Linus Torvalds wrote:
quoted text
>> I suspect the right thing to do is not to mark them for "IO", but mark 
>> them for "short-lived", and allow short-lived allocations that don't 
>> have extended lifetimes to succeed even when a "real" allocation 
>> wouldn't.

Pekka Enberg wrote:
quoted text
> Yeah, makes sense. We do have GFP_TEMPORARY so we could associate this 
> new semantics with that. But the real problem here is how to do the 
> "allocate harder" part which, btw, sounds very similar to what Peter's 
> kmalloc reserve patches try to do...

Actually, a trivial way to implement that is to have a few "emergency 
kmalloc" caches say for sizes 64, 128, 256, and 512 that have some 
pre-allocated pages into which these GFP_TEMPORARY allocations are 
allowed to dip into on OOM and OOM only.

		Pekka
--
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/