re: Locating managed objects within ObjectAlloc (was Re: Garbage collection, core data, and tight loops)

FROM : Ben Trumbull
DATE : Sun Nov 04 00:08:42 2007

> This sounds like caching behavior that I think I've observed.  My app
> create millions of managed objects during an import process, and I've
> seen large memory usage that doesn't go down when the objects are
> dealloced.


You mean RSIZE as reported by 'top', not malloc's free heap space as 
reported by 'heap'.

> Looking at backtraces in ObjectAlloc, I see a lot of
> mallocs in caching functions as various objects are dealloc'ed.  If I
> close the NSPersistentDocument, I get some memory reclaimed, but not
> what I expect.  But, if I repeated run the import (without quitting),
> the peak memory doesn't go up, so the memory is apparently reused.


Correct.  The memory is freed (heap space), but not returned to the 
kernel (VM mapping).  This is the behavior of malloc on OSX.  You will 
see a high watermark effect.

> This leads me to a question about using ObjectAlloc (in Instruments on
> Leopard) effectively, especially with respect to Core Data.  When
> looking at allocations, there are a large number of GeneralBlocks of
> various sizes, but no NSManagedObjects.  I have some custom
> subclasses, too, and those aren't identified either in the Categories
> column.  I want to observe the usage patterns of the various entities
> in my model, but I'm not sure how to do that.  Is there an easy way to
> filter managed objects?


No, allocation events for managed objects are not currently tracked by 
ObjectAlloc except as general blocks.  The 'heap' and 'leaks' tools 
both perceive managed objects.  You'll probably find the 'heap' tool 
useful for experimenting with these questions.  Also, 'malloc_history' 
is much improved on Leopard and often overlooked.

- Ben