Implement support for RPCSEC_GSS authentication to both the NFS client

and server. This replaces the RPC implementation of the NFS client and
server with the newer RPC implementation originally developed
(actually ported from the userland sunrpc code) to support the NFS
Lock Manager.  I have tested this code extensively and I believe it is
stable and that performance is at least equal to the legacy RPC
implementation.

The NFS code currently contains support for both the new RPC
implementation and the older legacy implementation inherited from the
original NFS codebase. The default is to use the new implementation -
add the NFS_LEGACYRPC option to fall back to the old code. When I
merge this support back to RELENG_7, I will probably change this so
that users have to 'opt in' to get the new code.

To use RPCSEC_GSS on either client or server, you must build a kernel
which includes the KGSSAPI option and the crypto device. On the
userland side, you must build at least a new libc, mountd, mount_nfs
and gssd. You must install new versions of /etc/rc.d/gssd and
/etc/rc.d/nfsd and add 'gssd_enable=YES' to /etc/rc.conf.

As long as gssd is running, you should be able to mount an NFS
filesystem from a server that requires RPCSEC_GSS authentication. The
mount itself can happen without any kerberos credentials but all
access to the filesystem will be denied unless the accessing user has
a valid ticket file in the standard place (/tmp/krb5cc_<uid>). There
is currently no support for situations where the ticket file is in a
different place, such as when the user logged in via SSH and has
delegated credentials from that login. This restriction is also
present in Solaris and Linux. In theory, we could improve this in
future, possibly using Brooks Davis' implementation of variant
symlinks.

Supporting RPCSEC_GSS on a server is nearly as simple. You must create
service creds for the server in the form 'nfs/<fqdn>@<REALM>' and
install them in /etc/krb5.keytab. The standard heimdal utility ktutil
makes this fairly easy. After the service creds have been created, you
can add a '-sec=krb5' option to /etc/exports and restart both mountd
and nfsd.

The only other difference an administrator should notice is that nfsd
doesn't fork to create service threads any more. In normal operation,
there will be two nfsd processes, one in userland waiting for TCP
connections and one in the kernel handling requests. The latter
process will create as many kthreads as required - these should be
visible via 'top -H'. The code has some support for varying the number
of service threads according to load but initially at least, nfsd uses
a fixed number of threads according to the value supplied to its '-n'
option.

Sponsored by:	Isilon Systems
MFC after:	1 month

1 files changed, 39 insertions, 28 deletions

diff --git a/sys/nfsserver/nfs_serv.c b/sys/nfsserver/nfs_serv.c
index a0a0308..d528769 100644
--- a/sys/nfsserver/nfs_serv.c
+++ b/sys/nfsserver/nfs_serv.c
@@ -142,8 +142,10 @@ SYSCTL_STRUCT(_vfs_nfsrv, NFS_NFSRVSTATS, nfsrvstats, CTLFLAG_RW,
 
 static int	nfsrv_access(struct vnode *, accmode_t, struct ucred *,
 		    int, int);
+#ifdef NFS_LEGACYRPC
 static void	nfsrvw_coalesce(struct nfsrv_descript *,
 		    struct nfsrv_descript *);
+#endif
 
 /*
  * Clear nameidata fields that are tested in nsfmout cleanup code prior
@@ -216,7 +218,7 @@ nfsrv3_access(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
 	fhp = &nfh.fh_generic;
 	nfsm_srvmtofh(fhp);
 	tl = nfsm_dissect_nonblock(u_int32_t *, NFSX_UNSIGNED);
-	error = nfsrv_fhtovp(fhp, 1, &vp, &vfslocked, cred, slp,
+	error = nfsrv_fhtovp(fhp, 1, &vp, &vfslocked, nfsd, slp,
 	    nam, &rdonly, TRUE);
 	if (error) {
 		nfsm_reply(NFSX_UNSIGNED);
@@ -283,7 +285,7 @@ nfsrv_getattr(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
 	vfslocked = 0;
 	fhp = &nfh.fh_generic;
 	nfsm_srvmtofh(fhp);
-	error = nfsrv_fhtovp(fhp, 1, &vp, &vfslocked, cred, slp, nam,
+	error = nfsrv_fhtovp(fhp, 1, &vp, &vfslocked, nfsd, slp, nam,
 	    &rdonly, TRUE);
 	if (error) {
 		nfsm_reply(0);
@@ -392,7 +394,7 @@ nfsrv_setattr(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
 	/*
 	 * Now that we have all the fields, lets do it.
 	 */
-	error = nfsrv_fhtovp(fhp, 1, &vp, &tvfslocked, cred, slp,
+	error = nfsrv_fhtovp(fhp, 1, &vp, &tvfslocked, nfsd, slp,
 	    nam, &rdonly, TRUE);
 	vfslocked = nfsrv_lockedpair(vfslocked, tvfslocked);
 	if (error) {
@@ -505,7 +507,7 @@ nfsrv_lookup(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
 	nd.ni_cnd.cn_cred = cred;
 	nd.ni_cnd.cn_nameiop = LOOKUP;
 	nd.ni_cnd.cn_flags = LOCKLEAF | SAVESTART | MPSAFE;
-	error = nfs_namei(&nd, fhp, len, slp, nam, &md, &dpos,
+	error = nfs_namei(&nd, nfsd, fhp, len, slp, nam, &md, &dpos,
 		&dirp, v3, &dirattr, &dirattr_ret, pubflag);
 	vfslocked = NDHASGIANT(&nd);
 
@@ -715,7 +717,7 @@ nfsrv_readlink(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
 	uiop->uio_rw = UIO_READ;
 	uiop->uio_segflg = UIO_SYSSPACE;
 	uiop->uio_td = NULL;
-	error = nfsrv_fhtovp(fhp, 1, &vp, &vfslocked, cred, slp,
+	error = nfsrv_fhtovp(fhp, 1, &vp, &vfslocked, nfsd, slp,
 	    nam, &rdonly, TRUE);
 	if (error) {
 		nfsm_reply(2 * NFSX_UNSIGNED);
@@ -811,7 +813,7 @@ nfsrv_read(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
 	 * as well.
 	 */
 
-	error = nfsrv_fhtovp(fhp, 1, &vp, &vfslocked, cred, slp,
+	error = nfsrv_fhtovp(fhp, 1, &vp, &vfslocked, nfsd, slp,
 	    nam, &rdonly, TRUE);
 	if (error) {
 		vp = NULL;
@@ -1112,7 +1114,7 @@ nfsrv_write(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
 		error = 0;
 		goto nfsmout;
 	}
-	error = nfsrv_fhtovp(fhp, 1, &vp, &tvfslocked, cred, slp,
+	error = nfsrv_fhtovp(fhp, 1, &vp, &tvfslocked, nfsd, slp,
 	    nam, &rdonly, TRUE);
 	vfslocked = nfsrv_lockedpair(vfslocked, tvfslocked);
 	if (error) {
@@ -1227,6 +1229,16 @@ nfsmout:
 	return(error);
 }
 
+#ifdef NFS_LEGACYRPC
+
+/*
+ * XXX dfr - write gathering isn't supported by the new RPC code since
+ * its really only useful for NFSv2. If there is a real need, we could
+ * attempt to fit it into the filehandle affinity system, e.g. by
+ * looking to see if there are queued write requests that overlap this
+ * one.
+ */
+
 /*
  * For the purposes of write gathering, we must decide if the credential
  * associated with two pending requests have equivilent privileges.  Since
@@ -1432,7 +1444,7 @@ loop1:
 		cred = nfsd->nd_cr;
 		v3 = (nfsd->nd_flag & ND_NFSV3);
 		forat_ret = aftat_ret = 1;
-		error = nfsrv_fhtovp(&nfsd->nd_fh, 1, &vp, &vfslocked, cred,
+		error = nfsrv_fhtovp(&nfsd->nd_fh, 1, &vp, &vfslocked, nfsd,
 		    slp, nfsd->nd_nam, &rdonly, TRUE);
 		if (!error) {
 		    if (v3)
@@ -1634,6 +1646,8 @@ nfsrvw_coalesce(struct nfsrv_descript *owp, struct nfsrv_descript *nfsd)
 	}
 }
 
+#endif
+
 /*
  * nfs create service
  * now does a truncate to 0 length via. setattr if it already exists
@@ -1697,7 +1711,7 @@ nfsrv_create(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
 	 * be valid at all if an error occurs so we have to invalidate it
 	 * prior to calling nfsm_reply ( which might goto nfsmout ).
 	 */
-	error = nfs_namei(&nd, fhp, len, slp, nam, &md, &dpos,
+	error = nfs_namei(&nd, nfsd, fhp, len, slp, nam, &md, &dpos,
 		&dirp, v3, &dirfor, &dirfor_ret, FALSE);
 	vfslocked = nfsrv_lockedpair_nd(vfslocked, &nd);
 	if (dirp && !v3) {
@@ -1987,7 +2001,7 @@ nfsrv_mknod(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
 	 * nfsmout.
 	 */
 
-	error = nfs_namei(&nd, fhp, len, slp, nam, &md, &dpos,
+	error = nfs_namei(&nd, nfsd, fhp, len, slp, nam, &md, &dpos,
 		&dirp, v3, &dirfor, &dirfor_ret, FALSE);
 	vfslocked = nfsrv_lockedpair_nd(vfslocked, &nd);
 	if (error) {
@@ -2169,7 +2183,7 @@ nfsrv_remove(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
 	nd.ni_cnd.cn_cred = cred;
 	nd.ni_cnd.cn_nameiop = DELETE;
 	nd.ni_cnd.cn_flags = LOCKPARENT | LOCKLEAF | MPSAFE;
-	error = nfs_namei(&nd, fhp, len, slp, nam, &md, &dpos,
+	error = nfs_namei(&nd, nfsd, fhp, len, slp, nam, &md, &dpos,
 		&dirp, v3,  &dirfor, &dirfor_ret, FALSE);
 	vfslocked = nfsrv_lockedpair_nd(vfslocked, &nd);
 	if (dirp && !v3) {
@@ -2296,7 +2310,7 @@ nfsrv_rename(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
 	fromnd.ni_cnd.cn_cred = cred;
 	fromnd.ni_cnd.cn_nameiop = DELETE;
 	fromnd.ni_cnd.cn_flags = WANTPARENT | SAVESTART | MPSAFE;
-	error = nfs_namei(&fromnd, ffhp, len, slp, nam, &md,
+	error = nfs_namei(&fromnd, nfsd, ffhp, len, slp, nam, &md,
 		&dpos, &fdirp, v3, &fdirfor, &fdirfor_ret, FALSE);
 	vfslocked = nfsrv_lockedpair_nd(vfslocked, &fromnd);
 	if (fdirp && !v3) {
@@ -2319,7 +2333,7 @@ nfsrv_rename(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
 	tond.ni_cnd.cn_cred = cred;
 	tond.ni_cnd.cn_nameiop = RENAME;
 	tond.ni_cnd.cn_flags = LOCKPARENT | LOCKLEAF | NOCACHE | SAVESTART | MPSAFE;
-	error = nfs_namei(&tond, tfhp, len2, slp, nam, &md,
+	error = nfs_namei(&tond, nfsd, tfhp, len2, slp, nam, &md,
 		&dpos, &tdirp, v3, &tdirfor, &tdirfor_ret, FALSE);
 	vfslocked = nfsrv_lockedpair_nd(vfslocked, &tond);
 	if (tdirp && !v3) {
@@ -2512,7 +2526,7 @@ nfsrv_link(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
 	nfsm_srvmtofh(dfhp);
 	nfsm_srvnamesiz(len);
 
-	error = nfsrv_fhtovp(fhp, TRUE, &vp, &tvfslocked, cred, slp,
+	error = nfsrv_fhtovp(fhp, TRUE, &vp, &tvfslocked, nfsd, slp,
 	    nam, &rdonly, TRUE);
 	vfslocked = nfsrv_lockedpair(vfslocked, tvfslocked);
 	if (error) {
@@ -2535,7 +2549,7 @@ nfsrv_link(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
 	nd.ni_cnd.cn_cred = cred;
 	nd.ni_cnd.cn_nameiop = CREATE;
 	nd.ni_cnd.cn_flags = LOCKPARENT | MPSAFE | MPSAFE;
-	error = nfs_namei(&nd, dfhp, len, slp, nam, &md, &dpos,
+	error = nfs_namei(&nd, nfsd, dfhp, len, slp, nam, &md, &dpos,
 		&dirp, v3, &dirfor, &dirfor_ret, FALSE);
 	vfslocked = nfsrv_lockedpair_nd(vfslocked, &nd);
 	if (dirp && !v3) {
@@ -2664,7 +2678,7 @@ nfsrv_symlink(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
 	nd.ni_cnd.cn_cred = cred;
 	nd.ni_cnd.cn_nameiop = CREATE;
 	nd.ni_cnd.cn_flags = LOCKPARENT | SAVESTART | MPSAFE;
-	error = nfs_namei(&nd, fhp, len, slp, nam, &md, &dpos,
+	error = nfs_namei(&nd, nfsd, fhp, len, slp, nam, &md, &dpos,
 		&dirp, v3, &dirfor, &dirfor_ret, FALSE);
 	vfslocked = nfsrv_lockedpair_nd(vfslocked, &nd);
 	if (error == 0) {
@@ -2847,7 +2861,7 @@ nfsrv_mkdir(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
 	nd.ni_cnd.cn_nameiop = CREATE;
 	nd.ni_cnd.cn_flags = LOCKPARENT | MPSAFE;
 
-	error = nfs_namei(&nd, fhp, len, slp, nam, &md, &dpos,
+	error = nfs_namei(&nd, nfsd, fhp, len, slp, nam, &md, &dpos,
 		&dirp, v3, &dirfor, &dirfor_ret, FALSE);
 	vfslocked = nfsrv_lockedpair_nd(vfslocked, &nd);
 	if (dirp && !v3) {
@@ -3005,7 +3019,7 @@ nfsrv_rmdir(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
 	nd.ni_cnd.cn_cred = cred;
 	nd.ni_cnd.cn_nameiop = DELETE;
 	nd.ni_cnd.cn_flags = LOCKPARENT | LOCKLEAF | MPSAFE;
-	error = nfs_namei(&nd, fhp, len, slp, nam, &md, &dpos,
+	error = nfs_namei(&nd, nfsd, fhp, len, slp, nam, &md, &dpos,
 		&dirp, v3, &dirfor, &dirfor_ret, FALSE);
 	vfslocked = nfsrv_lockedpair_nd(vfslocked, &nd);
 	if (dirp && !v3) {
@@ -3180,7 +3194,7 @@ nfsrv_readdir(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
 	if (siz > xfer)
 		siz = xfer;
 	fullsiz = siz;
-	error = nfsrv_fhtovp(fhp, 1, &vp, &vfslocked, cred, slp,
+	error = nfsrv_fhtovp(fhp, 1, &vp, &vfslocked, nfsd, slp,
 	    nam, &rdonly, TRUE);
 	if (!error && vp->v_type != VDIR) {
 		error = ENOTDIR;
@@ -3474,7 +3488,7 @@ nfsrv_readdirplus(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
 	if (siz > xfer)
 		siz = xfer;
 	fullsiz = siz;
-	error = nfsrv_fhtovp(fhp, 1, &vp, &vfslocked, cred, slp,
+	error = nfsrv_fhtovp(fhp, 1, &vp, &vfslocked, nfsd, slp,
 	    nam, &rdonly, TRUE);
 	if (!error && vp->v_type != VDIR) {
 		error = ENOTDIR;
@@ -3815,7 +3829,7 @@ nfsrv_commit(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
 	off = fxdr_hyper(tl);
 	tl += 2;
 	cnt = fxdr_unsigned(int, *tl);
-	error = nfsrv_fhtovp(fhp, 1, &vp, &tvfslocked, cred, slp,
+	error = nfsrv_fhtovp(fhp, 1, &vp, &tvfslocked, nfsd, slp,
 	    nam, &rdonly, TRUE);
 	vfslocked = nfsrv_lockedpair(vfslocked, tvfslocked);
 	if (error) {
@@ -3960,7 +3974,7 @@ nfsrv_statfs(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
 	vfslocked = 0;
 	fhp = &nfh.fh_generic;
 	nfsm_srvmtofh(fhp);
-	error = nfsrv_fhtovp(fhp, 1, &vp, &vfslocked, cred, slp,
+	error = nfsrv_fhtovp(fhp, 1, &vp, &vfslocked, nfsd, slp,
 	    nam, &rdonly, TRUE);
 	if (error) {
 		nfsm_reply(NFSX_UNSIGNED);
@@ -4055,7 +4069,7 @@ nfsrv_fsinfo(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
 	fhp = &nfh.fh_generic;
 	vfslocked = 0;
 	nfsm_srvmtofh(fhp);
-	error = nfsrv_fhtovp(fhp, 1, &vp, &vfslocked, cred, slp,
+	error = nfsrv_fhtovp(fhp, 1, &vp, &vfslocked, nfsd, slp,
 	    nam, &rdonly, TRUE);
 	if (error) {
 		nfsm_reply(NFSX_UNSIGNED);
@@ -4080,10 +4094,7 @@ nfsrv_fsinfo(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
 	 * There should be filesystem VFS OP(s) to get this information.
 	 * For now, assume ufs.
 	 */
-	if (slp->ns_so->so_type == SOCK_DGRAM)
-		pref = NFS_MAXDGRAMDATA;
-	else
-		pref = NFS_MAXDATA;
+	pref = NFS_SRVMAXDATA(nfsd);
 	sip->fs_rtmax = txdr_unsigned(pref);
 	sip->fs_rtpref = txdr_unsigned(pref);
 	sip->fs_rtmult = txdr_unsigned(NFS_FABLKSIZE);
@@ -4133,7 +4144,7 @@ nfsrv_pathconf(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
 	vfslocked = 0;
 	fhp = &nfh.fh_generic;
 	nfsm_srvmtofh(fhp);
-	error = nfsrv_fhtovp(fhp, 1, &vp, &vfslocked, cred, slp,
+	error = nfsrv_fhtovp(fhp, 1, &vp, &vfslocked, nfsd, slp,
 	    nam, &rdonly, TRUE);
 	if (error) {
 		nfsm_reply(NFSX_UNSIGNED);