Re: [patch 1/14] mm: opt rmqueue

[Nick Piggin <nickpiggin@yahoo.com.au>]

Andi Kleen wrote:
> Nick Piggin <nickpiggin@yahoo.com.au> writes:
> 
> 
>>1/14
>>
>>-- 
>>SUSE Labs, Novell Inc.
>>
>>Slightly optimise some page allocation and freeing functions by
>>taking advantage of knowing whether or not interrupts are disabled.
> 
> 
> Another thing that could optimize that would be to use local_t
> for the per zone statistics and the VM statistics (i have an
> old patch for the later, needs polishing up for the current
> kernel) 
> With an architecture optimized for it (like i386/x86-64) they
> generate much better code.
> 

Yes, all this turning on and off of interrupts does have a
significant cost here.

With the full patchset applied, most of the hot path statistics
get put under areas that already require interrupts to be off,
however there are still a few I didn't get around to doing.
zone_statistics on CONFIG_NUMA, for example.

I wonder if local_t is still good on architectures like ppc64
where it still requires an ll/sc sequence?

Nick

-- 
SUSE Labs, Novell Inc.

Send instant messages to your online friends http://au.messenger.yahoo.com