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:   Mon, 25 Nov 2019 12:47:08 +0100
From:   Michal Hocko <mhocko@kernel.org>
To:     David Rientjes <rientjes@google.com>
Cc:     Mel Gorman <mgorman@suse.de>,
        Andrew Morton <akpm@linux-foundation.org>,
        Vlastimil Babka <vbabka@suse.cz>,
        Linus Torvalds <torvalds@linux-foundation.org>,
        Andrea Arcangeli <aarcange@redhat.com>,
        "Kirill A. Shutemov" <kirill@shutemov.name>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        Linux-MM <linux-mm@kvack.org>
Subject: Re: [patch for-5.3 0/4] revert immediate fallback to remote hugepages
Message-ID: <20191125114708.GI31714@dhcp22.suse.cz>
References: <08a3f4dd-c3ce-0009-86c5-9ee51aba8557@suse.cz>
 <20191029151549.GO31513@dhcp22.suse.cz>
 <20191029143351.95f781f09a9fbf254163d728@linux-foundation.org>
 <alpine.DEB.2.21.1910291623050.9914@chino.kir.corp.google.com>
 <20191105130253.GO22672@dhcp22.suse.cz>
 <alpine.DEB.2.21.1911051659010.181254@chino.kir.corp.google.com>
 <20191106073521.GC8314@dhcp22.suse.cz>
 <alpine.DEB.2.21.1911061330030.155572@chino.kir.corp.google.com>
 <20191113112042.GG28938@suse.de>
 <alpine.DEB.2.21.1911241548340.192260@chino.kir.corp.google.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <alpine.DEB.2.21.1911241548340.192260@chino.kir.corp.google.com>
User-Agent: Mutt/1.10.1 (2018-07-13)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Sun 24-11-19 16:10:53, David Rientjes wrote:
[...]
> So my question would be: if we know the previous behavior that allowed 
> excessive swap and recalling into compaction was deemed harmful for the 
> local node, why do we now believe it cannot be harmful if done for all 
> system memory?

I have to say that I got lost in your explanation. I have already
pointed this out in a previous email you didn't reply to. But the main
difference to previous __GFP_THISNODE behavior is that it is used along
with __GFP_NORETRY and that reduces the overall effort of the reclaim
AFAIU. If that is not the case then please be _explicit_ why.

Having test results from Andrea would be really appreciated of course
but he seems to be too busy to do that (or maybe not interested
anymore). I do not see any real reason to hold on this patch based on
hand waving though. So either we have some good reasoning to argue
against the patch or a good testing results or we should go ahead.
As things stand right now, THP success rate went down after your last
changes for _very simple_ workloads. This needs addressing which I hope
we do agree on.
-- 
Michal Hocko
SUSE Labs