Re: [PATCH 5/6] mm: kswapd: Treat zone->all_unreclaimable insleeping_prematurely similar to balance_pgdat()

[Mel Gorman]

On Fri, Dec 10, 2010 at 10:23:37AM +0900, KAMEZAWA Hiroyuki wrote:
> On Thu,  9 Dec 2010 11:18:19 +0000
> Mel Gorman <mel@xxxxxxxxx> wrote:
> 
> > After DEF_PRIORITY, balance_pgdat() considers all_unreclaimable zones to
> > be balanced but sleeping_prematurely does not. This can force kswapd to
> > stay awake longer than it should. This patch fixes it.
> > 
> > Signed-off-by: Mel Gorman <mel@xxxxxxxxx>
> 
> Hmm, maybe the logic works well but I don't like very much.
> 
> How about adding below instead of pgdat->node_present_pages ?
> 
> static unsigned long required_balanced_pages(pgdat, classzone_idx)
> {
> 	unsigned long present = 0;
> 
> 	for_each_zone_in_node(zone, pgdat) {
> 		if (zone->all_unreclaimable) /* Ignore unreclaimable zone at checking balance */
> 			continue;
> 		if (zone_idx(zone) > classzone_idx)
> 			continue;
> 		present = zone->present_pages;
> 	}
> 	return present;
> }
> 

I'm afraid I do not really understand. After your earlier comments,
pgdat_balanced() now looks like

static bool pgdat_balanced(pg_data_t *pgdat, unsigned long balanced_pages,
                                                int classzone_idx)
{
        unsigned long present_pages = 0;
        int i;

        for (i = 0; i <= classzone_idx; i++)
                present_pages += pgdat->node_zones[i].present_pages;

        return balanced_pages > (present_pages >> 2);
}

so the classzone is being taken into account. I'm not sure what you're
asking for it to be changed to. Maybe it'll be clearer after V4 comes
out rebased on top of mmotm.


-- 
Mel Gorman
Part-time Phd Student                          Linux Technology Center
University of Limerick                         IBM Dublin Software Lab
--
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/