Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Wed, 24 Jun 2020 10:50:15 +0200
From:   Michal Hocko <mhocko@kernel.org>
To:     David Hildenbrand <david@redhat.com>
Cc:     Wei Yang <richard.weiyang@linux.alibaba.com>,
        akpm@linux-foundation.org, osalvador@suse.de,
        dan.j.williams@intel.com, linux-mm@kvack.org,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH] mm/spase: never partially remove memmap for early section
Message-ID: <20200624085015.GG1320@dhcp22.suse.cz>
References: <20200623094258.6705-1-richard.weiyang@linux.alibaba.com>
 <20200623151828.GA31426@dhcp22.suse.cz>
 <57d8a558-dcb3-b318-8e38-6c016b050414@redhat.com>
 <20200624081324.GA12461@L-31X9LVDL-1304.local>
 <2b2b722a-58de-4c6e-a04e-dc3516b1b8ab@redhat.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <2b2b722a-58de-4c6e-a04e-dc3516b1b8ab@redhat.com>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Wed 24-06-20 10:41:18, David Hildenbrand wrote:
[...]
> But nothing actually breaks .... because *drummroll* we use huge pages in the vmemmap,
> so the partial depopulate will not actually depopulate anything here. Huge page is 2M,
> the memmap of 128MB sections is exactly 2MB == one hugepages. Trying to depopulate a
> fraction (e.g., 16MB) of that won't do anything.
> 
> 
> Now, forcing a CPU without hugepages - PSE (QEMU: "-cpu host,pse=off)", I can trigger
> via  ndctl create-namespace --force --reconfig=namespace0.0 --mode=devdax --size=16M

OK, now I am following! Thanks a lot. That is something to be mentioned
in the changelog. Small pages might be used if the hotplug happens on a
system with fragmented memory so this is something we do care about.

Thanks a lot David!

-- 
Michal Hocko
SUSE Labs