and out of your UIViewController while your FetchImage object is still doing work on the background thread. You might not ever notice this when testing, but if you have hundreds of users, some of them will notice because the app will crash when your object tries to call a method on the SELF reference.
If your object uses the delegate design pattern, always have this to cleanup:
#pragma mark - dealloc - cleanup delegate references to prevent callbacks into deallocated objects (EXC_BAD_ACCESS / KERN_INVALID_ADDRESS)
- (void)dealloc
{
[_xmlParser setDelegate:nil];
// for non ARC based code you would also call: [super dealloc];
}
search every class in your project, if you have setDelegate:self or delegate = self then your users are most likely experiencing race condition crashes with your app if you don't have a dealloc cleanup method as described above. If you don't have the dealloc, add it even if you never see crashes when testing. -rrh

Receive memory warning and memory leak

I'm using ARC (automatic reference counting).
Is it ok if I set the IBOutlets to nil in the viewDidDisappear instead of viewDidUnload?
Such as these:
[self setTheImage:nil];
[self setBTNplay:nil];
[self setBTNstop:nil];
I'm writing a navigation based app which includes pageViewController, I tested my app in Instruments to see the memory leaks and I keep getting receive memory warning message.
I've even put a log code in the viewDidUnload method. But it doesn't seem to get called when I even pop to rootViewController!
One more thing: If each page has an audioPlayer, where should I set a #property (nonatomic, strong) AVAudioPlayer *audioPlayer; to nil?
Or how do I set it to weak instead of strong? Because it gives me a 'warning' then in this code line:
_audioPlayer = [[AVAudioPlayer alloc] initWithContentsOfURL:poemURL error:nil];
it says: Assigning retained object to weak variable
You don't need to nil out those values in viewDidUnload. Make sure you're using weak properties instead of strong or assign for IBOutlets. Received memory warning doesn't necessarily mean you're leaking. Received memory warning means that your app is consuming too much memory. Run Instruments and edit your question with how much memory your app uses.
The fact that you're using AVAudioPlayer makes me thing that maybe you're pulling down some massive audio files into memory.
Also by the way, initWithContentsOfURL:error: will get you rejected from the App Store because you're blocking the main thread. Try testing your app on an iPhone with only cellular enabled and go into a part of your office/house that has a bad internet connection. Also try with your phone switched to airplane mode. Your app will undoubtedly either freeze for a long time before the connection fails or it will simply crash.
Instead, you should be using grand central dispatch or downloading it via NSURLConnection's block or delegate methods.
First, do not set to nil your properties in viewDidDisappear cause your view is still loaded. You must always set them to nil in viewDidUnload. It's invoked in low memory situations and here you must clean-up all stuff that breaks system memory.
Apple's UIViewController reference for viewDidUnload
When a low-memory condition occurs and the current view controller’s
views are not needed, the system may opt to remove those views from
memory. This method is called after the view controller’s view has
been released and is your chance to perform any final cleanup.
Second , take a look at this tutorial where is explained very well ARC
Are you calling [[NSNotificationCenter defaultCenter] removeObserver:self]; from one of your view controller subclasses? If so, that would explain why you're not getting viewDidUnload called.
If that's the problem, you should remove yourself from specific notifications when needed rather than all notifications as above. (It's OK to call removeObserver:self from dealloc, though.)
Is it ok if I set the IBOutlets to nil in the viewDidDisappear instead
of viewDidUnload?
There are many things wrong from this statement.
First of all, you do not set IBOutlets to nil in viewDidDisappear. viewDidDisappear is called when the view "disappears" (e.g. when it's in a tab bar controller, and you switch to another tab; or it's on a navigation controller, and you push something on top of it); the view can then "appear" again without loading again. If you set IBOutlets to nil, they will not be set again when you appear. (They are only set when the view is loaded.)
Second, if you have a leak, and setting stuff to nil "fixes it", that means you are not releasing the instance variables. You must always release retained instance variables in dealloc.
I've even put a log code in the viewDidUnload method. But it doesn't
seem to get called when I even pop to rootViewController!
Yes, viewDidUnload is only called in low memory situations. It is not called normally in most situations. If you were depending for it to be called, you were using the wrong method.

UIViewController prevent view from unloading

When my iPhone app receives a memory warning the views of UIViewControllers that are not currently visible get unloaded. In one particular controller unloading the view and the outlets is rather fatal.
I'm looking for a way to prevent this view from being unloaded. I find this behavior rather stupid - I have a cache mechanism, so when a memory warning comes - I unload myself tons of data and I free enough memory, but I definitely need this view untouched.
I see UIViewController has a method unloadViewIfReloadable, which gets called when the memory warning comes. Does anybody know how to tell Cocoa Touch that my view is not reloadable?
Any other suggestions how to prevent my view from being unloaded on memory warning?
Thanks in advance
Apple docs about the view life cycle of a view controller says:
didReceiveMemoryWarning - The default
implementation releases the view only
if it determines that it is safe to do
so
Now ... I override the didReceiveMemoryWarning with an empty function which just calls NSLog to let me know a warning was received. However - the view gets unloaded anyway. Plus, on what criteria exactly is decided whether a view is safe to unload ... oh ! so many questions!
According to the docs, the default implementation of didReceiveMemoryWarning: releases the view if it is safe to do (ie: superview==nil).
To prevent the view from being released you could override didReceiveMemoryWarning: but in your implementation do not call [super didReceiveMemoryWarning]. That's where the view is released by default (if not visible).
The default didReceiveMemoryWarning releases the view by calling [viewcontroller setView:nil], so you could override that instead.
What appears to be working for me was to override setView: to ignore setting to nil. It's kludgy, but then, this is a kludgy issue, and this did the trick:
-(void)setView:(UIView*)view {
if(view != nil || self.okayToUnloadView) {
[super setView:view];
}
}
Could it be so simple?
Even though nowhere in the documentation this is mentioned, it seems that if I exclusively retain my view in viewDidLoad, then it does not get released on Memory Warning. I tried with several consecutive warnings in the simulator and all still seem good.
So ... the trick for the moment is "retain" in viewDidLoad, and a release in dealloc - this way the viewcontroller is "stuck" with the view until the time it needs to be released.
I'll test some more, and write about the results
I don't think any of these ideas work. I tried overriding [didReceiveMemoryWarning], and that worked for some phones, but found one phone unloaded the view BEFORE that method was even called (must have been in extremely low memory or something). Overriding [setView] produces loads of log warnings so I wouldn't risk that by Apple. Retaining the view will just leak that view - it'll prevent crashes but not really work - the view will replaced next time the controllers UI is loaded.
So really you've just got to plan on your views being unloaded any time they're off-screen, which is not ideal but there you go. The best patterns I've found to work with this are immediate commit so your UI is always up-to-date, or copy-edit-copy, where you copy your model to a temporary instance, populate your views and use immediate commit with that instance, then copy the changes back to your original model when the user hits 'save' or whatever.
Because the accepted solution has problems with viewDidUnload still getting called even though the view was blocked from being cleared, I'm using a different though still fragile approach. The system unloads the view using an unloadViewForced: message to the controller so I'm intercepting that to block the message. This prevents the confused call to viewDidUnload. Here's the code:
#interface UIViewController (Private)
- (void)unloadViewForced:(BOOL)forced;
#end
- (void)unloadViewForced:(BOOL)forced {
if (!_safeToUnloadView) {
return;
}
[super unloadViewForced:forced];
}
This has obvious problems since it's intercepting an undocumented message in UIViewController.
progrmr posted an answer above which recommends intercepting didReceiveMemoryWarning instead. Based on the stack traces I've seen, intercepting that should also work. I haven't tried that route though because I'm concerned there may be other memory cleanup which would also be blocked (such as causing it to not call child view controllers with the memory warning message).

Resources