linux-kernel - Re: [PATCHv2 5/8] khugepaged: Allow to callapse a page shared across fork

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: <20200413094204.a2gpsjhugy5dznjy@box>
Date:   Mon, 13 Apr 2020 12:42:04 +0300
From:   "Kirill A. Shutemov" <kirill@...temov.name>
To:     John Hubbard <jhubbard@...dia.com>
Cc:     akpm@...ux-foundation.org, Andrea Arcangeli <aarcange@...hat.com>,
        Zi Yan <ziy@...dia.com>, Yang Shi <yang.shi@...ux.alibaba.com>,
        linux-mm@...ck.org, linux-kernel@...r.kernel.org,
        "Kirill A. Shutemov" <kirill.shutemov@...ux.intel.com>
Subject: Re: [PATCHv2 5/8] khugepaged: Allow to callapse a page shared across
 fork

On Fri, Apr 10, 2020 at 01:59:22PM -0700, John Hubbard wrote:
> I think I understood what you were saying. The problem is that was ignoring
> a couple of points, especially in an RDMA situation: 1) the page can be
> pinned by various drivers, on behalf of other processes, even if the original
> process is being torn down, and 2) it doesn't really matter which process pins
> a page--the end result is that it's pinned.

Well, no. It is critical that nobody gets new pins after this point on
behalf of *this* process as we about change what is mapped on this virtual
address range. We must avoid the situation that khugepaged screws
legitimate GUP users and make what process see differs from what GUP see.

Pins on behalf of other processes after the point are not relevant to us.

I will keep the comment as is for now. As you can see I'm struggling
communicating my point. Any better wording is welcome.

-- 
 Kirill A. Shutemov