[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20151207161820.GB20774@dhcp22.suse.cz>
Date: Mon, 7 Dec 2015 17:18:21 +0100
From: Michal Hocko <mhocko@...nel.org>
To: "Huang, Ying" <ying.huang@...ux.intel.com>
Cc: Mel Gorman <mgorman@...hsingularity.net>, lkp@...org,
LKML <linux-kernel@...r.kernel.org>,
Andrew Morton <akpm@...ux-foundation.org>,
Rik van Riel <riel@...hat.com>,
Vitaly Wool <vitalywool@...il.com>,
David Rientjes <rientjes@...gle.com>,
Christoph Lameter <cl@...ux.com>,
Johannes Weiner <hannes@...xchg.org>,
Vlastimil Babka <vbabka@...e.cz>,
Will Deacon <will.deacon@....com>,
Linus Torvalds <torvalds@...ux-foundation.org>
Subject: Re: [lkp] [mm, page_alloc] d0164adc89: -100.0% fsmark.app_overhead
On Fri 04-12-15 09:53:35, Huang, Ying wrote:
> Mel Gorman <mgorman@...hsingularity.net> writes:
[...]
> > What is the result of the __GFP_HIGH patch to give it access to
> > reserves?
>
> Applied Michal's patch on v4.4-rc3 and tested again, now there is no
> page allocation failure.
I still think this is just a coincidence and __GFP_HIGH just
papers over it (it still makes sense for other reasons though). It
would be better imho to log all allocations which do not include
___GFP_KSWAPD_RECLAIM. This might help us to pin point a caller which
was previously waking kswapd up but it doesn't do so anymore. There
would have to be a very specific pattern to trigger this because most
allocations would simply wake kswapd up. I am just wondering that
any debugging would simply alter the timing and/or allocation pattern
and we would see a different results so I am not sure this is worth
spending a lot of time on.
--
Michal Hocko
SUSE Labs
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists