Re: [PATCH 12/30] mm: memory reserve management

On Mon, 2008-07-28 at 13:29 +0300, Pekka Enberg wrote:

quoted

quoted

quoted

+	}
+
+	obj = __kmalloc_node_track_caller(size, flags, node, ip);
+	WARN_ON(!obj);

Why don't we discharge from the reserve here if !obj?

Well, this allocation should never fail:
  - we reserved memory
  - we accounted/throttle its usage

Thus this allocation should always succeed.

But if it *does* fail, it doesn't help that we mess up the reservation
counts, no?

I guess you're right there. Will fix. Thanks!

quoted

quoted

quoted

+{
+	size_t size = ksize(obj);
+
+	kfree(obj);

We're trying to get rid of kfree() so I'd __kfree_reserve() could to
mm/sl?b.c. Matt, thoughts?

My issue with moving these helpers into mm/sl?b.c is that it would
require replicating all this code 3 times. Even though the functionality
is (or should) be invariant to the actual slab implementation.

Right, I guess we could just rename ksize() to something else then and
keep it internal to mm/.

That would be nice - we can stuff it into mm/internal.h or somesuch.

Also, you might have noticed, I still need to do everything SLOB. The
last time I rewrote all this code I was still hoping Linux would 'soon'
have a single slab allocator, but evidently we're still going with 3 for
now.. :-/

So I guess I can no longer hide behind that and will have to bite the
bullet and write the SLOB bits..

quoted

quoted

quoted

+	/*
+	 * ksize gives the full allocated size vs the requested size we used to
+	 * charge; however since we round up to the nearest power of two, this
+	 * should all work nicely.
+	 */
+	mem_reserve_kmalloc_charge(res, -size);
+}