• Open Menu Close Menu
  • Apple
  • Shopping Bag
  • Apple
  • Mac
  • iPad
  • iPhone
  • Watch
  • TV
  • Music
  • Support
  • Search apple.com
  • Shopping Bag

Lists

Open Menu Close Menu
  • Terms and Conditions
  • Lists hosted on this site
  • Email the Postmaster
  • Tips for posting to public mailing lists
Re: NSPersistentStoreCoordinator Wars Episode I: The Phantom Lock Menace
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: NSPersistentStoreCoordinator Wars Episode I: The Phantom Lock Menace


  • Subject: Re: NSPersistentStoreCoordinator Wars Episode I: The Phantom Lock Menace
  • From: sanchezm <email@hidden>
  • Date: Thu, 8 Jan 2009 22:12:12 -0800

Nick,

I can't be sure on what thread access pattern you are following in your application, but it sounds like you might be trying to access the same object and managed object context instances from different threads. This is tough to get right.
The preferred way of doing this is to have a separate context per thread pass object IDs between threads


http://developer.apple.com/DOCUMENTATION/Cocoa/Conceptual/CoreData/Articles/cdMultiThreading.html#/ /apple_ref/doc/uid/TP40003385-SW2

In the above document you will also see the statement "If you lock (or successfully tryLock) a context, that context must be retained until you invoke unlock. If you don’t properly retain a context in a multi-threaded environment, you may cause a deadlock."

Another thing you could try if have captured the deadlock in gdb is to try to determine which of the other active threads hasn't released the lock yet. You might do this by keeping a per thread counter that you increment when you lock and decrement when you unlock. When the deadlock is caught in gdb, you can inspect the other threads to see which one has the counter in a locked state.

Finally, if you want to load data in a background thread you might want to take a look at the example

/Developer/Examples/CoreData/BackgroundFetching

	I hope all this gives you some hints,

- Miguel

On Jan 8, 2009, at 6:14 PM, Nick Zitzmann wrote:

Yes, I did try searching and found nothing interesting...

I'm working on a project that uses CoreData objects on multiple threads at once, where it's not uncommon for the main thread to be loading data at the same time that an NSOperation running in the background is processing a different set of data.

Whenever I fetch or store data from an NSManagedObjectContext, I always make sure to lock and unlock it using the provided methods. But I'm having a problem where, if there are a lot of records in the model ("a lot" defined as 1,000+), some phantom object is leaving a stray lock open on the NSPersistentStoreCoordinator object, so I get a deadlock as one thread is waiting for the NSManagedObjectContext to unlock while the thread that has the lock is twiddling its thumbs waiting forever for the NSManagedObjectContext to come unlocked. And it's happening on all four CPU architectures.

How do I figure out who is locking the persistent store coordinator and not unlocking it properly? It's a real mystery, since I've audited my code several times now, and all locks appear to be balanced (and until very recently, I wasn't even locking the persistent store coordinator at all). I already tried setting breakpoints on -[_PFLock lock] and -[_PFLock unlock] that print the stack traces, but it's taking hours due to all of the locking and unlocking going on.

All I need is to get a stack trace to the code that locked the context but forgot to unlock it, which I'd normally do by subclassing the lock object and have the subclass keep track of stack traces, but I can't hack CoreData to do this (and yes, I did try posing, but it was too glitchy, and won't work on PPC64 or X86-64 anyway).

I've also tried Guard Malloc, which can often catch glitchy behavior, but so far it hasn't found anything either.

Or am I using the wrong locks in the first place?

Nick Zitzmann
<http://www.chronosnet.com/>

_______________________________________________

Cocoa-dev mailing list (email@hidden)

Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com

Help/Unsubscribe/Update your Subscription:
email@hidden


This email sent to email@hidden

_______________________________________________

Cocoa-dev mailing list (email@hidden)

Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com

Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden


References: 
 >NSPersistentStoreCoordinator Wars Episode I: The Phantom Lock Menace (From: Nick Zitzmann <email@hidden>)

  • Prev by Date: Re: [newbie] Use of accessors versus direct access of ivars
  • Next by Date: Re: [newbie] Use of accessors versus direct access of ivars
  • Previous by thread: NSPersistentStoreCoordinator Wars Episode I: The Phantom Lock Menace
  • Next by thread: setting accessibility links for cocoa applications in IB
  • Index(es):
    • Date
    • Thread