freebsd-dev/sys/rpc/rpcb_prot.c

247 lines
6.8 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: rpcb_prot.c,v 1.3 2000/07/14 08:40:42 fvdl 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.
*/
/* #ident "@(#)rpcb_prot.c 1.13 94/04/24 SMI" */
#if defined(LIBC_SCCS) && !defined(lint)
static char sccsid[] = "@(#)rpcb_prot.c 1.9 89/04/21 Copyr 1984 Sun Micro";
#endif
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
/*
* rpcb_prot.c
* XDR routines for the rpcbinder version 3.
*
* Copyright (C) 1984, 1988, Sun Microsystems, Inc.
*/
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/kernel.h>
#include <sys/malloc.h>
#include <rpc/rpc.h>
#include <rpc/rpc_com.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/rpcb_prot.h>
bool_t
xdr_portmap(XDR *xdrs, struct portmap *regs)
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 (xdr_u_long(xdrs, &regs->pm_prog) &&
xdr_u_long(xdrs, &regs->pm_vers) &&
xdr_u_long(xdrs, &regs->pm_prot))
return (xdr_u_long(xdrs, &regs->pm_port));
return (FALSE);
}
bool_t
xdr_rpcb(XDR *xdrs, RPCB *objp)
{
if (!xdr_uint32_t(xdrs, &objp->r_prog)) {
return (FALSE);
}
if (!xdr_uint32_t(xdrs, &objp->r_vers)) {
return (FALSE);
}
if (!xdr_string(xdrs, &objp->r_netid, RPC_MAXDATASIZE)) {
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
return (FALSE);
}
if (!xdr_string(xdrs, &objp->r_addr, RPC_MAXDATASIZE)) {
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
return (FALSE);
}
if (!xdr_string(xdrs, &objp->r_owner, RPC_MAXDATASIZE)) {
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
return (FALSE);
}
return (TRUE);
}
/*
* rpcblist_ptr implements a linked list. The RPCL definition from
* rpcb_prot.x is:
*
* struct rpcblist {
* rpcb rpcb_map;
* struct rpcblist *rpcb_next;
* };
* typedef rpcblist *rpcblist_ptr;
*
* Recall that "pointers" in XDR are encoded as a boolean, indicating whether
* there's any data behind the pointer, followed by the data (if any exists).
* The boolean can be interpreted as ``more data follows me''; if FALSE then
* nothing follows the boolean; if TRUE then the boolean is followed by an
* actual struct rpcb, and another rpcblist_ptr (declared in RPCL as "struct
* rpcblist *").
*
* This could be implemented via the xdr_pointer type, though this would
* result in one recursive call per element in the list. Rather than do that
* we can ``unwind'' the recursion into a while loop and use xdr_reference to
* serialize the rpcb elements.
*/
bool_t
xdr_rpcblist_ptr(XDR *xdrs, rpcblist_ptr *rp)
{
/*
* more_elements is pre-computed in case the direction is
* XDR_ENCODE or XDR_FREE. more_elements is overwritten by
* xdr_bool when the direction is XDR_DECODE.
*/
bool_t more_elements;
int freeing = (xdrs->x_op == XDR_FREE);
rpcblist_ptr next;
rpcblist_ptr next_copy;
next = NULL;
for (;;) {
more_elements = (bool_t)(*rp != NULL);
if (! xdr_bool(xdrs, &more_elements)) {
return (FALSE);
}
if (! more_elements) {
return (TRUE); /* we are done */
}
/*
* the unfortunate side effect of non-recursion is that in
* the case of freeing we must remember the next object
* before we free the current object ...
*/
if (freeing && *rp)
next = (*rp)->rpcb_next;
if (! xdr_reference(xdrs, (caddr_t *)rp,
(u_int)sizeof (RPCBLIST), (xdrproc_t)xdr_rpcb)) {
return (FALSE);
}
if (freeing) {
next_copy = next;
rp = &next_copy;
/*
* Note that in the subsequent iteration, next_copy
* gets nulled out by the xdr_reference
* but next itself survives.
*/
} else if (*rp) {
rp = &((*rp)->rpcb_next);
}
}
/*NOTREACHED*/
}
#if 0
/*
* xdr_rpcblist() is specified to take a RPCBLIST **, but is identical in
* functionality to xdr_rpcblist_ptr().
*/
bool_t
xdr_rpcblist(XDR *xdrs, RPCBLIST **rp)
{
bool_t dummy;
dummy = xdr_rpcblist_ptr(xdrs, (rpcblist_ptr *)rp);
return (dummy);
}
#endif
bool_t
xdr_rpcb_entry(XDR *xdrs, rpcb_entry *objp)
{
if (!xdr_string(xdrs, &objp->r_maddr, RPC_MAXDATASIZE)) {
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
return (FALSE);
}
if (!xdr_string(xdrs, &objp->r_nc_netid, RPC_MAXDATASIZE)) {
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
return (FALSE);
}
if (!xdr_uint32_t(xdrs, &objp->r_nc_semantics)) {
return (FALSE);
}
if (!xdr_string(xdrs, &objp->r_nc_protofmly, RPC_MAXDATASIZE)) {
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
return (FALSE);
}
if (!xdr_string(xdrs, &objp->r_nc_proto, RPC_MAXDATASIZE)) {
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
return (FALSE);
}
return (TRUE);
}
bool_t
xdr_rpcb_entry_list_ptr(XDR *xdrs, rpcb_entry_list_ptr *rp)
{
/*
* more_elements is pre-computed in case the direction is
* XDR_ENCODE or XDR_FREE. more_elements is overwritten by
* xdr_bool when the direction is XDR_DECODE.
*/
bool_t more_elements;
int freeing = (xdrs->x_op == XDR_FREE);
rpcb_entry_list_ptr next;
rpcb_entry_list_ptr next_copy;
next = NULL;
for (;;) {
more_elements = (bool_t)(*rp != NULL);
if (! xdr_bool(xdrs, &more_elements)) {
return (FALSE);
}
if (! more_elements) {
return (TRUE); /* we are done */
}
/*
* the unfortunate side effect of non-recursion is that in
* the case of freeing we must remember the next object
* before we free the current object ...
*/
if (freeing)
next = (*rp)->rpcb_entry_next;
if (! xdr_reference(xdrs, (caddr_t *)rp,
(u_int)sizeof (rpcb_entry_list),
(xdrproc_t)xdr_rpcb_entry)) {
return (FALSE);
}
if (freeing && *rp) {
next_copy = next;
rp = &next_copy;
/*
* Note that in the subsequent iteration, next_copy
* gets nulled out by the xdr_reference
* but next itself survives.
*/
} else if (*rp) {
rp = &((*rp)->rpcb_entry_next);
}
}
/*NOTREACHED*/
}