Mutex Memory Leak on 4.3.68

[barisyild]

In 4.3.2 there is no memory leak while in 4.3.68 there is memory leak.

I looked at both versions on xcode memory leak instrument.

[hughsando]

These objects should cleaned up by the garbage collector. Previously, the Deque and Semaphore were allocated internally, but now they are allocated using "new" which means they will show up on tools such as this. To get a correct reading, you should ensure that all thread pools are closed, and a full GC is run.

[barisyild]

These objects should cleaned up by the garbage collector. Previously, the Deque and Semaphore were allocated internally, but now they are allocated using "new" which means they will show up on tools such as this. To get a correct reading, you should ensure that all thread pools are closed, and a full GC is run.

As far as I know, xcode marks it as a leak if the object exists but has no reference left.

I think it's a memory leak.

https://stackoverflow.com/a/2320004

I should also add that this problem only occurs in the mutex, there is no other problem in another instances.

[hughsando]

It is still possible that the GC can find the object - it's just obfuscated from the tooling - because it is in the middle of a GC slab. This is why a final collect is useful. I don't think the behaviour has changed here. The object is not cleaned up, but I don't think it was before - you could just not see this because it did not use "new" to allocate the object.
If it leaks an object per thread, even after lots of collections, then I would be worried.