| Message ID | 1774242197.1390191.1374190255822.JavaMail.root@uoguelph.ca (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Thu, 2013-07-18 at 19:30 -0400, Rick Macklem wrote: > Andre Heider wrote: > > On Wed, Jul 17, 2013 at 11:59 PM, Trond Myklebust > > <Trond.Myklebust@netapp.com> wrote: > > > Technically, the Linux client is allowed by the NFSv4 spec to send > > > 3 word bitmaps as part of an OPEN request. However, this causes the > > > current FreeBSD server to return NFS4ERR_ATTRNOTSUPP errors. > > > > > > Fix the regression by making the Linux client use a 2 word bitmap > > > unless > > > doing NFSv4.2 with labeled NFS. > > > > > > Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com> > > > > Tested-by: Andre Heider <a.heider@gmail.com> > > > I've attached the patch I plan to commit to FreeBSD's head soon, which > fixes the server so that it checks for the high order bitmaps words > being non-zero before replying with NFS4_ERR_ATTRNOTSUPP. > > The patch is pretty straightforward, but if you can apply it to your > server and test it against the unpatched Linux client, that would be > appreciated. > > It will take a while for the patch to find its way to a FreeBSD release, > so having the workaround in the Linux client will be very helpful. Right. So, my plan is to send patch 1/2 as a bugfix within the next few days, to ensure that we end up with no gap in NFSv4 client functionality between the release versions of Linux 3.10 and 3.11. Since 2/2 is more of a cleanup, and actually changes the functionality to send only a 1-word bitmap for truncate() calls, I'll wait for the 3.12 merge window before committing that. Cheers Trond -- Trond Myklebust Linux NFS client maintainer NetApp Trond.Myklebust@netapp.com www.netapp.com
On Fri, Jul 19, 2013 at 1:30 AM, Rick Macklem <rmacklem@uoguelph.ca> wrote: > Andre Heider wrote: >> On Wed, Jul 17, 2013 at 11:59 PM, Trond Myklebust >> <Trond.Myklebust@netapp.com> wrote: >> > Technically, the Linux client is allowed by the NFSv4 spec to send >> > 3 word bitmaps as part of an OPEN request. However, this causes the >> > current FreeBSD server to return NFS4ERR_ATTRNOTSUPP errors. >> > >> > Fix the regression by making the Linux client use a 2 word bitmap >> > unless >> > doing NFSv4.2 with labeled NFS. >> > >> > Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com> >> >> Tested-by: Andre Heider <a.heider@gmail.com> >> > I've attached the patch I plan to commit to FreeBSD's head soon, which > fixes the server so that it checks for the high order bitmaps words > being non-zero before replying with NFS4_ERR_ATTRNOTSUPP. > > The patch is pretty straightforward, but if you can apply it to your > server and test it against the unpatched Linux client, that would be > appreciated. Confirmed, reverting the client patches and applying this server patch also works ;) Thanks, Andre -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
--- sys/fs/nfs/nfs_commonsubs.c.orig 2013-07-18 19:15:25.000000000 -0400 +++ sys/fs/nfs/nfs_commonsubs.c 2013-07-18 19:15:58.000000000 -0400 @@ -761,21 +761,21 @@ nfsrv_getattrbits(struct nfsrv_descript error = NFSERR_BADXDR; goto nfsmout; } - if (cnt > NFSATTRBIT_MAXWORDS) { + if (cnt > NFSATTRBIT_MAXWORDS) outcnt = NFSATTRBIT_MAXWORDS; - if (retnotsupp) - *retnotsupp = NFSERR_ATTRNOTSUPP; - } else { + else outcnt = cnt; - } NFSZERO_ATTRBIT(attrbitp); if (outcnt > 0) { NFSM_DISSECT(tl, u_int32_t *, outcnt * NFSX_UNSIGNED); for (i = 0; i < outcnt; i++) attrbitp->bits[i] = fxdr_unsigned(u_int32_t, *tl++); } - if (cnt > outcnt) - error = nfsm_advance(nd, (cnt - outcnt) * NFSX_UNSIGNED, -1); + for (i = 0; i < (cnt - outcnt); i++) { + NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); + if (retnotsupp != NULL && *tl != 0) + *retnotsupp = NFSERR_ATTRNOTSUPP; + } if (cntp) *cntp = NFSX_UNSIGNED + (cnt * NFSX_UNSIGNED); nfsmout: