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:   Christian Brauner <christian@brauner.io>
Date:   Fri, 2 Aug 2019 15:02:35 +0200
To:     Oleg Nesterov <oleg@redhat.com>
Cc:     Adrian Reber <areber@redhat.com>,
        Christian Brauner <christian@brauner.io>,
        Eric Biederman <ebiederm@xmission.com>,
        Pavel Emelianov <xemul@virtuozzo.com>,
        Jann Horn <jannh@google.com>,
        Dmitry Safonov <0x7f454c46@gmail.com>,
        linux-kernel@vger.kernel.org, Andrei Vagin <avagin@gmail.com>,
        Mike Rapoport <rppt@linux.ibm.com>,
        Radostin Stoyanov <rstoyanov1@gmail.com>
Subject: Re: [PATCH v2 1/2] fork: extend clone3() to support CLONE_SET_TID
Message-ID: <20190802130214.l6vporvni7nnbnif@brauner.io>
References: <20190731161223.2928-1-areber@redhat.com>
 <20190731174135.GA30225@redhat.com>
 <20190802072511.GD18263@dcbz.redhat.com>
 <20190802124738.GC20111@redhat.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <20190802124738.GC20111@redhat.com>
User-Agent: NeoMutt/20180716
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Fri, Aug 02, 2019 at 02:47:38PM +0200, Oleg Nesterov wrote:
> On 08/02, Adrian Reber wrote:
> >
> > On Wed, Jul 31, 2019 at 07:41:36PM +0200, Oleg Nesterov wrote:
> > > But the main question is how it can really help if ns->level > 0, unlikely
> > > CRIU will ever need to clone the process with the same pid_nr == set_tid
> > > in the ns->parent chain.
> >
> > Not sure I understand what you mean. For CRIU only the PID in the PID
> > namespace is relevant.
> 
> If it runs "inside" this namespace. But in this case alloc_pid() should
> use nr == set_tid only once in the main loop, when i == ns->level.
> 
> It doesn't need to have the same pid_nr in the parent pid namespace.
> 
> And in fact we should not allow criu (or anything else) to control the child's
> pid_nr in the parent(s) namespace.

This should definitely not be possible!

> 
> Right?
> 
> > > So may be kernel_clone_args->set_tid should be pid_t __user *set_tid_array?
> > > Or I missed something ?
> >
> > Not sure why and how an array would be needed. Could you give me some
> > more details why you think this is needed.
> 
> IIURC, criu can restore the process tree along with nested pid namespaces.

Hm, I'm not a fan of this array approach...

> 
> how can this patch help in this case?
> 
> But again, perhaps I missed something. I forgot everything about criu.
> 
> Oleg.
>