Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Wed, 4 Mar 2020 12:26:53 +0100
From:   Michal Hocko <mhocko@kernel.org>
To:     "Huang, Ying" <ying.huang@intel.com>
Cc:     Mel Gorman <mgorman@suse.de>, David Hildenbrand <david@redhat.com>,
        Johannes Weiner <hannes@cmpxchg.org>,
        Matthew Wilcox <willy@infradead.org>,
        Andrew Morton <akpm@linux-foundation.org>, linux-mm@kvack.org,
        linux-kernel@vger.kernel.org, Vlastimil Babka <vbabka@suse.cz>,
        Zi Yan <ziy@nvidia.com>, Peter Zijlstra <peterz@infradead.org>,
        Dave Hansen <dave.hansen@linux.intel.com>,
        Minchan Kim <minchan@kernel.org>,
        Hugh Dickins <hughd@google.com>,
        Alexander Duyck <alexander.duyck@gmail.com>
Subject: Re: [RFC 0/3] mm: Discard lazily freed pages when migrating
Message-ID: <20200304112653.GH16139@dhcp22.suse.cz>
References: <20200228094954.GB3772@suse.de>
 <87h7z76lwf.fsf@yhuang-dev.intel.com>
 <20200302151607.GC3772@suse.de>
 <87zhcy5hoj.fsf@yhuang-dev.intel.com>
 <20200303080945.GX4380@dhcp22.suse.cz>
 <87o8td4yf9.fsf@yhuang-dev.intel.com>
 <20200303085805.GB4380@dhcp22.suse.cz>
 <87ftep4pzy.fsf@yhuang-dev.intel.com>
 <20200304095802.GE16139@dhcp22.suse.cz>
 <87blpc2wxj.fsf@yhuang-dev.intel.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <87blpc2wxj.fsf@yhuang-dev.intel.com>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Wed 04-03-20 19:15:20, Huang, Ying wrote:
> Michal Hocko <mhocko@kernel.org> writes:
> 
> > On Tue 03-03-20 19:49:53, Huang, Ying wrote:
[...]
> >> Because the penalty difference is so large, I
> >> think it may be a good idea to always put clean MADV_FREE pages at the
> >> tail of the inactive file LRU list?
> >
> > You are again making assumptions without giving any actual real
> > examples. Reconstructing MADV_FREE pages cost can differ a lot.
> 
> In which situation the cost to reconstruct MADV_FREE pages can be higher
> than the cost to allocate file cache page and read from disk?  Heavy
> contention on mmap_sem?

Allocating a page might be really costly but even without that.
Reconstructnig the content of the page can be really high and actually
much larger than an IO to get a page from the storage. Consider
decompression or some other transformations to get the content.
Userspace has means to detect that the content is still up-to-date as
already has been mentioned so all those steps can be avoided.

-- 
Michal Hocko
SUSE Labs