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;
From:   ebiederm@xmission.com (Eric W. Biederman)
To:     Alexey Dobriyan <adobriyan@gmail.com>
Cc:     alban.crequy@gmail.com, linux-kernel@vger.kernel.org
References: <20180404174909.GB2540@avx2>
Date:   Wed, 04 Apr 2018 18:59:08 -0500
In-Reply-To: <20180404174909.GB2540@avx2> (Alexey Dobriyan's message of "Wed,
        4 Apr 2018 20:49:09 +0300")
Message-ID: <87k1tmed6b.fsf@xmission.com>
User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/25.1 (gnu/linux)
MIME-Version: 1.0
Content-Type: text/plain
Subject: Re: [PATCH] [RFC][WIP] namespace.c: Allow some unprivileged proc mounts when not fully visible
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

Alexey Dobriyan <adobriyan@gmail.com> writes:

>> The only option I have seen proposed that might qualify as something
>> general purpose and simple is a new filesystem that is just the process
>> directories of proc.
>
> While "mount -t pid" and "mount -t sysctl" are decades overdue, I don't
> think they cover everything.
>
> IIRC some gcc versions read /proc/meminfo on every invocation. Now
> imagine such program doesn't have a fallback if /proc/ doesn't exist
> (how many thousands such programs are there?) So user is going to ask
> for /proc with just /proc/meminfo only. At this point it is back to
> nearly full /proc.

To avoid falling susceptible to the kinds of checks in fs_fully_visible we
can only offer information about objects that root in the user namespace
has privilege over.  So "mount -t pid" good.  A "/proc/meminfo" bad.

Which in short means if "mount -t pid" isn't good enough.  There really
isn't anything the kernel can do.

Eric