linux-kernel - Re: [PATCH RFC] mm/memory_hotplug: make it possible to offline blocks with reserved pages

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20171109134241.4dz6wchxzx27dgsr@dhcp22.suse.cz>
Date:   Thu, 9 Nov 2017 14:42:41 +0100
From:   Michal Hocko <mhocko@...nel.org>
To:     Vitaly Kuznetsov <vkuznets@...hat.com>
Cc:     linux-mm@...ck.org, linux-kernel@...r.kernel.org,
        Andrew Morton <akpm@...ux-foundation.org>,
        Vlastimil Babka <vbabka@...e.cz>,
        Mel Gorman <mgorman@...hsingularity.net>,
        YASUAKI ISHIMATSU <yasu.isimatu@...il.com>,
        Hillf Danton <hillf.zj@...baba-inc.com>,
        Johannes Weiner <hannes@...xchg.org>,
        "K. Y. Srinivasan" <kys@...rosoft.com>,
        Stephen Hemminger <sthemmin@...rosoft.com>,
        Alex Ng <alexng@...rosoft.com>
Subject: Re: [PATCH RFC] mm/memory_hotplug: make it possible to offline
 blocks with reserved pages

On Thu 09-11-17 14:30:18, Vitaly Kuznetsov wrote:
> Michal Hocko <mhocko@...nel.org> writes:
[...]
> > How realistic is that the host gives only such a small amount of memory
> > btw?
> 
> It happens all the time, Hyper-V host will gradually increase guest's
> memory when Dynamic Memory is enabled. Moreover, there's a manual
> interface when host's admin can hotplug memory to guests (starting
> WS2016) with 2M granularity.

Sigh, this sounds quite insane but whatever. I am not sure we want to
make the generic hotplug code more complicated for this single usecase.
So I suspect you might be better off by implementing this feature on top
of hotplug. Just keep track of the partial sections and make the memory
which is not onlined yet reserved and unusable by the kernel. It sucks,
I know, but as long as there is not a wider demand for sub section
memory hotplug I would be rather reluctant to make the fragile code even
more complicated. Mem section granularity is hardcoded in way too many
places I am afraid.
-- 
Michal Hocko
SUSE Labs