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:   Thu, 8 Feb 2018 12:21:00 -0800
From:   Matthew Wilcox <willy@infradead.org>
To:     Daniel Micay <danielmicay@gmail.com>
Cc:     Jann Horn <jannh@google.com>, linux-mm@kvack.org,
        Kernel Hardening <kernel-hardening@lists.openwall.com>,
        kernel list <linux-kernel@vger.kernel.org>,
        "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Subject: Re: [RFC] Warn the user when they could overflow mapcount
Message-ID: <20180208202100.GB3424@bombadil.infradead.org>
References: <20180208021112.GB14918@bombadil.infradead.org>
 <CAG48ez2-MTJ2YrS5fPZi19RY6P_6NWuK1U5CcQpJ25=xrGSy_A@mail.gmail.com>
 <CA+DvKQLHDR0s=6r4uiHL8kw2_PnfJcwYfPxgQOmuLbc=5k39+g@mail.gmail.com>
 <20180208185648.GB9524@bombadil.infradead.org>
 <CA+DvKQLHcFc3+kW_SnD6hs53yyD5Zi+uAeSgDMm1tRzxqy-Opg@mail.gmail.com>
 <20180208194235.GA3424@bombadil.infradead.org>
 <CA+DvKQKba0iU+tydbmGkAJsxCxazORDnuoe32sy-2nggyagUxQ@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <CA+DvKQKba0iU+tydbmGkAJsxCxazORDnuoe32sy-2nggyagUxQ@mail.gmail.com>
User-Agent: Mutt/1.9.1 (2017-09-22)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Thu, Feb 08, 2018 at 02:48:52PM -0500, Daniel Micay wrote:
> I guess it could saturate and then switch to tracking the count via an
> object pointer -> count mapping with a global lock? Whatever the
> solution is should probably be a generic one since it's a recurring
> issue.

I was thinking of saturating _mapcount at 2 billion (allowing _refcount
the extra space to go into the 2-3 billion range).  Once saturated,
disallow all attempts at mapping it until _mapcount has gone below 2
billion again.  We can walk the page->mapping->i_mmap tree and find
tasks with more than, say, 10 mappings each, and kill them.

Now that I think about it, though, perhaps the simplest solution is not
to worry about checking whether _mapcount has saturated, and instead when
adding a new mmap, check whether this task already has it mapped 10 times.
If so, refuse the mapping.

Now we can argue that since pid_max is smaller than 400 million that
_mapcount will never overflow, and so we don't need to check it.
Convincing argument?