Non-deterministic crash in BlocksKit bk_apply block - objective-c

I have a function that constructs an NSMutableDictionary using bk_apply, a method provided by the third-party block utility library BlocksKit. The function's test suite usually passes just fine, but once every couple runs it crashes.
NSMutableDictionary *result = [[NSMutableDictionary alloc] init];
[inputSet bk_apply:^(NSString *property) {
NSString *localValueName = propertyToLocalName[property];
NSObject *localValue = [self valueForKey:localValueName];
result[property] = localValue ?: defaults[property]; // Crash
// Convert all dates in result to ISO 8601 strings
if ([result[property] isKindOfClass:[NSDate class]]) { // Crash
result[property] = ((NSDate *)result[property]).ISODateString; // Crash
}
}];
The crash always happens on a line where result is referenced, but it's not the same line every time.
Examining the contents of result in the debugger, I've seen very strange values like
po result
{
val1 = "Some reasonable value";
val2 = "Also reasonable value";
(null) = (null);
}
It's impossible for an NSDictionary to have null keys or values, so clearly some invariant is being violated.
What is causing this crash and how do I fix it?

From the BlocksKit documentation for bk_apply:
Enumeration will occur on appropriate background queues. This will
have a noticeable speed increase, especially on dual-core devices, but
you must be aware of the thread safety of the objects you message
from within the block.
The code above is highly unsafe with respect to threading, because it reads from and writes to a mutable variable on multiple threads.
The intermittent nature of the crash comes from the fact that the thread scheduler is non-deterministic. The crash won't happen when several threads accessing shared memory happen to have their execution scheduled in sequence rather than in parallel. It is therefore possible to "get lucky" some or even most of the time, but the code is still wrong.
The debugger printout is a good example of the danger. The thread that's paused is most likely reading from result while another thread performs an insertion.
NSMutableDictionary insertions are likely not atomic; example steps might be,
allocate memory for the new entry
copy the entry's key into the memory
copy the entry's value into the memory
If you read the dictionary from another thread between steps 1 and 2, you will see an entry for which memory has been allocated, but the memory contains no values.
The simplest fix is to switch to bk_each. bk_each does the same thing as bk_apply but it's implemented in a way that guarantees sequential execution.

Related

NSOperationQueue methods parameter are becoming empty

I am having problem with NSOperationQueue, if I am adding the same operation for 200 times method is behaving as expected.
But if I increase the for loop to 500 times, parameter are becoming empty when queue will start executing the task. Below is the code snippet.
- (void)someMethod:(char *)param1 {
NSBlockOperation *theOp = [NSBlockOperation blockOperationWithBlock: ^{
// use this paramter and do something
}];
[[MyQueueService sharedMyQueueService] operationQueue] addOperation:theOp];
}
This is how I am invoking the above method
for (int index = 1; index < 500; index++) {
MyClass *classInstance = [[MyClass alloc] init];
NSString *parm1 = [NSString stringWithFormat:#"%d", index];
[classInstance someMethod:(char *)[string cStringUsingEncoding:NSUTF8StringEncoding]];
}
Here is becoming empty i.e "", if i run the same method for 500 time, due to this I am unable to perform other operation. Please help me regarding this.
The problem is not with NSOperationQueue. The issue is the use of char *. As the documentation for cStringUsingEncoding says:
The returned C string is guaranteed to be valid only until either the receiver is freed, or until the current memory is emptied, whichever occurs first. You should copy the C string or use getCString:maxLength:encoding: if it needs to store the C string beyond this time.
Bottom line, simple C pointers like char * do not participate in (automatic) reference counting. Your code is using dangling pointers to unmanaged buffer pointers. This is exceedingly dangerous and when not done properly (as in this case), will lead to undefined behavior. The resulting behavior is dictated whether the memory in question happened to be reused for other purposes in the intervening time, which can lead to unpredictable behavior that changes based upon completely unrelated factors (e.g. the loop count or whatever).
You should try running your app with the "address sanitizer" turned on (found in the scheme settings under the "run" settings, on the diagnostics tab) and it will likely report some of these issues. E.g. when I ran your code with address sanitizer, it reported:
==15249==ERROR: AddressSanitizer: heap-use-after-free on address 0x60300003a458 at pc 0x00010ba3bde6 bp 0x70000e837880 sp 0x70000e837028
For more information, see Address Sanitizer documentation or its introductory video.
The easiest solution is going to be to eliminate the char * and instead use the int index value or use an object, such as NSString *.

Why does this ternary operation causes memory growth

The following line causes memory growth (no releases, only one malloc line in instruments) when testing with mark generation feature of allocations instrument
- (instancetype)initWithTitle:(NSString *)title andImageNamed:(NSString *)imageName andButtonProperties:(NOZSKButtonNodeProperties *)buttonProperties
{
...
textNode.text = buttonProperties.buttonTitleIsUppercase ?
[title uppercaseStringWithLocale:[NSLocale currentLocale] ] : title;
...
}
Here is the code that calls it
NOZSKButtonNodeProperties *props = [self getThreeButtonProps];
...
props.buttonTitleIsUppercase = YES;
...
// this initializer is calling the above initializer by passing nil for imageName arg
NOZSKButtonNode *btn = [[NOZSKButtonNode alloc] initWithTitle:#"Play Again" andButtonProperties:props];
-uppercaseStringWithLocale: makes an upper-case copy of the title string. This requires allocation. And although you didn't show this, I assume textNode both retains its text property and has a wider scope than the -initWithTitle:andImageNamed:andButtonProperties: method and therefore continues to live afterwards.
Without more complete code it is not possible to be sure, but here is a guess in case it helps.
It sounds like you might be chasing a ghost. Are you seeing increasing memory due to this code across different iterations of your run loop?
Explanation: Even with ARC some memory is placed in the "autorelease pool" rather than being immediately deallocated when no longer required. This is an unfortunate legacy of MRC. ARC is able to avoid some uses of the autorelease pool and deallocate memory quicker, but not all uses.
The autorelease pool is typically emptied once per iteration of your main run loop. If you allocate and release a lot of memory in response to a single event, say in a long loop, it can be worth while wrapping the offending code in an #autorelease { ... } block which creates, and empties on exit, a local autorelease pool. This helps keep the peak memory usage down - it doesn't deallocate any more memory overall, it just does it cleanup sooner.
When you altered your code and saw an apparent improvement you may just have reorganised your code in a way more amenable to ARC's optimisations which reduce use of the autorelease pool, and so memory is deallocated sooner.
You only need to be concerned if (a) memory is increasing across multiple events or (b) you are hitting a too high peak memory use. Under ARC (a) should be rare, while (b) requires locating the source and wrapping it in an #autorelease { ... } block.
HTH
textNode.text = buttonProperties.buttonTitleIsUppercase ?
[title uppercaseStringWithLocale:[NSLocale currentLocale] ] : title;
When you write [title uppercaseStringWithLocale:], this method has to create a new instance in order to change the given string to upper case. This is going to increase the memory usage of your program because this new string has to be allocated.
Also, it would have helped to know whether or not you are using ARC, because I don't think this should be an issue with ARC.

objective-c How to prevent an action while a thread is being executed

I've been using Multithreading for a while I thought I got it but my program is crashing now.
I have a method that has to download data for the server and access memory depending on the data, that process takes long, so I execute it from a secondary thread like this:
-(void)showPeople{
dispatch_queue_t pintaOcupantes = dispatch_queue_create("Pinta Ocupantes", NULL);
dispatch_async(pintaOcupantes, ^{
//BUNCH OF CODE
[self isPersonIn:jid];
//MORE CODE that include methods calling isPersonIn
});
Inside that block there's isPersonIn. It crashes if I press too fast the button that executes showPeople. IsPersonIn is something like:
-(int)isPersonIn:(XMPPJID *)jid{
int i = 0;
for(NSDictionary *card in self.listaGente){
NSLog(#"la jid es: %#", [card objectForKey:#"jid"]);
NSLog(#"la jid del usuario es: %#", jid.user);
if([[card objectForKey:#"jid"] isEqualToString:jid.user]){
return i;
}
i++;
}
return -1;
}
It compares a XMPPJID with an array which is a instance variable.
isPersonIn is called several times from different methods but all the methods that call it belong to the block, so as I understand it, all the executions of isPersonIn should be serialized, FIFO, right?
But if I press the button that executes showPeople, the one containing the block, many times very fast the app crashes on isPersonIn, sometimes without any message. I can see the threads when it crashes and I see at least 2 threads with isPersonIn last in the stack, which doesn`t make sense, since the block should be executed one at a time, not several threads at the same time, right?
Any help will be very much appreaciated.
Thanks!
[EDIT]
Also the instance array, self.listaGente, is modified outside the block.
I'm not a GCD expert, but I suspect the reason you're getting multiple threads is that you're creating a new dispatch queue each time showPeople is called.
So rather than having a single serial queue with multiple blocks, I think you are ending up with multiple queues each executing a single block.
[EDIT] If the collection is modified outside of the block but during execution of the block, this could be the source of your crash. From Fast Enumeration Documentation:
Enumeration is “safe”—the enumerator has a mutation guard so that if you attempt to modify the collection during enumeration, an exception is raised.
In this case protecting the array, that was provoking my app to crash, fixed the problem.
using:
#syncronized(theArray){
//CODE THAT WILL ACCESS OR WRITE IN THE ARRAY
}
This way threads will stop before if there's a thread already executing that code, like a mutex or semaphore

Object released with CFRelease causes obvious crash, but only rarely

I have the following method:
+ (NSString*) getMD5HashFromFile:(NSString*)filePath {
CFStringRef md5hash = FileMD5HashCreateWithPath((CFStringRef)filePath, FileHashDefaultChunkSizeForReadingData);
NSString *hashStr = (NSString*)md5hash;
CFRelease(md5hash);
return hashStr;
}
I was getting random crashes on the Simulator, about 1 in 20-30 executions. The fact that this wasn't consistent didn't help me dig deeper before.
Now that I see the code again, it seems obvious that md5hash gets released before being returned, which means the returned object is invalidated. The returned value is used in another method in a consistent way that crashes sometimes, but not always. My question is why this only happens rarely and not always.
Does it have something to do with the mix of Obj-C and C code and the way autorelease pools work?
Note: The bug seems to be fixed by using NSString *hashStr = [NSString stringWithString:(NSString*)md5hash], which makes total sense to me.
Just because a piece of memory is released and deallocated doesn't mean that it's immediately returned to the OS. Your application can hold onto it for an arbitrary period of time based on numerous factors and at several layers. The OS has more important things to do sometimes than reclaim every piece of memory you let go of and might ask for again in half a second. Accessing memory that you've called free() on, but technically own, does not generate a signal. This is why MallocScribble exists. It overwrites memory that you free with trash (0x55) so that it's more obvious when you use freed memory.
Try the following:
char *foo = malloc(100);
strcpy(foo, "stuff");
free(foo);
printf("%s", foo);
Most of the time that'll work fine, despite being completely wrong. Now, edit your Scheme>Diagnostics and Enable Scribble. Re-run and you'll see a bunch of "U" (0x55) indicating that you're reading nonsense. But it still won't crash.
You may be interested in Matt Gallagher's A look at how malloc works on the Mac for a bit more on the topic.
CFRelease argument must not be NULL.
If CFRelease argument is NULL, this will cause a runtime error and
your application will crash
if(md5hash)
CFRelease(md5hash);
+(NSString*) getMD5HashFromFile:(NSString*)filePath {
CFStringRef md5hash = FileMD5HashCreateWithPath((CFStringRef)filePath, FileHashDefaultChunkSizeForReadingData);
NSString *hashStr = [(NSString*)md5hash copy];
CFRelease(md5hash);
return [hashStr autorelease];
}
make sure to retain the returned value in the caller if you need to hang on to it for any length of time.

Why I'm getting memory leaks with xmlTextReaderConstValue?

I'm writing my own wrapper class for parsing XML data. Usually I use the Leak Performance Tool to detect suspicios behaviour through forgetting to release allocated memory.
At this time I figured out that the following code (the first line becomes marked by the tool) brings me an enormous memory leak (leaks more the bigger the XML data file becomes).
the following part is used to receive the text inside a Node.
NSString *currentTagValue = [NSString stringWithCString:(char *)xmlTextReaderConstValue(XMLReader) encoding:NSUTF8StringEncoding];
SEL selector = NSSelectorFromString([NSString stringWithFormat:#"set%#:", [currentTag capitalizedString]]);
[currentItem performSelector:selector withObject:currentTagValue];
If I add
[currentTagValue release]
the memory leaks are gone.
This seems strange to me, because I don't allocate memory for the NSString manually. That's why I thought it would be autoreleased.
The whole situation becomes stranger if I compare the upper code example with the part that is responsible for obtaining the node name.
NSString *currentTagName = [NSString stringWithCString:(char *)xmlTextReaderConstName(XMLReader) encoding:NSUTF8StringEncoding];
SEL selector = NSSelectorFromString([NSString stringWithFormat:#"set%#:", [currentTagName capitalizedString]]);
Here I dont't have to add a manual release, everything works fine and I'm getting no memory leak.
I'm not sure if my described problem is a side-effect of the xml...ConstValue function (the working part uses xml...ConstName) or if the reason is the performed selector afterwards.
Thanks for reading, I hope anyone can explain it to me.
Are you using libxml2? I haven't used libxml2 yet, but I googled quickly and found this:
http://xmlsoft.org/html/libxml-xmlreader.html
Function: xmlTextReaderConstValue
Returns: the string or NULL if not
available. The result will be
deallocated on the next Read()
operation.
Compare that with xmlTextReaderConstName
Function: xmlTextReaderConstName
Returns: the local name or NULL if not
available, the string is deallocated
with the reader.
It may be a leak in the lib, or a false alarm as the result seems to be on a delayed release (or something entirely different as I have no firsthand experience to say otherwise). Is the program crashing because of the leak or not? If it is not, maybe it's just a false alarm.
Hope it helps.