freebsd-nq/sys/rpc/rpcsec_gss/svc_rpcsec_gss.c

1530 lines
39 KiB
C
Raw Normal View History

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
/*-
* Copyright (c) 2008 Doug Rabson
* All rights reserved.
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
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. 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.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
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
* 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 AUTHOR OR CONTRIBUTORS BE LIABLE
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
* 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.
*/
/*
svc_rpcsec_gss.c
Copyright (c) 2000 The Regents of the University of Michigan.
All rights reserved.
Copyright (c) 2000 Dug Song <dugsong@UMICH.EDU>.
All rights reserved, all wrongs reversed.
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
1. Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
2. 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.
3. Neither the name of the University 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 ``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 REGENTS 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.
$Id: svc_auth_gss.c,v 1.27 2002/01/15 15:43:00 andros Exp $
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 <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/jail.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 <sys/kernel.h>
#include <sys/kobj.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 <sys/lock.h>
#include <sys/malloc.h>
#include <sys/mbuf.h>
#include <sys/mutex.h>
#include <sys/proc.h>
#include <sys/sx.h>
#include <sys/ucred.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/rpc.h>
#include <rpc/rpcsec_gss.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 "rpcsec_gss_int.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
static bool_t svc_rpc_gss_wrap(SVCAUTH *, struct mbuf **);
static bool_t svc_rpc_gss_unwrap(SVCAUTH *, struct mbuf **);
static void svc_rpc_gss_release(SVCAUTH *);
static enum auth_stat svc_rpc_gss(struct svc_req *, struct rpc_msg *);
static int rpc_gss_svc_getcred(struct svc_req *, struct ucred **, int *);
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 struct svc_auth_ops svc_auth_gss_ops = {
svc_rpc_gss_wrap,
svc_rpc_gss_unwrap,
svc_rpc_gss_release,
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
};
struct sx svc_rpc_gss_lock;
struct svc_rpc_gss_callback {
SLIST_ENTRY(svc_rpc_gss_callback) cb_link;
rpc_gss_callback_t cb_callback;
};
static SLIST_HEAD(svc_rpc_gss_callback_list, svc_rpc_gss_callback)
svc_rpc_gss_callbacks = SLIST_HEAD_INITIALIZER(svc_rpc_gss_callbacks);
struct svc_rpc_gss_svc_name {
SLIST_ENTRY(svc_rpc_gss_svc_name) sn_link;
char *sn_principal;
gss_OID sn_mech;
u_int sn_req_time;
gss_cred_id_t sn_cred;
u_int sn_program;
u_int sn_version;
};
static SLIST_HEAD(svc_rpc_gss_svc_name_list, svc_rpc_gss_svc_name)
svc_rpc_gss_svc_names = SLIST_HEAD_INITIALIZER(svc_rpc_gss_svc_names);
enum svc_rpc_gss_client_state {
CLIENT_NEW, /* still authenticating */
CLIENT_ESTABLISHED, /* context established */
CLIENT_STALE /* garbage to collect */
};
#define SVC_RPC_GSS_SEQWINDOW 128
#ifndef RPCAUTH_UNIXGIDS
#define RPCAUTH_UNIXGIDS 16
#endif
struct svc_rpc_gss_clientid {
unsigned long ci_hostid;
uint32_t ci_boottime;
uint32_t ci_id;
};
struct svc_rpc_gss_client {
TAILQ_ENTRY(svc_rpc_gss_client) cl_link;
TAILQ_ENTRY(svc_rpc_gss_client) cl_alllink;
volatile u_int cl_refs;
struct sx cl_lock;
struct svc_rpc_gss_clientid cl_id;
time_t cl_expiration; /* when to gc */
enum svc_rpc_gss_client_state cl_state; /* client state */
bool_t cl_locked; /* fixed service+qop */
gss_ctx_id_t cl_ctx; /* context id */
gss_cred_id_t cl_creds; /* delegated creds */
gss_name_t cl_cname; /* client name */
struct svc_rpc_gss_svc_name *cl_sname; /* server name used */
rpc_gss_rawcred_t cl_rawcred; /* raw credentials */
rpc_gss_ucred_t cl_ucred; /* unix-style credentials */
struct ucred *cl_cred; /* kernel-style credentials */
int cl_rpcflavor; /* RPC pseudo sec flavor */
bool_t cl_done_callback; /* TRUE after call */
void *cl_cookie; /* user cookie from callback */
gid_t cl_gid_storage[RPCAUTH_UNIXGIDS];
gss_OID cl_mech; /* mechanism */
gss_qop_t cl_qop; /* quality of protection */
uint32_t cl_seqlast; /* sequence window origin */
uint32_t cl_seqmask[SVC_RPC_GSS_SEQWINDOW/32]; /* bitmask of seqnums */
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
};
TAILQ_HEAD(svc_rpc_gss_client_list, svc_rpc_gss_client);
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
/*
* This structure holds enough information to unwrap arguments or wrap
* results for a given request. We use the rq_clntcred area for this
* (which is a per-request buffer).
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
*/
struct svc_rpc_gss_cookedcred {
struct svc_rpc_gss_client *cc_client;
rpc_gss_service_t cc_service;
uint32_t cc_seq;
};
#define CLIENT_HASH_SIZE 256
#define CLIENT_MAX 128
struct svc_rpc_gss_client_list svc_rpc_gss_client_hash[CLIENT_HASH_SIZE];
struct svc_rpc_gss_client_list svc_rpc_gss_clients;
static size_t svc_rpc_gss_client_count;
static uint32_t svc_rpc_gss_next_clientid = 1;
static void
svc_rpc_gss_init(void *arg)
{
int i;
for (i = 0; i < CLIENT_HASH_SIZE; i++)
TAILQ_INIT(&svc_rpc_gss_client_hash[i]);
TAILQ_INIT(&svc_rpc_gss_clients);
svc_auth_reg(RPCSEC_GSS, svc_rpc_gss, rpc_gss_svc_getcred);
sx_init(&svc_rpc_gss_lock, "gsslock");
}
SYSINIT(svc_rpc_gss_init, SI_SUB_KMEM, SI_ORDER_ANY, svc_rpc_gss_init, NULL);
bool_t
rpc_gss_set_callback(rpc_gss_callback_t *cb)
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
{
struct svc_rpc_gss_callback *scb;
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
scb = mem_alloc(sizeof(struct svc_rpc_gss_callback));
if (!scb) {
_rpc_gss_set_error(RPC_GSS_ER_SYSTEMERROR, ENOMEM);
return (FALSE);
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
}
scb->cb_callback = *cb;
sx_xlock(&svc_rpc_gss_lock);
SLIST_INSERT_HEAD(&svc_rpc_gss_callbacks, scb, cb_link);
sx_xunlock(&svc_rpc_gss_lock);
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
return (TRUE);
}
void
rpc_gss_clear_callback(rpc_gss_callback_t *cb)
{
struct svc_rpc_gss_callback *scb;
sx_xlock(&svc_rpc_gss_lock);
SLIST_FOREACH(scb, &svc_rpc_gss_callbacks, cb_link) {
if (scb->cb_callback.program == cb->program
&& scb->cb_callback.version == cb->version
&& scb->cb_callback.callback == cb->callback) {
SLIST_REMOVE(&svc_rpc_gss_callbacks, scb,
svc_rpc_gss_callback, cb_link);
sx_xunlock(&svc_rpc_gss_lock);
mem_free(scb, sizeof(*scb));
return;
}
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
}
sx_xunlock(&svc_rpc_gss_lock);
}
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
rpc_gss_acquire_svc_cred(struct svc_rpc_gss_svc_name *sname)
{
OM_uint32 maj_stat, min_stat;
gss_buffer_desc namebuf;
gss_name_t name;
gss_OID_set_desc oid_set;
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
oid_set.count = 1;
oid_set.elements = sname->sn_mech;
namebuf.value = (void *) sname->sn_principal;
namebuf.length = strlen(sname->sn_principal);
maj_stat = gss_import_name(&min_stat, &namebuf,
GSS_C_NT_HOSTBASED_SERVICE, &name);
if (maj_stat != GSS_S_COMPLETE)
return (FALSE);
if (sname->sn_cred != GSS_C_NO_CREDENTIAL)
gss_release_cred(&min_stat, &sname->sn_cred);
maj_stat = gss_acquire_cred(&min_stat, name,
sname->sn_req_time, &oid_set, GSS_C_ACCEPT, &sname->sn_cred,
NULL, NULL);
if (maj_stat != GSS_S_COMPLETE) {
gss_release_name(&min_stat, &name);
return (FALSE);
}
gss_release_name(&min_stat, &name);
return (TRUE);
}
bool_t
rpc_gss_set_svc_name(const char *principal, const char *mechanism,
u_int req_time, u_int program, u_int version)
{
struct svc_rpc_gss_svc_name *sname;
gss_OID mech_oid;
if (!rpc_gss_mech_to_oid(mechanism, &mech_oid))
return (FALSE);
sname = mem_alloc(sizeof(*sname));
if (!sname)
return (FALSE);
sname->sn_principal = strdup(principal, M_RPC);
sname->sn_mech = mech_oid;
sname->sn_req_time = req_time;
sname->sn_cred = GSS_C_NO_CREDENTIAL;
sname->sn_program = program;
sname->sn_version = version;
if (!rpc_gss_acquire_svc_cred(sname)) {
free(sname->sn_principal, M_RPC);
mem_free(sname, sizeof(*sname));
return (FALSE);
}
sx_xlock(&svc_rpc_gss_lock);
SLIST_INSERT_HEAD(&svc_rpc_gss_svc_names, sname, sn_link);
sx_xunlock(&svc_rpc_gss_lock);
return (TRUE);
}
void
rpc_gss_clear_svc_name(u_int program, u_int version)
{
OM_uint32 min_stat;
struct svc_rpc_gss_svc_name *sname;
sx_xlock(&svc_rpc_gss_lock);
SLIST_FOREACH(sname, &svc_rpc_gss_svc_names, sn_link) {
if (sname->sn_program == program
&& sname->sn_version == version) {
SLIST_REMOVE(&svc_rpc_gss_svc_names, sname,
svc_rpc_gss_svc_name, sn_link);
sx_xunlock(&svc_rpc_gss_lock);
gss_release_cred(&min_stat, &sname->sn_cred);
free(sname->sn_principal, M_RPC);
mem_free(sname, sizeof(*sname));
return;
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
}
}
sx_xunlock(&svc_rpc_gss_lock);
}
bool_t
rpc_gss_get_principal_name(rpc_gss_principal_t *principal,
const char *mech, const char *name, const char *node, const char *domain)
{
OM_uint32 maj_stat, min_stat;
gss_OID mech_oid;
size_t namelen;
gss_buffer_desc buf;
gss_name_t gss_name, gss_mech_name;
rpc_gss_principal_t result;
if (!rpc_gss_mech_to_oid(mech, &mech_oid))
return (FALSE);
/*
* Construct a gss_buffer containing the full name formatted
* as "name/node@domain" where node and domain are optional.
*/
namelen = strlen(name);
if (node) {
namelen += strlen(node) + 1;
}
if (domain) {
namelen += strlen(domain) + 1;
}
buf.value = mem_alloc(namelen);
buf.length = namelen;
strcpy((char *) buf.value, name);
if (node) {
strcat((char *) buf.value, "/");
strcat((char *) buf.value, node);
}
if (domain) {
strcat((char *) buf.value, "@");
strcat((char *) buf.value, domain);
}
/*
* Convert that to a gss_name_t and then convert that to a
* mechanism name in the selected mechanism.
*/
maj_stat = gss_import_name(&min_stat, &buf,
GSS_C_NT_USER_NAME, &gss_name);
mem_free(buf.value, buf.length);
if (maj_stat != GSS_S_COMPLETE) {
rpc_gss_log_status("gss_import_name", mech_oid, maj_stat, min_stat);
return (FALSE);
}
maj_stat = gss_canonicalize_name(&min_stat, gss_name, mech_oid,
&gss_mech_name);
if (maj_stat != GSS_S_COMPLETE) {
rpc_gss_log_status("gss_canonicalize_name", mech_oid, maj_stat,
min_stat);
gss_release_name(&min_stat, &gss_name);
return (FALSE);
}
gss_release_name(&min_stat, &gss_name);
/*
* Export the mechanism name and use that to construct the
* rpc_gss_principal_t result.
*/
maj_stat = gss_export_name(&min_stat, gss_mech_name, &buf);
if (maj_stat != GSS_S_COMPLETE) {
rpc_gss_log_status("gss_export_name", mech_oid, maj_stat, min_stat);
gss_release_name(&min_stat, &gss_mech_name);
return (FALSE);
}
gss_release_name(&min_stat, &gss_mech_name);
result = mem_alloc(sizeof(int) + buf.length);
if (!result) {
gss_release_buffer(&min_stat, &buf);
return (FALSE);
}
result->len = buf.length;
memcpy(result->name, buf.value, buf.length);
gss_release_buffer(&min_stat, &buf);
*principal = result;
return (TRUE);
}
bool_t
rpc_gss_getcred(struct svc_req *req, rpc_gss_rawcred_t **rcred,
rpc_gss_ucred_t **ucred, void **cookie)
{
struct svc_rpc_gss_cookedcred *cc;
struct svc_rpc_gss_client *client;
if (req->rq_cred.oa_flavor != RPCSEC_GSS)
return (FALSE);
cc = req->rq_clntcred;
client = cc->cc_client;
if (rcred)
*rcred = &client->cl_rawcred;
if (ucred)
*ucred = &client->cl_ucred;
if (cookie)
*cookie = client->cl_cookie;
return (TRUE);
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
}
/*
* This simpler interface is used by svc_getcred to copy the cred data
* into a kernel cred structure.
*/
static int
rpc_gss_svc_getcred(struct svc_req *req, struct ucred **crp, int *flavorp)
{
struct ucred *cr;
struct svc_rpc_gss_cookedcred *cc;
struct svc_rpc_gss_client *client;
rpc_gss_ucred_t *uc;
if (req->rq_cred.oa_flavor != RPCSEC_GSS)
return (FALSE);
cc = req->rq_clntcred;
client = cc->cc_client;
if (flavorp)
*flavorp = client->cl_rpcflavor;
if (client->cl_cred) {
*crp = crhold(client->cl_cred);
return (TRUE);
}
uc = &client->cl_ucred;
cr = client->cl_cred = crget();
cr->cr_uid = cr->cr_ruid = cr->cr_svuid = uc->uid;
cr->cr_rgid = cr->cr_svgid = uc->gid;
crsetgroups(cr, uc->gidlen, uc->gidlist);
cr->cr_prison = &prison0;
prison_hold(cr->cr_prison);
*crp = crhold(cr);
return (TRUE);
}
int
rpc_gss_svc_max_data_length(struct svc_req *req, int max_tp_unit_len)
{
struct svc_rpc_gss_cookedcred *cc = req->rq_clntcred;
struct svc_rpc_gss_client *client = cc->cc_client;
int want_conf;
OM_uint32 max;
OM_uint32 maj_stat, min_stat;
int result;
switch (client->cl_rawcred.service) {
case rpc_gss_svc_none:
return (max_tp_unit_len);
break;
case rpc_gss_svc_default:
case rpc_gss_svc_integrity:
want_conf = FALSE;
break;
case rpc_gss_svc_privacy:
want_conf = TRUE;
break;
default:
return (0);
}
maj_stat = gss_wrap_size_limit(&min_stat, client->cl_ctx, want_conf,
client->cl_qop, max_tp_unit_len, &max);
if (maj_stat == GSS_S_COMPLETE) {
result = (int) max;
if (result < 0)
result = 0;
return (result);
} else {
rpc_gss_log_status("gss_wrap_size_limit", client->cl_mech,
maj_stat, min_stat);
return (0);
}
}
static struct svc_rpc_gss_client *
svc_rpc_gss_find_client(struct svc_rpc_gss_clientid *id)
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
{
struct svc_rpc_gss_client *client;
struct svc_rpc_gss_client_list *list;
unsigned long hostid;
rpc_gss_log_debug("in svc_rpc_gss_find_client(%d)", id->ci_id);
getcredhostid(curthread->td_ucred, &hostid);
if (id->ci_hostid != hostid || id->ci_boottime != boottime.tv_sec)
return (NULL);
list = &svc_rpc_gss_client_hash[id->ci_id % CLIENT_HASH_SIZE];
sx_xlock(&svc_rpc_gss_lock);
TAILQ_FOREACH(client, list, cl_link) {
if (client->cl_id.ci_id == id->ci_id) {
/*
* Move this client to the front of the LRU
* list.
*/
TAILQ_REMOVE(&svc_rpc_gss_clients, client, cl_alllink);
TAILQ_INSERT_HEAD(&svc_rpc_gss_clients, client,
cl_alllink);
refcount_acquire(&client->cl_refs);
break;
}
}
sx_xunlock(&svc_rpc_gss_lock);
return (client);
}
static struct svc_rpc_gss_client *
svc_rpc_gss_create_client(void)
{
struct svc_rpc_gss_client *client;
struct svc_rpc_gss_client_list *list;
unsigned long hostid;
rpc_gss_log_debug("in svc_rpc_gss_create_client()");
client = mem_alloc(sizeof(struct svc_rpc_gss_client));
memset(client, 0, sizeof(struct svc_rpc_gss_client));
refcount_init(&client->cl_refs, 1);
sx_init(&client->cl_lock, "GSS-client");
getcredhostid(curthread->td_ucred, &hostid);
client->cl_id.ci_hostid = hostid;
client->cl_id.ci_boottime = boottime.tv_sec;
client->cl_id.ci_id = svc_rpc_gss_next_clientid++;
list = &svc_rpc_gss_client_hash[client->cl_id.ci_id % CLIENT_HASH_SIZE];
sx_xlock(&svc_rpc_gss_lock);
TAILQ_INSERT_HEAD(list, client, cl_link);
TAILQ_INSERT_HEAD(&svc_rpc_gss_clients, client, cl_alllink);
svc_rpc_gss_client_count++;
sx_xunlock(&svc_rpc_gss_lock);
/*
* Start the client off with a short expiration time. We will
* try to get a saner value from the client creds later.
*/
client->cl_state = CLIENT_NEW;
client->cl_locked = FALSE;
client->cl_expiration = time_uptime + 5*60;
return (client);
}
static void
svc_rpc_gss_destroy_client(struct svc_rpc_gss_client *client)
{
OM_uint32 min_stat;
rpc_gss_log_debug("in svc_rpc_gss_destroy_client()");
if (client->cl_ctx)
gss_delete_sec_context(&min_stat,
&client->cl_ctx, GSS_C_NO_BUFFER);
if (client->cl_cname)
gss_release_name(&min_stat, &client->cl_cname);
if (client->cl_rawcred.client_principal)
mem_free(client->cl_rawcred.client_principal,
sizeof(*client->cl_rawcred.client_principal)
+ client->cl_rawcred.client_principal->len);
if (client->cl_cred)
crfree(client->cl_cred);
sx_destroy(&client->cl_lock);
mem_free(client, sizeof(*client));
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
}
/*
* Drop a reference to a client and free it if that was the last reference.
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 void
svc_rpc_gss_release_client(struct svc_rpc_gss_client *client)
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 (!refcount_release(&client->cl_refs))
return;
svc_rpc_gss_destroy_client(client);
}
/*
* Remove a client from our global lists.
* Must be called with svc_rpc_gss_lock held.
*/
static void
svc_rpc_gss_forget_client_locked(struct svc_rpc_gss_client *client)
{
struct svc_rpc_gss_client_list *list;
sx_assert(&svc_rpc_gss_lock, SX_XLOCKED);
list = &svc_rpc_gss_client_hash[client->cl_id.ci_id % CLIENT_HASH_SIZE];
TAILQ_REMOVE(list, client, cl_link);
TAILQ_REMOVE(&svc_rpc_gss_clients, client, cl_alllink);
svc_rpc_gss_client_count--;
}
/*
* Remove a client from our global lists and free it if we can.
*/
static void
svc_rpc_gss_forget_client(struct svc_rpc_gss_client *client)
{
struct svc_rpc_gss_client_list *list;
struct svc_rpc_gss_client *tclient;
list = &svc_rpc_gss_client_hash[client->cl_id.ci_id % CLIENT_HASH_SIZE];
sx_xlock(&svc_rpc_gss_lock);
TAILQ_FOREACH(tclient, list, cl_link) {
/*
* Make sure this client has not already been removed
* from the lists by svc_rpc_gss_forget_client() or
* svc_rpc_gss_forget_client_locked().
*/
if (client == tclient) {
svc_rpc_gss_forget_client_locked(client);
sx_xunlock(&svc_rpc_gss_lock);
svc_rpc_gss_release_client(client);
return;
}
}
sx_xunlock(&svc_rpc_gss_lock);
}
static void
svc_rpc_gss_timeout_clients(void)
{
struct svc_rpc_gss_client *client;
time_t now = time_uptime;
rpc_gss_log_debug("in svc_rpc_gss_timeout_clients()");
/*
* First enforce the max client limit. We keep
* svc_rpc_gss_clients in LRU order.
*/
sx_xlock(&svc_rpc_gss_lock);
client = TAILQ_LAST(&svc_rpc_gss_clients, svc_rpc_gss_client_list);
while (svc_rpc_gss_client_count > CLIENT_MAX && client != NULL) {
svc_rpc_gss_forget_client_locked(client);
sx_xunlock(&svc_rpc_gss_lock);
svc_rpc_gss_release_client(client);
sx_xlock(&svc_rpc_gss_lock);
client = TAILQ_LAST(&svc_rpc_gss_clients,
svc_rpc_gss_client_list);
}
again:
TAILQ_FOREACH(client, &svc_rpc_gss_clients, cl_alllink) {
if (client->cl_state == CLIENT_STALE
|| now > client->cl_expiration) {
svc_rpc_gss_forget_client_locked(client);
sx_xunlock(&svc_rpc_gss_lock);
rpc_gss_log_debug("expiring client %p", client);
svc_rpc_gss_release_client(client);
sx_xlock(&svc_rpc_gss_lock);
goto again;
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
}
}
sx_xunlock(&svc_rpc_gss_lock);
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
}
#ifdef DEBUG
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
/*
* OID<->string routines. These are uuuuugly.
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 OM_uint32
gss_oid_to_str(OM_uint32 *minor_status, gss_OID oid, gss_buffer_t oid_str)
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
{
char numstr[128];
unsigned long number;
int numshift;
size_t string_length;
size_t i;
unsigned char *cp;
char *bp;
/* Decoded according to krb5/gssapi_krb5.c */
/* First determine the size of the string */
string_length = 0;
number = 0;
numshift = 0;
cp = (unsigned char *) oid->elements;
number = (unsigned long) cp[0];
sprintf(numstr, "%ld ", number/40);
string_length += strlen(numstr);
sprintf(numstr, "%ld ", number%40);
string_length += strlen(numstr);
for (i=1; i<oid->length; i++) {
if ( (size_t) (numshift+7) < (sizeof(unsigned long)*8)) {
number = (number << 7) | (cp[i] & 0x7f);
numshift += 7;
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
}
else {
*minor_status = 0;
return(GSS_S_FAILURE);
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 ((cp[i] & 0x80) == 0) {
sprintf(numstr, "%ld ", number);
string_length += strlen(numstr);
number = 0;
numshift = 0;
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 we get here, we've calculated the length of "n n n ... n ". Add 4
* here for "{ " and "}\0".
*/
string_length += 4;
if ((bp = (char *) mem_alloc(string_length))) {
strcpy(bp, "{ ");
number = (unsigned long) cp[0];
sprintf(numstr, "%ld ", number/40);
strcat(bp, numstr);
sprintf(numstr, "%ld ", number%40);
strcat(bp, numstr);
number = 0;
cp = (unsigned char *) oid->elements;
for (i=1; i<oid->length; i++) {
number = (number << 7) | (cp[i] & 0x7f);
if ((cp[i] & 0x80) == 0) {
sprintf(numstr, "%ld ", number);
strcat(bp, numstr);
number = 0;
}
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
}
strcat(bp, "}");
oid_str->length = strlen(bp)+1;
oid_str->value = (void *) bp;
*minor_status = 0;
return(GSS_S_COMPLETE);
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
}
*minor_status = 0;
return(GSS_S_FAILURE);
}
#endif
static void
svc_rpc_gss_build_ucred(struct svc_rpc_gss_client *client,
const gss_name_t name)
{
OM_uint32 maj_stat, min_stat;
rpc_gss_ucred_t *uc = &client->cl_ucred;
int numgroups;
uc->uid = 65534;
uc->gid = 65534;
uc->gidlist = client->cl_gid_storage;
numgroups = RPCAUTH_UNIXGIDS;
maj_stat = gss_pname_to_unix_cred(&min_stat, name, client->cl_mech,
&uc->uid, &uc->gid, &numgroups, &uc->gidlist[0]);
if (GSS_ERROR(maj_stat))
uc->gidlen = 0;
else
uc->gidlen = numgroups;
}
static void
svc_rpc_gss_set_flavor(struct svc_rpc_gss_client *client)
{
static gss_OID_desc krb5_mech_oid =
{9, (void *) "\x2a\x86\x48\x86\xf7\x12\x01\x02\x02" };
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
/*
* Attempt to translate mech type and service into a
* 'pseudo flavor'. Hardwire in krb5 support for now.
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 (kgss_oid_equal(client->cl_mech, &krb5_mech_oid)) {
switch (client->cl_rawcred.service) {
case rpc_gss_svc_default:
case rpc_gss_svc_none:
client->cl_rpcflavor = RPCSEC_GSS_KRB5;
break;
case rpc_gss_svc_integrity:
client->cl_rpcflavor = RPCSEC_GSS_KRB5I;
break;
case rpc_gss_svc_privacy:
client->cl_rpcflavor = RPCSEC_GSS_KRB5P;
break;
}
} else {
client->cl_rpcflavor = RPCSEC_GSS;
}
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
svc_rpc_gss_accept_sec_context(struct svc_rpc_gss_client *client,
struct svc_req *rqst,
struct rpc_gss_init_res *gr,
struct rpc_gss_cred *gc)
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
{
gss_buffer_desc recv_tok;
gss_OID mech;
OM_uint32 maj_stat = 0, min_stat = 0, ret_flags;
OM_uint32 cred_lifetime;
struct svc_rpc_gss_svc_name *sname;
rpc_gss_log_debug("in svc_rpc_gss_accept_context()");
/* Deserialize arguments. */
memset(&recv_tok, 0, sizeof(recv_tok));
if (!svc_getargs(rqst,
(xdrproc_t) xdr_gss_buffer_desc,
(caddr_t) &recv_tok)) {
client->cl_state = CLIENT_STALE;
return (FALSE);
}
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
/*
* First time round, try all the server names we have until
* one matches. Afterwards, stick with that one.
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
*/
sx_xlock(&svc_rpc_gss_lock);
if (!client->cl_sname) {
SLIST_FOREACH(sname, &svc_rpc_gss_svc_names, sn_link) {
if (sname->sn_program == rqst->rq_prog
&& sname->sn_version == rqst->rq_vers) {
retry:
gr->gr_major = gss_accept_sec_context(
&gr->gr_minor,
&client->cl_ctx,
sname->sn_cred,
&recv_tok,
GSS_C_NO_CHANNEL_BINDINGS,
&client->cl_cname,
&mech,
&gr->gr_token,
&ret_flags,
&cred_lifetime,
&client->cl_creds);
if (gr->gr_major ==
GSS_S_CREDENTIALS_EXPIRED) {
/*
* Either our creds really did
* expire or gssd was
* restarted.
*/
if (rpc_gss_acquire_svc_cred(sname))
goto retry;
}
client->cl_sname = sname;
break;
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 (!sname) {
xdr_free((xdrproc_t) xdr_gss_buffer_desc,
(char *) &recv_tok);
sx_xunlock(&svc_rpc_gss_lock);
return (FALSE);
}
} else {
gr->gr_major = gss_accept_sec_context(
&gr->gr_minor,
&client->cl_ctx,
client->cl_sname->sn_cred,
&recv_tok,
GSS_C_NO_CHANNEL_BINDINGS,
&client->cl_cname,
&mech,
&gr->gr_token,
&ret_flags,
&cred_lifetime,
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
}
sx_xunlock(&svc_rpc_gss_lock);
xdr_free((xdrproc_t) xdr_gss_buffer_desc, (char *) &recv_tok);
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 we get an error from gss_accept_sec_context, send the
* reply anyway so that the client gets a chance to see what
* is wrong.
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 (gr->gr_major != GSS_S_COMPLETE &&
gr->gr_major != GSS_S_CONTINUE_NEEDED) {
rpc_gss_log_status("accept_sec_context", client->cl_mech,
gr->gr_major, gr->gr_minor);
client->cl_state = CLIENT_STALE;
return (TRUE);
}
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
gr->gr_handle.value = &client->cl_id;
gr->gr_handle.length = sizeof(client->cl_id);
gr->gr_win = SVC_RPC_GSS_SEQWINDOW;
/* Save client info. */
client->cl_mech = mech;
client->cl_qop = GSS_C_QOP_DEFAULT;
client->cl_done_callback = FALSE;
if (gr->gr_major == GSS_S_COMPLETE) {
gss_buffer_desc export_name;
/*
* Change client expiration time to be near when the
* client creds expire (or 24 hours if we can't figure
* that out).
*/
if (cred_lifetime == GSS_C_INDEFINITE)
cred_lifetime = time_uptime + 24*60*60;
client->cl_expiration = time_uptime + cred_lifetime;
/*
* Fill in cred details in the rawcred structure.
*/
client->cl_rawcred.version = RPCSEC_GSS_VERSION;
rpc_gss_oid_to_mech(mech, &client->cl_rawcred.mechanism);
maj_stat = gss_export_name(&min_stat, client->cl_cname,
&export_name);
if (maj_stat != GSS_S_COMPLETE) {
rpc_gss_log_status("gss_export_name", client->cl_mech,
maj_stat, min_stat);
return (FALSE);
}
client->cl_rawcred.client_principal =
mem_alloc(sizeof(*client->cl_rawcred.client_principal)
+ export_name.length);
client->cl_rawcred.client_principal->len = export_name.length;
memcpy(client->cl_rawcred.client_principal->name,
export_name.value, export_name.length);
gss_release_buffer(&min_stat, &export_name);
client->cl_rawcred.svc_principal =
client->cl_sname->sn_principal;
client->cl_rawcred.service = gc->gc_svc;
/*
* Use gss_pname_to_uid to map to unix creds. For
* kerberos5, this uses krb5_aname_to_localname.
*/
svc_rpc_gss_build_ucred(client, client->cl_cname);
svc_rpc_gss_set_flavor(client);
gss_release_name(&min_stat, &client->cl_cname);
#ifdef DEBUG
{
gss_buffer_desc mechname;
gss_oid_to_str(&min_stat, mech, &mechname);
rpc_gss_log_debug("accepted context for %s with "
"<mech %.*s, qop %d, svc %d>",
client->cl_rawcred.client_principal->name,
mechname.length, (char *)mechname.value,
client->cl_qop, client->cl_rawcred.service);
gss_release_buffer(&min_stat, &mechname);
}
#endif /* DEBUG */
}
return (TRUE);
}
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
svc_rpc_gss_validate(struct svc_rpc_gss_client *client, struct rpc_msg *msg,
gss_qop_t *qop)
{
struct opaque_auth *oa;
gss_buffer_desc rpcbuf, checksum;
OM_uint32 maj_stat, min_stat;
gss_qop_t qop_state;
int32_t rpchdr[128 / sizeof(int32_t)];
int32_t *buf;
rpc_gss_log_debug("in svc_rpc_gss_validate()");
memset(rpchdr, 0, sizeof(rpchdr));
/* Reconstruct RPC header for signing (from xdr_callmsg). */
buf = rpchdr;
IXDR_PUT_LONG(buf, msg->rm_xid);
IXDR_PUT_ENUM(buf, msg->rm_direction);
IXDR_PUT_LONG(buf, msg->rm_call.cb_rpcvers);
IXDR_PUT_LONG(buf, msg->rm_call.cb_prog);
IXDR_PUT_LONG(buf, msg->rm_call.cb_vers);
IXDR_PUT_LONG(buf, msg->rm_call.cb_proc);
oa = &msg->rm_call.cb_cred;
IXDR_PUT_ENUM(buf, oa->oa_flavor);
IXDR_PUT_LONG(buf, oa->oa_length);
if (oa->oa_length) {
memcpy((caddr_t)buf, oa->oa_base, oa->oa_length);
buf += RNDUP(oa->oa_length) / sizeof(int32_t);
}
rpcbuf.value = rpchdr;
rpcbuf.length = (u_char *)buf - (u_char *)rpchdr;
checksum.value = msg->rm_call.cb_verf.oa_base;
checksum.length = msg->rm_call.cb_verf.oa_length;
maj_stat = gss_verify_mic(&min_stat, client->cl_ctx, &rpcbuf, &checksum,
&qop_state);
if (maj_stat != GSS_S_COMPLETE) {
rpc_gss_log_status("gss_verify_mic", client->cl_mech,
maj_stat, min_stat);
client->cl_state = CLIENT_STALE;
return (FALSE);
}
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
*qop = qop_state;
return (TRUE);
}
static bool_t
svc_rpc_gss_nextverf(struct svc_rpc_gss_client *client,
struct svc_req *rqst, u_int seq)
{
gss_buffer_desc signbuf;
gss_buffer_desc mic;
OM_uint32 maj_stat, min_stat;
uint32_t nseq;
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
rpc_gss_log_debug("in svc_rpc_gss_nextverf()");
nseq = htonl(seq);
signbuf.value = &nseq;
signbuf.length = sizeof(nseq);
maj_stat = gss_get_mic(&min_stat, client->cl_ctx, client->cl_qop,
&signbuf, &mic);
if (maj_stat != GSS_S_COMPLETE) {
rpc_gss_log_status("gss_get_mic", client->cl_mech, maj_stat, min_stat);
client->cl_state = CLIENT_STALE;
return (FALSE);
}
KASSERT(mic.length <= MAX_AUTH_BYTES,
("MIC too large for RPCSEC_GSS"));
rqst->rq_verf.oa_flavor = RPCSEC_GSS;
rqst->rq_verf.oa_length = mic.length;
bcopy(mic.value, rqst->rq_verf.oa_base, mic.length);
gss_release_buffer(&min_stat, &mic);
return (TRUE);
}
static bool_t
svc_rpc_gss_callback(struct svc_rpc_gss_client *client, struct svc_req *rqst)
{
struct svc_rpc_gss_callback *scb;
rpc_gss_lock_t lock;
void *cookie;
bool_t cb_res;
bool_t result;
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
/*
* See if we have a callback for this guy.
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
*/
result = TRUE;
SLIST_FOREACH(scb, &svc_rpc_gss_callbacks, cb_link) {
if (scb->cb_callback.program == rqst->rq_prog
&& scb->cb_callback.version == rqst->rq_vers) {
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
/*
* This one matches. Call the callback and see
* if it wants to veto or something.
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
*/
lock.locked = FALSE;
lock.raw_cred = &client->cl_rawcred;
cb_res = scb->cb_callback.callback(rqst,
client->cl_creds,
client->cl_ctx,
&lock,
&cookie);
if (!cb_res) {
client->cl_state = CLIENT_STALE;
result = FALSE;
break;
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
}
/*
* The callback accepted the connection - it
* is responsible for freeing client->cl_creds
* now.
*/
client->cl_creds = GSS_C_NO_CREDENTIAL;
client->cl_locked = lock.locked;
client->cl_cookie = cookie;
return (TRUE);
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
}
}
/*
* Either no callback exists for this program/version or one
* of the callbacks rejected the connection. We just need to
* clean up the delegated client creds, if any.
*/
if (client->cl_creds) {
OM_uint32 min_ver;
gss_release_cred(&min_ver, &client->cl_creds);
}
return (result);
}
static bool_t
svc_rpc_gss_check_replay(struct svc_rpc_gss_client *client, uint32_t seq)
{
u_int32_t offset;
int word, bit;
bool_t result;
sx_xlock(&client->cl_lock);
if (seq <= client->cl_seqlast) {
/*
* The request sequence number is less than
* the largest we have seen so far. If it is
* outside the window or if we have seen a
* request with this sequence before, silently
* discard it.
*/
offset = client->cl_seqlast - seq;
if (offset >= SVC_RPC_GSS_SEQWINDOW) {
result = FALSE;
goto out;
}
word = offset / 32;
bit = offset % 32;
if (client->cl_seqmask[word] & (1 << bit)) {
result = FALSE;
goto out;
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
}
}
result = TRUE;
out:
sx_xunlock(&client->cl_lock);
return (result);
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 void
svc_rpc_gss_update_seq(struct svc_rpc_gss_client *client, uint32_t seq)
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
{
int offset, i, word, bit;
uint32_t carry, newcarry;
sx_xlock(&client->cl_lock);
if (seq > client->cl_seqlast) {
/*
* This request has a sequence number greater
* than any we have seen so far. Advance the
* seq window and set bit zero of the window
* (which corresponds to the new sequence
* number)
*/
offset = seq - client->cl_seqlast;
while (offset > 32) {
for (i = (SVC_RPC_GSS_SEQWINDOW / 32) - 1;
i > 0; i--) {
client->cl_seqmask[i] = client->cl_seqmask[i-1];
}
client->cl_seqmask[0] = 0;
offset -= 32;
}
carry = 0;
for (i = 0; i < SVC_RPC_GSS_SEQWINDOW / 32; i++) {
newcarry = client->cl_seqmask[i] >> (32 - offset);
client->cl_seqmask[i] =
(client->cl_seqmask[i] << offset) | carry;
carry = newcarry;
}
client->cl_seqmask[0] |= 1;
client->cl_seqlast = seq;
} else {
offset = client->cl_seqlast - seq;
word = offset / 32;
bit = offset % 32;
client->cl_seqmask[word] |= (1 << bit);
}
sx_xunlock(&client->cl_lock);
}
enum auth_stat
svc_rpc_gss(struct svc_req *rqst, struct rpc_msg *msg)
{
OM_uint32 min_stat;
XDR xdrs;
struct svc_rpc_gss_cookedcred *cc;
struct svc_rpc_gss_client *client;
struct rpc_gss_cred gc;
struct rpc_gss_init_res gr;
gss_qop_t qop;
int call_stat;
enum auth_stat result;
rpc_gss_log_debug("in svc_rpc_gss()");
/* Garbage collect old clients. */
svc_rpc_gss_timeout_clients();
/* Initialize reply. */
rqst->rq_verf = _null_auth;
/* Deserialize client credentials. */
if (rqst->rq_cred.oa_length <= 0)
return (AUTH_BADCRED);
memset(&gc, 0, sizeof(gc));
xdrmem_create(&xdrs, rqst->rq_cred.oa_base,
rqst->rq_cred.oa_length, XDR_DECODE);
if (!xdr_rpc_gss_cred(&xdrs, &gc)) {
XDR_DESTROY(&xdrs);
return (AUTH_BADCRED);
}
XDR_DESTROY(&xdrs);
client = NULL;
/* Check version. */
if (gc.gc_version != RPCSEC_GSS_VERSION) {
result = AUTH_BADCRED;
goto out;
}
/* Check the proc and find the client (or create it) */
if (gc.gc_proc == RPCSEC_GSS_INIT) {
if (gc.gc_handle.length != 0) {
result = AUTH_BADCRED;
goto out;
}
client = svc_rpc_gss_create_client();
refcount_acquire(&client->cl_refs);
} else {
struct svc_rpc_gss_clientid *p;
if (gc.gc_handle.length != sizeof(*p)) {
result = AUTH_BADCRED;
goto out;
}
p = gc.gc_handle.value;
client = svc_rpc_gss_find_client(p);
if (!client) {
/*
* Can't find the client - we may have
* destroyed it - tell the other side to
* re-authenticate.
*/
result = RPCSEC_GSS_CREDPROBLEM;
goto out;
}
}
cc = rqst->rq_clntcred;
cc->cc_client = client;
cc->cc_service = gc.gc_svc;
cc->cc_seq = gc.gc_seq;
/*
* The service and sequence number must be ignored for
* RPCSEC_GSS_INIT and RPCSEC_GSS_CONTINUE_INIT.
*/
if (gc.gc_proc != RPCSEC_GSS_INIT
&& gc.gc_proc != RPCSEC_GSS_CONTINUE_INIT) {
/*
* Check for sequence number overflow.
*/
if (gc.gc_seq >= MAXSEQ) {
result = RPCSEC_GSS_CTXPROBLEM;
goto out;
}
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
/*
* Check for valid service.
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 (gc.gc_svc != rpc_gss_svc_none &&
gc.gc_svc != rpc_gss_svc_integrity &&
gc.gc_svc != rpc_gss_svc_privacy) {
result = AUTH_BADCRED;
goto out;
}
}
/* Handle RPCSEC_GSS control procedure. */
switch (gc.gc_proc) {
case RPCSEC_GSS_INIT:
case RPCSEC_GSS_CONTINUE_INIT:
if (rqst->rq_proc != NULLPROC) {
result = AUTH_REJECTEDCRED;
break;
}
memset(&gr, 0, sizeof(gr));
if (!svc_rpc_gss_accept_sec_context(client, rqst, &gr, &gc)) {
result = AUTH_REJECTEDCRED;
break;
}
if (gr.gr_major == GSS_S_COMPLETE) {
/*
* We borrow the space for the call verf to
* pack our reply verf.
*/
rqst->rq_verf = msg->rm_call.cb_verf;
if (!svc_rpc_gss_nextverf(client, rqst, gr.gr_win)) {
result = AUTH_REJECTEDCRED;
break;
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
}
} else {
rqst->rq_verf = _null_auth;
}
call_stat = svc_sendreply(rqst,
(xdrproc_t) xdr_rpc_gss_init_res,
(caddr_t) &gr);
gss_release_buffer(&min_stat, &gr.gr_token);
if (!call_stat) {
result = AUTH_FAILED;
break;
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 (gr.gr_major == GSS_S_COMPLETE)
client->cl_state = CLIENT_ESTABLISHED;
result = RPCSEC_GSS_NODISPATCH;
break;
case RPCSEC_GSS_DATA:
case RPCSEC_GSS_DESTROY:
if (!svc_rpc_gss_check_replay(client, gc.gc_seq)) {
result = RPCSEC_GSS_NODISPATCH;
break;
}
if (!svc_rpc_gss_validate(client, msg, &qop)) {
result = RPCSEC_GSS_CREDPROBLEM;
break;
}
/*
* We borrow the space for the call verf to pack our
* reply verf.
*/
rqst->rq_verf = msg->rm_call.cb_verf;
if (!svc_rpc_gss_nextverf(client, rqst, gc.gc_seq)) {
result = RPCSEC_GSS_CTXPROBLEM;
break;
}
svc_rpc_gss_update_seq(client, gc.gc_seq);
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
/*
* Change the SVCAUTH ops on the request to point at
* our own code so that we can unwrap the arguments
* and wrap the result. The caller will re-set this on
* every request to point to a set of null wrap/unwrap
* methods. Acquire an extra reference to the client
* which will be released by svc_rpc_gss_release()
* after the request has finished processing.
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
*/
refcount_acquire(&client->cl_refs);
rqst->rq_auth.svc_ah_ops = &svc_auth_gss_ops;
rqst->rq_auth.svc_ah_private = cc;
if (gc.gc_proc == RPCSEC_GSS_DATA) {
/*
* We might be ready to do a callback to the server to
* see if it wants to accept/reject the connection.
*/
sx_xlock(&client->cl_lock);
if (!client->cl_done_callback) {
client->cl_done_callback = TRUE;
client->cl_qop = qop;
client->cl_rawcred.qop = _rpc_gss_num_to_qop(
client->cl_rawcred.mechanism, qop);
if (!svc_rpc_gss_callback(client, rqst)) {
result = AUTH_REJECTEDCRED;
sx_xunlock(&client->cl_lock);
break;
}
}
sx_xunlock(&client->cl_lock);
/*
* If the server has locked this client to a
* particular service+qop pair, enforce that
* restriction now.
*/
if (client->cl_locked) {
if (client->cl_rawcred.service != gc.gc_svc) {
result = AUTH_FAILED;
break;
} else if (client->cl_qop != qop) {
result = AUTH_BADVERF;
break;
}
}
/*
* If the qop changed, look up the new qop
* name for rawcred.
*/
if (client->cl_qop != qop) {
client->cl_qop = qop;
client->cl_rawcred.qop = _rpc_gss_num_to_qop(
client->cl_rawcred.mechanism, qop);
}
/*
* Make sure we use the right service value
* for unwrap/wrap.
*/
if (client->cl_rawcred.service != gc.gc_svc) {
client->cl_rawcred.service = gc.gc_svc;
svc_rpc_gss_set_flavor(client);
}
result = AUTH_OK;
} else {
if (rqst->rq_proc != NULLPROC) {
result = AUTH_REJECTEDCRED;
break;
}
call_stat = svc_sendreply(rqst,
(xdrproc_t) xdr_void, (caddr_t) NULL);
if (!call_stat) {
result = AUTH_FAILED;
break;
}
svc_rpc_gss_forget_client(client);
result = RPCSEC_GSS_NODISPATCH;
break;
}
break;
default:
result = AUTH_BADCRED;
break;
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
}
out:
if (client)
svc_rpc_gss_release_client(client);
xdr_free((xdrproc_t) xdr_rpc_gss_cred, (char *) &gc);
return (result);
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
svc_rpc_gss_wrap(SVCAUTH *auth, struct mbuf **mp)
{
struct svc_rpc_gss_cookedcred *cc;
struct svc_rpc_gss_client *client;
rpc_gss_log_debug("in svc_rpc_gss_wrap()");
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
cc = (struct svc_rpc_gss_cookedcred *) auth->svc_ah_private;
client = cc->cc_client;
if (client->cl_state != CLIENT_ESTABLISHED
|| cc->cc_service == rpc_gss_svc_none || *mp == NULL) {
return (TRUE);
}
return (xdr_rpc_gss_wrap_data(mp,
client->cl_ctx, client->cl_qop,
cc->cc_service, cc->cc_seq));
}
static bool_t
svc_rpc_gss_unwrap(SVCAUTH *auth, struct mbuf **mp)
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
{
struct svc_rpc_gss_cookedcred *cc;
struct svc_rpc_gss_client *client;
rpc_gss_log_debug("in svc_rpc_gss_unwrap()");
cc = (struct svc_rpc_gss_cookedcred *) auth->svc_ah_private;
client = cc->cc_client;
if (client->cl_state != CLIENT_ESTABLISHED
|| cc->cc_service == rpc_gss_svc_none) {
return (TRUE);
}
return (xdr_rpc_gss_unwrap_data(mp,
client->cl_ctx, client->cl_qop,
cc->cc_service, cc->cc_seq));
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 void
svc_rpc_gss_release(SVCAUTH *auth)
{
struct svc_rpc_gss_cookedcred *cc;
struct svc_rpc_gss_client *client;
rpc_gss_log_debug("in svc_rpc_gss_release()");
cc = (struct svc_rpc_gss_cookedcred *) auth->svc_ah_private;
client = cc->cc_client;
svc_rpc_gss_release_client(client);
}