Martin J. Bligh wrote:

quoted

Could someone test that it works on real SMP?

Tested on 16-way NUMA-Q (shows up races quicker than anything ;-)). 
Boots, compiles the kernel 5 times OK. That's good enough for me. 
No performance regression, in fact was marginally faster (within 
experimental error though).

Thanks for the test. NUMA is on my TODO list, after figuring out 
where/how to drain cpu caches and the free list.

I've found one stupid bug with debugging enabled: the new debug code 
tries to poison NULL pointers, with limited success :-(

And one limitation might be important for arch specific code: 
kmem_cache_create() during mem_init() is not possible anymore.

--
	Manfred

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/