2008-03-26 15:23:12 +00:00
|
|
|
/* $NetBSD: auth_unix.c,v 1.18 2000/07/06 03:03:30 christos Exp $ */
|
|
|
|
|
2013-11-25 19:04:36 +00:00
|
|
|
/*-
|
2017-11-20 19:43:44 +00:00
|
|
|
* SPDX-License-Identifier: BSD-3-Clause
|
|
|
|
*
|
2013-11-25 19:04:36 +00:00
|
|
|
* Copyright (c) 2009, Sun Microsystems, Inc.
|
|
|
|
* All rights reserved.
|
2008-03-26 15:23:12 +00:00
|
|
|
*
|
2013-11-25 19:04:36 +00:00
|
|
|
* Redistribution and use in source and binary forms, with or without
|
|
|
|
* modification, are permitted provided that the following conditions are met:
|
|
|
|
* - Redistributions of source code must retain the above copyright notice,
|
|
|
|
* this list of conditions and the following disclaimer.
|
|
|
|
* - Redistributions in binary form must reproduce the above copyright notice,
|
|
|
|
* this list of conditions and the following disclaimer in the documentation
|
|
|
|
* and/or other materials provided with the distribution.
|
|
|
|
* - Neither the name of Sun Microsystems, Inc. nor the names of its
|
|
|
|
* contributors may be used to endorse or promote products derived
|
|
|
|
* from this software without specific prior written permission.
|
|
|
|
*
|
|
|
|
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
|
|
|
|
* AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
|
|
|
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
|
|
|
* ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE
|
|
|
|
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
|
|
|
|
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
|
|
|
|
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
|
|
|
|
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
|
|
|
|
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
|
|
|
|
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
|
|
|
|
* POSSIBILITY OF SUCH DAMAGE.
|
2008-03-26 15:23:12 +00:00
|
|
|
*/
|
|
|
|
|
|
|
|
#if defined(LIBC_SCCS) && !defined(lint)
|
|
|
|
static char *sccsid2 = "@(#)auth_unix.c 1.19 87/08/11 Copyr 1984 Sun Micro";
|
|
|
|
static char *sccsid = "@(#)auth_unix.c 2.2 88/08/01 4.0 RPCSRC";
|
|
|
|
#endif
|
|
|
|
#include <sys/cdefs.h>
|
|
|
|
/*
|
|
|
|
* auth_unix.c, Implements UNIX style authentication parameters.
|
|
|
|
*
|
|
|
|
* Copyright (C) 1984, Sun Microsystems, Inc.
|
|
|
|
*
|
|
|
|
* The system is very weak. The client uses no encryption for it's
|
|
|
|
* credentials and only sends null verifiers. The server sends backs
|
|
|
|
* null verifiers or optionally a verifier that suggests a new short hand
|
|
|
|
* for the credentials.
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <sys/param.h>
|
|
|
|
#include <sys/systm.h>
|
2008-06-26 10:21:54 +00:00
|
|
|
#include <sys/hash.h>
|
|
|
|
#include <sys/kernel.h>
|
2008-03-26 15:23:12 +00:00
|
|
|
#include <sys/lock.h>
|
|
|
|
#include <sys/malloc.h>
|
2008-06-27 14:35:05 +00:00
|
|
|
#include <sys/pcpu.h>
|
2008-07-23 09:18:08 +00:00
|
|
|
#include <sys/refcount.h>
|
2008-06-26 10:21:54 +00:00
|
|
|
#include <sys/sx.h>
|
2008-03-26 15:23:12 +00:00
|
|
|
#include <sys/ucred.h>
|
|
|
|
|
|
|
|
#include <rpc/types.h>
|
|
|
|
#include <rpc/xdr.h>
|
|
|
|
#include <rpc/auth.h>
|
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
2008-11-03 10:38:00 +00:00
|
|
|
#include <rpc/clnt.h>
|
2008-03-26 15:23:12 +00:00
|
|
|
|
2008-03-28 09:50:32 +00:00
|
|
|
#include <rpc/rpc_com.h>
|
2008-03-26 15:23:12 +00:00
|
|
|
|
|
|
|
/* auth_unix.c */
|
|
|
|
static void authunix_nextverf (AUTH *);
|
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
2008-11-03 10:38:00 +00:00
|
|
|
static bool_t authunix_marshal (AUTH *, uint32_t, XDR *, struct mbuf *);
|
|
|
|
static bool_t authunix_validate (AUTH *, uint32_t, struct opaque_auth *,
|
|
|
|
struct mbuf **);
|
2008-03-26 15:23:12 +00:00
|
|
|
static bool_t authunix_refresh (AUTH *, void *);
|
|
|
|
static void authunix_destroy (AUTH *);
|
|
|
|
static void marshal_new_auth (AUTH *);
|
|
|
|
|
2021-07-09 14:56:13 +00:00
|
|
|
static const struct auth_ops authunix_ops = {
|
2008-03-26 15:23:12 +00:00
|
|
|
.ah_nextverf = authunix_nextverf,
|
|
|
|
.ah_marshal = authunix_marshal,
|
|
|
|
.ah_validate = authunix_validate,
|
|
|
|
.ah_refresh = authunix_refresh,
|
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
2008-11-03 10:38:00 +00:00
|
|
|
.ah_destroy = authunix_destroy,
|
2008-03-26 15:23:12 +00:00
|
|
|
};
|
|
|
|
|
|
|
|
/*
|
|
|
|
* This struct is pointed to by the ah_private field of an auth_handle.
|
|
|
|
*/
|
|
|
|
struct audata {
|
2008-06-26 10:21:54 +00:00
|
|
|
TAILQ_ENTRY(audata) au_link;
|
|
|
|
TAILQ_ENTRY(audata) au_alllink;
|
2008-07-23 09:18:08 +00:00
|
|
|
volatile u_int au_refs;
|
2008-06-26 10:21:54 +00:00
|
|
|
struct xucred au_xcred;
|
2008-03-26 15:23:12 +00:00
|
|
|
struct opaque_auth au_origcred; /* original credentials */
|
|
|
|
struct opaque_auth au_shcred; /* short hand cred */
|
|
|
|
u_long au_shfaults; /* short hand cache faults */
|
|
|
|
char au_marshed[MAX_AUTH_BYTES];
|
|
|
|
u_int au_mpos; /* xdr pos at end of marshed */
|
2008-06-26 10:21:54 +00:00
|
|
|
AUTH *au_auth; /* link back to AUTH */
|
2008-03-26 15:23:12 +00:00
|
|
|
};
|
2008-06-26 10:21:54 +00:00
|
|
|
TAILQ_HEAD(audata_list, audata);
|
2008-03-26 15:23:12 +00:00
|
|
|
#define AUTH_PRIVATE(auth) ((struct audata *)auth->ah_private)
|
|
|
|
|
2008-06-26 10:21:54 +00:00
|
|
|
#define AUTH_UNIX_HASH_SIZE 16
|
|
|
|
#define AUTH_UNIX_MAX 256
|
|
|
|
static struct audata_list auth_unix_cache[AUTH_UNIX_HASH_SIZE];
|
|
|
|
static struct audata_list auth_unix_all;
|
|
|
|
static struct sx auth_unix_lock;
|
|
|
|
static int auth_unix_count;
|
|
|
|
|
|
|
|
static void
|
|
|
|
authunix_init(void *dummy)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
|
|
|
|
for (i = 0; i < AUTH_UNIX_HASH_SIZE; i++)
|
|
|
|
TAILQ_INIT(&auth_unix_cache[i]);
|
|
|
|
TAILQ_INIT(&auth_unix_all);
|
|
|
|
sx_init(&auth_unix_lock, "auth_unix_lock");
|
|
|
|
}
|
|
|
|
SYSINIT(authunix_init, SI_SUB_KMEM, SI_ORDER_ANY, authunix_init, NULL);
|
|
|
|
|
2008-03-26 15:23:12 +00:00
|
|
|
/*
|
|
|
|
* Create a unix style authenticator.
|
|
|
|
* Returns an auth handle with the given stuff in it.
|
|
|
|
*/
|
|
|
|
AUTH *
|
|
|
|
authunix_create(struct ucred *cred)
|
|
|
|
{
|
2008-06-26 10:21:54 +00:00
|
|
|
uint32_t h, th;
|
2008-03-26 15:23:12 +00:00
|
|
|
struct xucred xcr;
|
|
|
|
char mymem[MAX_AUTH_BYTES];
|
|
|
|
XDR xdrs;
|
|
|
|
AUTH *auth;
|
2008-06-26 10:21:54 +00:00
|
|
|
struct audata *au, *tau;
|
2008-03-26 15:23:12 +00:00
|
|
|
struct timeval now;
|
|
|
|
uint32_t time;
|
|
|
|
int len;
|
|
|
|
|
2008-06-26 10:21:54 +00:00
|
|
|
if (auth_unix_count > AUTH_UNIX_MAX) {
|
|
|
|
while (auth_unix_count > AUTH_UNIX_MAX) {
|
|
|
|
sx_xlock(&auth_unix_lock);
|
|
|
|
tau = TAILQ_FIRST(&auth_unix_all);
|
|
|
|
th = HASHSTEP(HASHINIT, tau->au_xcred.cr_uid)
|
|
|
|
% AUTH_UNIX_HASH_SIZE;
|
|
|
|
TAILQ_REMOVE(&auth_unix_cache[th], tau, au_link);
|
|
|
|
TAILQ_REMOVE(&auth_unix_all, tau, au_alllink);
|
|
|
|
auth_unix_count--;
|
|
|
|
sx_xunlock(&auth_unix_lock);
|
|
|
|
AUTH_DESTROY(tau->au_auth);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Hash the uid to see if we already have an AUTH with this cred.
|
|
|
|
*/
|
|
|
|
h = HASHSTEP(HASHINIT, cred->cr_uid) % AUTH_UNIX_HASH_SIZE;
|
|
|
|
cru2x(cred, &xcr);
|
|
|
|
again:
|
|
|
|
sx_slock(&auth_unix_lock);
|
|
|
|
TAILQ_FOREACH(au, &auth_unix_cache[h], au_link) {
|
|
|
|
if (!memcmp(&xcr, &au->au_xcred, sizeof(xcr))) {
|
2008-07-23 09:18:08 +00:00
|
|
|
refcount_acquire(&au->au_refs);
|
2008-06-26 10:21:54 +00:00
|
|
|
if (sx_try_upgrade(&auth_unix_lock)) {
|
|
|
|
/*
|
|
|
|
* Keep auth_unix_all LRU sorted.
|
|
|
|
*/
|
|
|
|
TAILQ_REMOVE(&auth_unix_all, au, au_alllink);
|
|
|
|
TAILQ_INSERT_TAIL(&auth_unix_all, au,
|
|
|
|
au_alllink);
|
|
|
|
sx_xunlock(&auth_unix_lock);
|
|
|
|
} else {
|
|
|
|
sx_sunlock(&auth_unix_lock);
|
|
|
|
}
|
2008-07-23 09:18:08 +00:00
|
|
|
return (au->au_auth);
|
2008-06-26 10:21:54 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2008-07-23 09:18:08 +00:00
|
|
|
sx_sunlock(&auth_unix_lock);
|
|
|
|
|
2008-03-26 15:23:12 +00:00
|
|
|
/*
|
|
|
|
* Allocate and set up auth handle
|
|
|
|
*/
|
|
|
|
au = NULL;
|
|
|
|
auth = mem_alloc(sizeof(*auth));
|
|
|
|
au = mem_alloc(sizeof(*au));
|
|
|
|
auth->ah_ops = &authunix_ops;
|
|
|
|
auth->ah_private = (caddr_t)au;
|
|
|
|
auth->ah_verf = au->au_shcred = _null_auth;
|
2008-07-23 09:18:08 +00:00
|
|
|
refcount_init(&au->au_refs, 1);
|
2008-06-26 10:21:54 +00:00
|
|
|
au->au_xcred = xcr;
|
2008-03-26 15:23:12 +00:00
|
|
|
au->au_shfaults = 0;
|
|
|
|
au->au_origcred.oa_base = NULL;
|
2008-06-26 10:21:54 +00:00
|
|
|
au->au_auth = auth;
|
2008-03-26 15:23:12 +00:00
|
|
|
|
|
|
|
getmicrotime(&now);
|
|
|
|
time = now.tv_sec;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Serialize the parameters into origcred
|
|
|
|
*/
|
|
|
|
xdrmem_create(&xdrs, mymem, MAX_AUTH_BYTES, XDR_ENCODE);
|
|
|
|
cru2x(cred, &xcr);
|
|
|
|
if (! xdr_authunix_parms(&xdrs, &time, &xcr))
|
|
|
|
panic("authunix_create: failed to encode creds");
|
|
|
|
au->au_origcred.oa_length = len = XDR_GETPOS(&xdrs);
|
|
|
|
au->au_origcred.oa_flavor = AUTH_UNIX;
|
|
|
|
au->au_origcred.oa_base = mem_alloc((u_int) len);
|
|
|
|
memcpy(au->au_origcred.oa_base, mymem, (size_t)len);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* set auth handle to reflect new cred.
|
|
|
|
*/
|
|
|
|
auth->ah_cred = au->au_origcred;
|
|
|
|
marshal_new_auth(auth);
|
2008-06-26 10:21:54 +00:00
|
|
|
|
2008-07-23 09:18:08 +00:00
|
|
|
sx_xlock(&auth_unix_lock);
|
|
|
|
TAILQ_FOREACH(tau, &auth_unix_cache[h], au_link) {
|
|
|
|
if (!memcmp(&xcr, &tau->au_xcred, sizeof(xcr))) {
|
|
|
|
/*
|
|
|
|
* We lost a race to create the AUTH that
|
|
|
|
* matches this cred.
|
|
|
|
*/
|
|
|
|
sx_xunlock(&auth_unix_lock);
|
|
|
|
AUTH_DESTROY(auth);
|
|
|
|
goto again;
|
|
|
|
}
|
2008-03-26 15:23:12 +00:00
|
|
|
}
|
2008-07-23 09:18:08 +00:00
|
|
|
|
|
|
|
auth_unix_count++;
|
|
|
|
TAILQ_INSERT_TAIL(&auth_unix_cache[h], au, au_link);
|
|
|
|
TAILQ_INSERT_TAIL(&auth_unix_all, au, au_alllink);
|
|
|
|
refcount_acquire(&au->au_refs); /* one for the cache, one for user */
|
|
|
|
sx_xunlock(&auth_unix_lock);
|
|
|
|
|
|
|
|
return (auth);
|
2008-03-26 15:23:12 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* authunix operations
|
|
|
|
*/
|
|
|
|
|
|
|
|
/* ARGSUSED */
|
|
|
|
static void
|
|
|
|
authunix_nextverf(AUTH *auth)
|
|
|
|
{
|
|
|
|
/* no action necessary */
|
|
|
|
}
|
|
|
|
|
|
|
|
static bool_t
|
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
2008-11-03 10:38:00 +00:00
|
|
|
authunix_marshal(AUTH *auth, uint32_t xid, XDR *xdrs, struct mbuf *args)
|
2008-03-26 15:23:12 +00:00
|
|
|
{
|
|
|
|
struct audata *au;
|
|
|
|
|
|
|
|
au = AUTH_PRIVATE(auth);
|
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
2008-11-03 10:38:00 +00:00
|
|
|
if (!XDR_PUTBYTES(xdrs, au->au_marshed, au->au_mpos))
|
|
|
|
return (FALSE);
|
|
|
|
|
|
|
|
xdrmbuf_append(xdrs, args);
|
|
|
|
|
|
|
|
return (TRUE);
|
2008-03-26 15:23:12 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static bool_t
|
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
2008-11-03 10:38:00 +00:00
|
|
|
authunix_validate(AUTH *auth, uint32_t xid, struct opaque_auth *verf,
|
|
|
|
struct mbuf **mrepp)
|
2008-03-26 15:23:12 +00:00
|
|
|
{
|
|
|
|
struct audata *au;
|
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
2008-11-03 10:38:00 +00:00
|
|
|
XDR txdrs;
|
|
|
|
|
|
|
|
if (!verf)
|
|
|
|
return (TRUE);
|
2008-03-26 15:23:12 +00:00
|
|
|
|
|
|
|
if (verf->oa_flavor == AUTH_SHORT) {
|
|
|
|
au = AUTH_PRIVATE(auth);
|
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
2008-11-03 10:38:00 +00:00
|
|
|
xdrmem_create(&txdrs, verf->oa_base, verf->oa_length,
|
2008-03-26 15:23:12 +00:00
|
|
|
XDR_DECODE);
|
|
|
|
|
|
|
|
if (au->au_shcred.oa_base != NULL) {
|
|
|
|
mem_free(au->au_shcred.oa_base,
|
|
|
|
au->au_shcred.oa_length);
|
|
|
|
au->au_shcred.oa_base = NULL;
|
|
|
|
}
|
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
2008-11-03 10:38:00 +00:00
|
|
|
if (xdr_opaque_auth(&txdrs, &au->au_shcred)) {
|
2008-03-26 15:23:12 +00:00
|
|
|
auth->ah_cred = au->au_shcred;
|
|
|
|
} else {
|
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
2008-11-03 10:38:00 +00:00
|
|
|
txdrs.x_op = XDR_FREE;
|
|
|
|
(void)xdr_opaque_auth(&txdrs, &au->au_shcred);
|
2008-03-26 15:23:12 +00:00
|
|
|
au->au_shcred.oa_base = NULL;
|
|
|
|
auth->ah_cred = au->au_origcred;
|
|
|
|
}
|
|
|
|
marshal_new_auth(auth);
|
|
|
|
}
|
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
2008-11-03 10:38:00 +00:00
|
|
|
|
2008-03-26 15:23:12 +00:00
|
|
|
return (TRUE);
|
|
|
|
}
|
|
|
|
|
|
|
|
static bool_t
|
|
|
|
authunix_refresh(AUTH *auth, void *dummy)
|
|
|
|
{
|
|
|
|
struct audata *au = AUTH_PRIVATE(auth);
|
|
|
|
struct xucred xcr;
|
|
|
|
uint32_t time;
|
|
|
|
struct timeval now;
|
|
|
|
XDR xdrs;
|
|
|
|
int stat;
|
|
|
|
|
|
|
|
if (auth->ah_cred.oa_base == au->au_origcred.oa_base) {
|
|
|
|
/* there is no hope. Punt */
|
|
|
|
return (FALSE);
|
|
|
|
}
|
|
|
|
au->au_shfaults ++;
|
|
|
|
|
|
|
|
/* first deserialize the creds back into a struct ucred */
|
|
|
|
xdrmem_create(&xdrs, au->au_origcred.oa_base,
|
|
|
|
au->au_origcred.oa_length, XDR_DECODE);
|
|
|
|
stat = xdr_authunix_parms(&xdrs, &time, &xcr);
|
|
|
|
if (! stat)
|
|
|
|
goto done;
|
|
|
|
|
|
|
|
/* update the time and serialize in place */
|
|
|
|
getmicrotime(&now);
|
|
|
|
time = now.tv_sec;
|
|
|
|
xdrs.x_op = XDR_ENCODE;
|
|
|
|
XDR_SETPOS(&xdrs, 0);
|
|
|
|
|
|
|
|
stat = xdr_authunix_parms(&xdrs, &time, &xcr);
|
|
|
|
if (! stat)
|
|
|
|
goto done;
|
|
|
|
auth->ah_cred = au->au_origcred;
|
|
|
|
marshal_new_auth(auth);
|
|
|
|
done:
|
|
|
|
XDR_DESTROY(&xdrs);
|
|
|
|
return (stat);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
authunix_destroy(AUTH *auth)
|
|
|
|
{
|
|
|
|
struct audata *au;
|
|
|
|
|
|
|
|
au = AUTH_PRIVATE(auth);
|
2008-06-26 10:21:54 +00:00
|
|
|
|
2008-07-23 09:18:08 +00:00
|
|
|
if (!refcount_release(&au->au_refs))
|
2008-06-26 10:21:54 +00:00
|
|
|
return;
|
|
|
|
|
2008-03-26 15:23:12 +00:00
|
|
|
mem_free(au->au_origcred.oa_base, au->au_origcred.oa_length);
|
|
|
|
|
|
|
|
if (au->au_shcred.oa_base != NULL)
|
|
|
|
mem_free(au->au_shcred.oa_base, au->au_shcred.oa_length);
|
|
|
|
|
|
|
|
mem_free(auth->ah_private, sizeof(struct audata));
|
|
|
|
|
|
|
|
if (auth->ah_verf.oa_base != NULL)
|
|
|
|
mem_free(auth->ah_verf.oa_base, auth->ah_verf.oa_length);
|
|
|
|
|
|
|
|
mem_free(auth, sizeof(*auth));
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Marshals (pre-serializes) an auth struct.
|
|
|
|
* sets private data, au_marshed and au_mpos
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
marshal_new_auth(AUTH *auth)
|
|
|
|
{
|
|
|
|
XDR xdr_stream;
|
|
|
|
XDR *xdrs = &xdr_stream;
|
|
|
|
struct audata *au;
|
|
|
|
|
|
|
|
au = AUTH_PRIVATE(auth);
|
|
|
|
xdrmem_create(xdrs, au->au_marshed, MAX_AUTH_BYTES, XDR_ENCODE);
|
|
|
|
if ((! xdr_opaque_auth(xdrs, &(auth->ah_cred))) ||
|
|
|
|
(! xdr_opaque_auth(xdrs, &(auth->ah_verf))))
|
|
|
|
printf("auth_none.c - Fatal marshalling problem");
|
|
|
|
else
|
|
|
|
au->au_mpos = XDR_GETPOS(xdrs);
|
|
|
|
XDR_DESTROY(xdrs);
|
|
|
|
}
|