On Tue, Aug 31, 2010 at 10:38 AM, KOSAKI Motohiro
<kosaki.motohiro@xxxxxxxxxxxxxx> wrote:
>> > I think both Ying's and Minchan's opnion are right and makes sense. however I _personally_
>> > like Ying version because 1) this version is simpler 2) swap full is very rarely event 3)
>> > no swap mounting is very common on HPC. so this version could have a chance to
>> > improvement hpc workload too.
>>
>> I agree.
>>
>> >
>> > In the other word, both avoiding unnecessary TLB flush and keeping proper page aging are
>> > performance matter. so when we are talking performance, we always need to think frequency
>> > of the event.
>>
>> Ying's one and mine both has a same effect.
>> Only difference happens swap is full. My version maintains old
>> behavior but Ying's one changes the behavior. I admit swap full is
>> rare event but I hoped not changed old behavior if we doesn't find any
>> problem.
>> If kswapd does aging when swap full happens, is it a problem?
>> We have been used to it from 2.6.28.
>>
>> If we regard a code consistency is more important than _unexpected_
>> result, Okay. I don't mind it. :)
>
> To be honest, I don't mind the difference between you and Ying's version. because
> _practically_ swap full occur mean the application has a bug. so, proper page aging
> doesn't help so much. That's the reason why I said I prefer simper. I don't have
> strong opinion. I think it's not big matter.
>
>
>> But at least we should do more thing to make the patch to compile out
>> for non-swap configurable system.
>
> Yes, It makes embedded happy :)
>
>
How about this?
(Not formal patch. If we agree, I will post it later when I have a SMTP).
Signed-off-by: Ying Han <yinghan@xxxxxxxxxx>
Signed-off-by: Minchan Kim <minchan.kim@xxxxxxxxx>
diff --git a/mm/vmscan.c b/mm/vmscan.c
index 3109ff7..c3c44a8 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -1579,7 +1579,7 @@ static void shrink_active_list(unsigned long
nr_pages, struct zone *zone,
__mod_zone_page_state(zone, NR_ISOLATED_ANON + file, -nr_taken);
spin_unlock_irq(&zone->lru_lock);
}
-
+#if CONFIG_SWAP
static int inactive_anon_is_low_global(struct zone *zone)
{
unsigned long active, inactive;
@@ -1605,12 +1605,21 @@ static int inactive_anon_is_low(struct zone
*zone, struct scan_control *sc)
{
int low;
+ if (nr_swap_pages)
+ return 0;
+
if (scanning_global_lru(sc))
low = inactive_anon_is_low_global(zone);
else
low = mem_cgroup_inactive_anon_is_low(sc->mem_cgroup);
return low;
}
+#else
+static inline int inactive_anon_is_low(struct zone *zone, struct
scan_control *sc)
+{
+ return 0;
+}
+#endif
static int inactive_file_is_low_global(struct zone *zone)
{
@@ -1856,7 +1865,7 @@ static void shrink_zone(int priority, struct zone *zone,
* Even if we did not try to evict anon pages at all, we want to
* rebalance the anon lru active/inactive ratio.
*/
- if (inactive_anon_is_low(zone, sc) && nr_swap_pages > 0)
+ if (inactive_anon_is_low(zone, sc))
shrink_active_list(SWAP_CLUSTER_MAX, zone, sc, priority, 0);
throttle_vm_writeout(sc->gfp_mask);
--
Kind regards,
Minchan Kim
--
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/ .
Don't email: <a href