Re: [PATCH 1/5] vmscan: remove all_unreclaimable check from direct reclaim path completely

[Minchan Kim <minchan.kim@gmail.com>]

On Wed, Mar 23, 2011 at 04:13:21PM +0900, KOSAKI Motohiro wrote:
> > Okay. I got it.
> > 
> > The problem is following as.
> > By the race the free_pcppages_bulk and balance_pgdat, it is possible
> > zone->all_unreclaimable = 1 and zone->pages_scanned = 0.
> > DMA zone have few LRU pages and in case of no-swap and big memory
> > pressure, there could be a just a page in inactive file list like your
> > example. (anon lru pages isn't important in case of non-swap system)
> > In such case, shrink_zones doesn't scan the page at all until priority
> > become 0 as get_scan_count does scan >>= priority(it's mostly zero).
> 
> Nope.
> 
>                         if (zone->all_unreclaimable && priority != DEF_PRIORITY)
>                                 continue;
> 
> This tow lines mean, all_unreclaimable prevent priority 0 reclaim.
> 

Yes. I missed it. Thanks. 

> 
> > And although priority become 0, nr_scan_try_batch returns zero until
> > saved pages become 32. So for scanning the page, at least, we need 32
> > times iteration of priority 12..0.  If system has fork-bomb, it is
> > almost livelock.
> 
> Therefore, 1000 times get_scan_count(DEF_PRIORITY) takes 1000 times no-op.
> 
> > 
> > If is is right, how about this?
> 
> Boo.
> You seems forgot why you introduced current all_unreclaimable() function.
> While hibernation, we can't trust all_unreclaimable.

Hmm. AFAIR, the why we add all_unreclaimable is when the hibernation is going on,
kswapd is freezed so it can't mark the zone->all_unreclaimable.
So I think hibernation can't be a problem.
Am I miss something?

-- 
Kind regards,
Minchan Kim