freebsd-dev/sys/rpc/rpc_generic.c

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

989 lines
21 KiB
C
Raw Normal View History

Add the new kernel-mode NFS Lock Manager. To use it instead of the user-mode lock manager, build a kernel with the NFSLOCKD option and add '-k' to 'rpc_lockd_flags' in rc.conf. Highlights include: * Thread-safe kernel RPC client - many threads can use the same RPC client handle safely with replies being de-multiplexed at the socket upcall (typically driven directly by the NIC interrupt) and handed off to whichever thread matches the reply. For UDP sockets, many RPC clients can share the same socket. This allows the use of a single privileged UDP port number to talk to an arbitrary number of remote hosts. * Single-threaded kernel RPC server. Adding support for multi-threaded server would be relatively straightforward and would follow approximately the Solaris KPI. A single thread should be sufficient for the NLM since it should rarely block in normal operation. * Kernel mode NLM server supporting cancel requests and granted callbacks. I've tested the NLM server reasonably extensively - it passes both my own tests and the NFS Connectathon locking tests running on Solaris, Mac OS X and Ubuntu Linux. * Userland NLM client supported. While the NLM server doesn't have support for the local NFS client's locking needs, it does have to field async replies and granted callbacks from remote NLMs that the local client has contacted. We relay these replies to the userland rpc.lockd over a local domain RPC socket. * Robust deadlock detection for the local lock manager. In particular it will detect deadlocks caused by a lock request that covers more than one blocking request. As required by the NLM protocol, all deadlock detection happens synchronously - a user is guaranteed that if a lock request isn't rejected immediately, the lock will eventually be granted. The old system allowed for a 'deferred deadlock' condition where a blocked lock request could wake up and find that some other deadlock-causing lock owner had beaten them to the lock. * Since both local and remote locks are managed by the same kernel locking code, local and remote processes can safely use file locks for mutual exclusion. Local processes have no fairness advantage compared to remote processes when contending to lock a region that has just been unlocked - the local lock manager enforces a strict first-come first-served model for both local and remote lockers. Sponsored by: Isilon Systems PR: 95247 107555 115524 116679 MFC after: 2 weeks
2008-03-26 15:23:12 +00:00
/* $NetBSD: rpc_generic.c,v 1.4 2000/09/28 09:07:04 kleink Exp $ */
/*-
* SPDX-License-Identifier: BSD-3-Clause
*
* Copyright (c) 2009, Sun Microsystems, Inc.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are met:
* - Redistributions of source code must retain the above copyright notice,
* this list of conditions and the following disclaimer.
* - Redistributions in binary form must reproduce the above copyright notice,
* this list of conditions and the following disclaimer in the documentation
* and/or other materials provided with the distribution.
* - Neither the name of Sun Microsystems, Inc. nor the names of its
* contributors may be used to endorse or promote products derived
* from this software without specific prior written permission.
Add the new kernel-mode NFS Lock Manager. To use it instead of the user-mode lock manager, build a kernel with the NFSLOCKD option and add '-k' to 'rpc_lockd_flags' in rc.conf. Highlights include: * Thread-safe kernel RPC client - many threads can use the same RPC client handle safely with replies being de-multiplexed at the socket upcall (typically driven directly by the NIC interrupt) and handed off to whichever thread matches the reply. For UDP sockets, many RPC clients can share the same socket. This allows the use of a single privileged UDP port number to talk to an arbitrary number of remote hosts. * Single-threaded kernel RPC server. Adding support for multi-threaded server would be relatively straightforward and would follow approximately the Solaris KPI. A single thread should be sufficient for the NLM since it should rarely block in normal operation. * Kernel mode NLM server supporting cancel requests and granted callbacks. I've tested the NLM server reasonably extensively - it passes both my own tests and the NFS Connectathon locking tests running on Solaris, Mac OS X and Ubuntu Linux. * Userland NLM client supported. While the NLM server doesn't have support for the local NFS client's locking needs, it does have to field async replies and granted callbacks from remote NLMs that the local client has contacted. We relay these replies to the userland rpc.lockd over a local domain RPC socket. * Robust deadlock detection for the local lock manager. In particular it will detect deadlocks caused by a lock request that covers more than one blocking request. As required by the NLM protocol, all deadlock detection happens synchronously - a user is guaranteed that if a lock request isn't rejected immediately, the lock will eventually be granted. The old system allowed for a 'deferred deadlock' condition where a blocked lock request could wake up and find that some other deadlock-causing lock owner had beaten them to the lock. * Since both local and remote locks are managed by the same kernel locking code, local and remote processes can safely use file locks for mutual exclusion. Local processes have no fairness advantage compared to remote processes when contending to lock a region that has just been unlocked - the local lock manager enforces a strict first-come first-served model for both local and remote lockers. Sponsored by: Isilon Systems PR: 95247 107555 115524 116679 MFC after: 2 weeks
2008-03-26 15:23:12 +00:00
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
* AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
Add the new kernel-mode NFS Lock Manager. To use it instead of the user-mode lock manager, build a kernel with the NFSLOCKD option and add '-k' to 'rpc_lockd_flags' in rc.conf. Highlights include: * Thread-safe kernel RPC client - many threads can use the same RPC client handle safely with replies being de-multiplexed at the socket upcall (typically driven directly by the NIC interrupt) and handed off to whichever thread matches the reply. For UDP sockets, many RPC clients can share the same socket. This allows the use of a single privileged UDP port number to talk to an arbitrary number of remote hosts. * Single-threaded kernel RPC server. Adding support for multi-threaded server would be relatively straightforward and would follow approximately the Solaris KPI. A single thread should be sufficient for the NLM since it should rarely block in normal operation. * Kernel mode NLM server supporting cancel requests and granted callbacks. I've tested the NLM server reasonably extensively - it passes both my own tests and the NFS Connectathon locking tests running on Solaris, Mac OS X and Ubuntu Linux. * Userland NLM client supported. While the NLM server doesn't have support for the local NFS client's locking needs, it does have to field async replies and granted callbacks from remote NLMs that the local client has contacted. We relay these replies to the userland rpc.lockd over a local domain RPC socket. * Robust deadlock detection for the local lock manager. In particular it will detect deadlocks caused by a lock request that covers more than one blocking request. As required by the NLM protocol, all deadlock detection happens synchronously - a user is guaranteed that if a lock request isn't rejected immediately, the lock will eventually be granted. The old system allowed for a 'deferred deadlock' condition where a blocked lock request could wake up and find that some other deadlock-causing lock owner had beaten them to the lock. * Since both local and remote locks are managed by the same kernel locking code, local and remote processes can safely use file locks for mutual exclusion. Local processes have no fairness advantage compared to remote processes when contending to lock a region that has just been unlocked - the local lock manager enforces a strict first-come first-served model for both local and remote lockers. Sponsored by: Isilon Systems PR: 95247 107555 115524 116679 MFC after: 2 weeks
2008-03-26 15:23:12 +00:00
*/
/*
* Copyright (c) 1986-1991 by Sun Microsystems Inc.
*/
/* #pragma ident "@(#)rpc_generic.c 1.17 94/04/24 SMI" */
#include <sys/cdefs.h>
/*
* rpc_generic.c, Miscl routines for RPC.
*
*/
#include "opt_inet6.h"
#include <sys/param.h>
#include <sys/kernel.h>
Add the new kernel-mode NFS Lock Manager. To use it instead of the user-mode lock manager, build a kernel with the NFSLOCKD option and add '-k' to 'rpc_lockd_flags' in rc.conf. Highlights include: * Thread-safe kernel RPC client - many threads can use the same RPC client handle safely with replies being de-multiplexed at the socket upcall (typically driven directly by the NIC interrupt) and handed off to whichever thread matches the reply. For UDP sockets, many RPC clients can share the same socket. This allows the use of a single privileged UDP port number to talk to an arbitrary number of remote hosts. * Single-threaded kernel RPC server. Adding support for multi-threaded server would be relatively straightforward and would follow approximately the Solaris KPI. A single thread should be sufficient for the NLM since it should rarely block in normal operation. * Kernel mode NLM server supporting cancel requests and granted callbacks. I've tested the NLM server reasonably extensively - it passes both my own tests and the NFS Connectathon locking tests running on Solaris, Mac OS X and Ubuntu Linux. * Userland NLM client supported. While the NLM server doesn't have support for the local NFS client's locking needs, it does have to field async replies and granted callbacks from remote NLMs that the local client has contacted. We relay these replies to the userland rpc.lockd over a local domain RPC socket. * Robust deadlock detection for the local lock manager. In particular it will detect deadlocks caused by a lock request that covers more than one blocking request. As required by the NLM protocol, all deadlock detection happens synchronously - a user is guaranteed that if a lock request isn't rejected immediately, the lock will eventually be granted. The old system allowed for a 'deferred deadlock' condition where a blocked lock request could wake up and find that some other deadlock-causing lock owner had beaten them to the lock. * Since both local and remote locks are managed by the same kernel locking code, local and remote processes can safely use file locks for mutual exclusion. Local processes have no fairness advantage compared to remote processes when contending to lock a region that has just been unlocked - the local lock manager enforces a strict first-come first-served model for both local and remote lockers. Sponsored by: Isilon Systems PR: 95247 107555 115524 116679 MFC after: 2 weeks
2008-03-26 15:23:12 +00:00
#include <sys/malloc.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/mbuf.h>
#include <sys/module.h>
Add the new kernel-mode NFS Lock Manager. To use it instead of the user-mode lock manager, build a kernel with the NFSLOCKD option and add '-k' to 'rpc_lockd_flags' in rc.conf. Highlights include: * Thread-safe kernel RPC client - many threads can use the same RPC client handle safely with replies being de-multiplexed at the socket upcall (typically driven directly by the NIC interrupt) and handed off to whichever thread matches the reply. For UDP sockets, many RPC clients can share the same socket. This allows the use of a single privileged UDP port number to talk to an arbitrary number of remote hosts. * Single-threaded kernel RPC server. Adding support for multi-threaded server would be relatively straightforward and would follow approximately the Solaris KPI. A single thread should be sufficient for the NLM since it should rarely block in normal operation. * Kernel mode NLM server supporting cancel requests and granted callbacks. I've tested the NLM server reasonably extensively - it passes both my own tests and the NFS Connectathon locking tests running on Solaris, Mac OS X and Ubuntu Linux. * Userland NLM client supported. While the NLM server doesn't have support for the local NFS client's locking needs, it does have to field async replies and granted callbacks from remote NLMs that the local client has contacted. We relay these replies to the userland rpc.lockd over a local domain RPC socket. * Robust deadlock detection for the local lock manager. In particular it will detect deadlocks caused by a lock request that covers more than one blocking request. As required by the NLM protocol, all deadlock detection happens synchronously - a user is guaranteed that if a lock request isn't rejected immediately, the lock will eventually be granted. The old system allowed for a 'deferred deadlock' condition where a blocked lock request could wake up and find that some other deadlock-causing lock owner had beaten them to the lock. * Since both local and remote locks are managed by the same kernel locking code, local and remote processes can safely use file locks for mutual exclusion. Local processes have no fairness advantage compared to remote processes when contending to lock a region that has just been unlocked - the local lock manager enforces a strict first-come first-served model for both local and remote lockers. Sponsored by: Isilon Systems PR: 95247 107555 115524 116679 MFC after: 2 weeks
2008-03-26 15:23:12 +00:00
#include <sys/proc.h>
#include <sys/protosw.h>
#include <sys/sbuf.h>
#include <sys/systm.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/syslog.h>
#include <net/vnet.h>
Add the new kernel-mode NFS Lock Manager. To use it instead of the user-mode lock manager, build a kernel with the NFSLOCKD option and add '-k' to 'rpc_lockd_flags' in rc.conf. Highlights include: * Thread-safe kernel RPC client - many threads can use the same RPC client handle safely with replies being de-multiplexed at the socket upcall (typically driven directly by the NIC interrupt) and handed off to whichever thread matches the reply. For UDP sockets, many RPC clients can share the same socket. This allows the use of a single privileged UDP port number to talk to an arbitrary number of remote hosts. * Single-threaded kernel RPC server. Adding support for multi-threaded server would be relatively straightforward and would follow approximately the Solaris KPI. A single thread should be sufficient for the NLM since it should rarely block in normal operation. * Kernel mode NLM server supporting cancel requests and granted callbacks. I've tested the NLM server reasonably extensively - it passes both my own tests and the NFS Connectathon locking tests running on Solaris, Mac OS X and Ubuntu Linux. * Userland NLM client supported. While the NLM server doesn't have support for the local NFS client's locking needs, it does have to field async replies and granted callbacks from remote NLMs that the local client has contacted. We relay these replies to the userland rpc.lockd over a local domain RPC socket. * Robust deadlock detection for the local lock manager. In particular it will detect deadlocks caused by a lock request that covers more than one blocking request. As required by the NLM protocol, all deadlock detection happens synchronously - a user is guaranteed that if a lock request isn't rejected immediately, the lock will eventually be granted. The old system allowed for a 'deferred deadlock' condition where a blocked lock request could wake up and find that some other deadlock-causing lock owner had beaten them to the lock. * Since both local and remote locks are managed by the same kernel locking code, local and remote processes can safely use file locks for mutual exclusion. Local processes have no fairness advantage compared to remote processes when contending to lock a region that has just been unlocked - the local lock manager enforces a strict first-come first-served model for both local and remote lockers. Sponsored by: Isilon Systems PR: 95247 107555 115524 116679 MFC after: 2 weeks
2008-03-26 15:23:12 +00:00
#include <rpc/rpc.h>
#include <rpc/nettype.h>
#include <rpc/rpcsec_gss.h>
Add TLS support to the kernel RPC. An internet draft titled "Towards Remote Procedure Call Encryption By Default" describes how TLS is to be used for Sun RPC, with NFS as an intended use case. This patch adds client and server support for this to the kernel RPC, using KERN_TLS and upcalls to daemons for the handshake, peer reset and other non-application data record cases. The upcalls to the daemons use three fields to uniquely identify the TCP connection. They are the time.tv_sec, time.tv_usec of the connection establshment, plus a 64bit sequence number. The time fields avoid problems with re-use of the sequence number after a daemon restart. For the server side, once a Null RPC with AUTH_TLS is received, kernel reception on the socket is blocked and an upcall to the rpctlssd(8) daemon is done to perform the TLS handshake. Upon completion, the completion status of the handshake is stored in xp_tls as flag bits and the reply to the Null RPC is sent. For the client, if CLSET_TLS has been set, a new TCP connection will send the Null RPC with AUTH_TLS to initiate the handshake. The client kernel RPC code will then block kernel I/O on the socket and do an upcall to the rpctlscd(8) daemon to perform the handshake. If the upcall is successful, ct_rcvstate will be maintained to indicate if/when an upcall is being done. If non-application data records are received, the code does an upcall to the appropriate daemon, which will do a SSL_read() of 0 length to handle the record(s). When the socket is being shut down, upcalls are done to the daemons, so that they can perform SSL_shutdown() calls to perform the "peer reset". The rpctlssd(8) and rpctlscd(8) daemons require a patched version of the openssl library and, as such, will not be committed to head at this time. Although the changes done by this patch are fairly numerous, there should be no semantics change to the kernel RPC at this time. A future commit to the NFS code will optionally enable use of TLS for NFS.
2020-08-22 03:57:55 +00:00
#include <rpc/rpcsec_tls.h>
Add the new kernel-mode NFS Lock Manager. To use it instead of the user-mode lock manager, build a kernel with the NFSLOCKD option and add '-k' to 'rpc_lockd_flags' in rc.conf. Highlights include: * Thread-safe kernel RPC client - many threads can use the same RPC client handle safely with replies being de-multiplexed at the socket upcall (typically driven directly by the NIC interrupt) and handed off to whichever thread matches the reply. For UDP sockets, many RPC clients can share the same socket. This allows the use of a single privileged UDP port number to talk to an arbitrary number of remote hosts. * Single-threaded kernel RPC server. Adding support for multi-threaded server would be relatively straightforward and would follow approximately the Solaris KPI. A single thread should be sufficient for the NLM since it should rarely block in normal operation. * Kernel mode NLM server supporting cancel requests and granted callbacks. I've tested the NLM server reasonably extensively - it passes both my own tests and the NFS Connectathon locking tests running on Solaris, Mac OS X and Ubuntu Linux. * Userland NLM client supported. While the NLM server doesn't have support for the local NFS client's locking needs, it does have to field async replies and granted callbacks from remote NLMs that the local client has contacted. We relay these replies to the userland rpc.lockd over a local domain RPC socket. * Robust deadlock detection for the local lock manager. In particular it will detect deadlocks caused by a lock request that covers more than one blocking request. As required by the NLM protocol, all deadlock detection happens synchronously - a user is guaranteed that if a lock request isn't rejected immediately, the lock will eventually be granted. The old system allowed for a 'deferred deadlock' condition where a blocked lock request could wake up and find that some other deadlock-causing lock owner had beaten them to the lock. * Since both local and remote locks are managed by the same kernel locking code, local and remote processes can safely use file locks for mutual exclusion. Local processes have no fairness advantage compared to remote processes when contending to lock a region that has just been unlocked - the local lock manager enforces a strict first-come first-served model for both local and remote lockers. Sponsored by: Isilon Systems PR: 95247 107555 115524 116679 MFC after: 2 weeks
2008-03-26 15:23:12 +00:00
#include <rpc/rpc_com.h>
Add TLS support to the kernel RPC. An internet draft titled "Towards Remote Procedure Call Encryption By Default" describes how TLS is to be used for Sun RPC, with NFS as an intended use case. This patch adds client and server support for this to the kernel RPC, using KERN_TLS and upcalls to daemons for the handshake, peer reset and other non-application data record cases. The upcalls to the daemons use three fields to uniquely identify the TCP connection. They are the time.tv_sec, time.tv_usec of the connection establshment, plus a 64bit sequence number. The time fields avoid problems with re-use of the sequence number after a daemon restart. For the server side, once a Null RPC with AUTH_TLS is received, kernel reception on the socket is blocked and an upcall to the rpctlssd(8) daemon is done to perform the TLS handshake. Upon completion, the completion status of the handshake is stored in xp_tls as flag bits and the reply to the Null RPC is sent. For the client, if CLSET_TLS has been set, a new TCP connection will send the Null RPC with AUTH_TLS to initiate the handshake. The client kernel RPC code will then block kernel I/O on the socket and do an upcall to the rpctlscd(8) daemon to perform the handshake. If the upcall is successful, ct_rcvstate will be maintained to indicate if/when an upcall is being done. If non-application data records are received, the code does an upcall to the appropriate daemon, which will do a SSL_read() of 0 length to handle the record(s). When the socket is being shut down, upcalls are done to the daemons, so that they can perform SSL_shutdown() calls to perform the "peer reset". The rpctlssd(8) and rpctlscd(8) daemons require a patched version of the openssl library and, as such, will not be committed to head at this time. Although the changes done by this patch are fairly numerous, there should be no semantics change to the kernel RPC at this time. A future commit to the NFS code will optionally enable use of TLS for NFS.
2020-08-22 03:57:55 +00:00
#include <rpc/krpc.h>
#include <vm/vm.h>
#include <vm/pmap.h>
#include <vm/vm_param.h>
extern u_long sb_max_adj; /* not defined in socketvar.h */
/* Provide an entry point hook for the rpcsec_gss module. */
struct rpc_gss_entries rpc_gss_entries;
Add the new kernel-mode NFS Lock Manager. To use it instead of the user-mode lock manager, build a kernel with the NFSLOCKD option and add '-k' to 'rpc_lockd_flags' in rc.conf. Highlights include: * Thread-safe kernel RPC client - many threads can use the same RPC client handle safely with replies being de-multiplexed at the socket upcall (typically driven directly by the NIC interrupt) and handed off to whichever thread matches the reply. For UDP sockets, many RPC clients can share the same socket. This allows the use of a single privileged UDP port number to talk to an arbitrary number of remote hosts. * Single-threaded kernel RPC server. Adding support for multi-threaded server would be relatively straightforward and would follow approximately the Solaris KPI. A single thread should be sufficient for the NLM since it should rarely block in normal operation. * Kernel mode NLM server supporting cancel requests and granted callbacks. I've tested the NLM server reasonably extensively - it passes both my own tests and the NFS Connectathon locking tests running on Solaris, Mac OS X and Ubuntu Linux. * Userland NLM client supported. While the NLM server doesn't have support for the local NFS client's locking needs, it does have to field async replies and granted callbacks from remote NLMs that the local client has contacted. We relay these replies to the userland rpc.lockd over a local domain RPC socket. * Robust deadlock detection for the local lock manager. In particular it will detect deadlocks caused by a lock request that covers more than one blocking request. As required by the NLM protocol, all deadlock detection happens synchronously - a user is guaranteed that if a lock request isn't rejected immediately, the lock will eventually be granted. The old system allowed for a 'deferred deadlock' condition where a blocked lock request could wake up and find that some other deadlock-causing lock owner had beaten them to the lock. * Since both local and remote locks are managed by the same kernel locking code, local and remote processes can safely use file locks for mutual exclusion. Local processes have no fairness advantage compared to remote processes when contending to lock a region that has just been unlocked - the local lock manager enforces a strict first-come first-served model for both local and remote lockers. Sponsored by: Isilon Systems PR: 95247 107555 115524 116679 MFC after: 2 weeks
2008-03-26 15:23:12 +00:00
struct handle {
NCONF_HANDLE *nhandle;
int nflag; /* Whether NETPATH or NETCONFIG */
int nettype;
};
static const struct _rpcnettype {
const char *name;
const int type;
} _rpctypelist[] = {
{ "netpath", _RPC_NETPATH },
{ "visible", _RPC_VISIBLE },
{ "circuit_v", _RPC_CIRCUIT_V },
{ "datagram_v", _RPC_DATAGRAM_V },
{ "circuit_n", _RPC_CIRCUIT_N },
{ "datagram_n", _RPC_DATAGRAM_N },
{ "tcp", _RPC_TCP },
{ "udp", _RPC_UDP },
{ 0, _RPC_NONE }
};
struct netid_af {
const char *netid;
int af;
int protocol;
};
static const struct netid_af na_cvt[] = {
{ "udp", AF_INET, IPPROTO_UDP },
{ "tcp", AF_INET, IPPROTO_TCP },
#ifdef INET6
{ "udp6", AF_INET6, IPPROTO_UDP },
{ "tcp6", AF_INET6, IPPROTO_TCP },
#endif
{ "local", AF_LOCAL, 0 }
};
struct rpc_createerr rpc_createerr;
/*
* Find the appropriate buffer size
*/
u_int
/*ARGSUSED*/
__rpc_get_t_size(int af, int proto, int size)
{
int defsize;
Add the new kernel-mode NFS Lock Manager. To use it instead of the user-mode lock manager, build a kernel with the NFSLOCKD option and add '-k' to 'rpc_lockd_flags' in rc.conf. Highlights include: * Thread-safe kernel RPC client - many threads can use the same RPC client handle safely with replies being de-multiplexed at the socket upcall (typically driven directly by the NIC interrupt) and handed off to whichever thread matches the reply. For UDP sockets, many RPC clients can share the same socket. This allows the use of a single privileged UDP port number to talk to an arbitrary number of remote hosts. * Single-threaded kernel RPC server. Adding support for multi-threaded server would be relatively straightforward and would follow approximately the Solaris KPI. A single thread should be sufficient for the NLM since it should rarely block in normal operation. * Kernel mode NLM server supporting cancel requests and granted callbacks. I've tested the NLM server reasonably extensively - it passes both my own tests and the NFS Connectathon locking tests running on Solaris, Mac OS X and Ubuntu Linux. * Userland NLM client supported. While the NLM server doesn't have support for the local NFS client's locking needs, it does have to field async replies and granted callbacks from remote NLMs that the local client has contacted. We relay these replies to the userland rpc.lockd over a local domain RPC socket. * Robust deadlock detection for the local lock manager. In particular it will detect deadlocks caused by a lock request that covers more than one blocking request. As required by the NLM protocol, all deadlock detection happens synchronously - a user is guaranteed that if a lock request isn't rejected immediately, the lock will eventually be granted. The old system allowed for a 'deferred deadlock' condition where a blocked lock request could wake up and find that some other deadlock-causing lock owner had beaten them to the lock. * Since both local and remote locks are managed by the same kernel locking code, local and remote processes can safely use file locks for mutual exclusion. Local processes have no fairness advantage compared to remote processes when contending to lock a region that has just been unlocked - the local lock manager enforces a strict first-come first-served model for both local and remote lockers. Sponsored by: Isilon Systems PR: 95247 107555 115524 116679 MFC after: 2 weeks
2008-03-26 15:23:12 +00:00
switch (proto) {
case IPPROTO_TCP:
defsize = 64 * 1024; /* XXX */
break;
case IPPROTO_UDP:
defsize = UDPMSGSIZE;
break;
default:
defsize = RPC_MAXDATASIZE;
break;
}
if (size == 0)
return defsize;
/* Check whether the value is within the upper max limit */
return (size > sb_max_adj ? (u_int)sb_max_adj : (u_int)size);
Add the new kernel-mode NFS Lock Manager. To use it instead of the user-mode lock manager, build a kernel with the NFSLOCKD option and add '-k' to 'rpc_lockd_flags' in rc.conf. Highlights include: * Thread-safe kernel RPC client - many threads can use the same RPC client handle safely with replies being de-multiplexed at the socket upcall (typically driven directly by the NIC interrupt) and handed off to whichever thread matches the reply. For UDP sockets, many RPC clients can share the same socket. This allows the use of a single privileged UDP port number to talk to an arbitrary number of remote hosts. * Single-threaded kernel RPC server. Adding support for multi-threaded server would be relatively straightforward and would follow approximately the Solaris KPI. A single thread should be sufficient for the NLM since it should rarely block in normal operation. * Kernel mode NLM server supporting cancel requests and granted callbacks. I've tested the NLM server reasonably extensively - it passes both my own tests and the NFS Connectathon locking tests running on Solaris, Mac OS X and Ubuntu Linux. * Userland NLM client supported. While the NLM server doesn't have support for the local NFS client's locking needs, it does have to field async replies and granted callbacks from remote NLMs that the local client has contacted. We relay these replies to the userland rpc.lockd over a local domain RPC socket. * Robust deadlock detection for the local lock manager. In particular it will detect deadlocks caused by a lock request that covers more than one blocking request. As required by the NLM protocol, all deadlock detection happens synchronously - a user is guaranteed that if a lock request isn't rejected immediately, the lock will eventually be granted. The old system allowed for a 'deferred deadlock' condition where a blocked lock request could wake up and find that some other deadlock-causing lock owner had beaten them to the lock. * Since both local and remote locks are managed by the same kernel locking code, local and remote processes can safely use file locks for mutual exclusion. Local processes have no fairness advantage compared to remote processes when contending to lock a region that has just been unlocked - the local lock manager enforces a strict first-come first-served model for both local and remote lockers. Sponsored by: Isilon Systems PR: 95247 107555 115524 116679 MFC after: 2 weeks
2008-03-26 15:23:12 +00:00
}
/*
* Find the appropriate address buffer size
*/
u_int
__rpc_get_a_size(int af)
Add the new kernel-mode NFS Lock Manager. To use it instead of the user-mode lock manager, build a kernel with the NFSLOCKD option and add '-k' to 'rpc_lockd_flags' in rc.conf. Highlights include: * Thread-safe kernel RPC client - many threads can use the same RPC client handle safely with replies being de-multiplexed at the socket upcall (typically driven directly by the NIC interrupt) and handed off to whichever thread matches the reply. For UDP sockets, many RPC clients can share the same socket. This allows the use of a single privileged UDP port number to talk to an arbitrary number of remote hosts. * Single-threaded kernel RPC server. Adding support for multi-threaded server would be relatively straightforward and would follow approximately the Solaris KPI. A single thread should be sufficient for the NLM since it should rarely block in normal operation. * Kernel mode NLM server supporting cancel requests and granted callbacks. I've tested the NLM server reasonably extensively - it passes both my own tests and the NFS Connectathon locking tests running on Solaris, Mac OS X and Ubuntu Linux. * Userland NLM client supported. While the NLM server doesn't have support for the local NFS client's locking needs, it does have to field async replies and granted callbacks from remote NLMs that the local client has contacted. We relay these replies to the userland rpc.lockd over a local domain RPC socket. * Robust deadlock detection for the local lock manager. In particular it will detect deadlocks caused by a lock request that covers more than one blocking request. As required by the NLM protocol, all deadlock detection happens synchronously - a user is guaranteed that if a lock request isn't rejected immediately, the lock will eventually be granted. The old system allowed for a 'deferred deadlock' condition where a blocked lock request could wake up and find that some other deadlock-causing lock owner had beaten them to the lock. * Since both local and remote locks are managed by the same kernel locking code, local and remote processes can safely use file locks for mutual exclusion. Local processes have no fairness advantage compared to remote processes when contending to lock a region that has just been unlocked - the local lock manager enforces a strict first-come first-served model for both local and remote lockers. Sponsored by: Isilon Systems PR: 95247 107555 115524 116679 MFC after: 2 weeks
2008-03-26 15:23:12 +00:00
{
switch (af) {
case AF_INET:
return sizeof (struct sockaddr_in);
#ifdef INET6
case AF_INET6:
return sizeof (struct sockaddr_in6);
#endif
case AF_LOCAL:
return sizeof (struct sockaddr_un);
default:
break;
}
return ((u_int)RPC_MAXADDRSIZE);
}
#if 0
/*
* Used to ping the NULL procedure for clnt handle.
* Returns NULL if fails, else a non-NULL pointer.
*/
void *
rpc_nullproc(clnt)
CLIENT *clnt;
{
struct timeval TIMEOUT = {25, 0};
if (clnt_call(clnt, NULLPROC, (xdrproc_t) xdr_void, NULL,
(xdrproc_t) xdr_void, NULL, TIMEOUT) != RPC_SUCCESS) {
return (NULL);
}
return ((void *) clnt);
}
#endif
int
__rpc_socket2sockinfo(struct socket *so, struct __rpc_sockinfo *sip)
{
int type, proto;
struct sockaddr *sa;
sa_family_t family;
struct sockopt opt;
int error;
CURVNET_SET(so->so_vnet);
error = so->so_proto->pr_sockaddr(so, &sa);
CURVNET_RESTORE();
Add the new kernel-mode NFS Lock Manager. To use it instead of the user-mode lock manager, build a kernel with the NFSLOCKD option and add '-k' to 'rpc_lockd_flags' in rc.conf. Highlights include: * Thread-safe kernel RPC client - many threads can use the same RPC client handle safely with replies being de-multiplexed at the socket upcall (typically driven directly by the NIC interrupt) and handed off to whichever thread matches the reply. For UDP sockets, many RPC clients can share the same socket. This allows the use of a single privileged UDP port number to talk to an arbitrary number of remote hosts. * Single-threaded kernel RPC server. Adding support for multi-threaded server would be relatively straightforward and would follow approximately the Solaris KPI. A single thread should be sufficient for the NLM since it should rarely block in normal operation. * Kernel mode NLM server supporting cancel requests and granted callbacks. I've tested the NLM server reasonably extensively - it passes both my own tests and the NFS Connectathon locking tests running on Solaris, Mac OS X and Ubuntu Linux. * Userland NLM client supported. While the NLM server doesn't have support for the local NFS client's locking needs, it does have to field async replies and granted callbacks from remote NLMs that the local client has contacted. We relay these replies to the userland rpc.lockd over a local domain RPC socket. * Robust deadlock detection for the local lock manager. In particular it will detect deadlocks caused by a lock request that covers more than one blocking request. As required by the NLM protocol, all deadlock detection happens synchronously - a user is guaranteed that if a lock request isn't rejected immediately, the lock will eventually be granted. The old system allowed for a 'deferred deadlock' condition where a blocked lock request could wake up and find that some other deadlock-causing lock owner had beaten them to the lock. * Since both local and remote locks are managed by the same kernel locking code, local and remote processes can safely use file locks for mutual exclusion. Local processes have no fairness advantage compared to remote processes when contending to lock a region that has just been unlocked - the local lock manager enforces a strict first-come first-served model for both local and remote lockers. Sponsored by: Isilon Systems PR: 95247 107555 115524 116679 MFC after: 2 weeks
2008-03-26 15:23:12 +00:00
if (error)
return 0;
sip->si_alen = sa->sa_len;
family = sa->sa_family;
free(sa, M_SONAME);
opt.sopt_dir = SOPT_GET;
opt.sopt_level = SOL_SOCKET;
opt.sopt_name = SO_TYPE;
opt.sopt_val = &type;
opt.sopt_valsize = sizeof type;
opt.sopt_td = NULL;
error = sogetopt(so, &opt);
if (error)
return 0;
/* XXX */
if (family != AF_LOCAL) {
if (type == SOCK_STREAM)
proto = IPPROTO_TCP;
else if (type == SOCK_DGRAM)
proto = IPPROTO_UDP;
else
return 0;
} else
proto = 0;
sip->si_af = family;
sip->si_proto = proto;
sip->si_socktype = type;
return 1;
}
/*
* Linear search, but the number of entries is small.
*/
int
__rpc_nconf2sockinfo(const struct netconfig *nconf, struct __rpc_sockinfo *sip)
{
int i;
for (i = 0; i < (sizeof na_cvt) / (sizeof (struct netid_af)); i++)
if (strcmp(na_cvt[i].netid, nconf->nc_netid) == 0 || (
strcmp(nconf->nc_netid, "unix") == 0 &&
strcmp(na_cvt[i].netid, "local") == 0)) {
sip->si_af = na_cvt[i].af;
sip->si_proto = na_cvt[i].protocol;
sip->si_socktype =
__rpc_seman2socktype((int)nconf->nc_semantics);
if (sip->si_socktype == -1)
return 0;
sip->si_alen = __rpc_get_a_size(sip->si_af);
return 1;
}
return 0;
}
struct socket *
__rpc_nconf2socket(const struct netconfig *nconf)
{
struct __rpc_sockinfo si;
struct socket *so;
int error;
if (!__rpc_nconf2sockinfo(nconf, &si))
return 0;
so = NULL;
error = socreate(si.si_af, &so, si.si_socktype, si.si_proto,
curthread->td_ucred, curthread);
if (error)
return NULL;
else
return so;
}
char *
taddr2uaddr(const struct netconfig *nconf, const struct netbuf *nbuf)
{
struct __rpc_sockinfo si;
if (!__rpc_nconf2sockinfo(nconf, &si))
return NULL;
return __rpc_taddr2uaddr_af(si.si_af, nbuf);
}
struct netbuf *
uaddr2taddr(const struct netconfig *nconf, const char *uaddr)
{
struct __rpc_sockinfo si;
if (!__rpc_nconf2sockinfo(nconf, &si))
return NULL;
return __rpc_uaddr2taddr_af(si.si_af, uaddr);
}
char *
__rpc_taddr2uaddr_af(int af, const struct netbuf *nbuf)
{
char *ret;
struct sbuf sb;
struct sockaddr_in *sin;
struct sockaddr_un *sun;
char namebuf[INET_ADDRSTRLEN];
#ifdef INET6
struct sockaddr_in6 *sin6;
char namebuf6[INET6_ADDRSTRLEN];
#endif
uint16_t port;
Add the new kernel-mode NFS Lock Manager. To use it instead of the user-mode lock manager, build a kernel with the NFSLOCKD option and add '-k' to 'rpc_lockd_flags' in rc.conf. Highlights include: * Thread-safe kernel RPC client - many threads can use the same RPC client handle safely with replies being de-multiplexed at the socket upcall (typically driven directly by the NIC interrupt) and handed off to whichever thread matches the reply. For UDP sockets, many RPC clients can share the same socket. This allows the use of a single privileged UDP port number to talk to an arbitrary number of remote hosts. * Single-threaded kernel RPC server. Adding support for multi-threaded server would be relatively straightforward and would follow approximately the Solaris KPI. A single thread should be sufficient for the NLM since it should rarely block in normal operation. * Kernel mode NLM server supporting cancel requests and granted callbacks. I've tested the NLM server reasonably extensively - it passes both my own tests and the NFS Connectathon locking tests running on Solaris, Mac OS X and Ubuntu Linux. * Userland NLM client supported. While the NLM server doesn't have support for the local NFS client's locking needs, it does have to field async replies and granted callbacks from remote NLMs that the local client has contacted. We relay these replies to the userland rpc.lockd over a local domain RPC socket. * Robust deadlock detection for the local lock manager. In particular it will detect deadlocks caused by a lock request that covers more than one blocking request. As required by the NLM protocol, all deadlock detection happens synchronously - a user is guaranteed that if a lock request isn't rejected immediately, the lock will eventually be granted. The old system allowed for a 'deferred deadlock' condition where a blocked lock request could wake up and find that some other deadlock-causing lock owner had beaten them to the lock. * Since both local and remote locks are managed by the same kernel locking code, local and remote processes can safely use file locks for mutual exclusion. Local processes have no fairness advantage compared to remote processes when contending to lock a region that has just been unlocked - the local lock manager enforces a strict first-come first-served model for both local and remote lockers. Sponsored by: Isilon Systems PR: 95247 107555 115524 116679 MFC after: 2 weeks
2008-03-26 15:23:12 +00:00
sbuf_new(&sb, NULL, 0, SBUF_AUTOEXTEND);
switch (af) {
case AF_INET:
if (nbuf->len < sizeof(*sin))
return NULL;
Add the new kernel-mode NFS Lock Manager. To use it instead of the user-mode lock manager, build a kernel with the NFSLOCKD option and add '-k' to 'rpc_lockd_flags' in rc.conf. Highlights include: * Thread-safe kernel RPC client - many threads can use the same RPC client handle safely with replies being de-multiplexed at the socket upcall (typically driven directly by the NIC interrupt) and handed off to whichever thread matches the reply. For UDP sockets, many RPC clients can share the same socket. This allows the use of a single privileged UDP port number to talk to an arbitrary number of remote hosts. * Single-threaded kernel RPC server. Adding support for multi-threaded server would be relatively straightforward and would follow approximately the Solaris KPI. A single thread should be sufficient for the NLM since it should rarely block in normal operation. * Kernel mode NLM server supporting cancel requests and granted callbacks. I've tested the NLM server reasonably extensively - it passes both my own tests and the NFS Connectathon locking tests running on Solaris, Mac OS X and Ubuntu Linux. * Userland NLM client supported. While the NLM server doesn't have support for the local NFS client's locking needs, it does have to field async replies and granted callbacks from remote NLMs that the local client has contacted. We relay these replies to the userland rpc.lockd over a local domain RPC socket. * Robust deadlock detection for the local lock manager. In particular it will detect deadlocks caused by a lock request that covers more than one blocking request. As required by the NLM protocol, all deadlock detection happens synchronously - a user is guaranteed that if a lock request isn't rejected immediately, the lock will eventually be granted. The old system allowed for a 'deferred deadlock' condition where a blocked lock request could wake up and find that some other deadlock-causing lock owner had beaten them to the lock. * Since both local and remote locks are managed by the same kernel locking code, local and remote processes can safely use file locks for mutual exclusion. Local processes have no fairness advantage compared to remote processes when contending to lock a region that has just been unlocked - the local lock manager enforces a strict first-come first-served model for both local and remote lockers. Sponsored by: Isilon Systems PR: 95247 107555 115524 116679 MFC after: 2 weeks
2008-03-26 15:23:12 +00:00
sin = nbuf->buf;
if (inet_ntop(af, &sin->sin_addr, namebuf, sizeof namebuf)
Add the new kernel-mode NFS Lock Manager. To use it instead of the user-mode lock manager, build a kernel with the NFSLOCKD option and add '-k' to 'rpc_lockd_flags' in rc.conf. Highlights include: * Thread-safe kernel RPC client - many threads can use the same RPC client handle safely with replies being de-multiplexed at the socket upcall (typically driven directly by the NIC interrupt) and handed off to whichever thread matches the reply. For UDP sockets, many RPC clients can share the same socket. This allows the use of a single privileged UDP port number to talk to an arbitrary number of remote hosts. * Single-threaded kernel RPC server. Adding support for multi-threaded server would be relatively straightforward and would follow approximately the Solaris KPI. A single thread should be sufficient for the NLM since it should rarely block in normal operation. * Kernel mode NLM server supporting cancel requests and granted callbacks. I've tested the NLM server reasonably extensively - it passes both my own tests and the NFS Connectathon locking tests running on Solaris, Mac OS X and Ubuntu Linux. * Userland NLM client supported. While the NLM server doesn't have support for the local NFS client's locking needs, it does have to field async replies and granted callbacks from remote NLMs that the local client has contacted. We relay these replies to the userland rpc.lockd over a local domain RPC socket. * Robust deadlock detection for the local lock manager. In particular it will detect deadlocks caused by a lock request that covers more than one blocking request. As required by the NLM protocol, all deadlock detection happens synchronously - a user is guaranteed that if a lock request isn't rejected immediately, the lock will eventually be granted. The old system allowed for a 'deferred deadlock' condition where a blocked lock request could wake up and find that some other deadlock-causing lock owner had beaten them to the lock. * Since both local and remote locks are managed by the same kernel locking code, local and remote processes can safely use file locks for mutual exclusion. Local processes have no fairness advantage compared to remote processes when contending to lock a region that has just been unlocked - the local lock manager enforces a strict first-come first-served model for both local and remote lockers. Sponsored by: Isilon Systems PR: 95247 107555 115524 116679 MFC after: 2 weeks
2008-03-26 15:23:12 +00:00
== NULL)
return NULL;
port = ntohs(sin->sin_port);
if (sbuf_printf(&sb, "%s.%u.%u", namebuf,
((uint32_t)port) >> 8,
port & 0xff) < 0)
return NULL;
break;
#ifdef INET6
case AF_INET6:
if (nbuf->len < sizeof(*sin6))
return NULL;
Add the new kernel-mode NFS Lock Manager. To use it instead of the user-mode lock manager, build a kernel with the NFSLOCKD option and add '-k' to 'rpc_lockd_flags' in rc.conf. Highlights include: * Thread-safe kernel RPC client - many threads can use the same RPC client handle safely with replies being de-multiplexed at the socket upcall (typically driven directly by the NIC interrupt) and handed off to whichever thread matches the reply. For UDP sockets, many RPC clients can share the same socket. This allows the use of a single privileged UDP port number to talk to an arbitrary number of remote hosts. * Single-threaded kernel RPC server. Adding support for multi-threaded server would be relatively straightforward and would follow approximately the Solaris KPI. A single thread should be sufficient for the NLM since it should rarely block in normal operation. * Kernel mode NLM server supporting cancel requests and granted callbacks. I've tested the NLM server reasonably extensively - it passes both my own tests and the NFS Connectathon locking tests running on Solaris, Mac OS X and Ubuntu Linux. * Userland NLM client supported. While the NLM server doesn't have support for the local NFS client's locking needs, it does have to field async replies and granted callbacks from remote NLMs that the local client has contacted. We relay these replies to the userland rpc.lockd over a local domain RPC socket. * Robust deadlock detection for the local lock manager. In particular it will detect deadlocks caused by a lock request that covers more than one blocking request. As required by the NLM protocol, all deadlock detection happens synchronously - a user is guaranteed that if a lock request isn't rejected immediately, the lock will eventually be granted. The old system allowed for a 'deferred deadlock' condition where a blocked lock request could wake up and find that some other deadlock-causing lock owner had beaten them to the lock. * Since both local and remote locks are managed by the same kernel locking code, local and remote processes can safely use file locks for mutual exclusion. Local processes have no fairness advantage compared to remote processes when contending to lock a region that has just been unlocked - the local lock manager enforces a strict first-come first-served model for both local and remote lockers. Sponsored by: Isilon Systems PR: 95247 107555 115524 116679 MFC after: 2 weeks
2008-03-26 15:23:12 +00:00
sin6 = nbuf->buf;
if (inet_ntop(af, &sin6->sin6_addr, namebuf6, sizeof namebuf6)
Add the new kernel-mode NFS Lock Manager. To use it instead of the user-mode lock manager, build a kernel with the NFSLOCKD option and add '-k' to 'rpc_lockd_flags' in rc.conf. Highlights include: * Thread-safe kernel RPC client - many threads can use the same RPC client handle safely with replies being de-multiplexed at the socket upcall (typically driven directly by the NIC interrupt) and handed off to whichever thread matches the reply. For UDP sockets, many RPC clients can share the same socket. This allows the use of a single privileged UDP port number to talk to an arbitrary number of remote hosts. * Single-threaded kernel RPC server. Adding support for multi-threaded server would be relatively straightforward and would follow approximately the Solaris KPI. A single thread should be sufficient for the NLM since it should rarely block in normal operation. * Kernel mode NLM server supporting cancel requests and granted callbacks. I've tested the NLM server reasonably extensively - it passes both my own tests and the NFS Connectathon locking tests running on Solaris, Mac OS X and Ubuntu Linux. * Userland NLM client supported. While the NLM server doesn't have support for the local NFS client's locking needs, it does have to field async replies and granted callbacks from remote NLMs that the local client has contacted. We relay these replies to the userland rpc.lockd over a local domain RPC socket. * Robust deadlock detection for the local lock manager. In particular it will detect deadlocks caused by a lock request that covers more than one blocking request. As required by the NLM protocol, all deadlock detection happens synchronously - a user is guaranteed that if a lock request isn't rejected immediately, the lock will eventually be granted. The old system allowed for a 'deferred deadlock' condition where a blocked lock request could wake up and find that some other deadlock-causing lock owner had beaten them to the lock. * Since both local and remote locks are managed by the same kernel locking code, local and remote processes can safely use file locks for mutual exclusion. Local processes have no fairness advantage compared to remote processes when contending to lock a region that has just been unlocked - the local lock manager enforces a strict first-come first-served model for both local and remote lockers. Sponsored by: Isilon Systems PR: 95247 107555 115524 116679 MFC after: 2 weeks
2008-03-26 15:23:12 +00:00
== NULL)
return NULL;
port = ntohs(sin6->sin6_port);
if (sbuf_printf(&sb, "%s.%u.%u", namebuf6,
((uint32_t)port) >> 8,
port & 0xff) < 0)
return NULL;
break;
#endif
case AF_LOCAL:
sun = nbuf->buf;
if (sbuf_printf(&sb, "%.*s", (int)(sun->sun_len -
offsetof(struct sockaddr_un, sun_path)),
sun->sun_path) < 0)
return (NULL);
break;
default:
return NULL;
}
sbuf_finish(&sb);
ret = strdup(sbuf_data(&sb), M_RPC);
sbuf_delete(&sb);
return ret;
}
struct netbuf *
__rpc_uaddr2taddr_af(int af, const char *uaddr)
{
struct netbuf *ret = NULL;
char *addrstr, *p;
unsigned port, portlo, porthi;
struct sockaddr_in *sin;
#ifdef INET6
struct sockaddr_in6 *sin6;
#endif
struct sockaddr_un *sun;
port = 0;
sin = NULL;
if (uaddr == NULL)
return NULL;
Add the new kernel-mode NFS Lock Manager. To use it instead of the user-mode lock manager, build a kernel with the NFSLOCKD option and add '-k' to 'rpc_lockd_flags' in rc.conf. Highlights include: * Thread-safe kernel RPC client - many threads can use the same RPC client handle safely with replies being de-multiplexed at the socket upcall (typically driven directly by the NIC interrupt) and handed off to whichever thread matches the reply. For UDP sockets, many RPC clients can share the same socket. This allows the use of a single privileged UDP port number to talk to an arbitrary number of remote hosts. * Single-threaded kernel RPC server. Adding support for multi-threaded server would be relatively straightforward and would follow approximately the Solaris KPI. A single thread should be sufficient for the NLM since it should rarely block in normal operation. * Kernel mode NLM server supporting cancel requests and granted callbacks. I've tested the NLM server reasonably extensively - it passes both my own tests and the NFS Connectathon locking tests running on Solaris, Mac OS X and Ubuntu Linux. * Userland NLM client supported. While the NLM server doesn't have support for the local NFS client's locking needs, it does have to field async replies and granted callbacks from remote NLMs that the local client has contacted. We relay these replies to the userland rpc.lockd over a local domain RPC socket. * Robust deadlock detection for the local lock manager. In particular it will detect deadlocks caused by a lock request that covers more than one blocking request. As required by the NLM protocol, all deadlock detection happens synchronously - a user is guaranteed that if a lock request isn't rejected immediately, the lock will eventually be granted. The old system allowed for a 'deferred deadlock' condition where a blocked lock request could wake up and find that some other deadlock-causing lock owner had beaten them to the lock. * Since both local and remote locks are managed by the same kernel locking code, local and remote processes can safely use file locks for mutual exclusion. Local processes have no fairness advantage compared to remote processes when contending to lock a region that has just been unlocked - the local lock manager enforces a strict first-come first-served model for both local and remote lockers. Sponsored by: Isilon Systems PR: 95247 107555 115524 116679 MFC after: 2 weeks
2008-03-26 15:23:12 +00:00
addrstr = strdup(uaddr, M_RPC);
if (addrstr == NULL)
return NULL;
/*
* AF_LOCAL addresses are expected to be absolute
* pathnames, anything else will be AF_INET or AF_INET6.
*/
if (*addrstr != '/') {
p = strrchr(addrstr, '.');
if (p == NULL)
goto out;
portlo = (unsigned)strtol(p + 1, NULL, 10);
*p = '\0';
p = strrchr(addrstr, '.');
if (p == NULL)
goto out;
porthi = (unsigned)strtol(p + 1, NULL, 10);
*p = '\0';
port = (porthi << 8) | portlo;
}
ret = (struct netbuf *)malloc(sizeof *ret, M_RPC, M_WAITOK);
switch (af) {
case AF_INET:
sin = (struct sockaddr_in *)malloc(sizeof *sin, M_RPC,
M_WAITOK);
memset(sin, 0, sizeof *sin);
sin->sin_family = AF_INET;
sin->sin_port = htons(port);
if (inet_pton(AF_INET, addrstr, &sin->sin_addr) <= 0) {
Add the new kernel-mode NFS Lock Manager. To use it instead of the user-mode lock manager, build a kernel with the NFSLOCKD option and add '-k' to 'rpc_lockd_flags' in rc.conf. Highlights include: * Thread-safe kernel RPC client - many threads can use the same RPC client handle safely with replies being de-multiplexed at the socket upcall (typically driven directly by the NIC interrupt) and handed off to whichever thread matches the reply. For UDP sockets, many RPC clients can share the same socket. This allows the use of a single privileged UDP port number to talk to an arbitrary number of remote hosts. * Single-threaded kernel RPC server. Adding support for multi-threaded server would be relatively straightforward and would follow approximately the Solaris KPI. A single thread should be sufficient for the NLM since it should rarely block in normal operation. * Kernel mode NLM server supporting cancel requests and granted callbacks. I've tested the NLM server reasonably extensively - it passes both my own tests and the NFS Connectathon locking tests running on Solaris, Mac OS X and Ubuntu Linux. * Userland NLM client supported. While the NLM server doesn't have support for the local NFS client's locking needs, it does have to field async replies and granted callbacks from remote NLMs that the local client has contacted. We relay these replies to the userland rpc.lockd over a local domain RPC socket. * Robust deadlock detection for the local lock manager. In particular it will detect deadlocks caused by a lock request that covers more than one blocking request. As required by the NLM protocol, all deadlock detection happens synchronously - a user is guaranteed that if a lock request isn't rejected immediately, the lock will eventually be granted. The old system allowed for a 'deferred deadlock' condition where a blocked lock request could wake up and find that some other deadlock-causing lock owner had beaten them to the lock. * Since both local and remote locks are managed by the same kernel locking code, local and remote processes can safely use file locks for mutual exclusion. Local processes have no fairness advantage compared to remote processes when contending to lock a region that has just been unlocked - the local lock manager enforces a strict first-come first-served model for both local and remote lockers. Sponsored by: Isilon Systems PR: 95247 107555 115524 116679 MFC after: 2 weeks
2008-03-26 15:23:12 +00:00
free(sin, M_RPC);
free(ret, M_RPC);
ret = NULL;
goto out;
}
sin->sin_len = ret->maxlen = ret->len = sizeof *sin;
ret->buf = sin;
break;
#ifdef INET6
case AF_INET6:
sin6 = (struct sockaddr_in6 *)malloc(sizeof *sin6, M_RPC,
M_WAITOK);
memset(sin6, 0, sizeof *sin6);
sin6->sin6_family = AF_INET6;
sin6->sin6_port = htons(port);
if (inet_pton(AF_INET6, addrstr, &sin6->sin6_addr) <= 0) {
Add the new kernel-mode NFS Lock Manager. To use it instead of the user-mode lock manager, build a kernel with the NFSLOCKD option and add '-k' to 'rpc_lockd_flags' in rc.conf. Highlights include: * Thread-safe kernel RPC client - many threads can use the same RPC client handle safely with replies being de-multiplexed at the socket upcall (typically driven directly by the NIC interrupt) and handed off to whichever thread matches the reply. For UDP sockets, many RPC clients can share the same socket. This allows the use of a single privileged UDP port number to talk to an arbitrary number of remote hosts. * Single-threaded kernel RPC server. Adding support for multi-threaded server would be relatively straightforward and would follow approximately the Solaris KPI. A single thread should be sufficient for the NLM since it should rarely block in normal operation. * Kernel mode NLM server supporting cancel requests and granted callbacks. I've tested the NLM server reasonably extensively - it passes both my own tests and the NFS Connectathon locking tests running on Solaris, Mac OS X and Ubuntu Linux. * Userland NLM client supported. While the NLM server doesn't have support for the local NFS client's locking needs, it does have to field async replies and granted callbacks from remote NLMs that the local client has contacted. We relay these replies to the userland rpc.lockd over a local domain RPC socket. * Robust deadlock detection for the local lock manager. In particular it will detect deadlocks caused by a lock request that covers more than one blocking request. As required by the NLM protocol, all deadlock detection happens synchronously - a user is guaranteed that if a lock request isn't rejected immediately, the lock will eventually be granted. The old system allowed for a 'deferred deadlock' condition where a blocked lock request could wake up and find that some other deadlock-causing lock owner had beaten them to the lock. * Since both local and remote locks are managed by the same kernel locking code, local and remote processes can safely use file locks for mutual exclusion. Local processes have no fairness advantage compared to remote processes when contending to lock a region that has just been unlocked - the local lock manager enforces a strict first-come first-served model for both local and remote lockers. Sponsored by: Isilon Systems PR: 95247 107555 115524 116679 MFC after: 2 weeks
2008-03-26 15:23:12 +00:00
free(sin6, M_RPC);
free(ret, M_RPC);
ret = NULL;
goto out;
}
sin6->sin6_len = ret->maxlen = ret->len = sizeof *sin6;
ret->buf = sin6;
break;
#endif
case AF_LOCAL:
sun = (struct sockaddr_un *)malloc(sizeof *sun, M_RPC,
M_WAITOK);
memset(sun, 0, sizeof *sun);
sun->sun_family = AF_LOCAL;
strncpy(sun->sun_path, addrstr, sizeof(sun->sun_path) - 1);
ret->len = ret->maxlen = sun->sun_len = SUN_LEN(sun);
ret->buf = sun;
break;
default:
break;
}
out:
free(addrstr, M_RPC);
return ret;
}
int
__rpc_seman2socktype(int semantics)
{
switch (semantics) {
case NC_TPI_CLTS:
return SOCK_DGRAM;
case NC_TPI_COTS_ORD:
return SOCK_STREAM;
case NC_TPI_RAW:
return SOCK_RAW;
default:
break;
}
return -1;
}
int
__rpc_socktype2seman(int socktype)
{
switch (socktype) {
case SOCK_DGRAM:
return NC_TPI_CLTS;
case SOCK_STREAM:
return NC_TPI_COTS_ORD;
case SOCK_RAW:
return NC_TPI_RAW;
default:
break;
}
return -1;
}
/*
* Returns the type of the network as defined in <rpc/nettype.h>
* If nettype is NULL, it defaults to NETPATH.
*/
static int
getnettype(const char *nettype)
{
int i;
if ((nettype == NULL) || (nettype[0] == 0)) {
return (_RPC_NETPATH); /* Default */
}
#if 0
nettype = strlocase(nettype);
#endif
for (i = 0; _rpctypelist[i].name; i++)
if (strcasecmp(nettype, _rpctypelist[i].name) == 0) {
return (_rpctypelist[i].type);
}
return (_rpctypelist[i].type);
}
/*
* For the given nettype (tcp or udp only), return the first structure found.
* This should be freed by calling freenetconfigent()
*/
struct netconfig *
__rpc_getconfip(const char *nettype)
{
char *netid;
static char *netid_tcp = (char *) NULL;
static char *netid_udp = (char *) NULL;
struct netconfig *dummy;
if (!netid_udp && !netid_tcp) {
struct netconfig *nconf;
void *confighandle;
if (!(confighandle = setnetconfig())) {
log(LOG_ERR, "rpc: failed to open " NETCONFIG);
return (NULL);
}
while ((nconf = getnetconfig(confighandle)) != NULL) {
if (strcmp(nconf->nc_protofmly, NC_INET) == 0) {
if (strcmp(nconf->nc_proto, NC_TCP) == 0) {
netid_tcp = strdup(nconf->nc_netid,
M_RPC);
} else
if (strcmp(nconf->nc_proto, NC_UDP) == 0) {
netid_udp = strdup(nconf->nc_netid,
M_RPC);
}
}
}
endnetconfig(confighandle);
}
if (strcmp(nettype, "udp") == 0)
netid = netid_udp;
else if (strcmp(nettype, "tcp") == 0)
netid = netid_tcp;
else {
return (NULL);
}
if ((netid == NULL) || (netid[0] == 0)) {
return (NULL);
}
dummy = getnetconfigent(netid);
return (dummy);
}
/*
* Returns the type of the nettype, which should then be used with
* __rpc_getconf().
*
* For simplicity in the kernel, we don't support the NETPATH
* environment variable. We behave as userland would then NETPATH is
* unset, i.e. iterate over all visible entries in netconfig.
*/
void *
__rpc_setconf(const char *nettype)
Add the new kernel-mode NFS Lock Manager. To use it instead of the user-mode lock manager, build a kernel with the NFSLOCKD option and add '-k' to 'rpc_lockd_flags' in rc.conf. Highlights include: * Thread-safe kernel RPC client - many threads can use the same RPC client handle safely with replies being de-multiplexed at the socket upcall (typically driven directly by the NIC interrupt) and handed off to whichever thread matches the reply. For UDP sockets, many RPC clients can share the same socket. This allows the use of a single privileged UDP port number to talk to an arbitrary number of remote hosts. * Single-threaded kernel RPC server. Adding support for multi-threaded server would be relatively straightforward and would follow approximately the Solaris KPI. A single thread should be sufficient for the NLM since it should rarely block in normal operation. * Kernel mode NLM server supporting cancel requests and granted callbacks. I've tested the NLM server reasonably extensively - it passes both my own tests and the NFS Connectathon locking tests running on Solaris, Mac OS X and Ubuntu Linux. * Userland NLM client supported. While the NLM server doesn't have support for the local NFS client's locking needs, it does have to field async replies and granted callbacks from remote NLMs that the local client has contacted. We relay these replies to the userland rpc.lockd over a local domain RPC socket. * Robust deadlock detection for the local lock manager. In particular it will detect deadlocks caused by a lock request that covers more than one blocking request. As required by the NLM protocol, all deadlock detection happens synchronously - a user is guaranteed that if a lock request isn't rejected immediately, the lock will eventually be granted. The old system allowed for a 'deferred deadlock' condition where a blocked lock request could wake up and find that some other deadlock-causing lock owner had beaten them to the lock. * Since both local and remote locks are managed by the same kernel locking code, local and remote processes can safely use file locks for mutual exclusion. Local processes have no fairness advantage compared to remote processes when contending to lock a region that has just been unlocked - the local lock manager enforces a strict first-come first-served model for both local and remote lockers. Sponsored by: Isilon Systems PR: 95247 107555 115524 116679 MFC after: 2 weeks
2008-03-26 15:23:12 +00:00
{
struct handle *handle;
handle = (struct handle *) malloc(sizeof (struct handle),
M_RPC, M_WAITOK);
switch (handle->nettype = getnettype(nettype)) {
case _RPC_NETPATH:
case _RPC_CIRCUIT_N:
case _RPC_DATAGRAM_N:
if (!(handle->nhandle = setnetconfig()))
goto failed;
handle->nflag = TRUE;
break;
case _RPC_VISIBLE:
case _RPC_CIRCUIT_V:
case _RPC_DATAGRAM_V:
case _RPC_TCP:
case _RPC_UDP:
if (!(handle->nhandle = setnetconfig())) {
log(LOG_ERR, "rpc: failed to open " NETCONFIG);
goto failed;
}
handle->nflag = FALSE;
break;
default:
goto failed;
}
return (handle);
failed:
free(handle, M_RPC);
return (NULL);
}
/*
* Returns the next netconfig struct for the given "net" type.
* __rpc_setconf() should have been called previously.
*/
struct netconfig *
__rpc_getconf(void *vhandle)
{
struct handle *handle;
struct netconfig *nconf;
handle = (struct handle *)vhandle;
if (handle == NULL) {
return (NULL);
}
for (;;) {
if (handle->nflag) {
nconf = getnetconfig(handle->nhandle);
if (nconf && !(nconf->nc_flag & NC_VISIBLE))
continue;
} else {
nconf = getnetconfig(handle->nhandle);
}
if (nconf == NULL)
break;
if ((nconf->nc_semantics != NC_TPI_CLTS) &&
(nconf->nc_semantics != NC_TPI_COTS) &&
(nconf->nc_semantics != NC_TPI_COTS_ORD))
continue;
switch (handle->nettype) {
case _RPC_VISIBLE:
if (!(nconf->nc_flag & NC_VISIBLE))
continue;
/* FALLTHROUGH */
case _RPC_NETPATH: /* Be happy */
break;
case _RPC_CIRCUIT_V:
if (!(nconf->nc_flag & NC_VISIBLE))
continue;
/* FALLTHROUGH */
case _RPC_CIRCUIT_N:
if ((nconf->nc_semantics != NC_TPI_COTS) &&
(nconf->nc_semantics != NC_TPI_COTS_ORD))
continue;
break;
case _RPC_DATAGRAM_V:
if (!(nconf->nc_flag & NC_VISIBLE))
continue;
/* FALLTHROUGH */
case _RPC_DATAGRAM_N:
if (nconf->nc_semantics != NC_TPI_CLTS)
continue;
break;
case _RPC_TCP:
if (((nconf->nc_semantics != NC_TPI_COTS) &&
(nconf->nc_semantics != NC_TPI_COTS_ORD)) ||
(strcmp(nconf->nc_protofmly, NC_INET)
#ifdef INET6
&& strcmp(nconf->nc_protofmly, NC_INET6))
#else
)
#endif
||
strcmp(nconf->nc_proto, NC_TCP))
continue;
break;
case _RPC_UDP:
if ((nconf->nc_semantics != NC_TPI_CLTS) ||
(strcmp(nconf->nc_protofmly, NC_INET)
#ifdef INET6
&& strcmp(nconf->nc_protofmly, NC_INET6))
#else
)
#endif
||
strcmp(nconf->nc_proto, NC_UDP))
continue;
break;
}
break;
}
return (nconf);
}
void
__rpc_endconf(void *vhandle)
Add the new kernel-mode NFS Lock Manager. To use it instead of the user-mode lock manager, build a kernel with the NFSLOCKD option and add '-k' to 'rpc_lockd_flags' in rc.conf. Highlights include: * Thread-safe kernel RPC client - many threads can use the same RPC client handle safely with replies being de-multiplexed at the socket upcall (typically driven directly by the NIC interrupt) and handed off to whichever thread matches the reply. For UDP sockets, many RPC clients can share the same socket. This allows the use of a single privileged UDP port number to talk to an arbitrary number of remote hosts. * Single-threaded kernel RPC server. Adding support for multi-threaded server would be relatively straightforward and would follow approximately the Solaris KPI. A single thread should be sufficient for the NLM since it should rarely block in normal operation. * Kernel mode NLM server supporting cancel requests and granted callbacks. I've tested the NLM server reasonably extensively - it passes both my own tests and the NFS Connectathon locking tests running on Solaris, Mac OS X and Ubuntu Linux. * Userland NLM client supported. While the NLM server doesn't have support for the local NFS client's locking needs, it does have to field async replies and granted callbacks from remote NLMs that the local client has contacted. We relay these replies to the userland rpc.lockd over a local domain RPC socket. * Robust deadlock detection for the local lock manager. In particular it will detect deadlocks caused by a lock request that covers more than one blocking request. As required by the NLM protocol, all deadlock detection happens synchronously - a user is guaranteed that if a lock request isn't rejected immediately, the lock will eventually be granted. The old system allowed for a 'deferred deadlock' condition where a blocked lock request could wake up and find that some other deadlock-causing lock owner had beaten them to the lock. * Since both local and remote locks are managed by the same kernel locking code, local and remote processes can safely use file locks for mutual exclusion. Local processes have no fairness advantage compared to remote processes when contending to lock a region that has just been unlocked - the local lock manager enforces a strict first-come first-served model for both local and remote lockers. Sponsored by: Isilon Systems PR: 95247 107555 115524 116679 MFC after: 2 weeks
2008-03-26 15:23:12 +00:00
{
struct handle *handle;
handle = (struct handle *) vhandle;
if (handle == NULL) {
return;
}
endnetconfig(handle->nhandle);
free(handle, M_RPC);
}
int
__rpc_sockisbound(struct socket *so)
{
struct sockaddr *sa;
int error, bound;
CURVNET_SET(so->so_vnet);
error = so->so_proto->pr_sockaddr(so, &sa);
CURVNET_RESTORE();
Add the new kernel-mode NFS Lock Manager. To use it instead of the user-mode lock manager, build a kernel with the NFSLOCKD option and add '-k' to 'rpc_lockd_flags' in rc.conf. Highlights include: * Thread-safe kernel RPC client - many threads can use the same RPC client handle safely with replies being de-multiplexed at the socket upcall (typically driven directly by the NIC interrupt) and handed off to whichever thread matches the reply. For UDP sockets, many RPC clients can share the same socket. This allows the use of a single privileged UDP port number to talk to an arbitrary number of remote hosts. * Single-threaded kernel RPC server. Adding support for multi-threaded server would be relatively straightforward and would follow approximately the Solaris KPI. A single thread should be sufficient for the NLM since it should rarely block in normal operation. * Kernel mode NLM server supporting cancel requests and granted callbacks. I've tested the NLM server reasonably extensively - it passes both my own tests and the NFS Connectathon locking tests running on Solaris, Mac OS X and Ubuntu Linux. * Userland NLM client supported. While the NLM server doesn't have support for the local NFS client's locking needs, it does have to field async replies and granted callbacks from remote NLMs that the local client has contacted. We relay these replies to the userland rpc.lockd over a local domain RPC socket. * Robust deadlock detection for the local lock manager. In particular it will detect deadlocks caused by a lock request that covers more than one blocking request. As required by the NLM protocol, all deadlock detection happens synchronously - a user is guaranteed that if a lock request isn't rejected immediately, the lock will eventually be granted. The old system allowed for a 'deferred deadlock' condition where a blocked lock request could wake up and find that some other deadlock-causing lock owner had beaten them to the lock. * Since both local and remote locks are managed by the same kernel locking code, local and remote processes can safely use file locks for mutual exclusion. Local processes have no fairness advantage compared to remote processes when contending to lock a region that has just been unlocked - the local lock manager enforces a strict first-come first-served model for both local and remote lockers. Sponsored by: Isilon Systems PR: 95247 107555 115524 116679 MFC after: 2 weeks
2008-03-26 15:23:12 +00:00
if (error)
return (0);
switch (sa->sa_family) {
case AF_INET:
bound = (((struct sockaddr_in *) sa)->sin_port != 0);
break;
#ifdef INET6
case AF_INET6:
bound = (((struct sockaddr_in6 *) sa)->sin6_port != 0);
break;
#endif
case AF_LOCAL:
/* XXX check this */
bound = (((struct sockaddr_un *) sa)->sun_path[0] != '\0');
break;
default:
bound = FALSE;
break;
}
free(sa, M_SONAME);
return bound;
}
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
/*
* Implement XDR-style API for RPC call.
*/
enum clnt_stat
clnt_call_private(
CLIENT *cl, /* client handle */
struct rpc_callextra *ext, /* call metadata */
rpcproc_t proc, /* procedure number */
xdrproc_t xargs, /* xdr routine for args */
void *argsp, /* pointer to args */
xdrproc_t xresults, /* xdr routine for results */
void *resultsp, /* pointer to results */
struct timeval utimeout) /* seconds to wait before giving up */
{
XDR xdrs;
struct mbuf *mreq;
struct mbuf *mrep;
enum clnt_stat stat;
mreq = m_getcl(M_WAITOK, MT_DATA, 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
xdrmbuf_create(&xdrs, mreq, XDR_ENCODE);
if (!xargs(&xdrs, argsp)) {
m_freem(mreq);
return (RPC_CANTENCODEARGS);
}
XDR_DESTROY(&xdrs);
stat = CLNT_CALL_MBUF(cl, ext, proc, mreq, &mrep, utimeout);
m_freem(mreq);
if (stat == RPC_SUCCESS) {
xdrmbuf_create(&xdrs, mrep, XDR_DECODE);
if (!xresults(&xdrs, resultsp)) {
XDR_DESTROY(&xdrs);
return (RPC_CANTDECODERES);
}
XDR_DESTROY(&xdrs);
}
return (stat);
}
/*
* Bind a socket to a privileged IP port
*/
int
bindresvport(struct socket *so, struct sockaddr *sa)
{
int old, error, af;
bool_t freesa = FALSE;
struct sockaddr_in *sin;
#ifdef INET6
struct sockaddr_in6 *sin6;
#endif
struct sockopt opt;
int proto, portrange, portlow;
uint16_t *portp;
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
socklen_t salen;
if (sa == NULL) {
CURVNET_SET(so->so_vnet);
error = so->so_proto->pr_sockaddr(so, &sa);
CURVNET_RESTORE();
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 (error)
return (error);
freesa = TRUE;
af = sa->sa_family;
salen = sa->sa_len;
memset(sa, 0, sa->sa_len);
} else {
af = sa->sa_family;
salen = sa->sa_len;
}
switch (af) {
case AF_INET:
proto = IPPROTO_IP;
portrange = IP_PORTRANGE;
portlow = IP_PORTRANGE_LOW;
sin = (struct sockaddr_in *)sa;
portp = &sin->sin_port;
break;
#ifdef INET6
case AF_INET6:
proto = IPPROTO_IPV6;
portrange = IPV6_PORTRANGE;
portlow = IPV6_PORTRANGE_LOW;
sin6 = (struct sockaddr_in6 *)sa;
portp = &sin6->sin6_port;
break;
#endif
default:
return (EPFNOSUPPORT);
}
sa->sa_family = af;
sa->sa_len = salen;
if (*portp == 0) {
bzero(&opt, sizeof(opt));
opt.sopt_dir = SOPT_GET;
opt.sopt_level = proto;
opt.sopt_name = portrange;
opt.sopt_val = &old;
opt.sopt_valsize = sizeof(old);
error = sogetopt(so, &opt);
if (error) {
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
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
opt.sopt_dir = SOPT_SET;
opt.sopt_val = &portlow;
error = sosetopt(so, &opt);
if (error)
goto out;
}
error = sobind(so, sa, curthread);
if (*portp == 0) {
if (error) {
opt.sopt_dir = SOPT_SET;
opt.sopt_val = &old;
sosetopt(so, &opt);
}
}
out:
if (freesa)
free(sa, M_SONAME);
return (error);
}
Add TLS support to the kernel RPC. An internet draft titled "Towards Remote Procedure Call Encryption By Default" describes how TLS is to be used for Sun RPC, with NFS as an intended use case. This patch adds client and server support for this to the kernel RPC, using KERN_TLS and upcalls to daemons for the handshake, peer reset and other non-application data record cases. The upcalls to the daemons use three fields to uniquely identify the TCP connection. They are the time.tv_sec, time.tv_usec of the connection establshment, plus a 64bit sequence number. The time fields avoid problems with re-use of the sequence number after a daemon restart. For the server side, once a Null RPC with AUTH_TLS is received, kernel reception on the socket is blocked and an upcall to the rpctlssd(8) daemon is done to perform the TLS handshake. Upon completion, the completion status of the handshake is stored in xp_tls as flag bits and the reply to the Null RPC is sent. For the client, if CLSET_TLS has been set, a new TCP connection will send the Null RPC with AUTH_TLS to initiate the handshake. The client kernel RPC code will then block kernel I/O on the socket and do an upcall to the rpctlscd(8) daemon to perform the handshake. If the upcall is successful, ct_rcvstate will be maintained to indicate if/when an upcall is being done. If non-application data records are received, the code does an upcall to the appropriate daemon, which will do a SSL_read() of 0 length to handle the record(s). When the socket is being shut down, upcalls are done to the daemons, so that they can perform SSL_shutdown() calls to perform the "peer reset". The rpctlssd(8) and rpctlscd(8) daemons require a patched version of the openssl library and, as such, will not be committed to head at this time. Although the changes done by this patch are fairly numerous, there should be no semantics change to the kernel RPC at this time. A future commit to the NFS code will optionally enable use of TLS for NFS.
2020-08-22 03:57:55 +00:00
/*
* Make sure an mbuf list is made up entirely of ext_pgs mbufs.
* This is needed for sosend() when KERN_TLS is being used.
* (There might also be a performance improvement for certain
* network interfaces that handle ext_pgs mbufs efficiently.)
* It expects at least one non-ext_pgs mbuf followed by zero
* or more ext_pgs mbufs. It does not handle the case where
* non-ext_pgs mbuf(s) follow ext_pgs ones.
* It also performs sanity checks on the resultant list.
* The "mp" argument list is consumed.
* The "maxextsiz" argument is the upper bound on the data
* size for each mbuf (usually 16K for KERN_TLS).
*/
struct mbuf *
_rpc_copym_into_ext_pgs(struct mbuf *mp, int maxextsiz)
{
struct mbuf *m, *m2, *m3, *mhead;
int tlen;
KASSERT((mp->m_flags & (M_EXT | M_EXTPG)) !=
(M_EXT | M_EXTPG), ("_rpc_copym_into_ext_pgs:"
" first mbuf is an ext_pgs"));
/*
* Find the last non-ext_pgs mbuf and the total
* length of the non-ext_pgs mbuf(s).
* The first mbuf must always be a non-ext_pgs
* mbuf.
*/
tlen = mp->m_len;
m2 = mp;
for (m = mp->m_next; m != NULL; m = m->m_next) {
if ((m->m_flags & M_EXTPG) != 0)
break;
tlen += m->m_len;
m2 = m;
}
/*
* Copy the non-ext_pgs mbuf(s) into an ext_pgs
* mbuf list.
*/
m2->m_next = NULL;
mhead = mb_mapped_to_unmapped(mp, tlen, maxextsiz,
M_WAITOK, &m2);
/*
* Link the ext_pgs list onto the newly copied
* list and free up the non-ext_pgs mbuf(s).
*/
m2->m_next = m;
m_freem(mp);
/*
* Sanity check the resultant mbuf list. Check for and
* remove any 0 length mbufs in the list, since the
* KERN_TLS code does not expect any 0 length mbuf(s)
* in the list.
*/
m3 = NULL;
m2 = mhead;
tlen = 0;
while (m2 != NULL) {
KASSERT(m2->m_len >= 0, ("_rpc_copym_into_ext_pgs:"
" negative m_len"));
KASSERT((m2->m_flags & (M_EXT | M_EXTPG)) ==
(M_EXT | M_EXTPG), ("_rpc_copym_into_ext_pgs:"
" non-nomap mbuf in list"));
if (m2->m_len == 0) {
if (m3 != NULL)
m3->m_next = m2->m_next;
else
m = m2->m_next;
m2->m_next = NULL;
m_free(m2);
if (m3 != NULL)
m2 = m3->m_next;
else
m2 = m;
} else {
MBUF_EXT_PGS_ASSERT_SANITY(m2);
m3 = m2;
tlen += m2->m_len;
m2 = m2->m_next;
}
}
return (mhead);
}
/*
* Kernel module glue
*/
static int
krpc_modevent(module_t mod, int type, void *data)
{
Add TLS support to the kernel RPC. An internet draft titled "Towards Remote Procedure Call Encryption By Default" describes how TLS is to be used for Sun RPC, with NFS as an intended use case. This patch adds client and server support for this to the kernel RPC, using KERN_TLS and upcalls to daemons for the handshake, peer reset and other non-application data record cases. The upcalls to the daemons use three fields to uniquely identify the TCP connection. They are the time.tv_sec, time.tv_usec of the connection establshment, plus a 64bit sequence number. The time fields avoid problems with re-use of the sequence number after a daemon restart. For the server side, once a Null RPC with AUTH_TLS is received, kernel reception on the socket is blocked and an upcall to the rpctlssd(8) daemon is done to perform the TLS handshake. Upon completion, the completion status of the handshake is stored in xp_tls as flag bits and the reply to the Null RPC is sent. For the client, if CLSET_TLS has been set, a new TCP connection will send the Null RPC with AUTH_TLS to initiate the handshake. The client kernel RPC code will then block kernel I/O on the socket and do an upcall to the rpctlscd(8) daemon to perform the handshake. If the upcall is successful, ct_rcvstate will be maintained to indicate if/when an upcall is being done. If non-application data records are received, the code does an upcall to the appropriate daemon, which will do a SSL_read() of 0 length to handle the record(s). When the socket is being shut down, upcalls are done to the daemons, so that they can perform SSL_shutdown() calls to perform the "peer reset". The rpctlssd(8) and rpctlscd(8) daemons require a patched version of the openssl library and, as such, will not be committed to head at this time. Although the changes done by this patch are fairly numerous, there should be no semantics change to the kernel RPC at this time. A future commit to the NFS code will optionally enable use of TLS for NFS.
2020-08-22 03:57:55 +00:00
int error = 0;
Add TLS support to the kernel RPC. An internet draft titled "Towards Remote Procedure Call Encryption By Default" describes how TLS is to be used for Sun RPC, with NFS as an intended use case. This patch adds client and server support for this to the kernel RPC, using KERN_TLS and upcalls to daemons for the handshake, peer reset and other non-application data record cases. The upcalls to the daemons use three fields to uniquely identify the TCP connection. They are the time.tv_sec, time.tv_usec of the connection establshment, plus a 64bit sequence number. The time fields avoid problems with re-use of the sequence number after a daemon restart. For the server side, once a Null RPC with AUTH_TLS is received, kernel reception on the socket is blocked and an upcall to the rpctlssd(8) daemon is done to perform the TLS handshake. Upon completion, the completion status of the handshake is stored in xp_tls as flag bits and the reply to the Null RPC is sent. For the client, if CLSET_TLS has been set, a new TCP connection will send the Null RPC with AUTH_TLS to initiate the handshake. The client kernel RPC code will then block kernel I/O on the socket and do an upcall to the rpctlscd(8) daemon to perform the handshake. If the upcall is successful, ct_rcvstate will be maintained to indicate if/when an upcall is being done. If non-application data records are received, the code does an upcall to the appropriate daemon, which will do a SSL_read() of 0 length to handle the record(s). When the socket is being shut down, upcalls are done to the daemons, so that they can perform SSL_shutdown() calls to perform the "peer reset". The rpctlssd(8) and rpctlscd(8) daemons require a patched version of the openssl library and, as such, will not be committed to head at this time. Although the changes done by this patch are fairly numerous, there should be no semantics change to the kernel RPC at this time. A future commit to the NFS code will optionally enable use of TLS for NFS.
2020-08-22 03:57:55 +00:00
switch (type) {
case MOD_LOAD:
error = rpctls_init();
break;
case MOD_UNLOAD:
/*
* Cannot be unloaded, since the rpctlssd or rpctlscd daemons
* might be performing a rpctls syscall.
*/
/* FALLTHROUGH */
default:
error = EOPNOTSUPP;
}
return (error);
}
static moduledata_t krpc_mod = {
"krpc",
krpc_modevent,
NULL,
};
DECLARE_MODULE(krpc, krpc_mod, SI_SUB_VFS, SI_ORDER_ANY);
/* So that loader and kldload(2) can find us, wherever we are.. */
MODULE_VERSION(krpc, 1);
MODULE_DEPEND(krpc, xdr, 1, 1, 1);