On Thu, 2013-04-04 at 16:34 -0700, Chuck Lever wrote: > > On Apr 4, 2013, at 1:55 PM, "Myklebust, Trond" <Trond.Myklebust@xxxxxxxxxx> wrote: > > > On Thu, 2013-04-04 at 13:29 -0700, Chuck Lever wrote: > >> On Apr 4, 2013, at 1:17 PM, Trond Myklebust <Trond.Myklebust@xxxxxxxxxx> wrote: > >> > >>> If the rpcsec_gss_krb5 module cannot be loaded, the attempt to create > >>> an rpc_client in nfs4_init_client will currently fail with an EINVAL. > >>> Fix is to retry with AUTH_NULL. > >>> > >>> Regression introduced by the commit "NFS: Use "krb5i" to establish NFSv4 > >>> state whenever possible" > >>> > >>> Signed-off-by: Trond Myklebust <Trond.Myklebust@xxxxxxxxxx> > >>> Cc: Chuck Lever <chuck.lever@xxxxxxxxxx> > >>> Cc: Bryan Schumaker <bjschuma@xxxxxxxxxx> > >>> --- > >>> fs/nfs/nfs4client.c | 2 ++ > >>> 1 file changed, 2 insertions(+) > >>> > >>> diff --git a/fs/nfs/nfs4client.c b/fs/nfs/nfs4client.c > >>> index bb9789e..a4f2100 100644 > >>> --- a/fs/nfs/nfs4client.c > >>> +++ b/fs/nfs/nfs4client.c > >>> @@ -200,6 +200,8 @@ struct nfs_client *nfs4_init_client(struct nfs_client *clp, > >>> > >>> __set_bit(NFS_CS_DISCRTRY, &clp->cl_flags); > >>> error = nfs_create_rpc_client(clp, timeparms, RPC_AUTH_GSS_KRB5I); > >> > >> The parameters passed to nfs_create_rpc_client() are all valid here, so -EINVAL is not a good error value in this situation. > >> > >> rpcauth_create(), I think, should return -EACCES in this case, rather than -EINVAL. > > > > I disagree. This isn't an issue of something being denied permission to > > access a resource: it is that the flavour does not correspond to > > anything in the kernel's list of valid auth flavours. > > > > Either way, that is a separate issue that shouldn't affect this patch. > > Fair enough. -ENOENT might be a stronger choice in this case. But yes, changing the API contract of rpcauth_auth() can be done later in an independent change. > > However -EACCES is returned in some cases from the flavor's create method. Instead of checking for -EINVAL in particular, how about retrying nfs_create_rpc_client() with AUTH_NULL if error < 0 ? No. Why bother if the error is ECONNREFUSED. ENETUNREACH, or something similar? The whole point of returning different errors is so that we can distinguish between those cases that are worth retrying, and those that are not. -- Trond Myklebust Linux NFS client maintainer NetApp Trond.Myklebust@xxxxxxxxxx www.netapp.com -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
