>I'm not a mm expert (far from it really) but the last two lines sound
>suspicios to me. Isn't this what kswapd is made for? Are you duplicating
>its work? IOW, it seems not so clear to me.
kswapd has to do the background pageout. If at __get_free_pages() time we
are short in memory we can' t wait kswapd so we are _used_ to try to do
some work by hand. This was so in 2.0 and 2.1. Look at the patch and you'
ll notice that __get_free_pages() was running try_to_free_pages in some
cases also before my patch. Note also that with my patch
try_to_free_pages() is not run at every __get_free_pages but only if
there' s no _one_ free page. You must follow the tricky macro used
in __get_free_pages to notice this.
Andrea[s] Arcangeli
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/