Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Thu, 8 Sep 2022 11:42:18 +0200
From:   Vincent Whitchurch <vincent.whitchurch@axis.com>
To:     Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
CC:     Alexey Dobriyan <adobriyan@gmail.com>,
        Andrew Morton <akpm@linux-foundation.org>,
        Vlastimil Babka <vbabka@suse.cz>,
        Kees Cook <keescook@chromium.org>,
        Matthew Wilcox <willy@infradead.org>,
        <linux-kernel@vger.kernel.org>, <linux-fsdevel@vger.kernel.org>
Subject: Re: [RFC] process /proc/PID/smaps vs /proc/PID/smaps_rollup
Message-ID: <Yxm4+rGMLtHLUmMU@axis.com>
References: <20200929020520.GC871730@jagdpanzerIV.localdomain>
 <20200929024018.GA529@jagdpanzerIV.localdomain>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Disposition: inline
In-Reply-To: <20200929024018.GA529@jagdpanzerIV.localdomain>
Precedence: bulk

On Tue, Sep 29, 2020 at 11:40:18AM +0900, Sergey Senozhatsky wrote:
> On (20/09/29 11:05), Sergey Senozhatsky wrote:
> > One of our unprivileged daemon process needs process PSS info. That
> > info is usually available in /proc/PID/smaps on per-vma basis, on
> > in /proc/PID/smaps_rollup as a bunch of accumulated per-vma values.
> > The latter one is much faster and simpler to get, but, unlike smaps,
> > smaps_rollup requires PTRACE_MODE_READ, which we don't want to
> > grant to our unprivileged daemon.
> > 
> > So the question is - can we get, somehow, accumulated PSS info from
> > a non-privileged process? (Iterating through all process' smaps
> > vma-s consumes quite a bit of CPU time). This is related to another
> > question - why do smaps and smaps_rollup have different permission
> > requirements?
> 
> Hold on, seems that I misread something, /proc/PID/smaps is also
> unavailable. So the question is, then, how do we get PSS info of
> a random user-space process from an unprivileged daemon?

smaps contains a lot of sensitive information, but perhaps smaps_rollup
could be allowed without ptrace rights if the range information is
masked.  I've posted a patch here:

 https://lore.kernel.org/linux-mm/20220908093919.843346-1-vincent.whitchurch@axis.com/