> root@dragon:/proc/sys/vm# cat pagecache
> 1 1 1
> root@dragon:/proc/sys/vm# cat freepages
> 2000 5000 10000
Freepages isn't used any more (and will be re-used RSN).
> Am I interpreting the docs right? (using 1 2 3 for pagecache don' t make
> differences) It should be the the normal beaviour?
It seems there's a typo in mm/vmscan.c, I'll fix this
too.
> >The change we'll probably need is changing the swap policy
> >to SC_ROUNDROBIN from SC_whatwehavenow.
> >(see the 1.2.xx source code for references :)
>
> I really don' t think the problems is the swap code. I think the problem
> is the cache allocating code. Only it. Then the swapping code try to free
> memory swapping all out. As ever I never got into mm details so I can be
> wrong.
You are :) Kswapd frees pagecache & buffermem pages first
when one of them is above maximum...
When both (one of them?) is under borrowpercent, kswapd
should free one cache/buffer page and one normal page in
turn.
(which it currently doesn't do)
Rik.
+-------------------------------------------+--------------------------+
| Linux: - LinuxHQ MM-patches page | Scouting webmaster |
| - kswapd ask-him & complain-to guy | Vries cubscout leader |
| http://www.phys.uu.nl/~riel/ | <H.H.vanRiel@phys.uu.nl> |
+-------------------------------------------+--------------------------+
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu