Thread safety in Objective-C categories - ios

I have a category on NSString:
- (CGSize) agb_sizeWithFont:(UIFont *)font width:(CGFloat)width lineBreakMode:(NSLineBreakMode)lineBreakMode {
if (!font) return CGSizeZero;
NSMutableParagraphStyle *paragraphStyle = [[NSMutableParagraphStyle alloc] init];
paragraphStyle.lineBreakMode = lineBreakMode;
NSDictionary *attributes = [NSDictionary dictionaryWithObjectsAndKeys:font, NSFontAttributeName, paragraphStyle, NSParagraphStyleAttributeName, nil];
// this line is not threadsafe
NSAttributedString *as = [[NSAttributedString alloc] initWithString:self attributes:attributes];
CGRect bounds = [as boundingRectWithSize:CGSizeMake(width, 10000) options:(NSStringDrawingUsesLineFragmentOrigin) context:nil];
return CGSizeMake(width, bounds.size.height);
}
I've observed intermittent EXC_BAD_ACCESS crashes when running this code on multiple threads (sometimes in initWithString: attributes:, sometimes in boundingRectWithSize:options:context:).
I believe my code is not thread-safe because self might deallocate on one thread while initWithString: attributes: is executing on the other.
Is my conclusion about this method's thread safety correct?
Would this code make it thread-safe?
NSString *strongSelf = self;
NSAttributedString *as = [[NSAttributedString alloc] initWithString:strongSelf attributes:attributes];
(By maintaining a reference to self, I attempt to ensure the object in memory is not deallocated while I'm using it.)
Is there any way to declare this method as not thread-safe? I'd love to generate a warning, for example, if a string declared as a nonatomic property is passed into this method.
Note: In case it's not obvious, I'm using ARC.

Any NSMutableString is-a NSString, too. Any method you add to NSString in a category can therefore be invoked on an NSMutableString, too. It is not safe to operate on a mutable string on one thread if it might be being mutated at the same time on another thread. It is safe to read a mutable string while it is being read on another thread.
Since your invocation of -[NSAttributedString initWithString:attributes:] is reading from the string self, that's not safe to do if self is a mutable string and might be mutated on another thread at the same time.
You might be tempted to review all calls of -agb_sizeWithFont:width:lineBreakMode: to see if the type of the receiver is NSMutableString*. However, that's not sufficient. An instance of NSMutableString may be pointed to by a variable of type NSString* (or other things like NSObject* or id), too.
You won't be able to make -agb_sizeWithFont:width:lineBreakMode: thread-safe, as such. Thread-safety doesn't decompose. It will have to be designed into the code which manages whatever strings it might be invoked on. That is, all of the code which touches a given string object and might invoke your method and, in other places, might mutate it, must be in charge of making sure that only one of those things is happening at a time.

Your code looks like ARC enabled so I assume it is enabled.
self might deallocate on one thread while initWithString: attributes: is executing on the other.
this is not possible with ARC, because in order to execute agb_sizeWithFont... method, you must doing something like this
NSString *str = /*...*/; // this will create a strong reference to str
[str agb_sizeWithFont:/*...*/]; // even str is released in other thread at this point, it won't be deallocated
NSString *strongSelf = self;
this code just create an extra strong reference to self, but since self must be retained somewhere already, it won't change anything
I don't think you can declare anything to be "not thread-safe" (actually everything are not thread-safe by default) and I don't understand how you want to use such warning.
methods in category have no difference compare to method in #implementation regards to thread-safety.

Related

Nonatomatic assign and memory leak issue

I have one property say
#property(nonatomic,assign) NSString *str;
Now I am having something like
self.str = [[NSString alloc] init];
self.str = #"test";
NSLog(#"%#",str);
[self.str release];
When I run I can see a leak "Potential leak of memory".
Why its showing me leak ?
Please guide me I am leaning phase of iOS
In other words (I'm just expanding on Anoop's answer here), you have two strings, not one.
self.str = [[NSString alloc] init];
self.str = #"test";
The thing on the right side of the first line is a string: [[NSString alloc] init]. But in the second line you throw it away, replacing it with a different string, namely #"test". Now there is NO REFERENCE pointing to the first string. Thus it leaks, since it can never be released nor can anything else ever be done to or for it.
The situation of the string created in the first line is, after the second line, like the situation of "thing1" in the second panel of this diagram:
No one is pointing to it, so its memory cannot be managed, and it lives forever in isolation (leak).
self.str = [[NSString alloc] init]; //1st
One string is allocated not used.
self.str = #"test"; // 2nd
Another constant string "test" is allocated but released.
So first one is a leak.
When you use factory method or create object using alloc,new,retain,copy,mutableCopy your object has +1 retain count every time. You own object in this case. You are responsible for releasing it. So you need to release object after you finish using object which cause -1 retain count to object.
It is not necessary to alloc you string again as you are already creating its property.
And it has its getter setter + you are allocating it again so gives you a leak.

What is difference between initilizing an NSString simply and with retain?

I want to know that difference between following two lines
name1 = [[NSString stringWithUTF8String:(char *)sqlite3_column_text(compiledStatement,1)] retain];
name1 = [NSString stringWithUTF8String:(char *)sqlite3_column_text(compiledStatement,1)];
What will be effect on name1 if I do use retain at the last,
I face once problem, and couldn't use name1 in a method that is being call by NSTimer, and when I use retain, they it worked fine for me.
If I do call value from database in viewDidLoad, and use in a method that is being called by NSTimer after each second, then it will give bad-exec, but when I do use retain then it will work properly,
I want to know the reason
Here is the difference
- (void)func1 {
name1 = [[NSString stringWithUTF8String:...] retain];
name2 = [NSString stringWithUTF8String:...];
}
- (void)func2 {
NSLog(#"%#", name1); //OK, name1 is still there
NSLog(#"%#", name2); //Would be crashed because name2 could be released anytime after func1 is finished.
}
I wrote this answer to another question, but it explains what you're asking:
Objects in objective c have a retain count. If this retain count is greater that 0 when the object goes out of scope (when you stop using it), it leaks.
The following things increase the retain count
[[alloc] init]
new
copy
[retain]
adding an object to an array
adding an object as a child (e.g. views)
There are likely more, but you don't appear to use any others in your code
The following decrease the retain count
[release]
removing an object from an array
if you dealloc an array, all of its objects are released
You should go through your code and ensure each of the retains or additions to an array are matched with a corresponding release. (You can release member variables in the dealloc method).
Another user made a valid point that my answer doesn't
Once you add an object to an array, it takes ownership and will release the object when it is done with it. All you need to do is make sure you release anything you own according to the memory management rules
There are also autorelease objects, have a look at this example;
-(init){
...
stagePickerArray = [[NSMutableArray alloc] init];
for (int i = 0; i < 3; i++)
{
//this string is autoreleased, you don't have call release on it.
//methods with the format [CLASS CLASSwithsomething] tend to be autorelease
NSString *s = [NSString stringWithFormat:#"%d", i);
[stagePickerArray addObject:s];
}
...
}
Your issue is that when you come to use your string later, it has a retain count of zero and has been released. By calling retain on it, you're saying 'I want to use this later'. Don't forget to match every retain with a release or you're objects will 'leak'
I bet your code wouldn't crash if your name1 was a property - either (nonatomic, retain) or just (copy) depending on your needs.
Second condition is to have name1 initialized to sth meaningful at the time your other function tries to do sth with it.
EDIT:
With a property you'd have to use synthesized setter in this case with: self.name1 = #"your string";.
Normally you don't have to manually retain/release a string created with stringWith... methods since there's nothing you created in memory yourself by using explicit alloc. Also please note that with code:
NSString *str = [NSString stringWithUTF8String:#"your string"];
your str (if not used to set a property) will stop being available when the function gets out of scope (iOS eventloop will autorelease it).

Need clarification in iPhone memory management

I need few clarification in iPhone memory management.
Here is an example for setters;
1).
-(void)setValue:(NSString*)input{
[value autorelease];
value = [input retain];
}
In this example, why we have to use autorelease?
Can we use like as follows?
if(value)
[value release];
value = [input retain];
In the first example, Why we should not release the memory for input
2).
If I use following statement; what is the retain count for value
NSString *value;
value = #"welcome";
After the above statement, just I am trying to set one more value. Then what will happen?
eg:
value = #"For testing";
3).
What is the different between 2) and 3)?
NSString *value;
value = [input1 retain];
...
...
value = [input2 retain];// Now what is the retain count for value
4).
If I use following statement, why the app is getting crash?
NSString *value = [[[NSString alloc] init] autorelease];
...
...
Thanks in advance..
If "input" is the exact same object as "value" then calling [value release] could dealloc the object. So you must retain the new input value, release the old value, then assign the new value to the ivar:
[input retain];
[value release];
value = input;
After each of 2) and 3), the NSString *value points to a literal NSString object, the retain count will be 1 in each case, and releasing it is probably not a good idea
After this code:
value = [input2 retain];
value is an alias to the input2 object. The thing to realize is that objects have retain counts, variables do not.
As for your last case,
NSString *value = [[[NSString alloc] init] autorelease];
It creates an autoreleased empty string. If you reference that object again once the autorelease actually happens, you may get a crash because you'll be referring to an object that doesn't exist any more.
If you release a value before you retain the new value then you can have problems if the same value is being set twice. This happens if the caller hasn't retained their own copy, such as when they get the value from the same object they try to set it on, like this:
object.value = object.value;
That statement will cause the object to be released before it's retained again which could lead to the memory being deallocated and result in a dangling pointer being retained. By doing the autorelease it ensures that copying the same pointer onto itself will work correctly.
I usually write my setters as
- (void)setValue:(NString *)input {
if (value != input) {
[value release];
value = [input retain];
}
}
which avoids the problem of input and value both being the same object. If you just release it without checking then you might completely free it and the next line will try to retain an object that doesn't exist anymore.
However, it's best practice to copy strings instead of retaining them :)
(assuming you're working in a non-gc env)
1) you can create a temporary variable and release the temp after the retain/assign. otherwise, you'd need to compare the pointers. deferred release may also mask threading errors (whether you consider that good or bad is...)
2) technically, the NSString literals are valid for the life of your program. that is: while (1) [#"why won't i die?" release]; yields an infinite loop.
3) with explicit retain, alloc+init, new, and copy, you must counterbalance the retain count using release or autorelease. since you did not release value1, the static analysis may (correctly) spot that as a leak. since the string constants never die, the two aren't comparable in this regard.
4) there's nothing wrong with that staement. the problem lies elsewhere in your program. either you are assigning it to an ivar without retaining, or releasing it later.
try using static analysis often, and try reducing how much you use autorelease (you can't avoid it entirely).
none of this is magic, but you can at least reduce the location of many problems to the callsites (or very close) by not using autorelease all over the place. just use manual retain/release where possible.
lastly, check for leaks and run with NSZombies enabled.
1)
You will have to do autorelease because of the following:
When input is the same object as value and you will release value it's retain count will reach zero and get deallocated before you can retain it again though the input.
You can do it with retain but you have to change your code:
-(void)setValue:(NSString*)input{
if (value != input) {
[value autorelease];
value = [input retain];
}
}
2)
I believe #"Text" will be treated as a constant. When you want a object which you do not want any memory management with use:
NSString *value = [NSString stringWithString:#"Text"];
This will return an autoreleased object.
3) In this example it is not about the retain count of value, but about the retain count of both objects where value one is referenced to.
When you dont release input1 before you leave that method, you will have a memory management problem.
4) This statement should work. No points to argue. You rather use [NSString string].
Note:
For memory management: when you use alloc new or copy, you also have to use release or autorelease on the same object in the same scope.

Why myInstance = nil instead of self.myInstance = nil?

Why would I use (inside my dealloc method)?
[myInstance release] instead of [self.myInstance release]
myInstance = nil instead of self.myInstance = nil
Although we use self.myInstance = [[[AClass alloc] init] autorelease] instead of myInstance = [[[AClass alloc] init] autorelease]?
Those practices are from numerous examples I see on the web.
1) [myInstance release] instead of [self.myInstance release]
prefer the former.
the returned value of self.myInstance is defined by implementation when a subclass has overridden the method myInstance. you're not interested in the behaviour of the interface of a constructed object during dealloc (since a subclass may override and return something other than your ivar).
what you are interested in dealloc is releasing the references you own before your object is destroyed. if the subclass has overridden myInstance, then it could:
a) return an ivar (declared in the subclass) that's already been released
or
b) the implementation of the override may return a newly created autoreleased object
either a or b could lead to an over-release and a crash (assuming everything else is correctly retained/released). this also suggests why you should assign nil to the ivar after releasing it.
this is also a classic example of how to trigger object resurrection. object resurrection occurs when an implementation of the getter/setter you call recreates its state after it's already been deallocated. the least offensive side-effect would cause a harmless leak.
2) myInstance = nil instead of self.myInstance = nil
again, prefer the former.
a formal response would look much like the response to #1 -- the rationale, side-effects and dangers apply here as well.
the safest way to handle this is to access the ivar directly:
[myInstance release], myInstance = nil;
because there may be really nasty side-effects (crashes, leaks, resurrection) which may be difficult to reproduce.
these dangers may be easily avoided and your code will be far easier to maintain. on the other hand, if people encounter the side-effects when using your programs, they will probably avoid (re)using it wherever they can.
good luck
Calling self.myInstance = uses the auto generated setter method. Calling [self.myInstance release]; calls release on the object returned by your getter method. It all depends on how your properties were set up (retain, assign?). There is no necessarily right or wrong answer to your question, since it all depends on the property in question. I suggest you read up on Objective C properties to get a better feel for this kind of thing.
And, unless myInstance was declared with assign, you wouldn't want to call self.myInstance = [[AClass alloc] init] You'd be much better off with self.myInstance = [[[AClass alloc] init] autorelease]
Note that using
myInstance = nil
instead of
self.myInstance = nil
Is incorrect (in the context of say a viewDidUnload method in a UIViewController subclass) if myInstance is a retain property, since if myInstance points to an object, it will be leaked!
This depends on a property that you defined in interface. For example if you define retain property:
#property (nonatomic, retain) NSObject *property;
then you may use just self.property = nil; in dealloc method, because it equals to:
[property release]; // releases previous property
property = [nil retain]; // [nil retain] returns just nil
The very same thing with self.property = [[A alloc] init];. This equals to
[property release]; // releases previous property
property = [[[A alloc] init] retain];
in case of property = [[A alloc] init]; property won't be retained.
Here's a full properties guide form Apple.
Actually using
self.myInstance = [[AClass alloc] init];
will lead in a memory leak, cause self.myInstance is using setter methods which leads in retain +1 along with alloc/init retain +1. So you'll get a retain count +2;
... = self.myInstance
and
self.myInstance = ...
are actually subroutine or method calls to getters and setters, which depending on how you define these subroutines, or have Objective C Properties create them, could do almost anything.
If the case of retain properties, the subroutines might play with the retain counts. If you do your own getters and setters, you could have them control the lights in your house, turning them on for none zero sets and turning the lights out when setting something to zero or nil. There doesn't even need to be a backing variable named "instance" which could be set by:
instance = ...

Objective C NSString* property retain count oddity

I have the following example class:
Test.h:
#interface Test : UIButton {
NSString *value;
}
- (id)initWithValue:(NSString *)newValue;
#property(copy) NSString *value;
Test.m:
#implementation Test
#synthesize value;
- (id)initWithValue:(NSString *)newValue {
[super init];
NSLog(#"before nil value has retain count of %d", [value retainCount]);
value = nil;
NSLog(#"on nil value has retain count of %d", [value retainCount]);
value = newValue;
NSLog(#"after init value has retain count of %d", [value retainCount]);
return self;
}
Which produces the following output:
2008-12-31 09:31:41.755 Concentration[18604:20b] before nil value has retain count of 0
2008-12-31 09:31:41.756 Concentration[18604:20b] on nil value has retain count of 0
2008-12-31 09:31:41.757 Concentration[18604:20b] after init value has retain count of 2147483647
I am calling it like:
Test *test = [[Test alloc] initWithValue:#"some text"];
Shouldn't value have a retain count of 1? What am I missing?
Thanks for your help.
Don't look at retain counts. They're not useful and will only mislead you — you can't be certain that nothing else is retaining an object, that an object you get from somewhere isn't shared.
Instead, concentrate on object ownership and follow the Cocoa memory management rules to the letter. That way your memory management will be correct no matter what optimizations Cocoa may be doing behind the scenes for you. (For example, implementing -copy as just -retain for immutable objects.)
Furthermore, it's critical to understand the difference between properties of your objects and instance variables within your objects. In your question's code, you are assigning a value to an instance variable. That instance variable is just that: a variable. Assigning to it will behave like any other variable assignment. To use the property, you must use either dot syntax or bracket syntax to actually invoke the property's setter method:
self.value = newValue; // this is exactly equivalent to the next line
[self setValue:newValue]; // this is exactly equivalent to the previous line
The code generated for the dot syntax and the bracket syntax is identical, and neither will access the instance variable directly.
You are passing in a literal string. The compiler probably allocates it in static memory and sets the retain count to the maximum possible value.
Try a dynamically allocated string instead and see what happens.
NSString* string = [[NSString alloc] initWithString: #"some text"];
Test* test = [[Test alloc] initWithValue: string];
You've got a reference to an immutable string. Assignment doesn't need to copy the value (the string data) since it's immutable. If you do a mutable operation, like value = [newValue uppercaseString] then it should copy the bits into value, and value's retain count incremented.
You're passing in a string constant, which can't really be deallocated. I think that 2147483647 is probably UINT_MAX, which basically means that the object can't be released.
I think you want to do this:
self.value = newValue;
which will invoke the property setter and cause the copy to occur. "value = newValue" simply assigns a pointer value to the instance variable.
You shouldn't be paying attention to the retain counts, just follow the Cocoa memory management rules. http://iamleeg.blogspot.com/2008/12/cocoa-memory-management.html
hmm.. we're getting closer.
it appears that newValue's retain count is also 2147483647.
I tried dynamically allocating the string instead with the same retain count results.
I found a helpful article here: http://www.cocoadev.com/index.pl?NSString
FTA:
Does the NSString returned by #"" need to be released, or is it autoreleased?
Neither. #""-strings are of class NSConstantString?, and thus act like atoms in lisp; they hang around. That is, if you use #"cow" in two separate places in your code, they will be referencing the very same object.
I don't think -release or -autorelease does anything to either of them.
If I have "copy" on the property though, shouldn't it copy the contents of the target memory into new memory with a retain count of 1? It would seem the copy attribute does nothing in this case?
#import <Foundation/Foundation.h>
int main (int argc, const char * argv[]) {
NSAutoreleasePool * pool = [[NSAutoreleasePool alloc] init];
char *cstr = "this is a c string";
NSString *str = [[NSString alloc] initWithUTF8String:cstr];
NSLog(#"rc1: %d", [str retainCount]);
[pool drain];
return 0;
}
If you run the above code, it will display a retain count of 1
In Cocoa, many immutable objects will simply retain themselves when you ask for a copy within the same zone. If the object is guaranteed not to change (i.e. its immutableness) then an exact duplicate is redundant.
In Objective-C, the constant string class is separate to Cocoa's NSString class, although it may be a subclass of NSString (I'm not too sure). This constant string class may override NSObject's methods like retain, release and dealloc so that they do nothing, and also override retainCount so that it always returns the same number, UINT_MAX or so. This is because an Objective-C constant string is created in static memory. It must have the overall general behaviour of a Cocoa object (when using Cocoa) so that it can be added to arrays, used as keys to a dictionary etc, except in regards to its memory management, since it was allocated differently.
Disclaimer: I don't actually know what I'm talking about.

Resources