mirror/freebsd
1
0
Fork 0
mirror of https://git.FreeBSD.org/src.git synced 2024-12-20 11:11:24 +00:00
Code Issues Releases Activity
3724136741
freebsd/sys/nfsserver/nfs_serv.c

4069 lines
97 KiB
C
Raw Normal View History

BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Copyright (c) 1989, 1993
* The Regents of the University of California. All rights reserved.
*
* This code is derived from software contributed to Berkeley by
* Rick Macklem at The University of Guelph.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by the University of
* California, Berkeley and its contributors.
* 4. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS 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 REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
NFS Jumbo commit part 1. Cosmetic and structural changes only. The aim of this part of commits is to minimize unnecessary differences between the other NFS's of similar origin. Yes, there are gratuitous changes here that the style folks won't like, but it makes the catch-up less difficult.
1998-05-31 17:27:58 +00:00
* @(#)nfs_serv.c 8.8 (Berkeley) 7/31/95
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
* nfs version 2 and 3 server calls to vnode ops
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
* - these routines generally have 3 phases
* 1 - break down and validate rpc request in mbuf list
* 2 - do the vnode ops for the request
* (surprisingly ?? many are very similar to syscalls in vfs_syscalls.c)
* 3 - build the rpc reply in an mbuf list
* nb:
* - do not mix the phases, since the nfsm_?? macros can return failures
* on a bad rpc or similar and do not do any vrele() or vput()'s
*
* - the nfsm_reply() macro generates an nfs rpc reply with the nfs
* error number iff error != 0 whereas
* returning an error from the server function implies a fatal error
* such as a badly constructed rpc request that should be dropped without
* a reply.
Fix a few NFSv2 issues that slipped in during the big cleanup. The semantics of the nfsm_reply() macro were changed so that the caller has to explicitly handle the V2 error case, whereas before, nfsm_reply() did a `goto nfsmout' then. A few server ops (setattr, readlink, create, mkdir) weren't updated to match, so errors in the V2 case could cause protocol hangs and leaked mbufs. Correct some comments that describe the old nfsm_reply behaviour. [older, harmless nit] Remove the unnecessary `nfsmreply0' label in nfsrv_create(), since for its users, the main `ereply' label does the same thing.
2002-01-12 03:57:25 +00:00
* For nfsm_reply(), the case where error == EBADRPC is treated
* specially; after constructing a reply, it does an immediate
* `goto nfsmout' to avoid getting any V3 post-op status appended.
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
*
* Other notes:
* Warning: always pay careful attention to resource cleanup on return
* and note that nfsm_*() macros can terminate a procedure on certain
* errors.
*
Remove WILLRELE from VOP_SYMLINK Note: Previous commit to these files (except coda_vnops and devfs_vnops) that claimed to remove WILLRELE from VOP_RENAME actually removed it from VOP_MKNOD.
1999-11-13 20:58:17 +00:00
* lookup() and namei()
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
* may return garbage in various structural fields/return elements
* if an error is returned, and may garbage up nd.ni_dvp even if no
* error is returned and you did not request LOCKPARENT or WANTPARENT.
*
* We use the ni_cnd.cn_flags 'HASBUF' flag to track whether the name
Introduce NDFREE (and remove VOP_ABORTOP)
1999-12-15 23:02:35 +00:00
* buffer has been freed or not.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/proc.h>
#include <sys/namei.h>
Eliminated nested include of <sys/unistd.h> in <sys/file.h> in the kernel. Include it directly in the few places where it is used. Reduced some #includes of <sys/file.h> to #includes of <sys/fcntl.h> or nothing.
1996-09-03 14:25:27 +00:00
#include <sys/unistd.h>
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#include <sys/vnode.h>
#include <sys/mount.h>
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
#include <sys/socket.h>
#include <sys/socketvar.h>
Added used #include - don't depend on <sys/mbuf.h> including <sys/malloc.h> (unless we only use the bogusly shared M*WAIT flags).
1997-09-02 01:19:47 +00:00
#include <sys/malloc.h>
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#include <sys/mbuf.h>
#include <sys/dirent.h>
#include <sys/stat.h>
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
#include <sys/kernel.h>
Staticize.
1995-12-17 21:14:36 +00:00
#include <sys/sysctl.h>
Separate the struct bio related stuff out of <sys/buf.h> into <sys/bio.h>. <sys/bio.h> is now a prerequisite for <sys/buf.h> but it shall not be made a nested include according to bdes teachings on the subject of nested includes. Diskdrivers and similar stuff below specfs::strategy() should no longer need to include <sys/buf.> unless they need caching of data. Still a few bogus uses of struct buf to track down. Repocopy by: peter
2000-05-05 09:59:14 +00:00
#include <sys/bio.h>
Asynchronized client-side nfs_commit. NFS commit operations were previously issued synchronously even if async daemons (nfsiod's) were available. The commit has been moved from the strategy code to the doio code in order to asynchronize it. Removed use of lastr in preparation for removal of vnode->v_lastr. It has been replaced with seqcount, which is already supported by the system and, in fact, gives us a better heuristic for sequential detection then lastr ever did. Made major performance improvements to the server side commit. The server previously fsync'd the entire file for each commit rpc. The server now bawrite()s only those buffers related to the offset/size specified in the commit rpc. Note that we do not commit the meta-data yet. This works still needs to be done. Note that a further optimization can be done (and has not yet been done) on the client: we can merge multiple potential commit rpc's into a single rpc with a greater file offset/size range and greatly reduce rpc traffic. Reviewed by: Alan Cox <alc@cs.rice.edu>, David Greenman <dg@root.com>
1999-09-17 05:57:57 +00:00
#include <sys/buf.h>
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#include <vm/vm.h>
Untangled the vm.h include file spaghetti.
1995-12-07 12:48:31 +00:00
#include <vm/vm_extern.h>
Change the busy page mgmt, so that when pages are freed, they MUST be PG_BUSY. It is bogus to free a page that isn't busy, because it is in a state of being "unavailable" when being freed. The additional advantage is that the page_remove code has a better cross-check that the page should be busy and unavailable for other use. There were some minor problems with the collapse code, and this plugs those subtile "holes." Also, the vfs_bio code wasn't checking correctly for PG_BUSY pages. I am going to develop a more consistant scheme for grabbing pages, busy or otherwise. For now, we are stuck with the current morass.
1998-01-31 11:56:53 +00:00
#include <vm/vm_object.h>
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
#include <nfs/nfsproto.h>
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#include <nfs/rpcv2.h>
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
#include <nfsserver/nfs.h>
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#include <nfs/xdr_subs.h>
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
#include <nfsserver/nfsm_subs.h>
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
#ifdef NFSRV_DEBUG
#define nfsdbprintf(info) printf info
#else
#define nfsdbprintf(info)
#endif
Asynchronized client-side nfs_commit. NFS commit operations were previously issued synchronously even if async daemons (nfsiod's) were available. The commit has been moved from the strategy code to the doio code in order to asynchronize it. Removed use of lastr in preparation for removal of vnode->v_lastr. It has been replaced with seqcount, which is already supported by the system and, in fact, gives us a better heuristic for sequential detection then lastr ever did. Made major performance improvements to the server side commit. The server previously fsync'd the entire file for each commit rpc. The server now bawrite()s only those buffers related to the offset/size specified in the commit rpc. Note that we do not commit the meta-data yet. This works still needs to be done. Note that a further optimization can be done (and has not yet been done) on the client: we can merge multiple potential commit rpc's into a single rpc with a greater file offset/size range and greatly reduce rpc traffic. Reviewed by: Alan Cox <alc@cs.rice.edu>, David Greenman <dg@root.com>
1999-09-17 05:57:57 +00:00
#define MAX_COMMIT_COUNT (1024 * 1024)
Add a readahead heuristic to the NFS server side code. While the server cannot unilaterally pass data to a client it can reduce the physical disk transaction overhead by reading larger blocks. This results in better pipelining of requests/responses over the network and an almost 100% increase in cpu efficiency on the server. On a 100BaseTX network NFS read performance increases from 8.5 MBytes/sec to 10 MB/sec (maxed out), and cpu efficiency increases from 72% idle to 80% idle on the server. Reviewed by: Alfred Perlstein <bright@wintelcom.net>
1999-12-13 17:34:45 +00:00
#define NUM_HEURISTIC 64
#define NHUSE_INIT 64
#define NHUSE_INC 16
#define NHUSE_MAX 2048
static struct nfsheur {
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct vnode *nh_vp; /* vp to match (unreferenced pointer) */
off_t nh_nextr; /* next offset for sequential detection */
int nh_use; /* use count for selection */
int nh_seqcount; /* heuristic */
Add a readahead heuristic to the NFS server side code. While the server cannot unilaterally pass data to a client it can reduce the physical disk transaction overhead by reading larger blocks. This results in better pipelining of requests/responses over the network and an almost 100% increase in cpu efficiency on the server. On a 100BaseTX network NFS read performance increases from 8.5 MBytes/sec to 10 MB/sec (maxed out), and cpu efficiency increases from 72% idle to 80% idle on the server. Reviewed by: Alfred Perlstein <bright@wintelcom.net>
1999-12-13 17:34:45 +00:00
} nfsheur[NUM_HEURISTIC];
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/* Global vars */
Add an option NFS_NOSERVER which saves 100K in the install kernel (or any other kernel that uses it). Use with option NFS.
1996-01-13 23:27:58 +00:00
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
int nfsrvw_procrastinate = NFS_GATHERDELAY * 1000;
Implement a separate control for write gathering on NFSv3. This is turned off for NFSv3 by default since write gathering seems to reduce performance for NFSv3 by up to 60%. Add sysctl knobs to control both variables.
1997-05-10 16:59:36 +00:00
int nfsrvw_procrastinate_v3 = 0;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Have NFS use a snapshot of boottime instead of boottime itself to generate the NFSv3 Version id. boottime itself may change, sometimes once every tick if you are running xntpd, which really throws off clients. Clients will tend to throw away what they believe to be stale data too often, and can get into long loops rewriting the same data over and over again because they believe the server has rebooted over and over again due to the changing version id. Approved by: jkh
1999-12-16 17:01:32 +00:00
static struct timeval nfsver = { 0 };
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
SYSCTL_NODE(_vfs, OID_AUTO, nfsrv, CTLFLAG_RW, 0, "NFS server");
* Change sysctl from using linker_set to construct its tree using SLISTs. This makes it possible to change the sysctl tree at runtime. * Change KLD to find and register any sysctl nodes contained in the loaded file and to unregister them when the file is unloaded. Reviewed by: Archie Cobbs <archie@whistle.com>, Peter Wemm <peter@netplex.com.au> (well they looked at it anyway)
1999-02-16 10:49:55 +00:00
Staticize.
1998-02-09 06:11:36 +00:00
static int nfs_async;
Asynchronized client-side nfs_commit. NFS commit operations were previously issued synchronously even if async daemons (nfsiod's) were available. The commit has been moved from the strategy code to the doio code in order to asynchronize it. Removed use of lastr in preparation for removal of vnode->v_lastr. It has been replaced with seqcount, which is already supported by the system and, in fact, gives us a better heuristic for sequential detection then lastr ever did. Made major performance improvements to the server side commit. The server previously fsync'd the entire file for each commit rpc. The server now bawrite()s only those buffers related to the offset/size specified in the commit rpc. Note that we do not commit the meta-data yet. This works still needs to be done. Note that a further optimization can be done (and has not yet been done) on the client: we can merge multiple potential commit rpc's into a single rpc with a greater file offset/size range and greatly reduce rpc traffic. Reviewed by: Alan Cox <alc@cs.rice.edu>, David Greenman <dg@root.com>
1999-09-17 05:57:57 +00:00
static int nfs_commit_blks;
static int nfs_commit_miss;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
SYSCTL_INT(_vfs_nfsrv, OID_AUTO, async, CTLFLAG_RW, &nfs_async, 0, "");
SYSCTL_INT(_vfs_nfsrv, OID_AUTO, commit_blks, CTLFLAG_RW, &nfs_commit_blks, 0, "");
SYSCTL_INT(_vfs_nfsrv, OID_AUTO, commit_miss, CTLFLAG_RW, &nfs_commit_miss, 0, "");
struct nfsrvstats nfsrvstats;
SYSCTL_STRUCT(_vfs_nfsrv, NFS_NFSRVSTATS, nfsrvstats, CTLFLAG_RD,
&nfsrvstats, nfsrvstats, "S,nfsrvstats");
Staticize.
1995-12-17 21:14:36 +00:00
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
static int nfsrv_access(struct vnode *, int, struct ucred *, int,
struct thread *, int);
static void nfsrvw_coalesce(struct nfsrv_descript *,
struct nfsrv_descript *);
Added NFS_ASYNC kernel option. It only has an effect for NFSv2.
1995-08-24 11:39:31 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/*
* Clear nameidata fields that are tested in nsfmout cleanup code prior
* to using first nfsm macro (that might jump to the cleanup code).
*/
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
static __inline void
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
ndclear(struct nameidata *nd)
{
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nd->ni_cnd.cn_flags = 0;
nd->ni_vp = NULL;
nd->ni_dvp = NULL;
nd->ni_startdir = NULL;
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
* nfs v3 access service
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrv3_access(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
struct thread *td, struct mbuf **mrq)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md;
Fix all areas of the system (or at least all those in LINT) to avoid storing socket addresses in mbufs. (Socket buffers are the one exception.) A number of kernel APIs needed to get fixed in order to make this happen. Also, fix three protocol families which kept PCBs in mbufs to not malloc them instead. Delete some old compatibility cruft while we're at it, and add some new routines in the in_cksum family.
1997-08-16 19:16:27 +00:00
struct sockaddr *nam = nfsd->nd_nam;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
caddr_t dpos = nfsd->nd_dpos;
struct ucred *cred = &nfsd->nd_cr;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
struct vnode *vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsfh_t nfh;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhandle_t *fhp;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
u_int32_t *tl;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
caddr_t bpos;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int error = 0, rdonly, getret;
struct mbuf *mb, *mreq;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct vattr vattr, *vap = &vattr;
u_long testmode, nfsmode;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int v3 = (nfsd->nd_flag & ND_NFSV3);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsdbprintf(("%s %d\n", __FILE__, __LINE__));
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (!v3)
panic("nfsrv3_access: v3 proc called on a v2 connection");
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhp = &nfh.fh_generic;
nfsm_srvmtofh(fhp);
Make nfsm_dissect() have an obvious return value.
2001-09-27 22:40:38 +00:00
tl = nfsm_dissect(u_int32_t *, NFSX_UNSIGNED);
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
error = nfsrv_fhtovp(fhp, 1, &vp, cred, slp, nam, &rdonly, TRUE);
NFS Jumbo commit part 1. Cosmetic and structural changes only. The aim of this part of commits is to minimize unnecessary differences between the other NFS's of similar origin. Yes, there are gratuitous changes here that the style folks won't like, but it makes the catch-up less difficult.
1998-05-31 17:27:58 +00:00
if (error) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_UNSIGNED);
Convert old style (type foo *)0 casts to NULLs PR: kern/40360 Requested by: Hiten PAndya via direct email
2002-07-11 17:54:58 +00:00
nfsm_srvpostop_attr(1, NULL);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
For the on-the-wire protocol, u_long -> u_int32_t; long -> int32_t; int -> int32_t; u_short -> u_int16_t. Also, use mode_t instead of u_short for storing modes (mode_t is a u_int16_t). Obtained from: NetBSD
1998-05-31 20:09:01 +00:00
nfsmode = fxdr_unsigned(u_int32_t, *tl);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if ((nfsmode & NFSV3ACCESS_READ) &&
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
nfsrv_access(vp, VREAD, cred, rdonly, td, 0))
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsmode &= ~NFSV3ACCESS_READ;
if (vp->v_type == VDIR)
testmode = (NFSV3ACCESS_MODIFY | NFSV3ACCESS_EXTEND |
NFSV3ACCESS_DELETE);
else
testmode = (NFSV3ACCESS_MODIFY | NFSV3ACCESS_EXTEND);
if ((nfsmode & testmode) &&
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
nfsrv_access(vp, VWRITE, cred, rdonly, td, 0))
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsmode &= ~testmode;
if (vp->v_type == VDIR)
testmode = NFSV3ACCESS_LOOKUP;
else
testmode = NFSV3ACCESS_EXECUTE;
if ((nfsmode & testmode) &&
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
nfsrv_access(vp, VEXEC, cred, rdonly, td, 0))
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsmode &= ~testmode;
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
getret = VOP_GETATTR(vp, vap, cred, td);
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vput(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_POSTOPATTR(1) + NFSX_UNSIGNED);
nfsm_srvpostop_attr(getret, vap);
Tidy up nfsm_build usage. This is only partially finished.
2001-09-27 02:33:36 +00:00
tl = nfsm_build(u_int32_t *, NFSX_UNSIGNED);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
*tl = txdr_unsigned(nfsmode);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsmout:
if (vp)
vput(vp);
return(error);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
/*
* nfs getattr service
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrv_getattr(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
struct thread *td, struct mbuf **mrq)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md;
Fix all areas of the system (or at least all those in LINT) to avoid storing socket addresses in mbufs. (Socket buffers are the one exception.) A number of kernel APIs needed to get fixed in order to make this happen. Also, fix three protocol families which kept PCBs in mbufs to not malloc them instead. Delete some old compatibility cruft while we're at it, and add some new routines in the in_cksum family.
1997-08-16 19:16:27 +00:00
struct sockaddr *nam = nfsd->nd_nam;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
caddr_t dpos = nfsd->nd_dpos;
struct ucred *cred = &nfsd->nd_cr;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct nfs_fattr *fp;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct vattr va;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct vattr *vap = &va;
Fix a number of server-side issues related to aborting badly formed NFS packets, mainly initializing structure pointers to NULL which are conditionally freed prior to return. PR: kern/15249 Submitted by: Ian Dowse <iedowse@maths.tcd.ie>
1999-12-12 07:06:39 +00:00
struct vnode *vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsfh_t nfh;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhandle_t *fhp;
caddr_t bpos;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int error = 0, rdonly;
struct mbuf *mb, *mreq;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsdbprintf(("%s %d\n", __FILE__, __LINE__));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhp = &nfh.fh_generic;
nfsm_srvmtofh(fhp);
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
error = nfsrv_fhtovp(fhp, 1, &vp, cred, slp, nam, &rdonly, TRUE);
NFS Jumbo commit part 1. Cosmetic and structural changes only. The aim of this part of commits is to minimize unnecessary differences between the other NFS's of similar origin. Yes, there are gratuitous changes here that the style folks won't like, but it makes the catch-up less difficult.
1998-05-31 17:27:58 +00:00
if (error) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nfsm_reply(0);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
error = VOP_GETATTR(vp, vap, cred, td);
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vput(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_FATTR(nfsd->nd_flag & ND_NFSV3));
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (error) {
error = 0;
goto nfsmout;
}
Tidy up nfsm_build usage. This is only partially finished.
2001-09-27 02:33:36 +00:00
fp = nfsm_build(struct nfs_fattr *,
NFSX_FATTR(nfsd->nd_flag & ND_NFSV3));
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_srvfillattr(vap, fp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/* fall through */
nfsmout:
if (vp)
vput(vp);
return(error);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
/*
* nfs setattr service
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrv_setattr(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
struct thread *td, struct mbuf **mrq)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md;
Fix all areas of the system (or at least all those in LINT) to avoid storing socket addresses in mbufs. (Socket buffers are the one exception.) A number of kernel APIs needed to get fixed in order to make this happen. Also, fix three protocol families which kept PCBs in mbufs to not malloc them instead. Delete some old compatibility cruft while we're at it, and add some new routines in the in_cksum family.
1997-08-16 19:16:27 +00:00
struct sockaddr *nam = nfsd->nd_nam;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
caddr_t dpos = nfsd->nd_dpos;
struct ucred *cred = &nfsd->nd_cr;
struct vattr va, preat;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct vattr *vap = &va;
struct nfsv2_sattr *sp;
struct nfs_fattr *fp;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
struct vnode *vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsfh_t nfh;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhandle_t *fhp;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
u_int32_t *tl;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
caddr_t bpos;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int error = 0, rdonly, preat_ret = 1, postat_ret = 1;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
int v3 = (nfsd->nd_flag & ND_NFSV3), gcheck = 0;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct mbuf *mb, *mreq;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct timespec guard;
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
struct mount *mp = NULL;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsdbprintf(("%s %d\n", __FILE__, __LINE__));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhp = &nfh.fh_generic;
nfsm_srvmtofh(fhp);
Fix some problems that were introduced in revision 1.97. Instead of returning an error code to the caller, NFS server op routines must themselves build an error reply and return 0 to the caller. This is achieved by replacing the erroneous return statements with code that jumps forward to the op function's reply code. We need to be careful to ensure that the 'struct mount' pointer is NULL though, so that the final vn_finished_write() call becomes a no-op. Reviewed by: mckusick, dillon
2001-02-09 13:24:06 +00:00
if ((mp = vfs_getvfs(&fhp->fh_fsid)) == NULL) {
error = ESTALE;
goto out;
}
if ((error = VFS_FHTOVP(mp, &fhp->fh_fid, &vp)) != 0) {
mp = NULL;
goto out;
}
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
(void) vn_start_write(vp, &mp, V_WAIT);
vput(vp);
vp = NULL;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
VATTR_NULL(vap);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (v3) {
nfsm_srvsattr(vap);
Make nfsm_dissect() have an obvious return value.
2001-09-27 22:40:38 +00:00
tl = nfsm_dissect(u_int32_t *, NFSX_UNSIGNED);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
gcheck = fxdr_unsigned(int, *tl);
if (gcheck) {
Make nfsm_dissect() have an obvious return value.
2001-09-27 22:40:38 +00:00
tl = nfsm_dissect(u_int32_t *, 2 * NFSX_UNSIGNED);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
fxdr_nfsv3time(tl, &guard);
}
} else {
Make nfsm_dissect() have an obvious return value.
2001-09-27 22:40:38 +00:00
sp = nfsm_dissect(struct nfsv2_sattr *, NFSX_V2SATTR);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
/*
* Nah nah nah nah na nah
* There is a bug in the Sun client that puts 0xffff in the mode
* field of sattr when it should put in 0xffffffff. The u_short
* doesn't sign extend.
* --> check the low order 2 bytes for 0xffff
*/
if ((fxdr_unsigned(int, sp->sa_mode) & 0xffff) != 0xffff)
vap->va_mode = nfstov_mode(sp->sa_mode);
Rename some variables that end up shadowing their namesakes in the NFS client code. Reviewed by: peter
2002-01-08 19:41:06 +00:00
if (sp->sa_uid != nfsrv_nfs_xdrneg1)
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
vap->va_uid = fxdr_unsigned(uid_t, sp->sa_uid);
Rename some variables that end up shadowing their namesakes in the NFS client code. Reviewed by: peter
2002-01-08 19:41:06 +00:00
if (sp->sa_gid != nfsrv_nfs_xdrneg1)
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
vap->va_gid = fxdr_unsigned(gid_t, sp->sa_gid);
Rename some variables that end up shadowing their namesakes in the NFS client code. Reviewed by: peter
2002-01-08 19:41:06 +00:00
if (sp->sa_size != nfsrv_nfs_xdrneg1)
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
vap->va_size = fxdr_unsigned(u_quad_t, sp->sa_size);
Rename some variables that end up shadowing their namesakes in the NFS client code. Reviewed by: peter
2002-01-08 19:41:06 +00:00
if (sp->sa_atime.nfsv2_sec != nfsrv_nfs_xdrneg1) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#ifdef notyet
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
fxdr_nfsv2time(&sp->sa_atime, &vap->va_atime);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#else
In sys/time.h, struct timespec is defined as: /* * Structure defined by POSIX.4 to be like a timeval. */ struct timespec { time_t ts_sec; /* seconds */ long ts_nsec; /* and nanoseconds */ }; The correct names of the fields are tv_sec and tv_nsec. Reminded by: James Drobina <jdrobina@infinet.com>
1996-09-19 18:21:32 +00:00
vap->va_atime.tv_sec =
For the on-the-wire protocol, u_long -> u_int32_t; long -> int32_t; int -> int32_t; u_short -> u_int16_t. Also, use mode_t instead of u_short for storing modes (mode_t is a u_int16_t). Obtained from: NetBSD
1998-05-31 20:09:01 +00:00
fxdr_unsigned(int32_t, sp->sa_atime.nfsv2_sec);
In sys/time.h, struct timespec is defined as: /* * Structure defined by POSIX.4 to be like a timeval. */ struct timespec { time_t ts_sec; /* seconds */ long ts_nsec; /* and nanoseconds */ }; The correct names of the fields are tv_sec and tv_nsec. Reminded by: James Drobina <jdrobina@infinet.com>
1996-09-19 18:21:32 +00:00
vap->va_atime.tv_nsec = 0;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#endif
}
Rename some variables that end up shadowing their namesakes in the NFS client code. Reviewed by: peter
2002-01-08 19:41:06 +00:00
if (sp->sa_mtime.nfsv2_sec != nfsrv_nfs_xdrneg1)
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
fxdr_nfsv2time(&sp->sa_mtime, &vap->va_mtime);
}
/*
* Now that we have all the fields, lets do it.
*/
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
error = nfsrv_fhtovp(fhp, 1, &vp, cred, slp, nam, &rdonly, TRUE);
NFS Jumbo commit part 1. Cosmetic and structural changes only. The aim of this part of commits is to minimize unnecessary differences between the other NFS's of similar origin. Yes, there are gratuitous changes here that the style folks won't like, but it makes the catch-up less difficult.
1998-05-31 17:27:58 +00:00
if (error) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(2 * NFSX_UNSIGNED);
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (v3)
nfsm_srvwcc_data(preat_ret, &preat, postat_ret, vap);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/*
* vp now an active resource, pay careful attention to cleanup
*/
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (v3) {
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
error = preat_ret = VOP_GETATTR(vp, &preat, cred, td);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (!error && gcheck &&
In sys/time.h, struct timespec is defined as: /* * Structure defined by POSIX.4 to be like a timeval. */ struct timespec { time_t ts_sec; /* seconds */ long ts_nsec; /* and nanoseconds */ }; The correct names of the fields are tv_sec and tv_nsec. Reminded by: James Drobina <jdrobina@infinet.com>
1996-09-19 18:21:32 +00:00
(preat.va_ctime.tv_sec != guard.tv_sec ||
preat.va_ctime.tv_nsec != guard.tv_nsec))
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
error = NFSERR_NOT_SYNC;
if (error) {
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vput(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_WCCDATA(v3));
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (v3)
nfsm_srvwcc_data(preat_ret, &preat, postat_ret, vap);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
/*
* If the size is being changed write acces is required, otherwise
More s/file system/filesystem/g
2002-05-16 21:28:32 +00:00
* just check for a read only filesystem.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
if (vap->va_size == ((u_quad_t)((quad_t) -1))) {
if (rdonly || (vp->v_mount->mnt_flag & MNT_RDONLY)) {
error = EROFS;
goto out;
}
} else {
if (vp->v_type == VDIR) {
error = EISDIR;
goto out;
NFS Jumbo commit part 1. Cosmetic and structural changes only. The aim of this part of commits is to minimize unnecessary differences between the other NFS's of similar origin. Yes, there are gratuitous changes here that the style folks won't like, but it makes the catch-up less difficult.
1998-05-31 17:27:58 +00:00
} else if ((error = nfsrv_access(vp, VWRITE, cred, rdonly,
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
td, 0)) != 0)
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
goto out;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
error = VOP_SETATTR(vp, vap, cred, td);
postat_ret = VOP_GETATTR(vp, vap, cred, td);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (!error)
error = postat_ret;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
out:
Fix some problems that were introduced in revision 1.97. Instead of returning an error code to the caller, NFS server op routines must themselves build an error reply and return 0 to the caller. This is achieved by replacing the erroneous return statements with code that jumps forward to the op function's reply code. We need to be careful to ensure that the 'struct mount' pointer is NULL though, so that the final vn_finished_write() call becomes a no-op. Reviewed by: mckusick, dillon
2001-02-09 13:24:06 +00:00
if (vp != NULL)
vput(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_WCCORFATTR(v3));
if (v3) {
nfsm_srvwcc_data(preat_ret, &preat, postat_ret, vap);
Fix a few NFSv2 issues that slipped in during the big cleanup. The semantics of the nfsm_reply() macro were changed so that the caller has to explicitly handle the V2 error case, whereas before, nfsm_reply() did a `goto nfsmout' then. A few server ops (setattr, readlink, create, mkdir) weren't updated to match, so errors in the V2 case could cause protocol hangs and leaked mbufs. Correct some comments that describe the old nfsm_reply behaviour. [older, harmless nit] Remove the unnecessary `nfsmreply0' label in nfsrv_create(), since for its users, the main `ereply' label does the same thing.
2002-01-12 03:57:25 +00:00
} else if (!error) {
/* v2 non-error case. */
Tidy up nfsm_build usage. This is only partially finished.
2001-09-27 02:33:36 +00:00
fp = nfsm_build(struct nfs_fattr *, NFSX_V2FATTR);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_srvfillattr(vap, fp);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Fix a few NFSv2 issues that slipped in during the big cleanup. The semantics of the nfsm_reply() macro were changed so that the caller has to explicitly handle the V2 error case, whereas before, nfsm_reply() did a `goto nfsmout' then. A few server ops (setattr, readlink, create, mkdir) weren't updated to match, so errors in the V2 case could cause protocol hangs and leaked mbufs. Correct some comments that describe the old nfsm_reply behaviour. [older, harmless nit] Remove the unnecessary `nfsmreply0' label in nfsrv_create(), since for its users, the main `ereply' label does the same thing.
2002-01-12 03:57:25 +00:00
error = 0;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/* fall through */
nfsmout:
if (vp)
vput(vp);
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
vn_finished_write(mp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
return(error);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
/*
* nfs lookup rpc
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrv_lookup(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
struct thread *td, struct mbuf **mrq)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md;
Fix all areas of the system (or at least all those in LINT) to avoid storing socket addresses in mbufs. (Socket buffers are the one exception.) A number of kernel APIs needed to get fixed in order to make this happen. Also, fix three protocol families which kept PCBs in mbufs to not malloc them instead. Delete some old compatibility cruft while we're at it, and add some new routines in the in_cksum family.
1997-08-16 19:16:27 +00:00
struct sockaddr *nam = nfsd->nd_nam;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
caddr_t dpos = nfsd->nd_dpos;
struct ucred *cred = &nfsd->nd_cr;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct nfs_fattr *fp;
Merge WebNFS changes from NetBSD. Obtained from: NetBSD
1997-07-16 09:06:30 +00:00
struct nameidata nd, ind, *ndp = &nd;
Fix a number of server-side issues related to aborting badly formed NFS packets, mainly initializing structure pointers to NULL which are conditionally freed prior to return. PR: kern/15249 Submitted by: Ian Dowse <iedowse@maths.tcd.ie>
1999-12-12 07:06:39 +00:00
struct vnode *vp, *dirp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsfh_t nfh;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhandle_t *fhp;
caddr_t bpos;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int error = 0, len, dirattr_ret = 1;
Merge WebNFS changes from NetBSD. Obtained from: NetBSD
1997-07-16 09:06:30 +00:00
int v3 = (nfsd->nd_flag & ND_NFSV3), pubflag;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct mbuf *mb, *mreq;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct vattr va, dirattr, *vap = &va;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsdbprintf(("%s %d\n", __FILE__, __LINE__));
ndclear(&nd);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhp = &nfh.fh_generic;
nfsm_srvmtofh(fhp);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_srvnamesiz(len);
Merge WebNFS changes from NetBSD. Obtained from: NetBSD
1997-07-16 09:06:30 +00:00
pubflag = nfs_ispublicfh(fhp);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nd.ni_cnd.cn_cred = cred;
nd.ni_cnd.cn_nameiop = LOOKUP;
nd.ni_cnd.cn_flags = LOCKLEAF | SAVESTART;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
error = nfs_namei(&nd, fhp, len, slp, nam, &md, &dpos,
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
&dirp, td, pubflag);
Merge WebNFS changes from NetBSD. Obtained from: NetBSD
1997-07-16 09:06:30 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/*
* namei failure, only dirp to cleanup. Clear out garbarge from
* structure in case macros jump to nfsmout.
*/
if (error) {
if (dirp) {
if (v3)
dirattr_ret = VOP_GETATTR(dirp, &dirattr, cred,
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
td);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vrele(dirp);
dirp = NULL;
}
nfsm_reply(NFSX_POSTOPATTR(v3));
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (v3)
nfsm_srvpostop_attr(dirattr_ret, &dirattr);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
}
/*
* Locate index file for public filehandle
*
* error is 0 on entry and 0 on exit from this block.
*/
if (pubflag) {
Merge WebNFS changes from NetBSD. Obtained from: NetBSD
1997-07-16 09:06:30 +00:00
if (nd.ni_vp->v_type == VDIR && nfs_pub.np_index != NULL) {
/*
* Setup call to lookup() to see if we can find
* the index file. Arguably, this doesn't belong
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
* in a kernel.. Ugh. If an error occurs, do not
* try to install an index file and then clear the
* error.
*
* When we replace nd with ind and redirect ndp,
* maintenance of ni_startdir and ni_vp shift to
* ind and we have to clean them up in the old nd.
* However, the cnd resource continues to be maintained
* via the original nd. Confused? You aren't alone!
Merge WebNFS changes from NetBSD. Obtained from: NetBSD
1997-07-16 09:06:30 +00:00
*/
ind = nd;
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
VOP_UNLOCK(nd.ni_vp, 0, td);
Merge WebNFS changes from NetBSD. Obtained from: NetBSD
1997-07-16 09:06:30 +00:00
ind.ni_pathlen = strlen(nfs_pub.np_index);
ind.ni_cnd.cn_nameptr = ind.ni_cnd.cn_pnbuf =
nfs_pub.np_index;
ind.ni_startdir = nd.ni_vp;
VREF(ind.ni_startdir);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
Merge WebNFS changes from NetBSD. Obtained from: NetBSD
1997-07-16 09:06:30 +00:00
error = lookup(&ind);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
ind.ni_dvp = NULL;
if (error == 0) {
Merge WebNFS changes from NetBSD. Obtained from: NetBSD
1997-07-16 09:06:30 +00:00
/*
* Found an index file. Get rid of
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
* the old references. transfer nd.ni_vp'
Merge WebNFS changes from NetBSD. Obtained from: NetBSD
1997-07-16 09:06:30 +00:00
*/
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (dirp)
Merge WebNFS changes from NetBSD. Obtained from: NetBSD
1997-07-16 09:06:30 +00:00
vrele(dirp);
dirp = nd.ni_vp;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nd.ni_vp = NULL;
Merge WebNFS changes from NetBSD. Obtained from: NetBSD
1997-07-16 09:06:30 +00:00
vrele(nd.ni_startdir);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nd.ni_startdir = NULL;
Merge WebNFS changes from NetBSD. Obtained from: NetBSD
1997-07-16 09:06:30 +00:00
ndp = &ind;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
}
error = 0;
Merge WebNFS changes from NetBSD. Obtained from: NetBSD
1997-07-16 09:06:30 +00:00
}
/*
* If the public filehandle was used, check that this lookup
* didn't result in a filehandle outside the publicly exported
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
* filesystem. We clear the poor vp here to avoid lockups due
* to NFS I/O.
Merge WebNFS changes from NetBSD. Obtained from: NetBSD
1997-07-16 09:06:30 +00:00
*/
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (ndp->ni_vp->v_mount != nfs_pub.np_mount) {
Merge WebNFS changes from NetBSD. Obtained from: NetBSD
1997-07-16 09:06:30 +00:00
vput(nd.ni_vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nd.ni_vp = NULL;
Merge WebNFS changes from NetBSD. Obtained from: NetBSD
1997-07-16 09:06:30 +00:00
error = EPERM;
}
}
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (dirp) {
if (v3)
dirattr_ret = VOP_GETATTR(dirp, &dirattr, cred,
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
td);
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(dirp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
dirp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
Merge WebNFS changes from NetBSD. Obtained from: NetBSD
1997-07-16 09:06:30 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/*
* Resources at this point:
* ndp->ni_vp may not be NULL
*
*/
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (error) {
nfsm_reply(NFSX_POSTOPATTR(v3));
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (v3)
nfsm_srvpostop_attr(dirattr_ret, &dirattr);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
Merge WebNFS changes from NetBSD. Obtained from: NetBSD
1997-07-16 09:06:30 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/*
* Clear out some resources prior to potentially blocking. This
* is not as critical as ni_dvp resources in other routines, but
* it helps.
*/
Merge WebNFS changes from NetBSD. Obtained from: NetBSD
1997-07-16 09:06:30 +00:00
vrele(ndp->ni_startdir);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
ndp->ni_startdir = NULL;
Introduce NDFREE (and remove VOP_ABORTOP)
1999-12-15 23:02:35 +00:00
NDFREE(&nd, NDF_ONLY_PNBUF);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/*
* Get underlying attribute, then release remaining resources ( for
* the same potential blocking reason ) and reply.
*/
Merge WebNFS changes from NetBSD. Obtained from: NetBSD
1997-07-16 09:06:30 +00:00
vp = ndp->ni_vp;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
bzero((caddr_t)fhp, sizeof(nfh));
fhp->fh_fsid = vp->v_mount->mnt_stat.f_fsid;
Prototyping and general gcc-shutting up. Gcc has one warning now which looks bad, I will get to it eventually, unless somebody beats me to it.
1994-10-02 17:27:07 +00:00
error = VFS_VPTOFH(vp, &fhp->fh_fid);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (!error)
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
error = VOP_GETATTR(vp, vap, cred, td);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vput(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
ndp->ni_vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_SRVFH(v3) + NFSX_POSTOPORFATTR(v3) + NFSX_POSTOPATTR(v3));
Prototyping and general gcc-shutting up. Gcc has one warning now which looks bad, I will get to it eventually, unless somebody beats me to it.
1994-10-02 17:27:07 +00:00
if (error) {
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (v3)
nfsm_srvpostop_attr(dirattr_ret, &dirattr);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_srvfhtom(fhp, v3);
if (v3) {
nfsm_srvpostop_attr(0, vap);
nfsm_srvpostop_attr(dirattr_ret, &dirattr);
} else {
Tidy up nfsm_build usage. This is only partially finished.
2001-09-27 02:33:36 +00:00
fp = nfsm_build(struct nfs_fattr *, NFSX_V2FATTR);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_srvfillattr(vap, fp);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsmout:
if (dirp)
vrele(dirp);
Introduce NDFREE (and remove VOP_ABORTOP)
1999-12-15 23:02:35 +00:00
NDFREE(&nd, NDF_ONLY_PNBUF);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (ndp->ni_startdir)
vrele(ndp->ni_startdir);
if (ndp->ni_vp)
vput(ndp->ni_vp);
return (error);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
/*
* nfs readlink service
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrv_readlink(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
struct thread *td, struct mbuf **mrq)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md;
Fix all areas of the system (or at least all those in LINT) to avoid storing socket addresses in mbufs. (Socket buffers are the one exception.) A number of kernel APIs needed to get fixed in order to make this happen. Also, fix three protocol families which kept PCBs in mbufs to not malloc them instead. Delete some old compatibility cruft while we're at it, and add some new routines in the in_cksum family.
1997-08-16 19:16:27 +00:00
struct sockaddr *nam = nfsd->nd_nam;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
caddr_t dpos = nfsd->nd_dpos;
struct ucred *cred = &nfsd->nd_cr;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct iovec iv[(NFS_MAXPATHLEN+MLEN-1)/MLEN];
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct iovec *ivp = iv;
struct mbuf *mp;
u_int32_t *tl;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
caddr_t bpos;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int error = 0, rdonly, i, tlen, len, getret;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
int v3 = (nfsd->nd_flag & ND_NFSV3);
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct mbuf *mb, *mp3, *nmp, *mreq;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
struct vnode *vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct vattr attr;
nfsfh_t nfh;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhandle_t *fhp;
struct uio io, *uiop = &io;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsdbprintf(("%s %d\n", __FILE__, __LINE__));
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
#ifndef nolint
Convert old style (type foo *)0 casts to NULLs PR: kern/40360 Requested by: Hiten PAndya via direct email
2002-07-11 17:54:58 +00:00
mp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
#endif
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
mp3 = NULL;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhp = &nfh.fh_generic;
nfsm_srvmtofh(fhp);
len = 0;
i = 0;
while (len < NFS_MAXPATHLEN) {
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
MGET(nmp, M_TRYWAIT, MT_DATA);
MCLGET(nmp, M_TRYWAIT);
nmp->m_len = NFSMSIZ(nmp);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (len == 0)
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
mp3 = mp = nmp;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
else {
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
mp->m_next = nmp;
mp = nmp;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if ((len + mp->m_len) > NFS_MAXPATHLEN) {
mp->m_len = NFS_MAXPATHLEN - len;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
len = NFS_MAXPATHLEN;
} else
len += mp->m_len;
ivp->iov_base = mtod(mp, caddr_t);
ivp->iov_len = mp->m_len;
i++;
ivp++;
}
uiop->uio_iov = iv;
uiop->uio_iovcnt = i;
uiop->uio_offset = 0;
uiop->uio_resid = len;
uiop->uio_rw = UIO_READ;
uiop->uio_segflg = UIO_SYSSPACE;
Convert old style (type foo *)0 casts to NULLs PR: kern/40360 Requested by: Hiten PAndya via direct email
2002-07-11 17:54:58 +00:00
uiop->uio_td = NULL;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
error = nfsrv_fhtovp(fhp, 1, &vp, cred, slp, nam, &rdonly, TRUE);
NFS Jumbo commit part 1. Cosmetic and structural changes only. The aim of this part of commits is to minimize unnecessary differences between the other NFS's of similar origin. Yes, there are gratuitous changes here that the style folks won't like, but it makes the catch-up less difficult.
1998-05-31 17:27:58 +00:00
if (error) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(2 * NFSX_UNSIGNED);
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (v3)
Convert old style (type foo *)0 casts to NULLs PR: kern/40360 Requested by: Hiten PAndya via direct email
2002-07-11 17:54:58 +00:00
nfsm_srvpostop_attr(1, NULL);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
if (vp->v_type != VLNK) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (v3)
error = EINVAL;
else
error = ENXIO;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto out;
}
error = VOP_READLINK(vp, uiop, cred);
out:
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
getret = VOP_GETATTR(vp, &attr, cred, td);
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vput(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_POSTOPATTR(v3) + NFSX_UNSIGNED);
Fix a few NFSv2 issues that slipped in during the big cleanup. The semantics of the nfsm_reply() macro were changed so that the caller has to explicitly handle the V2 error case, whereas before, nfsm_reply() did a `goto nfsmout' then. A few server ops (setattr, readlink, create, mkdir) weren't updated to match, so errors in the V2 case could cause protocol hangs and leaked mbufs. Correct some comments that describe the old nfsm_reply behaviour. [older, harmless nit] Remove the unnecessary `nfsmreply0' label in nfsrv_create(), since for its users, the main `ereply' label does the same thing.
2002-01-12 03:57:25 +00:00
if (v3)
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_srvpostop_attr(getret, &attr);
Fix a few NFSv2 issues that slipped in during the big cleanup. The semantics of the nfsm_reply() macro were changed so that the caller has to explicitly handle the V2 error case, whereas before, nfsm_reply() did a `goto nfsmout' then. A few server ops (setattr, readlink, create, mkdir) weren't updated to match, so errors in the V2 case could cause protocol hangs and leaked mbufs. Correct some comments that describe the old nfsm_reply behaviour. [older, harmless nit] Remove the unnecessary `nfsmreply0' label in nfsrv_create(), since for its users, the main `ereply' label does the same thing.
2002-01-12 03:57:25 +00:00
if (error) {
error = 0;
goto nfsmout;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (uiop->uio_resid > 0) {
len -= uiop->uio_resid;
tlen = nfsm_rndup(len);
nfsm_adj(mp3, NFS_MAXPATHLEN-tlen, tlen-len);
}
Tidy up nfsm_build usage. This is only partially finished.
2001-09-27 02:33:36 +00:00
tl = nfsm_build(u_int32_t *, NFSX_UNSIGNED);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*tl = txdr_unsigned(len);
mb->m_next = mp3;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
mp3 = NULL;
nfsmout:
if (mp3)
m_freem(mp3);
if (vp)
vput(vp);
return(error);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
/*
* nfs read service
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrv_read(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
struct thread *td, struct mbuf **mrq)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md;
Fix all areas of the system (or at least all those in LINT) to avoid storing socket addresses in mbufs. (Socket buffers are the one exception.) A number of kernel APIs needed to get fixed in order to make this happen. Also, fix three protocol families which kept PCBs in mbufs to not malloc them instead. Delete some old compatibility cruft while we're at it, and add some new routines in the in_cksum family.
1997-08-16 19:16:27 +00:00
struct sockaddr *nam = nfsd->nd_nam;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
caddr_t dpos = nfsd->nd_dpos;
struct ucred *cred = &nfsd->nd_cr;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct iovec *iv;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct iovec *iv2;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct mbuf *m;
struct nfs_fattr *fp;
u_int32_t *tl;
int i;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
caddr_t bpos;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int error = 0, rdonly, cnt, len, left, siz, tlen, getret;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
int v3 = (nfsd->nd_flag & ND_NFSV3), reqlen;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct mbuf *mb, *mreq;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct mbuf *m2;
Fix a number of server-side issues related to aborting badly formed NFS packets, mainly initializing structure pointers to NULL which are conditionally freed prior to return. PR: kern/15249 Submitted by: Ian Dowse <iedowse@maths.tcd.ie>
1999-12-12 07:06:39 +00:00
struct vnode *vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsfh_t nfh;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhandle_t *fhp;
struct uio io, *uiop = &io;
struct vattr va, *vap = &va;
Add a readahead heuristic to the NFS server side code. While the server cannot unilaterally pass data to a client it can reduce the physical disk transaction overhead by reading larger blocks. This results in better pipelining of requests/responses over the network and an almost 100% increase in cpu efficiency on the server. On a 100BaseTX network NFS read performance increases from 8.5 MBytes/sec to 10 MB/sec (maxed out), and cpu efficiency increases from 72% idle to 80% idle on the server. Reviewed by: Alfred Perlstein <bright@wintelcom.net>
1999-12-13 17:34:45 +00:00
struct nfsheur *nh;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
off_t off;
Add a readahead heuristic to the NFS server side code. While the server cannot unilaterally pass data to a client it can reduce the physical disk transaction overhead by reading larger blocks. This results in better pipelining of requests/responses over the network and an almost 100% increase in cpu efficiency on the server. On a 100BaseTX network NFS read performance increases from 8.5 MBytes/sec to 10 MB/sec (maxed out), and cpu efficiency increases from 72% idle to 80% idle on the server. Reviewed by: Alfred Perlstein <bright@wintelcom.net>
1999-12-13 17:34:45 +00:00
int ioflag = 0;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsdbprintf(("%s %d\n", __FILE__, __LINE__));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhp = &nfh.fh_generic;
nfsm_srvmtofh(fhp);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (v3) {
Make nfsm_dissect() have an obvious return value.
2001-09-27 22:40:38 +00:00
tl = nfsm_dissect(u_int32_t *, 2 * NFSX_UNSIGNED);
Various changes lifted from the OpenBSD cvs tree: txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge. nfs_serv.c: don't call nfsm_adj() with negative values, windows clients could crash servers when doing a readdir of a large directory. nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin loop trying to bind(). Don't clobber a mbuf pointer or we get panics on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf. nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB. Obtained from: OpenBSD
1999-06-05 05:35:03 +00:00
off = fxdr_hyper(tl);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
} else {
Make nfsm_dissect() have an obvious return value.
2001-09-27 22:40:38 +00:00
tl = nfsm_dissect(u_int32_t *, NFSX_UNSIGNED);
For the on-the-wire protocol, u_long -> u_int32_t; long -> int32_t; int -> int32_t; u_short -> u_int16_t. Also, use mode_t instead of u_short for storing modes (mode_t is a u_int16_t). Obtained from: NetBSD
1998-05-31 20:09:01 +00:00
off = (off_t)fxdr_unsigned(u_int32_t, *tl);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
nfsm_srvstrsiz(reqlen, NFS_SRVMAXDATA(nfsd));
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/*
* Reference vp. If an error occurs, vp will be invalid, but we
* have to NULL it just in case. The macros might goto nfsmout
* as well.
*/
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
error = nfsrv_fhtovp(fhp, 1, &vp, cred, slp, nam, &rdonly, TRUE);
NFS Jumbo commit part 1. Cosmetic and structural changes only. The aim of this part of commits is to minimize unnecessary differences between the other NFS's of similar origin. Yes, there are gratuitous changes here that the style folks won't like, but it makes the catch-up less difficult.
1998-05-31 17:27:58 +00:00
if (error) {
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(2 * NFSX_UNSIGNED);
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (v3)
Convert old style (type foo *)0 casts to NULLs PR: kern/40360 Requested by: Hiten PAndya via direct email
2002-07-11 17:54:58 +00:00
nfsm_srvpostop_attr(1, NULL);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (vp->v_type != VREG) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (v3)
error = EINVAL;
else
error = (vp->v_type == VDIR) ? EISDIR : EACCES;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (!error) {
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if ((error = nfsrv_access(vp, VREAD, cred, rdonly, td, 1)) != 0)
error = nfsrv_access(vp, VEXEC, cred, rdonly, td, 1);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
getret = VOP_GETATTR(vp, vap, cred, td);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (!error)
error = getret;
Prototyping and general gcc-shutting up. Gcc has one warning now which looks bad, I will get to it eventually, unless somebody beats me to it.
1994-10-02 17:27:07 +00:00
if (error) {
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vput(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_POSTOPATTR(v3));
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (v3)
nfsm_srvpostop_attr(getret, vap);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Add a readahead heuristic to the NFS server side code. While the server cannot unilaterally pass data to a client it can reduce the physical disk transaction overhead by reading larger blocks. This results in better pipelining of requests/responses over the network and an almost 100% increase in cpu efficiency on the server. On a 100BaseTX network NFS read performance increases from 8.5 MBytes/sec to 10 MB/sec (maxed out), and cpu efficiency increases from 72% idle to 80% idle on the server. Reviewed by: Alfred Perlstein <bright@wintelcom.net>
1999-12-13 17:34:45 +00:00
/*
* Calculate byte count to read
*/
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (off >= vap->va_size)
cnt = 0;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
else if ((off + reqlen) > vap->va_size)
Various changes lifted from the OpenBSD cvs tree: txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge. nfs_serv.c: don't call nfsm_adj() with negative values, windows clients could crash servers when doing a readdir of a large directory. nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin loop trying to bind(). Don't clobber a mbuf pointer or we get panics on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf. nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB. Obtained from: OpenBSD
1999-06-05 05:35:03 +00:00
cnt = vap->va_size - off;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
else
cnt = reqlen;
Add a readahead heuristic to the NFS server side code. While the server cannot unilaterally pass data to a client it can reduce the physical disk transaction overhead by reading larger blocks. This results in better pipelining of requests/responses over the network and an almost 100% increase in cpu efficiency on the server. On a 100BaseTX network NFS read performance increases from 8.5 MBytes/sec to 10 MB/sec (maxed out), and cpu efficiency increases from 72% idle to 80% idle on the server. Reviewed by: Alfred Perlstein <bright@wintelcom.net>
1999-12-13 17:34:45 +00:00
/*
* Calculate seqcount for heuristic
*/
{
int hi;
int try = 4;
/*
* Locate best candidate
*/
Fix compilation warning on alpha when converting pointer to integer to generate hash index. Reviewed by: Andrew Gallatin <gallatin@cs.duke.edu>
1999-12-18 19:20:05 +00:00
hi = ((int)(vm_offset_t)vp / sizeof(struct vnode)) & (NUM_HEURISTIC - 1);
Add a readahead heuristic to the NFS server side code. While the server cannot unilaterally pass data to a client it can reduce the physical disk transaction overhead by reading larger blocks. This results in better pipelining of requests/responses over the network and an almost 100% increase in cpu efficiency on the server. On a 100BaseTX network NFS read performance increases from 8.5 MBytes/sec to 10 MB/sec (maxed out), and cpu efficiency increases from 72% idle to 80% idle on the server. Reviewed by: Alfred Perlstein <bright@wintelcom.net>
1999-12-13 17:34:45 +00:00
nh = &nfsheur[hi];
while (try--) {
if (nfsheur[hi].nh_vp == vp) {
nh = &nfsheur[hi];
break;
}
if (nfsheur[hi].nh_use > 0)
--nfsheur[hi].nh_use;
hi = (hi + 1) & (NUM_HEURISTIC - 1);
if (nfsheur[hi].nh_use < nh->nh_use)
nh = &nfsheur[hi];
}
if (nh->nh_vp != vp) {
nh->nh_vp = vp;
nh->nh_nextr = off;
nh->nh_use = NHUSE_INIT;
if (off == 0)
nh->nh_seqcount = 4;
else
nh->nh_seqcount = 1;
}
/*
* Calculate heuristic
*/
if ((off == 0 && nh->nh_seqcount > 0) || off == nh->nh_nextr) {
Abstract-out the constants for the sequential heuristic. No operational changes. MFC after: 1 day
2002-12-28 20:28:10 +00:00
if (++nh->nh_seqcount > IO_SEQMAX)
nh->nh_seqcount = IO_SEQMAX;
Add a readahead heuristic to the NFS server side code. While the server cannot unilaterally pass data to a client it can reduce the physical disk transaction overhead by reading larger blocks. This results in better pipelining of requests/responses over the network and an almost 100% increase in cpu efficiency on the server. On a 100BaseTX network NFS read performance increases from 8.5 MBytes/sec to 10 MB/sec (maxed out), and cpu efficiency increases from 72% idle to 80% idle on the server. Reviewed by: Alfred Perlstein <bright@wintelcom.net>
1999-12-13 17:34:45 +00:00
} else if (nh->nh_seqcount > 1) {
nh->nh_seqcount = 1;
} else {
nh->nh_seqcount = 0;
}
nh->nh_use += NHUSE_INC;
if (nh->nh_use > NHUSE_MAX)
nh->nh_use = NHUSE_MAX;
Abstract-out the constants for the sequential heuristic. No operational changes. MFC after: 1 day
2002-12-28 20:28:10 +00:00
ioflag |= nh->nh_seqcount << IO_SEQSHIFT;
Add a readahead heuristic to the NFS server side code. While the server cannot unilaterally pass data to a client it can reduce the physical disk transaction overhead by reading larger blocks. This results in better pipelining of requests/responses over the network and an almost 100% increase in cpu efficiency on the server. On a 100BaseTX network NFS read performance increases from 8.5 MBytes/sec to 10 MB/sec (maxed out), and cpu efficiency increases from 72% idle to 80% idle on the server. Reviewed by: Alfred Perlstein <bright@wintelcom.net>
1999-12-13 17:34:45 +00:00
}
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_POSTOPORFATTR(v3) + 3 * NFSX_UNSIGNED+nfsm_rndup(cnt));
if (v3) {
Tidy up nfsm_build usage. This is only partially finished.
2001-09-27 02:33:36 +00:00
tl = nfsm_build(u_int32_t *, NFSX_V3FATTR + 4 * NFSX_UNSIGNED);
Rename some variables that end up shadowing their namesakes in the NFS client code. Reviewed by: peter
2002-01-08 19:41:06 +00:00
*tl++ = nfsrv_nfs_true;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
fp = (struct nfs_fattr *)tl;
For the on-the-wire protocol, u_long -> u_int32_t; long -> int32_t; int -> int32_t; u_short -> u_int16_t. Also, use mode_t instead of u_short for storing modes (mode_t is a u_int16_t). Obtained from: NetBSD
1998-05-31 20:09:01 +00:00
tl += (NFSX_V3FATTR / sizeof (u_int32_t));
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
} else {
Tidy up nfsm_build usage. This is only partially finished.
2001-09-27 02:33:36 +00:00
tl = nfsm_build(u_int32_t *, NFSX_V2FATTR + NFSX_UNSIGNED);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
fp = (struct nfs_fattr *)tl;
For the on-the-wire protocol, u_long -> u_int32_t; long -> int32_t; int -> int32_t; u_short -> u_int16_t. Also, use mode_t instead of u_short for storing modes (mode_t is a u_int16_t). Obtained from: NetBSD
1998-05-31 20:09:01 +00:00
tl += (NFSX_V2FATTR / sizeof (u_int32_t));
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
Various changes lifted from the OpenBSD cvs tree: txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge. nfs_serv.c: don't call nfsm_adj() with negative values, windows clients could crash servers when doing a readdir of a large directory. nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin loop trying to bind(). Don't clobber a mbuf pointer or we get panics on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf. nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB. Obtained from: OpenBSD
1999-06-05 05:35:03 +00:00
len = left = nfsm_rndup(cnt);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (cnt > 0) {
/*
* Generate the mbuf list with the uio_iov ref. to it.
*/
i = 0;
m = m2 = mb;
while (left > 0) {
siz = min(M_TRAILINGSPACE(m), left);
if (siz > 0) {
left -= siz;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
i++;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
if (left > 0) {
* Rename M_WAIT mbuf subsystem flag to M_TRYWAIT. This is because calls with M_WAIT (now M_TRYWAIT) may not wait forever when nothing is available for allocation, and may end up returning NULL. Hopefully we now communicate more of the right thing to developers and make it very clear that it's necessary to check whether calls with M_(TRY)WAIT also resulted in a failed allocation. M_TRYWAIT basically means "try harder, block if necessary, but don't necessarily wait forever." The time spent blocking is tunable with the kern.ipc.mbuf_wait sysctl. M_WAIT is now deprecated but still defined for the next little while. * Fix a typo in a comment in mbuf.h * Fix some code that was actually passing the mbuf subsystem's M_WAIT to malloc(). Made it pass M_WAITOK instead. If we were ever to redefine the value of the M_WAIT flag, this could have became a big problem.
2000-12-21 21:44:31 +00:00
MGET(m, M_TRYWAIT, MT_DATA);
MCLGET(m, M_TRYWAIT);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
m->m_len = 0;
m2->m_next = m;
m2 = m;
}
}
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
MALLOC(iv, struct iovec *, i * sizeof (struct iovec),
M_TEMP, M_WAITOK);
uiop->uio_iov = iv2 = iv;
m = mb;
Various changes lifted from the OpenBSD cvs tree: txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge. nfs_serv.c: don't call nfsm_adj() with negative values, windows clients could crash servers when doing a readdir of a large directory. nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin loop trying to bind(). Don't clobber a mbuf pointer or we get panics on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf. nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB. Obtained from: OpenBSD
1999-06-05 05:35:03 +00:00
left = len;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
i = 0;
while (left > 0) {
if (m == NULL)
panic("nfsrv_read iov");
siz = min(M_TRAILINGSPACE(m), left);
if (siz > 0) {
iv->iov_base = mtod(m, caddr_t) + m->m_len;
iv->iov_len = siz;
m->m_len += siz;
left -= siz;
iv++;
i++;
}
m = m->m_next;
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
uiop->uio_iovcnt = i;
uiop->uio_offset = off;
Various changes lifted from the OpenBSD cvs tree: txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge. nfs_serv.c: don't call nfsm_adj() with negative values, windows clients could crash servers when doing a readdir of a large directory. nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin loop trying to bind(). Don't clobber a mbuf pointer or we get panics on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf. nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB. Obtained from: OpenBSD
1999-06-05 05:35:03 +00:00
uiop->uio_resid = len;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
uiop->uio_rw = UIO_READ;
uiop->uio_segflg = UIO_SYSSPACE;
Add a readahead heuristic to the NFS server side code. While the server cannot unilaterally pass data to a client it can reduce the physical disk transaction overhead by reading larger blocks. This results in better pipelining of requests/responses over the network and an almost 100% increase in cpu efficiency on the server. On a 100BaseTX network NFS read performance increases from 8.5 MBytes/sec to 10 MB/sec (maxed out), and cpu efficiency increases from 72% idle to 80% idle on the server. Reviewed by: Alfred Perlstein <bright@wintelcom.net>
1999-12-13 17:34:45 +00:00
error = VOP_READ(vp, uiop, IO_NODELOCKED | ioflag, cred);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
off = uiop->uio_offset;
Add a readahead heuristic to the NFS server side code. While the server cannot unilaterally pass data to a client it can reduce the physical disk transaction overhead by reading larger blocks. This results in better pipelining of requests/responses over the network and an almost 100% increase in cpu efficiency on the server. On a 100BaseTX network NFS read performance increases from 8.5 MBytes/sec to 10 MB/sec (maxed out), and cpu efficiency increases from 72% idle to 80% idle on the server. Reviewed by: Alfred Perlstein <bright@wintelcom.net>
1999-12-13 17:34:45 +00:00
nh->nh_nextr = off;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
FREE((caddr_t)iv2, M_TEMP);
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
if (error || (getret = VOP_GETATTR(vp, vap, cred, td))) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (!error)
error = getret;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
m_freem(mreq);
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vput(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_POSTOPATTR(v3));
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (v3)
nfsm_srvpostop_attr(getret, vap);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
} else {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
uiop->uio_resid = 0;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
}
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vput(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_srvfillattr(vap, fp);
Various changes lifted from the OpenBSD cvs tree: txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge. nfs_serv.c: don't call nfsm_adj() with negative values, windows clients could crash servers when doing a readdir of a large directory. nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin loop trying to bind(). Don't clobber a mbuf pointer or we get panics on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf. nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB. Obtained from: OpenBSD
1999-06-05 05:35:03 +00:00
tlen = len - uiop->uio_resid;
cnt = cnt < tlen ? cnt : tlen;
tlen = nfsm_rndup(cnt);
if (len != tlen || tlen != cnt)
nfsm_adj(mb, len - tlen, tlen - cnt);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (v3) {
Various changes lifted from the OpenBSD cvs tree: txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge. nfs_serv.c: don't call nfsm_adj() with negative values, windows clients could crash servers when doing a readdir of a large directory. nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin loop trying to bind(). Don't clobber a mbuf pointer or we get panics on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf. nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB. Obtained from: OpenBSD
1999-06-05 05:35:03 +00:00
*tl++ = txdr_unsigned(cnt);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (len < reqlen)
Rename some variables that end up shadowing their namesakes in the NFS client code. Reviewed by: peter
2002-01-08 19:41:06 +00:00
*tl++ = nfsrv_nfs_true;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
else
Rename some variables that end up shadowing their namesakes in the NFS client code. Reviewed by: peter
2002-01-08 19:41:06 +00:00
*tl++ = nfsrv_nfs_false;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
Various changes lifted from the OpenBSD cvs tree: txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge. nfs_serv.c: don't call nfsm_adj() with negative values, windows clients could crash servers when doing a readdir of a large directory. nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin loop trying to bind(). Don't clobber a mbuf pointer or we get panics on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf. nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB. Obtained from: OpenBSD
1999-06-05 05:35:03 +00:00
*tl = txdr_unsigned(cnt);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsmout:
if (vp)
vput(vp);
return(error);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
/*
* nfs write service
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrv_write(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
struct thread *td, struct mbuf **mrq)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md;
Fix all areas of the system (or at least all those in LINT) to avoid storing socket addresses in mbufs. (Socket buffers are the one exception.) A number of kernel APIs needed to get fixed in order to make this happen. Also, fix three protocol families which kept PCBs in mbufs to not malloc them instead. Delete some old compatibility cruft while we're at it, and add some new routines in the in_cksum family.
1997-08-16 19:16:27 +00:00
struct sockaddr *nam = nfsd->nd_nam;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
caddr_t dpos = nfsd->nd_dpos;
struct ucred *cred = &nfsd->nd_cr;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct iovec *ivp;
int i, cnt;
struct mbuf *mp;
struct nfs_fattr *fp;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct iovec *iv;
struct vattr va, forat;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct vattr *vap = &va;
u_int32_t *tl;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
caddr_t bpos;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int error = 0, rdonly, len, forat_ret = 1;
Fix some problems that were introduced in revision 1.97. Instead of returning an error code to the caller, NFS server op routines must themselves build an error reply and return 0 to the caller. This is achieved by replacing the erroneous return statements with code that jumps forward to the op function's reply code. We need to be careful to ensure that the 'struct mount' pointer is NULL though, so that the final vn_finished_write() call becomes a no-op. Reviewed by: mckusick, dillon
2001-02-09 13:24:06 +00:00
int ioflags, aftat_ret = 1, retlen = 0, zeroing, adjust;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
int stable = NFSV3WRITE_FILESYNC;
int v3 = (nfsd->nd_flag & ND_NFSV3);
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct mbuf *mb, *mreq;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
struct vnode *vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsfh_t nfh;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhandle_t *fhp;
struct uio io, *uiop = &io;
off_t off;
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
struct mount *mntp = NULL;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsdbprintf(("%s %d\n", __FILE__, __LINE__));
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (mrep == NULL) {
*mrq = NULL;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhp = &nfh.fh_generic;
nfsm_srvmtofh(fhp);
Fix some problems that were introduced in revision 1.97. Instead of returning an error code to the caller, NFS server op routines must themselves build an error reply and return 0 to the caller. This is achieved by replacing the erroneous return statements with code that jumps forward to the op function's reply code. We need to be careful to ensure that the 'struct mount' pointer is NULL though, so that the final vn_finished_write() call becomes a no-op. Reviewed by: mckusick, dillon
2001-02-09 13:24:06 +00:00
if ((mntp = vfs_getvfs(&fhp->fh_fsid)) == NULL) {
error = ESTALE;
goto ereply;
}
if ((error = VFS_FHTOVP(mntp, &fhp->fh_fid, &vp)) != 0) {
mntp = NULL;
goto ereply;
}
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
(void) vn_start_write(vp, &mntp, V_WAIT);
vput(vp);
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (v3) {
Make nfsm_dissect() have an obvious return value.
2001-09-27 22:40:38 +00:00
tl = nfsm_dissect(u_int32_t *, 5 * NFSX_UNSIGNED);
Various changes lifted from the OpenBSD cvs tree: txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge. nfs_serv.c: don't call nfsm_adj() with negative values, windows clients could crash servers when doing a readdir of a large directory. nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin loop trying to bind(). Don't clobber a mbuf pointer or we get panics on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf. nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB. Obtained from: OpenBSD
1999-06-05 05:35:03 +00:00
off = fxdr_hyper(tl);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
tl += 3;
stable = fxdr_unsigned(int, *tl++);
} else {
Make nfsm_dissect() have an obvious return value.
2001-09-27 22:40:38 +00:00
tl = nfsm_dissect(u_int32_t *, 4 * NFSX_UNSIGNED);
For the on-the-wire protocol, u_long -> u_int32_t; long -> int32_t; int -> int32_t; u_short -> u_int16_t. Also, use mode_t instead of u_short for storing modes (mode_t is a u_int16_t). Obtained from: NetBSD
1998-05-31 20:09:01 +00:00
off = (off_t)fxdr_unsigned(u_int32_t, *++tl);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
tl += 2;
Added NFS_ASYNC kernel option. It only has an effect for NFSv2.
1995-08-24 11:39:31 +00:00
if (nfs_async)
stable = NFSV3WRITE_UNSTABLE;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
For the on-the-wire protocol, u_long -> u_int32_t; long -> int32_t; int -> int32_t; u_short -> u_int16_t. Also, use mode_t instead of u_short for storing modes (mode_t is a u_int16_t). Obtained from: NetBSD
1998-05-31 20:09:01 +00:00
retlen = len = fxdr_unsigned(int32_t, *tl);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
cnt = i = 0;
/*
* For NFS Version 2, it is not obvious what a write of zero length
* should do, but I might as well be consistent with Version 3,
* which is to return ok so long as there are no permission problems.
*/
if (len > 0) {
zeroing = 1;
mp = mrep;
while (mp) {
if (mp == md) {
zeroing = 0;
adjust = dpos - mtod(mp, caddr_t);
mp->m_len -= adjust;
if (mp->m_len > 0 && adjust > 0)
Unwind some more macros. NFSMADV() was kinda silly since it was right next to equivalent m_len adjustments. Move the nfsm_subs.h macros into groups depending on which phase they are used in, since that affects the error recovery requirements. Collect some of the common error checking into a single macro as preparation for unwinding some more. Have nfs_rephead return a value instead of secretly modifying args. Remove some unused function arguments that were being passed around. Clarify nfsm_reply()'s error handling (I hope).
2001-09-28 04:37:08 +00:00
mp->m_data += adjust;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (zeroing)
mp->m_len = 0;
else if (mp->m_len > 0) {
i += mp->m_len;
if (i > len) {
mp->m_len -= (i - len);
zeroing = 1;
}
if (mp->m_len > 0)
cnt++;
}
mp = mp->m_next;
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (len > NFS_MAXDATA || len < 0 || i < len) {
error = EIO;
nfsm_reply(2 * NFSX_UNSIGNED);
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (v3)
nfsm_srvwcc_data(forat_ret, &forat, aftat_ret, vap);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
error = nfsrv_fhtovp(fhp, 1, &vp, cred, slp, nam, &rdonly, TRUE);
NFS Jumbo commit part 1. Cosmetic and structural changes only. The aim of this part of commits is to minimize unnecessary differences between the other NFS's of similar origin. Yes, there are gratuitous changes here that the style folks won't like, but it makes the catch-up less difficult.
1998-05-31 17:27:58 +00:00
if (error) {
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(2 * NFSX_UNSIGNED);
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (v3)
nfsm_srvwcc_data(forat_ret, &forat, aftat_ret, vap);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
if (v3)
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
forat_ret = VOP_GETATTR(vp, &forat, cred, td);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (vp->v_type != VREG) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (v3)
error = EINVAL;
else
error = (vp->v_type == VDIR) ? EISDIR : EACCES;
}
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (!error)
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
error = nfsrv_access(vp, VWRITE, cred, rdonly, td, 1);
Prototyping and general gcc-shutting up. Gcc has one warning now which looks bad, I will get to it eventually, unless somebody beats me to it.
1994-10-02 17:27:07 +00:00
if (error) {
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vput(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_WCCDATA(v3));
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (v3)
nfsm_srvwcc_data(forat_ret, &forat, aftat_ret, vap);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (len > 0) {
MALLOC(ivp, struct iovec *, cnt * sizeof (struct iovec), M_TEMP,
M_WAITOK);
uiop->uio_iov = iv = ivp;
uiop->uio_iovcnt = cnt;
mp = mrep;
while (mp) {
if (mp->m_len > 0) {
ivp->iov_base = mtod(mp, caddr_t);
ivp->iov_len = mp->m_len;
ivp++;
}
mp = mp->m_next;
}
/*
* XXX
* The IO_METASYNC flag indicates that all metadata (and not just
* enough to ensure data integrity) mus be written to stable storage
* synchronously.
* (IO_METASYNC is not yet implemented in 4.4BSD-Lite.)
*/
if (stable == NFSV3WRITE_UNSTABLE)
ioflags = IO_NODELOCKED;
else if (stable == NFSV3WRITE_DATASYNC)
ioflags = (IO_SYNC | IO_NODELOCKED);
else
ioflags = (IO_METASYNC | IO_SYNC | IO_NODELOCKED);
uiop->uio_resid = len;
uiop->uio_rw = UIO_WRITE;
uiop->uio_segflg = UIO_SYSSPACE;
Convert old style (type foo *)0 casts to NULLs PR: kern/40360 Requested by: Hiten PAndya via direct email
2002-07-11 17:54:58 +00:00
uiop->uio_td = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
uiop->uio_offset = off;
error = VOP_WRITE(vp, uiop, ioflags, cred);
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrvstats.srvvop_writes++;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
FREE((caddr_t)iv, M_TEMP);
}
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
aftat_ret = VOP_GETATTR(vp, vap, cred, td);
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vput(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (!error)
error = aftat_ret;
Fix some problems that were introduced in revision 1.97. Instead of returning an error code to the caller, NFS server op routines must themselves build an error reply and return 0 to the caller. This is achieved by replacing the erroneous return statements with code that jumps forward to the op function's reply code. We need to be careful to ensure that the 'struct mount' pointer is NULL though, so that the final vn_finished_write() call becomes a no-op. Reviewed by: mckusick, dillon
2001-02-09 13:24:06 +00:00
ereply:
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_PREOPATTR(v3) + NFSX_POSTOPORFATTR(v3) +
2 * NFSX_UNSIGNED + NFSX_WRITEVERF(v3));
if (v3) {
nfsm_srvwcc_data(forat_ret, &forat, aftat_ret, vap);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (error) {
error = 0;
goto nfsmout;
}
Tidy up nfsm_build usage. This is only partially finished.
2001-09-27 02:33:36 +00:00
tl = nfsm_build(u_int32_t *, 4 * NFSX_UNSIGNED);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
*tl++ = txdr_unsigned(retlen);
Implement the async mount option for NFSv3. This makes NFS pretend that all writes sent to the server were synchronous and therefore no commits are needed. This is the same as the vfs.nfs.async variable on the server but allows each client to choose whether to work this way. Also make the vfs.nfs.async variable do the 'right' thing for NFSv3, i.e. pretend that the write was synchronous.
1997-06-03 13:56:55 +00:00
/*
* If nfs_async is set, then pretend the write was FILESYNC.
*/
if (stable == NFSV3WRITE_UNSTABLE && !nfs_async)
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
*tl++ = txdr_unsigned(stable);
else
*tl++ = txdr_unsigned(NFSV3WRITE_FILESYNC);
/*
* Actually, there is no need to txdr these fields,
* but it may make the values more human readable,
* for debugging purposes.
*/
Have NFS use a snapshot of boottime instead of boottime itself to generate the NFSv3 Version id. boottime itself may change, sometimes once every tick if you are running xntpd, which really throws off clients. Clients will tend to throw away what they believe to be stale data too often, and can get into long loops rewriting the same data over and over again because they believe the server has rebooted over and over again due to the changing version id. Approved by: jkh
1999-12-16 17:01:32 +00:00
if (nfsver.tv_sec == 0)
nfsver = boottime;
*tl++ = txdr_unsigned(nfsver.tv_sec);
*tl = txdr_unsigned(nfsver.tv_usec);
Fix a few NFSv2 issues that slipped in during the big cleanup. The semantics of the nfsm_reply() macro were changed so that the caller has to explicitly handle the V2 error case, whereas before, nfsm_reply() did a `goto nfsmout' then. A few server ops (setattr, readlink, create, mkdir) weren't updated to match, so errors in the V2 case could cause protocol hangs and leaked mbufs. Correct some comments that describe the old nfsm_reply behaviour. [older, harmless nit] Remove the unnecessary `nfsmreply0' label in nfsrv_create(), since for its users, the main `ereply' label does the same thing.
2002-01-12 03:57:25 +00:00
} else if (!error) {
/* v2 non-error case. */
Tidy up nfsm_build usage. This is only partially finished.
2001-09-27 02:33:36 +00:00
fp = nfsm_build(struct nfs_fattr *, NFSX_V2FATTR);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_srvfillattr(vap, fp);
}
Fix a few NFSv2 issues that slipped in during the big cleanup. The semantics of the nfsm_reply() macro were changed so that the caller has to explicitly handle the V2 error case, whereas before, nfsm_reply() did a `goto nfsmout' then. A few server ops (setattr, readlink, create, mkdir) weren't updated to match, so errors in the V2 case could cause protocol hangs and leaked mbufs. Correct some comments that describe the old nfsm_reply behaviour. [older, harmless nit] Remove the unnecessary `nfsmreply0' label in nfsrv_create(), since for its users, the main `ereply' label does the same thing.
2002-01-12 03:57:25 +00:00
error = 0;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsmout:
if (vp)
vput(vp);
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
vn_finished_write(mntp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
return(error);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
/*
* NFS write service with write gathering support. Called when
* nfsrvw_procrastinate > 0.
* See: Chet Juszczak, "Improving the Write Performance of an NFS Server",
* in Proc. of the Winter 1994 Usenix Conference, pg. 247-259, San Franscisco,
* Jan. 1994.
*/
int
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrv_writegather(struct nfsrv_descript **ndp, struct nfssvc_sock *slp,
struct thread *td, struct mbuf **mrq)
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
{
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct iovec *ivp;
struct mbuf *mp;
struct nfsrv_descript *wp, *nfsd, *owp, *swp;
struct nfs_fattr *fp;
int i;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct iovec *iov;
struct nfsrvw_delayhash *wpp;
struct ucred *cred;
struct vattr va, forat;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
u_int32_t *tl;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
caddr_t bpos, dpos;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int error = 0, rdonly, len, forat_ret = 1;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
int ioflags, aftat_ret = 1, s, adjust, v3, zeroing;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct mbuf *mb, *mreq, *mrep, *md;
Fix a number of server-side issues related to aborting badly formed NFS packets, mainly initializing structure pointers to NULL which are conditionally freed prior to return. PR: kern/15249 Submitted by: Ian Dowse <iedowse@maths.tcd.ie>
1999-12-12 07:06:39 +00:00
struct vnode *vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct uio io, *uiop = &io;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
u_quad_t cur_usec;
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
struct mount *mntp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsdbprintf(("%s %d\n", __FILE__, __LINE__));
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
#ifndef nolint
i = 0;
len = 0;
#endif
*mrq = NULL;
if (*ndp) {
nfsd = *ndp;
*ndp = NULL;
mrep = nfsd->nd_mrep;
md = nfsd->nd_md;
dpos = nfsd->nd_dpos;
cred = &nfsd->nd_cr;
v3 = (nfsd->nd_flag & ND_NFSV3);
LIST_INIT(&nfsd->nd_coalesce);
nfsd->nd_mreq = NULL;
nfsd->nd_stable = NFSV3WRITE_FILESYNC;
Eradicate the variable "time" from the kernel, using various measures. "time" wasn't a atomic variable, so splfoo() protection were needed around any access to it, unless you just wanted the seconds part. Most uses of time.tv_sec now uses the new variable time_second instead. gettime() changed to getmicrotime(0. Remove a couple of unneeded splfoo() protections, the new getmicrotime() is atomic, (until Bruce sets a breakpoint in it). A couple of places needed random data, so use read_random() instead of mucking about with time which isn't random. Add a new nfs_curusec() function. Mark a couple of bogosities involving the now disappeard time variable. Update ffs_update() to avoid the weird "== &time" checks, by fixing the one remaining call that passwd &time as args. Change profiling in ncr.c to use ticks instead of time. Resolution is the same. Add new function "tvtohz()" to avoid the bogus "splfoo(), add time, call hzto() which subtracts time" sequences. Reviewed by: bde
1998-03-30 09:56:58 +00:00
cur_usec = nfs_curusec();
Implement a separate control for write gathering on NFSv3. This is turned off for NFSv3 by default since write gathering seems to reduce performance for NFSv3 by up to 60%. Add sysctl knobs to control both variables.
1997-05-10 16:59:36 +00:00
nfsd->nd_time = cur_usec +
(v3 ? nfsrvw_procrastinate_v3 : nfsrvw_procrastinate);
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
/*
* Now, get the write header..
*/
nfsm_srvmtofh(&nfsd->nd_fh);
if (v3) {
Make nfsm_dissect() have an obvious return value.
2001-09-27 22:40:38 +00:00
tl = nfsm_dissect(u_int32_t *, 5 * NFSX_UNSIGNED);
Various changes lifted from the OpenBSD cvs tree: txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge. nfs_serv.c: don't call nfsm_adj() with negative values, windows clients could crash servers when doing a readdir of a large directory. nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin loop trying to bind(). Don't clobber a mbuf pointer or we get panics on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf. nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB. Obtained from: OpenBSD
1999-06-05 05:35:03 +00:00
nfsd->nd_off = fxdr_hyper(tl);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
tl += 3;
nfsd->nd_stable = fxdr_unsigned(int, *tl++);
} else {
Make nfsm_dissect() have an obvious return value.
2001-09-27 22:40:38 +00:00
tl = nfsm_dissect(u_int32_t *, 4 * NFSX_UNSIGNED);
For the on-the-wire protocol, u_long -> u_int32_t; long -> int32_t; int -> int32_t; u_short -> u_int16_t. Also, use mode_t instead of u_short for storing modes (mode_t is a u_int16_t). Obtained from: NetBSD
1998-05-31 20:09:01 +00:00
nfsd->nd_off = (off_t)fxdr_unsigned(u_int32_t, *++tl);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
tl += 2;
Added NFS_ASYNC kernel option. It only has an effect for NFSv2.
1995-08-24 11:39:31 +00:00
if (nfs_async)
nfsd->nd_stable = NFSV3WRITE_UNSTABLE;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
For the on-the-wire protocol, u_long -> u_int32_t; long -> int32_t; int -> int32_t; u_short -> u_int16_t. Also, use mode_t instead of u_short for storing modes (mode_t is a u_int16_t). Obtained from: NetBSD
1998-05-31 20:09:01 +00:00
len = fxdr_unsigned(int32_t, *tl);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsd->nd_len = len;
nfsd->nd_eoff = nfsd->nd_off + len;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
/*
* Trim the header out of the mbuf list and trim off any trailing
* junk so that the mbuf list has only the write data.
*/
zeroing = 1;
i = 0;
mp = mrep;
while (mp) {
if (mp == md) {
zeroing = 0;
adjust = dpos - mtod(mp, caddr_t);
mp->m_len -= adjust;
if (mp->m_len > 0 && adjust > 0)
Unwind some more macros. NFSMADV() was kinda silly since it was right next to equivalent m_len adjustments. Move the nfsm_subs.h macros into groups depending on which phase they are used in, since that affects the error recovery requirements. Collect some of the common error checking into a single macro as preparation for unwinding some more. Have nfs_rephead return a value instead of secretly modifying args. Remove some unused function arguments that were being passed around. Clarify nfsm_reply()'s error handling (I hope).
2001-09-28 04:37:08 +00:00
mp->m_data += adjust;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
if (zeroing)
mp->m_len = 0;
else {
i += mp->m_len;
if (i > len) {
mp->m_len -= (i - len);
zeroing = 1;
}
}
mp = mp->m_next;
}
if (len > NFS_MAXDATA || len < 0 || i < len) {
nfsmout:
m_freem(mrep);
error = EIO;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsm_writereply(2 * NFSX_UNSIGNED);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (v3)
nfsm_srvwcc_data(forat_ret, &forat, aftat_ret, &va);
nfsd->nd_mreq = mreq;
nfsd->nd_mrep = NULL;
nfsd->nd_time = 0;
}
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
/*
* Add this entry to the hash and time queues.
*/
s = splsoftclock();
owp = NULL;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
wp = LIST_FIRST(&slp->ns_tq);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
while (wp && wp->nd_time < nfsd->nd_time) {
owp = wp;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
wp = LIST_NEXT(wp, nd_tq);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
Fix a nasty hang connected with write gathering. Also add debug print statements to bits of the server which helped me find the hang.
1997-05-10 16:12:03 +00:00
NFS_DPF(WG, ("Q%03x", nfsd->nd_retxid & 0xfff));
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (owp) {
LIST_INSERT_AFTER(owp, nfsd, nd_tq);
} else {
LIST_INSERT_HEAD(&slp->ns_tq, nfsd, nd_tq);
}
if (nfsd->nd_mrep) {
wpp = NWDELAYHASH(slp, nfsd->nd_fh.fh_fid.fid_data);
owp = NULL;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
wp = LIST_FIRST(wpp);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
while (wp &&
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
bcmp((caddr_t)&nfsd->nd_fh,(caddr_t)&wp->nd_fh, NFSX_V3FH)){
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
owp = wp;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
wp = LIST_NEXT(wp, nd_hash);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
while (wp && wp->nd_off < nfsd->nd_off &&
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
!bcmp((caddr_t)&nfsd->nd_fh,(caddr_t)&wp->nd_fh, NFSX_V3FH)) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
owp = wp;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
wp = LIST_NEXT(wp, nd_hash);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
if (owp) {
LIST_INSERT_AFTER(owp, nfsd, nd_hash);
/*
* Search the hash list for overlapping entries and
* coalesce.
*/
for(; nfsd && NFSW_CONTIG(owp, nfsd); nfsd = wp) {
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
wp = LIST_NEXT(nfsd, nd_hash);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (NFSW_SAMECRED(owp, nfsd))
nfsrvw_coalesce(owp, nfsd);
}
} else {
LIST_INSERT_HEAD(wpp, nfsd, nd_hash);
}
}
splx(s);
}
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
* Now, do VOP_WRITE()s for any one(s) that need to be done now
* and generate the associated reply mbuf list(s).
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
loop1:
Eradicate the variable "time" from the kernel, using various measures. "time" wasn't a atomic variable, so splfoo() protection were needed around any access to it, unless you just wanted the seconds part. Most uses of time.tv_sec now uses the new variable time_second instead. gettime() changed to getmicrotime(0. Remove a couple of unneeded splfoo() protections, the new getmicrotime() is atomic, (until Bruce sets a breakpoint in it). A couple of places needed random data, so use read_random() instead of mucking about with time which isn't random. Add a new nfs_curusec() function. Mark a couple of bogosities involving the now disappeard time variable. Update ffs_update() to avoid the weird "== &time" checks, by fixing the one remaining call that passwd &time as args. Change profiling in ncr.c to use ticks instead of time. Resolution is the same. Add new function "tvtohz()" to avoid the bogus "splfoo(), add time, call hzto() which subtracts time" sequences. Reviewed by: bde
1998-03-30 09:56:58 +00:00
cur_usec = nfs_curusec();
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
s = splsoftclock();
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
for (nfsd = LIST_FIRST(&slp->ns_tq); nfsd; nfsd = owp) {
owp = LIST_NEXT(nfsd, nd_tq);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (nfsd->nd_time > cur_usec)
break;
if (nfsd->nd_mreq)
continue;
Fix a nasty hang connected with write gathering. Also add debug print statements to bits of the server which helped me find the hang.
1997-05-10 16:12:03 +00:00
NFS_DPF(WG, ("P%03x", nfsd->nd_retxid & 0xfff));
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
LIST_REMOVE(nfsd, nd_tq);
LIST_REMOVE(nfsd, nd_hash);
splx(s);
mrep = nfsd->nd_mrep;
nfsd->nd_mrep = NULL;
cred = &nfsd->nd_cr;
v3 = (nfsd->nd_flag & ND_NFSV3);
forat_ret = aftat_ret = 1;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
error = nfsrv_fhtovp(&nfsd->nd_fh, 1, &vp, cred, slp,
nfsd->nd_nam, &rdonly, TRUE);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (!error) {
if (v3)
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
forat_ret = VOP_GETATTR(vp, &forat, cred, td);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (vp->v_type != VREG) {
if (v3)
error = EINVAL;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
else
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
error = (vp->v_type == VDIR) ? EISDIR : EACCES;
}
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
} else {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
vp = NULL;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
}
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (!error)
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
error = nfsrv_access(vp, VWRITE, cred, rdonly, td, 1);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (nfsd->nd_stable == NFSV3WRITE_UNSTABLE)
ioflags = IO_NODELOCKED;
else if (nfsd->nd_stable == NFSV3WRITE_DATASYNC)
ioflags = (IO_SYNC | IO_NODELOCKED);
else
ioflags = (IO_METASYNC | IO_SYNC | IO_NODELOCKED);
uiop->uio_rw = UIO_WRITE;
uiop->uio_segflg = UIO_SYSSPACE;
Convert old style (type foo *)0 casts to NULLs PR: kern/40360 Requested by: Hiten PAndya via direct email
2002-07-11 17:54:58 +00:00
uiop->uio_td = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
uiop->uio_offset = nfsd->nd_off;
uiop->uio_resid = nfsd->nd_eoff - nfsd->nd_off;
if (uiop->uio_resid > 0) {
mp = mrep;
i = 0;
while (mp) {
if (mp->m_len > 0)
i++;
mp = mp->m_next;
}
uiop->uio_iovcnt = i;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
MALLOC(iov, struct iovec *, i * sizeof (struct iovec),
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
M_TEMP, M_WAITOK);
uiop->uio_iov = ivp = iov;
mp = mrep;
while (mp) {
if (mp->m_len > 0) {
ivp->iov_base = mtod(mp, caddr_t);
ivp->iov_len = mp->m_len;
ivp++;
}
mp = mp->m_next;
}
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
if (!error) {
if (vn_start_write(vp, &mntp, V_NOWAIT) != 0) {
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
VOP_UNLOCK(vp, 0, td);
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
error = vn_start_write(NULL, &mntp, V_WAIT);
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
}
}
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (!error) {
error = VOP_WRITE(vp, uiop, ioflags, cred);
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrvstats.srvvop_writes++;
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
vn_finished_write(mntp);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
FREE((caddr_t)iov, M_TEMP);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
m_freem(mrep);
if (vp) {
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
aftat_ret = VOP_GETATTR(vp, &va, cred, td);
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vput(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
/*
* Loop around generating replies for all write rpcs that have
* now been completed.
*/
swp = nfsd;
do {
Fix a nasty hang connected with write gathering. Also add debug print statements to bits of the server which helped me find the hang.
1997-05-10 16:12:03 +00:00
NFS_DPF(WG, ("R%03x", nfsd->nd_retxid & 0xfff));
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (error) {
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsm_writereply(NFSX_WCCDATA(v3));
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (v3) {
nfsm_srvwcc_data(forat_ret, &forat, aftat_ret, &va);
}
} else {
nfsm_writereply(NFSX_PREOPATTR(v3) +
NFSX_POSTOPORFATTR(v3) + 2 * NFSX_UNSIGNED +
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
NFSX_WRITEVERF(v3));
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (v3) {
nfsm_srvwcc_data(forat_ret, &forat, aftat_ret, &va);
Tidy up nfsm_build usage. This is only partially finished.
2001-09-27 02:33:36 +00:00
tl = nfsm_build(u_int32_t *, 4 * NFSX_UNSIGNED);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
*tl++ = txdr_unsigned(nfsd->nd_len);
*tl++ = txdr_unsigned(swp->nd_stable);
/*
* Actually, there is no need to txdr these fields,
* but it may make the values more human readable,
* for debugging purposes.
*/
Have NFS use a snapshot of boottime instead of boottime itself to generate the NFSv3 Version id. boottime itself may change, sometimes once every tick if you are running xntpd, which really throws off clients. Clients will tend to throw away what they believe to be stale data too often, and can get into long loops rewriting the same data over and over again because they believe the server has rebooted over and over again due to the changing version id. Approved by: jkh
1999-12-16 17:01:32 +00:00
if (nfsver.tv_sec == 0)
nfsver = boottime;
*tl++ = txdr_unsigned(nfsver.tv_sec);
*tl = txdr_unsigned(nfsver.tv_usec);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
} else {
Tidy up nfsm_build usage. This is only partially finished.
2001-09-27 02:33:36 +00:00
fp = nfsm_build(struct nfs_fattr *, NFSX_V2FATTR);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_srvfillattr(&va, fp);
}
}
nfsd->nd_mreq = mreq;
if (nfsd->nd_mrep)
panic("nfsrv_write: nd_mrep not free");
/*
* Done. Put it at the head of the timer queue so that
* the final phase can return the reply.
*/
s = splsoftclock();
if (nfsd != swp) {
nfsd->nd_time = 0;
LIST_INSERT_HEAD(&slp->ns_tq, nfsd, nd_tq);
}
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsd = LIST_FIRST(&swp->nd_coalesce);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (nfsd) {
LIST_REMOVE(nfsd, nd_tq);
}
splx(s);
} while (nfsd);
s = splsoftclock();
swp->nd_time = 0;
LIST_INSERT_HEAD(&slp->ns_tq, swp, nd_tq);
splx(s);
goto loop1;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
splx(s);
/*
* Search for a reply to return.
*/
s = splsoftclock();
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
LIST_FOREACH(nfsd, &slp->ns_tq, nd_tq)
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (nfsd->nd_mreq) {
Fix a nasty hang connected with write gathering. Also add debug print statements to bits of the server which helped me find the hang.
1997-05-10 16:12:03 +00:00
NFS_DPF(WG, ("X%03x", nfsd->nd_retxid & 0xfff));
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
LIST_REMOVE(nfsd, nd_tq);
*mrq = nfsd->nd_mreq;
*ndp = nfsd;
break;
}
splx(s);
return (0);
}
/*
* Coalesce the write request nfsd into owp. To do this we must:
* - remove nfsd from the queues
* - merge nfsd->nd_mrep into owp->nd_mrep
* - update the nd_eoff and nd_stable for owp
* - put nfsd on owp's nd_coalesce list
* NB: Must be called at splsoftclock().
*/
Staticize.
1995-12-17 21:14:36 +00:00
static void
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrvw_coalesce(struct nfsrv_descript *owp, struct nfsrv_descript *nfsd)
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
{
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int overlap;
struct mbuf *mp;
Fix a nasty hang connected with write gathering. Also add debug print statements to bits of the server which helped me find the hang.
1997-05-10 16:12:03 +00:00
struct nfsrv_descript *p;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
Fix a nasty hang connected with write gathering. Also add debug print statements to bits of the server which helped me find the hang.
1997-05-10 16:12:03 +00:00
NFS_DPF(WG, ("C%03x-%03x",
nfsd->nd_retxid & 0xfff, owp->nd_retxid & 0xfff));
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
LIST_REMOVE(nfsd, nd_hash);
LIST_REMOVE(nfsd, nd_tq);
if (owp->nd_eoff < nfsd->nd_eoff) {
overlap = owp->nd_eoff - nfsd->nd_off;
if (overlap < 0)
panic("nfsrv_coalesce: bad off");
if (overlap > 0)
m_adj(nfsd->nd_mrep, overlap);
mp = owp->nd_mrep;
while (mp->m_next)
mp = mp->m_next;
mp->m_next = nfsd->nd_mrep;
owp->nd_eoff = nfsd->nd_eoff;
} else
m_freem(nfsd->nd_mrep);
nfsd->nd_mrep = NULL;
if (nfsd->nd_stable == NFSV3WRITE_FILESYNC)
owp->nd_stable = NFSV3WRITE_FILESYNC;
else if (nfsd->nd_stable == NFSV3WRITE_DATASYNC &&
owp->nd_stable == NFSV3WRITE_UNSTABLE)
owp->nd_stable = NFSV3WRITE_DATASYNC;
LIST_INSERT_HEAD(&owp->nd_coalesce, nfsd, nd_tq);
Fix a nasty hang connected with write gathering. Also add debug print statements to bits of the server which helped me find the hang.
1997-05-10 16:12:03 +00:00
/*
* If nfsd had anything else coalesced into it, transfer them
* to owp, otherwise their replies will never get sent.
*/
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
for (p = LIST_FIRST(&nfsd->nd_coalesce); p;
p = LIST_FIRST(&nfsd->nd_coalesce)) {
Fix a nasty hang connected with write gathering. Also add debug print statements to bits of the server which helped me find the hang.
1997-05-10 16:12:03 +00:00
LIST_REMOVE(p, nd_tq);
LIST_INSERT_HEAD(&owp->nd_coalesce, p, nd_tq);
}
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* nfs create service
* now does a truncate to 0 length via. setattr if it already exists
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrv_create(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
struct thread *td, struct mbuf **mrq)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md;
Fix all areas of the system (or at least all those in LINT) to avoid storing socket addresses in mbufs. (Socket buffers are the one exception.) A number of kernel APIs needed to get fixed in order to make this happen. Also, fix three protocol families which kept PCBs in mbufs to not malloc them instead. Delete some old compatibility cruft while we're at it, and add some new routines in the in_cksum family.
1997-08-16 19:16:27 +00:00
struct sockaddr *nam = nfsd->nd_nam;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
caddr_t dpos = nfsd->nd_dpos;
struct ucred *cred = &nfsd->nd_cr;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct nfs_fattr *fp;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct vattr va, dirfor, diraft;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct vattr *vap = &va;
struct nfsv2_sattr *sp;
u_int32_t *tl;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct nameidata nd;
caddr_t bpos;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int error = 0, rdev, len, tsize, dirfor_ret = 1, diraft_ret = 1;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
int v3 = (nfsd->nd_flag & ND_NFSV3), how, exclusive_flag = 0;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
caddr_t cp;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct mbuf *mb, *mreq;
Convert old style (type foo *)0 casts to NULLs PR: kern/40360 Requested by: Hiten PAndya via direct email
2002-07-11 17:54:58 +00:00
struct vnode *dirp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsfh_t nfh;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhandle_t *fhp;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
u_quad_t tempsize;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
u_char cverf[NFSX_V3CREATEVERF];
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
struct mount *mp = NULL;
struct vnode *vp;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsdbprintf(("%s %d\n", __FILE__, __LINE__));
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
#ifndef nolint
rdev = 0;
#endif
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
ndclear(&nd);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
fhp = &nfh.fh_generic;
nfsm_srvmtofh(fhp);
Fix some problems that were introduced in revision 1.97. Instead of returning an error code to the caller, NFS server op routines must themselves build an error reply and return 0 to the caller. This is achieved by replacing the erroneous return statements with code that jumps forward to the op function's reply code. We need to be careful to ensure that the 'struct mount' pointer is NULL though, so that the final vn_finished_write() call becomes a no-op. Reviewed by: mckusick, dillon
2001-02-09 13:24:06 +00:00
if ((mp = vfs_getvfs(&fhp->fh_fsid)) == NULL) {
error = ESTALE;
goto ereply;
}
if ((error = VFS_FHTOVP(mp, &fhp->fh_fid, &vp)) != 0) {
mp = NULL;
goto ereply;
}
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
(void) vn_start_write(vp, &mp, V_WAIT);
vput(vp);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_srvnamesiz(len);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nd.ni_cnd.cn_cred = cred;
nd.ni_cnd.cn_nameiop = CREATE;
nd.ni_cnd.cn_flags = LOCKPARENT | LOCKLEAF | SAVESTART;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/*
* Call namei and do initial cleanup to get a few things
* out of the way. If we get an initial error we cleanup
* and return here to avoid special-casing the invalid nd
* structure through the rest of the case. dirp may be
* set even if an error occurs, but the nd structure will not
* be valid at all if an error occurs so we have to invalidate it
* prior to calling nfsm_reply ( which might goto nfsmout ).
*/
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
error = nfs_namei(&nd, fhp, len, slp, nam, &md, &dpos,
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
&dirp, td, FALSE);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (dirp) {
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (v3) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
dirfor_ret = VOP_GETATTR(dirp, &dirfor, cred,
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
td);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
} else {
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(dirp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
dirp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
}
if (error) {
nfsm_reply(NFSX_WCCDATA(v3));
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (v3)
nfsm_srvwcc_data(dirfor_ret, &dirfor, diraft_ret, &diraft);
Submitted by: "David E. Cross" <crossd@cs.rpi.edu> Matt missed a line..
1999-06-30 04:29:13 +00:00
error = 0;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
goto nfsmout;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/*
* No error. Continue. State:
*
* startdir is valid ( we release this immediately )
* dirp may be valid
* nd.ni_vp may be valid
* nd.ni_dvp is valid
*
* The error state is set through the code and we may also do some
* opportunistic releasing of vnodes to avoid holding locks through
* NFS I/O. The cleanup at the end is a catch-all
*/
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
VATTR_NULL(vap);
if (v3) {
Make nfsm_dissect() have an obvious return value.
2001-09-27 22:40:38 +00:00
tl = nfsm_dissect(u_int32_t *, NFSX_UNSIGNED);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
how = fxdr_unsigned(int, *tl);
switch (how) {
case NFSV3CREATE_GUARDED:
if (nd.ni_vp) {
error = EEXIST;
break;
}
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/* fall through */
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
case NFSV3CREATE_UNCHECKED:
nfsm_srvsattr(vap);
break;
case NFSV3CREATE_EXCLUSIVE:
Make nfsm_dissect() have an obvious return value.
2001-09-27 22:40:38 +00:00
cp = nfsm_dissect(caddr_t, NFSX_V3CREATEVERF);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
bcopy(cp, cverf, NFSX_V3CREATEVERF);
exclusive_flag = 1;
break;
};
vap->va_type = VREG;
} else {
Make nfsm_dissect() have an obvious return value.
2001-09-27 22:40:38 +00:00
sp = nfsm_dissect(struct nfsv2_sattr *, NFSX_V2SATTR);
For the on-the-wire protocol, u_long -> u_int32_t; long -> int32_t; int -> int32_t; u_short -> u_int16_t. Also, use mode_t instead of u_short for storing modes (mode_t is a u_int16_t). Obtained from: NetBSD
1998-05-31 20:09:01 +00:00
vap->va_type = IFTOVT(fxdr_unsigned(u_int32_t, sp->sa_mode));
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (vap->va_type == VNON)
vap->va_type = VREG;
vap->va_mode = nfstov_mode(sp->sa_mode);
switch (vap->va_type) {
case VREG:
For the on-the-wire protocol, u_long -> u_int32_t; long -> int32_t; int -> int32_t; u_short -> u_int16_t. Also, use mode_t instead of u_short for storing modes (mode_t is a u_int16_t). Obtained from: NetBSD
1998-05-31 20:09:01 +00:00
tsize = fxdr_unsigned(int32_t, sp->sa_size);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (tsize != -1)
vap->va_size = (u_quad_t)tsize;
break;
case VCHR:
case VBLK:
case VFIFO:
rdev = fxdr_unsigned(long, sp->sa_size);
break;
NFS Jumbo commit part 1. Cosmetic and structural changes only. The aim of this part of commits is to minimize unnecessary differences between the other NFS's of similar origin. Yes, there are gratuitous changes here that the style folks won't like, but it makes the catch-up less difficult.
1998-05-31 17:27:58 +00:00
default:
break;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
};
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Iff doesn't exist, create it
* otherwise just truncate to 0 length
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
* should I set the mode too ?
*
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
* The only possible error we can have at this point is EEXIST.
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
* nd.ni_vp will also be non-NULL in that case.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
if (nd.ni_vp == NULL) {
It is required by VOP_CREATE, VOP_MKNOD, VOP_SYMLINK and VOP_MKDIR that va_mode of the supplied attributes is filled in with a valid file mode (i.e not VNOVAL, and only ALLPERM bits set). However, some NFS server op functions didn't guarantee this for all possible request messages: If a V3 client chose not include to a mode specification, we could end up creating an ffs inode with mode 0177777, requiring a manual fsck on the next reboot. Fix this by setting va_mode to 0 before calling the VOP if a mode hasn't been supplied by the client. In nfsrv_symlink(), S_IFMT bits supplied by a V2 client could end up in the va_mode passed to VOP_SYMLINK with similar effects. We now use the macro nfstov_mode() to correctly mask the bits.
2002-01-13 05:36:05 +00:00
if (vap->va_mode == (mode_t)VNOVAL)
vap->va_mode = 0;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (vap->va_type == VREG || vap->va_type == VSOCK) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
error = VOP_CREATE(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, vap);
Introduce NDFREE (and remove VOP_ABORTOP)
1999-12-15 23:02:35 +00:00
if (error)
NDFREE(&nd, NDF_ONLY_PNBUF);
else {
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
nfsrv_object_create(nd.ni_vp);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (exclusive_flag) {
exclusive_flag = 0;
VATTR_NULL(vap);
bcopy(cverf, (caddr_t)&vap->va_atime,
NFSX_V3CREATEVERF);
error = VOP_SETATTR(nd.ni_vp, vap, cred,
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
td);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
}
Fix a few NFSv2 issues that slipped in during the big cleanup. The semantics of the nfsm_reply() macro were changed so that the caller has to explicitly handle the V2 error case, whereas before, nfsm_reply() did a `goto nfsmout' then. A few server ops (setattr, readlink, create, mkdir) weren't updated to match, so errors in the V2 case could cause protocol hangs and leaked mbufs. Correct some comments that describe the old nfsm_reply behaviour. [older, harmless nit] Remove the unnecessary `nfsmreply0' label in nfsrv_create(), since for its users, the main `ereply' label does the same thing.
2002-01-12 03:57:25 +00:00
} else if (vap->va_type == VCHR || vap->va_type == VBLK ||
vap->va_type == VFIFO) {
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/*
Fix a few NFSv2 issues that slipped in during the big cleanup. The semantics of the nfsm_reply() macro were changed so that the caller has to explicitly handle the V2 error case, whereas before, nfsm_reply() did a `goto nfsmout' then. A few server ops (setattr, readlink, create, mkdir) weren't updated to match, so errors in the V2 case could cause protocol hangs and leaked mbufs. Correct some comments that describe the old nfsm_reply behaviour. [older, harmless nit] Remove the unnecessary `nfsmreply0' label in nfsrv_create(), since for its users, the main `ereply' label does the same thing.
2002-01-12 03:57:25 +00:00
* NFSv2-specific code for creating device nodes
* and fifos.
*
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
* Handle SysV FIFO node special cases. All other
* devices require super user to access.
*/
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (vap->va_type == VCHR && rdev == 0xffffffff)
vap->va_type = VFIFO;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (vap->va_type != VFIFO &&
Change the suser() API to take advantage of td_ucred as well as do a general cleanup of the API. The entire API now consists of two functions similar to the pre-KSE API. The suser() function takes a thread pointer as its only argument. The td_ucred member of this thread must be valid so the only valid thread pointers are curthread and a few kernel threads such as thread0. The suser_cred() function takes a pointer to a struct ucred as its first argument and an integer flag as its second argument. The flag is currently only used for the PRISON_ROOT flag. Discussed on: smp@
2002-04-01 21:31:13 +00:00
(error = suser_cred(cred, 0))) {
Fix a few NFSv2 issues that slipped in during the big cleanup. The semantics of the nfsm_reply() macro were changed so that the caller has to explicitly handle the V2 error case, whereas before, nfsm_reply() did a `goto nfsmout' then. A few server ops (setattr, readlink, create, mkdir) weren't updated to match, so errors in the V2 case could cause protocol hangs and leaked mbufs. Correct some comments that describe the old nfsm_reply behaviour. [older, harmless nit] Remove the unnecessary `nfsmreply0' label in nfsrv_create(), since for its users, the main `ereply' label does the same thing.
2002-01-12 03:57:25 +00:00
goto ereply;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
}
vap->va_rdev = rdev;
In the words of the submitter: --------- Make callers of namei() responsible for releasing references or locks instead of having the underlying filesystems do it. This eliminates redundancy in all terminal filesystems and makes it possible for stacked transport layers such as umapfs or nullfs to operate correctly. Quality testing was done with testvn, and lat_fs from the lmbench suite. Some NFS client testing courtesy of Patrik Kudo. vop_mknod and vop_symlink still release the returned vpp. vop_rename still releases 4 vnode arguments before it returns. These remaining cases will be corrected in the next set of patches. --------- Submitted by: Michael Hancock <michaelh@cet.co.jp>
1998-05-07 04:58:58 +00:00
error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, vap);
if (error) {
Introduce NDFREE (and remove VOP_ABORTOP)
1999-12-15 23:02:35 +00:00
NDFREE(&nd, NDF_ONLY_PNBUF);
Fix a few NFSv2 issues that slipped in during the big cleanup. The semantics of the nfsm_reply() macro were changed so that the caller has to explicitly handle the V2 error case, whereas before, nfsm_reply() did a `goto nfsmout' then. A few server ops (setattr, readlink, create, mkdir) weren't updated to match, so errors in the V2 case could cause protocol hangs and leaked mbufs. Correct some comments that describe the old nfsm_reply behaviour. [older, harmless nit] Remove the unnecessary `nfsmreply0' label in nfsrv_create(), since for its users, the main `ereply' label does the same thing.
2002-01-12 03:57:25 +00:00
goto ereply;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Remove WILLRELE from VOP_RENAME
1999-11-12 03:34:28 +00:00
vput(nd.ni_vp);
nd.ni_vp = NULL;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/*
* release dvp prior to lookup
*/
vput(nd.ni_dvp);
nd.ni_dvp = NULL;
/*
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
* Setup for lookup.
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
*
* Even though LOCKPARENT was cleared, ni_dvp may
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
* be garbage.
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
*/
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nd.ni_cnd.cn_nameiop = LOOKUP;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nd.ni_cnd.cn_flags &= ~(LOCKPARENT);
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
nd.ni_cnd.cn_thread = td;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nd.ni_cnd.cn_cred = cred;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = lookup(&nd);
nd.ni_dvp = NULL;
Fix a few NFSv2 issues that slipped in during the big cleanup. The semantics of the nfsm_reply() macro were changed so that the caller has to explicitly handle the V2 error case, whereas before, nfsm_reply() did a `goto nfsmout' then. A few server ops (setattr, readlink, create, mkdir) weren't updated to match, so errors in the V2 case could cause protocol hangs and leaked mbufs. Correct some comments that describe the old nfsm_reply behaviour. [older, harmless nit] Remove the unnecessary `nfsmreply0' label in nfsrv_create(), since for its users, the main `ereply' label does the same thing.
2002-01-12 03:57:25 +00:00
if (error)
goto ereply;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
nfsrv_object_create(nd.ni_vp);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (nd.ni_cnd.cn_flags & ISSYMLINK) {
error = EINVAL;
Fix a few NFSv2 issues that slipped in during the big cleanup. The semantics of the nfsm_reply() macro were changed so that the caller has to explicitly handle the V2 error case, whereas before, nfsm_reply() did a `goto nfsmout' then. A few server ops (setattr, readlink, create, mkdir) weren't updated to match, so errors in the V2 case could cause protocol hangs and leaked mbufs. Correct some comments that describe the old nfsm_reply behaviour. [older, harmless nit] Remove the unnecessary `nfsmreply0' label in nfsrv_create(), since for its users, the main `ereply' label does the same thing.
2002-01-12 03:57:25 +00:00
goto ereply;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
} else {
error = ENXIO;
}
} else {
if (vap->va_size != -1) {
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = nfsrv_access(nd.ni_vp, VWRITE, cred,
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
(nd.ni_cnd.cn_flags & RDONLY), td, 0);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (!error) {
tempsize = vap->va_size;
VATTR_NULL(vap);
vap->va_size = tempsize;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = VOP_SETATTR(nd.ni_vp, vap, cred,
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
td);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
}
}
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (!error) {
bzero((caddr_t)fhp, sizeof(nfh));
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
fhp->fh_fsid = nd.ni_vp->v_mount->mnt_stat.f_fsid;
error = VFS_VPTOFH(nd.ni_vp, &fhp->fh_fid);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (!error)
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
error = VOP_GETATTR(nd.ni_vp, vap, cred, td);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (v3) {
if (exclusive_flag && !error &&
bcmp(cverf, (caddr_t)&vap->va_atime, NFSX_V3CREATEVERF))
error = EEXIST;
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
diraft_ret = VOP_GETATTR(dirp, &diraft, cred, td);
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(dirp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
dirp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
Fix some problems that were introduced in revision 1.97. Instead of returning an error code to the caller, NFS server op routines must themselves build an error reply and return 0 to the caller. This is achieved by replacing the erroneous return statements with code that jumps forward to the op function's reply code. We need to be careful to ensure that the 'struct mount' pointer is NULL though, so that the final vn_finished_write() call becomes a no-op. Reviewed by: mckusick, dillon
2001-02-09 13:24:06 +00:00
ereply:
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_SRVFH(v3) + NFSX_FATTR(v3) + NFSX_WCCDATA(v3));
if (v3) {
if (!error) {
nfsm_srvpostop_fh(fhp);
nfsm_srvpostop_attr(0, vap);
}
nfsm_srvwcc_data(dirfor_ret, &dirfor, diraft_ret, &diraft);
Fix a few NFSv2 issues that slipped in during the big cleanup. The semantics of the nfsm_reply() macro were changed so that the caller has to explicitly handle the V2 error case, whereas before, nfsm_reply() did a `goto nfsmout' then. A few server ops (setattr, readlink, create, mkdir) weren't updated to match, so errors in the V2 case could cause protocol hangs and leaked mbufs. Correct some comments that describe the old nfsm_reply behaviour. [older, harmless nit] Remove the unnecessary `nfsmreply0' label in nfsrv_create(), since for its users, the main `ereply' label does the same thing.
2002-01-12 03:57:25 +00:00
} else if (!error) {
/* v2 non-error case. */
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_srvfhtom(fhp, v3);
Tidy up nfsm_build usage. This is only partially finished.
2001-09-27 02:33:36 +00:00
fp = nfsm_build(struct nfs_fattr *, NFSX_V2FATTR);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_srvfillattr(vap, fp);
}
Fix a few NFSv2 issues that slipped in during the big cleanup. The semantics of the nfsm_reply() macro were changed so that the caller has to explicitly handle the V2 error case, whereas before, nfsm_reply() did a `goto nfsmout' then. A few server ops (setattr, readlink, create, mkdir) weren't updated to match, so errors in the V2 case could cause protocol hangs and leaked mbufs. Correct some comments that describe the old nfsm_reply behaviour. [older, harmless nit] Remove the unnecessary `nfsmreply0' label in nfsrv_create(), since for its users, the main `ereply' label does the same thing.
2002-01-12 03:57:25 +00:00
error = 0;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nfsmout:
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (nd.ni_startdir) {
vrele(nd.ni_startdir);
nd.ni_startdir = NULL;
}
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (dirp)
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(dirp);
Introduce NDFREE (and remove VOP_ABORTOP)
1999-12-15 23:02:35 +00:00
NDFREE(&nd, NDF_ONLY_PNBUF);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (nd.ni_dvp) {
if (nd.ni_dvp == nd.ni_vp)
vrele(nd.ni_dvp);
else
vput(nd.ni_dvp);
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (nd.ni_vp)
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vput(nd.ni_vp);
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
vn_finished_write(mp);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return (error);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
/*
* nfs v3 mknod service
*/
int
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrv_mknod(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
struct thread *td, struct mbuf **mrq)
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
{
struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md;
Fix all areas of the system (or at least all those in LINT) to avoid storing socket addresses in mbufs. (Socket buffers are the one exception.) A number of kernel APIs needed to get fixed in order to make this happen. Also, fix three protocol families which kept PCBs in mbufs to not malloc them instead. Delete some old compatibility cruft while we're at it, and add some new routines in the in_cksum family.
1997-08-16 19:16:27 +00:00
struct sockaddr *nam = nfsd->nd_nam;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
caddr_t dpos = nfsd->nd_dpos;
struct ucred *cred = &nfsd->nd_cr;
struct vattr va, dirfor, diraft;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct vattr *vap = &va;
u_int32_t *tl;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct nameidata nd;
caddr_t bpos;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int error = 0, len, dirfor_ret = 1, diraft_ret = 1;
For the on-the-wire protocol, u_long -> u_int32_t; long -> int32_t; int -> int32_t; u_short -> u_int16_t. Also, use mode_t instead of u_short for storing modes (mode_t is a u_int16_t). Obtained from: NetBSD
1998-05-31 20:09:01 +00:00
u_int32_t major, minor;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
enum vtype vtyp;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct mbuf *mb, *mreq;
Convert old style (type foo *)0 casts to NULLs PR: kern/40360 Requested by: Hiten PAndya via direct email
2002-07-11 17:54:58 +00:00
struct vnode *vp, *dirp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsfh_t nfh;
fhandle_t *fhp;
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
struct mount *mp = NULL;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int v3 = (nfsd->nd_flag & ND_NFSV3);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsdbprintf(("%s %d\n", __FILE__, __LINE__));
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (!v3)
panic("nfsrv_mknod: v3 proc called on a v2 connection");
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
ndclear(&nd);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
fhp = &nfh.fh_generic;
nfsm_srvmtofh(fhp);
Fix some problems that were introduced in revision 1.97. Instead of returning an error code to the caller, NFS server op routines must themselves build an error reply and return 0 to the caller. This is achieved by replacing the erroneous return statements with code that jumps forward to the op function's reply code. We need to be careful to ensure that the 'struct mount' pointer is NULL though, so that the final vn_finished_write() call becomes a no-op. Reviewed by: mckusick, dillon
2001-02-09 13:24:06 +00:00
if ((mp = vfs_getvfs(&fhp->fh_fsid)) == NULL) {
error = ESTALE;
goto ereply;
}
if ((error = VFS_FHTOVP(mp, &fhp->fh_fid, &vp)) != 0) {
mp = NULL;
goto ereply;
}
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
(void) vn_start_write(vp, &mp, V_WAIT);
vput(vp);
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_srvnamesiz(len);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nd.ni_cnd.cn_cred = cred;
nd.ni_cnd.cn_nameiop = CREATE;
nd.ni_cnd.cn_flags = LOCKPARENT | LOCKLEAF | SAVESTART;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/*
* Handle nfs_namei() call. If an error occurs, the nd structure
* is not valid. However, nfsm_*() routines may still jump to
* nfsmout.
*/
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
error = nfs_namei(&nd, fhp, len, slp, nam, &md, &dpos,
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
&dirp, td, FALSE);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (dirp)
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
dirfor_ret = VOP_GETATTR(dirp, &dirfor, cred, td);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (error) {
nfsm_reply(NFSX_WCCDATA(1));
nfsm_srvwcc_data(dirfor_ret, &dirfor, diraft_ret, &diraft);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
Make nfsm_dissect() have an obvious return value.
2001-09-27 22:40:38 +00:00
tl = nfsm_dissect(u_int32_t *, NFSX_UNSIGNED);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
vtyp = nfsv3tov_type(*tl);
if (vtyp != VCHR && vtyp != VBLK && vtyp != VSOCK && vtyp != VFIFO) {
error = NFSERR_BADTYPE;
goto out;
}
VATTR_NULL(vap);
nfsm_srvsattr(vap);
if (vtyp == VCHR || vtyp == VBLK) {
Make nfsm_dissect() have an obvious return value.
2001-09-27 22:40:38 +00:00
tl = nfsm_dissect(u_int32_t *, 2 * NFSX_UNSIGNED);
For the on-the-wire protocol, u_long -> u_int32_t; long -> int32_t; int -> int32_t; u_short -> u_int16_t. Also, use mode_t instead of u_short for storing modes (mode_t is a u_int16_t). Obtained from: NetBSD
1998-05-31 20:09:01 +00:00
major = fxdr_unsigned(u_int32_t, *tl++);
minor = fxdr_unsigned(u_int32_t, *tl);
I have not one single time remembered the name of this function correctly so obviously I gave it the wrong name. s/umakedev/makeudev/g
1999-07-17 18:43:50 +00:00
vap->va_rdev = makeudev(major, minor);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
/*
* Iff doesn't exist, create it.
*/
if (nd.ni_vp) {
error = EEXIST;
goto out;
}
vap->va_type = vtyp;
It is required by VOP_CREATE, VOP_MKNOD, VOP_SYMLINK and VOP_MKDIR that va_mode of the supplied attributes is filled in with a valid file mode (i.e not VNOVAL, and only ALLPERM bits set). However, some NFS server op functions didn't guarantee this for all possible request messages: If a V3 client chose not include to a mode specification, we could end up creating an ffs inode with mode 0177777, requiring a manual fsck on the next reboot. Fix this by setting va_mode to 0 before calling the VOP if a mode hasn't been supplied by the client. In nfsrv_symlink(), S_IFMT bits supplied by a V2 client could end up in the va_mode passed to VOP_SYMLINK with similar effects. We now use the macro nfstov_mode() to correctly mask the bits.
2002-01-13 05:36:05 +00:00
if (vap->va_mode == (mode_t)VNOVAL)
vap->va_mode = 0;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (vtyp == VSOCK) {
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(nd.ni_startdir);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nd.ni_startdir = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
error = VOP_CREATE(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, vap);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (error)
Introduce NDFREE (and remove VOP_ABORTOP)
1999-12-15 23:02:35 +00:00
NDFREE(&nd, NDF_ONLY_PNBUF);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
} else {
Change the suser() API to take advantage of td_ucred as well as do a general cleanup of the API. The entire API now consists of two functions similar to the pre-KSE API. The suser() function takes a thread pointer as its only argument. The td_ucred member of this thread must be valid so the only valid thread pointers are curthread and a few kernel threads such as thread0. The suser_cred() function takes a pointer to a struct ucred as its first argument and an integer flag as its second argument. The flag is currently only used for the PRISON_ROOT flag. Discussed on: smp@
2002-04-01 21:31:13 +00:00
if (vtyp != VFIFO && (error = suser_cred(cred, 0)))
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
goto out;
In the words of the submitter: --------- Make callers of namei() responsible for releasing references or locks instead of having the underlying filesystems do it. This eliminates redundancy in all terminal filesystems and makes it possible for stacked transport layers such as umapfs or nullfs to operate correctly. Quality testing was done with testvn, and lat_fs from the lmbench suite. Some NFS client testing courtesy of Patrik Kudo. vop_mknod and vop_symlink still release the returned vpp. vop_rename still releases 4 vnode arguments before it returns. These remaining cases will be corrected in the next set of patches. --------- Submitted by: Michael Hancock <michaelh@cet.co.jp>
1998-05-07 04:58:58 +00:00
error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, vap);
if (error) {
Introduce NDFREE (and remove VOP_ABORTOP)
1999-12-15 23:02:35 +00:00
NDFREE(&nd, NDF_ONLY_PNBUF);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
goto out;
}
Remove WILLRELE from VOP_RENAME
1999-11-12 03:34:28 +00:00
vput(nd.ni_vp);
nd.ni_vp = NULL;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/*
* Release dvp prior to lookup
*/
vput(nd.ni_dvp);
nd.ni_dvp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nd.ni_cnd.cn_nameiop = LOOKUP;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nd.ni_cnd.cn_flags &= ~(LOCKPARENT);
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
nd.ni_cnd.cn_thread = td;
Simple p_ucred -> td_ucred changes to start using the per-thread ucred reference.
2002-02-27 18:32:23 +00:00
nd.ni_cnd.cn_cred = td->td_ucred;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
error = lookup(&nd);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nd.ni_dvp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (error)
goto out;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (nd.ni_cnd.cn_flags & ISSYMLINK)
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
error = EINVAL;
}
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/*
* send response, cleanup, return.
*/
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
out:
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (nd.ni_startdir) {
vrele(nd.ni_startdir);
nd.ni_startdir = NULL;
}
Introduce NDFREE (and remove VOP_ABORTOP)
1999-12-15 23:02:35 +00:00
NDFREE(&nd, NDF_ONLY_PNBUF);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (nd.ni_dvp) {
if (nd.ni_dvp == nd.ni_vp)
vrele(nd.ni_dvp);
else
vput(nd.ni_dvp);
nd.ni_dvp = NULL;
}
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
vp = nd.ni_vp;
if (!error) {
bzero((caddr_t)fhp, sizeof(nfh));
fhp->fh_fsid = vp->v_mount->mnt_stat.f_fsid;
error = VFS_VPTOFH(vp, &fhp->fh_fid);
if (!error)
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
error = VOP_GETATTR(vp, vap, cred, td);
The vnode was not being vput()'d in the EEXIST mknod case on the nfs server side. This can lead to a system deadlock. Reviewed by: iedowse Tested by: Alexey G Misurenko <mag@caravan.ru>, iedowse Bug found with help by: Alexey G Misurenko <mag@caravan.ru> MFC at: earliest convenience
2002-01-14 19:14:08 +00:00
}
if (vp) {
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vput(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
nd.ni_vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
diraft_ret = VOP_GETATTR(dirp, &diraft, cred, td);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (dirp) {
vrele(dirp);
dirp = NULL;
}
Fix some problems that were introduced in revision 1.97. Instead of returning an error code to the caller, NFS server op routines must themselves build an error reply and return 0 to the caller. This is achieved by replacing the erroneous return statements with code that jumps forward to the op function's reply code. We need to be careful to ensure that the 'struct mount' pointer is NULL though, so that the final vn_finished_write() call becomes a no-op. Reviewed by: mckusick, dillon
2001-02-09 13:24:06 +00:00
ereply:
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_SRVFH(1) + NFSX_POSTOPATTR(1) + NFSX_WCCDATA(1));
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (v3) {
if (!error) {
nfsm_srvpostop_fh(fhp);
nfsm_srvpostop_attr(0, vap);
}
nfsm_srvwcc_data(dirfor_ret, &dirfor, diraft_ret, &diraft);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
vn_finished_write(mp);
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
return (0);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsmout:
if (dirp)
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(dirp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (nd.ni_startdir)
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(nd.ni_startdir);
Introduce NDFREE (and remove VOP_ABORTOP)
1999-12-15 23:02:35 +00:00
NDFREE(&nd, NDF_ONLY_PNBUF);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (nd.ni_dvp) {
if (nd.ni_dvp == nd.ni_vp)
vrele(nd.ni_dvp);
else
vput(nd.ni_dvp);
}
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (nd.ni_vp)
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vput(nd.ni_vp);
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
vn_finished_write(mp);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
return (error);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
/*
* nfs remove service
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrv_remove(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
struct thread *td, struct mbuf **mrq)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md;
Fix all areas of the system (or at least all those in LINT) to avoid storing socket addresses in mbufs. (Socket buffers are the one exception.) A number of kernel APIs needed to get fixed in order to make this happen. Also, fix three protocol families which kept PCBs in mbufs to not malloc them instead. Delete some old compatibility cruft while we're at it, and add some new routines in the in_cksum family.
1997-08-16 19:16:27 +00:00
struct sockaddr *nam = nfsd->nd_nam;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
caddr_t dpos = nfsd->nd_dpos;
struct ucred *cred = &nfsd->nd_cr;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct nameidata nd;
caddr_t bpos;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int error = 0, len, dirfor_ret = 1, diraft_ret = 1;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
int v3 = (nfsd->nd_flag & ND_NFSV3);
Second batch of cleanup changes. This time mostly making a lot of things static and some unused variables here and there.
1995-10-29 15:33:36 +00:00
struct mbuf *mb, *mreq;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
struct vnode *dirp;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct vattr dirfor, diraft;
nfsfh_t nfh;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhandle_t *fhp;
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
struct mount *mp = NULL;
struct vnode *vp;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsdbprintf(("%s %d\n", __FILE__, __LINE__));
ndclear(&nd);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhp = &nfh.fh_generic;
nfsm_srvmtofh(fhp);
Fix some problems that were introduced in revision 1.97. Instead of returning an error code to the caller, NFS server op routines must themselves build an error reply and return 0 to the caller. This is achieved by replacing the erroneous return statements with code that jumps forward to the op function's reply code. We need to be careful to ensure that the 'struct mount' pointer is NULL though, so that the final vn_finished_write() call becomes a no-op. Reviewed by: mckusick, dillon
2001-02-09 13:24:06 +00:00
if ((mp = vfs_getvfs(&fhp->fh_fsid)) == NULL) {
error = ESTALE;
goto ereply;
}
if ((error = VFS_FHTOVP(mp, &fhp->fh_fid, &vp)) != 0) {
mp = NULL;
goto ereply;
}
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
(void) vn_start_write(vp, &mp, V_WAIT);
vput(vp);
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_srvnamesiz(len);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nd.ni_cnd.cn_cred = cred;
nd.ni_cnd.cn_nameiop = DELETE;
nd.ni_cnd.cn_flags = LOCKPARENT | LOCKLEAF;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
error = nfs_namei(&nd, fhp, len, slp, nam, &md, &dpos,
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
&dirp, td, FALSE);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (dirp) {
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (v3) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
dirfor_ret = VOP_GETATTR(dirp, &dirfor, cred,
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
td);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
} else {
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(dirp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
dirp = NULL;
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (error == 0) {
if (nd.ni_vp->v_type == VDIR) {
The local fs interface does not allow link()/unlink() of directories, do not allow a remote nfs client to cause local fs corruption either.
1997-03-25 05:08:28 +00:00
error = EPERM; /* POSIX */
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
goto out;
The local fs interface does not allow link()/unlink() of directories, do not allow a remote nfs client to cause local fs corruption either.
1997-03-25 05:08:28 +00:00
}
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
/*
* The root of a mounted filesystem cannot be deleted.
*/
- Replace v_flag with v_iflag and v_vflag - v_vflag is protected by the vnode lock and is used when synchronization with VOP calls is needed. - v_iflag is protected by interlock and is used for dealing with vnode management issues. These flags include X/O LOCK, FREE, DOOMED, etc. - All accesses to v_iflag and v_vflag have either been locked or marked with mp_fixme's. - Many ASSERT_VOP_LOCKED calls have been added where the locking was not clear. - Many functions in vfs_subr.c were restructured to provide for stronger locking. Idea stolen from: BSD/OS
2002-08-04 10:29:36 +00:00
if (nd.ni_vp->v_vflag & VV_ROOT) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
error = EBUSY;
goto out;
}
out:
if (!error) {
error = VOP_REMOVE(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
Introduce NDFREE (and remove VOP_ABORTOP)
1999-12-15 23:02:35 +00:00
NDFREE(&nd, NDF_ONLY_PNBUF);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
}
if (dirp && v3) {
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
diraft_ret = VOP_GETATTR(dirp, &diraft, cred, td);
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(dirp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
dirp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
Fix some problems that were introduced in revision 1.97. Instead of returning an error code to the caller, NFS server op routines must themselves build an error reply and return 0 to the caller. This is achieved by replacing the erroneous return statements with code that jumps forward to the op function's reply code. We need to be careful to ensure that the 'struct mount' pointer is NULL though, so that the final vn_finished_write() call becomes a no-op. Reviewed by: mckusick, dillon
2001-02-09 13:24:06 +00:00
ereply:
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_WCCDATA(v3));
if (v3) {
nfsm_srvwcc_data(dirfor_ret, &dirfor, diraft_ret, &diraft);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsmout:
Introduce NDFREE (and remove VOP_ABORTOP)
1999-12-15 23:02:35 +00:00
NDFREE(&nd, NDF_ONLY_PNBUF);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (nd.ni_dvp) {
if (nd.ni_dvp == nd.ni_vp)
vrele(nd.ni_dvp);
else
vput(nd.ni_dvp);
}
if (nd.ni_vp)
vput(nd.ni_vp);
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
vn_finished_write(mp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
return(error);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
/*
* nfs rename service
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrv_rename(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
struct thread *td, struct mbuf **mrq)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md;
Fix all areas of the system (or at least all those in LINT) to avoid storing socket addresses in mbufs. (Socket buffers are the one exception.) A number of kernel APIs needed to get fixed in order to make this happen. Also, fix three protocol families which kept PCBs in mbufs to not malloc them instead. Delete some old compatibility cruft while we're at it, and add some new routines in the in_cksum family.
1997-08-16 19:16:27 +00:00
struct sockaddr *nam = nfsd->nd_nam;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
caddr_t dpos = nfsd->nd_dpos;
struct ucred *cred = &nfsd->nd_cr;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
caddr_t bpos;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int error = 0, len, len2, fdirfor_ret = 1, fdiraft_ret = 1;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
int tdirfor_ret = 1, tdiraft_ret = 1;
int v3 = (nfsd->nd_flag & ND_NFSV3);
Second batch of cleanup changes. This time mostly making a lot of things static and some unused variables here and there.
1995-10-29 15:33:36 +00:00
struct mbuf *mb, *mreq;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct nameidata fromnd, tond;
Convert old style (type foo *)0 casts to NULLs PR: kern/40360 Requested by: Hiten PAndya via direct email
2002-07-11 17:54:58 +00:00
struct vnode *fvp, *tvp, *tdvp, *fdirp = NULL;
struct vnode *tdirp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct vattr fdirfor, fdiraft, tdirfor, tdiraft;
nfsfh_t fnfh, tnfh;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhandle_t *ffhp, *tfhp;
uid_t saved_uid;
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
struct mount *mp = NULL;
struct vnode *vp;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsdbprintf(("%s %d\n", __FILE__, __LINE__));
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
#ifndef nolint
Convert old style (type foo *)0 casts to NULLs PR: kern/40360 Requested by: Hiten PAndya via direct email
2002-07-11 17:54:58 +00:00
fvp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
#endif
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
ffhp = &fnfh.fh_generic;
tfhp = &tnfh.fh_generic;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/*
* Clear fields incase goto nfsmout occurs from macro.
*/
ndclear(&fromnd);
ndclear(&tond);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nfsm_srvmtofh(ffhp);
Fix some problems that were introduced in revision 1.97. Instead of returning an error code to the caller, NFS server op routines must themselves build an error reply and return 0 to the caller. This is achieved by replacing the erroneous return statements with code that jumps forward to the op function's reply code. We need to be careful to ensure that the 'struct mount' pointer is NULL though, so that the final vn_finished_write() call becomes a no-op. Reviewed by: mckusick, dillon
2001-02-09 13:24:06 +00:00
if ((mp = vfs_getvfs(&ffhp->fh_fsid)) == NULL) {
error = ESTALE;
goto out1;
}
if ((error = VFS_FHTOVP(mp, &ffhp->fh_fid, &vp)) != 0) {
mp = NULL;
goto out1;
}
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
(void) vn_start_write(vp, &mp, V_WAIT);
vput(vp);
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_srvnamesiz(len);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Remember our original uid so that we can reset cr_uid before
* the second nfs_namei() call, in case it is remapped.
*/
saved_uid = cred->cr_uid;
fromnd.ni_cnd.cn_cred = cred;
fromnd.ni_cnd.cn_nameiop = DELETE;
fromnd.ni_cnd.cn_flags = WANTPARENT | SAVESTART;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
error = nfs_namei(&fromnd, ffhp, len, slp, nam, &md,
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
&dpos, &fdirp, td, FALSE);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (fdirp) {
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (v3) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
fdirfor_ret = VOP_GETATTR(fdirp, &fdirfor, cred,
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
td);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
} else {
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(fdirp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
fdirp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
}
if (error) {
nfsm_reply(2 * NFSX_WCCDATA(v3));
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (v3) {
nfsm_srvwcc_data(fdirfor_ret, &fdirfor, fdiraft_ret, &fdiraft);
nfsm_srvwcc_data(tdirfor_ret, &tdirfor, tdiraft_ret, &tdiraft);
}
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fvp = fromnd.ni_vp;
nfsm_srvmtofh(tfhp);
- Introduce a new macro, since that's what nfs loves, called nfsm_srvpathsiz. This macro plucks a length out of an rpc request and verifies that its size does not exceed NFS_MAXPATHLEN. If it does it generates an ENAMETOOLONG response. - Use this macro, and the existing nfsm_srvnamsiz macro in two places where we deal with paths passed in by the client. This fixes a linux interoperability bug. Linux was sending oversized path components which would cause us to ignore the request all together. This causes linux to hang indefinitly while it waits for a response. This could still happen in other cases where we error out with EBADRPC. Sponsored by: Isilon Systems, Inc. Reviewed by: alfred, fabbri@isilon.com, neal@isilon.com
2002-10-31 22:35:03 +00:00
nfsm_srvnamesiz(len2);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
cred->cr_uid = saved_uid;
tond.ni_cnd.cn_cred = cred;
tond.ni_cnd.cn_nameiop = RENAME;
tond.ni_cnd.cn_flags = LOCKPARENT | LOCKLEAF | NOCACHE | SAVESTART;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
error = nfs_namei(&tond, tfhp, len2, slp, nam, &md,
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
&dpos, &tdirp, td, FALSE);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (tdirp) {
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (v3) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
tdirfor_ret = VOP_GETATTR(tdirp, &tdirfor, cred,
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
td);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
} else {
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(tdirp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
tdirp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
}
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (error)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto out1;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
tdvp = tond.ni_dvp;
tvp = tond.ni_vp;
if (tvp != NULL) {
if (fvp->v_type == VDIR && tvp->v_type != VDIR) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (v3)
error = EEXIST;
else
error = EISDIR;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto out;
} else if (fvp->v_type != VDIR && tvp->v_type == VDIR) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (v3)
error = EEXIST;
else
error = ENOTDIR;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto out;
}
if (tvp->v_type == VDIR && tvp->v_mountedhere) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (v3)
error = EXDEV;
else
error = ENOTEMPTY;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto out;
}
}
if (fvp->v_type == VDIR && fvp->v_mountedhere) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (v3)
error = EXDEV;
else
error = ENOTEMPTY;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto out;
}
if (fvp->v_mount != tdvp->v_mount) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (v3)
error = EXDEV;
else
error = ENOTEMPTY;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto out;
}
Add sufficient braces to keep egcs happy about potentially ambiguous if/else nesting.
1999-05-06 18:13:11 +00:00
if (fvp == tdvp) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (v3)
error = EINVAL;
else
error = ENOTEMPTY;
Add sufficient braces to keep egcs happy about potentially ambiguous if/else nesting.
1999-05-06 18:13:11 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* If source is the same as the destination (that is the
* same vnode with the same name in the same directory),
* then there is nothing to do.
*/
if (fvp == tvp && fromnd.ni_dvp == tdvp &&
fromnd.ni_cnd.cn_namelen == tond.ni_cnd.cn_namelen &&
!bcmp(fromnd.ni_cnd.cn_nameptr, tond.ni_cnd.cn_nameptr,
fromnd.ni_cnd.cn_namelen))
error = -1;
out:
if (!error) {
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/*
* The VOP_RENAME function releases all vnode references &
* locks prior to returning so we need to clear the pointers
* to bypass cleanup code later on.
*/
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
error = VOP_RENAME(fromnd.ni_dvp, fromnd.ni_vp, &fromnd.ni_cnd,
tond.ni_dvp, tond.ni_vp, &tond.ni_cnd);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
fromnd.ni_dvp = NULL;
fromnd.ni_vp = NULL;
tond.ni_dvp = NULL;
tond.ni_vp = NULL;
if (error) {
fromnd.ni_cnd.cn_flags &= ~HASBUF;
tond.ni_cnd.cn_flags &= ~HASBUF;
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
} else {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (error == -1)
error = 0;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/* fall through */
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
out1:
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (fdirp)
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
fdiraft_ret = VOP_GETATTR(fdirp, &fdiraft, cred, td);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (tdirp)
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
tdiraft_ret = VOP_GETATTR(tdirp, &tdiraft, cred, td);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(2 * NFSX_WCCDATA(v3));
if (v3) {
nfsm_srvwcc_data(fdirfor_ret, &fdirfor, fdiraft_ret, &fdiraft);
nfsm_srvwcc_data(tdirfor_ret, &tdirfor, tdiraft_ret, &tdiraft);
}
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
/* fall through */
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nfsmout:
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/*
* Clear out tond related fields
*/
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (tdirp)
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(tdirp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (tond.ni_startdir)
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(tond.ni_startdir);
Introduce NDFREE (and remove VOP_ABORTOP)
1999-12-15 23:02:35 +00:00
NDFREE(&tond, NDF_ONLY_PNBUF);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (tond.ni_dvp) {
if (tond.ni_dvp == tond.ni_vp)
vrele(tond.ni_dvp);
else
vput(tond.ni_dvp);
}
if (tond.ni_vp)
vput(tond.ni_vp);
/*
* Clear out fromnd related fields
*/
if (fdirp)
vrele(fdirp);
if (fromnd.ni_startdir)
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(fromnd.ni_startdir);
Introduce NDFREE (and remove VOP_ABORTOP)
1999-12-15 23:02:35 +00:00
NDFREE(&fromnd, NDF_ONLY_PNBUF);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (fromnd.ni_dvp)
vrele(fromnd.ni_dvp);
if (fromnd.ni_vp)
vrele(fromnd.ni_vp);
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
vn_finished_write(mp);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return (error);
}
/*
* nfs link service
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrv_link(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
struct thread *td, struct mbuf **mrq)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md;
Fix all areas of the system (or at least all those in LINT) to avoid storing socket addresses in mbufs. (Socket buffers are the one exception.) A number of kernel APIs needed to get fixed in order to make this happen. Also, fix three protocol families which kept PCBs in mbufs to not malloc them instead. Delete some old compatibility cruft while we're at it, and add some new routines in the in_cksum family.
1997-08-16 19:16:27 +00:00
struct sockaddr *nam = nfsd->nd_nam;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
caddr_t dpos = nfsd->nd_dpos;
struct ucred *cred = &nfsd->nd_cr;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct nameidata nd;
caddr_t bpos;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int error = 0, rdonly, len, dirfor_ret = 1, diraft_ret = 1;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
int getret = 1, v3 = (nfsd->nd_flag & ND_NFSV3);
Second batch of cleanup changes. This time mostly making a lot of things static and some unused variables here and there.
1995-10-29 15:33:36 +00:00
struct mbuf *mb, *mreq;
Convert old style (type foo *)0 casts to NULLs PR: kern/40360 Requested by: Hiten PAndya via direct email
2002-07-11 17:54:58 +00:00
struct vnode *vp = NULL, *xp, *dirp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct vattr dirfor, diraft, at;
nfsfh_t nfh, dnfh;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhandle_t *fhp, *dfhp;
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
struct mount *mp = NULL;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsdbprintf(("%s %d\n", __FILE__, __LINE__));
ndclear(&nd);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhp = &nfh.fh_generic;
dfhp = &dnfh.fh_generic;
nfsm_srvmtofh(fhp);
Fix some problems that were introduced in revision 1.97. Instead of returning an error code to the caller, NFS server op routines must themselves build an error reply and return 0 to the caller. This is achieved by replacing the erroneous return statements with code that jumps forward to the op function's reply code. We need to be careful to ensure that the 'struct mount' pointer is NULL though, so that the final vn_finished_write() call becomes a no-op. Reviewed by: mckusick, dillon
2001-02-09 13:24:06 +00:00
if ((mp = vfs_getvfs(&fhp->fh_fsid)) == NULL) {
error = ESTALE;
goto ereply;
}
if ((error = VFS_FHTOVP(mp, &fhp->fh_fid, &vp)) != 0) {
mp = NULL;
goto ereply;
}
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
(void) vn_start_write(vp, &mp, V_WAIT);
vput(vp);
vp = NULL;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nfsm_srvmtofh(dfhp);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_srvnamesiz(len);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
error = nfsrv_fhtovp(fhp, FALSE, &vp, cred, slp, nam, &rdonly, TRUE);
NFS Jumbo commit part 1. Cosmetic and structural changes only. The aim of this part of commits is to minimize unnecessary differences between the other NFS's of similar origin. Yes, there are gratuitous changes here that the style folks won't like, but it makes the catch-up less difficult.
1998-05-31 17:27:58 +00:00
if (error) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_POSTOPATTR(v3) + NFSX_WCCDATA(v3));
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (v3) {
nfsm_srvpostop_attr(getret, &at);
nfsm_srvwcc_data(dirfor_ret, &dirfor, diraft_ret, &diraft);
}
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
error = 0;
goto nfsmout;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
The local fs interface does not allow link()/unlink() of directories, do not allow a remote nfs client to cause local fs corruption either.
1997-03-25 05:08:28 +00:00
if (vp->v_type == VDIR) {
error = EPERM; /* POSIX */
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto out1;
The local fs interface does not allow link()/unlink() of directories, do not allow a remote nfs client to cause local fs corruption either.
1997-03-25 05:08:28 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nd.ni_cnd.cn_cred = cred;
nd.ni_cnd.cn_nameiop = CREATE;
nd.ni_cnd.cn_flags = LOCKPARENT;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
error = nfs_namei(&nd, dfhp, len, slp, nam, &md, &dpos,
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
&dirp, td, FALSE);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (dirp) {
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (v3) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
dirfor_ret = VOP_GETATTR(dirp, &dirfor, cred,
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
td);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
} else {
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(dirp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
dirp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
}
Prototyping and general gcc-shutting up. Gcc has one warning now which looks bad, I will get to it eventually, unless somebody beats me to it.
1994-10-02 17:27:07 +00:00
if (error)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto out1;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
xp = nd.ni_vp;
if (xp != NULL) {
error = EEXIST;
goto out;
}
xp = nd.ni_dvp;
if (vp->v_mount != xp->v_mount)
error = EXDEV;
out:
if (!error) {
Use the correct (relative to the implementation) ordering of args in the VOP_LINK() calls, Closes PR#3064 Submitted by: bde
1997-03-25 05:13:40 +00:00
error = VOP_LINK(nd.ni_dvp, vp, &nd.ni_cnd);
Introduce NDFREE (and remove VOP_ABORTOP)
1999-12-15 23:02:35 +00:00
NDFREE(&nd, NDF_ONLY_PNBUF);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/* fall through */
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
out1:
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (v3)
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
getret = VOP_GETATTR(vp, &at, cred, td);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (dirp)
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
diraft_ret = VOP_GETATTR(dirp, &diraft, cred, td);
Fix some problems that were introduced in revision 1.97. Instead of returning an error code to the caller, NFS server op routines must themselves build an error reply and return 0 to the caller. This is achieved by replacing the erroneous return statements with code that jumps forward to the op function's reply code. We need to be careful to ensure that the 'struct mount' pointer is NULL though, so that the final vn_finished_write() call becomes a no-op. Reviewed by: mckusick, dillon
2001-02-09 13:24:06 +00:00
ereply:
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_POSTOPATTR(v3) + NFSX_WCCDATA(v3));
if (v3) {
nfsm_srvpostop_attr(getret, &at);
nfsm_srvwcc_data(dirfor_ret, &dirfor, diraft_ret, &diraft);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/* fall through */
nfsmout:
Introduce NDFREE (and remove VOP_ABORTOP)
1999-12-15 23:02:35 +00:00
NDFREE(&nd, NDF_ONLY_PNBUF);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (dirp)
vrele(dirp);
if (vp)
vrele(vp);
if (nd.ni_dvp) {
if (nd.ni_dvp == nd.ni_vp)
vrele(nd.ni_dvp);
else
vput(nd.ni_dvp);
}
if (nd.ni_vp)
vrele(nd.ni_vp);
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
vn_finished_write(mp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
return(error);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
/*
* nfs symbolic link service
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrv_symlink(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
struct thread *td, struct mbuf **mrq)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md;
Fix all areas of the system (or at least all those in LINT) to avoid storing socket addresses in mbufs. (Socket buffers are the one exception.) A number of kernel APIs needed to get fixed in order to make this happen. Also, fix three protocol families which kept PCBs in mbufs to not malloc them instead. Delete some old compatibility cruft while we're at it, and add some new routines in the in_cksum family.
1997-08-16 19:16:27 +00:00
struct sockaddr *nam = nfsd->nd_nam;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
caddr_t dpos = nfsd->nd_dpos;
struct ucred *cred = &nfsd->nd_cr;
struct vattr va, dirfor, diraft;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct nameidata nd;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct vattr *vap = &va;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct nfsv2_sattr *sp;
Convert old style (type foo *)0 casts to NULLs PR: kern/40360 Requested by: Hiten PAndya via direct email
2002-07-11 17:54:58 +00:00
char *bpos, *pathcp = NULL;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct uio io;
struct iovec iv;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int error = 0, len, len2, dirfor_ret = 1, diraft_ret = 1;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
int v3 = (nfsd->nd_flag & ND_NFSV3);
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct mbuf *mb, *mreq;
Convert old style (type foo *)0 casts to NULLs PR: kern/40360 Requested by: Hiten PAndya via direct email
2002-07-11 17:54:58 +00:00
struct vnode *dirp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsfh_t nfh;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhandle_t *fhp;
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
struct mount *mp = NULL;
struct vnode *vp;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsdbprintf(("%s %d\n", __FILE__, __LINE__));
ndclear(&nd);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhp = &nfh.fh_generic;
nfsm_srvmtofh(fhp);
Fix some problems that were introduced in revision 1.97. Instead of returning an error code to the caller, NFS server op routines must themselves build an error reply and return 0 to the caller. This is achieved by replacing the erroneous return statements with code that jumps forward to the op function's reply code. We need to be careful to ensure that the 'struct mount' pointer is NULL though, so that the final vn_finished_write() call becomes a no-op. Reviewed by: mckusick, dillon
2001-02-09 13:24:06 +00:00
if ((mp = vfs_getvfs(&fhp->fh_fsid)) == NULL) {
error = ESTALE;
goto out;
}
if ((error = VFS_FHTOVP(mp, &fhp->fh_fid, &vp)) != 0) {
mp = NULL;
goto out;
}
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
(void) vn_start_write(vp, &mp, V_WAIT);
vput(vp);
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_srvnamesiz(len);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nd.ni_cnd.cn_cred = cred;
nd.ni_cnd.cn_nameiop = CREATE;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nd.ni_cnd.cn_flags = LOCKPARENT | SAVESTART;
error = nfs_namei(&nd, fhp, len, slp, nam, &md, &dpos,
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
&dirp, td, FALSE);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (dirp) {
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (v3) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
dirfor_ret = VOP_GETATTR(dirp, &dirfor, cred,
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
td);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
} else {
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(dirp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
dirp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
}
Prototyping and general gcc-shutting up. Gcc has one warning now which looks bad, I will get to it eventually, unless somebody beats me to it.
1994-10-02 17:27:07 +00:00
if (error)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto out;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
VATTR_NULL(vap);
if (v3)
nfsm_srvsattr(vap);
- Introduce a new macro, since that's what nfs loves, called nfsm_srvpathsiz. This macro plucks a length out of an rpc request and verifies that its size does not exceed NFS_MAXPATHLEN. If it does it generates an ENAMETOOLONG response. - Use this macro, and the existing nfsm_srvnamsiz macro in two places where we deal with paths passed in by the client. This fixes a linux interoperability bug. Linux was sending oversized path components which would cause us to ignore the request all together. This causes linux to hang indefinitly while it waits for a response. This could still happen in other cases where we error out with EBADRPC. Sponsored by: Isilon Systems, Inc. Reviewed by: alfred, fabbri@isilon.com, neal@isilon.com
2002-10-31 22:35:03 +00:00
nfsm_srvpathsiz(len2);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
MALLOC(pathcp, caddr_t, len2 + 1, M_TEMP, M_WAITOK);
iv.iov_base = pathcp;
iv.iov_len = len2;
io.uio_resid = len2;
io.uio_offset = 0;
io.uio_iov = &iv;
io.uio_iovcnt = 1;
io.uio_segflg = UIO_SYSSPACE;
io.uio_rw = UIO_READ;
Convert old style (type foo *)0 casts to NULLs PR: kern/40360 Requested by: Hiten PAndya via direct email
2002-07-11 17:54:58 +00:00
io.uio_td = NULL;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nfsm_mtouio(&io, len2);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (!v3) {
Make nfsm_dissect() have an obvious return value.
2001-09-27 22:40:38 +00:00
sp = nfsm_dissect(struct nfsv2_sattr *, NFSX_V2SATTR);
It is required by VOP_CREATE, VOP_MKNOD, VOP_SYMLINK and VOP_MKDIR that va_mode of the supplied attributes is filled in with a valid file mode (i.e not VNOVAL, and only ALLPERM bits set). However, some NFS server op functions didn't guarantee this for all possible request messages: If a V3 client chose not include to a mode specification, we could end up creating an ffs inode with mode 0177777, requiring a manual fsck on the next reboot. Fix this by setting va_mode to 0 before calling the VOP if a mode hasn't been supplied by the client. In nfsrv_symlink(), S_IFMT bits supplied by a V2 client could end up in the va_mode passed to VOP_SYMLINK with similar effects. We now use the macro nfstov_mode() to correctly mask the bits.
2002-01-13 05:36:05 +00:00
vap->va_mode = nfstov_mode(sp->sa_mode);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*(pathcp + len2) = '\0';
if (nd.ni_vp) {
error = EEXIST;
goto out;
}
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/*
* issue symlink op. SAVESTART is set so the underlying path component
Remove WILLRELE from VOP_SYMLINK Note: Previous commit to these files (except coda_vnops and devfs_vnops) that claimed to remove WILLRELE from VOP_RENAME actually removed it from VOP_MKNOD.
1999-11-13 20:58:17 +00:00
* is only freed by the VOP if an error occurs.
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
*/
It is required by VOP_CREATE, VOP_MKNOD, VOP_SYMLINK and VOP_MKDIR that va_mode of the supplied attributes is filled in with a valid file mode (i.e not VNOVAL, and only ALLPERM bits set). However, some NFS server op functions didn't guarantee this for all possible request messages: If a V3 client chose not include to a mode specification, we could end up creating an ffs inode with mode 0177777, requiring a manual fsck on the next reboot. Fix this by setting va_mode to 0 before calling the VOP if a mode hasn't been supplied by the client. In nfsrv_symlink(), S_IFMT bits supplied by a V2 client could end up in the va_mode passed to VOP_SYMLINK with similar effects. We now use the macro nfstov_mode() to correctly mask the bits.
2002-01-13 05:36:05 +00:00
if (vap->va_mode == (mode_t)VNOVAL)
vap->va_mode = 0;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
error = VOP_SYMLINK(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, vap, pathcp);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (error)
Introduce NDFREE (and remove VOP_ABORTOP)
1999-12-15 23:02:35 +00:00
NDFREE(&nd, NDF_ONLY_PNBUF);
When VOP_SYMLINK fails, the value of *vpp is junk, so we must NULL out nd.ni_vp to prevent the resource cleanup code at the end of nfsrv_symlink from trying to vrele it. This fixes a "vrele: negative ref cnt" panic that can occur when a symlink is attempted on an NFS filesystem with no free space. Found locally, but the symptoms correspond to those in the PR referenced below. PR: kern/26878 MFC after: 3 days
2001-12-04 16:53:42 +00:00
else
Remove WILLRELE from VOP_SYMLINK Note: Previous commit to these files (except coda_vnops and devfs_vnops) that claimed to remove WILLRELE from VOP_RENAME actually removed it from VOP_MKNOD.
1999-11-13 20:58:17 +00:00
vput(nd.ni_vp);
When VOP_SYMLINK fails, the value of *vpp is junk, so we must NULL out nd.ni_vp to prevent the resource cleanup code at the end of nfsrv_symlink from trying to vrele it. This fixes a "vrele: negative ref cnt" panic that can occur when a symlink is attempted on an NFS filesystem with no free space. Found locally, but the symptoms correspond to those in the PR referenced below. PR: kern/26878 MFC after: 3 days
2001-12-04 16:53:42 +00:00
nd.ni_vp = NULL;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/*
* releases directory prior to potential lookup op.
*/
vput(nd.ni_dvp);
nd.ni_dvp = NULL;
if (error == 0) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (v3) {
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/*
* Issue lookup. Leave SAVESTART set so we can easily free
* the name buffer later on.
*
* since LOCKPARENT is not set, ni_dvp will be garbage on
* return whether an error occurs or not.
*/
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nd.ni_cnd.cn_nameiop = LOOKUP;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nd.ni_cnd.cn_flags &= ~(LOCKPARENT | FOLLOW);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nd.ni_cnd.cn_flags |= (NOFOLLOW | LOCKLEAF);
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
nd.ni_cnd.cn_thread = td;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nd.ni_cnd.cn_cred = cred;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
error = lookup(&nd);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nd.ni_dvp = NULL;
if (error == 0) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
bzero((caddr_t)fhp, sizeof(nfh));
fhp->fh_fsid = nd.ni_vp->v_mount->mnt_stat.f_fsid;
error = VFS_VPTOFH(nd.ni_vp, &fhp->fh_fid);
if (!error)
error = VOP_GETATTR(nd.ni_vp, vap, cred,
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
td);
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vput(nd.ni_vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nd.ni_vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
}
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
out:
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/*
* These releases aren't strictly required, does even doing them
* make any sense? XXX can nfsm_reply() block?
*/
if (pathcp) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
FREE(pathcp, M_TEMP);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
pathcp = NULL;
}
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (dirp) {
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
diraft_ret = VOP_GETATTR(dirp, &diraft, cred, td);
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(dirp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
dirp = NULL;
}
if (nd.ni_startdir) {
vrele(nd.ni_startdir);
nd.ni_startdir = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
nfsm_reply(NFSX_SRVFH(v3) + NFSX_POSTOPATTR(v3) + NFSX_WCCDATA(v3));
if (v3) {
if (!error) {
nfsm_srvpostop_fh(fhp);
nfsm_srvpostop_attr(0, vap);
}
nfsm_srvwcc_data(dirfor_ret, &dirfor, diraft_ret, &diraft);
}
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
/* fall through */
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nfsmout:
Introduce NDFREE (and remove VOP_ABORTOP)
1999-12-15 23:02:35 +00:00
NDFREE(&nd, NDF_ONLY_PNBUF);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (nd.ni_dvp) {
if (nd.ni_dvp == nd.ni_vp)
vrele(nd.ni_dvp);
else
vput(nd.ni_dvp);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (nd.ni_vp)
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(nd.ni_vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (nd.ni_startdir)
vrele(nd.ni_startdir);
if (dirp)
vrele(dirp);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (pathcp)
FREE(pathcp, M_TEMP);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
vn_finished_write(mp);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return (error);
}
/*
* nfs mkdir service
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrv_mkdir(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
struct thread *td, struct mbuf **mrq)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md;
Fix all areas of the system (or at least all those in LINT) to avoid storing socket addresses in mbufs. (Socket buffers are the one exception.) A number of kernel APIs needed to get fixed in order to make this happen. Also, fix three protocol families which kept PCBs in mbufs to not malloc them instead. Delete some old compatibility cruft while we're at it, and add some new routines in the in_cksum family.
1997-08-16 19:16:27 +00:00
struct sockaddr *nam = nfsd->nd_nam;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
caddr_t dpos = nfsd->nd_dpos;
struct ucred *cred = &nfsd->nd_cr;
struct vattr va, dirfor, diraft;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct vattr *vap = &va;
struct nfs_fattr *fp;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct nameidata nd;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
u_int32_t *tl;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
caddr_t bpos;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int error = 0, len, dirfor_ret = 1, diraft_ret = 1;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
int v3 = (nfsd->nd_flag & ND_NFSV3);
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct mbuf *mb, *mreq;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
struct vnode *dirp = NULL;
int vpexcl = 0;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsfh_t nfh;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhandle_t *fhp;
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
struct mount *mp = NULL;
struct vnode *vp;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsdbprintf(("%s %d\n", __FILE__, __LINE__));
ndclear(&nd);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhp = &nfh.fh_generic;
nfsm_srvmtofh(fhp);
Fix some problems that were introduced in revision 1.97. Instead of returning an error code to the caller, NFS server op routines must themselves build an error reply and return 0 to the caller. This is achieved by replacing the erroneous return statements with code that jumps forward to the op function's reply code. We need to be careful to ensure that the 'struct mount' pointer is NULL though, so that the final vn_finished_write() call becomes a no-op. Reviewed by: mckusick, dillon
2001-02-09 13:24:06 +00:00
if ((mp = vfs_getvfs(&fhp->fh_fsid)) == NULL) {
error = ESTALE;
goto out;
}
if ((error = VFS_FHTOVP(mp, &fhp->fh_fid, &vp)) != 0) {
mp = NULL;
goto out;
}
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
(void) vn_start_write(vp, &mp, V_WAIT);
vput(vp);
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_srvnamesiz(len);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nd.ni_cnd.cn_cred = cred;
nd.ni_cnd.cn_nameiop = CREATE;
nd.ni_cnd.cn_flags = LOCKPARENT;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
error = nfs_namei(&nd, fhp, len, slp, nam, &md, &dpos,
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
&dirp, td, FALSE);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (dirp) {
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (v3) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
dirfor_ret = VOP_GETATTR(dirp, &dirfor, cred,
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
td);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
} else {
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(dirp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
dirp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
}
if (error) {
nfsm_reply(NFSX_WCCDATA(v3));
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (v3)
nfsm_srvwcc_data(dirfor_ret, &dirfor, diraft_ret, &diraft);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
VATTR_NULL(vap);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (v3) {
nfsm_srvsattr(vap);
} else {
Make nfsm_dissect() have an obvious return value.
2001-09-27 22:40:38 +00:00
tl = nfsm_dissect(u_int32_t *, NFSX_UNSIGNED);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
vap->va_mode = nfstov_mode(*tl++);
}
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/*
* At this point nd.ni_dvp is referenced and exclusively locked and
* nd.ni_vp, if it exists, is referenced but not locked.
*/
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
vap->va_type = VDIR;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (nd.ni_vp != NULL) {
Introduce NDFREE (and remove VOP_ABORTOP)
1999-12-15 23:02:35 +00:00
NDFREE(&nd, NDF_ONLY_PNBUF);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
error = EEXIST;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
goto out;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/*
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
* Issue mkdir op. Since SAVESTART is not set, the pathname
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
* component is freed by the VOP call. This will fill-in
* nd.ni_vp, reference, and exclusively lock it.
*/
It is required by VOP_CREATE, VOP_MKNOD, VOP_SYMLINK and VOP_MKDIR that va_mode of the supplied attributes is filled in with a valid file mode (i.e not VNOVAL, and only ALLPERM bits set). However, some NFS server op functions didn't guarantee this for all possible request messages: If a V3 client chose not include to a mode specification, we could end up creating an ffs inode with mode 0177777, requiring a manual fsck on the next reboot. Fix this by setting va_mode to 0 before calling the VOP if a mode hasn't been supplied by the client. In nfsrv_symlink(), S_IFMT bits supplied by a V2 client could end up in the va_mode passed to VOP_SYMLINK with similar effects. We now use the macro nfstov_mode() to correctly mask the bits.
2002-01-13 05:36:05 +00:00
if (vap->va_mode == (mode_t)VNOVAL)
vap->va_mode = 0;
Prototyping and general gcc-shutting up. Gcc has one warning now which looks bad, I will get to it eventually, unless somebody beats me to it.
1994-10-02 17:27:07 +00:00
error = VOP_MKDIR(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, vap);
Introduce NDFREE (and remove VOP_ABORTOP)
1999-12-15 23:02:35 +00:00
NDFREE(&nd, NDF_ONLY_PNBUF);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vpexcl = 1;
In the words of the submitter: --------- Make callers of namei() responsible for releasing references or locks instead of having the underlying filesystems do it. This eliminates redundancy in all terminal filesystems and makes it possible for stacked transport layers such as umapfs or nullfs to operate correctly. Quality testing was done with testvn, and lat_fs from the lmbench suite. Some NFS client testing courtesy of Patrik Kudo. vop_mknod and vop_symlink still release the returned vpp. vop_rename still releases 4 vnode arguments before it returns. These remaining cases will be corrected in the next set of patches. --------- Submitted by: Michael Hancock <michaelh@cet.co.jp>
1998-05-07 04:58:58 +00:00
vput(nd.ni_dvp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nd.ni_dvp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (!error) {
bzero((caddr_t)fhp, sizeof(nfh));
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
fhp->fh_fsid = nd.ni_vp->v_mount->mnt_stat.f_fsid;
error = VFS_VPTOFH(nd.ni_vp, &fhp->fh_fid);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (!error)
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
error = VOP_GETATTR(nd.ni_vp, vap, cred, td);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
out:
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (dirp)
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
diraft_ret = VOP_GETATTR(dirp, &diraft, cred, td);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_SRVFH(v3) + NFSX_POSTOPATTR(v3) + NFSX_WCCDATA(v3));
if (v3) {
if (!error) {
nfsm_srvpostop_fh(fhp);
nfsm_srvpostop_attr(0, vap);
}
nfsm_srvwcc_data(dirfor_ret, &dirfor, diraft_ret, &diraft);
Fix a few NFSv2 issues that slipped in during the big cleanup. The semantics of the nfsm_reply() macro were changed so that the caller has to explicitly handle the V2 error case, whereas before, nfsm_reply() did a `goto nfsmout' then. A few server ops (setattr, readlink, create, mkdir) weren't updated to match, so errors in the V2 case could cause protocol hangs and leaked mbufs. Correct some comments that describe the old nfsm_reply behaviour. [older, harmless nit] Remove the unnecessary `nfsmreply0' label in nfsrv_create(), since for its users, the main `ereply' label does the same thing.
2002-01-12 03:57:25 +00:00
} else if (!error) {
/* v2 non-error case. */
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_srvfhtom(fhp, v3);
Tidy up nfsm_build usage. This is only partially finished.
2001-09-27 02:33:36 +00:00
fp = nfsm_build(struct nfs_fattr *, NFSX_V2FATTR);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_srvfillattr(vap, fp);
}
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
/* fall through */
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nfsmout:
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (dirp)
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(dirp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (nd.ni_dvp) {
Introduce NDFREE (and remove VOP_ABORTOP)
1999-12-15 23:02:35 +00:00
NDFREE(&nd, NDF_ONLY_PNBUF);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (nd.ni_dvp == nd.ni_vp && vpexcl)
vrele(nd.ni_dvp);
else
vput(nd.ni_dvp);
}
if (nd.ni_vp) {
if (vpexcl)
vput(nd.ni_vp);
else
vrele(nd.ni_vp);
}
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
vn_finished_write(mp);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return (error);
}
/*
* nfs rmdir service
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrv_rmdir(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
struct thread *td, struct mbuf **mrq)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md;
Fix all areas of the system (or at least all those in LINT) to avoid storing socket addresses in mbufs. (Socket buffers are the one exception.) A number of kernel APIs needed to get fixed in order to make this happen. Also, fix three protocol families which kept PCBs in mbufs to not malloc them instead. Delete some old compatibility cruft while we're at it, and add some new routines in the in_cksum family.
1997-08-16 19:16:27 +00:00
struct sockaddr *nam = nfsd->nd_nam;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
caddr_t dpos = nfsd->nd_dpos;
struct ucred *cred = &nfsd->nd_cr;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
caddr_t bpos;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int error = 0, len, dirfor_ret = 1, diraft_ret = 1;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
int v3 = (nfsd->nd_flag & ND_NFSV3);
Second batch of cleanup changes. This time mostly making a lot of things static and some unused variables here and there.
1995-10-29 15:33:36 +00:00
struct mbuf *mb, *mreq;
Convert old style (type foo *)0 casts to NULLs PR: kern/40360 Requested by: Hiten PAndya via direct email
2002-07-11 17:54:58 +00:00
struct vnode *vp, *dirp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct vattr dirfor, diraft;
nfsfh_t nfh;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhandle_t *fhp;
struct nameidata nd;
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
struct mount *mp = NULL;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsdbprintf(("%s %d\n", __FILE__, __LINE__));
ndclear(&nd);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhp = &nfh.fh_generic;
nfsm_srvmtofh(fhp);
Fix some problems that were introduced in revision 1.97. Instead of returning an error code to the caller, NFS server op routines must themselves build an error reply and return 0 to the caller. This is achieved by replacing the erroneous return statements with code that jumps forward to the op function's reply code. We need to be careful to ensure that the 'struct mount' pointer is NULL though, so that the final vn_finished_write() call becomes a no-op. Reviewed by: mckusick, dillon
2001-02-09 13:24:06 +00:00
if ((mp = vfs_getvfs(&fhp->fh_fsid)) == NULL) {
error = ESTALE;
goto out;
}
if ((error = VFS_FHTOVP(mp, &fhp->fh_fid, &vp)) != 0) {
mp = NULL;
goto out;
}
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
(void) vn_start_write(vp, &mp, V_WAIT);
vput(vp);
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_srvnamesiz(len);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nd.ni_cnd.cn_cred = cred;
nd.ni_cnd.cn_nameiop = DELETE;
nd.ni_cnd.cn_flags = LOCKPARENT | LOCKLEAF;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
error = nfs_namei(&nd, fhp, len, slp, nam, &md, &dpos,
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
&dirp, td, FALSE);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (dirp) {
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (v3) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
dirfor_ret = VOP_GETATTR(dirp, &dirfor, cred,
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
td);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
} else {
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(dirp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
dirp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
}
if (error) {
nfsm_reply(NFSX_WCCDATA(v3));
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (v3)
nfsm_srvwcc_data(dirfor_ret, &dirfor, diraft_ret, &diraft);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
vp = nd.ni_vp;
if (vp->v_type != VDIR) {
error = ENOTDIR;
goto out;
}
/*
* No rmdir "." please.
*/
if (nd.ni_dvp == vp) {
error = EINVAL;
goto out;
}
/*
* The root of a mounted filesystem cannot be deleted.
*/
- Replace v_flag with v_iflag and v_vflag - v_vflag is protected by the vnode lock and is used when synchronization with VOP calls is needed. - v_iflag is protected by interlock and is used for dealing with vnode management issues. These flags include X/O LOCK, FREE, DOOMED, etc. - All accesses to v_iflag and v_vflag have either been locked or marked with mp_fixme's. - Many ASSERT_VOP_LOCKED calls have been added where the locking was not clear. - Many functions in vfs_subr.c were restructured to provide for stronger locking. Idea stolen from: BSD/OS
2002-08-04 10:29:36 +00:00
if (vp->v_vflag & VV_ROOT)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
error = EBUSY;
out:
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/*
* Issue or abort op. Since SAVESTART is not set, path name
* component is freed by the VOP after either.
*/
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (!error)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
error = VOP_RMDIR(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
Introduce NDFREE (and remove VOP_ABORTOP)
1999-12-15 23:02:35 +00:00
NDFREE(&nd, NDF_ONLY_PNBUF);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (dirp)
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
diraft_ret = VOP_GETATTR(dirp, &diraft, cred, td);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_WCCDATA(v3));
Correct a problem wherein NFS servers running NFSv2 would not return certain classes of failure responses to the client during a failed remove operation. Submitted by: Ian Dowse <iedowse@maths.tcd.ie>
2002-10-03 21:50:37 +00:00
error = 0;
if (v3)
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_srvwcc_data(dirfor_ret, &dirfor, diraft_ret, &diraft);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/* fall through */
nfsmout:
Introduce NDFREE (and remove VOP_ABORTOP)
1999-12-15 23:02:35 +00:00
NDFREE(&nd, NDF_ONLY_PNBUF);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (dirp)
vrele(dirp);
if (nd.ni_dvp) {
if (nd.ni_dvp == nd.ni_vp)
vrele(nd.ni_dvp);
else
vput(nd.ni_dvp);
}
if (nd.ni_vp)
vput(nd.ni_vp);
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
vn_finished_write(mp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
return(error);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
/*
* nfs readdir service
* - mallocs what it thinks is enough to read
* count rounded up to a multiple of NFS_DIRBLKSIZ <= NFS_MAXREADDIR
* - calls VOP_READDIR()
* - loops around building the reply
* if the output generated exceeds count break out of loop
* The nfsm_clget macro is used here so that the reply will be packed
* tightly in mbuf clusters.
* - it only knows that it has encountered eof when the VOP_READDIR()
* reads nothing
* - as such one readdir rpc will return eof false although you are there
* and then the next will return eof
* - it trims out records with d_fileno == 0
* this doesn't matter for Unix clients, but they might confuse clients
* for other os'.
* NB: It is tempting to set eof to true if the VOP_READDIR() reads less
* than requested, but this may not apply to all filesystems. For
* example, client NFS does not { although it is never remote mounted
* anyhow }
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
* The alternate call nfsrv_readdirplus() does lookups as well.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
* PS: The NFS protocol spec. does not clarify what the "count" byte
* argument is a count of.. just name strings and file id's or the
* entire reply rpc or ...
* I tried just file name and id sizes and it confused the Sun client,
* so I am using the full rpc size now. The "paranoia.." comment refers
* to including the status longwords that are not a part of the dir.
* "entry" structures, but are in the rpc.
*/
struct flrep {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsuint64 fl_off;
For the on-the-wire protocol, u_long -> u_int32_t; long -> int32_t; int -> int32_t; u_short -> u_int16_t. Also, use mode_t instead of u_short for storing modes (mode_t is a u_int16_t). Obtained from: NetBSD
1998-05-31 20:09:01 +00:00
u_int32_t fl_postopok;
u_int32_t fl_fattr[NFSX_V3FATTR / sizeof (u_int32_t)];
u_int32_t fl_fhok;
u_int32_t fl_fhsize;
u_int32_t fl_nfh[NFSX_V3FH / sizeof (u_int32_t)];
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
};
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrv_readdir(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
struct thread *td, struct mbuf **mrq)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md;
Fix all areas of the system (or at least all those in LINT) to avoid storing socket addresses in mbufs. (Socket buffers are the one exception.) A number of kernel APIs needed to get fixed in order to make this happen. Also, fix three protocol families which kept PCBs in mbufs to not malloc them instead. Delete some old compatibility cruft while we're at it, and add some new routines in the in_cksum family.
1997-08-16 19:16:27 +00:00
struct sockaddr *nam = nfsd->nd_nam;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
caddr_t dpos = nfsd->nd_dpos;
struct ucred *cred = &nfsd->nd_cr;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
char *bp, *be;
struct mbuf *mp;
struct dirent *dp;
caddr_t cp;
u_int32_t *tl;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
caddr_t bpos;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct mbuf *mb, *mreq;
char *cpos, *cend, *rbuf;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
struct vnode *vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct vattr at;
nfsfh_t nfh;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhandle_t *fhp;
struct uio io;
struct iovec iv;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
int len, nlen, rem, xfer, tsiz, i, error = 0, getret = 1;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int siz, cnt, fullsiz, eofflag, rdonly, ncookies;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
int v3 = (nfsd->nd_flag & ND_NFSV3);
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
u_quad_t off, toff, verf;
For the on-the-wire protocol, u_long -> u_int32_t; long -> int32_t; int -> int32_t; u_short -> u_int16_t. Also, use mode_t instead of u_short for storing modes (mode_t is a u_int16_t). Obtained from: NetBSD
1998-05-31 20:09:01 +00:00
u_long *cookies = NULL, *cookiep; /* needs to be int64_t or off_t */
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsdbprintf(("%s %d\n", __FILE__, __LINE__));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhp = &nfh.fh_generic;
nfsm_srvmtofh(fhp);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (v3) {
Make nfsm_dissect() have an obvious return value.
2001-09-27 22:40:38 +00:00
tl = nfsm_dissect(u_int32_t *, 5 * NFSX_UNSIGNED);
Various changes lifted from the OpenBSD cvs tree: txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge. nfs_serv.c: don't call nfsm_adj() with negative values, windows clients could crash servers when doing a readdir of a large directory. nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin loop trying to bind(). Don't clobber a mbuf pointer or we get panics on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf. nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB. Obtained from: OpenBSD
1999-06-05 05:35:03 +00:00
toff = fxdr_hyper(tl);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
tl += 2;
Various changes lifted from the OpenBSD cvs tree: txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge. nfs_serv.c: don't call nfsm_adj() with negative values, windows clients could crash servers when doing a readdir of a large directory. nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin loop trying to bind(). Don't clobber a mbuf pointer or we get panics on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf. nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB. Obtained from: OpenBSD
1999-06-05 05:35:03 +00:00
verf = fxdr_hyper(tl);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
tl += 2;
} else {
Make nfsm_dissect() have an obvious return value.
2001-09-27 22:40:38 +00:00
tl = nfsm_dissect(u_int32_t *, 2 * NFSX_UNSIGNED);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
toff = fxdr_unsigned(u_quad_t, *tl++);
Various changes lifted from the OpenBSD cvs tree: txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge. nfs_serv.c: don't call nfsm_adj() with negative values, windows clients could crash servers when doing a readdir of a large directory. nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin loop trying to bind(). Don't clobber a mbuf pointer or we get panics on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf. nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB. Obtained from: OpenBSD
1999-06-05 05:35:03 +00:00
verf = 0; /* shut up gcc */
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
off = toff;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
cnt = fxdr_unsigned(int, *tl);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
siz = ((cnt + DIRBLKSIZ - 1) & ~(DIRBLKSIZ - 1));
xfer = NFS_SRVMAXDATA(nfsd);
Limit to the maximum allowed reply size the amount of data that nfsrv_readdir and nfsrv_readdirplus can return. A client request containing an over-large `count' field could trigger the "Bad nfs svc reply" panic in nfs_syscalls.c. Spotted while trying to reproduce kern/37304, which turned out to be fixed in FreeBSD a long time ago. MFC after: 1 week
2002-04-21 16:14:54 +00:00
if (cnt > xfer)
cnt = xfer;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (siz > xfer)
siz = xfer;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fullsiz = siz;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
error = nfsrv_fhtovp(fhp, 1, &vp, cred, slp, nam, &rdonly, TRUE);
Refuse READDIR / READDIRPLUS rpc's for non-directories Obtained from: NetBSD
1998-05-31 17:54:18 +00:00
if (!error && vp->v_type != VDIR) {
error = ENOTDIR;
vput(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
Refuse READDIR / READDIRPLUS rpc's for non-directories Obtained from: NetBSD
1998-05-31 17:54:18 +00:00
}
NFS Jumbo commit part 1. Cosmetic and structural changes only. The aim of this part of commits is to minimize unnecessary differences between the other NFS's of similar origin. Yes, there are gratuitous changes here that the style folks won't like, but it makes the catch-up less difficult.
1998-05-31 17:27:58 +00:00
if (error) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_UNSIGNED);
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (v3)
nfsm_srvpostop_attr(getret, &at);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/*
* Obtain lock on vnode for this section of the code
*/
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (v3) {
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
error = getret = VOP_GETATTR(vp, &at, cred, td);
Make FreeBSD less conservative in determining when to return a cookie error for a directory. I have made this change after a great deal of review although I cannot be absolutely sure that this meets the spec. The issue devolves into whether changes in an underlying (UFS) directory can cause NFS directory blocks to be renumbered. My read of the code indicates that NFS directory blocks will not be renumbered, which means that the cookies should still remain valid after a change is made to the underlying directory. This being the case, a cookie error should not be returned when a change is made to the underlying directory and, instead, the NFS client should rely on mtime detection to invalidate and reload the directory. The use of mtime is problematic in of itself, due to insufficient resolution, which is why I believe the original conservative error handling was done. Still, there have been dozens of bug reports by people needing solaris<->FreeBSD interoperability and these have to be accomodated.
1999-09-29 17:14:58 +00:00
#if 0
NetBSD has a comment that Solaris 2.5 doesn't do verifiers correctly, we have weakened this test already for Digital Unix, so it may be enough for Solaris. It needs to be checked again. Obtained from: NetBSD
1998-05-31 19:07:47 +00:00
/*
* XXX This check may be too strict for Solaris 2.5 clients.
*/
Allow NULL cookie verifiers for non-NULL offsets. This is needed for Digital Unix boxes since they appear to always send null verifiers.
1997-07-22 15:35:15 +00:00
if (!error && toff && verf && verf != at.va_filerev)
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
error = NFSERR_BAD_COOKIE;
Make FreeBSD less conservative in determining when to return a cookie error for a directory. I have made this change after a great deal of review although I cannot be absolutely sure that this meets the spec. The issue devolves into whether changes in an underlying (UFS) directory can cause NFS directory blocks to be renumbered. My read of the code indicates that NFS directory blocks will not be renumbered, which means that the cookies should still remain valid after a change is made to the underlying directory. This being the case, a cookie error should not be returned when a change is made to the underlying directory and, instead, the NFS client should rely on mtime detection to invalidate and reload the directory. The use of mtime is problematic in of itself, due to insufficient resolution, which is why I believe the original conservative error handling was done. Still, there have been dozens of bug reports by people needing solaris<->FreeBSD interoperability and these have to be accomodated.
1999-09-29 17:14:58 +00:00
#endif
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
if (!error)
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
error = nfsrv_access(vp, VEXEC, cred, rdonly, td, 0);
Prototyping and general gcc-shutting up. Gcc has one warning now which looks bad, I will get to it eventually, unless somebody beats me to it.
1994-10-02 17:27:07 +00:00
if (error) {
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vput(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_POSTOPATTR(v3));
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (v3)
nfsm_srvpostop_attr(getret, &at);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
VOP_UNLOCK(vp, 0, td);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
/*
* end section. Allocate rbuf and continue
*/
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
MALLOC(rbuf, caddr_t, siz, M_TEMP, M_WAITOK);
again:
iv.iov_base = rbuf;
iv.iov_len = fullsiz;
io.uio_iov = &iv;
io.uio_iovcnt = 1;
io.uio_offset = (off_t)off;
io.uio_resid = fullsiz;
io.uio_segflg = UIO_SYSSPACE;
io.uio_rw = UIO_READ;
Convert old style (type foo *)0 casts to NULLs PR: kern/40360 Requested by: Hiten PAndya via direct email
2002-07-11 17:54:58 +00:00
io.uio_td = NULL;
Make NFS ask the filesystems for directory cookies instead of making them itself.
1994-09-28 16:45:22 +00:00
eofflag = 0;
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (cookies) {
free((caddr_t)cookies, M_TEMP);
cookies = NULL;
}
Make NFS ask the filesystems for directory cookies instead of making them itself.
1994-09-28 16:45:22 +00:00
error = VOP_READDIR(vp, &io, cred, &eofflag, &ncookies, &cookies);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
off = (off_t)io.uio_offset;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (!cookies && !error)
error = NFSERR_PERM;
if (v3) {
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
getret = VOP_GETATTR(vp, &at, cred, td);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (!error)
error = getret;
}
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
VOP_UNLOCK(vp, 0, td);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (error) {
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
free((caddr_t)rbuf, M_TEMP);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (cookies)
free((caddr_t)cookies, M_TEMP);
nfsm_reply(NFSX_POSTOPATTR(v3));
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (v3)
nfsm_srvpostop_attr(getret, &at);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
Make NFS ask the filesystems for directory cookies instead of making them itself.
1994-09-28 16:45:22 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (io.uio_resid) {
siz -= io.uio_resid;
/*
* If nothing read, return eof
* rpc reply
*/
if (siz == 0) {
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_POSTOPATTR(v3) + NFSX_COOKIEVERF(v3) +
2 * NFSX_UNSIGNED);
if (v3) {
nfsm_srvpostop_attr(getret, &at);
Tidy up nfsm_build usage. This is only partially finished.
2001-09-27 02:33:36 +00:00
tl = nfsm_build(u_int32_t *, 4 * NFSX_UNSIGNED);
Various changes lifted from the OpenBSD cvs tree: txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge. nfs_serv.c: don't call nfsm_adj() with negative values, windows clients could crash servers when doing a readdir of a large directory. nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin loop trying to bind(). Don't clobber a mbuf pointer or we get panics on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf. nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB. Obtained from: OpenBSD
1999-06-05 05:35:03 +00:00
txdr_hyper(at.va_filerev, tl);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
tl += 2;
} else
Tidy up nfsm_build usage. This is only partially finished.
2001-09-27 02:33:36 +00:00
tl = nfsm_build(u_int32_t *, 2 * NFSX_UNSIGNED);
Rename some variables that end up shadowing their namesakes in the NFS client code. Reviewed by: peter
2002-01-08 19:41:06 +00:00
*tl++ = nfsrv_nfs_false;
*tl = nfsrv_nfs_true;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
FREE((caddr_t)rbuf, M_TEMP);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
FREE((caddr_t)cookies, M_TEMP);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
}
/*
* Check for degenerate cases of nothing useful read.
* If so go try again
*/
Make NFS ask the filesystems for directory cookies instead of making them itself.
1994-09-28 16:45:22 +00:00
cpos = rbuf;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
cend = rbuf + siz;
dp = (struct dirent *)cpos;
Make NFS ask the filesystems for directory cookies instead of making them itself.
1994-09-28 16:45:22 +00:00
cookiep = cookies;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
/*
* For some reason FreeBSD's ufs_readdir() chooses to back the
* directory offset up to a block boundary, so it is necessary to
Preceed/preceeding are not english words. Use precede and preceding.
2001-02-18 10:43:53 +00:00
* skip over the records that precede the requested offset. This
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
* requires the assumption that file offset cookies monotonically
* increase.
*/
while (cpos < cend && ncookies > 0 &&
Hide whiteouts from NFS, since the protocol doesn't support them. Obtained from: NetBSD
1998-05-31 19:10:52 +00:00
(dp->d_fileno == 0 || dp->d_type == DT_WHT ||
((u_quad_t)(*cookiep)) <= toff)) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
cpos += dp->d_reclen;
dp = (struct dirent *)cpos;
Make NFS ask the filesystems for directory cookies instead of making them itself.
1994-09-28 16:45:22 +00:00
cookiep++;
ncookies--;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (cpos >= cend || ncookies == 0) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
toff = off;
siz = fullsiz;
goto again;
}
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
len = 3 * NFSX_UNSIGNED; /* paranoia, probably can be 0 */
nfsm_reply(NFSX_POSTOPATTR(v3) + NFSX_COOKIEVERF(v3) + siz);
if (v3) {
nfsm_srvpostop_attr(getret, &at);
Tidy up nfsm_build usage. This is only partially finished.
2001-09-27 02:33:36 +00:00
tl = nfsm_build(u_int32_t *, 2 * NFSX_UNSIGNED);
Various changes lifted from the OpenBSD cvs tree: txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge. nfs_serv.c: don't call nfsm_adj() with negative values, windows clients could crash servers when doing a readdir of a large directory. nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin loop trying to bind(). Don't clobber a mbuf pointer or we get panics on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf. nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB. Obtained from: OpenBSD
1999-06-05 05:35:03 +00:00
txdr_hyper(at.va_filerev, tl);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
mp = mb;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
bp = bpos;
be = bp + M_TRAILINGSPACE(mp);
/* Loop through the records and build reply */
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
while (cpos < cend && ncookies > 0) {
Hide whiteouts from NFS, since the protocol doesn't support them. Obtained from: NetBSD
1998-05-31 19:10:52 +00:00
if (dp->d_fileno != 0 && dp->d_type != DT_WHT) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nlen = dp->d_namlen;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
rem = nfsm_rndup(nlen) - nlen;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
len += (4 * NFSX_UNSIGNED + nlen + rem);
if (v3)
len += 2 * NFSX_UNSIGNED;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (len > cnt) {
eofflag = 0;
break;
}
/*
* Build the directory record xdr from
* the dirent entry.
*/
nfsm_clget;
Rename some variables that end up shadowing their namesakes in the NFS client code. Reviewed by: peter
2002-01-08 19:41:06 +00:00
*tl = nfsrv_nfs_true;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
bp += NFSX_UNSIGNED;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (v3) {
nfsm_clget;
*tl = 0;
bp += NFSX_UNSIGNED;
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nfsm_clget;
*tl = txdr_unsigned(dp->d_fileno);
bp += NFSX_UNSIGNED;
nfsm_clget;
*tl = txdr_unsigned(nlen);
bp += NFSX_UNSIGNED;
Remove trailing whitespace.
1995-05-30 08:16:23 +00:00
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/* And loop around copying the name */
xfer = nlen;
cp = dp->d_name;
while (xfer > 0) {
nfsm_clget;
if ((bp+xfer) > be)
tsiz = be-bp;
else
tsiz = xfer;
bcopy(cp, bp, tsiz);
bp += tsiz;
xfer -= tsiz;
if (xfer > 0)
cp += tsiz;
}
For the on-the-wire protocol, u_long -> u_int32_t; long -> int32_t; int -> int32_t; u_short -> u_int16_t. Also, use mode_t instead of u_short for storing modes (mode_t is a u_int16_t). Obtained from: NetBSD
1998-05-31 20:09:01 +00:00
/* And null pad to a int32_t boundary */
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
for (i = 0; i < rem; i++)
*bp++ = '\0';
nfsm_clget;
Remove trailing whitespace.
1995-05-30 08:16:23 +00:00
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/* Finish off the record */
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (v3) {
*tl = 0;
bp += NFSX_UNSIGNED;
nfsm_clget;
}
Make NFS ask the filesystems for directory cookies instead of making them itself.
1994-09-28 16:45:22 +00:00
*tl = txdr_unsigned(*cookiep);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
bp += NFSX_UNSIGNED;
Make NFS ask the filesystems for directory cookies instead of making them itself.
1994-09-28 16:45:22 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
cpos += dp->d_reclen;
dp = (struct dirent *)cpos;
Make NFS ask the filesystems for directory cookies instead of making them itself.
1994-09-28 16:45:22 +00:00
cookiep++;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
ncookies--;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nfsm_clget;
Rename some variables that end up shadowing their namesakes in the NFS client code. Reviewed by: peter
2002-01-08 19:41:06 +00:00
*tl = nfsrv_nfs_false;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
bp += NFSX_UNSIGNED;
nfsm_clget;
if (eofflag)
Rename some variables that end up shadowing their namesakes in the NFS client code. Reviewed by: peter
2002-01-08 19:41:06 +00:00
*tl = nfsrv_nfs_true;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
else
Rename some variables that end up shadowing their namesakes in the NFS client code. Reviewed by: peter
2002-01-08 19:41:06 +00:00
*tl = nfsrv_nfs_false;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
bp += NFSX_UNSIGNED;
if (mp != mb) {
if (bp < be)
mp->m_len = bp - mtod(mp, caddr_t);
} else
mp->m_len += bp - bpos;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
FREE((caddr_t)rbuf, M_TEMP);
FREE((caddr_t)cookies, M_TEMP);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsmout:
if (vp)
vrele(vp);
return(error);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrv_readdirplus(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
struct thread *td, struct mbuf **mrq)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md;
Fix all areas of the system (or at least all those in LINT) to avoid storing socket addresses in mbufs. (Socket buffers are the one exception.) A number of kernel APIs needed to get fixed in order to make this happen. Also, fix three protocol families which kept PCBs in mbufs to not malloc them instead. Delete some old compatibility cruft while we're at it, and add some new routines in the in_cksum family.
1997-08-16 19:16:27 +00:00
struct sockaddr *nam = nfsd->nd_nam;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
caddr_t dpos = nfsd->nd_dpos;
struct ucred *cred = &nfsd->nd_cr;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
char *bp, *be;
struct mbuf *mp;
struct dirent *dp;
caddr_t cp;
u_int32_t *tl;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
caddr_t bpos;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct mbuf *mb, *mreq;
char *cpos, *cend, *rbuf;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
struct vnode *vp = NULL, *nvp;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct flrep fl;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsfh_t nfh;
fhandle_t *fhp, *nfhp = (fhandle_t *)fl.fl_nfh;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct uio io;
struct iovec iv;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct vattr va, at, *vap = &va;
struct nfs_fattr *fp;
int len, nlen, rem, xfer, tsiz, i, error = 0, getret = 1;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int siz, cnt, fullsiz, eofflag, rdonly, dirlen, ncookies;
u_quad_t off, toff, verf;
For the on-the-wire protocol, u_long -> u_int32_t; long -> int32_t; int -> int32_t; u_short -> u_int16_t. Also, use mode_t instead of u_short for storing modes (mode_t is a u_int16_t). Obtained from: NetBSD
1998-05-31 20:09:01 +00:00
u_long *cookies = NULL, *cookiep; /* needs to be int64_t or off_t */
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int v3 = (nfsd->nd_flag & ND_NFSV3);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsdbprintf(("%s %d\n", __FILE__, __LINE__));
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (!v3)
panic("nfsrv_readdirplus: v3 proc called on a v2 connection");
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhp = &nfh.fh_generic;
nfsm_srvmtofh(fhp);
Make nfsm_dissect() have an obvious return value.
2001-09-27 22:40:38 +00:00
tl = nfsm_dissect(u_int32_t *, 6 * NFSX_UNSIGNED);
Various changes lifted from the OpenBSD cvs tree: txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge. nfs_serv.c: don't call nfsm_adj() with negative values, windows clients could crash servers when doing a readdir of a large directory. nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin loop trying to bind(). Don't clobber a mbuf pointer or we get panics on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf. nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB. Obtained from: OpenBSD
1999-06-05 05:35:03 +00:00
toff = fxdr_hyper(tl);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
tl += 2;
Various changes lifted from the OpenBSD cvs tree: txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge. nfs_serv.c: don't call nfsm_adj() with negative values, windows clients could crash servers when doing a readdir of a large directory. nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin loop trying to bind(). Don't clobber a mbuf pointer or we get panics on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf. nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB. Obtained from: OpenBSD
1999-06-05 05:35:03 +00:00
verf = fxdr_hyper(tl);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
tl += 2;
siz = fxdr_unsigned(int, *tl++);
cnt = fxdr_unsigned(int, *tl);
off = toff;
siz = ((siz + DIRBLKSIZ - 1) & ~(DIRBLKSIZ - 1));
xfer = NFS_SRVMAXDATA(nfsd);
Limit to the maximum allowed reply size the amount of data that nfsrv_readdir and nfsrv_readdirplus can return. A client request containing an over-large `count' field could trigger the "Bad nfs svc reply" panic in nfs_syscalls.c. Spotted while trying to reproduce kern/37304, which turned out to be fixed in FreeBSD a long time ago. MFC after: 1 week
2002-04-21 16:14:54 +00:00
if (cnt > xfer)
cnt = xfer;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (siz > xfer)
siz = xfer;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fullsiz = siz;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
error = nfsrv_fhtovp(fhp, 1, &vp, cred, slp, nam, &rdonly, TRUE);
Refuse READDIR / READDIRPLUS rpc's for non-directories Obtained from: NetBSD
1998-05-31 17:54:18 +00:00
if (!error && vp->v_type != VDIR) {
error = ENOTDIR;
vput(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
Refuse READDIR / READDIRPLUS rpc's for non-directories Obtained from: NetBSD
1998-05-31 17:54:18 +00:00
}
NFS Jumbo commit part 1. Cosmetic and structural changes only. The aim of this part of commits is to minimize unnecessary differences between the other NFS's of similar origin. Yes, there are gratuitous changes here that the style folks won't like, but it makes the catch-up less difficult.
1998-05-31 17:27:58 +00:00
if (error) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_UNSIGNED);
nfsm_srvpostop_attr(getret, &at);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
error = getret = VOP_GETATTR(vp, &at, cred, td);
Make FreeBSD less conservative in determining when to return a cookie error for a directory. I have made this change after a great deal of review although I cannot be absolutely sure that this meets the spec. The issue devolves into whether changes in an underlying (UFS) directory can cause NFS directory blocks to be renumbered. My read of the code indicates that NFS directory blocks will not be renumbered, which means that the cookies should still remain valid after a change is made to the underlying directory. This being the case, a cookie error should not be returned when a change is made to the underlying directory and, instead, the NFS client should rely on mtime detection to invalidate and reload the directory. The use of mtime is problematic in of itself, due to insufficient resolution, which is why I believe the original conservative error handling was done. Still, there have been dozens of bug reports by people needing solaris<->FreeBSD interoperability and these have to be accomodated.
1999-09-29 17:14:58 +00:00
#if 0
NetBSD has a comment that Solaris 2.5 doesn't do verifiers correctly, we have weakened this test already for Digital Unix, so it may be enough for Solaris. It needs to be checked again. Obtained from: NetBSD
1998-05-31 19:07:47 +00:00
/*
* XXX This check may be too strict for Solaris 2.5 clients.
*/
Allow NULL cookie verifiers for non-NULL offsets. This is needed for Digital Unix boxes since they appear to always send null verifiers.
1997-07-22 15:35:15 +00:00
if (!error && toff && verf && verf != at.va_filerev)
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
error = NFSERR_BAD_COOKIE;
Make FreeBSD less conservative in determining when to return a cookie error for a directory. I have made this change after a great deal of review although I cannot be absolutely sure that this meets the spec. The issue devolves into whether changes in an underlying (UFS) directory can cause NFS directory blocks to be renumbered. My read of the code indicates that NFS directory blocks will not be renumbered, which means that the cookies should still remain valid after a change is made to the underlying directory. This being the case, a cookie error should not be returned when a change is made to the underlying directory and, instead, the NFS client should rely on mtime detection to invalidate and reload the directory. The use of mtime is problematic in of itself, due to insufficient resolution, which is why I believe the original conservative error handling was done. Still, there have been dozens of bug reports by people needing solaris<->FreeBSD interoperability and these have to be accomodated.
1999-09-29 17:14:58 +00:00
#endif
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (!error)
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
error = nfsrv_access(vp, VEXEC, cred, rdonly, td, 0);
Prototyping and general gcc-shutting up. Gcc has one warning now which looks bad, I will get to it eventually, unless somebody beats me to it.
1994-10-02 17:27:07 +00:00
if (error) {
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vput(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_V3POSTOPATTR);
nfsm_srvpostop_attr(getret, &at);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
VOP_UNLOCK(vp, 0, td);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
MALLOC(rbuf, caddr_t, siz, M_TEMP, M_WAITOK);
again:
iv.iov_base = rbuf;
iv.iov_len = fullsiz;
io.uio_iov = &iv;
io.uio_iovcnt = 1;
io.uio_offset = (off_t)off;
io.uio_resid = fullsiz;
io.uio_segflg = UIO_SYSSPACE;
io.uio_rw = UIO_READ;
Convert old style (type foo *)0 casts to NULLs PR: kern/40360 Requested by: Hiten PAndya via direct email
2002-07-11 17:54:58 +00:00
io.uio_td = NULL;
Make NFS ask the filesystems for directory cookies instead of making them itself.
1994-09-28 16:45:22 +00:00
eofflag = 0;
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, td);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (cookies) {
free((caddr_t)cookies, M_TEMP);
cookies = NULL;
}
Make NFS ask the filesystems for directory cookies instead of making them itself.
1994-09-28 16:45:22 +00:00
error = VOP_READDIR(vp, &io, cred, &eofflag, &ncookies, &cookies);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
off = (u_quad_t)io.uio_offset;
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
getret = VOP_GETATTR(vp, &at, cred, td);
VOP_UNLOCK(vp, 0, td);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (!cookies && !error)
error = NFSERR_PERM;
if (!error)
error = getret;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (error) {
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (cookies)
free((caddr_t)cookies, M_TEMP);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
free((caddr_t)rbuf, M_TEMP);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_V3POSTOPATTR);
nfsm_srvpostop_attr(getret, &at);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
Make NFS ask the filesystems for directory cookies instead of making them itself.
1994-09-28 16:45:22 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (io.uio_resid) {
siz -= io.uio_resid;
/*
* If nothing read, return eof
* rpc reply
*/
if (siz == 0) {
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_V3POSTOPATTR + NFSX_V3COOKIEVERF +
2 * NFSX_UNSIGNED);
nfsm_srvpostop_attr(getret, &at);
Tidy up nfsm_build usage. This is only partially finished.
2001-09-27 02:33:36 +00:00
tl = nfsm_build(u_int32_t *, 4 * NFSX_UNSIGNED);
Various changes lifted from the OpenBSD cvs tree: txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge. nfs_serv.c: don't call nfsm_adj() with negative values, windows clients could crash servers when doing a readdir of a large directory. nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin loop trying to bind(). Don't clobber a mbuf pointer or we get panics on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf. nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB. Obtained from: OpenBSD
1999-06-05 05:35:03 +00:00
txdr_hyper(at.va_filerev, tl);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
tl += 2;
Rename some variables that end up shadowing their namesakes in the NFS client code. Reviewed by: peter
2002-01-08 19:41:06 +00:00
*tl++ = nfsrv_nfs_false;
*tl = nfsrv_nfs_true;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
FREE((caddr_t)cookies, M_TEMP);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
FREE((caddr_t)rbuf, M_TEMP);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
}
/*
* Check for degenerate cases of nothing useful read.
* If so go try again
*/
Make NFS ask the filesystems for directory cookies instead of making them itself.
1994-09-28 16:45:22 +00:00
cpos = rbuf;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
cend = rbuf + siz;
dp = (struct dirent *)cpos;
Make NFS ask the filesystems for directory cookies instead of making them itself.
1994-09-28 16:45:22 +00:00
cookiep = cookies;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
/*
* For some reason FreeBSD's ufs_readdir() chooses to back the
* directory offset up to a block boundary, so it is necessary to
Preceed/preceeding are not english words. Use precede and preceding.
2001-02-18 10:43:53 +00:00
* skip over the records that precede the requested offset. This
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
* requires the assumption that file offset cookies monotonically
* increase.
*/
while (cpos < cend && ncookies > 0 &&
Cut-n-paste glitch
1998-05-31 19:43:34 +00:00
(dp->d_fileno == 0 || dp->d_type == DT_WHT ||
Hide whiteouts from NFS, since the protocol doesn't support them. Obtained from: NetBSD
1998-05-31 19:10:52 +00:00
((u_quad_t)(*cookiep)) <= toff)) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
cpos += dp->d_reclen;
dp = (struct dirent *)cpos;
Make NFS ask the filesystems for directory cookies instead of making them itself.
1994-09-28 16:45:22 +00:00
cookiep++;
ncookies--;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (cpos >= cend || ncookies == 0) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
toff = off;
siz = fullsiz;
goto again;
}
Slight changes to locking around VOP_READRIR. Detect in nfsrv_readdirplus when a filesystem soes not support VFS_VGET and return NFSERR_NOTSUPP so that the client will use ordinary readdir instead.
1995-08-03 12:14:16 +00:00
/*
* Probe one of the directory entries to see if the filesystem
* supports VGET.
*/
Add a flags parameter to VFS_VGET to pass through the desired locking flags when acquiring a vnode. The immediate purpose is to allow polling lock requests (LK_NOWAIT) needed by soft updates to avoid deadlock when enlisting other processes to help with the background cleanup. For the future it will allow the use of shared locks for read access to vnodes. This change touches a lot of files as it affects most filesystems within the system. It has been well tested on FFS, loopback, and CD-ROM filesystems. only lightly on the others, so if you find a problem there, please let me (mckusick@mckusick.com) know.
2002-03-17 01:25:47 +00:00
if (VFS_VGET(vp->v_mount, dp->d_fileno, LK_EXCLUSIVE, &nvp) ==
EOPNOTSUPP) {
Slight changes to locking around VOP_READRIR. Detect in nfsrv_readdirplus when a filesystem soes not support VFS_VGET and return NFSERR_NOTSUPP so that the client will use ordinary readdir instead.
1995-08-03 12:14:16 +00:00
error = NFSERR_NOTSUPP;
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
Slight changes to locking around VOP_READRIR. Detect in nfsrv_readdirplus when a filesystem soes not support VFS_VGET and return NFSERR_NOTSUPP so that the client will use ordinary readdir instead.
1995-08-03 12:14:16 +00:00
free((caddr_t)cookies, M_TEMP);
free((caddr_t)rbuf, M_TEMP);
nfsm_reply(NFSX_V3POSTOPATTR);
nfsm_srvpostop_attr(getret, &at);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
Slight changes to locking around VOP_READRIR. Detect in nfsrv_readdirplus when a filesystem soes not support VFS_VGET and return NFSERR_NOTSUPP so that the client will use ordinary readdir instead.
1995-08-03 12:14:16 +00:00
}
Release an unneeded reference to a vnode that was gained in a VFS_VGET(). Fixes a readdirplus panic. Submitted by: Doug Rabson <dfr@render.com>
1996-09-05 07:58:04 +00:00
vput(nvp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nvp = NULL;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
Tidy up nfsm_build usage. This is only partially finished.
2001-09-27 02:33:36 +00:00
dirlen = len = NFSX_V3POSTOPATTR + NFSX_V3COOKIEVERF +
2 * NFSX_UNSIGNED;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(cnt);
nfsm_srvpostop_attr(getret, &at);
Tidy up nfsm_build usage. This is only partially finished.
2001-09-27 02:33:36 +00:00
tl = nfsm_build(u_int32_t *, 2 * NFSX_UNSIGNED);
Various changes lifted from the OpenBSD cvs tree: txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge. nfs_serv.c: don't call nfsm_adj() with negative values, windows clients could crash servers when doing a readdir of a large directory. nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin loop trying to bind(). Don't clobber a mbuf pointer or we get panics on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf. nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB. Obtained from: OpenBSD
1999-06-05 05:35:03 +00:00
txdr_hyper(at.va_filerev, tl);
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
mp = mb;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
bp = bpos;
be = bp + M_TRAILINGSPACE(mp);
/* Loop through the records and build reply */
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
while (cpos < cend && ncookies > 0) {
Hide whiteouts from NFS, since the protocol doesn't support them. Obtained from: NetBSD
1998-05-31 19:10:52 +00:00
if (dp->d_fileno != 0 && dp->d_type != DT_WHT) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nlen = dp->d_namlen;
rem = nfsm_rndup(nlen)-nlen;
Remove trailing whitespace.
1995-05-30 08:16:23 +00:00
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* For readdir_and_lookup get the vnode using
* the file number.
*/
Add a flags parameter to VFS_VGET to pass through the desired locking flags when acquiring a vnode. The immediate purpose is to allow polling lock requests (LK_NOWAIT) needed by soft updates to avoid deadlock when enlisting other processes to help with the background cleanup. For the future it will allow the use of shared locks for read access to vnodes. This change touches a lot of files as it affects most filesystems within the system. It has been well tested on FFS, loopback, and CD-ROM filesystems. only lightly on the others, so if you find a problem there, please let me (mckusick@mckusick.com) know.
2002-03-17 01:25:47 +00:00
if (VFS_VGET(vp->v_mount, dp->d_fileno, LK_EXCLUSIVE,
&nvp))
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto invalid;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
bzero((caddr_t)nfhp, NFSX_V3FH);
nfhp->fh_fsid =
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nvp->v_mount->mnt_stat.f_fsid;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (VFS_VPTOFH(nvp, &nfhp->fh_fid)) {
Changed some (incorrect) nfsrv_vput()'s back into regular vput()'s. This fixes the last of the known NQNFS problems (until I find more, that is :-)).
1995-03-17 07:45:19 +00:00
vput(nvp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nvp = NULL;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto invalid;
}
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
if (VOP_GETATTR(nvp, vap, cred, td)) {
Changed some (incorrect) nfsrv_vput()'s back into regular vput()'s. This fixes the last of the known NQNFS problems (until I find more, that is :-)).
1995-03-17 07:45:19 +00:00
vput(nvp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nvp = NULL;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto invalid;
}
Changed some (incorrect) nfsrv_vput()'s back into regular vput()'s. This fixes the last of the known NQNFS problems (until I find more, that is :-)).
1995-03-17 07:45:19 +00:00
vput(nvp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nvp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
/*
* If either the dircount or maxcount will be
* exceeded, get out now. Both of these lengths
* are calculated conservatively, including all
* XDR overheads.
*/
Correct the sanity test length calculation in nfsrv_readdirplus(): len is being incremented by 4 bytes too few each time through the loop, which allows more data into the mbuf chain that we really want. In the worst case, when we're using 32K read/write sizes with a TCP client, this causes readdirplus replies to sometimes exceed NFS_MAXPACKET which leads to a panic. This problem cropped up for me using an IRIX 6.5.4 NFSv3 TCP client with 32K read/write sizes, however supposedly it can be triggered by WinNT NFS servers too. In theory, it can probably be triggered by any NFS v3 implementation using TCP as long as it's using the maxiumum block size. Reviewed by: Matthew Dillon <dillon@backplane.com>
1999-07-29 21:42:57 +00:00
len += (8 * NFSX_UNSIGNED + nlen + rem + NFSX_V3FH +
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
NFSX_V3POSTOPATTR);
dirlen += (6 * NFSX_UNSIGNED + nlen + rem);
if (len > cnt || dirlen > fullsiz) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
eofflag = 0;
break;
}
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Build the directory record xdr from
* the dirent entry.
*/
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
fp = (struct nfs_fattr *)&fl.fl_fattr;
nfsm_srvfillattr(vap, fp);
fl.fl_fhsize = txdr_unsigned(NFSX_V3FH);
Rename some variables that end up shadowing their namesakes in the NFS client code. Reviewed by: peter
2002-01-08 19:41:06 +00:00
fl.fl_fhok = nfsrv_nfs_true;
fl.fl_postopok = nfsrv_nfs_true;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
fl.fl_off.nfsuquad[0] = 0;
fl.fl_off.nfsuquad[1] = txdr_unsigned(*cookiep);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nfsm_clget;
Rename some variables that end up shadowing their namesakes in the NFS client code. Reviewed by: peter
2002-01-08 19:41:06 +00:00
*tl = nfsrv_nfs_true;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
bp += NFSX_UNSIGNED;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_clget;
*tl = 0;
bp += NFSX_UNSIGNED;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nfsm_clget;
*tl = txdr_unsigned(dp->d_fileno);
bp += NFSX_UNSIGNED;
nfsm_clget;
*tl = txdr_unsigned(nlen);
bp += NFSX_UNSIGNED;
Remove trailing whitespace.
1995-05-30 08:16:23 +00:00
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/* And loop around copying the name */
xfer = nlen;
cp = dp->d_name;
while (xfer > 0) {
nfsm_clget;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if ((bp + xfer) > be)
tsiz = be - bp;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
else
tsiz = xfer;
bcopy(cp, bp, tsiz);
bp += tsiz;
xfer -= tsiz;
if (xfer > 0)
cp += tsiz;
}
For the on-the-wire protocol, u_long -> u_int32_t; long -> int32_t; int -> int32_t; u_short -> u_int16_t. Also, use mode_t instead of u_short for storing modes (mode_t is a u_int16_t). Obtained from: NetBSD
1998-05-31 20:09:01 +00:00
/* And null pad to a int32_t boundary */
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
for (i = 0; i < rem; i++)
*bp++ = '\0';
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
/*
* Now copy the flrep structure out.
*/
xfer = sizeof (struct flrep);
cp = (caddr_t)&fl;
while (xfer > 0) {
nfsm_clget;
if ((bp + xfer) > be)
tsiz = be - bp;
else
tsiz = xfer;
bcopy(cp, bp, tsiz);
bp += tsiz;
xfer -= tsiz;
if (xfer > 0)
cp += tsiz;
}
Make NFS ask the filesystems for directory cookies instead of making them itself.
1994-09-28 16:45:22 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
invalid:
cpos += dp->d_reclen;
dp = (struct dirent *)cpos;
Make NFS ask the filesystems for directory cookies instead of making them itself.
1994-09-28 16:45:22 +00:00
cookiep++;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
ncookies--;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vrele(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nfsm_clget;
Rename some variables that end up shadowing their namesakes in the NFS client code. Reviewed by: peter
2002-01-08 19:41:06 +00:00
*tl = nfsrv_nfs_false;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
bp += NFSX_UNSIGNED;
nfsm_clget;
if (eofflag)
Rename some variables that end up shadowing their namesakes in the NFS client code. Reviewed by: peter
2002-01-08 19:41:06 +00:00
*tl = nfsrv_nfs_true;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
else
Rename some variables that end up shadowing their namesakes in the NFS client code. Reviewed by: peter
2002-01-08 19:41:06 +00:00
*tl = nfsrv_nfs_false;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
bp += NFSX_UNSIGNED;
if (mp != mb) {
if (bp < be)
mp->m_len = bp - mtod(mp, caddr_t);
} else
mp->m_len += bp - bpos;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
FREE((caddr_t)cookies, M_TEMP);
FREE((caddr_t)rbuf, M_TEMP);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsmout:
if (vp)
vrele(vp);
return(error);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
/*
* nfs commit service
*/
int
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrv_commit(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
struct thread *td, struct mbuf **mrq)
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
{
struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md;
Fix all areas of the system (or at least all those in LINT) to avoid storing socket addresses in mbufs. (Socket buffers are the one exception.) A number of kernel APIs needed to get fixed in order to make this happen. Also, fix three protocol families which kept PCBs in mbufs to not malloc them instead. Delete some old compatibility cruft while we're at it, and add some new routines in the in_cksum family.
1997-08-16 19:16:27 +00:00
struct sockaddr *nam = nfsd->nd_nam;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
caddr_t dpos = nfsd->nd_dpos;
struct ucred *cred = &nfsd->nd_cr;
struct vattr bfor, aft;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
struct vnode *vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsfh_t nfh;
fhandle_t *fhp;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
u_int32_t *tl;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
caddr_t bpos;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int error = 0, rdonly, for_ret = 1, aft_ret = 1, cnt;
struct mbuf *mb, *mreq;
u_quad_t off;
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
struct mount *mp = NULL;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int v3 = (nfsd->nd_flag & ND_NFSV3);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsdbprintf(("%s %d\n", __FILE__, __LINE__));
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (!v3)
panic("nfsrv_commit: v3 proc called on a v2 connection");
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
fhp = &nfh.fh_generic;
nfsm_srvmtofh(fhp);
Fix some problems that were introduced in revision 1.97. Instead of returning an error code to the caller, NFS server op routines must themselves build an error reply and return 0 to the caller. This is achieved by replacing the erroneous return statements with code that jumps forward to the op function's reply code. We need to be careful to ensure that the 'struct mount' pointer is NULL though, so that the final vn_finished_write() call becomes a no-op. Reviewed by: mckusick, dillon
2001-02-09 13:24:06 +00:00
if ((mp = vfs_getvfs(&fhp->fh_fsid)) == NULL) {
error = ESTALE;
goto ereply;
}
if ((error = VFS_FHTOVP(mp, &fhp->fh_fid, &vp)) != 0) {
mp = NULL;
goto ereply;
}
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
(void) vn_start_write(vp, &mp, V_WAIT);
vput(vp);
vp = NULL;
Make nfsm_dissect() have an obvious return value.
2001-09-27 22:40:38 +00:00
tl = nfsm_dissect(u_int32_t *, 3 * NFSX_UNSIGNED);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
/*
* XXX At this time VOP_FSYNC() does not accept offset and byte
* count parameters, so these arguments are useless (someday maybe).
*/
Various changes lifted from the OpenBSD cvs tree: txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge. nfs_serv.c: don't call nfsm_adj() with negative values, windows clients could crash servers when doing a readdir of a large directory. nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin loop trying to bind(). Don't clobber a mbuf pointer or we get panics on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf. nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB. Obtained from: OpenBSD
1999-06-05 05:35:03 +00:00
off = fxdr_hyper(tl);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
tl += 2;
cnt = fxdr_unsigned(int, *tl);
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
error = nfsrv_fhtovp(fhp, 1, &vp, cred, slp, nam, &rdonly, TRUE);
NFS Jumbo commit part 1. Cosmetic and structural changes only. The aim of this part of commits is to minimize unnecessary differences between the other NFS's of similar origin. Yes, there are gratuitous changes here that the style folks won't like, but it makes the catch-up less difficult.
1998-05-31 17:27:58 +00:00
if (error) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(2 * NFSX_UNSIGNED);
nfsm_srvwcc_data(for_ret, &bfor, aft_ret, &aft);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
for_ret = VOP_GETATTR(vp, &bfor, cred, td);
Asynchronized client-side nfs_commit. NFS commit operations were previously issued synchronously even if async daemons (nfsiod's) were available. The commit has been moved from the strategy code to the doio code in order to asynchronize it. Removed use of lastr in preparation for removal of vnode->v_lastr. It has been replaced with seqcount, which is already supported by the system and, in fact, gives us a better heuristic for sequential detection then lastr ever did. Made major performance improvements to the server side commit. The server previously fsync'd the entire file for each commit rpc. The server now bawrite()s only those buffers related to the offset/size specified in the commit rpc. Note that we do not commit the meta-data yet. This works still needs to be done. Note that a further optimization can be done (and has not yet been done) on the client: we can merge multiple potential commit rpc's into a single rpc with a greater file offset/size range and greatly reduce rpc traffic. Reviewed by: Alan Cox <alc@cs.rice.edu>, David Greenman <dg@root.com>
1999-09-17 05:57:57 +00:00
if (cnt > MAX_COMMIT_COUNT) {
/*
* Give up and do the whole thing
*/
if (vp->v_object &&
(vp->v_object->flags & OBJ_MIGHTBEDIRTY)) {
vm_object_page_clean() last arg changed from TRUE to OBJPC_SYNC. I'm not sure that this is necessary to be a sync write here since a VOP_FSYNC() follows and it will schedule, sort and complete the writes that the vm_object_page_clean() started (as I think I understand things).
1998-10-31 15:39:31 +00:00
vm_object_page_clean(vp->v_object, 0, 0, OBJPC_SYNC);
Asynchronized client-side nfs_commit. NFS commit operations were previously issued synchronously even if async daemons (nfsiod's) were available. The commit has been moved from the strategy code to the doio code in order to asynchronize it. Removed use of lastr in preparation for removal of vnode->v_lastr. It has been replaced with seqcount, which is already supported by the system and, in fact, gives us a better heuristic for sequential detection then lastr ever did. Made major performance improvements to the server side commit. The server previously fsync'd the entire file for each commit rpc. The server now bawrite()s only those buffers related to the offset/size specified in the commit rpc. Note that we do not commit the meta-data yet. This works still needs to be done. Note that a further optimization can be done (and has not yet been done) on the client: we can merge multiple potential commit rpc's into a single rpc with a greater file offset/size range and greatly reduce rpc traffic. Reviewed by: Alan Cox <alc@cs.rice.edu>, David Greenman <dg@root.com>
1999-09-17 05:57:57 +00:00
}
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
error = VOP_FSYNC(vp, cred, MNT_WAIT, td);
Asynchronized client-side nfs_commit. NFS commit operations were previously issued synchronously even if async daemons (nfsiod's) were available. The commit has been moved from the strategy code to the doio code in order to asynchronize it. Removed use of lastr in preparation for removal of vnode->v_lastr. It has been replaced with seqcount, which is already supported by the system and, in fact, gives us a better heuristic for sequential detection then lastr ever did. Made major performance improvements to the server side commit. The server previously fsync'd the entire file for each commit rpc. The server now bawrite()s only those buffers related to the offset/size specified in the commit rpc. Note that we do not commit the meta-data yet. This works still needs to be done. Note that a further optimization can be done (and has not yet been done) on the client: we can merge multiple potential commit rpc's into a single rpc with a greater file offset/size range and greatly reduce rpc traffic. Reviewed by: Alan Cox <alc@cs.rice.edu>, David Greenman <dg@root.com>
1999-09-17 05:57:57 +00:00
} else {
/*
* Locate and synchronously write any buffers that fall
* into the requested range. Note: we are assuming that
* f_iosize is a power of 2.
*/
int iosize = vp->v_mount->mnt_stat.f_iosize;
int iomask = iosize - 1;
int s;
daddr_t lblkno;
/*
* Align to iosize boundry, super-align to page boundry.
*/
if (off & iomask) {
cnt += off & iomask;
off &= ~(u_quad_t)iomask;
}
if (off & PAGE_MASK) {
cnt += off & PAGE_MASK;
off &= ~(u_quad_t)PAGE_MASK;
}
lblkno = off / iosize;
if (vp->v_object &&
(vp->v_object->flags & OBJ_MIGHTBEDIRTY)) {
vm_object_page_clean(vp->v_object, off / PAGE_SIZE, (cnt + PAGE_MASK) / PAGE_SIZE, OBJPC_SYNC);
}
s = splbio();
while (cnt > 0) {
struct buf *bp;
/*
* If we have a buffer and it is marked B_DELWRI we
* have to lock and write it. Otherwise the prior
* write is assumed to have already been committed.
Replace the global buffer hash table with per-vnode splay trees using a methodology similar to the vm_map_entry splay and the VM splay that Alan Cox is working on. Extensive testing has appeared to have shown no increase in overhead. Disadvantages Dirties more cache lines during lookups. Not as fast as a hash table lookup (but still N log N and optimal when there is locality of reference). Advantages vnode->v_dirtyblkhd is now perfectly sorted, making fsync/sync/filesystem syncer operate more efficiently. I get to rip out all the old hacks (some of which were mine) that tried to keep the v_dirtyblkhd tailq sorted. The per-vnode splay tree should be easier to lock / SMPng pushdown on vnodes will be easier. This commit along with another that Alan is working on for the VM page global hash table will allow me to implement ranged fsync(), optimize server-side nfs commit rpcs, and implement partial syncs by the filesystem syncer (aka filesystem syncer would detect that someone is trying to get the vnode lock, remembers its place, and skip to the next vnode). Note that the buffer cache splay is somewhat more complex then other splays due to special handling of background bitmap writes (multiple buffers with the same lblkno in the same vnode), and B_INVAL discontinuities between the old hash table and the existence of the buffer on the v_cleanblkhd list. Suggested by: alc
2002-07-10 17:02:32 +00:00
*
* gbincore() can return invalid buffers now so we
* have to check that bit as well (though B_DELWRI
* should not be set if B_INVAL is set there could be
* a race here since we haven't locked the buffer).
Asynchronized client-side nfs_commit. NFS commit operations were previously issued synchronously even if async daemons (nfsiod's) were available. The commit has been moved from the strategy code to the doio code in order to asynchronize it. Removed use of lastr in preparation for removal of vnode->v_lastr. It has been replaced with seqcount, which is already supported by the system and, in fact, gives us a better heuristic for sequential detection then lastr ever did. Made major performance improvements to the server side commit. The server previously fsync'd the entire file for each commit rpc. The server now bawrite()s only those buffers related to the offset/size specified in the commit rpc. Note that we do not commit the meta-data yet. This works still needs to be done. Note that a further optimization can be done (and has not yet been done) on the client: we can merge multiple potential commit rpc's into a single rpc with a greater file offset/size range and greatly reduce rpc traffic. Reviewed by: Alan Cox <alc@cs.rice.edu>, David Greenman <dg@root.com>
1999-09-17 05:57:57 +00:00
*/
- Use incore() instead of gbincore() so we don't have to acquire the vnode interlock.
2002-09-25 02:39:39 +00:00
if ((bp = incore(vp, lblkno)) != NULL &&
Replace the global buffer hash table with per-vnode splay trees using a methodology similar to the vm_map_entry splay and the VM splay that Alan Cox is working on. Extensive testing has appeared to have shown no increase in overhead. Disadvantages Dirties more cache lines during lookups. Not as fast as a hash table lookup (but still N log N and optimal when there is locality of reference). Advantages vnode->v_dirtyblkhd is now perfectly sorted, making fsync/sync/filesystem syncer operate more efficiently. I get to rip out all the old hacks (some of which were mine) that tried to keep the v_dirtyblkhd tailq sorted. The per-vnode splay tree should be easier to lock / SMPng pushdown on vnodes will be easier. This commit along with another that Alan is working on for the VM page global hash table will allow me to implement ranged fsync(), optimize server-side nfs commit rpcs, and implement partial syncs by the filesystem syncer (aka filesystem syncer would detect that someone is trying to get the vnode lock, remembers its place, and skip to the next vnode). Note that the buffer cache splay is somewhat more complex then other splays due to special handling of background bitmap writes (multiple buffers with the same lblkno in the same vnode), and B_INVAL discontinuities between the old hash table and the existence of the buffer on the v_cleanblkhd list. Suggested by: alc
2002-07-10 17:02:32 +00:00
(bp->b_flags & (B_DELWRI|B_INVAL)) == B_DELWRI) {
Asynchronized client-side nfs_commit. NFS commit operations were previously issued synchronously even if async daemons (nfsiod's) were available. The commit has been moved from the strategy code to the doio code in order to asynchronize it. Removed use of lastr in preparation for removal of vnode->v_lastr. It has been replaced with seqcount, which is already supported by the system and, in fact, gives us a better heuristic for sequential detection then lastr ever did. Made major performance improvements to the server side commit. The server previously fsync'd the entire file for each commit rpc. The server now bawrite()s only those buffers related to the offset/size specified in the commit rpc. Note that we do not commit the meta-data yet. This works still needs to be done. Note that a further optimization can be done (and has not yet been done) on the client: we can merge multiple potential commit rpc's into a single rpc with a greater file offset/size range and greatly reduce rpc traffic. Reviewed by: Alan Cox <alc@cs.rice.edu>, David Greenman <dg@root.com>
1999-09-17 05:57:57 +00:00
if (BUF_LOCK(bp, LK_EXCLUSIVE | LK_NOWAIT)) {
BUF_LOCK(bp, LK_EXCLUSIVE | LK_SLEEPFAIL);
continue; /* retry */
}
bremfree(bp);
bp->b_flags &= ~B_ASYNC;
Rename the existing BUF_STRATEGY() to DEV_STRATEGY() substitute BUF_WRITE(foo) for VOP_BWRITE(foo->b_vp, foo) substitute BUF_STRATEGY(foo) for VOP_STRATEGY(foo->b_vp, foo) This patch is machine generated except for the ccd.c and buf.h parts.
2000-03-20 11:29:10 +00:00
BUF_WRITE(bp);
Asynchronized client-side nfs_commit. NFS commit operations were previously issued synchronously even if async daemons (nfsiod's) were available. The commit has been moved from the strategy code to the doio code in order to asynchronize it. Removed use of lastr in preparation for removal of vnode->v_lastr. It has been replaced with seqcount, which is already supported by the system and, in fact, gives us a better heuristic for sequential detection then lastr ever did. Made major performance improvements to the server side commit. The server previously fsync'd the entire file for each commit rpc. The server now bawrite()s only those buffers related to the offset/size specified in the commit rpc. Note that we do not commit the meta-data yet. This works still needs to be done. Note that a further optimization can be done (and has not yet been done) on the client: we can merge multiple potential commit rpc's into a single rpc with a greater file offset/size range and greatly reduce rpc traffic. Reviewed by: Alan Cox <alc@cs.rice.edu>, David Greenman <dg@root.com>
1999-09-17 05:57:57 +00:00
++nfs_commit_miss;
}
++nfs_commit_blks;
if (cnt < iosize)
break;
cnt -= iosize;
++lblkno;
}
splx(s);
Change the busy page mgmt, so that when pages are freed, they MUST be PG_BUSY. It is bogus to free a page that isn't busy, because it is in a state of being "unavailable" when being freed. The additional advantage is that the page_remove code has a better cross-check that the page should be busy and unavailable for other use. There were some minor problems with the collapse code, and this plugs those subtile "holes." Also, the vfs_bio code wasn't checking correctly for PG_BUSY pages. I am going to develop a more consistant scheme for grabbing pages, busy or otherwise. For now, we are stuck with the current morass.
1998-01-31 11:56:53 +00:00
}
Asynchronized client-side nfs_commit. NFS commit operations were previously issued synchronously even if async daemons (nfsiod's) were available. The commit has been moved from the strategy code to the doio code in order to asynchronize it. Removed use of lastr in preparation for removal of vnode->v_lastr. It has been replaced with seqcount, which is already supported by the system and, in fact, gives us a better heuristic for sequential detection then lastr ever did. Made major performance improvements to the server side commit. The server previously fsync'd the entire file for each commit rpc. The server now bawrite()s only those buffers related to the offset/size specified in the commit rpc. Note that we do not commit the meta-data yet. This works still needs to be done. Note that a further optimization can be done (and has not yet been done) on the client: we can merge multiple potential commit rpc's into a single rpc with a greater file offset/size range and greatly reduce rpc traffic. Reviewed by: Alan Cox <alc@cs.rice.edu>, David Greenman <dg@root.com>
1999-09-17 05:57:57 +00:00
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
aft_ret = VOP_GETATTR(vp, &aft, cred, td);
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vput(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
Fix some problems that were introduced in revision 1.97. Instead of returning an error code to the caller, NFS server op routines must themselves build an error reply and return 0 to the caller. This is achieved by replacing the erroneous return statements with code that jumps forward to the op function's reply code. We need to be careful to ensure that the 'struct mount' pointer is NULL though, so that the final vn_finished_write() call becomes a no-op. Reviewed by: mckusick, dillon
2001-02-09 13:24:06 +00:00
ereply:
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_V3WCCDATA + NFSX_V3WRITEVERF);
nfsm_srvwcc_data(for_ret, &bfor, aft_ret, &aft);
if (!error) {
Tidy up nfsm_build usage. This is only partially finished.
2001-09-27 02:33:36 +00:00
tl = nfsm_build(u_int32_t *, NFSX_V3WRITEVERF);
Have NFS use a snapshot of boottime instead of boottime itself to generate the NFSv3 Version id. boottime itself may change, sometimes once every tick if you are running xntpd, which really throws off clients. Clients will tend to throw away what they believe to be stale data too often, and can get into long loops rewriting the same data over and over again because they believe the server has rebooted over and over again due to the changing version id. Approved by: jkh
1999-12-16 17:01:32 +00:00
if (nfsver.tv_sec == 0)
nfsver = boottime;
*tl++ = txdr_unsigned(nfsver.tv_sec);
*tl = txdr_unsigned(nfsver.tv_usec);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
} else {
error = 0;
}
nfsmout:
if (vp)
vput(vp);
Add snapshots to the fast filesystem. Most of the changes support the gating of system calls that cause modifications to the underlying filesystem. The gating can be enabled by any filesystem that needs to consistently suspend operations by adding the vop_stdgetwritemount to their set of vnops. Once gating is enabled, the function vfs_write_suspend stops all new write operations to a filesystem, allows any filesystem modifying system calls already in progress to complete, then sync's the filesystem to disk and returns. The function vfs_write_resume allows the suspended write operations to begin again. Gating is not added by default for all filesystems as for SMP systems it adds two extra locks to such critical kernel paths as the write system call. Thus, gating should only be added as needed. Details on the use and current status of snapshots in FFS can be found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness is not included here. Unless and until you create a snapshot file, these changes should have no effect on your system (famous last words).
2000-07-11 22:07:57 +00:00
vn_finished_write(mp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
return(error);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
/*
* nfs statfs service
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrv_statfs(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
struct thread *td, struct mbuf **mrq)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md;
Fix all areas of the system (or at least all those in LINT) to avoid storing socket addresses in mbufs. (Socket buffers are the one exception.) A number of kernel APIs needed to get fixed in order to make this happen. Also, fix three protocol families which kept PCBs in mbufs to not malloc them instead. Delete some old compatibility cruft while we're at it, and add some new routines in the in_cksum family.
1997-08-16 19:16:27 +00:00
struct sockaddr *nam = nfsd->nd_nam;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
caddr_t dpos = nfsd->nd_dpos;
struct ucred *cred = &nfsd->nd_cr;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct statfs *sf;
struct nfs_statfs *sfp;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
caddr_t bpos;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int error = 0, rdonly, getret = 1;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
int v3 = (nfsd->nd_flag & ND_NFSV3);
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct mbuf *mb, *mreq;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
struct vnode *vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct vattr at;
nfsfh_t nfh;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhandle_t *fhp;
struct statfs statfs;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
u_quad_t tval;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsdbprintf(("%s %d\n", __FILE__, __LINE__));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fhp = &nfh.fh_generic;
nfsm_srvmtofh(fhp);
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
error = nfsrv_fhtovp(fhp, 1, &vp, cred, slp, nam, &rdonly, TRUE);
NFS Jumbo commit part 1. Cosmetic and structural changes only. The aim of this part of commits is to minimize unnecessary differences between the other NFS's of similar origin. Yes, there are gratuitous changes here that the style folks won't like, but it makes the catch-up less difficult.
1998-05-31 17:27:58 +00:00
if (error) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_UNSIGNED);
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (v3)
nfsm_srvpostop_attr(getret, &at);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
sf = &statfs;
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
error = VFS_STATFS(vp->v_mount, sf, td);
getret = VOP_GETATTR(vp, &at, cred, td);
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vput(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_POSTOPATTR(v3) + NFSX_STATFS(v3));
if (v3)
nfsm_srvpostop_attr(getret, &at);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (error) {
error = 0;
goto nfsmout;
}
Tidy up nfsm_build usage. This is only partially finished.
2001-09-27 02:33:36 +00:00
sfp = nfsm_build(struct nfs_statfs *, NFSX_STATFS(v3));
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
if (v3) {
tval = (u_quad_t)sf->f_blocks;
tval *= (u_quad_t)sf->f_bsize;
Various changes lifted from the OpenBSD cvs tree: txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge. nfs_serv.c: don't call nfsm_adj() with negative values, windows clients could crash servers when doing a readdir of a large directory. nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin loop trying to bind(). Don't clobber a mbuf pointer or we get panics on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf. nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB. Obtained from: OpenBSD
1999-06-05 05:35:03 +00:00
txdr_hyper(tval, &sfp->sf_tbytes);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
tval = (u_quad_t)sf->f_bfree;
tval *= (u_quad_t)sf->f_bsize;
Various changes lifted from the OpenBSD cvs tree: txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge. nfs_serv.c: don't call nfsm_adj() with negative values, windows clients could crash servers when doing a readdir of a large directory. nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin loop trying to bind(). Don't clobber a mbuf pointer or we get panics on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf. nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB. Obtained from: OpenBSD
1999-06-05 05:35:03 +00:00
txdr_hyper(tval, &sfp->sf_fbytes);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
tval = (u_quad_t)sf->f_bavail;
tval *= (u_quad_t)sf->f_bsize;
Various changes lifted from the OpenBSD cvs tree: txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge. nfs_serv.c: don't call nfsm_adj() with negative values, windows clients could crash servers when doing a readdir of a large directory. nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin loop trying to bind(). Don't clobber a mbuf pointer or we get panics on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf. nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB. Obtained from: OpenBSD
1999-06-05 05:35:03 +00:00
txdr_hyper(tval, &sfp->sf_abytes);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
sfp->sf_tfiles.nfsuquad[0] = 0;
sfp->sf_tfiles.nfsuquad[1] = txdr_unsigned(sf->f_files);
sfp->sf_ffiles.nfsuquad[0] = 0;
sfp->sf_ffiles.nfsuquad[1] = txdr_unsigned(sf->f_ffree);
sfp->sf_afiles.nfsuquad[0] = 0;
sfp->sf_afiles.nfsuquad[1] = txdr_unsigned(sf->f_ffree);
sfp->sf_invarsec = 0;
} else {
sfp->sf_tsize = txdr_unsigned(NFS_MAXDGRAMDATA);
sfp->sf_bsize = txdr_unsigned(sf->f_bsize);
sfp->sf_blocks = txdr_unsigned(sf->f_blocks);
sfp->sf_bfree = txdr_unsigned(sf->f_bfree);
sfp->sf_bavail = txdr_unsigned(sf->f_bavail);
}
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsmout:
if (vp)
vput(vp);
return(error);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
/*
* nfs fsinfo service
*/
int
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrv_fsinfo(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
struct thread *td, struct mbuf **mrq)
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
{
struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md;
Fix all areas of the system (or at least all those in LINT) to avoid storing socket addresses in mbufs. (Socket buffers are the one exception.) A number of kernel APIs needed to get fixed in order to make this happen. Also, fix three protocol families which kept PCBs in mbufs to not malloc them instead. Delete some old compatibility cruft while we're at it, and add some new routines in the in_cksum family.
1997-08-16 19:16:27 +00:00
struct sockaddr *nam = nfsd->nd_nam;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
caddr_t dpos = nfsd->nd_dpos;
struct ucred *cred = &nfsd->nd_cr;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct nfsv3_fsinfo *sip;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
caddr_t bpos;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int error = 0, rdonly, getret = 1, pref;
struct mbuf *mb, *mreq;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
struct vnode *vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct vattr at;
nfsfh_t nfh;
fhandle_t *fhp;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
u_quad_t maxfsize;
When using NFSv3, use the remote server's idea of the maximum file size rather than assuming 2^64. It may not like files that big. :-) On the nfs server, calculate and report the max file size as the point that the block numbers in the cache would turn negative. (ie: 1099511627775 bytes (1TB)). One of the things I'm worried about however, is that directory offsets are really cookies on a NFSv3 server and can be rather large, especially when/if the server generates the opaque directory cookies by using a local filesystem offset in what comes out as the upper 32 bits of the 64 bit cookie. (a server is free to do this, it could save byte swapping depending on the native 64 bit byte order) Obtained from: NetBSD
1998-05-30 16:33:58 +00:00
struct statfs sb;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int v3 = (nfsd->nd_flag & ND_NFSV3);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsdbprintf(("%s %d\n", __FILE__, __LINE__));
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (!v3)
panic("nfsrv_fsinfo: v3 proc called on a v2 connection");
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
fhp = &nfh.fh_generic;
nfsm_srvmtofh(fhp);
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
error = nfsrv_fhtovp(fhp, 1, &vp, cred, slp, nam, &rdonly, TRUE);
When using NFSv3, use the remote server's idea of the maximum file size rather than assuming 2^64. It may not like files that big. :-) On the nfs server, calculate and report the max file size as the point that the block numbers in the cache would turn negative. (ie: 1099511627775 bytes (1TB)). One of the things I'm worried about however, is that directory offsets are really cookies on a NFSv3 server and can be rather large, especially when/if the server generates the opaque directory cookies by using a local filesystem offset in what comes out as the upper 32 bits of the 64 bit cookie. (a server is free to do this, it could save byte swapping depending on the native 64 bit byte order) Obtained from: NetBSD
1998-05-30 16:33:58 +00:00
if (error) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_UNSIGNED);
nfsm_srvpostop_attr(getret, &at);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
When using NFSv3, use the remote server's idea of the maximum file size rather than assuming 2^64. It may not like files that big. :-) On the nfs server, calculate and report the max file size as the point that the block numbers in the cache would turn negative. (ie: 1099511627775 bytes (1TB)). One of the things I'm worried about however, is that directory offsets are really cookies on a NFSv3 server and can be rather large, especially when/if the server generates the opaque directory cookies by using a local filesystem offset in what comes out as the upper 32 bits of the 64 bit cookie. (a server is free to do this, it could save byte swapping depending on the native 64 bit byte order) Obtained from: NetBSD
1998-05-30 16:33:58 +00:00
/* XXX Try to make a guess on the max file size. */
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
VFS_STATFS(vp->v_mount, &sb, td);
When using NFSv3, use the remote server's idea of the maximum file size rather than assuming 2^64. It may not like files that big. :-) On the nfs server, calculate and report the max file size as the point that the block numbers in the cache would turn negative. (ie: 1099511627775 bytes (1TB)). One of the things I'm worried about however, is that directory offsets are really cookies on a NFSv3 server and can be rather large, especially when/if the server generates the opaque directory cookies by using a local filesystem offset in what comes out as the upper 32 bits of the 64 bit cookie. (a server is free to do this, it could save byte swapping depending on the native 64 bit byte order) Obtained from: NetBSD
1998-05-30 16:33:58 +00:00
maxfsize = (u_quad_t)0x80000000 * sb.f_bsize - 1;
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
getret = VOP_GETATTR(vp, &at, cred, td);
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vput(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_V3POSTOPATTR + NFSX_V3FSINFO);
nfsm_srvpostop_attr(getret, &at);
Tidy up nfsm_build usage. This is only partially finished.
2001-09-27 02:33:36 +00:00
sip = nfsm_build(struct nfsv3_fsinfo *, NFSX_V3FSINFO);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
/*
* XXX
More s/file system/filesystem/g
2002-05-16 21:28:32 +00:00
* There should be filesystem VFS OP(s) to get this information.
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
* For now, assume ufs.
*/
if (slp->ns_so->so_type == SOCK_DGRAM)
pref = NFS_MAXDGRAMDATA;
else
pref = NFS_MAXDATA;
In the NFSv3 `fsinfo' procedure reply, don't claim that we support 32k read and write operations on datagram sockets when in fact we reject requests larger than 16k. It must be the case that virtually all clients use data sizes of 16k or less for UDP transport (FreeBSD's client defaults to 8k and never exceeds 16k), as this bug has been present ever since NFSv3 support was added. Reported by: Senthil <lihtnes78@netscape.net> Reviewed by: dillon Approved by: re MFC-after: 1 week
2002-12-05 16:58:11 +00:00
sip->fs_rtmax = txdr_unsigned(pref);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
sip->fs_rtpref = txdr_unsigned(pref);
sip->fs_rtmult = txdr_unsigned(NFS_FABLKSIZE);
In the NFSv3 `fsinfo' procedure reply, don't claim that we support 32k read and write operations on datagram sockets when in fact we reject requests larger than 16k. It must be the case that virtually all clients use data sizes of 16k or less for UDP transport (FreeBSD's client defaults to 8k and never exceeds 16k), as this bug has been present ever since NFSv3 support was added. Reported by: Senthil <lihtnes78@netscape.net> Reviewed by: dillon Approved by: re MFC-after: 1 week
2002-12-05 16:58:11 +00:00
sip->fs_wtmax = txdr_unsigned(pref);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
sip->fs_wtpref = txdr_unsigned(pref);
sip->fs_wtmult = txdr_unsigned(NFS_FABLKSIZE);
sip->fs_dtpref = txdr_unsigned(pref);
Various changes lifted from the OpenBSD cvs tree: txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge. nfs_serv.c: don't call nfsm_adj() with negative values, windows clients could crash servers when doing a readdir of a large directory. nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin loop trying to bind(). Don't clobber a mbuf pointer or we get panics on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf. nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB. Obtained from: OpenBSD
1999-06-05 05:35:03 +00:00
txdr_hyper(maxfsize, &sip->fs_maxfilesize);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
sip->fs_timedelta.nfsv3_sec = 0;
sip->fs_timedelta.nfsv3_nsec = txdr_unsigned(1);
sip->fs_properties = txdr_unsigned(NFSV3FSINFO_LINK |
NFSV3FSINFO_SYMLINK | NFSV3FSINFO_HOMOGENEOUS |
NFSV3FSINFO_CANSETTIME);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsmout:
if (vp)
vput(vp);
return(error);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
}
/*
* nfs pathconf service
*/
int
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrv_pathconf(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
struct thread *td, struct mbuf **mrq)
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
{
struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md;
Fix all areas of the system (or at least all those in LINT) to avoid storing socket addresses in mbufs. (Socket buffers are the one exception.) A number of kernel APIs needed to get fixed in order to make this happen. Also, fix three protocol families which kept PCBs in mbufs to not malloc them instead. Delete some old compatibility cruft while we're at it, and add some new routines in the in_cksum family.
1997-08-16 19:16:27 +00:00
struct sockaddr *nam = nfsd->nd_nam;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
caddr_t dpos = nfsd->nd_dpos;
struct ucred *cred = &nfsd->nd_cr;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct nfsv3_pathconf *pc;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
caddr_t bpos;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int error = 0, rdonly, getret = 1;
This commit fixes various 64bit portability problems required for FreeBSD/alpha. The most significant item is to change the command argument to ioctl functions from int to u_long. This change brings us inline with various other BSD versions. Driver writers may like to use (__FreeBSD_version == 300003) to detect this change. The prototype FreeBSD/alpha machdep will follow in a couple of days time.
1998-06-07 17:13:14 +00:00
register_t linkmax, namemax, chownres, notrunc;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
struct mbuf *mb, *mreq;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
struct vnode *vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
struct vattr at;
nfsfh_t nfh;
fhandle_t *fhp;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int v3 = (nfsd->nd_flag & ND_NFSV3);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsdbprintf(("%s %d\n", __FILE__, __LINE__));
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (!v3)
panic("nfsrv_pathconf: v3 proc called on a v2 connection");
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
fhp = &nfh.fh_generic;
nfsm_srvmtofh(fhp);
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
error = nfsrv_fhtovp(fhp, 1, &vp, cred, slp, nam, &rdonly, TRUE);
NFS Jumbo commit part 1. Cosmetic and structural changes only. The aim of this part of commits is to minimize unnecessary differences between the other NFS's of similar origin. Yes, there are gratuitous changes here that the style folks won't like, but it makes the catch-up less difficult.
1998-05-31 17:27:58 +00:00
if (error) {
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_UNSIGNED);
nfsm_srvpostop_attr(getret, &at);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
error = 0;
goto nfsmout;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
error = VOP_PATHCONF(vp, _PC_LINK_MAX, &linkmax);
if (!error)
error = VOP_PATHCONF(vp, _PC_NAME_MAX, &namemax);
if (!error)
error = VOP_PATHCONF(vp, _PC_CHOWN_RESTRICTED, &chownres);
if (!error)
error = VOP_PATHCONF(vp, _PC_NO_TRUNC, &notrunc);
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
getret = VOP_GETATTR(vp, &at, cred, td);
Even though this looks like it, this is not a complex code change. The interface into the "VMIO" system has changed to be more consistant and robust. Essentially, it is now no longer necessary to call vn_open to get merged VM/Buffer cache operation, and exceptional conditions such as merged operation of VBLK devices is simpler and more correct. This code corrects a potentially large set of problems including the problems with ktrace output and loaded systems, file create/deletes, etc. Most of the changes to NFS are cosmetic and name changes, eliminating a layer of subroutine calls. The direct calls to vput/vrele have been re-instituted for better cross platform compatibility. Reviewed by: davidg
1996-08-21 21:56:23 +00:00
vput(vp);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vp = NULL;
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
nfsm_reply(NFSX_V3POSTOPATTR + NFSX_V3PATHCONF);
nfsm_srvpostop_attr(getret, &at);
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
if (error) {
error = 0;
goto nfsmout;
}
Tidy up nfsm_build usage. This is only partially finished.
2001-09-27 02:33:36 +00:00
pc = nfsm_build(struct nfsv3_pathconf *, NFSX_V3PATHCONF);
Changes to support version 3 of the NFS protocol. The version 2 support has been tested (client+server) against FreeBSD-2.0, IRIX 5.3 and FreeBSD-current (using a loopback mount). The version 2 support is stable AFAIK. The version 3 support has been tested with a loopback mount and minimally against an IRIX 5.3 server. It needs more testing and may have problems. I have patched amd to support the new variable length filehandles although it will still only use version 2 of the protocol. Before booting a kernel with these changes, nfs clients will need to at least build and install /usr/sbin/mount_nfs. Servers will need to build and install /usr/sbin/mountd. NFS diskless support is untested. Obtained from: Rick Macklem <rick@snowhite.cis.uoguelph.ca>
1995-06-27 11:07:30 +00:00
pc->pc_linkmax = txdr_unsigned(linkmax);
pc->pc_namemax = txdr_unsigned(namemax);
pc->pc_notrunc = txdr_unsigned(notrunc);
pc->pc_chownrestricted = txdr_unsigned(chownres);
/*
* These should probably be supported by VOP_PATHCONF(), but
* until msdosfs is exportable (why would you want to?), the
* Unix defaults should be ok.
*/
Rename some variables that end up shadowing their namesakes in the NFS client code. Reviewed by: peter
2002-01-08 19:41:06 +00:00
pc->pc_caseinsensitive = nfsrv_nfs_false;
pc->pc_casepreserving = nfsrv_nfs_true;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsmout:
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
if (vp)
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
vput(vp);
return(error);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
/*
* Null operation, used by clients to ping server
*/
/* ARGSUSED */
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrv_null(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
struct thread *td, struct mbuf **mrq)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Second batch of cleanup changes. This time mostly making a lot of things static and some unused variables here and there.
1995-10-29 15:33:36 +00:00
struct mbuf *mrep = nfsd->nd_mrep;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
caddr_t bpos;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int error = NFSERR_RETVOID;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct mbuf *mb, *mreq;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsdbprintf(("%s %d\n", __FILE__, __LINE__));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nfsm_reply(0);
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsmout:
return (error);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
/*
* No operation, used for obsolete procedures
*/
/* ARGSUSED */
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrv_noop(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp,
struct thread *td, struct mbuf **mrq)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Second batch of cleanup changes. This time mostly making a lot of things static and some unused variables here and there.
1995-10-29 15:33:36 +00:00
struct mbuf *mrep = nfsd->nd_mrep;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
caddr_t bpos;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
int error;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct mbuf *mb, *mreq;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsdbprintf(("%s %d\n", __FILE__, __LINE__));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (nfsd->nd_repstat)
error = nfsd->nd_repstat;
else
error = EPROCUNAVAIL;
nfsm_reply(0);
Now that nfsm_reply() does not usually set 'error' to 0, we need to do it explicitly in nfsrv_noop so that the reply gets sent back to the client. This fixes the generation of a selection of RPC error replies (RPC_PROGMISMATCH, RPC_PROGUNAVAIL, RPC_PROCUNAVAIL etc.) that are used by some clients to detect support for optional protocols and features. Reviewed by: peter Reported by: Thomas Quinot <quinot@inf.enst.fr> PR: kern/31479
2001-10-25 19:07:56 +00:00
error = 0;
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsmout:
return (error);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
/*
* Perform access checking for vnodes obtained from file handles that would
* refer to files already opened by a Unix client. You cannot just use
* vn_writechk() and VOP_ACCESS() for two reasons.
* 1 - You must check for exported rdonly as well as MNT_RDONLY for the write case
Only ignore "owner" permissions selectively rather than always. In some cases we ignore it (eg: read/write) to maintain chmod-after-open semantics but in other cases we do care, eg: creating files, access() etc. Never ignore errors from VOP_ACCESS() on immutable files. This apparently comes from BSDI (from Keith Bostic) via NetBSD. PR: 5148 Submitted by: Yoshiro MIHIRA <sanpei@yy.cs.keio.ac.jp>
1998-05-20 09:05:48 +00:00
* 2 - The owner is to be given access irrespective of mode bits for some
* operations, so that processes that chmod after opening a file don't
* break. I don't like this because it opens a security hole, but since
* the nfs server opens a security hole the size of a barn door anyhow,
* what the heck.
*
* The exception to rule 2 is EPERM. If a file is IMMUTABLE, VOP_ACCESS()
* will return EPERM instead of EACCESS. EPERM is always an error.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
Staticize.
1995-12-17 21:14:36 +00:00
static int
Cleanup and split of nfs client and server code. This builds on the top of several repo-copies.
2001-09-18 23:32:09 +00:00
nfsrv_access(struct vnode *vp, int flags, struct ucred *cred, int rdonly,
struct thread *td, int override)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
struct vattr vattr;
int error;
Matt's NFS fixes. Submitted by: Matt Dillon Reviewed by: David Cross, Julian Elischer, Mike Smith, Drew Gallatin 3.2 version to follow when tested
1999-06-23 04:44:14 +00:00
nfsdbprintf(("%s %d\n", __FILE__, __LINE__));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (flags & VWRITE) {
/* Just vn_writechk() changed to check rdonly */
/*
More s/file system/filesystem/g
2002-05-16 21:28:32 +00:00
* Disallow write attempts on read-only filesystems;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
* unless the file is a socket or a block or character
More s/file system/filesystem/g
2002-05-16 21:28:32 +00:00
* device resident on the filesystem.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
if (rdonly || (vp->v_mount->mnt_flag & MNT_RDONLY)) {
switch (vp->v_type) {
NFS Jumbo commit part 1. Cosmetic and structural changes only. The aim of this part of commits is to minimize unnecessary differences between the other NFS's of similar origin. Yes, there are gratuitous changes here that the style folks won't like, but it makes the catch-up less difficult.
1998-05-31 17:27:58 +00:00
case VREG:
case VDIR:
case VLNK:
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return (EROFS);
NFS Jumbo commit part 1. Cosmetic and structural changes only. The aim of this part of commits is to minimize unnecessary differences between the other NFS's of similar origin. Yes, there are gratuitous changes here that the style folks won't like, but it makes the catch-up less difficult.
1998-05-31 17:27:58 +00:00
default:
break;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
}
/*
* If there's shared text associated with
Removed unnecessary call to vnode_pager_uncache(). We automatically clear the VTEXT flag after all mappers have finished with the object.
1995-03-19 12:08:03 +00:00
* the inode, we can't allow writing.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
- Replace v_flag with v_iflag and v_vflag - v_vflag is protected by the vnode lock and is used when synchronization with VOP calls is needed. - v_iflag is protected by interlock and is used for dealing with vnode management issues. These flags include X/O LOCK, FREE, DOOMED, etc. - All accesses to v_iflag and v_vflag have either been locked or marked with mp_fixme's. - Many ASSERT_VOP_LOCKED calls have been added where the locking was not clear. - Many functions in vfs_subr.c were restructured to provide for stronger locking. Idea stolen from: BSD/OS
2002-08-04 10:29:36 +00:00
if (vp->v_vflag & VV_TEXT)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return (ETXTBSY);
}
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
error = VOP_GETATTR(vp, &vattr, cred, td);
NFS Jumbo commit part 1. Cosmetic and structural changes only. The aim of this part of commits is to minimize unnecessary differences between the other NFS's of similar origin. Yes, there are gratuitous changes here that the style folks won't like, but it makes the catch-up less difficult.
1998-05-31 17:27:58 +00:00
if (error)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return (error);
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
error = VOP_ACCESS(vp, flags, cred, td);
Only ignore "owner" permissions selectively rather than always. In some cases we ignore it (eg: read/write) to maintain chmod-after-open semantics but in other cases we do care, eg: creating files, access() etc. Never ignore errors from VOP_ACCESS() on immutable files. This apparently comes from BSDI (from Keith Bostic) via NetBSD. PR: 5148 Submitted by: Yoshiro MIHIRA <sanpei@yy.cs.keio.ac.jp>
1998-05-20 09:05:48 +00:00
/*
* Allow certain operations for the owner (reads and writes
* on files that are already open).
*/
if (override && error == EACCES && cred->cr_uid == vattr.va_uid)
error = 0;
return error;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Reference in New Issue Copy Permalink