Re: [PATCH] mmtom: Prevent shrinking of active anon lru list in case of no swap space V2

[Minchan Kim]

On Thu, May 14, 2009 at 10:18 PM, Rik van Riel <riel@xxxxxxxxxx> wrote:
> Minchan Kim wrote:
>>
>> HI, Rik
>>
>> Thanks for careful review. :)
>>
>> On Thu, May 14, 2009 at 9:58 PM, Rik van Riel <riel@xxxxxxxxxx> wrote:
>>>
>>> Minchan Kim wrote:
>>>
>>>> Now shrink_active_list is called several places.
>>>> But if we don't have a swap space, we can't reclaim anon pages.
>>>
>>> If swap space has run out, get_scan_ratio() will return
>>> 0 for the anon scan ratio, meaning we do not scan the
>>> anon lists.
>>
>> I think get_scan_ration can't prevent scanning of anon pages in no
>> swap system(like embedded system).
>> That's because in shrink_zone, you add following as
>>
>> Â Â Â Â/*
>> Â Â Â Â * Even if we did not try to evict anon pages at all, we want to
>> Â Â Â Â * rebalance the anon lru active/inactive ratio.
>> Â Â Â Â */
>> Â Â Â Âif (inactive_anon_is_low(zone, sc))
>> Â Â Â Â Â Â Â Âshrink_active_list(SWAP_CLUSTER_MAX, zone, sc, priority,
>> 0);
>
> That's a fair point.
>
> How about we change this to:
>
> Â Â Â Âif (inactive_anon_is_low(zone, sc) && nr_swap_pages >= 0)
> That way GCC will statically optimize away this branch on
> systems with CONFIG_SWAP=n.
>
> Does that look reasonable?

Now inactive_anon_is_low called following as.

1. shrink_zone => Looks good since your idea.
2. balance_pgdat => Looks good since aging.
3. shrink_list
shrink_list is called at two places.
1. shrink_zone => It's OK since get_scan_ratio can't prevent it.
2. shrink_all_zones. => It's OK since we can't suspend without swap space.

So, Okay I will do that in next version.
Thanks for good review. Rik :)
Could I add your ack in next version ?

> --
> All rights reversed.
>



-- 
Kinds regards,
Minchan Kim
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/