Re: [PATCH] 2.6.4-rc2-mm1: vm-split-active-lists

[Mike Fedyk]

Jamie Lokier wrote:
> Nick Piggin wrote:
>
> > In Linux, all reclaim is driven by a memory shortage. Often it
> > is just because more memory is being requested for more file
> > cache.
>
>
> Is reclaim the same as swapping, though?  I'd expect pages to be
> written to the swapfile speculatively, before they are needed for
> reclaim.  Is that one of those behaviours which everyone agrees is
> sensible, but it's yet to be implemented in the 2.6 VM?

Nobody has mentioned the swap cache yet.  If a page is in ram, and swap 
and not dirty, it's counted in the swap cache.

> > But presumably if you are running into memory pressure, you really
> > will need to free those free list pages, requiring the page to be
> > read from disk when it is used again.
>
>
> The idea is that you write pages to swap _before_ the memory pressure
> arrives, which makes those pages available immediately when memory
> pressure does arrive, provided they are still clean.  It's speculative.
>
> I thought Linux did this already, but I don't know the current VM well.

You're saying all anon memory should become swap_cache eventually 
(though, it should be a background "task" so it doesn't block userspace 
memory requests).

That would have other side benefits.  If the anon page matches (I'm not 
calling it "!dirty" since that might have other semantics in the current 
VM) what is in swap, it can be cleaned without performing any IO.  Also, 
 suspending will have much less IO to perform before completion.

Though there would have to be swap recycling algo if swap size < ram.

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