// Then I create a repeating timer
[NSTimer scheduledTimerWithTimeInterval: 5
target: self
selector: @selector(performDirectoryTasksWithTimer:)
userInfo: nil
repeats: YES];
return self
// When the timer fires, it calls performDirectoryTasksWithTimer
which is defined like so:
- (void) performDirectoryTasksWithTimer:(NSTimer *) inTimer
{
[NSThread detachNewThreadSelector: @selector(scanDirectoriesInThread:)
toTarget: self
}
// I create the thread because the scanning is pretty time consuming
(12 minutes on average) and I want the machine to be usable during
this time
// Here's the scanDirectoriesInThread definition (where I allocate an
autorelease pool)
- (void) scanDirectoriesInThread:(NSString *) inString
{
// allocate a new autorelease pool
NSAutoreleasePool *pool = [[NSAutoreleasePool alloc] init];
// map the directories
[self initDirectories];
// release the pool
[pool release];
}
// Which in turn calls the method that actually performs the scan
- (void) initDirectories
{
// The malloc/free happens in here and the data in the malloc'd
pointer is packed into an NSData and put in "holdingBin" allocated in
the init routine
[ holdingBin addObject: [NSData dataWithBytes: ptr length: len]];
// When I try to free pointers in here, I get those errors
}
See any obvious allocation related gotchas here?
Thanks
Ken
On Jul 29, 2006, at 6:14 PM, stephen joseph butler wrote:
2006/7/29, Ken Tozier <email@hidden>:
On Jul 29, 2006, at 4:47 PM, Ken Tozier wrote:
> I have a process that spawns threads and inside these threads I
> malloc/free some memory but I'm getting either "xxx has exited due
> to signal 11 (SIGSEGV)" or "sigbus 10" errors. I know from past
> experience that these most likely indicate a non-retained object,
> problem is though that I checked very carefully and everything that
> I want to keep around is definitely retained. Could the fact that
> some objects were created inside a thread's NSAutorelease alloc/
> release override the retain commands?
I'm confused as to what malloc/free has to do with NSAutoreleasePool.
Nothing returned by malloc should ever end up in the autorelease pool.
In fact, nothing returned from malloc should ever be used as an
Objective-C object. All Objective-C objects come from either
alloc/init (which are not autoreleased, unless you send the
autorelease message) or a convience message (which are autoreleased).
It seems like the NSAutorelease pool destroys all objects created
within it regardless of whether they have a retain or not. That's the
only thing I can think of. Is that what it does?
No. It's a delayed release. It says "call release on this object
sometime after my function ends."
Since you mentioned multithreading, I should point out that if you're
going to autorelease objects inside a new thread then you need to
create your own autorelease pools. See http://developer.apple.com/documentation/Cocoa/Conceptual/
Multithreading/articles/CocoaDetaching.html#//apple_ref/doc/uid/
20000738-97249
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Cocoa-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/cocoa-dev/email@hidden
