Hello, On Tue, Nov 16, 2010 at 12:57:26PM +0900, KAMEZAWA Hiroyuki wrote: > Hmm. I think this patch is troublesome. > > This patch will make memcg's pageout routine _not_ throttoled even when the whole > system vmscan's pageout is throttoled. > > So, one idea is.... > > Make this change > == > +++ b/mm/vmscan.c > @@ -1844,7 +1844,7 @@ static void shrink_zone(int priority, struct zone *zone, > if (inactive_anon_is_low(zone, sc)) > shrink_active_list(SWAP_CLUSTER_MAX, zone, sc, priority, 0); > > - throttle_vm_writeout(sc->gfp_mask); > + throttle_vm_writeout(sc->gfp_mask, sc->mem_cgroup); > } > == > as > > == > > if (!sc->mem_cgroup || throttle_vm_writeout(sc->gfp_mask, sc->mem_cgroup) == not throttled) > throttole_vm_writeout(sc->gfp_mask, NULL); > > Then, both of memcg and global dirty thresh will be checked. Good point, both limits should apply. I'd prefer to stuff it all into throttle_vm_writeout() and not encode memcg-specific behaviour into the caller, though. -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxxx For more info on Linux MM, see: http://www.linux-mm.org/ . Fight unfair telecom policy in Canada: sign http://dissolvethecrtc.ca/ Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>