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:   Mon, 11 Nov 2019 16:40:28 +0100
From:   Adrian Reber <areber@redhat.com>
To:     Oleg Nesterov <oleg@redhat.com>
Cc:     Christian Brauner <christian.brauner@ubuntu.com>,
        Eric Biederman <ebiederm@xmission.com>,
        Pavel Emelyanov <ovzxemul@gmail.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 v7 1/2] fork: extend clone3() to support setting a PID
Message-ID: <20191111154028.GF514519@dcbz.redhat.com>
References: <20191111131704.656169-1-areber@redhat.com>
 <20191111152514.GA11389@redhat.com>
MIME-Version: 1.0
In-Reply-To: <20191111152514.GA11389@redhat.com>
Organization: Red Hat
User-Agent: Mutt/1.12.1 (2019-06-15)
Content-Type: text/plain; charset=WINDOWS-1252
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Mon, Nov 11, 2019 at 04:25:15PM +0100, Oleg Nesterov wrote:
> On 11/11, Adrian Reber wrote:
> >
> > v7:
> >  - changed set_tid to be an array to set the PID of a process
> >    in multiple nested PID namespaces at the same time as discussed
> >    at LPC 2019 (container MC)
>=20
> cough... iirc you convinced me this is not needed when we discussed
> the previous version ;) Nevermind, probably my memory fools me.

You are right. You suggested the same thing and we didn't listen ;)

> So far I only have some cosmetic nits,

Thanks for the quick review. I will try to apply your suggestions.

> > @@ -175,6 +187,18 @@ struct pid *alloc_pid(struct pid_namespace *ns)
> >
> >  =09for (i =3D ns->level; i >=3D 0; i--) {
> >  =09=09int pid_min =3D 1;
> > +=09=09int t_pos =3D 0;
>                     ^^^^^
>=20
> I won't insist, but I'd suggest to cache set_tid[t_pos] instead to make
> the code a bit more simple.
>=20
> > @@ -186,12 +210,24 @@ struct pid *alloc_pid(struct pid_namespace *ns)
> >  =09=09if (idr_get_cursor(&tmp->idr) > RESERVED_PIDS)
> >  =09=09=09pid_min =3D RESERVED_PIDS;
>=20
> You can probably move this code into the "else" branch below.
>=20
> IOW, something like
>=20
>=20
> =09for (i =3D ns->level; i >=3D 0; i--) {
> =09=09int xxx =3D 0;
>=20
> =09=09if (set_tid_size) {
> =09=09=09int pos =3D ns->level - i;
>=20
> =09=09=09xxx =3D set_tid[pos];
> =09=09=09if (xxx < 1 || xxx >=3D pid_max)
> =09=09=09=09return ERR_PTR(-EINVAL);
> =09=09=09/* Also fail if a PID !=3D 1 is requested and no PID 1 exists */
> =09=09=09if (xxx !=3D 1 && !tmp->child_reaper)
> =09=09=09=09return ERR_PTR(-EINVAL);
> =09=09=09if (!ns_capable(tmp->user_ns, CAP_SYS_ADMIN))
> =09=09=09=09return ERR_PTR(-EPERM);
> =09=09=09set_tid_size--;
> =09=09}
>=20
> =09=09idr_preload(GFP_KERNEL);
> =09=09spin_lock_irq(&pidmap_lock);
>=20
> =09=09if (xxx) {
> =09=09=09nr =3D idr_alloc(&tmp->idr, NULL, xxx, xxx + 1,
> =09=09=09=09=09GFP_ATOMIC);
> =09=09=09/*
> =09=09=09 * If ENOSPC is returned it means that the PID is
> =09=09=09 * alreay in use. Return EEXIST in that case.
> =09=09=09 */
> =09=09=09if (nr =3D=3D -ENOSPC)
> =09=09=09=09nr =3D -EEXIST;
> =09=09} else {
> =09=09=09int pid_min =3D 1;
> =09=09=09/*
> =09=09=09 * init really needs pid 1, but after reaching the
> =09=09=09 * maximum wrap back to RESERVED_PIDS
> =09=09=09 */
> =09=09=09if (idr_get_cursor(&tmp->idr) > RESERVED_PIDS)
> =09=09=09=09pid_min =3D RESERVED_PIDS;
> =09=09=09/*
> =09=09=09 * Store a null pointer so find_pid_ns does not find
> =09=09=09 * a partially initialized PID (see below).
> =09=09=09 */
> =09=09=09nr =3D idr_alloc_cyclic(&tmp->idr, NULL, pid_min,
> =09=09=09=09=09      pid_max, GFP_ATOMIC);
> =09=09}
>=20
> =09=09...
>=20
> This way only the "if (set_tid_size)" block has to play with set_tid_size=
/set_tid.
>=20
> note also that this way we can easily allow set_tid[some_level] =3D=3D 0,=
 we can
> simply do
>=20
> =09-=09if (xxx < 1 || xxx >=3D pid_max)
> =09+=09if (xxx < 0 || xxx >=3D pid_max)
>=20
> although I don't think this is really useful.

Yes. I explicitly didn't allow 0 as a PID as I didn't thought it would
be useful (or maybe even valid).

=09=09Adrian