Date: Thu, 2 Jul 2015 17:41:15 -0400 (EDT) From: Rick Macklem <rmacklem@uoguelph.ca> To: Ahmed Kamal <email.ahmedkamal@googlemail.com> Cc: Julian Elischer <julian@freebsd.org>, Freebsd fs <freebsd-fs@freebsd.org>, d@delphij.net Subject: Re: [nfsv4] Is "skip by 1" allowed for the NFSv4.0 seqid? Message-ID: <1919954909.3441620.1435873275001.JavaMail.zimbra@uoguelph.ca> In-Reply-To: <CADaq8jcB-1XMGz0ud_9rRphVrK5Mq0G_wqxvaAVVZp1Obbt4sQ@mail.gmail.com> References: <1427974645.2786896.1435800121257.JavaMail.zimbra@uoguelph.ca> <CADaq8jcB-1XMGz0ud_9rRphVrK5Mq0G_wqxvaAVVZp1Obbt4sQ@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Here is the response on nfsv4@ietf.org. I think the Redhat client is broken from reading this. It will be interesting to see if anyone working on the Linux client responds. rick ----- Original Message ----- > > Is "skip by 1" allowed for NFv4.0 seqid? >=20 > No it isn't. As you indicated RFC7530, which obsoletes RFC3530,, indicate= s > you are' > only allowed to increment by one, with the exception of the wraparound ca= se. >=20 > > I'll admit the wording in RFC-3530 isn't ideal, >=20 > I think the text "(r =3D=3D L+2)" is intended to be read as "(e.g. r =3D= =3D L+2)" but > the existing text > is unclear. >=20 > > but I've never heard of this interpretation before . >=20 > Me either. >=20 > > Is an NFSv4.0 server supposed to accept both 0 and 1 as the > correct next sequence # after UINT32_MAX? >=20 > I would say it isn't.=C2=A0 RFC7530 says "If th e sequence value received= is any > other > value, it is rejected with the return of error NFS4ERR_BAD_SEQID" In thi= s > case > the correct value is 1. On the other hand, this isn't a "MUST". If ther= e > are clients out > out there that don't do seqid wraparound correctly, it is seems like a > reasonable > accommodation for a server to accept the incorrect value zero in this cas= e., >=20 >=20 >=20 >=20 >=20 >=20 > On Wed, Jul 1, 2015 at 9:22 PM, Rick Macklem < rmacklem@uoguelph.ca > wro= te: >=20 >=20 > Hi, >=20 > If you look here (on page #6) it seems to indicate that > incrementing the seqid by 2 is allowed by the RFC. > (I'll admit the wording in RFC-3530 isn't ideal, but I've > never heard of this interpretation before.) > https://dl.dropboxusercontent.com/u/51939288/nfs4-bad-seq.pdf >=20 > RFC-7530 seems clear that it should only be incremented by 1. > However, I do notice that a wraparound of the seqid is supposed > to skip 0 and go to 1. I don't see any mention of this in RFC-3530. > --> Is an NFSv4.0 server supposed to accept both 0 and 1 as the > correct next sequence # after UINT32_MAX? >=20 > Thanks for any help clarifying this, rick >=20 > _______________________________________________ > nfsv4 mailing list > nfsv4@ietf.org > https://www.ietf.org/mailman/listinfo/nfsv4 >=20 >=20 > _______________________________________________ > nfsv4 mailing list > nfsv4@ietf.org > https://www.ietf.org/mailman/listinfo/nfsv4 >=20
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1919954909.3441620.1435873275001.JavaMail.zimbra>