This question follows on from my other question on why my app isn't being brought down by exceptions.
The Problem
When an exception is thrown on the main thread via an Action, the app still doesn't crash.
As per Dave's answer to my original question, I've implemented the reportException category on NSApplication and set the uncaught exception handler.
Code
I've got the following in my app delegate, which I've hooked up to a button in my UI to test.
-(IBAction)crashOnMainThread:(id)sender {
[self performSelectorOnMainThread:#selector(crash) withObject:nil waitUntilDone:YES];
}
-(void)crash {
// To test out the exception handling
[NSException raise:NSInternalInconsistencyException format:#"This should crash the app."];
}
When I press the button, my app doesn't crash. When I look at the console log, I see this:
06/09/2010 14:12:25 EHTest1[26384] HIToolbox: ignoring exception 'This should crash the app.' that raised inside Carbon event dispatch
(
0 CoreFoundation 0x00007fff80ab4cc4 __exceptionPreprocess + 180
1 libobjc.A.dylib 0x00007fff819560f3 objc_exception_throw + 45
2 CoreFoundation 0x00007fff80ab4ae7 +[NSException raise:format:arguments:] + 103
3 CoreFoundation 0x00007fff80ab4a74 +[NSException raise:format:] + 148
4 EHTest1 0x00000001000010e3 -[EHTest1_AppDelegate crashLapsus] + 63
5 Foundation 0x00007fff88957c25 -[NSObject(NSThreadPerformAdditions) performSelector:onThread:withObject:waitUntilDone:modes:] + 234
6 Foundation 0x00007fff8896ad48 -[NSObject(NSThreadPerformAdditions) performSelectorOnMainThread:withObject:waitUntilDone:] + 143
7 EHTest1 0x0000000100001030 -[EHTest1_AppDelegate crashOnMainThread:] + 60
8 AppKit 0x00007fff85c7e152 -[NSApplication sendAction:to:from:] + 95
9 AppKit 0x00007fff85ca26be -[NSMenuItem _corePerformAction] + 365
** Snip **
It looks like Carbon is catching the exception, which is really annoying.
This suggests that for any action code, you need to immediately run it in a background thread so any exceptions are registered as uncaught. Huh? I've not seen any code that's structured like this.
What I've tried
Crashing the app with a delay so it's not connected to a UI element. It crashes fine.
I've tried installing a custom NSExceptionHandler using this in my app delegate:
-(BOOL)exceptionHandler:(NSExceptionHandler *)sender
shouldHandleException:(NSException *)exception
mask:(unsigned int)aMask {
abort();
return YES;
}
-(void)applicationWillFinishLaunching:(NSNotification *)aNotification {
NSExceptionHandler *handler = [NSExceptionHandler defaultExceptionHandler];
[handler setExceptionHandlingMask:NSLogAndHandleEveryExceptionMask];
[handler setDelegate:self];
}
the problem here is it crashes on every exception, whether it's caught or not.
If I try and check the mask and don't crash on a caught exception, I'm back to square 1 as it seems that HIToolbox catches the exception in exactly the same way as a try/catch block would.
Questions
How can I stop HIToolbox catching the exceptions so that my app uses the uncaught exception handler and crashes?
Is it OK to be running code that's in the same call stack as an action? Surely this is OK?
If it's not OK, what's the alternative?
This is driving me up the wall, so any help would be much appreciated.
I answered your last question on this subject, and ran into the same problem with Carbon's HIToolbox catching exceptions thrown by IBActions.
First, undo everything I mentioned in my previous answer. It doesn't work with IBActions for some reason. My hunch is that HIToolbox lives lower on the exception-handling-chain, and gets any IBAction/GUI exceptions before NSApplication has the opportunity to. Any custom exception-handling function you can register with NSSetUncaughtExceptionHandler() lives (I believe) at the top of the chain.
You're on the right track with NSExceptionHandling:
Add the ExceptionHandling.framework to your Xcode Project
#import "ExceptionHandlerDelegate.h" into your AppDelegate.m (or a custom Singleton exception class)
Inside AppDelegate.m:
// Very first message sent to class
+ (void)initialize
{
NSExceptionHandler *exceptionHandler = [NSExceptionHandler defaultExceptionHandler];
unsigned int handlingMask = NSLogAndHandleEveryExceptionMask;
[exceptionHandler setExceptionHandlingMask:handlingMask];
[exceptionHandler setDelegate:self];
// ...
}
#pragma mark -
#pragma mark NSExceptionHandler Delegate Methods
// Called 1st: Should NSExceptionHandler log the exception?
- (BOOL)exceptionHandler:(NSExceptionHandler *)sender shouldLogException:(NSException *)exception mask:(unsigned int)aMask
{
// ...log NSException if desired...
return NO; // Changes to YES if NSExceptionHandler should handle logging
}
// Called 2nd: Should NSExceptionHandler handle the exception?
- (BOOL)exceptionHandler:(NSExceptionHandler *)sender shouldHandleException:(NSException *)exception mask:(unsigned int)aMask
{
// ...crash your app here (e.g. [NSApp terminate:self]; )
// ...or handle the NSException and return YES/NO accordingly
return NO; // If app crashed, never gets returned - should crash before that
}
The NSLogAndHandleEveryExceptionMask flag tells NSExceptionHandler to capture every exception it can (for your app only, I believe), regardless of where on the exception chain it exists.
This means that #catch/#try/#finally blocks won't work, because the NSHandleOtherExceptionMask flag tells NSExceptionHandler to capture "everything below it" on the exception-handler chain. You can remove that flag, but then HIToolKit will probably get any IBAction exceptions again, since it appears to be lower on said chain.
Apple's docs have info about the flags: NSExceptionHandler docs
So when an NSException is raised (anywhere in your app AFAIK), and NSLogAndHandleEveryExceptionMask is set, these are called in the delegate in-order:
-exceptionHandler:shouldLogException:mask: is called first.
-exceptionHandler:shouldHandleException:mask: is called second.
Just put your "crash code" inside the second delegate method and you should be OK.
Helpful article: Understanding Exceptions and Handlers in Cocoa
The reason I think you were having trouble getting NSExceptionHandler's delegate to work is because it's NOT compatible with a custom method set with NSSetUncaughtExceptionHandler(), which was part of the answer in my previous question. Per Apple:
The NSExceptionHandler class provides
facilities for monitoring and
debugging exceptional conditions in
Objective-C programs. It works by
installing a special uncaught
exception handler via the
NSSetUncaughtExceptionHandler
function. Consequently, to use the
services of NSExceptionHandler, you
must not install your own custom
uncaught exception handler.
It also probably doesn't work well when you override NSApplication's -reportException: method.
Lastly, there doesn't appear to be a need to use #catch/#try/#finally (also part of my previous answer). Configuring NSExceptionHandler inside +initialize appears to "kick in" immediately, unlike overriding NSApplication's -reportException: method.
You can’t reliably. Throwing exceptions across API boundaries is not supported unless explicitly documented (and I can’t think of any cases that are explicitly documented).
Related
After updating to 10.10.3 the WebView component started to crash after dealloc
- (void)dealloc {
[self.webView.windowScriptObject setValue:nil forKey:#"CocoaApp"];
[[self.webView mainFrame] stopLoading];
[self.webView setUIDelegate:nil];
[self.webView setEditingDelegate:nil];
[self.webView setFrameLoadDelegate:nil];
[self.webView setPolicyDelegate:nil];
[self.webView removeFromSuperview];
}
The crash happens somewhere deep in WebView
EXC_BAD_ACCESS
1 0x7fff910bae9e WebDocumentLoaderMac::detachFromFrame()
2 0x7fff920288c0 WebCore::FrameLoader::detachFromParent()
3 0x7fff910d0e55 -[WebView(WebPrivate) _close]
4 0x7fff910d0c49 -[WebView dealloc]
5 0x7fff8b1cf89c objc_object::sidetable_release(bool)
6 0x7fff8b1b5e8f (anonymous namespace)::AutoreleasePoolPage::pop(void*)
7 0x7fff912b26f2 _CFAutoreleasePoolPop
8 0x7fff8830e762 -[NSAutoreleasePool drain]
9 0x7fff8e3f0cc1 -[NSApplication run]
10 0x7fff8e36d354 NSApplicationMain
11 0x1000ebb12 main
12 0x7fff8c81e5c9 start
13 0x3
Any ideas? Is this a Apple bug? It started AFTER 10.10.3?
It doesn't crash when NSZombie is enabled!
I noticed you're using your own policy delegate:
[self.webView setPolicyDelegate:nil];
There's a known bug related to policy delegates in WebKit (only very recently fixed):
https://bugs.webkit.org/show_bug.cgi?id=144975
The short version is that you're probably hitting this assertion (which crashes the process with an intentional segfault):
https://github.com/WebKit/webkit/blob/24b1ae89efc10a4e6a6057b429c8e1d8d138a32f/Source/WebCore/loader/DocumentLoader.cpp#L935
because your policy handler (i.e. decidePolicyForMIMEType:request:frame:decisionListener:) is failing to make a policy decision (i.e not use, ignore, or download). The decision hangs around unmade, and when the loader eventually detaches it asserts that there are no pending policy decisions, which fails since the view is still waiting for a decision.
The fix i made, is not to release the webview, but hold a static reference into it (this is far from solving it and i contacted Apple regarding this issue)
#warning HOTFIX
{
//this is because of http://stackoverflow.com/questions/29746074/osx-10-10-3-crashes-webview-on-dealloc
static NSMutableArray * LIVE_FOR_EVER_WEBVIEW;
if (LIVE_FOR_EVER_WEBVIEW == nil) {
LIVE_FOR_EVER_WEBVIEW = [NSMutableArray new];
}
if (self.webView) {
[LIVE_FOR_EVER_WEBVIEW addObject:self.webView];
}
}
I have implemented a concurrent nsoperation and have ARC enabled. Now my customer is experiencing a crash which I cannot reproduce. He sent me the follow crash log :
Date/Time: 2013-04-24 12:23:34.925 -0400
OS Version: Mac OS X 10.8.3 (12D78)
Report Version: 10
Interval Since Last Report: 30946 sec
Crashes Since Last Report: 1
Per-App Interval Since Last Report: 33196 sec
Per-App Crashes Since Last Report: 1
Anonymous UUID: FB8460EE-5199-C6FB-55DC-F927D7F81A80
Crashed Thread: 15 Dispatch queue: com.apple.root.default-priority
Exception Type: EXC_BAD_ACCESS (SIGSEGV)
Exception Codes: EXC_I386_GPFLT
Application Specific Information:
objc_msgSend() selector name: isCancelled
Thread 15 Crashed:: Dispatch queue: com.apple.root.default-priority
0 libobjc.A.dylib 0x00007fff877f1250 objc_msgSend + 16
1 Myapp 0x000000010a608807 0x10a601000 + 30727
2 Myapp 0x000000010a650575 0x10a601000 + 324981
3 com.apple.Foundation 0x00007fff8b66212f -[NSBlockOperation main] + 124
4 com.apple.Foundation 0x00007fff8b638036 -[__NSOperationInternal start] + 684
5 com.apple.Foundation 0x00007fff8b63f861 __block_global_6 + 129
6 libdispatch.dylib 0x00007fff832d0f01 _dispatch_call_block_and_release + 15
7 libdispatch.dylib 0x00007fff832cd0b6 _dispatch_client_callout + 8
8 libdispatch.dylib 0x00007fff832ce1fa _dispatch_worker_thread2 + 304
9 libsystem_c.dylib 0x00007fff87d19d0b _pthread_wqthread + 404
10 libsystem_c.dylib 0x00007fff87d041d1 start_wqthread + 13
My code looks like this:
-(void)start
{
// Always check for cancellation before launching the task.
if ([self isCancelled])
{
// Must move the operation to the finished state if it is canceled.
[self onCancelSyncOperation];
return;
}
// If the operation is not canceled, begin executing the task.
[self willChangeValueForKey:#"isExecuting"];
[NSThread detachNewThreadSelector:#selector(main) toTarget:self withObject:nil];
executing = YES;
[self didChangeValueForKey:#"isExecuting"];
}
- (void)onCancelSyncOperation
{
[self willChangeValueForKey:#"isFinished"];
[self willChangeValueForKey:#"isExecuting"];
executing = NO;
finished = YES;
[self didChangeValueForKey:#"isExecuting"];
[self didChangeValueForKey:#"isFinished"];
}
It seems like the nsoperation is already released? when it tries to check for isCancelled?
Is this possible?
I don't think anyone can tell you why your app crashes by looking at this log. I haven't looked at your code, but you cut your crash-log, so it only shows system modules (lib.dispatch..., com.apple...). Typically the error is in the first occurrence of "com.myname...".
If this kind of crash EXC_BAD_ACCESS (SIGSEGV) appears along with an objc_msgSend(), it probably means you're trying to message to an object (or in other words: call a method of an object) that isn't there anymore. If you call that object, chances are very good that you'll find it, if you call it delayed or on another Thread or from a block, it will be a bit more complicated.
Your best chance to find the cause of this is to inspect your app with Instrument, using the Allocations or Leaks tool with NSZombies enabled (which is the default). You can launch Instruments from within Xcode. Then try to reproduce your crash. If you succeed, you might be able to find the class and location where the crash occurs.
If you don't know what to do with this answer, then check out the WWDC Developer Videos from Apple, there are some that show how to profile your app with Instruments (you'll need a [free] Dev Account to access the videos).
Good luck!
Simple question: what happens if I do this:
- (void)viewDidLoad
{
[self performSelectorInBackground:#selector(myBGMethod) withObject:nil];
}
-(void)myBGMethod
{
[self myOtherMethod];
}
-(void)myOtherMethod
{
NSLog(#"This is not Inception");
//some more code here
}
Will the NSLog() and other code in myOtherMethod be run on the main thread or in the background?
It'll be run in the background thread.
You can confirm this by calling NSLog inside all your methods. By default, NSLog prints the thread number along the process ID (pid).
It'll be run in the background. Once you make the call to myBGMethod in another thread, whatever it calls is made on that same thread unless it specifically requests another thread.
As a side note, depending on which version of iOS you want to support, you might want to learn more about Grand Central Dispatch. It makes multithreading a lot simpler.
If you are ever curious about what thread a particular line of code is executing on, you can put a breakpoint on that line and check the Debug Navigator pane in Xcode:
In this case, I put a breakpoint on NSLog(...)
dispatch_async(dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_DEFAULT, 0), ^{
NSLog(#"HI!");
});
and you can see that we're in Thread 2 com.apple.root.default-priority
An iPad app that runs fine under IOS3 fails under IOS4.2 It has a class that runs an http session from an operation queue and the failure is linked to this activity. Here is the console output:
Program received signal: “EXC_BAD_ACCESS”.
[Switching to thread 11523]
Running NSZombies enabled didn't reveal anything so I have been putting NSLog statements in the code and found that the crash occurs when a local variable is changed. Here is the code section:
self.currentOperation = [[[DeduceAccessOperation alloc] init] autorelease];
[self.currentOperation addObserver:self forKeyPath:#"isFinished"
options:(NSKeyValueObservingOptionNew | NSKeyValueObservingOptionOld)
context:NULL];
NSLog (#"Start observer added");
[operationQueue addOperation:self.currentOperation];
NSLog (#"Start operation added");
NSLog(#"State is %d", self.status);
self.status = IEnablerServiceUpdating;
NSLog (#"State updated");
And here is the console log output:
2010-12-08 21:26:44.548 UCiEnabler[5180:307] Start observer added
2010-12-08 21:26:44.550 UCiEnabler[5180:307] Start operation added
2010-12-08 21:26:44.552 UCiEnabler[5180:307] State is 1
Program received signal: “EXC_BAD_ACCESS”.
[Switching to thread 11523]
It is like status has become read-only (It's property is declared as atomic and readwrite).
The other relevant piece of information is that a sub-view has just been changed and it triggers the call on the above routine. It's code is:
//Start the update
UCiEnablerAppDelegate *controller = (UCiEnablerAppDelegate *)[[UIApplication sharedApplication] delegate];
[controller deduceIEnablerServiceAccess];
controller.serviceBusy = TRUE; //1.04
Has anyone seen anything like this?
Has anyone ideas where to look next?
Regards
Robin
Here's the stack trace:
#0 0x34a80464 in objc_msgSend
#1 0x3119543e in NSKVOPendingNotificationCreate
#2 0x3119535a in NSKeyValuePushPendingNotificationPerThread
#3 0x3117009a in NSKeyValueWillChange
#4 0x311682c6 in -[NSObject(NSKeyValueObserverNotification) willChangeValueForKey:]
#5 0x311cc718 in _NSSetIntValueAndNotify
#6 0x000097ce in -[IEnablerService startDeducingAccessState] at IEnablerService.m:55
#7 0x00002bc0 in -[UCiEnablerAppDelegate deduceIEnablerServiceAccess] at UCiEnablerAppDelegate.m:100
#8 0x0000a33e in -[RootViewControlleriPad animationDidStop:finished:context:] at RootViewController-iPad.m:43
#9 0x341bb336 in -[UIViewAnimationState sendDelegateAnimationDidStop:finished:]
NSOperationQueue in iOS 4.2 now uses GrandCentralDispatch when starting NSOperations. There's a Technical Q&A here
Pretty much the queue now calls the NSOperation subclass's start method in a new thread regardless of the isConcurrent property. In my experience it means that if you are using NSURLConnection inside a start method your code will no longer run because the thread start is running on doesn't have an NSRunLoop.
Apple says that this change doesn't break compatibility because you were supposed to spawn a new thread in the start method anyway.
For backwards compatibility you should change your subclass from using start to using run.
I actually have two questions regarding exception/error handling in the iPhone app that I am making:
The app uses Internet, but when there's no connection, the app just dies (during launch). How can I handle this to print some infomsg to the user, instead of just getting thrown back to the springboard?
Can someone show me an example of how to handle for instance a "page not found" or "no contact with server" error, so I can give some sort of info to the user in the same way as above?
For crashes, the first step is to use error messages and the debugger to figure out what call is causing the problem. If the problem is caused by an uncaught exception, read this Apple article on exception handling. The specific answer really depends on your code and exactly what is causing the crash, so I won't speculate about a particular solution.
As far as detecting server error response codes (such as 404), that's more specific to WebKit. I assume you're using UIWebView on iPhone, and you've probably noticed that none of the primary methods return errors. This is because it uses a delegate model to report progress or errors asynchronously. (It makes sense because you don't want your UI code to be at the mercy of a slow-loading (or non-existent) webpage. To be notified of such errors, there are a few steps.
Adopt the UIWebViewDelegate protocol, usually in the same class that will start the webpage load for convenience.
Set that object as the delegate of the UIWebView instance. (It has a delegate property, so you can use something like either uiView.delegate = self or [uiView setDelegate:self] based on what you prefer.)
Implement the webView:didFailLoadWithError: method in that class. (You can be notified when the load finishing by implementing webViewDidFinishLoad: as well.) This is where you include the logic of what should happen when an error occurs.
I didn't see any detailed documentation on the content of any particular errors handed back via this delegate method, but it's a standard NSError object, and I recommend checking out the contents by calling its methods, such as -localizedDescription and -userInfo.
Here is some sample code with #import statements excluded for brevity.
MyClass.h
#interface MyClass : NSObject <UIWebViewDelegate> {
IBOutlet UIWebView* myWebView;
}
-(void)webView:(UIWebView*)webView didFailLoadWithError:(NSError *)error;
#end
MyClass.m
#implementation MyClass
- (id) init {
if ((self = [super init]) == nil)
return nil;
// initialize myWebView
myWebView.delegate = self;
return self;
}
- (void) webView:(UIWebView*)webView didFailLoadWithError:(NSError*)error {
...
}
#end
Testing for a connection is pretty easy...
NSString * test = [NSString stringWithContentsOfURL:[NSURL URLWithString:#"http://www.stackoverflow.com"]];
if (test == nil) {
//display an alertview saying you don't have an internet connection
}
Using a URL to test for a connection is not a good idea, it is not robust enough to determine if the internet connection is down, the website is down or some other network issue etc and above all it adds an overhead to the call as far as network traffic.
Look at the Reachability demo on the Apple site, it uses the correct way to determine connectivity, including whether you are on wifi etc.