Re: kswapd's priority

Rik van Riel (H.H.vanRiel@phys.uu.nl)
Fri, 26 Jun 1998 13:41:34 +0200 (CEST)

On Fri, 26 Jun 1998, Stephen C. Tweedie wrote:
> <H.H.vanRiel@phys.uu.nl> said:
>
> > There's only one problem with that: tasks don't wait for
> > memory. If the memory's there they grab it, otherwise they
> > try to free some theirselves; and if that fails, they die.
>
> They don't wait yet. Arguably, they should do, or else we effectively
> have multiple copies of kswapd active at once, which is not
> necessarily a Good Thing. The interactions between the multiple
> instances of try_to_free_page may seriously impede our attempts to
> defragment and cluster the swap IOs.

They don't. All try_to_free_page() calls are made with
WAIT=0. Only after a certain number of tries the async
swap pages are synced to disk, this is going to give
relatively good swap clustering.

Besides, when there are parrallel programs calling
try_to_free_page(), this just means that we're very
low on memory and have to free loads of it...
This might just be a Good Thing :)

Rik.
+-------------------------------------------------------------------+
| Linux memory management tour guide. H.H.vanRiel@phys.uu.nl |
| Scouting Vries cubscout leader. http://www.phys.uu.nl/~riel/ |
+-------------------------------------------------------------------+

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu