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:   Fri, 22 Nov 2019 08:00:42 -0500
From:   Vivek Goyal <vgoyal@redhat.com>
To:     Stefan Hajnoczi <stefanha@redhat.com>
Cc:     virtio-fs@redhat.com, linux-fsdevel@vger.kernel.org,
        linux-kernel@vger.kernel.org, dgilbert@redhat.com,
        miklos@szeredi.hu
Subject: Re: [PATCH 4/4] virtiofs: Support blocking posix locks
 (fcntl(F_SETLKW))
Message-ID: <20191122130042.GB8636@redhat.com>
References: <20191115205705.2046-1-vgoyal@redhat.com>
 <20191115205705.2046-5-vgoyal@redhat.com>
 <20191121170020.GE445244@stefanha-x1.localdomain>
MIME-Version: 1.0
In-Reply-To: <20191121170020.GE445244@stefanha-x1.localdomain>
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 Thu, Nov 21, 2019 at 05:00:20PM +0000, Stefan Hajnoczi wrote:

[..]
> > +static int virtio_fs_handle_notify(struct virtio_fs *vfs,
> > +=09=09=09=09   struct virtio_fs_notify *notify)
> > +{
> > +=09int ret =3D 0;
> > +=09struct fuse_out_header *oh =3D &notify->out_hdr;
> > +=09struct fuse_notify_lock_out *lo;
> > +
> > +=09/*
> > +=09 * For notifications, oh.unique is 0 and oh->error contains code
> > +=09 * for which notification as arrived.
> > +=09 */
> > +=09switch(oh->error) {
> > +=09case FUSE_NOTIFY_LOCK:
> > +=09=09lo =3D (struct fuse_notify_lock_out *) &notify->outarg;
> > +=09=09notify_complete_waiting_req(vfs, lo);
> > +=09=09break;
> > +=09default:
> > +=09=09printk("virtio-fs: Unexpected notification %d\n", oh->error);
> > +=09}
> > +=09return ret;
> > +}
>=20
> Is this specific to virtio or can be it handled in common code?

This is not specific to virtio_fs. In principle, regular fuse daemon could
implement something similar. Though they might not have to because client
can just block without introducing deadlock possibilities.

Anyway, I will look into moving this code into fuse common.

[..]
> > diff --git a/include/uapi/linux/fuse.h b/include/uapi/linux/fuse.h
> > index 373cada89815..45f0c4efec8e 100644
> > --- a/include/uapi/linux/fuse.h
> > +++ b/include/uapi/linux/fuse.h
> > @@ -481,6 +481,7 @@ enum fuse_notify_code {
> >  =09FUSE_NOTIFY_STORE =3D 4,
> >  =09FUSE_NOTIFY_RETRIEVE =3D 5,
> >  =09FUSE_NOTIFY_DELETE =3D 6,
> > +=09FUSE_NOTIFY_LOCK =3D 7,
> >  =09FUSE_NOTIFY_CODE_MAX,
> >  };
> > =20
> > @@ -868,6 +869,12 @@ struct fuse_notify_retrieve_in {
> >  =09uint64_t=09dummy4;
> >  };
> > =20
> > +struct fuse_notify_lock_out {
> > +=09uint64_t=09id;
>=20
> Please call this field "unique" or "lock_unique" so it's clear this
> identifier is the fuse_header_in->unique value of the lock request.

Ok, will do.

Vivek