From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mx1.redhat.com ([209.132.183.28]:10859 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932260Ab0IGUa1 (ORCPT ); Tue, 7 Sep 2010 16:30:27 -0400 Message-ID: <4C86A0E0.6050003@RedHat.com> Date: Tue, 07 Sep 2010 16:30:24 -0400 From: Steve Dickson To: Chuck Lever CC: linux-nfs@vger.kernel.org Subject: Re: [PATCH 4/4] mount.nfs: Prepare way for "vers=4,rdma" mounts References: <20100907162156.3392.90376.stgit@seurat.1015granger.net> <20100907162638.3392.14786.stgit@seurat.1015granger.net> <4C869B70.4020405@RedHat.com> <8653371E-0798-4F57-9289-684A8D09B1C7@oracle.com> In-Reply-To: <8653371E-0798-4F57-9289-684A8D09B1C7@oracle.com> Content-Type: text/plain; charset=ISO-8859-1 Sender: linux-nfs-owner@vger.kernel.org List-ID: MIME-Version: 1.0 On 09/07/2010 04:14 PM, Chuck Lever wrote: > > On Sep 7, 2010, at 4:07 PM, Steve Dickson wrote: > >> >> >> On 09/07/2010 12:26 PM, Chuck Lever wrote: >>> At some point, when the kernel starts to support "vers=4,rdma" mounts, >>> we will want the mount.nfs command to pass "vers=4,rdma" mounts >>> instead of rejecting them. >>> >>> Assuming that the kernel will reject these today with EPROTONOSUPPORT, >>> that would cause the version fallback logic to go to "vers=3,rdma" >>> automatically. So the extra check we have now is not needed anyway. >>> >>> Signed-off-by: Chuck Lever >>> --- >>> >>> utils/mount/stropts.c | 7 ------- >>> 1 files changed, 0 insertions(+), 7 deletions(-) >>> >>> diff --git a/utils/mount/stropts.c b/utils/mount/stropts.c >>> index 9695c73..a8b22ce 100644 >>> --- a/utils/mount/stropts.c >>> +++ b/utils/mount/stropts.c >>> @@ -302,13 +302,6 @@ static int nfs_set_version(struct nfsmount_info *mi) >>> >>> if (strncmp(mi->type, "nfs4", 4) == 0) >>> mi->version = 4; >>> - else { >>> - unsigned long protocol; >>> - if (!nfs_nfs_protocol(mi->options, &protocol)) >>> - return 0; >>> - if (protocol == NFSPROTO_RDMA) >>> - mi->version = 3; >>> - } >>> >>> /* >>> * If we still don't know, check for version-specific >>> >> This is a show stopper... We can default to v4 because the callbacks. > > Again, this patch was for comments only, it isn't required. I guess I missed that in your description. > But I don't understand your comment. We can not default to v4 because of the callbacks. > > We don't default to NFSv4 for RDMA today, because of the if clause I remove here. > If no version is specified, it defaults to vers=3 for RDMA mounts. no... this is not the case with the above patched applied... I do have a way to test... > > I suggest we make it default to v4 in this case. What is the callback requirement? Its not clear they work... steved.