Received-SPF: pass (google.com: domain of linux-nfs-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <969a2ddc66df3ba05952fb14352ccee08bd84149.camel@kernel.org>
Subject: Re: [PATCH RFC] nfsd: set missing after_change as before_change + 1
From:   Jeff Layton <jlayton@kernel.org>
To:     Chuck Lever <chuck.lever@oracle.com>
Cc:     Neil Brown <neilb@suse.de>, Olga Kornievskaia <kolga@netapp.com>,
        Dai Ngo <Dai.Ngo@oracle.com>, Tom Talpey <tom@talpey.com>,
        linux-nfs@vger.kernel.org, linux-kernel@vger.kernel.org
Date:   Mon, 24 Jul 2023 12:21:47 -0400
In-Reply-To: <ZL6W0GqBSdlvVL2Y@tissot.1015granger.net>
References: <20230724-bz2223560-v1-1-b6da868c0fc6@kernel.org>
         <ZL6W0GqBSdlvVL2Y@tissot.1015granger.net>
Content-Type: text/plain; charset="ISO-8859-15"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.48.4 (3.48.4-1.fc38) 
MIME-Version: 1.0
Precedence: bulk

On Mon, 2023-07-24 at 11:20 -0400, Chuck Lever wrote:
> On Mon, Jul 24, 2023 at 10:53:39AM -0400, Jeff Layton wrote:
> > In the event that we can't fetch post_op_attr attributes, we still need
> > to set a value for the after_change. The operation has already happened=
,
> > so we're not able to return an error at that point, but we do want to
> > ensure that the client knows that its cache should be invalidated.
> >=20
> > If we weren't able to fetch post-op attrs, then just set the
> > after_change to before_change + 1. The atomic flag should already be
> > clear in this case.
> >=20
> > Suggested-by: Neil Brown <neilb@suse.de>
> > Signed-off-by: Jeff Layton <jlayton@kernel.org>
> > ---
> >  fs/nfsd/nfs4proc.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
>=20
> I'm not sure this change makes any difference. The client would
> possibly see the change value move forward then back. I'd think a
> false "atomic" field and using the /same/ pre- and post-change would
> be safer...?
>=20
> But I'm intrigued enough to apply this to nfsd-next provisionally,
> at least for testing and further review. It will appear a little
> later today.
>=20
>=20

Thanks. I think there really are no great choices here.

This is a rather unlikely error case that should only come into play
when there are problems with the underlying filesystem, but only when
fetching the post-op attrs.

We don't have a way to just opt out of providing a post-op attribute and
I think this is probably the least bad option of what to put in there.

> > diff --git a/fs/nfsd/nfs4proc.c b/fs/nfsd/nfs4proc.c
> > index 3f6710c9c5c9..f0f318e78630 100644
> > --- a/fs/nfsd/nfs4proc.c
> > +++ b/fs/nfsd/nfs4proc.c
> > @@ -411,7 +411,7 @@ set_change_info(struct nfsd4_change_info *cinfo, st=
ruct svc_fh *fhp)
> >  	if (WARN_ON_ONCE(!fhp->fh_pre_saved))
> >  		cinfo->before_change =3D 0;
> >  	if (!fhp->fh_post_saved)
> > -		cinfo->after_change =3D 0;
> > +		cinfo->after_change =3D cinfo->before_change + 1;
> >  }
> > =20
> >  static __be32
> >=20
> > ---
> > base-commit: 97a5d0146ef443df148805a4e9c3c44111f14ab1
> > change-id: 20230724-bz2223560-5ed6bc3a5db7
> >=20
> > Best regards,
> > --=20
> > Jeff Layton <jlayton@kernel.org>
> >=20
>=20

--=20
Jeff Layton <jlayton@kernel.org>