Nick Piggin wrote:

Here are a few of the more harmless mm patches I have been sitting on
for a while. They've had some testing in my tree (which does get used
by a handful of people).


------------------------------------------------------------------------


This reworks alloc_pages a bit.

Previously the ->protection[] logic was broken. It was difficult to follow
and basically didn't use the asynch reclaim watermarks properly.

This one uses ->protection only for lower-zone protection, and gives the
allocator flexibility to add the watermarks as desired.

Note that this patch strictly enforces the lower zone protection (which is
currently disabled anyway) instead of allowing GFP_ATOMIC allocations to
get at them.

It also does a few minor things like not taking rt_task into account during
the first loop (because the kswapd reclaim watermarks shouldn't depend on
that), and also only checking rt_task if !in_interrupt.

The biggest thing it does is use the kswapd watermarks correctly - and I've
generally observed lower allocstall, and kswapd being more productive.
--
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:"aart@kvack.org"> aart@kvack.org </a>