On Fri, Aug 02, 2013 at 11:37:23AM -0400, Johannes Weiner wrote:

Changes in version 2:

v2 looks great to me.

                zone->alloc_batch -= 1U << order;
    3147:       d3 e0                   shl    %cl,%eax
    3149:       29 42 54                sub    %eax,0x54(%rdx)

gcc builds it as one asm insn too.

Considering we depend on gcc to be optimal and to update ptes in a
single insn (and if it doesn't we'll corrupt memory), keeping it in C
should always provide the update in a single insn.

I believe the error introduced when mulptiple CPUs of the same NUMA
node step on each other is going to be unmeasurable.

ACK the whole series.

Signed-off-by: Andrea Arcangeli <redacted>

Thanks,
Andrea

--
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/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>