d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
e2835957c4
freebsd-dev/sys/fs/nfs/nfs_commonsubs.c

3863 lines
99 KiB
C
Raw Normal View History

Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +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.
* 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.
*
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
/*
* These functions support the macros and help fiddle mbuf chains for
* the nfs op functions. They do things like create the rpc header and
* copy data between mbuf chains and uio lists.
*/
#ifndef APPLEKEXT
opt_inet6.h was missing from some files in the new NFS subsystem. The effect of this was, for clients mounted via inet6 addresses, that the DRC cache would never have a hit in the server. It also broke NFSv4 callbacks when an inet6 address was the only one available in the client. This patch fixes the above, plus deletes opt_inet6.h from a couple of files it is not needed for. MFC after: 2 weeks
2012-01-08 01:54:46 +00:00
#include "opt_inet6.h"
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
#include <fs/nfs/nfsport.h>
/*
* Data items converted to xdr at startup, since they are constant
* This is kinda hokey, but may save a little time doing byte swaps
*/
u_int32_t newnfs_true, newnfs_false, newnfs_xdrneg1;
/* And other global data */
nfstype nfsv34_type[9] = { NFNON, NFREG, NFDIR, NFBLK, NFCHR, NFLNK, NFSOCK,
NFFIFO, NFNON };
enum vtype newnv2tov_type[8] = { VNON, VREG, VDIR, VBLK, VCHR, VLNK, VNON, VNON };
enum vtype nv34tov_type[8]={ VNON, VREG, VDIR, VBLK, VCHR, VLNK, VSOCK, VFIFO };
struct timeval nfsboottime; /* Copy boottime once, so it never changes */
int nfscl_ticks;
int nfsrv_useacl = 1;
struct nfssockreq nfsrv_nfsuserdsock;
int nfsrv_nfsuserd = 0;
struct nfsreqhead nfsd_reqq;
uid_t nfsrv_defaultuid;
gid_t nfsrv_defaultgid;
int nfsrv_lease = NFSRV_LEASE;
int ncl_mbuf_mlen = MLEN;
The new draft specification for NFSv4.0 specifies that a server should either accept owner and owner_group strings that are just the digits of the uid/gid or return NFS4ERR_BADOWNER. This patch adds a sysctl vfs.nfsd.enable_stringtouid, which can be set to enable the server w.r.t. accepting numeric string. It also ensures that NFS4ERR_BADOWNER is returned if numeric uid/gid strings are not enabled. This fixes the server for recent Linux nfs4 clients that use numeric uid/gid strings by default. Reported and tested by: craigyk@gmail.com MFC after: 2 weeks
2014-05-03 00:13:45 +00:00
int nfsd_enable_stringtouid = 0;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
NFSNAMEIDMUTEX;
NFSSOCKMUTEX;
/*
* This array of structures indicates, for V4:
* retfh - which of 3 types of calling args are used
* 0 - doesn't change cfh or use a sfh
* 1 - replaces cfh with a new one (unless it returns an error status)
* 2 - uses cfh and sfh
* needscfh - if the op wants a cfh and premtime
* 0 - doesn't use a cfh
* 1 - uses a cfh, but doesn't want pre-op attributes
* 2 - uses a cfh and wants pre-op attributes
* savereply - indicates a non-idempotent Op
* 0 - not non-idempotent
* 1 - non-idempotent
* Ops that are ordered via seqid# are handled separately from these
* non-idempotent Ops.
* Define it here, since it is used by both the client and server.
*/
Move the NFSv4.1 client patches over from projects/nfsv4.1-client to head. I don't think the NFS client behaviour will change unless the new "minorversion=1" mount option is used. It includes basic NFSv4.1 support plus support for pNFS using the Files Layout only. All problems detecting during an NFSv4.1 Bakeathon testing event in June 2012 have been resolved in this code and it has been tested against the NFSv4.1 server available to me. Although not reviewed, I believe that kib@ has looked at it.
2012-12-08 22:52:39 +00:00
struct nfsv4_opflag nfsv4_opflag[NFSV41_NOPS] = {
{ 0, 0, 0, 0, LK_EXCLUSIVE, 1 }, /* undef */
{ 0, 0, 0, 0, LK_EXCLUSIVE, 1 }, /* undef */
{ 0, 0, 0, 0, LK_EXCLUSIVE, 1 }, /* undef */
{ 0, 1, 0, 0, LK_SHARED, 1 }, /* Access */
{ 0, 1, 0, 0, LK_EXCLUSIVE, 1 }, /* Close */
{ 0, 2, 0, 1, LK_EXCLUSIVE, 1 }, /* Commit */
{ 1, 2, 1, 1, LK_EXCLUSIVE, 1 }, /* Create */
{ 0, 0, 0, 0, LK_EXCLUSIVE, 1 }, /* Delegpurge */
{ 0, 1, 0, 0, LK_EXCLUSIVE, 1 }, /* Delegreturn */
{ 0, 1, 0, 0, LK_SHARED, 1 }, /* Getattr */
{ 0, 1, 0, 0, LK_EXCLUSIVE, 1 }, /* GetFH */
{ 2, 1, 1, 1, LK_EXCLUSIVE, 1 }, /* Link */
{ 0, 1, 0, 0, LK_EXCLUSIVE, 1 }, /* Lock */
{ 0, 1, 0, 0, LK_EXCLUSIVE, 1 }, /* LockT */
{ 0, 1, 0, 0, LK_EXCLUSIVE, 1 }, /* LockU */
Modify the Lookup RPC for NFSv4 so that it acquires directory attributes. This allows the client to cache directory names when they are looked up, reducing the Lookup RPC count by about 40% for software builds. MFC after: 2 weeks
2014-04-18 22:05:34 +00:00
{ 1, 2, 0, 0, LK_EXCLUSIVE, 1 }, /* Lookup */
{ 1, 2, 0, 0, LK_EXCLUSIVE, 1 }, /* Lookupp */
Move the NFSv4.1 client patches over from projects/nfsv4.1-client to head. I don't think the NFS client behaviour will change unless the new "minorversion=1" mount option is used. It includes basic NFSv4.1 support plus support for pNFS using the Files Layout only. All problems detecting during an NFSv4.1 Bakeathon testing event in June 2012 have been resolved in this code and it has been tested against the NFSv4.1 server available to me. Although not reviewed, I believe that kib@ has looked at it.
2012-12-08 22:52:39 +00:00
{ 0, 1, 0, 0, LK_EXCLUSIVE, 1 }, /* NVerify */
{ 1, 1, 0, 1, LK_EXCLUSIVE, 1 }, /* Open */
{ 1, 1, 0, 0, LK_EXCLUSIVE, 1 }, /* OpenAttr */
{ 0, 1, 0, 0, LK_EXCLUSIVE, 1 }, /* OpenConfirm */
{ 0, 1, 0, 0, LK_EXCLUSIVE, 1 }, /* OpenDowngrade */
{ 1, 0, 0, 0, LK_EXCLUSIVE, 1 }, /* PutFH */
{ 1, 0, 0, 0, LK_EXCLUSIVE, 1 }, /* PutPubFH */
{ 1, 0, 0, 0, LK_EXCLUSIVE, 1 }, /* PutRootFH */
{ 0, 1, 0, 0, LK_SHARED, 1 }, /* Read */
{ 0, 1, 0, 0, LK_SHARED, 1 }, /* Readdir */
{ 0, 1, 0, 0, LK_SHARED, 1 }, /* ReadLink */
{ 0, 2, 1, 1, LK_EXCLUSIVE, 1 }, /* Remove */
{ 2, 1, 1, 1, LK_EXCLUSIVE, 1 }, /* Rename */
{ 0, 0, 0, 0, LK_EXCLUSIVE, 1 }, /* Renew */
{ 0, 0, 0, 0, LK_EXCLUSIVE, 1 }, /* RestoreFH */
{ 0, 1, 0, 0, LK_EXCLUSIVE, 1 }, /* SaveFH */
{ 0, 1, 0, 0, LK_EXCLUSIVE, 1 }, /* SecInfo */
{ 0, 2, 1, 1, LK_EXCLUSIVE, 1 }, /* Setattr */
{ 0, 0, 0, 0, LK_EXCLUSIVE, 1 }, /* SetClientID */
{ 0, 0, 0, 0, LK_EXCLUSIVE, 1 }, /* SetClientIDConfirm */
{ 0, 1, 0, 0, LK_EXCLUSIVE, 1 }, /* Verify */
{ 0, 2, 1, 1, LK_EXCLUSIVE, 1 }, /* Write */
{ 0, 0, 0, 0, LK_EXCLUSIVE, 1 }, /* ReleaseLockOwner */
{ 0, 0, 0, 0, LK_EXCLUSIVE, 1 }, /* Backchannel Ctrl */
{ 0, 0, 0, 0, LK_EXCLUSIVE, 1 }, /* Bind Conn to Sess */
{ 0, 0, 0, 0, LK_EXCLUSIVE, 0 }, /* Exchange ID */
{ 0, 0, 0, 0, LK_EXCLUSIVE, 0 }, /* Create Session */
{ 0, 0, 0, 0, LK_EXCLUSIVE, 0 }, /* Destroy Session */
{ 0, 0, 0, 0, LK_EXCLUSIVE, 1 }, /* Free StateID */
{ 0, 0, 0, 0, LK_EXCLUSIVE, 1 }, /* Get Dir Deleg */
{ 0, 0, 0, 0, LK_EXCLUSIVE, 1 }, /* Get Device Info */
{ 0, 0, 0, 0, LK_EXCLUSIVE, 1 }, /* Get Device List */
{ 0, 1, 0, 0, LK_EXCLUSIVE, 1 }, /* Layout Commit */
{ 0, 1, 0, 0, LK_EXCLUSIVE, 1 }, /* Layout Get */
{ 0, 1, 0, 0, LK_EXCLUSIVE, 1 }, /* Layout Return */
{ 0, 0, 0, 0, LK_EXCLUSIVE, 1 }, /* Secinfo No name */
{ 0, 0, 0, 0, LK_EXCLUSIVE, 1 }, /* Sequence */
{ 0, 0, 0, 0, LK_EXCLUSIVE, 1 }, /* Set SSV */
{ 0, 0, 0, 0, LK_EXCLUSIVE, 1 }, /* Test StateID */
{ 0, 0, 0, 0, LK_EXCLUSIVE, 1 }, /* Want Delegation */
{ 0, 0, 0, 0, LK_EXCLUSIVE, 0 }, /* Destroy ClientID */
{ 0, 0, 0, 0, LK_EXCLUSIVE, 1 }, /* Reclaim Complete */
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
};
#endif /* !APPLEKEXT */
static int ncl_mbuf_mhlen = MHLEN;
static int nfsrv_usercnt = 0;
static int nfsrv_dnsnamelen;
static u_char *nfsrv_dnsname = NULL;
static int nfsrv_usermax = 999999999;
static struct nfsuserhashhead nfsuserhash[NFSUSERHASHSIZE];
static struct nfsuserhashhead nfsusernamehash[NFSUSERHASHSIZE];
static struct nfsuserhashhead nfsgrouphash[NFSGROUPHASHSIZE];
static struct nfsuserhashhead nfsgroupnamehash[NFSGROUPHASHSIZE];
static struct nfsuserlruhead nfsuserlruhead;
/*
* This static array indicates whether or not the RPC generates a large
* reply. This is used by nfs_reply() to decide whether or not an mbuf
* cluster should be allocated. (If a cluster is required by an RPC
* marked 0 in this array, the code will still work, just not quite as
* efficiently.)
*/
Move the NFSv4.1 client patches over from projects/nfsv4.1-client to head. I don't think the NFS client behaviour will change unless the new "minorversion=1" mount option is used. It includes basic NFSv4.1 support plus support for pNFS using the Files Layout only. All problems detecting during an NFSv4.1 Bakeathon testing event in June 2012 have been resolved in this code and it has been tested against the NFSv4.1 server available to me. Although not reviewed, I believe that kib@ has looked at it.
2012-12-08 22:52:39 +00:00
int nfs_bigreply[NFSV41_NPROCS] = { 0, 0, 0, 1, 0, 1, 1, 0, 0, 0, 0,
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
Move the NFSv4.1 client patches over from projects/nfsv4.1-client to head. I don't think the NFS client behaviour will change unless the new "minorversion=1" mount option is used. It includes basic NFSv4.1 support plus support for pNFS using the Files Layout only. All problems detecting during an NFSv4.1 Bakeathon testing event in June 2012 have been resolved in this code and it has been tested against the NFSv4.1 server available to me. Although not reviewed, I believe that kib@ has looked at it.
2012-12-08 22:52:39 +00:00
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0 };
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
/* local functions */
static int nfsrv_skipace(struct nfsrv_descript *nd, int *acesizep);
static void nfsv4_wanted(struct nfsv4lock *lp);
static int nfsrv_cmpmixedcase(u_char *cp, u_char *cp2, int len);
static int nfsrv_getuser(int procnum, uid_t uid, gid_t gid, char *name,
NFSPROC_T *p);
static void nfsrv_removeuser(struct nfsusrgrp *usrp);
static int nfsrv_getrefstr(struct nfsrv_descript *, u_char **, u_char **,
int *, int *);
static void nfsrv_refstrbigenough(int, u_char **, u_char **, int *);
#ifndef APPLE
/*
* copies mbuf chain to the uio scatter/gather list
*/
int
nfsm_mbufuio(struct nfsrv_descript *nd, struct uio *uiop, int siz)
{
char *mbufcp, *uiocp;
int xfer, left, len;
mbuf_t mp;
long uiosiz, rem;
int error = 0;
mp = nd->nd_md;
mbufcp = nd->nd_dpos;
len = NFSMTOD(mp, caddr_t) + mbuf_len(mp) - mbufcp;
rem = NFSM_RNDUP(siz) - siz;
while (siz > 0) {
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
if (uiop->uio_iovcnt <= 0 || uiop->uio_iov == NULL) {
error = EBADRPC;
goto out;
}
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
left = uiop->uio_iov->iov_len;
uiocp = uiop->uio_iov->iov_base;
if (left > siz)
left = siz;
uiosiz = left;
while (left > 0) {
while (len == 0) {
mp = mbuf_next(mp);
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
if (mp == NULL) {
error = EBADRPC;
goto out;
}
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
mbufcp = NFSMTOD(mp, caddr_t);
len = mbuf_len(mp);
Adjust the test of a KASSERT to better match the intent. This assertion was added in r246213 as a guard against corrupted mbufs arriving from drivers, the key distinguishing factor of said mbufs being that they had a negative length. Given we're in a while loop specifically designed to skip over zero-length mbufs, panicking on a zero-length mbuf seems incorrect. No objection from: kib
2014-12-19 19:09:22 +00:00
KASSERT(len >= 0,
("len %d, corrupted mbuf?", len));
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
}
xfer = (left > len) ? len : left;
#ifdef notdef
/* Not Yet.. */
if (uiop->uio_iov->iov_op != NULL)
(*(uiop->uio_iov->iov_op))
(mbufcp, uiocp, xfer);
else
#endif
if (uiop->uio_segflg == UIO_SYSSPACE)
NFSBCOPY(mbufcp, uiocp, xfer);
else
copyout(mbufcp, CAST_USER_ADDR_T(uiocp), xfer);
left -= xfer;
len -= xfer;
mbufcp += xfer;
uiocp += xfer;
uiop->uio_offset += xfer;
uiop->uio_resid -= xfer;
}
if (uiop->uio_iov->iov_len <= siz) {
uiop->uio_iovcnt--;
uiop->uio_iov++;
} else {
uiop->uio_iov->iov_base = (void *)
((char *)uiop->uio_iov->iov_base + uiosiz);
uiop->uio_iov->iov_len -= uiosiz;
}
siz -= uiosiz;
}
nd->nd_dpos = mbufcp;
nd->nd_md = mp;
if (rem > 0) {
if (len < rem)
error = nfsm_advance(nd, rem, len);
else
nd->nd_dpos += rem;
}
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
out:
NFSEXITCODE2(error, nd);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
return (error);
}
#endif /* !APPLE */
/*
* Help break down an mbuf chain by setting the first siz bytes contiguous
* pointed to by returned val.
* This is used by the macro NFSM_DISSECT for tough
* cases.
*/
APPLESTATIC void *
Revamp the old NFS server's File Handle Affinity (FHA) code so that it will work with either the old or new server. The FHA code keeps a cache of currently active file handles for NFSv2 and v3 requests, so that read and write requests for the same file are directed to the same group of threads (reads) or thread (writes). It does not currently work for NFSv4 requests. They are more complex, and will take more work to support. This improves read-ahead performance, especially with ZFS, if the FHA tuning parameters are configured appropriately. Without the FHA code, concurrent reads that are part of a sequential read from a file will be directed to separate NFS threads. This has the effect of confusing the ZFS zfetch (prefetch) code and makes sequential reads significantly slower with clients like Linux that do a lot of prefetching. The FHA code has also been updated to direct write requests to nearby file offsets to the same thread in the same way it batches reads, and the FHA code will now also send writes to multiple threads when needed. This improves sequential write performance in ZFS, because writes to a file are now more ordered. Since NFS writes (generally less than 64K) are smaller than the typical ZFS record size (usually 128K), out of order NFS writes to the same block can trigger a read in ZFS. Sending them down the same thread increases the odds of their being in order. In order for multiple write threads per file in the FHA code to be useful, writes in the NFS server have been changed to use a LK_SHARED vnode lock, and upgrade that to LK_EXCLUSIVE if the filesystem doesn't allow multiple writers to a file at once. ZFS is currently the only filesystem that allows multiple writers to a file, because it has internal file range locking. This change does not affect the NFSv4 code. This improves random write performance to a single file in ZFS, since we can now have multiple writers inside ZFS at one time. I have changed the default tuning parameters to a 22 bit (4MB) window size (from 256K) and unlimited commands per thread as a result of my benchmarking with ZFS. The FHA code has been updated to allow configuring the tuning parameters from loader tunable variables in addition to sysctl variables. The read offset window calculation has been slightly modified as well. Instead of having separate bins, each file handle has a rolling window of bin_shift size. This minimizes glitches in throughput when shifting from one bin to another. sys/conf/files: Add nfs_fha_new.c and nfs_fha_old.c. Compile nfs_fha.c when either the old or the new NFS server is built. sys/fs/nfs/nfsport.h, sys/fs/nfs/nfs_commonport.c: Bring in changes from Rick Macklem to newnfs_realign that allow it to operate in blocking (M_WAITOK) or non-blocking (M_NOWAIT) mode. sys/fs/nfs/nfs_commonsubs.c, sys/fs/nfs/nfs_var.h: Bring in a change from Rick Macklem to allow telling nfsm_dissect() whether or not to wait for mallocs. sys/fs/nfs/nfsm_subs.h: Bring in changes from Rick Macklem to create a new nfsm_dissect_nonblock() inline function and NFSM_DISSECT_NONBLOCK() macro. sys/fs/nfs/nfs_commonkrpc.c, sys/fs/nfsclient/nfs_clkrpc.c: Add the malloc wait flag to a newnfs_realign() call. sys/fs/nfsserver/nfs_nfsdkrpc.c: Setup the new NFS server's RPC thread pool so that it will call the FHA code. Add the malloc flag argument to newnfs_realign(). Unstaticize newnfs_nfsv3_procid[] so that we can use it in the FHA code. sys/fs/nfsserver/nfs_nfsdsocket.c: In nfsrvd_dorpc(), add NFSPROC_WRITE to the list of RPC types that use the LK_SHARED lock type. sys/fs/nfsserver/nfs_nfsdport.c: In nfsd_fhtovp(), if we're starting a write, check to see whether the underlying filesystem supports shared writes. If not, upgrade the lock type from LK_SHARED to LK_EXCLUSIVE. sys/nfsserver/nfs_fha.c: Remove all code that is specific to the NFS server implementation. Anything that is server-specific is now accessed through a callback supplied by that server's FHA shim in the new softc. There are now separate sysctls and tunables for the FHA implementations for the old and new NFS servers. The new NFS server has its tunables under vfs.nfsd.fha, the old NFS server's tunables are under vfs.nfsrv.fha as before. In fha_extract_info(), use callouts for all server-specific code. Getting file handles and offsets is now done in the individual server's shim module. In fha_hash_entry_choose_thread(), change the way we decide whether two reads are in proximity to each other. Previously, the calculation was a simple shift operation to see whether the offsets were in the same power of 2 bucket. The issue was that there would be a bucket (and therefore thread) transition, even if the reads were in close proximity. When there is a thread transition, reads wind up going somewhat out of order, and ZFS gets confused. The new calculation simply tries to see whether the offsets are within 1 << bin_shift of each other. If they are, the reads will be sent to the same thread. The effect of this change is that for sequential reads, if the client doesn't exceed the max_reqs_per_nfsd parameter and the bin_shift is set to a reasonable value (22, or 4MB works well in my tests), the reads in any sequential stream will largely be confined to a single thread. Change fha_assign() so that it takes a softc argument. It is now called from the individual server's shim code, which will pass in the softc. Change fhe_stats_sysctl() so that it takes a softc parameter. It is now called from the individual server's shim code. Add the current offset to the list of things printed out about each active thread. Change the num_reads and num_writes counters in the fha_hash_entry structure to 32-bit values, and rename them num_rw and num_exclusive, respectively, to reflect their changed usage. Add an enable sysctl and tunable that allows the user to disable the FHA code (when vfs.XXX.fha.enable = 0). This is useful for before/after performance comparisons. nfs_fha.h: Move most structure definitions out of nfs_fha.c and into the header file, so that the individual server shims can see them. Change the default bin_shift to 22 (4MB) instead of 18 (256K). Allow unlimited commands per thread. sys/nfsserver/nfs_fha_old.c, sys/nfsserver/nfs_fha_old.h, sys/fs/nfsserver/nfs_fha_new.c, sys/fs/nfsserver/nfs_fha_new.h: Add shims for the old and new NFS servers to interface with the FHA code, and callbacks for the The shims contain all of the code and definitions that are specific to the NFS servers. They setup the server-specific callbacks and set the server name for the sysctl and loader tunable variables. sys/nfsserver/nfs_srvkrpc.c: Configure the RPC code to call fhaold_assign() instead of fha_assign(). sys/modules/nfsd/Makefile: Add nfs_fha.c and nfs_fha_new.c. sys/modules/nfsserver/Makefile: Add nfs_fha_old.c. Reviewed by: rmacklem Sponsored by: Spectra Logic MFC after: 2 weeks
2013-04-17 21:00:22 +00:00
nfsm_dissct(struct nfsrv_descript *nd, int siz, int how)
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
{
mbuf_t mp2;
int siz2, xfer;
caddr_t p;
int left;
caddr_t retp;
retp = NULL;
left = NFSMTOD(nd->nd_md, caddr_t) + mbuf_len(nd->nd_md) - nd->nd_dpos;
while (left == 0) {
nd->nd_md = mbuf_next(nd->nd_md);
if (nd->nd_md == NULL)
return (retp);
left = mbuf_len(nd->nd_md);
nd->nd_dpos = NFSMTOD(nd->nd_md, caddr_t);
}
if (left >= siz) {
retp = nd->nd_dpos;
nd->nd_dpos += siz;
} else if (mbuf_next(nd->nd_md) == NULL) {
return (retp);
} else if (siz > ncl_mbuf_mhlen) {
panic("nfs S too big");
} else {
Revamp the old NFS server's File Handle Affinity (FHA) code so that it will work with either the old or new server. The FHA code keeps a cache of currently active file handles for NFSv2 and v3 requests, so that read and write requests for the same file are directed to the same group of threads (reads) or thread (writes). It does not currently work for NFSv4 requests. They are more complex, and will take more work to support. This improves read-ahead performance, especially with ZFS, if the FHA tuning parameters are configured appropriately. Without the FHA code, concurrent reads that are part of a sequential read from a file will be directed to separate NFS threads. This has the effect of confusing the ZFS zfetch (prefetch) code and makes sequential reads significantly slower with clients like Linux that do a lot of prefetching. The FHA code has also been updated to direct write requests to nearby file offsets to the same thread in the same way it batches reads, and the FHA code will now also send writes to multiple threads when needed. This improves sequential write performance in ZFS, because writes to a file are now more ordered. Since NFS writes (generally less than 64K) are smaller than the typical ZFS record size (usually 128K), out of order NFS writes to the same block can trigger a read in ZFS. Sending them down the same thread increases the odds of their being in order. In order for multiple write threads per file in the FHA code to be useful, writes in the NFS server have been changed to use a LK_SHARED vnode lock, and upgrade that to LK_EXCLUSIVE if the filesystem doesn't allow multiple writers to a file at once. ZFS is currently the only filesystem that allows multiple writers to a file, because it has internal file range locking. This change does not affect the NFSv4 code. This improves random write performance to a single file in ZFS, since we can now have multiple writers inside ZFS at one time. I have changed the default tuning parameters to a 22 bit (4MB) window size (from 256K) and unlimited commands per thread as a result of my benchmarking with ZFS. The FHA code has been updated to allow configuring the tuning parameters from loader tunable variables in addition to sysctl variables. The read offset window calculation has been slightly modified as well. Instead of having separate bins, each file handle has a rolling window of bin_shift size. This minimizes glitches in throughput when shifting from one bin to another. sys/conf/files: Add nfs_fha_new.c and nfs_fha_old.c. Compile nfs_fha.c when either the old or the new NFS server is built. sys/fs/nfs/nfsport.h, sys/fs/nfs/nfs_commonport.c: Bring in changes from Rick Macklem to newnfs_realign that allow it to operate in blocking (M_WAITOK) or non-blocking (M_NOWAIT) mode. sys/fs/nfs/nfs_commonsubs.c, sys/fs/nfs/nfs_var.h: Bring in a change from Rick Macklem to allow telling nfsm_dissect() whether or not to wait for mallocs. sys/fs/nfs/nfsm_subs.h: Bring in changes from Rick Macklem to create a new nfsm_dissect_nonblock() inline function and NFSM_DISSECT_NONBLOCK() macro. sys/fs/nfs/nfs_commonkrpc.c, sys/fs/nfsclient/nfs_clkrpc.c: Add the malloc wait flag to a newnfs_realign() call. sys/fs/nfsserver/nfs_nfsdkrpc.c: Setup the new NFS server's RPC thread pool so that it will call the FHA code. Add the malloc flag argument to newnfs_realign(). Unstaticize newnfs_nfsv3_procid[] so that we can use it in the FHA code. sys/fs/nfsserver/nfs_nfsdsocket.c: In nfsrvd_dorpc(), add NFSPROC_WRITE to the list of RPC types that use the LK_SHARED lock type. sys/fs/nfsserver/nfs_nfsdport.c: In nfsd_fhtovp(), if we're starting a write, check to see whether the underlying filesystem supports shared writes. If not, upgrade the lock type from LK_SHARED to LK_EXCLUSIVE. sys/nfsserver/nfs_fha.c: Remove all code that is specific to the NFS server implementation. Anything that is server-specific is now accessed through a callback supplied by that server's FHA shim in the new softc. There are now separate sysctls and tunables for the FHA implementations for the old and new NFS servers. The new NFS server has its tunables under vfs.nfsd.fha, the old NFS server's tunables are under vfs.nfsrv.fha as before. In fha_extract_info(), use callouts for all server-specific code. Getting file handles and offsets is now done in the individual server's shim module. In fha_hash_entry_choose_thread(), change the way we decide whether two reads are in proximity to each other. Previously, the calculation was a simple shift operation to see whether the offsets were in the same power of 2 bucket. The issue was that there would be a bucket (and therefore thread) transition, even if the reads were in close proximity. When there is a thread transition, reads wind up going somewhat out of order, and ZFS gets confused. The new calculation simply tries to see whether the offsets are within 1 << bin_shift of each other. If they are, the reads will be sent to the same thread. The effect of this change is that for sequential reads, if the client doesn't exceed the max_reqs_per_nfsd parameter and the bin_shift is set to a reasonable value (22, or 4MB works well in my tests), the reads in any sequential stream will largely be confined to a single thread. Change fha_assign() so that it takes a softc argument. It is now called from the individual server's shim code, which will pass in the softc. Change fhe_stats_sysctl() so that it takes a softc parameter. It is now called from the individual server's shim code. Add the current offset to the list of things printed out about each active thread. Change the num_reads and num_writes counters in the fha_hash_entry structure to 32-bit values, and rename them num_rw and num_exclusive, respectively, to reflect their changed usage. Add an enable sysctl and tunable that allows the user to disable the FHA code (when vfs.XXX.fha.enable = 0). This is useful for before/after performance comparisons. nfs_fha.h: Move most structure definitions out of nfs_fha.c and into the header file, so that the individual server shims can see them. Change the default bin_shift to 22 (4MB) instead of 18 (256K). Allow unlimited commands per thread. sys/nfsserver/nfs_fha_old.c, sys/nfsserver/nfs_fha_old.h, sys/fs/nfsserver/nfs_fha_new.c, sys/fs/nfsserver/nfs_fha_new.h: Add shims for the old and new NFS servers to interface with the FHA code, and callbacks for the The shims contain all of the code and definitions that are specific to the NFS servers. They setup the server-specific callbacks and set the server name for the sysctl and loader tunable variables. sys/nfsserver/nfs_srvkrpc.c: Configure the RPC code to call fhaold_assign() instead of fha_assign(). sys/modules/nfsd/Makefile: Add nfs_fha.c and nfs_fha_new.c. sys/modules/nfsserver/Makefile: Add nfs_fha_old.c. Reviewed by: rmacklem Sponsored by: Spectra Logic MFC after: 2 weeks
2013-04-17 21:00:22 +00:00
MGET(mp2, MT_DATA, how);
if (mp2 == NULL)
return (NULL);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
mbuf_setnext(mp2, mbuf_next(nd->nd_md));
mbuf_setnext(nd->nd_md, mp2);
mbuf_setlen(nd->nd_md, mbuf_len(nd->nd_md) - left);
nd->nd_md = mp2;
retp = p = NFSMTOD(mp2, caddr_t);
NFSBCOPY(nd->nd_dpos, p, left); /* Copy what was left */
siz2 = siz - left;
p += left;
mp2 = mbuf_next(mp2);
/* Loop around copying up the siz2 bytes */
while (siz2 > 0) {
if (mp2 == NULL)
return (NULL);
xfer = (siz2 > mbuf_len(mp2)) ? mbuf_len(mp2) : siz2;
if (xfer > 0) {
NFSBCOPY(NFSMTOD(mp2, caddr_t), p, xfer);
NFSM_DATAP(mp2, xfer);
mbuf_setlen(mp2, mbuf_len(mp2) - xfer);
p += xfer;
siz2 -= xfer;
}
if (siz2 > 0)
mp2 = mbuf_next(mp2);
}
mbuf_setlen(nd->nd_md, siz);
nd->nd_md = mp2;
nd->nd_dpos = NFSMTOD(mp2, caddr_t);
}
return (retp);
}
/*
* Advance the position in the mbuf chain.
* If offs == 0, this is a no-op, but it is simpler to just return from
* here than check for offs > 0 for all calls to nfsm_advance.
* If left == -1, it should be calculated here.
*/
APPLESTATIC int
nfsm_advance(struct nfsrv_descript *nd, int offs, int left)
{
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
int error = 0;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
if (offs == 0)
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
goto out;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
/*
* A negative offs should be considered a serious problem.
*/
if (offs < 0)
panic("nfsrv_advance");
/*
* If left == -1, calculate it here.
*/
if (left == -1)
left = NFSMTOD(nd->nd_md, caddr_t) + mbuf_len(nd->nd_md) -
nd->nd_dpos;
/*
* Loop around, advancing over the mbuf data.
*/
while (offs > left) {
offs -= left;
nd->nd_md = mbuf_next(nd->nd_md);
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
if (nd->nd_md == NULL) {
error = EBADRPC;
goto out;
}
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
left = mbuf_len(nd->nd_md);
nd->nd_dpos = NFSMTOD(nd->nd_md, caddr_t);
}
nd->nd_dpos += offs;
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
out:
NFSEXITCODE(error);
return (error);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
}
/*
* Copy a string into mbuf(s).
* Return the number of bytes output, including XDR overheads.
*/
APPLESTATIC int
nfsm_strtom(struct nfsrv_descript *nd, const char *cp, int siz)
{
mbuf_t m2;
int xfer, left;
mbuf_t m1;
int rem, bytesize;
u_int32_t *tl;
char *cp2;
NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED);
*tl = txdr_unsigned(siz);
rem = NFSM_RNDUP(siz) - siz;
bytesize = NFSX_UNSIGNED + siz + rem;
m2 = nd->nd_mb;
cp2 = nd->nd_bpos;
left = M_TRAILINGSPACE(m2);
/*
* Loop around copying the string to mbuf(s).
*/
while (siz > 0) {
if (left == 0) {
if (siz > ncl_mbuf_mlen)
Mechanically substitute flags from historic mbuf allocator with malloc(9) flags within sys. Exceptions: - sys/contrib not touched - sys/mbuf.h edited manually
2012-12-05 08:04:20 +00:00
NFSMCLGET(m1, M_WAITOK);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
else
NFSMGET(m1);
mbuf_setlen(m1, 0);
mbuf_setnext(m2, m1);
m2 = m1;
cp2 = NFSMTOD(m2, caddr_t);
left = M_TRAILINGSPACE(m2);
}
if (left >= siz)
xfer = siz;
else
xfer = left;
NFSBCOPY(cp, cp2, xfer);
cp += xfer;
mbuf_setlen(m2, mbuf_len(m2) + xfer);
siz -= xfer;
left -= xfer;
if (siz == 0 && rem) {
if (left < rem)
panic("nfsm_strtom");
NFSBZERO(cp2 + xfer, rem);
mbuf_setlen(m2, mbuf_len(m2) + rem);
}
}
nd->nd_mb = m2;
nd->nd_bpos = NFSMTOD(m2, caddr_t) + mbuf_len(m2);
return (bytesize);
}
/*
* Called once to initialize data structures...
*/
APPLESTATIC void
newnfs_init(void)
{
static int nfs_inited = 0;
if (nfs_inited)
return;
nfs_inited = 1;
newnfs_true = txdr_unsigned(TRUE);
newnfs_false = txdr_unsigned(FALSE);
newnfs_xdrneg1 = txdr_unsigned(-1);
nfscl_ticks = (hz * NFS_TICKINTVL + 500) / 1000;
if (nfscl_ticks < 1)
nfscl_ticks = 1;
NFSSETBOOTTIME(nfsboottime);
/*
* Initialize reply list and start timer
*/
TAILQ_INIT(&nfsd_reqq);
NFS_TIMERINIT;
}
/*
* Put a file handle in an mbuf list.
* If the size argument == 0, just use the default size.
* set_true == 1 if there should be an newnfs_true prepended on the file handle.
* Return the number of bytes output, including XDR overhead.
*/
APPLESTATIC int
nfsm_fhtom(struct nfsrv_descript *nd, u_int8_t *fhp, int size, int set_true)
{
u_int32_t *tl;
u_int8_t *cp;
int fullsiz, rem, bytesize = 0;
if (size == 0)
size = NFSX_MYFH;
switch (nd->nd_flag & (ND_NFSV2 | ND_NFSV3 | ND_NFSV4)) {
case ND_NFSV2:
if (size > NFSX_V2FH)
panic("fh size > NFSX_V2FH for NFSv2");
NFSM_BUILD(cp, u_int8_t *, NFSX_V2FH);
NFSBCOPY(fhp, cp, size);
if (size < NFSX_V2FH)
NFSBZERO(cp + size, NFSX_V2FH - size);
bytesize = NFSX_V2FH;
break;
case ND_NFSV3:
case ND_NFSV4:
fullsiz = NFSM_RNDUP(size);
rem = fullsiz - size;
if (set_true) {
bytesize = 2 * NFSX_UNSIGNED + fullsiz;
NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED);
*tl = newnfs_true;
} else {
bytesize = NFSX_UNSIGNED + fullsiz;
}
(void) nfsm_strtom(nd, fhp, size);
break;
};
return (bytesize);
}
/*
* This function compares two net addresses by family and returns TRUE
* if they are the same host.
* If there is any doubt, return FALSE.
* The AF_INET family is handled as a special case so that address mbufs
* don't need to be saved to store "struct in_addr", which is only 4 bytes.
*/
APPLESTATIC int
nfsaddr_match(int family, union nethostaddr *haddr, NFSSOCKADDR_T nam)
{
struct sockaddr_in *inetaddr;
switch (family) {
case AF_INET:
inetaddr = NFSSOCKADDR(nam, struct sockaddr_in *);
if (inetaddr->sin_family == AF_INET &&
inetaddr->sin_addr.s_addr == haddr->had_inet.s_addr)
return (1);
break;
#ifdef INET6
case AF_INET6:
{
struct sockaddr_in6 *inetaddr6;
inetaddr6 = NFSSOCKADDR(nam, struct sockaddr_in6 *);
/* XXX - should test sin6_scope_id ? */
if (inetaddr6->sin6_family == AF_INET6 &&
IN6_ARE_ADDR_EQUAL(&inetaddr6->sin6_addr,
&haddr->had_inet6))
return (1);
}
break;
#endif
};
return (0);
}
/*
* Similar to the above, but takes to NFSSOCKADDR_T args.
*/
APPLESTATIC int
nfsaddr2_match(NFSSOCKADDR_T nam1, NFSSOCKADDR_T nam2)
{
struct sockaddr_in *addr1, *addr2;
struct sockaddr *inaddr;
inaddr = NFSSOCKADDR(nam1, struct sockaddr *);
switch (inaddr->sa_family) {
case AF_INET:
addr1 = NFSSOCKADDR(nam1, struct sockaddr_in *);
addr2 = NFSSOCKADDR(nam2, struct sockaddr_in *);
if (addr2->sin_family == AF_INET &&
addr1->sin_addr.s_addr == addr2->sin_addr.s_addr)
return (1);
break;
#ifdef INET6
case AF_INET6:
{
struct sockaddr_in6 *inet6addr1, *inet6addr2;
inet6addr1 = NFSSOCKADDR(nam1, struct sockaddr_in6 *);
inet6addr2 = NFSSOCKADDR(nam2, struct sockaddr_in6 *);
/* XXX - should test sin6_scope_id ? */
if (inet6addr2->sin6_family == AF_INET6 &&
IN6_ARE_ADDR_EQUAL(&inet6addr1->sin6_addr,
&inet6addr2->sin6_addr))
return (1);
}
break;
#endif
};
return (0);
}
/*
* Trim the stuff already dissected off the mbuf list.
*/
APPLESTATIC void
newnfs_trimleading(nd)
struct nfsrv_descript *nd;
{
mbuf_t m, n;
int offs;
/*
* First, free up leading mbufs.
*/
if (nd->nd_mrep != nd->nd_md) {
m = nd->nd_mrep;
while (mbuf_next(m) != nd->nd_md) {
if (mbuf_next(m) == NULL)
panic("nfsm trim leading");
m = mbuf_next(m);
}
mbuf_setnext(m, NULL);
mbuf_freem(nd->nd_mrep);
}
m = nd->nd_md;
/*
* Now, adjust this mbuf, based on nd_dpos.
*/
offs = nd->nd_dpos - NFSMTOD(m, caddr_t);
if (offs == mbuf_len(m)) {
n = m;
m = mbuf_next(m);
if (m == NULL)
panic("nfsm trim leading2");
mbuf_setnext(n, NULL);
mbuf_freem(n);
} else if (offs > 0) {
mbuf_setlen(m, mbuf_len(m) - offs);
NFSM_DATAP(m, offs);
} else if (offs < 0)
panic("nfsm trimleading offs");
nd->nd_mrep = m;
nd->nd_md = m;
nd->nd_dpos = NFSMTOD(m, caddr_t);
}
/*
* Trim trailing data off the mbuf list being built.
*/
APPLESTATIC void
newnfs_trimtrailing(nd, mb, bpos)
struct nfsrv_descript *nd;
mbuf_t mb;
caddr_t bpos;
{
if (mbuf_next(mb)) {
mbuf_freem(mbuf_next(mb));
mbuf_setnext(mb, NULL);
}
mbuf_setlen(mb, bpos - NFSMTOD(mb, caddr_t));
nd->nd_mb = mb;
nd->nd_bpos = bpos;
}
/*
* Dissect a file handle on the client.
*/
APPLESTATIC int
nfsm_getfh(struct nfsrv_descript *nd, struct nfsfh **nfhpp)
{
u_int32_t *tl;
struct nfsfh *nfhp;
int error, len;
*nfhpp = NULL;
if (nd->nd_flag & (ND_NFSV3 | ND_NFSV4)) {
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
if ((len = fxdr_unsigned(int, *tl)) <= 0 ||
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
len > NFSX_FHMAX) {
error = EBADRPC;
goto nfsmout;
}
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
} else
len = NFSX_V2FH;
MALLOC(nfhp, struct nfsfh *, sizeof (struct nfsfh) + len,
M_NFSFH, M_WAITOK);
error = nfsrv_mtostr(nd, nfhp->nfh_fh, len);
if (error) {
FREE((caddr_t)nfhp, M_NFSFH);
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
goto nfsmout;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
}
nfhp->nfh_len = len;
*nfhpp = nfhp;
nfsmout:
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
NFSEXITCODE2(error, nd);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
return (error);
}
/*
* Break down the nfsv4 acl.
* If the aclp == NULL or won't fit in an acl, just discard the acl info.
*/
APPLESTATIC int
nfsrv_dissectacl(struct nfsrv_descript *nd, NFSACL_T *aclp, int *aclerrp,
int *aclsizep, __unused NFSPROC_T *p)
{
u_int32_t *tl;
int i, aclsize;
int acecnt, error = 0, aceerr = 0, acesize;
*aclerrp = 0;
if (aclp)
aclp->acl_cnt = 0;
/*
* Parse out the ace entries and expect them to conform to
* what can be supported by R/W/X bits.
*/
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
aclsize = NFSX_UNSIGNED;
acecnt = fxdr_unsigned(int, *tl);
if (acecnt > ACL_MAX_ENTRIES)
Small acl patch to return the aclerror that comes back from nfsrv_dissectacl(). This fixes a problem where ATTRNOTSUPP was being returned instead of BADOWNER. Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:04:57 +00:00
aceerr = NFSERR_ATTRNOTSUPP;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
if (nfsrv_useacl == 0)
Small acl patch to return the aclerror that comes back from nfsrv_dissectacl(). This fixes a problem where ATTRNOTSUPP was being returned instead of BADOWNER. Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:04:57 +00:00
aceerr = NFSERR_ATTRNOTSUPP;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
for (i = 0; i < acecnt; i++) {
if (aclp && !aceerr)
error = nfsrv_dissectace(nd, &aclp->acl_entry[i],
&aceerr, &acesize, p);
else
error = nfsrv_skipace(nd, &acesize);
if (error)
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
goto nfsmout;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
aclsize += acesize;
}
if (aclp && !aceerr)
aclp->acl_cnt = acecnt;
if (aceerr)
*aclerrp = aceerr;
if (aclsizep)
*aclsizep = aclsize;
nfsmout:
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
NFSEXITCODE2(error, nd);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
return (error);
}
/*
* Skip over an NFSv4 ace entry. Just dissect the xdr and discard it.
*/
static int
nfsrv_skipace(struct nfsrv_descript *nd, int *acesizep)
{
u_int32_t *tl;
int error, len = 0;
NFSM_DISSECT(tl, u_int32_t *, 4 * NFSX_UNSIGNED);
len = fxdr_unsigned(int, *(tl + 3));
error = nfsm_advance(nd, NFSM_RNDUP(len), -1);
nfsmout:
*acesizep = NFSM_RNDUP(len) + (4 * NFSX_UNSIGNED);
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
NFSEXITCODE2(error, nd);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
return (error);
}
/*
* Get attribute bits from an mbuf list.
* Returns EBADRPC for a parsing error, 0 otherwise.
* If the clearinvalid flag is set, clear the bits not supported.
*/
APPLESTATIC int
nfsrv_getattrbits(struct nfsrv_descript *nd, nfsattrbit_t *attrbitp, int *cntp,
int *retnotsupp)
{
u_int32_t *tl;
int cnt, i, outcnt;
int error = 0;
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
cnt = fxdr_unsigned(int, *tl);
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
if (cnt < 0) {
error = NFSERR_BADXDR;
goto nfsmout;
}
The NFSv4 server incorrectly assumed that the high order words of the attribute bitmap argument would be non-zero. This caused an interoperability problem for a recent patch to the Linux NFSv4 client. The Linux folks have changed their patch to avoid this, but this patch fixes the problem on the server. Reported and tested by: Andre Heider (a.heider@gmail.com) MFC after: 3 days
2013-07-20 22:35:32 +00:00
if (cnt > NFSATTRBIT_MAXWORDS)
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
outcnt = NFSATTRBIT_MAXWORDS;
The NFSv4 server incorrectly assumed that the high order words of the attribute bitmap argument would be non-zero. This caused an interoperability problem for a recent patch to the Linux NFSv4 client. The Linux folks have changed their patch to avoid this, but this patch fixes the problem on the server. Reported and tested by: Andre Heider (a.heider@gmail.com) MFC after: 3 days
2013-07-20 22:35:32 +00:00
else
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
outcnt = cnt;
NFSZERO_ATTRBIT(attrbitp);
if (outcnt > 0) {
NFSM_DISSECT(tl, u_int32_t *, outcnt * NFSX_UNSIGNED);
for (i = 0; i < outcnt; i++)
attrbitp->bits[i] = fxdr_unsigned(u_int32_t, *tl++);
}
The NFSv4 server incorrectly assumed that the high order words of the attribute bitmap argument would be non-zero. This caused an interoperability problem for a recent patch to the Linux NFSv4 client. The Linux folks have changed their patch to avoid this, but this patch fixes the problem on the server. Reported and tested by: Andre Heider (a.heider@gmail.com) MFC after: 3 days
2013-07-20 22:35:32 +00:00
for (i = 0; i < (cnt - outcnt); i++) {
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
if (retnotsupp != NULL && *tl != 0)
*retnotsupp = NFSERR_ATTRNOTSUPP;
}
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
if (cntp)
*cntp = NFSX_UNSIGNED + (cnt * NFSX_UNSIGNED);
nfsmout:
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
NFSEXITCODE2(error, nd);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
return (error);
}
/*
* Get the attributes for V4.
* If the compare flag is true, test for any attribute changes,
* otherwise return the attribute values.
* These attributes cover fields in "struct vattr", "struct statfs",
* "struct nfsfsinfo", the file handle and the lease duration.
* The value of retcmpp is set to 1 if all attributes are the same,
* and 0 otherwise.
* Returns EBADRPC if it can't be parsed, 0 otherwise.
*/
APPLESTATIC int
nfsv4_loadattr(struct nfsrv_descript *nd, vnode_t vp,
struct nfsvattr *nap, struct nfsfh **nfhpp, fhandle_t *fhp, int fhsize,
struct nfsv3_pathconf *pc, struct statfs *sbp, struct nfsstatfs *sfp,
struct nfsfsinfo *fsp, NFSACL_T *aclp, int compare, int *retcmpp,
u_int32_t *leasep, u_int32_t *rderrp, NFSPROC_T *p, struct ucred *cred)
{
u_int32_t *tl;
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
int i = 0, j, k, l = 0, m, bitpos, attrsum = 0;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
int error, tfhsize, aceerr, attrsize, cnt, retnotsup;
u_char *cp, *cp2, namestr[NFSV4_SMALLSTR + 1];
nfsattrbit_t attrbits, retattrbits, checkattrbits;
struct nfsfh *tnfhp;
struct nfsreferral *refp;
u_quad_t tquad;
nfsquad_t tnfsquad;
struct timespec temptime;
uid_t uid;
gid_t gid;
long fid;
u_int32_t freenum = 0, tuint;
u_int64_t uquad = 0, thyp, thyp2;
#ifdef QUOTA
struct dqblk dqb;
uid_t savuid;
#endif
if (compare) {
retnotsup = 0;
error = nfsrv_getattrbits(nd, &attrbits, NULL, &retnotsup);
} else {
error = nfsrv_getattrbits(nd, &attrbits, NULL, NULL);
}
if (error)
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
goto nfsmout;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
if (compare) {
*retcmpp = retnotsup;
} else {
/*
* Just set default values to some of the important ones.
*/
if (nap != NULL) {
nap->na_type = VREG;
nap->na_mode = 0;
nap->na_rdev = (NFSDEV_T)0;
nap->na_mtime.tv_sec = 0;
nap->na_mtime.tv_nsec = 0;
nap->na_gen = 0;
nap->na_flags = 0;
nap->na_blocksize = NFS_FABLKSIZE;
}
if (sbp != NULL) {
sbp->f_bsize = NFS_FABLKSIZE;
sbp->f_blocks = 0;
sbp->f_bfree = 0;
sbp->f_bavail = 0;
sbp->f_files = 0;
sbp->f_ffree = 0;
}
if (fsp != NULL) {
fsp->fs_rtmax = 8192;
fsp->fs_rtpref = 8192;
fsp->fs_maxname = NFS_MAXNAMLEN;
fsp->fs_wtmax = 8192;
fsp->fs_wtpref = 8192;
fsp->fs_wtmult = NFS_FABLKSIZE;
fsp->fs_dtpref = 8192;
fsp->fs_maxfilesize = 0xffffffffffffffffull;
fsp->fs_timedelta.tv_sec = 0;
fsp->fs_timedelta.tv_nsec = 1;
fsp->fs_properties = (NFSV3_FSFLINK | NFSV3_FSFSYMLINK |
NFSV3_FSFHOMOGENEOUS | NFSV3_FSFCANSETTIME);
}
if (pc != NULL) {
pc->pc_linkmax = LINK_MAX;
pc->pc_namemax = NAME_MAX;
pc->pc_notrunc = 0;
pc->pc_chownrestricted = 0;
pc->pc_caseinsensitive = 0;
pc->pc_casepreserving = 1;
}
}
/*
* Loop around getting the attributes.
*/
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
attrsize = fxdr_unsigned(int, *tl);
for (bitpos = 0; bitpos < NFSATTRBIT_MAX; bitpos++) {
if (attrsum > attrsize) {
error = NFSERR_BADXDR;
goto nfsmout;
}
if (NFSISSET_ATTRBIT(&attrbits, bitpos))
switch (bitpos) {
case NFSATTRBIT_SUPPORTEDATTRS:
retnotsup = 0;
if (compare || nap == NULL)
error = nfsrv_getattrbits(nd, &retattrbits,
&cnt, &retnotsup);
else
error = nfsrv_getattrbits(nd, &nap->na_suppattr,
&cnt, &retnotsup);
if (error)
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
goto nfsmout;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
if (compare && !(*retcmpp)) {
NFSSETSUPP_ATTRBIT(&checkattrbits);
Fix failures and warnings reported by newpynfs20090424 test tool. This fix addresses only issues with the pynfs reports, none of these issues are know to create problems for extant real clients. Submitted by: Bart Hsiao <bart.hsiao@gmail.com> Reworked by: myself Reviewed by: rmacklem Approved by: rmacklem Sponsored by: QNAP Systems Inc.
2014-10-03 02:24:41 +00:00
/* Some filesystem do not support NFSv4ACL */
if (nfsrv_useacl == 0 || nfs_supportsnfsv4acls(vp) == 0) {
NFSCLRBIT_ATTRBIT(&checkattrbits, NFSATTRBIT_ACL);
NFSCLRBIT_ATTRBIT(&checkattrbits, NFSATTRBIT_ACLSUPPORT);
}
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
if (!NFSEQUAL_ATTRBIT(&retattrbits, &checkattrbits)
|| retnotsup)
*retcmpp = NFSERR_NOTSAME;
}
attrsum += cnt;
break;
case NFSATTRBIT_TYPE:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
if (compare) {
if (!(*retcmpp)) {
if (nap->na_type != nfsv34tov_type(*tl))
*retcmpp = NFSERR_NOTSAME;
}
} else if (nap != NULL) {
nap->na_type = nfsv34tov_type(*tl);
}
attrsum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_FHEXPIRETYPE:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
if (compare && !(*retcmpp)) {
if (fxdr_unsigned(int, *tl) !=
NFSV4FHTYPE_PERSISTENT)
*retcmpp = NFSERR_NOTSAME;
}
attrsum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_CHANGE:
NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER);
if (compare) {
if (!(*retcmpp)) {
if (nap->na_filerev != fxdr_hyper(tl))
*retcmpp = NFSERR_NOTSAME;
}
} else if (nap != NULL) {
nap->na_filerev = fxdr_hyper(tl);
}
attrsum += NFSX_HYPER;
break;
case NFSATTRBIT_SIZE:
NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER);
if (compare) {
if (!(*retcmpp)) {
if (nap->na_size != fxdr_hyper(tl))
*retcmpp = NFSERR_NOTSAME;
}
} else if (nap != NULL) {
nap->na_size = fxdr_hyper(tl);
}
attrsum += NFSX_HYPER;
break;
case NFSATTRBIT_LINKSUPPORT:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
if (compare) {
if (!(*retcmpp)) {
if (fsp->fs_properties & NFSV3_FSFLINK) {
if (*tl == newnfs_false)
*retcmpp = NFSERR_NOTSAME;
} else {
if (*tl == newnfs_true)
*retcmpp = NFSERR_NOTSAME;
}
}
} else if (fsp != NULL) {
if (*tl == newnfs_true)
fsp->fs_properties |= NFSV3_FSFLINK;
else
fsp->fs_properties &= ~NFSV3_FSFLINK;
}
attrsum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_SYMLINKSUPPORT:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
if (compare) {
if (!(*retcmpp)) {
if (fsp->fs_properties & NFSV3_FSFSYMLINK) {
if (*tl == newnfs_false)
*retcmpp = NFSERR_NOTSAME;
} else {
if (*tl == newnfs_true)
*retcmpp = NFSERR_NOTSAME;
}
}
} else if (fsp != NULL) {
if (*tl == newnfs_true)
fsp->fs_properties |= NFSV3_FSFSYMLINK;
else
fsp->fs_properties &= ~NFSV3_FSFSYMLINK;
}
attrsum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_NAMEDATTR:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
if (compare && !(*retcmpp)) {
if (*tl != newnfs_false)
*retcmpp = NFSERR_NOTSAME;
}
attrsum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_FSID:
NFSM_DISSECT(tl, u_int32_t *, 4 * NFSX_UNSIGNED);
thyp = fxdr_hyper(tl);
tl += 2;
thyp2 = fxdr_hyper(tl);
if (compare) {
if (*retcmpp == 0) {
if (thyp != (u_int64_t)
vfs_statfs(vnode_mount(vp))->f_fsid.val[0] ||
thyp2 != (u_int64_t)
vfs_statfs(vnode_mount(vp))->f_fsid.val[1])
*retcmpp = NFSERR_NOTSAME;
}
} else if (nap != NULL) {
nap->na_filesid[0] = thyp;
nap->na_filesid[1] = thyp2;
}
attrsum += (4 * NFSX_UNSIGNED);
break;
case NFSATTRBIT_UNIQUEHANDLES:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
if (compare && !(*retcmpp)) {
if (*tl != newnfs_true)
*retcmpp = NFSERR_NOTSAME;
}
attrsum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_LEASETIME:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
if (compare) {
if (fxdr_unsigned(int, *tl) != nfsrv_lease &&
!(*retcmpp))
*retcmpp = NFSERR_NOTSAME;
} else if (leasep != NULL) {
*leasep = fxdr_unsigned(u_int32_t, *tl);
}
attrsum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_RDATTRERROR:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
if (compare) {
if (!(*retcmpp))
*retcmpp = NFSERR_INVAL;
} else if (rderrp != NULL) {
*rderrp = fxdr_unsigned(u_int32_t, *tl);
}
attrsum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_ACL:
if (compare) {
if (!(*retcmpp)) {
Fix failures and warnings reported by newpynfs20090424 test tool. This fix addresses only issues with the pynfs reports, none of these issues are know to create problems for extant real clients. Submitted by: Bart Hsiao <bart.hsiao@gmail.com> Reworked by: myself Reviewed by: rmacklem Approved by: rmacklem Sponsored by: QNAP Systems Inc.
2014-10-03 02:24:41 +00:00
if (nfsrv_useacl && nfs_supportsnfsv4acls(vp)) {
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
NFSACL_T *naclp;
Fix the experimental nfs subsystem so that it builds with the current NFSv4 ACLs, as defined in sys/acl.h. It still needs a way to test a mount point for NFSv4 ACL support before it will work. Until then, the NFSHASNFS4ACL() macro just always returns 0. Approved by: kib (mentor)
2009-05-26 22:21:53 +00:00
naclp = acl_alloc(M_WAITOK);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
error = nfsrv_dissectacl(nd, naclp, &aceerr,
&cnt, p);
if (error) {
acl_free(naclp);
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
goto nfsmout;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
}
Revert revision 224079 as Rick pointed out that I would be calling VOP_PATHCONF without the vnode lock held. Implicitly approved by: zml (mentor)
2011-07-17 03:44:05 +00:00
if (aceerr || aclp == NULL ||
nfsrv_compareacl(aclp, naclp))
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
*retcmpp = NFSERR_NOTSAME;
acl_free(naclp);
Remove unneeded ifdefs. Reviewed by: rmacklem
2009-12-03 18:03:42 +00:00
} else {
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
error = nfsrv_dissectacl(nd, NULL, &aceerr,
&cnt, p);
*retcmpp = NFSERR_ATTRNOTSUPP;
}
}
} else {
Fix failures and warnings reported by newpynfs20090424 test tool. This fix addresses only issues with the pynfs reports, none of these issues are know to create problems for extant real clients. Submitted by: Bart Hsiao <bart.hsiao@gmail.com> Reworked by: myself Reviewed by: rmacklem Approved by: rmacklem Sponsored by: QNAP Systems Inc.
2014-10-03 02:24:41 +00:00
if (vp != NULL && aclp != NULL)
error = nfsrv_dissectacl(nd, aclp, &aceerr,
&cnt, p);
else
error = nfsrv_dissectacl(nd, NULL, &aceerr,
&cnt, p);
if (error)
goto nfsmout;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
}
Fix failures and warnings reported by newpynfs20090424 test tool. This fix addresses only issues with the pynfs reports, none of these issues are know to create problems for extant real clients. Submitted by: Bart Hsiao <bart.hsiao@gmail.com> Reworked by: myself Reviewed by: rmacklem Approved by: rmacklem Sponsored by: QNAP Systems Inc.
2014-10-03 02:24:41 +00:00
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
attrsum += cnt;
break;
case NFSATTRBIT_ACLSUPPORT:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
if (compare && !(*retcmpp)) {
Fix failures and warnings reported by newpynfs20090424 test tool. This fix addresses only issues with the pynfs reports, none of these issues are know to create problems for extant real clients. Submitted by: Bart Hsiao <bart.hsiao@gmail.com> Reworked by: myself Reviewed by: rmacklem Approved by: rmacklem Sponsored by: QNAP Systems Inc.
2014-10-03 02:24:41 +00:00
if (nfsrv_useacl && nfs_supportsnfsv4acls(vp)) {
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
if (fxdr_unsigned(u_int32_t, *tl) !=
NFSV4ACE_SUPTYPES)
*retcmpp = NFSERR_NOTSAME;
} else {
*retcmpp = NFSERR_ATTRNOTSUPP;
}
}
attrsum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_ARCHIVE:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
if (compare && !(*retcmpp))
*retcmpp = NFSERR_ATTRNOTSUPP;
attrsum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_CANSETTIME:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
if (compare) {
if (!(*retcmpp)) {
if (fsp->fs_properties & NFSV3_FSFCANSETTIME) {
if (*tl == newnfs_false)
*retcmpp = NFSERR_NOTSAME;
} else {
if (*tl == newnfs_true)
*retcmpp = NFSERR_NOTSAME;
}
}
} else if (fsp != NULL) {
if (*tl == newnfs_true)
fsp->fs_properties |= NFSV3_FSFCANSETTIME;
else
fsp->fs_properties &= ~NFSV3_FSFCANSETTIME;
}
attrsum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_CASEINSENSITIVE:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
if (compare) {
if (!(*retcmpp)) {
if (*tl != newnfs_false)
*retcmpp = NFSERR_NOTSAME;
}
} else if (pc != NULL) {
pc->pc_caseinsensitive =
fxdr_unsigned(u_int32_t, *tl);
}
attrsum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_CASEPRESERVING:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
if (compare) {
if (!(*retcmpp)) {
if (*tl != newnfs_true)
*retcmpp = NFSERR_NOTSAME;
}
} else if (pc != NULL) {
pc->pc_casepreserving =
fxdr_unsigned(u_int32_t, *tl);
}
attrsum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_CHOWNRESTRICTED:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
if (compare) {
if (!(*retcmpp)) {
Revert revision 224079 as Rick pointed out that I would be calling VOP_PATHCONF without the vnode lock held. Implicitly approved by: zml (mentor)
2011-07-17 03:44:05 +00:00
if (*tl != newnfs_true)
*retcmpp = NFSERR_NOTSAME;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
}
} else if (pc != NULL) {
pc->pc_chownrestricted =
fxdr_unsigned(u_int32_t, *tl);
}
attrsum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_FILEHANDLE:
error = nfsm_getfh(nd, &tnfhp);
if (error)
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
goto nfsmout;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
tfhsize = tnfhp->nfh_len;
if (compare) {
if (!(*retcmpp) &&
!NFSRV_CMPFH(tnfhp->nfh_fh, tfhsize,
fhp, fhsize))
*retcmpp = NFSERR_NOTSAME;
FREE((caddr_t)tnfhp, M_NFSFH);
} else if (nfhpp != NULL) {
*nfhpp = tnfhp;
} else {
FREE((caddr_t)tnfhp, M_NFSFH);
}
attrsum += (NFSX_UNSIGNED + NFSM_RNDUP(tfhsize));
break;
case NFSATTRBIT_FILEID:
NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER);
thyp = fxdr_hyper(tl);
if (compare) {
if (!(*retcmpp)) {
if ((u_int64_t)nap->na_fileid != thyp)
*retcmpp = NFSERR_NOTSAME;
}
} else if (nap != NULL) {
if (*tl++)
printf("NFSv4 fileid > 32bits\n");
nap->na_fileid = thyp;
}
attrsum += NFSX_HYPER;
break;
case NFSATTRBIT_FILESAVAIL:
NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER);
if (compare) {
if (!(*retcmpp) &&
sfp->sf_afiles != fxdr_hyper(tl))
*retcmpp = NFSERR_NOTSAME;
} else if (sfp != NULL) {
sfp->sf_afiles = fxdr_hyper(tl);
}
attrsum += NFSX_HYPER;
break;
case NFSATTRBIT_FILESFREE:
NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER);
if (compare) {
if (!(*retcmpp) &&
sfp->sf_ffiles != fxdr_hyper(tl))
*retcmpp = NFSERR_NOTSAME;
} else if (sfp != NULL) {
sfp->sf_ffiles = fxdr_hyper(tl);
}
attrsum += NFSX_HYPER;
break;
case NFSATTRBIT_FILESTOTAL:
NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER);
if (compare) {
if (!(*retcmpp) &&
sfp->sf_tfiles != fxdr_hyper(tl))
*retcmpp = NFSERR_NOTSAME;
} else if (sfp != NULL) {
sfp->sf_tfiles = fxdr_hyper(tl);
}
attrsum += NFSX_HYPER;
break;
case NFSATTRBIT_FSLOCATIONS:
error = nfsrv_getrefstr(nd, &cp, &cp2, &l, &m);
if (error)
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
goto nfsmout;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
attrsum += l;
if (compare && !(*retcmpp)) {
refp = nfsv4root_getreferral(vp, NULL, 0);
if (refp != NULL) {
if (cp == NULL || cp2 == NULL ||
strcmp(cp, "/") ||
strcmp(cp2, refp->nfr_srvlist))
*retcmpp = NFSERR_NOTSAME;
} else if (m == 0) {
*retcmpp = NFSERR_NOTSAME;
}
}
if (cp != NULL)
free(cp, M_NFSSTRING);
if (cp2 != NULL)
free(cp2, M_NFSSTRING);
break;
case NFSATTRBIT_HIDDEN:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
if (compare && !(*retcmpp))
*retcmpp = NFSERR_ATTRNOTSUPP;
attrsum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_HOMOGENEOUS:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
if (compare) {
if (!(*retcmpp)) {
if (fsp->fs_properties &
NFSV3_FSFHOMOGENEOUS) {
if (*tl == newnfs_false)
*retcmpp = NFSERR_NOTSAME;
} else {
if (*tl == newnfs_true)
*retcmpp = NFSERR_NOTSAME;
}
}
} else if (fsp != NULL) {
if (*tl == newnfs_true)
fsp->fs_properties |= NFSV3_FSFHOMOGENEOUS;
else
fsp->fs_properties &= ~NFSV3_FSFHOMOGENEOUS;
}
attrsum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_MAXFILESIZE:
NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER);
tnfsquad.qval = fxdr_hyper(tl);
if (compare) {
if (!(*retcmpp)) {
tquad = NFSRV_MAXFILESIZE;
if (tquad != tnfsquad.qval)
*retcmpp = NFSERR_NOTSAME;
}
} else if (fsp != NULL) {
fsp->fs_maxfilesize = tnfsquad.qval;
}
attrsum += NFSX_HYPER;
break;
case NFSATTRBIT_MAXLINK:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
if (compare) {
if (!(*retcmpp)) {
if (fxdr_unsigned(int, *tl) != LINK_MAX)
*retcmpp = NFSERR_NOTSAME;
}
} else if (pc != NULL) {
pc->pc_linkmax = fxdr_unsigned(u_int32_t, *tl);
}
attrsum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_MAXNAME:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
if (compare) {
if (!(*retcmpp)) {
if (fsp->fs_maxname !=
fxdr_unsigned(u_int32_t, *tl))
*retcmpp = NFSERR_NOTSAME;
}
} else {
tuint = fxdr_unsigned(u_int32_t, *tl);
/*
* Some Linux NFSv4 servers report this
* as 0 or 4billion, so I'll set it to
* NFS_MAXNAMLEN. If a server actually creates
* a name longer than NFS_MAXNAMLEN, it will
* get an error back.
*/
if (tuint == 0 || tuint > NFS_MAXNAMLEN)
tuint = NFS_MAXNAMLEN;
if (fsp != NULL)
fsp->fs_maxname = tuint;
if (pc != NULL)
pc->pc_namemax = tuint;
}
attrsum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_MAXREAD:
NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER);
if (compare) {
if (!(*retcmpp)) {
if (fsp->fs_rtmax != fxdr_unsigned(u_int32_t,
*(tl + 1)) || *tl != 0)
*retcmpp = NFSERR_NOTSAME;
}
} else if (fsp != NULL) {
fsp->fs_rtmax = fxdr_unsigned(u_int32_t, *++tl);
fsp->fs_rtpref = fsp->fs_rtmax;
fsp->fs_dtpref = fsp->fs_rtpref;
}
attrsum += NFSX_HYPER;
break;
case NFSATTRBIT_MAXWRITE:
NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER);
if (compare) {
if (!(*retcmpp)) {
if (fsp->fs_wtmax != fxdr_unsigned(u_int32_t,
*(tl + 1)) || *tl != 0)
*retcmpp = NFSERR_NOTSAME;
}
} else if (fsp != NULL) {
fsp->fs_wtmax = fxdr_unsigned(int, *++tl);
fsp->fs_wtpref = fsp->fs_wtmax;
}
attrsum += NFSX_HYPER;
break;
case NFSATTRBIT_MIMETYPE:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
i = fxdr_unsigned(int, *tl);
attrsum += (NFSX_UNSIGNED + NFSM_RNDUP(i));
error = nfsm_advance(nd, NFSM_RNDUP(i), -1);
if (error)
goto nfsmout;
if (compare && !(*retcmpp))
*retcmpp = NFSERR_ATTRNOTSUPP;
break;
case NFSATTRBIT_MODE:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
if (compare) {
if (!(*retcmpp)) {
if (nap->na_mode != nfstov_mode(*tl))
*retcmpp = NFSERR_NOTSAME;
}
} else if (nap != NULL) {
nap->na_mode = nfstov_mode(*tl);
}
attrsum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_NOTRUNC:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
if (compare) {
if (!(*retcmpp)) {
if (*tl != newnfs_true)
*retcmpp = NFSERR_NOTSAME;
}
} else if (pc != NULL) {
pc->pc_notrunc = fxdr_unsigned(u_int32_t, *tl);
}
attrsum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_NUMLINKS:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
tuint = fxdr_unsigned(u_int32_t, *tl);
if (compare) {
if (!(*retcmpp)) {
if ((u_int32_t)nap->na_nlink != tuint)
*retcmpp = NFSERR_NOTSAME;
}
} else if (nap != NULL) {
nap->na_nlink = tuint;
}
attrsum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_OWNER:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
j = fxdr_unsigned(int, *tl);
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
if (j < 0) {
error = NFSERR_BADXDR;
goto nfsmout;
}
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
attrsum += (NFSX_UNSIGNED + NFSM_RNDUP(j));
if (j > NFSV4_SMALLSTR)
cp = malloc(j + 1, M_NFSSTRING, M_WAITOK);
else
cp = namestr;
error = nfsrv_mtostr(nd, cp, j);
if (error) {
if (j > NFSV4_SMALLSTR)
free(cp, M_NFSSTRING);
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
goto nfsmout;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
}
if (compare) {
if (!(*retcmpp)) {
Modify the NFSv4 client so that it can handle owner and owner_group strings that consist entirely of digits, interpreting them as the uid/gid number. This change was needed since new (>= 3.3) Linux servers reply with these strings by default. This change is mandated by the rfc3530bis draft. Reported on freebsd-stable@ under the Subject heading "Problem with Linux >= 3.3 as NFSv4 server" by Norbert Aschendorff on Aug. 20, 2012. Tested by: norbert.aschendorff at yahoo.de Reviewed by: jhb MFC after: 2 weeks
2012-09-20 02:49:25 +00:00
if (nfsv4_strtouid(nd, cp, j, &uid, p) ||
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
nap->na_uid != uid)
*retcmpp = NFSERR_NOTSAME;
}
} else if (nap != NULL) {
Modify the NFSv4 client so that it can handle owner and owner_group strings that consist entirely of digits, interpreting them as the uid/gid number. This change was needed since new (>= 3.3) Linux servers reply with these strings by default. This change is mandated by the rfc3530bis draft. Reported on freebsd-stable@ under the Subject heading "Problem with Linux >= 3.3 as NFSv4 server" by Norbert Aschendorff on Aug. 20, 2012. Tested by: norbert.aschendorff at yahoo.de Reviewed by: jhb MFC after: 2 weeks
2012-09-20 02:49:25 +00:00
if (nfsv4_strtouid(nd, cp, j, &uid, p))
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
nap->na_uid = nfsrv_defaultuid;
else
nap->na_uid = uid;
}
if (j > NFSV4_SMALLSTR)
free(cp, M_NFSSTRING);
break;
case NFSATTRBIT_OWNERGROUP:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
j = fxdr_unsigned(int, *tl);
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
if (j < 0) {
error = NFSERR_BADXDR;
goto nfsmout;
}
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
attrsum += (NFSX_UNSIGNED + NFSM_RNDUP(j));
if (j > NFSV4_SMALLSTR)
cp = malloc(j + 1, M_NFSSTRING, M_WAITOK);
else
cp = namestr;
error = nfsrv_mtostr(nd, cp, j);
if (error) {
if (j > NFSV4_SMALLSTR)
free(cp, M_NFSSTRING);
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
goto nfsmout;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
}
if (compare) {
if (!(*retcmpp)) {
Modify the NFSv4 client so that it can handle owner and owner_group strings that consist entirely of digits, interpreting them as the uid/gid number. This change was needed since new (>= 3.3) Linux servers reply with these strings by default. This change is mandated by the rfc3530bis draft. Reported on freebsd-stable@ under the Subject heading "Problem with Linux >= 3.3 as NFSv4 server" by Norbert Aschendorff on Aug. 20, 2012. Tested by: norbert.aschendorff at yahoo.de Reviewed by: jhb MFC after: 2 weeks
2012-09-20 02:49:25 +00:00
if (nfsv4_strtogid(nd, cp, j, &gid, p) ||
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
nap->na_gid != gid)
*retcmpp = NFSERR_NOTSAME;
}
} else if (nap != NULL) {
Modify the NFSv4 client so that it can handle owner and owner_group strings that consist entirely of digits, interpreting them as the uid/gid number. This change was needed since new (>= 3.3) Linux servers reply with these strings by default. This change is mandated by the rfc3530bis draft. Reported on freebsd-stable@ under the Subject heading "Problem with Linux >= 3.3 as NFSv4 server" by Norbert Aschendorff on Aug. 20, 2012. Tested by: norbert.aschendorff at yahoo.de Reviewed by: jhb MFC after: 2 weeks
2012-09-20 02:49:25 +00:00
if (nfsv4_strtogid(nd, cp, j, &gid, p))
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
nap->na_gid = nfsrv_defaultgid;
else
nap->na_gid = gid;
}
if (j > NFSV4_SMALLSTR)
free(cp, M_NFSSTRING);
break;
case NFSATTRBIT_QUOTAHARD:
NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER);
if (sbp != NULL) {
if (priv_check_cred(cred, PRIV_VFS_EXCEEDQUOTA, 0))
freenum = sbp->f_bfree;
else
freenum = sbp->f_bavail;
#ifdef QUOTA
/*
* ufs_quotactl() insists that the uid argument
* equal p_ruid for non-root quota access, so
* we'll just make sure that's the case.
*/
savuid = p->p_cred->p_ruid;
p->p_cred->p_ruid = cred->cr_uid;
if (!VFS_QUOTACTL(vnode_mount(vp),QCMD(Q_GETQUOTA,
Remove the thread argument from the FSD (File-System Dependent) parts of the VFS. Now all the VFS_* functions and relating parts don't want the context as long as it always refers to curthread. In some points, in particular when dealing with VOPs and functions living in the same namespace (eg. vflush) which still need to be converted, pass curthread explicitly in order to retain the old behaviour. Such loose ends will be fixed ASAP. While here fix a bug: now, UFS_EXTATTR can be compiled alone without the UFS_EXTATTR_AUTOSTART option. VFS KPI is heavilly changed by this commit so thirdy parts modules needs to be recompiled. Bump __FreeBSD_version in order to signal such situation.
2009-05-11 15:33:26 +00:00
USRQUOTA), cred->cr_uid, (caddr_t)&dqb))
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
freenum = min(dqb.dqb_bhardlimit, freenum);
p->p_cred->p_ruid = savuid;
#endif /* QUOTA */
uquad = (u_int64_t)freenum;
NFSQUOTABLKTOBYTE(uquad, sbp->f_bsize);
}
if (compare && !(*retcmpp)) {
if (uquad != fxdr_hyper(tl))
*retcmpp = NFSERR_NOTSAME;
}
attrsum += NFSX_HYPER;
break;
case NFSATTRBIT_QUOTASOFT:
NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER);
if (sbp != NULL) {
if (priv_check_cred(cred, PRIV_VFS_EXCEEDQUOTA, 0))
freenum = sbp->f_bfree;
else
freenum = sbp->f_bavail;
#ifdef QUOTA
/*
* ufs_quotactl() insists that the uid argument
* equal p_ruid for non-root quota access, so
* we'll just make sure that's the case.
*/
savuid = p->p_cred->p_ruid;
p->p_cred->p_ruid = cred->cr_uid;
if (!VFS_QUOTACTL(vnode_mount(vp),QCMD(Q_GETQUOTA,
Remove the thread argument from the FSD (File-System Dependent) parts of the VFS. Now all the VFS_* functions and relating parts don't want the context as long as it always refers to curthread. In some points, in particular when dealing with VOPs and functions living in the same namespace (eg. vflush) which still need to be converted, pass curthread explicitly in order to retain the old behaviour. Such loose ends will be fixed ASAP. While here fix a bug: now, UFS_EXTATTR can be compiled alone without the UFS_EXTATTR_AUTOSTART option. VFS KPI is heavilly changed by this commit so thirdy parts modules needs to be recompiled. Bump __FreeBSD_version in order to signal such situation.
2009-05-11 15:33:26 +00:00
USRQUOTA), cred->cr_uid, (caddr_t)&dqb))
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
freenum = min(dqb.dqb_bsoftlimit, freenum);
p->p_cred->p_ruid = savuid;
#endif /* QUOTA */
uquad = (u_int64_t)freenum;
NFSQUOTABLKTOBYTE(uquad, sbp->f_bsize);
}
if (compare && !(*retcmpp)) {
if (uquad != fxdr_hyper(tl))
*retcmpp = NFSERR_NOTSAME;
}
attrsum += NFSX_HYPER;
break;
case NFSATTRBIT_QUOTAUSED:
NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER);
if (sbp != NULL) {
freenum = 0;
#ifdef QUOTA
/*
* ufs_quotactl() insists that the uid argument
* equal p_ruid for non-root quota access, so
* we'll just make sure that's the case.
*/
savuid = p->p_cred->p_ruid;
p->p_cred->p_ruid = cred->cr_uid;
if (!VFS_QUOTACTL(vnode_mount(vp),QCMD(Q_GETQUOTA,
Remove the thread argument from the FSD (File-System Dependent) parts of the VFS. Now all the VFS_* functions and relating parts don't want the context as long as it always refers to curthread. In some points, in particular when dealing with VOPs and functions living in the same namespace (eg. vflush) which still need to be converted, pass curthread explicitly in order to retain the old behaviour. Such loose ends will be fixed ASAP. While here fix a bug: now, UFS_EXTATTR can be compiled alone without the UFS_EXTATTR_AUTOSTART option. VFS KPI is heavilly changed by this commit so thirdy parts modules needs to be recompiled. Bump __FreeBSD_version in order to signal such situation.
2009-05-11 15:33:26 +00:00
USRQUOTA), cred->cr_uid, (caddr_t)&dqb))
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
freenum = dqb.dqb_curblocks;
p->p_cred->p_ruid = savuid;
#endif /* QUOTA */
uquad = (u_int64_t)freenum;
NFSQUOTABLKTOBYTE(uquad, sbp->f_bsize);
}
if (compare && !(*retcmpp)) {
if (uquad != fxdr_hyper(tl))
*retcmpp = NFSERR_NOTSAME;
}
attrsum += NFSX_HYPER;
break;
case NFSATTRBIT_RAWDEV:
NFSM_DISSECT(tl, u_int32_t *, NFSX_V4SPECDATA);
j = fxdr_unsigned(int, *tl++);
k = fxdr_unsigned(int, *tl);
if (compare) {
if (!(*retcmpp)) {
if (nap->na_rdev != NFSMAKEDEV(j, k))
*retcmpp = NFSERR_NOTSAME;
}
} else if (nap != NULL) {
nap->na_rdev = NFSMAKEDEV(j, k);
}
attrsum += NFSX_V4SPECDATA;
break;
case NFSATTRBIT_SPACEAVAIL:
NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER);
if (compare) {
if (!(*retcmpp) &&
sfp->sf_abytes != fxdr_hyper(tl))
*retcmpp = NFSERR_NOTSAME;
} else if (sfp != NULL) {
sfp->sf_abytes = fxdr_hyper(tl);
}
attrsum += NFSX_HYPER;
break;
case NFSATTRBIT_SPACEFREE:
NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER);
if (compare) {
if (!(*retcmpp) &&
sfp->sf_fbytes != fxdr_hyper(tl))
*retcmpp = NFSERR_NOTSAME;
} else if (sfp != NULL) {
sfp->sf_fbytes = fxdr_hyper(tl);
}
attrsum += NFSX_HYPER;
break;
case NFSATTRBIT_SPACETOTAL:
NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER);
if (compare) {
if (!(*retcmpp) &&
sfp->sf_tbytes != fxdr_hyper(tl))
*retcmpp = NFSERR_NOTSAME;
} else if (sfp != NULL) {
sfp->sf_tbytes = fxdr_hyper(tl);
}
attrsum += NFSX_HYPER;
break;
case NFSATTRBIT_SPACEUSED:
NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER);
thyp = fxdr_hyper(tl);
if (compare) {
if (!(*retcmpp)) {
if ((u_int64_t)nap->na_bytes != thyp)
*retcmpp = NFSERR_NOTSAME;
}
} else if (nap != NULL) {
nap->na_bytes = thyp;
}
attrsum += NFSX_HYPER;
break;
case NFSATTRBIT_SYSTEM:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
if (compare && !(*retcmpp))
*retcmpp = NFSERR_ATTRNOTSUPP;
attrsum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_TIMEACCESS:
NFSM_DISSECT(tl, u_int32_t *, NFSX_V4TIME);
fxdr_nfsv4time(tl, &temptime);
if (compare) {
if (!(*retcmpp)) {
if (!NFS_CMPTIME(temptime, nap->na_atime))
*retcmpp = NFSERR_NOTSAME;
}
} else if (nap != NULL) {
nap->na_atime = temptime;
}
attrsum += NFSX_V4TIME;
break;
case NFSATTRBIT_TIMEACCESSSET:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
attrsum += NFSX_UNSIGNED;
i = fxdr_unsigned(int, *tl);
if (i == NFSV4SATTRTIME_TOCLIENT) {
NFSM_DISSECT(tl, u_int32_t *, NFSX_V4TIME);
attrsum += NFSX_V4TIME;
}
if (compare && !(*retcmpp))
*retcmpp = NFSERR_INVAL;
break;
case NFSATTRBIT_TIMEBACKUP:
NFSM_DISSECT(tl, u_int32_t *, NFSX_V4TIME);
if (compare && !(*retcmpp))
*retcmpp = NFSERR_ATTRNOTSUPP;
attrsum += NFSX_V4TIME;
break;
case NFSATTRBIT_TIMECREATE:
NFSM_DISSECT(tl, u_int32_t *, NFSX_V4TIME);
if (compare && !(*retcmpp))
*retcmpp = NFSERR_ATTRNOTSUPP;
attrsum += NFSX_V4TIME;
break;
case NFSATTRBIT_TIMEDELTA:
NFSM_DISSECT(tl, u_int32_t *, NFSX_V4TIME);
if (fsp != NULL) {
if (compare) {
if (!(*retcmpp)) {
if ((u_int32_t)fsp->fs_timedelta.tv_sec !=
fxdr_unsigned(u_int32_t, *(tl + 1)) ||
(u_int32_t)fsp->fs_timedelta.tv_nsec !=
(fxdr_unsigned(u_int32_t, *(tl + 2)) %
1000000000) ||
*tl != 0)
*retcmpp = NFSERR_NOTSAME;
}
} else {
fxdr_nfsv4time(tl, &fsp->fs_timedelta);
}
}
attrsum += NFSX_V4TIME;
break;
case NFSATTRBIT_TIMEMETADATA:
NFSM_DISSECT(tl, u_int32_t *, NFSX_V4TIME);
fxdr_nfsv4time(tl, &temptime);
if (compare) {
if (!(*retcmpp)) {
if (!NFS_CMPTIME(temptime, nap->na_ctime))
*retcmpp = NFSERR_NOTSAME;
}
} else if (nap != NULL) {
nap->na_ctime = temptime;
}
attrsum += NFSX_V4TIME;
break;
case NFSATTRBIT_TIMEMODIFY:
NFSM_DISSECT(tl, u_int32_t *, NFSX_V4TIME);
fxdr_nfsv4time(tl, &temptime);
if (compare) {
if (!(*retcmpp)) {
if (!NFS_CMPTIME(temptime, nap->na_mtime))
*retcmpp = NFSERR_NOTSAME;
}
} else if (nap != NULL) {
nap->na_mtime = temptime;
}
attrsum += NFSX_V4TIME;
break;
case NFSATTRBIT_TIMEMODIFYSET:
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
attrsum += NFSX_UNSIGNED;
i = fxdr_unsigned(int, *tl);
if (i == NFSV4SATTRTIME_TOCLIENT) {
NFSM_DISSECT(tl, u_int32_t *, NFSX_V4TIME);
attrsum += NFSX_V4TIME;
}
if (compare && !(*retcmpp))
*retcmpp = NFSERR_INVAL;
break;
case NFSATTRBIT_MOUNTEDONFILEID:
NFSM_DISSECT(tl, u_int32_t *, NFSX_HYPER);
thyp = fxdr_hyper(tl);
if (compare) {
if (!(*retcmpp)) {
if (*tl++) {
*retcmpp = NFSERR_NOTSAME;
} else {
if (!vp || !nfsrv_atroot(vp, &fid))
fid = nap->na_fileid;
if ((u_int64_t)fid != thyp)
*retcmpp = NFSERR_NOTSAME;
}
}
} else if (nap != NULL) {
if (*tl++)
printf("NFSv4 mounted on fileid > 32bits\n");
nap->na_mntonfileno = thyp;
}
attrsum += NFSX_HYPER;
break;
Merge the NFSv4.1 server code in projects/nfsv4.1-server over into head. The code is not believed to have any effect on the semantics of non-NFSv4.1 server behaviour. It is a rather large merge, but I am hoping that there will not be any regressions for the NFS server. MFC after: 1 month
2014-07-01 20:47:16 +00:00
case NFSATTRBIT_SUPPATTREXCLCREAT:
retnotsup = 0;
error = nfsrv_getattrbits(nd, &retattrbits,
&cnt, &retnotsup);
if (error)
goto nfsmout;
if (compare && !(*retcmpp)) {
NFSSETSUPP_ATTRBIT(&checkattrbits);
NFSCLRNOTSETABLE_ATTRBIT(&checkattrbits);
NFSCLRBIT_ATTRBIT(&checkattrbits,
NFSATTRBIT_TIMEACCESSSET);
if (!NFSEQUAL_ATTRBIT(&retattrbits, &checkattrbits)
|| retnotsup)
*retcmpp = NFSERR_NOTSAME;
}
attrsum += cnt;
break;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
default:
printf("EEK! nfsv4_loadattr unknown attr=%d\n",
bitpos);
if (compare && !(*retcmpp))
*retcmpp = NFSERR_ATTRNOTSUPP;
/*
* and get out of the loop, since we can't parse
* the unknown attrbute data.
*/
bitpos = NFSATTRBIT_MAX;
break;
};
}
/*
* some clients pad the attrlist, so we need to skip over the
* padding.
*/
if (attrsum > attrsize) {
error = NFSERR_BADXDR;
} else {
attrsize = NFSM_RNDUP(attrsize);
if (attrsum < attrsize)
error = nfsm_advance(nd, attrsize - attrsum, -1);
}
nfsmout:
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
NFSEXITCODE2(error, nd);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
return (error);
}
/*
* Implement sleep locks for newnfs. The nfslock_usecnt allows for a
* shared lock and the NFSXXX_LOCK flag permits an exclusive lock.
* The first argument is a pointer to an nfsv4lock structure.
* The second argument is 1 iff a blocking lock is wanted.
* If this argument is 0, the call waits until no thread either wants nor
* holds an exclusive lock.
* It returns 1 if the lock was acquired, 0 otherwise.
* If several processes call this function concurrently wanting the exclusive
* lock, one will get the lock and the rest will return without getting the
* lock. (If the caller must have the lock, it simply calls this function in a
* loop until the function returns 1 to indicate the lock was acquired.)
* Any usecnt must be decremented by calling nfsv4_relref() before
* calling nfsv4_lock(). It was done this way, so nfsv4_lock() could
* be called in a loop.
Fix the new NFS client so that it handles NFSv4 state correctly during a forced dismount. This required that the exclusive and shared (refcnt) sleep lock functions check for MNTK_UMOUNTF before sleeping, so that they won't block while nfscl_umount() is getting rid of the state. As such, a "struct mount *" argument was added to the locking functions. I believe the only remaining case where a forced dismount can get hung in the kernel is when a thread is already attempting to do a TCP connect to a dead server when the krpc client structure called nr_client is NULL. This will only happen just after a "mount -u" with options that force a new TCP connection is done, so it shouldn't be a problem in practice. MFC after: 2 weeks
2011-05-27 22:05:10 +00:00
* The isleptp argument is set to indicate if the call slept, iff not NULL
* and the mp argument indicates to check for a forced dismount, iff not
* NULL.
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
*/
APPLESTATIC int
nfsv4_lock(struct nfsv4lock *lp, int iwantlock, int *isleptp,
Fix the new NFS client so that it handles NFSv4 state correctly during a forced dismount. This required that the exclusive and shared (refcnt) sleep lock functions check for MNTK_UMOUNTF before sleeping, so that they won't block while nfscl_umount() is getting rid of the state. As such, a "struct mount *" argument was added to the locking functions. I believe the only remaining case where a forced dismount can get hung in the kernel is when a thread is already attempting to do a TCP connect to a dead server when the krpc client structure called nr_client is NULL. This will only happen just after a "mount -u" with options that force a new TCP connection is done, so it shouldn't be a problem in practice. MFC after: 2 weeks
2011-05-27 22:05:10 +00:00
void *mutex, struct mount *mp)
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
{
if (isleptp)
*isleptp = 0;
/*
* If a lock is wanted, loop around until the lock is acquired by
* someone and then released. If I want the lock, try to acquire it.
* For a lock to be issued, no lock must be in force and the usecnt
* must be zero.
*/
if (iwantlock) {
if (!(lp->nfslock_lock & NFSV4LOCK_LOCK) &&
lp->nfslock_usecnt == 0) {
lp->nfslock_lock &= ~NFSV4LOCK_LOCKWANTED;
lp->nfslock_lock |= NFSV4LOCK_LOCK;
return (1);
}
lp->nfslock_lock |= NFSV4LOCK_LOCKWANTED;
}
while (lp->nfslock_lock & (NFSV4LOCK_LOCK | NFSV4LOCK_LOCKWANTED)) {
Fix the new NFS client so that it handles NFSv4 state correctly during a forced dismount. This required that the exclusive and shared (refcnt) sleep lock functions check for MNTK_UMOUNTF before sleeping, so that they won't block while nfscl_umount() is getting rid of the state. As such, a "struct mount *" argument was added to the locking functions. I believe the only remaining case where a forced dismount can get hung in the kernel is when a thread is already attempting to do a TCP connect to a dead server when the krpc client structure called nr_client is NULL. This will only happen just after a "mount -u" with options that force a new TCP connection is done, so it shouldn't be a problem in practice. MFC after: 2 weeks
2011-05-27 22:05:10 +00:00
if (mp != NULL && (mp->mnt_kern_flag & MNTK_UNMOUNTF) != 0) {
lp->nfslock_lock &= ~NFSV4LOCK_LOCKWANTED;
return (0);
}
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
lp->nfslock_lock |= NFSV4LOCK_WANTED;
if (isleptp)
*isleptp = 1;
(void) nfsmsleep(&lp->nfslock_lock, mutex,
PZERO - 1, "nfsv4lck", NULL);
if (iwantlock && !(lp->nfslock_lock & NFSV4LOCK_LOCK) &&
lp->nfslock_usecnt == 0) {
lp->nfslock_lock &= ~NFSV4LOCK_LOCKWANTED;
lp->nfslock_lock |= NFSV4LOCK_LOCK;
return (1);
}
}
return (0);
}
/*
* Release the lock acquired by nfsv4_lock().
* The second argument is set to 1 to indicate the nfslock_usecnt should be
* incremented, as well.
*/
APPLESTATIC void
nfsv4_unlock(struct nfsv4lock *lp, int incref)
{
lp->nfslock_lock &= ~NFSV4LOCK_LOCK;
if (incref)
lp->nfslock_usecnt++;
nfsv4_wanted(lp);
}
/*
* Release a reference cnt.
*/
APPLESTATIC void
nfsv4_relref(struct nfsv4lock *lp)
{
if (lp->nfslock_usecnt <= 0)
panic("nfsv4root ref cnt");
lp->nfslock_usecnt--;
if (lp->nfslock_usecnt == 0)
nfsv4_wanted(lp);
}
/*
* Get a reference cnt.
* This function will wait for any exclusive lock to be released, but will
* not wait for threads that want the exclusive lock. If priority needs
* to be given to threads that need the exclusive lock, a call to nfsv4_lock()
* with the 2nd argument == 0 should be done before calling nfsv4_getref().
Fix the new NFS client so that it handles NFSv4 state correctly during a forced dismount. This required that the exclusive and shared (refcnt) sleep lock functions check for MNTK_UMOUNTF before sleeping, so that they won't block while nfscl_umount() is getting rid of the state. As such, a "struct mount *" argument was added to the locking functions. I believe the only remaining case where a forced dismount can get hung in the kernel is when a thread is already attempting to do a TCP connect to a dead server when the krpc client structure called nr_client is NULL. This will only happen just after a "mount -u" with options that force a new TCP connection is done, so it shouldn't be a problem in practice. MFC after: 2 weeks
2011-05-27 22:05:10 +00:00
* If the mp argument is not NULL, check for MNTK_UNMOUNTF being set and
* return without getting a refcnt for that case.
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
*/
APPLESTATIC void
Fix the new NFS client so that it handles NFSv4 state correctly during a forced dismount. This required that the exclusive and shared (refcnt) sleep lock functions check for MNTK_UMOUNTF before sleeping, so that they won't block while nfscl_umount() is getting rid of the state. As such, a "struct mount *" argument was added to the locking functions. I believe the only remaining case where a forced dismount can get hung in the kernel is when a thread is already attempting to do a TCP connect to a dead server when the krpc client structure called nr_client is NULL. This will only happen just after a "mount -u" with options that force a new TCP connection is done, so it shouldn't be a problem in practice. MFC after: 2 weeks
2011-05-27 22:05:10 +00:00
nfsv4_getref(struct nfsv4lock *lp, int *isleptp, void *mutex,
struct mount *mp)
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
{
if (isleptp)
*isleptp = 0;
/*
* Wait for a lock held.
*/
while (lp->nfslock_lock & NFSV4LOCK_LOCK) {
Fix the new NFS client so that it handles NFSv4 state correctly during a forced dismount. This required that the exclusive and shared (refcnt) sleep lock functions check for MNTK_UMOUNTF before sleeping, so that they won't block while nfscl_umount() is getting rid of the state. As such, a "struct mount *" argument was added to the locking functions. I believe the only remaining case where a forced dismount can get hung in the kernel is when a thread is already attempting to do a TCP connect to a dead server when the krpc client structure called nr_client is NULL. This will only happen just after a "mount -u" with options that force a new TCP connection is done, so it shouldn't be a problem in practice. MFC after: 2 weeks
2011-05-27 22:05:10 +00:00
if (mp != NULL && (mp->mnt_kern_flag & MNTK_UNMOUNTF) != 0)
return;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
lp->nfslock_lock |= NFSV4LOCK_WANTED;
if (isleptp)
*isleptp = 1;
(void) nfsmsleep(&lp->nfslock_lock, mutex,
Move the NFSv4.1 client patches over from projects/nfsv4.1-client to head. I don't think the NFS client behaviour will change unless the new "minorversion=1" mount option is used. It includes basic NFSv4.1 support plus support for pNFS using the Files Layout only. All problems detecting during an NFSv4.1 Bakeathon testing event in June 2012 have been resolved in this code and it has been tested against the NFSv4.1 server available to me. Although not reviewed, I believe that kib@ has looked at it.
2012-12-08 22:52:39 +00:00
PZERO - 1, "nfsv4gr", NULL);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
}
Fix the new NFS client so that it handles NFSv4 state correctly during a forced dismount. This required that the exclusive and shared (refcnt) sleep lock functions check for MNTK_UMOUNTF before sleeping, so that they won't block while nfscl_umount() is getting rid of the state. As such, a "struct mount *" argument was added to the locking functions. I believe the only remaining case where a forced dismount can get hung in the kernel is when a thread is already attempting to do a TCP connect to a dead server when the krpc client structure called nr_client is NULL. This will only happen just after a "mount -u" with options that force a new TCP connection is done, so it shouldn't be a problem in practice. MFC after: 2 weeks
2011-05-27 22:05:10 +00:00
if (mp != NULL && (mp->mnt_kern_flag & MNTK_UNMOUNTF) != 0)
return;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
lp->nfslock_usecnt++;
}
The timer routine in the experimental NFS server did not acquire the correct mutex when checking nfsv4root_lock. Although this could be fixed by adding mutex lock/unlock calls, zack.kirsch at isilon.com suggested a better fix that uses a non-blocking acquisition of a reference count on nfsv4root_lock. This fix allows the weird NFSLOCKSTATE(); NFSUNLOCKSTATE(); synchronization to be deleted. This patch applies this fix. Tested by: zack.kirsch at isilon.com MFC after: 2 weeks
2010-08-28 21:41:18 +00:00
/*
* Get a reference as above, but return failure instead of sleeping if
* an exclusive lock is held.
*/
APPLESTATIC int
nfsv4_getref_nonblock(struct nfsv4lock *lp)
{
if ((lp->nfslock_lock & NFSV4LOCK_LOCK) != 0)
return (0);
lp->nfslock_usecnt++;
return (1);
}
This patch should fix handling of byte range locks locally on the server for the experimental nfs server. When enabled by setting vfs.newnfs.locallocks_enable to non-zero, the experimental nfs server will now acquire byte range locks on the file on behalf of NFSv4 clients, such that lock conflicts between the NFSv4 clients and processes running locally on the server, will be recognized and handled correctly. MFC after: 2 weeks
2010-03-30 23:11:50 +00:00
/*
* Test for a lock. Return 1 if locked, 0 otherwise.
*/
APPLESTATIC int
nfsv4_testlock(struct nfsv4lock *lp)
{
if ((lp->nfslock_lock & NFSV4LOCK_LOCK) == 0 &&
lp->nfslock_usecnt == 0)
return (0);
return (1);
}
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
/*
* Wake up anyone sleeping, waiting for this lock.
*/
static void
nfsv4_wanted(struct nfsv4lock *lp)
{
if (lp->nfslock_lock & NFSV4LOCK_WANTED) {
lp->nfslock_lock &= ~NFSV4LOCK_WANTED;
wakeup((caddr_t)&lp->nfslock_lock);
}
}
/*
* Copy a string from an mbuf list into a character array.
* Return EBADRPC if there is an mbuf error,
* 0 otherwise.
*/
APPLESTATIC int
nfsrv_mtostr(struct nfsrv_descript *nd, char *str, int siz)
{
char *cp;
int xfer, len;
mbuf_t mp;
int rem, error = 0;
mp = nd->nd_md;
cp = nd->nd_dpos;
len = NFSMTOD(mp, caddr_t) + mbuf_len(mp) - cp;
rem = NFSM_RNDUP(siz) - siz;
while (siz > 0) {
if (len > siz)
xfer = siz;
else
xfer = len;
NFSBCOPY(cp, str, xfer);
str += xfer;
siz -= xfer;
if (siz > 0) {
mp = mbuf_next(mp);
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
if (mp == NULL) {
error = EBADRPC;
goto out;
}
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
cp = NFSMTOD(mp, caddr_t);
len = mbuf_len(mp);
} else {
cp += xfer;
len -= xfer;
}
}
*str = '\0';
nd->nd_dpos = cp;
nd->nd_md = mp;
if (rem > 0) {
if (len < rem)
error = nfsm_advance(nd, rem, len);
else
nd->nd_dpos += rem;
}
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
out:
NFSEXITCODE2(error, nd);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
return (error);
}
/*
* Fill in the attributes as marked by the bitmap (V4).
*/
APPLESTATIC int
Modify the experimental NFSv4 server so that it handles crossing of server mount points properly. The functions nfsvno_fillattr() and nfsv4_fillattr() were modified to take the extra arguments that are the mount point, a flag to indicate that it is a file system root and the mounted on fileno. The mount point argument needs to be busy when nfsvno_fillattr() is called, since the vp argument is not locked. Reviewed by: kib MFC after: 2 weeks
2011-04-14 21:49:52 +00:00
nfsv4_fillattr(struct nfsrv_descript *nd, struct mount *mp, vnode_t vp,
NFSACL_T *saclp, struct vattr *vap, fhandle_t *fhp, int rderror,
nfsattrbit_t *attrbitp, struct ucred *cred, NFSPROC_T *p, int isdgram,
Fix the experimental NFSv4 server so that it uses VOP_PATHCONF() to determine if a file system supports NFSv4 ACLs. Since VOP_PATHCONF() must be called with a locked vnode, the function is called before nfsvno_fillattr() and the result is passed in as an extra argument. MFC after: 2 weeks
2011-04-14 23:46:15 +00:00
int reterr, int supports_nfsv4acls, int at_root, uint64_t mounted_on_fileno)
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
{
int bitpos, retnum = 0;
u_int32_t *tl;
int siz, prefixnum, error;
u_char *cp, namestr[NFSV4_SMALLSTR];
nfsattrbit_t attrbits, retbits;
nfsattrbit_t *retbitp = &retbits;
u_int32_t freenum, *retnump;
u_int64_t uquad;
struct statfs fs;
struct nfsfsinfo fsinf;
struct timespec temptime;
NFSACL_T *aclp, *naclp = NULL;
#ifdef QUOTA
struct dqblk dqb;
uid_t savuid;
#endif
/*
* First, set the bits that can be filled and get fsinfo.
*/
NFSSET_ATTRBIT(retbitp, attrbitp);
The NFSv4 client was passing both the p and cred arguments to nfsv4_fillattr() as NULLs for the Getattr callback. This caused nfsv4_fillattr() to not fill in the Change attribute for the reply. I believe this was a violation of the RFC, but had little effect on server behaviour. This patch passes a non-NULL p argument to fix this. MFC after: 1 week
2013-12-24 00:48:39 +00:00
/*
* If both p and cred are NULL, it is a client side setattr call.
* If both p and cred are not NULL, it is a server side reply call.
* If p is not NULL and cred is NULL, it is a client side callback
* reply call.
*/
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
if (p == NULL && cred == NULL) {
NFSCLRNOTSETABLE_ATTRBIT(retbitp);
aclp = saclp;
} else {
NFSCLRNOTFILLABLE_ATTRBIT(retbitp);
Fix the experimental nfs subsystem so that it builds with the current NFSv4 ACLs, as defined in sys/acl.h. It still needs a way to test a mount point for NFSv4 ACL support before it will work. Until then, the NFSHASNFS4ACL() macro just always returns 0. Approved by: kib (mentor)
2009-05-26 22:21:53 +00:00
naclp = acl_alloc(M_WAITOK);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
aclp = naclp;
}
nfsvno_getfs(&fsinf, isdgram);
#ifndef APPLE
/*
* Get the VFS_STATFS(), since some attributes need them.
*/
if (NFSISSETSTATFS_ATTRBIT(retbitp)) {
Modify the experimental NFSv4 server so that it handles crossing of server mount points properly. The functions nfsvno_fillattr() and nfsv4_fillattr() were modified to take the extra arguments that are the mount point, a flag to indicate that it is a file system root and the mounted on fileno. The mount point argument needs to be busy when nfsvno_fillattr() is called, since the vp argument is not locked. Reviewed by: kib MFC after: 2 weeks
2011-04-14 21:49:52 +00:00
error = VFS_STATFS(mp, &fs);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
if (error != 0) {
if (reterr) {
nd->nd_repstat = NFSERR_ACCES;
return (0);
}
NFSCLRSTATFS_ATTRBIT(retbitp);
}
}
#endif
/*
* And the NFSv4 ACL...
*/
Remove unneeded ifdefs. Reviewed by: rmacklem
2009-12-03 18:03:42 +00:00
if (NFSISSET_ATTRBIT(retbitp, NFSATTRBIT_ACLSUPPORT) &&
(nfsrv_useacl == 0 || ((cred != NULL || p != NULL) &&
Fix the experimental NFSv4 server so that it uses VOP_PATHCONF() to determine if a file system supports NFSv4 ACLs. Since VOP_PATHCONF() must be called with a locked vnode, the function is called before nfsvno_fillattr() and the result is passed in as an extra argument. MFC after: 2 weeks
2011-04-14 23:46:15 +00:00
supports_nfsv4acls == 0))) {
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
NFSCLRBIT_ATTRBIT(retbitp, NFSATTRBIT_ACLSUPPORT);
}
if (NFSISSET_ATTRBIT(retbitp, NFSATTRBIT_ACL)) {
if (nfsrv_useacl == 0 || ((cred != NULL || p != NULL) &&
Fix the experimental NFSv4 server so that it uses VOP_PATHCONF() to determine if a file system supports NFSv4 ACLs. Since VOP_PATHCONF() must be called with a locked vnode, the function is called before nfsvno_fillattr() and the result is passed in as an extra argument. MFC after: 2 weeks
2011-04-14 23:46:15 +00:00
supports_nfsv4acls == 0)) {
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
NFSCLRBIT_ATTRBIT(retbitp, NFSATTRBIT_ACL);
} else if (naclp != NULL) {
Simple find/replace of vn_lock -> NFSVOPLOCK. This is done so that NFSVOPLOCK can be modified later to add enhanced logging and assertions. Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:05:31 +00:00
if (NFSVOPLOCK(vp, LK_SHARED) == 0) {
Fix the experimental NFSv4 server so that it uses VOP_ACCESSX() to check for VREAD_ACL instead of VOP_ACCESS(). MFC after: 3 days
2011-01-18 14:34:45 +00:00
error = VOP_ACCESSX(vp, VREAD_ACL, cred, p);
Modify the experimental NFS server so that it uses LK_SHARED for RPC operations when it can. Since VFS_FHTOVP() currently always gets an exclusively locked vnode and is usually called at the beginning of each RPC, the RPCs for a given vnode will still be serialized. As such, passing a lock type argument to VFS_FHTOVP() would be preferable to doing the vn_lock() with LK_DOWNGRADE after the VFS_FHTOVP() call. Reviewed by: kib MFC after: 2 weeks
2010-12-25 21:56:25 +00:00
if (error == 0)
error = VOP_GETACL(vp, ACL_TYPE_NFS4,
naclp, cred, p);
Simple find/replace of VOP_UNLOCK -> NFSVOPUNLOCK. This is done so that NFSVOPUNLOCK can be modified later to add enhanced logging and assertions. Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:05:36 +00:00
NFSVOPUNLOCK(vp, 0);
Modify the experimental NFS server so that it uses LK_SHARED for RPC operations when it can. Since VFS_FHTOVP() currently always gets an exclusively locked vnode and is usually called at the beginning of each RPC, the RPCs for a given vnode will still be serialized. As such, passing a lock type argument to VFS_FHTOVP() would be preferable to doing the vn_lock() with LK_DOWNGRADE after the VFS_FHTOVP() call. Reviewed by: kib MFC after: 2 weeks
2010-12-25 21:56:25 +00:00
} else
error = NFSERR_PERM;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
if (error != 0) {
if (reterr) {
nd->nd_repstat = NFSERR_ACCES;
return (0);
}
NFSCLRBIT_ATTRBIT(retbitp, NFSATTRBIT_ACL);
}
}
}
Fix failures and warnings reported by newpynfs20090424 test tool. This fix addresses only issues with the pynfs reports, none of these issues are know to create problems for extant real clients. Submitted by: Bart Hsiao <bart.hsiao@gmail.com> Reworked by: myself Reviewed by: rmacklem Approved by: rmacklem Sponsored by: QNAP Systems Inc.
2014-10-03 02:24:41 +00:00
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
/*
* Put out the attribute bitmap for the ones being filled in
* and get the field for the number of attributes returned.
*/
prefixnum = nfsrv_putattrbit(nd, retbitp);
NFSM_BUILD(retnump, u_int32_t *, NFSX_UNSIGNED);
prefixnum += NFSX_UNSIGNED;
/*
* Now, loop around filling in the attributes for each bit set.
*/
for (bitpos = 0; bitpos < NFSATTRBIT_MAX; bitpos++) {
if (NFSISSET_ATTRBIT(retbitp, bitpos)) {
switch (bitpos) {
case NFSATTRBIT_SUPPORTEDATTRS:
NFSSETSUPP_ATTRBIT(&attrbits);
if (nfsrv_useacl == 0 || ((cred != NULL || p != NULL)
Fix the experimental NFSv4 server so that it uses VOP_PATHCONF() to determine if a file system supports NFSv4 ACLs. Since VOP_PATHCONF() must be called with a locked vnode, the function is called before nfsvno_fillattr() and the result is passed in as an extra argument. MFC after: 2 weeks
2011-04-14 23:46:15 +00:00
&& supports_nfsv4acls == 0)) {
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
NFSCLRBIT_ATTRBIT(&attrbits,NFSATTRBIT_ACLSUPPORT);
NFSCLRBIT_ATTRBIT(&attrbits,NFSATTRBIT_ACL);
}
retnum += nfsrv_putattrbit(nd, &attrbits);
break;
case NFSATTRBIT_TYPE:
NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED);
*tl = vtonfsv34_type(vap->va_type);
retnum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_FHEXPIRETYPE:
NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED);
*tl = txdr_unsigned(NFSV4FHTYPE_PERSISTENT);
retnum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_CHANGE:
NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER);
txdr_hyper(vap->va_filerev, tl);
retnum += NFSX_HYPER;
break;
case NFSATTRBIT_SIZE:
NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER);
txdr_hyper(vap->va_size, tl);
retnum += NFSX_HYPER;
break;
case NFSATTRBIT_LINKSUPPORT:
NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED);
if (fsinf.fs_properties & NFSV3FSINFO_LINK)
*tl = newnfs_true;
else
*tl = newnfs_false;
retnum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_SYMLINKSUPPORT:
NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED);
if (fsinf.fs_properties & NFSV3FSINFO_SYMLINK)
*tl = newnfs_true;
else
*tl = newnfs_false;
retnum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_NAMEDATTR:
NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED);
*tl = newnfs_false;
retnum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_FSID:
NFSM_BUILD(tl, u_int32_t *, NFSX_V4FSID);
*tl++ = 0;
Modify the experimental NFSv4 server so that it handles crossing of server mount points properly. The functions nfsvno_fillattr() and nfsv4_fillattr() were modified to take the extra arguments that are the mount point, a flag to indicate that it is a file system root and the mounted on fileno. The mount point argument needs to be busy when nfsvno_fillattr() is called, since the vp argument is not locked. Reviewed by: kib MFC after: 2 weeks
2011-04-14 21:49:52 +00:00
*tl++ = txdr_unsigned(mp->mnt_stat.f_fsid.val[0]);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
*tl++ = 0;
Modify the experimental NFSv4 server so that it handles crossing of server mount points properly. The functions nfsvno_fillattr() and nfsv4_fillattr() were modified to take the extra arguments that are the mount point, a flag to indicate that it is a file system root and the mounted on fileno. The mount point argument needs to be busy when nfsvno_fillattr() is called, since the vp argument is not locked. Reviewed by: kib MFC after: 2 weeks
2011-04-14 21:49:52 +00:00
*tl = txdr_unsigned(mp->mnt_stat.f_fsid.val[1]);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
retnum += NFSX_V4FSID;
break;
case NFSATTRBIT_UNIQUEHANDLES:
NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED);
*tl = newnfs_true;
retnum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_LEASETIME:
NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED);
*tl = txdr_unsigned(nfsrv_lease);
retnum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_RDATTRERROR:
NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED);
*tl = txdr_unsigned(rderror);
retnum += NFSX_UNSIGNED;
break;
/*
* Recommended Attributes. (Only the supported ones.)
*/
case NFSATTRBIT_ACL:
retnum += nfsrv_buildacl(nd, aclp, vnode_vtype(vp), p);
break;
case NFSATTRBIT_ACLSUPPORT:
NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED);
*tl = txdr_unsigned(NFSV4ACE_SUPTYPES);
retnum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_CANSETTIME:
NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED);
if (fsinf.fs_properties & NFSV3FSINFO_CANSETTIME)
*tl = newnfs_true;
else
*tl = newnfs_false;
retnum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_CASEINSENSITIVE:
NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED);
*tl = newnfs_false;
retnum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_CASEPRESERVING:
NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED);
*tl = newnfs_true;
retnum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_CHOWNRESTRICTED:
NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED);
Revert revision 224079 as Rick pointed out that I would be calling VOP_PATHCONF without the vnode lock held. Implicitly approved by: zml (mentor)
2011-07-17 03:44:05 +00:00
*tl = newnfs_true;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
retnum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_FILEHANDLE:
retnum += nfsm_fhtom(nd, (u_int8_t *)fhp, 0, 0);
break;
case NFSATTRBIT_FILEID:
NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER);
*tl++ = 0;
*tl = txdr_unsigned(vap->va_fileid);
retnum += NFSX_HYPER;
break;
case NFSATTRBIT_FILESAVAIL:
/*
* Check quota and use min(quota, f_ffree).
*/
freenum = fs.f_ffree;
#ifdef QUOTA
/*
* ufs_quotactl() insists that the uid argument
* equal p_ruid for non-root quota access, so
* we'll just make sure that's the case.
*/
savuid = p->p_cred->p_ruid;
p->p_cred->p_ruid = cred->cr_uid;
Modify the experimental NFSv4 server so that it handles crossing of server mount points properly. The functions nfsvno_fillattr() and nfsv4_fillattr() were modified to take the extra arguments that are the mount point, a flag to indicate that it is a file system root and the mounted on fileno. The mount point argument needs to be busy when nfsvno_fillattr() is called, since the vp argument is not locked. Reviewed by: kib MFC after: 2 weeks
2011-04-14 21:49:52 +00:00
if (!VFS_QUOTACTL(mp, QCMD(Q_GETQUOTA,USRQUOTA),
Remove the thread argument from the FSD (File-System Dependent) parts of the VFS. Now all the VFS_* functions and relating parts don't want the context as long as it always refers to curthread. In some points, in particular when dealing with VOPs and functions living in the same namespace (eg. vflush) which still need to be converted, pass curthread explicitly in order to retain the old behaviour. Such loose ends will be fixed ASAP. While here fix a bug: now, UFS_EXTATTR can be compiled alone without the UFS_EXTATTR_AUTOSTART option. VFS KPI is heavilly changed by this commit so thirdy parts modules needs to be recompiled. Bump __FreeBSD_version in order to signal such situation.
2009-05-11 15:33:26 +00:00
cred->cr_uid, (caddr_t)&dqb))
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
freenum = min(dqb.dqb_isoftlimit-dqb.dqb_curinodes,
freenum);
p->p_cred->p_ruid = savuid;
#endif /* QUOTA */
NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER);
*tl++ = 0;
*tl = txdr_unsigned(freenum);
retnum += NFSX_HYPER;
break;
case NFSATTRBIT_FILESFREE:
NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER);
*tl++ = 0;
*tl = txdr_unsigned(fs.f_ffree);
retnum += NFSX_HYPER;
break;
case NFSATTRBIT_FILESTOTAL:
NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER);
*tl++ = 0;
*tl = txdr_unsigned(fs.f_files);
retnum += NFSX_HYPER;
break;
case NFSATTRBIT_FSLOCATIONS:
NFSM_BUILD(tl, u_int32_t *, 2 * NFSX_UNSIGNED);
*tl++ = 0;
*tl = 0;
retnum += 2 * NFSX_UNSIGNED;
break;
case NFSATTRBIT_HOMOGENEOUS:
NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED);
if (fsinf.fs_properties & NFSV3FSINFO_HOMOGENEOUS)
*tl = newnfs_true;
else
*tl = newnfs_false;
retnum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_MAXFILESIZE:
NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER);
uquad = NFSRV_MAXFILESIZE;
txdr_hyper(uquad, tl);
retnum += NFSX_HYPER;
break;
case NFSATTRBIT_MAXLINK:
NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED);
*tl = txdr_unsigned(LINK_MAX);
retnum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_MAXNAME:
NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED);
*tl = txdr_unsigned(NFS_MAXNAMLEN);
retnum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_MAXREAD:
NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER);
*tl++ = 0;
*tl = txdr_unsigned(fsinf.fs_rtmax);
retnum += NFSX_HYPER;
break;
case NFSATTRBIT_MAXWRITE:
NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER);
*tl++ = 0;
*tl = txdr_unsigned(fsinf.fs_wtmax);
retnum += NFSX_HYPER;
break;
case NFSATTRBIT_MODE:
NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED);
*tl = vtonfsv34_mode(vap->va_mode);
retnum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_NOTRUNC:
NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED);
*tl = newnfs_true;
retnum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_NUMLINKS:
NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED);
*tl = txdr_unsigned(vap->va_nlink);
retnum += NFSX_UNSIGNED;
break;
case NFSATTRBIT_OWNER:
cp = namestr;
nfsv4_uidtostr(vap->va_uid, &cp, &siz, p);
retnum += nfsm_strtom(nd, cp, siz);
if (cp != namestr)
free(cp, M_NFSSTRING);
break;
case NFSATTRBIT_OWNERGROUP:
cp = namestr;
nfsv4_gidtostr(vap->va_gid, &cp, &siz, p);
retnum += nfsm_strtom(nd, cp, siz);
if (cp != namestr)
free(cp, M_NFSSTRING);
break;
case NFSATTRBIT_QUOTAHARD:
if (priv_check_cred(cred, PRIV_VFS_EXCEEDQUOTA, 0))
freenum = fs.f_bfree;
else
freenum = fs.f_bavail;
#ifdef QUOTA
/*
* ufs_quotactl() insists that the uid argument
* equal p_ruid for non-root quota access, so
* we'll just make sure that's the case.
*/
savuid = p->p_cred->p_ruid;
p->p_cred->p_ruid = cred->cr_uid;
Modify the experimental NFSv4 server so that it handles crossing of server mount points properly. The functions nfsvno_fillattr() and nfsv4_fillattr() were modified to take the extra arguments that are the mount point, a flag to indicate that it is a file system root and the mounted on fileno. The mount point argument needs to be busy when nfsvno_fillattr() is called, since the vp argument is not locked. Reviewed by: kib MFC after: 2 weeks
2011-04-14 21:49:52 +00:00
if (!VFS_QUOTACTL(mp, QCMD(Q_GETQUOTA,USRQUOTA),
Remove the thread argument from the FSD (File-System Dependent) parts of the VFS. Now all the VFS_* functions and relating parts don't want the context as long as it always refers to curthread. In some points, in particular when dealing with VOPs and functions living in the same namespace (eg. vflush) which still need to be converted, pass curthread explicitly in order to retain the old behaviour. Such loose ends will be fixed ASAP. While here fix a bug: now, UFS_EXTATTR can be compiled alone without the UFS_EXTATTR_AUTOSTART option. VFS KPI is heavilly changed by this commit so thirdy parts modules needs to be recompiled. Bump __FreeBSD_version in order to signal such situation.
2009-05-11 15:33:26 +00:00
cred->cr_uid, (caddr_t)&dqb))
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
freenum = min(dqb.dqb_bhardlimit, freenum);
p->p_cred->p_ruid = savuid;
#endif /* QUOTA */
NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER);
uquad = (u_int64_t)freenum;
NFSQUOTABLKTOBYTE(uquad, fs.f_bsize);
txdr_hyper(uquad, tl);
retnum += NFSX_HYPER;
break;
case NFSATTRBIT_QUOTASOFT:
if (priv_check_cred(cred, PRIV_VFS_EXCEEDQUOTA, 0))
freenum = fs.f_bfree;
else
freenum = fs.f_bavail;
#ifdef QUOTA
/*
* ufs_quotactl() insists that the uid argument
* equal p_ruid for non-root quota access, so
* we'll just make sure that's the case.
*/
savuid = p->p_cred->p_ruid;
p->p_cred->p_ruid = cred->cr_uid;
Modify the experimental NFSv4 server so that it handles crossing of server mount points properly. The functions nfsvno_fillattr() and nfsv4_fillattr() were modified to take the extra arguments that are the mount point, a flag to indicate that it is a file system root and the mounted on fileno. The mount point argument needs to be busy when nfsvno_fillattr() is called, since the vp argument is not locked. Reviewed by: kib MFC after: 2 weeks
2011-04-14 21:49:52 +00:00
if (!VFS_QUOTACTL(mp, QCMD(Q_GETQUOTA,USRQUOTA),
Remove the thread argument from the FSD (File-System Dependent) parts of the VFS. Now all the VFS_* functions and relating parts don't want the context as long as it always refers to curthread. In some points, in particular when dealing with VOPs and functions living in the same namespace (eg. vflush) which still need to be converted, pass curthread explicitly in order to retain the old behaviour. Such loose ends will be fixed ASAP. While here fix a bug: now, UFS_EXTATTR can be compiled alone without the UFS_EXTATTR_AUTOSTART option. VFS KPI is heavilly changed by this commit so thirdy parts modules needs to be recompiled. Bump __FreeBSD_version in order to signal such situation.
2009-05-11 15:33:26 +00:00
cred->cr_uid, (caddr_t)&dqb))
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
freenum = min(dqb.dqb_bsoftlimit, freenum);
p->p_cred->p_ruid = savuid;
#endif /* QUOTA */
NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER);
uquad = (u_int64_t)freenum;
NFSQUOTABLKTOBYTE(uquad, fs.f_bsize);
txdr_hyper(uquad, tl);
retnum += NFSX_HYPER;
break;
case NFSATTRBIT_QUOTAUSED:
freenum = 0;
#ifdef QUOTA
/*
* ufs_quotactl() insists that the uid argument
* equal p_ruid for non-root quota access, so
* we'll just make sure that's the case.
*/
savuid = p->p_cred->p_ruid;
p->p_cred->p_ruid = cred->cr_uid;
Modify the experimental NFSv4 server so that it handles crossing of server mount points properly. The functions nfsvno_fillattr() and nfsv4_fillattr() were modified to take the extra arguments that are the mount point, a flag to indicate that it is a file system root and the mounted on fileno. The mount point argument needs to be busy when nfsvno_fillattr() is called, since the vp argument is not locked. Reviewed by: kib MFC after: 2 weeks
2011-04-14 21:49:52 +00:00
if (!VFS_QUOTACTL(mp, QCMD(Q_GETQUOTA,USRQUOTA),
Remove the thread argument from the FSD (File-System Dependent) parts of the VFS. Now all the VFS_* functions and relating parts don't want the context as long as it always refers to curthread. In some points, in particular when dealing with VOPs and functions living in the same namespace (eg. vflush) which still need to be converted, pass curthread explicitly in order to retain the old behaviour. Such loose ends will be fixed ASAP. While here fix a bug: now, UFS_EXTATTR can be compiled alone without the UFS_EXTATTR_AUTOSTART option. VFS KPI is heavilly changed by this commit so thirdy parts modules needs to be recompiled. Bump __FreeBSD_version in order to signal such situation.
2009-05-11 15:33:26 +00:00
cred->cr_uid, (caddr_t)&dqb))
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
freenum = dqb.dqb_curblocks;
p->p_cred->p_ruid = savuid;
#endif /* QUOTA */
NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER);
uquad = (u_int64_t)freenum;
NFSQUOTABLKTOBYTE(uquad, fs.f_bsize);
txdr_hyper(uquad, tl);
retnum += NFSX_HYPER;
break;
case NFSATTRBIT_RAWDEV:
NFSM_BUILD(tl, u_int32_t *, NFSX_V4SPECDATA);
*tl++ = txdr_unsigned(NFSMAJOR(vap->va_rdev));
*tl = txdr_unsigned(NFSMINOR(vap->va_rdev));
retnum += NFSX_V4SPECDATA;
break;
case NFSATTRBIT_SPACEAVAIL:
NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER);
if (priv_check_cred(cred, PRIV_VFS_BLOCKRESERVE, 0))
uquad = (u_int64_t)fs.f_bfree;
else
uquad = (u_int64_t)fs.f_bavail;
uquad *= fs.f_bsize;
txdr_hyper(uquad, tl);
retnum += NFSX_HYPER;
break;
case NFSATTRBIT_SPACEFREE:
NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER);
uquad = (u_int64_t)fs.f_bfree;
uquad *= fs.f_bsize;
txdr_hyper(uquad, tl);
retnum += NFSX_HYPER;
break;
case NFSATTRBIT_SPACETOTAL:
NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER);
uquad = (u_int64_t)fs.f_blocks;
uquad *= fs.f_bsize;
txdr_hyper(uquad, tl);
retnum += NFSX_HYPER;
break;
case NFSATTRBIT_SPACEUSED:
NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER);
txdr_hyper(vap->va_bytes, tl);
retnum += NFSX_HYPER;
break;
case NFSATTRBIT_TIMEACCESS:
NFSM_BUILD(tl, u_int32_t *, NFSX_V4TIME);
txdr_nfsv4time(&vap->va_atime, tl);
retnum += NFSX_V4TIME;
break;
case NFSATTRBIT_TIMEACCESSSET:
Use the VA_UTIMES_NULL flag to detect when NULL was passed to utimes() instead of comparing the desired time against the current time as a heuristic. Reviewed by: rmacklem MFC after: 1 week
2013-01-16 21:52:31 +00:00
if ((vap->va_vaflags & VA_UTIMES_NULL) == 0) {
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
NFSM_BUILD(tl, u_int32_t *, NFSX_V4SETTIME);
*tl++ = txdr_unsigned(NFSV4SATTRTIME_TOCLIENT);
txdr_nfsv4time(&vap->va_atime, tl);
retnum += NFSX_V4SETTIME;
} else {
NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED);
*tl = txdr_unsigned(NFSV4SATTRTIME_TOSERVER);
retnum += NFSX_UNSIGNED;
}
break;
case NFSATTRBIT_TIMEDELTA:
NFSM_BUILD(tl, u_int32_t *, NFSX_V4TIME);
temptime.tv_sec = 0;
temptime.tv_nsec = 1000000000 / hz;
txdr_nfsv4time(&temptime, tl);
retnum += NFSX_V4TIME;
break;
case NFSATTRBIT_TIMEMETADATA:
NFSM_BUILD(tl, u_int32_t *, NFSX_V4TIME);
txdr_nfsv4time(&vap->va_ctime, tl);
retnum += NFSX_V4TIME;
break;
case NFSATTRBIT_TIMEMODIFY:
NFSM_BUILD(tl, u_int32_t *, NFSX_V4TIME);
txdr_nfsv4time(&vap->va_mtime, tl);
retnum += NFSX_V4TIME;
break;
case NFSATTRBIT_TIMEMODIFYSET:
Use the VA_UTIMES_NULL flag to detect when NULL was passed to utimes() instead of comparing the desired time against the current time as a heuristic. Reviewed by: rmacklem MFC after: 1 week
2013-01-16 21:52:31 +00:00
if ((vap->va_vaflags & VA_UTIMES_NULL) == 0) {
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
NFSM_BUILD(tl, u_int32_t *, NFSX_V4SETTIME);
*tl++ = txdr_unsigned(NFSV4SATTRTIME_TOCLIENT);
txdr_nfsv4time(&vap->va_mtime, tl);
retnum += NFSX_V4SETTIME;
} else {
NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED);
*tl = txdr_unsigned(NFSV4SATTRTIME_TOSERVER);
retnum += NFSX_UNSIGNED;
}
break;
case NFSATTRBIT_MOUNTEDONFILEID:
NFSM_BUILD(tl, u_int32_t *, NFSX_HYPER);
Modify the experimental NFSv4 server so that it handles crossing of server mount points properly. The functions nfsvno_fillattr() and nfsv4_fillattr() were modified to take the extra arguments that are the mount point, a flag to indicate that it is a file system root and the mounted on fileno. The mount point argument needs to be busy when nfsvno_fillattr() is called, since the vp argument is not locked. Reviewed by: kib MFC after: 2 weeks
2011-04-14 21:49:52 +00:00
if (at_root != 0)
uquad = mounted_on_fileno;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
else
Modify the experimental NFSv4 server so that it handles crossing of server mount points properly. The functions nfsvno_fillattr() and nfsv4_fillattr() were modified to take the extra arguments that are the mount point, a flag to indicate that it is a file system root and the mounted on fileno. The mount point argument needs to be busy when nfsvno_fillattr() is called, since the vp argument is not locked. Reviewed by: kib MFC after: 2 weeks
2011-04-14 21:49:52 +00:00
uquad = (u_int64_t)vap->va_fileid;
txdr_hyper(uquad, tl);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
retnum += NFSX_HYPER;
break;
Merge the NFSv4.1 server code in projects/nfsv4.1-server over into head. The code is not believed to have any effect on the semantics of non-NFSv4.1 server behaviour. It is a rather large merge, but I am hoping that there will not be any regressions for the NFS server. MFC after: 1 month
2014-07-01 20:47:16 +00:00
case NFSATTRBIT_SUPPATTREXCLCREAT:
NFSSETSUPP_ATTRBIT(&attrbits);
NFSCLRNOTSETABLE_ATTRBIT(&attrbits);
NFSCLRBIT_ATTRBIT(&attrbits, NFSATTRBIT_TIMEACCESSSET);
retnum += nfsrv_putattrbit(nd, &attrbits);
break;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
default:
printf("EEK! Bad V4 attribute bitpos=%d\n", bitpos);
};
}
}
if (naclp != NULL)
acl_free(naclp);
*retnump = txdr_unsigned(retnum);
return (retnum + prefixnum);
}
/*
* Put the attribute bits onto an mbuf list.
* Return the number of bytes of output generated.
*/
APPLESTATIC int
nfsrv_putattrbit(struct nfsrv_descript *nd, nfsattrbit_t *attrbitp)
{
u_int32_t *tl;
int cnt, i, bytesize;
for (cnt = NFSATTRBIT_MAXWORDS; cnt > 0; cnt--)
if (attrbitp->bits[cnt - 1])
break;
bytesize = (cnt + 1) * NFSX_UNSIGNED;
NFSM_BUILD(tl, u_int32_t *, bytesize);
*tl++ = txdr_unsigned(cnt);
for (i = 0; i < cnt; i++)
*tl++ = txdr_unsigned(attrbitp->bits[i]);
return (bytesize);
}
/*
* Convert a uid to a string.
* If the lookup fails, just output the digits.
* uid - the user id
* cpp - points to a buffer of size NFSV4_SMALLSTR
* (malloc a larger one, as required)
* retlenp - pointer to length to be returned
*/
APPLESTATIC void
nfsv4_uidtostr(uid_t uid, u_char **cpp, int *retlenp, NFSPROC_T *p)
{
int i;
struct nfsusrgrp *usrp;
u_char *cp = *cpp;
uid_t tmp;
int cnt, hasampersand, len = NFSV4_SMALLSTR, ret;
cnt = 0;
tryagain:
NFSLOCKNAMEID();
if (nfsrv_dnsname) {
/*
* Always map nfsrv_defaultuid to "nobody".
*/
if (uid == nfsrv_defaultuid) {
i = nfsrv_dnsnamelen + 7;
if (i > len) {
NFSUNLOCKNAMEID();
if (len > NFSV4_SMALLSTR)
free(cp, M_NFSSTRING);
cp = malloc(i, M_NFSSTRING, M_WAITOK);
*cpp = cp;
len = i;
goto tryagain;
}
*retlenp = i;
NFSBCOPY("nobody@", cp, 7);
cp += 7;
NFSBCOPY(nfsrv_dnsname, cp, nfsrv_dnsnamelen);
NFSUNLOCKNAMEID();
return;
}
hasampersand = 0;
LIST_FOREACH(usrp, NFSUSERHASH(uid), lug_numhash) {
if (usrp->lug_uid == uid) {
if (usrp->lug_expiry < NFSD_MONOSEC)
break;
/*
* If the name doesn't already have an '@'
* in it, append @domainname to it.
*/
for (i = 0; i < usrp->lug_namelen; i++) {
if (usrp->lug_name[i] == '@') {
hasampersand = 1;
break;
}
}
if (hasampersand)
i = usrp->lug_namelen;
else
i = usrp->lug_namelen +
nfsrv_dnsnamelen + 1;
if (i > len) {
NFSUNLOCKNAMEID();
if (len > NFSV4_SMALLSTR)
free(cp, M_NFSSTRING);
cp = malloc(i, M_NFSSTRING, M_WAITOK);
*cpp = cp;
len = i;
goto tryagain;
}
*retlenp = i;
NFSBCOPY(usrp->lug_name, cp, usrp->lug_namelen);
if (!hasampersand) {
cp += usrp->lug_namelen;
*cp++ = '@';
NFSBCOPY(nfsrv_dnsname, cp, nfsrv_dnsnamelen);
}
TAILQ_REMOVE(&nfsuserlruhead, usrp, lug_lru);
TAILQ_INSERT_TAIL(&nfsuserlruhead, usrp, lug_lru);
NFSUNLOCKNAMEID();
return;
}
}
NFSUNLOCKNAMEID();
cnt++;
ret = nfsrv_getuser(RPCNFSUSERD_GETUID, uid, (gid_t)0,
NULL, p);
if (ret == 0 && cnt < 2)
goto tryagain;
} else {
NFSUNLOCKNAMEID();
}
/*
* No match, just return a string of digits.
*/
tmp = uid;
i = 0;
while (tmp || i == 0) {
tmp /= 10;
i++;
}
len = (i > len) ? len : i;
*retlenp = len;
cp += (len - 1);
tmp = uid;
for (i = 0; i < len; i++) {
*cp-- = '0' + (tmp % 10);
tmp /= 10;
}
return;
}
/*
* Convert a string to a uid.
* If no conversion is possible return NFSERR_BADOWNER, otherwise
* return 0.
Modify the NFSv4 client so that it can handle owner and owner_group strings that consist entirely of digits, interpreting them as the uid/gid number. This change was needed since new (>= 3.3) Linux servers reply with these strings by default. This change is mandated by the rfc3530bis draft. Reported on freebsd-stable@ under the Subject heading "Problem with Linux >= 3.3 as NFSv4 server" by Norbert Aschendorff on Aug. 20, 2012. Tested by: norbert.aschendorff at yahoo.de Reviewed by: jhb MFC after: 2 weeks
2012-09-20 02:49:25 +00:00
* If this is called from a client side mount using AUTH_SYS and the
* string is made up entirely of digits, just convert the string to
* a number.
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
*/
APPLESTATIC int
Modify the NFSv4 client so that it can handle owner and owner_group strings that consist entirely of digits, interpreting them as the uid/gid number. This change was needed since new (>= 3.3) Linux servers reply with these strings by default. This change is mandated by the rfc3530bis draft. Reported on freebsd-stable@ under the Subject heading "Problem with Linux >= 3.3 as NFSv4 server" by Norbert Aschendorff on Aug. 20, 2012. Tested by: norbert.aschendorff at yahoo.de Reviewed by: jhb MFC after: 2 weeks
2012-09-20 02:49:25 +00:00
nfsv4_strtouid(struct nfsrv_descript *nd, u_char *str, int len, uid_t *uidp,
NFSPROC_T *p)
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
{
int i;
Modify the NFSv4 client so that it can handle owner and owner_group strings that consist entirely of digits, interpreting them as the uid/gid number. This change was needed since new (>= 3.3) Linux servers reply with these strings by default. This change is mandated by the rfc3530bis draft. Reported on freebsd-stable@ under the Subject heading "Problem with Linux >= 3.3 as NFSv4 server" by Norbert Aschendorff on Aug. 20, 2012. Tested by: norbert.aschendorff at yahoo.de Reviewed by: jhb MFC after: 2 weeks
2012-09-20 02:49:25 +00:00
char *cp, *endstr, *str0;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
struct nfsusrgrp *usrp;
int cnt, ret;
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
int error = 0;
Modify the NFSv4 client so that it can handle owner and owner_group strings that consist entirely of digits, interpreting them as the uid/gid number. This change was needed since new (>= 3.3) Linux servers reply with these strings by default. This change is mandated by the rfc3530bis draft. Reported on freebsd-stable@ under the Subject heading "Problem with Linux >= 3.3 as NFSv4 server" by Norbert Aschendorff on Aug. 20, 2012. Tested by: norbert.aschendorff at yahoo.de Reviewed by: jhb MFC after: 2 weeks
2012-09-20 02:49:25 +00:00
uid_t tuid;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
if (len == 0) {
error = NFSERR_BADOWNER;
goto out;
}
Modify the NFSv4 client so that it can handle owner and owner_group strings that consist entirely of digits, interpreting them as the uid/gid number. This change was needed since new (>= 3.3) Linux servers reply with these strings by default. This change is mandated by the rfc3530bis draft. Reported on freebsd-stable@ under the Subject heading "Problem with Linux >= 3.3 as NFSv4 server" by Norbert Aschendorff on Aug. 20, 2012. Tested by: norbert.aschendorff at yahoo.de Reviewed by: jhb MFC after: 2 weeks
2012-09-20 02:49:25 +00:00
/* If a string of digits and an AUTH_SYS mount, just convert it. */
str0 = str;
tuid = (uid_t)strtoul(str0, &endstr, 10);
The new draft specification for NFSv4.0 specifies that a server should either accept owner and owner_group strings that are just the digits of the uid/gid or return NFS4ERR_BADOWNER. This patch adds a sysctl vfs.nfsd.enable_stringtouid, which can be set to enable the server w.r.t. accepting numeric string. It also ensures that NFS4ERR_BADOWNER is returned if numeric uid/gid strings are not enabled. This fixes the server for recent Linux nfs4 clients that use numeric uid/gid strings by default. Reported and tested by: craigyk@gmail.com MFC after: 2 weeks
2014-05-03 00:13:45 +00:00
if ((endstr - str0) == len) {
/* A numeric string. */
if ((nd->nd_flag & ND_KERBV) == 0 &&
((nd->nd_flag & ND_NFSCL) != 0 ||
nfsd_enable_stringtouid != 0))
*uidp = tuid;
else
error = NFSERR_BADOWNER;
Modify the NFSv4 client so that it can handle owner and owner_group strings that consist entirely of digits, interpreting them as the uid/gid number. This change was needed since new (>= 3.3) Linux servers reply with these strings by default. This change is mandated by the rfc3530bis draft. Reported on freebsd-stable@ under the Subject heading "Problem with Linux >= 3.3 as NFSv4 server" by Norbert Aschendorff on Aug. 20, 2012. Tested by: norbert.aschendorff at yahoo.de Reviewed by: jhb MFC after: 2 weeks
2012-09-20 02:49:25 +00:00
goto out;
}
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
/*
* Look for an '@'.
*/
Modify the NFSv4 client so that it can handle owner and owner_group strings that consist entirely of digits, interpreting them as the uid/gid number. This change was needed since new (>= 3.3) Linux servers reply with these strings by default. This change is mandated by the rfc3530bis draft. Reported on freebsd-stable@ under the Subject heading "Problem with Linux >= 3.3 as NFSv4 server" by Norbert Aschendorff on Aug. 20, 2012. Tested by: norbert.aschendorff at yahoo.de Reviewed by: jhb MFC after: 2 weeks
2012-09-20 02:49:25 +00:00
cp = strchr(str0, '@');
if (cp != NULL)
i = (int)(cp++ - str0);
else
i = len;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
cnt = 0;
tryagain:
NFSLOCKNAMEID();
/*
* If an '@' is found and the domain name matches, search for the name
* with dns stripped off.
* Mixed case alpahbetics will match for the domain name, but all
* upper case will not.
*/
if (cnt == 0 && i < len && i > 0 && nfsrv_dnsname &&
(len - 1 - i) == nfsrv_dnsnamelen &&
!nfsrv_cmpmixedcase(cp, nfsrv_dnsname, nfsrv_dnsnamelen)) {
len -= (nfsrv_dnsnamelen + 1);
*(cp - 1) = '\0';
}
/*
* Check for the special case of "nobody".
*/
if (len == 6 && !NFSBCMP(str, "nobody", 6)) {
*uidp = nfsrv_defaultuid;
NFSUNLOCKNAMEID();
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
error = 0;
goto out;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
}
LIST_FOREACH(usrp, NFSUSERNAMEHASH(str, len), lug_namehash) {
if (usrp->lug_namelen == len &&
!NFSBCMP(usrp->lug_name, str, len)) {
if (usrp->lug_expiry < NFSD_MONOSEC)
break;
*uidp = usrp->lug_uid;
TAILQ_REMOVE(&nfsuserlruhead, usrp, lug_lru);
TAILQ_INSERT_TAIL(&nfsuserlruhead, usrp, lug_lru);
NFSUNLOCKNAMEID();
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
error = 0;
goto out;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
}
}
NFSUNLOCKNAMEID();
cnt++;
ret = nfsrv_getuser(RPCNFSUSERD_GETUSER, (uid_t)0, (gid_t)0,
str, p);
if (ret == 0 && cnt < 2)
goto tryagain;
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
error = NFSERR_BADOWNER;
out:
NFSEXITCODE(error);
return (error);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
}
/*
* Convert a gid to a string.
* gid - the group id
* cpp - points to a buffer of size NFSV4_SMALLSTR
* (malloc a larger one, as required)
* retlenp - pointer to length to be returned
*/
APPLESTATIC void
nfsv4_gidtostr(gid_t gid, u_char **cpp, int *retlenp, NFSPROC_T *p)
{
int i;
struct nfsusrgrp *usrp;
u_char *cp = *cpp;
gid_t tmp;
int cnt, hasampersand, len = NFSV4_SMALLSTR, ret;
cnt = 0;
tryagain:
NFSLOCKNAMEID();
if (nfsrv_dnsname) {
/*
* Always map nfsrv_defaultgid to "nogroup".
*/
if (gid == nfsrv_defaultgid) {
i = nfsrv_dnsnamelen + 8;
if (i > len) {
NFSUNLOCKNAMEID();
if (len > NFSV4_SMALLSTR)
free(cp, M_NFSSTRING);
cp = malloc(i, M_NFSSTRING, M_WAITOK);
*cpp = cp;
len = i;
goto tryagain;
}
*retlenp = i;
NFSBCOPY("nogroup@", cp, 8);
cp += 8;
NFSBCOPY(nfsrv_dnsname, cp, nfsrv_dnsnamelen);
NFSUNLOCKNAMEID();
return;
}
hasampersand = 0;
LIST_FOREACH(usrp, NFSGROUPHASH(gid), lug_numhash) {
if (usrp->lug_gid == gid) {
if (usrp->lug_expiry < NFSD_MONOSEC)
break;
/*
* If the name doesn't already have an '@'
* in it, append @domainname to it.
*/
for (i = 0; i < usrp->lug_namelen; i++) {
if (usrp->lug_name[i] == '@') {
hasampersand = 1;
break;
}
}
if (hasampersand)
i = usrp->lug_namelen;
else
i = usrp->lug_namelen +
nfsrv_dnsnamelen + 1;
if (i > len) {
NFSUNLOCKNAMEID();
if (len > NFSV4_SMALLSTR)
free(cp, M_NFSSTRING);
cp = malloc(i, M_NFSSTRING, M_WAITOK);
*cpp = cp;
len = i;
goto tryagain;
}
*retlenp = i;
NFSBCOPY(usrp->lug_name, cp, usrp->lug_namelen);
if (!hasampersand) {
cp += usrp->lug_namelen;
*cp++ = '@';
NFSBCOPY(nfsrv_dnsname, cp, nfsrv_dnsnamelen);
}
TAILQ_REMOVE(&nfsuserlruhead, usrp, lug_lru);
TAILQ_INSERT_TAIL(&nfsuserlruhead, usrp, lug_lru);
NFSUNLOCKNAMEID();
return;
}
}
NFSUNLOCKNAMEID();
cnt++;
ret = nfsrv_getuser(RPCNFSUSERD_GETGID, (uid_t)0, gid,
NULL, p);
if (ret == 0 && cnt < 2)
goto tryagain;
} else {
NFSUNLOCKNAMEID();
}
/*
* No match, just return a string of digits.
*/
tmp = gid;
i = 0;
while (tmp || i == 0) {
tmp /= 10;
i++;
}
len = (i > len) ? len : i;
*retlenp = len;
cp += (len - 1);
tmp = gid;
for (i = 0; i < len; i++) {
*cp-- = '0' + (tmp % 10);
tmp /= 10;
}
return;
}
/*
* Convert a string to a gid.
Modify the NFSv4 client so that it can handle owner and owner_group strings that consist entirely of digits, interpreting them as the uid/gid number. This change was needed since new (>= 3.3) Linux servers reply with these strings by default. This change is mandated by the rfc3530bis draft. Reported on freebsd-stable@ under the Subject heading "Problem with Linux >= 3.3 as NFSv4 server" by Norbert Aschendorff on Aug. 20, 2012. Tested by: norbert.aschendorff at yahoo.de Reviewed by: jhb MFC after: 2 weeks
2012-09-20 02:49:25 +00:00
* If no conversion is possible return NFSERR_BADOWNER, otherwise
* return 0.
* If this is called from a client side mount using AUTH_SYS and the
* string is made up entirely of digits, just convert the string to
* a number.
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
*/
APPLESTATIC int
Modify the NFSv4 client so that it can handle owner and owner_group strings that consist entirely of digits, interpreting them as the uid/gid number. This change was needed since new (>= 3.3) Linux servers reply with these strings by default. This change is mandated by the rfc3530bis draft. Reported on freebsd-stable@ under the Subject heading "Problem with Linux >= 3.3 as NFSv4 server" by Norbert Aschendorff on Aug. 20, 2012. Tested by: norbert.aschendorff at yahoo.de Reviewed by: jhb MFC after: 2 weeks
2012-09-20 02:49:25 +00:00
nfsv4_strtogid(struct nfsrv_descript *nd, u_char *str, int len, gid_t *gidp,
NFSPROC_T *p)
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
{
int i;
Modify the NFSv4 client so that it can handle owner and owner_group strings that consist entirely of digits, interpreting them as the uid/gid number. This change was needed since new (>= 3.3) Linux servers reply with these strings by default. This change is mandated by the rfc3530bis draft. Reported on freebsd-stable@ under the Subject heading "Problem with Linux >= 3.3 as NFSv4 server" by Norbert Aschendorff on Aug. 20, 2012. Tested by: norbert.aschendorff at yahoo.de Reviewed by: jhb MFC after: 2 weeks
2012-09-20 02:49:25 +00:00
char *cp, *endstr, *str0;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
struct nfsusrgrp *usrp;
int cnt, ret;
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
int error = 0;
Modify the NFSv4 client so that it can handle owner and owner_group strings that consist entirely of digits, interpreting them as the uid/gid number. This change was needed since new (>= 3.3) Linux servers reply with these strings by default. This change is mandated by the rfc3530bis draft. Reported on freebsd-stable@ under the Subject heading "Problem with Linux >= 3.3 as NFSv4 server" by Norbert Aschendorff on Aug. 20, 2012. Tested by: norbert.aschendorff at yahoo.de Reviewed by: jhb MFC after: 2 weeks
2012-09-20 02:49:25 +00:00
gid_t tgid;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
if (len == 0) {
error = NFSERR_BADOWNER;
goto out;
}
Modify the NFSv4 client so that it can handle owner and owner_group strings that consist entirely of digits, interpreting them as the uid/gid number. This change was needed since new (>= 3.3) Linux servers reply with these strings by default. This change is mandated by the rfc3530bis draft. Reported on freebsd-stable@ under the Subject heading "Problem with Linux >= 3.3 as NFSv4 server" by Norbert Aschendorff on Aug. 20, 2012. Tested by: norbert.aschendorff at yahoo.de Reviewed by: jhb MFC after: 2 weeks
2012-09-20 02:49:25 +00:00
/* If a string of digits and an AUTH_SYS mount, just convert it. */
str0 = str;
tgid = (gid_t)strtoul(str0, &endstr, 10);
The new draft specification for NFSv4.0 specifies that a server should either accept owner and owner_group strings that are just the digits of the uid/gid or return NFS4ERR_BADOWNER. This patch adds a sysctl vfs.nfsd.enable_stringtouid, which can be set to enable the server w.r.t. accepting numeric string. It also ensures that NFS4ERR_BADOWNER is returned if numeric uid/gid strings are not enabled. This fixes the server for recent Linux nfs4 clients that use numeric uid/gid strings by default. Reported and tested by: craigyk@gmail.com MFC after: 2 weeks
2014-05-03 00:13:45 +00:00
if ((endstr - str0) == len) {
/* A numeric string. */
if ((nd->nd_flag & ND_KERBV) == 0 &&
((nd->nd_flag & ND_NFSCL) != 0 ||
nfsd_enable_stringtouid != 0))
*gidp = tgid;
else
error = NFSERR_BADOWNER;
Modify the NFSv4 client so that it can handle owner and owner_group strings that consist entirely of digits, interpreting them as the uid/gid number. This change was needed since new (>= 3.3) Linux servers reply with these strings by default. This change is mandated by the rfc3530bis draft. Reported on freebsd-stable@ under the Subject heading "Problem with Linux >= 3.3 as NFSv4 server" by Norbert Aschendorff on Aug. 20, 2012. Tested by: norbert.aschendorff at yahoo.de Reviewed by: jhb MFC after: 2 weeks
2012-09-20 02:49:25 +00:00
goto out;
}
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
/*
* Look for an '@'.
*/
Modify the NFSv4 client so that it can handle owner and owner_group strings that consist entirely of digits, interpreting them as the uid/gid number. This change was needed since new (>= 3.3) Linux servers reply with these strings by default. This change is mandated by the rfc3530bis draft. Reported on freebsd-stable@ under the Subject heading "Problem with Linux >= 3.3 as NFSv4 server" by Norbert Aschendorff on Aug. 20, 2012. Tested by: norbert.aschendorff at yahoo.de Reviewed by: jhb MFC after: 2 weeks
2012-09-20 02:49:25 +00:00
cp = strchr(str0, '@');
if (cp != NULL)
i = (int)(cp++ - str0);
else
i = len;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
cnt = 0;
tryagain:
NFSLOCKNAMEID();
/*
* If an '@' is found and the dns name matches, search for the name
* with the dns stripped off.
*/
if (cnt == 0 && i < len && i > 0 && nfsrv_dnsname &&
(len - 1 - i) == nfsrv_dnsnamelen &&
!nfsrv_cmpmixedcase(cp, nfsrv_dnsname, nfsrv_dnsnamelen)) {
len -= (nfsrv_dnsnamelen + 1);
*(cp - 1) = '\0';
}
/*
* Check for the special case of "nogroup".
*/
if (len == 7 && !NFSBCMP(str, "nogroup", 7)) {
*gidp = nfsrv_defaultgid;
NFSUNLOCKNAMEID();
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
error = 0;
goto out;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
}
LIST_FOREACH(usrp, NFSGROUPNAMEHASH(str, len), lug_namehash) {
if (usrp->lug_namelen == len &&
!NFSBCMP(usrp->lug_name, str, len)) {
if (usrp->lug_expiry < NFSD_MONOSEC)
break;
*gidp = usrp->lug_gid;
TAILQ_REMOVE(&nfsuserlruhead, usrp, lug_lru);
TAILQ_INSERT_TAIL(&nfsuserlruhead, usrp, lug_lru);
NFSUNLOCKNAMEID();
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
error = 0;
goto out;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
}
}
NFSUNLOCKNAMEID();
cnt++;
ret = nfsrv_getuser(RPCNFSUSERD_GETGROUP, (uid_t)0, (gid_t)0,
str, p);
if (ret == 0 && cnt < 2)
goto tryagain;
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
error = NFSERR_BADOWNER;
out:
NFSEXITCODE(error);
return (error);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
}
/*
* Cmp len chars, allowing mixed case in the first argument to match lower
* case in the second, but not if the first argument is all upper case.
* Return 0 for a match, 1 otherwise.
*/
static int
nfsrv_cmpmixedcase(u_char *cp, u_char *cp2, int len)
{
int i;
u_char tmp;
int fndlower = 0;
for (i = 0; i < len; i++) {
if (*cp >= 'A' && *cp <= 'Z') {
tmp = *cp++ + ('a' - 'A');
} else {
tmp = *cp++;
if (tmp >= 'a' && tmp <= 'z')
fndlower = 1;
}
if (tmp != *cp2++)
return (1);
}
if (fndlower)
return (0);
else
return (1);
}
/*
* Set the port for the nfsuserd.
*/
APPLESTATIC int
nfsrv_nfsuserdport(u_short port, NFSPROC_T *p)
{
struct nfssockreq *rp;
struct sockaddr_in *ad;
int error;
NFSLOCKNAMEID();
if (nfsrv_nfsuserd) {
NFSUNLOCKNAMEID();
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
error = EPERM;
goto out;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
}
nfsrv_nfsuserd = 1;
NFSUNLOCKNAMEID();
/*
* Set up the socket record and connect.
*/
rp = &nfsrv_nfsuserdsock;
rp->nr_client = NULL;
rp->nr_sotype = SOCK_DGRAM;
rp->nr_soproto = IPPROTO_UDP;
rp->nr_lock = (NFSR_RESERVEDPORT | NFSR_LOCALHOST);
rp->nr_cred = NULL;
NFSSOCKADDRALLOC(rp->nr_nam);
NFSSOCKADDRSIZE(rp->nr_nam, sizeof (struct sockaddr_in));
ad = NFSSOCKADDR(rp->nr_nam, struct sockaddr_in *);
ad->sin_family = AF_INET;
ad->sin_addr.s_addr = htonl((u_int32_t)0x7f000001); /* 127.0.0.1 */
ad->sin_port = port;
rp->nr_prog = RPCPROG_NFSUSERD;
rp->nr_vers = RPCNFSUSERD_VERS;
error = newnfs_connect(NULL, rp, NFSPROCCRED(p), p, 0);
if (error) {
NFSSOCKADDRFREE(rp->nr_nam);
nfsrv_nfsuserd = 0;
}
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
out:
NFSEXITCODE(error);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
return (error);
}
/*
* Delete the nfsuserd port.
*/
APPLESTATIC void
nfsrv_nfsuserddelport(void)
{
NFSLOCKNAMEID();
if (nfsrv_nfsuserd == 0) {
NFSUNLOCKNAMEID();
return;
}
nfsrv_nfsuserd = 0;
NFSUNLOCKNAMEID();
newnfs_disconnect(&nfsrv_nfsuserdsock);
NFSSOCKADDRFREE(nfsrv_nfsuserdsock.nr_nam);
}
/*
* Do upcalls to the nfsuserd, for cache misses of the owner/ownergroup
* name<-->id cache.
* Returns 0 upon success, non-zero otherwise.
*/
static int
nfsrv_getuser(int procnum, uid_t uid, gid_t gid, char *name, NFSPROC_T *p)
{
u_int32_t *tl;
struct nfsrv_descript *nd;
int len;
struct nfsrv_descript nfsd;
struct ucred *cred;
int error;
NFSLOCKNAMEID();
if (nfsrv_nfsuserd == 0) {
NFSUNLOCKNAMEID();
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
error = EPERM;
goto out;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
}
NFSUNLOCKNAMEID();
nd = &nfsd;
cred = newnfs_getcred();
nd->nd_flag = ND_GSSINITREPLY;
nfsrvd_rephead(nd);
nd->nd_procnum = procnum;
if (procnum == RPCNFSUSERD_GETUID || procnum == RPCNFSUSERD_GETGID) {
NFSM_BUILD(tl, u_int32_t *, NFSX_UNSIGNED);
if (procnum == RPCNFSUSERD_GETUID)
*tl = txdr_unsigned(uid);
else
*tl = txdr_unsigned(gid);
} else {
len = strlen(name);
(void) nfsm_strtom(nd, name, len);
}
error = newnfs_request(nd, NULL, NULL, &nfsrv_nfsuserdsock, NULL, NULL,
Move the NFSv4.1 client patches over from projects/nfsv4.1-client to head. I don't think the NFS client behaviour will change unless the new "minorversion=1" mount option is used. It includes basic NFSv4.1 support plus support for pNFS using the Files Layout only. All problems detecting during an NFSv4.1 Bakeathon testing event in June 2012 have been resolved in this code and it has been tested against the NFSv4.1 server available to me. Although not reviewed, I believe that kib@ has looked at it.
2012-12-08 22:52:39 +00:00
cred, RPCPROG_NFSUSERD, RPCNFSUSERD_VERS, NULL, 0, NULL, NULL);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
NFSFREECRED(cred);
if (!error) {
mbuf_freem(nd->nd_mrep);
error = nd->nd_repstat;
}
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
out:
NFSEXITCODE(error);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
return (error);
}
/*
* This function is called from the nfssvc(2) system call, to update the
* kernel user/group name list(s) for the V4 owner and ownergroup attributes.
*/
APPLESTATIC int
nfssvc_idname(struct nfsd_idargs *nidp)
{
struct nfsusrgrp *nusrp, *usrp, *newusrp;
struct nfsuserhashhead *hp;
int i;
int error = 0;
u_char *cp;
if (nidp->nid_flag & NFSID_INITIALIZE) {
cp = (u_char *)malloc(nidp->nid_namelen + 1,
M_NFSSTRING, M_WAITOK);
error = copyin(CAST_USER_ADDR_T(nidp->nid_name), cp,
nidp->nid_namelen);
NFSLOCKNAMEID();
if (nfsrv_dnsname) {
/*
* Free up all the old stuff and reinitialize hash lists.
*/
TAILQ_FOREACH_SAFE(usrp, &nfsuserlruhead, lug_lru, nusrp) {
nfsrv_removeuser(usrp);
}
free(nfsrv_dnsname, M_NFSSTRING);
nfsrv_dnsname = NULL;
}
TAILQ_INIT(&nfsuserlruhead);
for (i = 0; i < NFSUSERHASHSIZE; i++)
LIST_INIT(&nfsuserhash[i]);
for (i = 0; i < NFSGROUPHASHSIZE; i++)
LIST_INIT(&nfsgrouphash[i]);
for (i = 0; i < NFSUSERHASHSIZE; i++)
LIST_INIT(&nfsusernamehash[i]);
for (i = 0; i < NFSGROUPHASHSIZE; i++)
LIST_INIT(&nfsgroupnamehash[i]);
/*
* Put name in "DNS" string.
*/
if (!error) {
nfsrv_dnsname = cp;
nfsrv_dnsnamelen = nidp->nid_namelen;
nfsrv_defaultuid = nidp->nid_uid;
nfsrv_defaultgid = nidp->nid_gid;
nfsrv_usercnt = 0;
nfsrv_usermax = nidp->nid_usermax;
}
NFSUNLOCKNAMEID();
if (error)
free(cp, M_NFSSTRING);
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
goto out;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
}
/*
* malloc the new one now, so any potential sleep occurs before
* manipulation of the lists.
*/
MALLOC(newusrp, struct nfsusrgrp *, sizeof (struct nfsusrgrp) +
nidp->nid_namelen, M_NFSUSERGROUP, M_WAITOK);
error = copyin(CAST_USER_ADDR_T(nidp->nid_name), newusrp->lug_name,
nidp->nid_namelen);
if (error) {
free((caddr_t)newusrp, M_NFSUSERGROUP);
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
goto out;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
}
newusrp->lug_namelen = nidp->nid_namelen;
NFSLOCKNAMEID();
/*
* Delete old entries, as required.
*/
if (nidp->nid_flag & (NFSID_DELUID | NFSID_ADDUID)) {
hp = NFSUSERHASH(nidp->nid_uid);
LIST_FOREACH_SAFE(usrp, hp, lug_numhash, nusrp) {
if (usrp->lug_uid == nidp->nid_uid)
nfsrv_removeuser(usrp);
}
}
if (nidp->nid_flag & (NFSID_DELUSERNAME | NFSID_ADDUSERNAME)) {
hp = NFSUSERNAMEHASH(newusrp->lug_name, newusrp->lug_namelen);
LIST_FOREACH_SAFE(usrp, hp, lug_namehash, nusrp) {
if (usrp->lug_namelen == newusrp->lug_namelen &&
!NFSBCMP(usrp->lug_name, newusrp->lug_name,
usrp->lug_namelen))
nfsrv_removeuser(usrp);
}
}
if (nidp->nid_flag & (NFSID_DELGID | NFSID_ADDGID)) {
hp = NFSGROUPHASH(nidp->nid_gid);
LIST_FOREACH_SAFE(usrp, hp, lug_numhash, nusrp) {
if (usrp->lug_gid == nidp->nid_gid)
nfsrv_removeuser(usrp);
}
}
if (nidp->nid_flag & (NFSID_DELGROUPNAME | NFSID_ADDGROUPNAME)) {
hp = NFSGROUPNAMEHASH(newusrp->lug_name, newusrp->lug_namelen);
LIST_FOREACH_SAFE(usrp, hp, lug_namehash, nusrp) {
if (usrp->lug_namelen == newusrp->lug_namelen &&
!NFSBCMP(usrp->lug_name, newusrp->lug_name,
usrp->lug_namelen))
nfsrv_removeuser(usrp);
}
}
TAILQ_FOREACH_SAFE(usrp, &nfsuserlruhead, lug_lru, nusrp) {
if (usrp->lug_expiry < NFSD_MONOSEC)
nfsrv_removeuser(usrp);
}
while (nfsrv_usercnt >= nfsrv_usermax) {
usrp = TAILQ_FIRST(&nfsuserlruhead);
nfsrv_removeuser(usrp);
}
/*
* Now, we can add the new one.
*/
if (nidp->nid_usertimeout)
newusrp->lug_expiry = NFSD_MONOSEC + nidp->nid_usertimeout;
else
newusrp->lug_expiry = NFSD_MONOSEC + 5;
if (nidp->nid_flag & (NFSID_ADDUID | NFSID_ADDUSERNAME)) {
newusrp->lug_uid = nidp->nid_uid;
LIST_INSERT_HEAD(NFSUSERHASH(newusrp->lug_uid), newusrp,
lug_numhash);
LIST_INSERT_HEAD(NFSUSERNAMEHASH(newusrp->lug_name,
newusrp->lug_namelen), newusrp, lug_namehash);
TAILQ_INSERT_TAIL(&nfsuserlruhead, newusrp, lug_lru);
nfsrv_usercnt++;
} else if (nidp->nid_flag & (NFSID_ADDGID | NFSID_ADDGROUPNAME)) {
newusrp->lug_gid = nidp->nid_gid;
LIST_INSERT_HEAD(NFSGROUPHASH(newusrp->lug_gid), newusrp,
lug_numhash);
LIST_INSERT_HEAD(NFSGROUPNAMEHASH(newusrp->lug_name,
newusrp->lug_namelen), newusrp, lug_namehash);
TAILQ_INSERT_TAIL(&nfsuserlruhead, newusrp, lug_lru);
nfsrv_usercnt++;
} else
FREE((caddr_t)newusrp, M_NFSUSERGROUP);
NFSUNLOCKNAMEID();
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
out:
NFSEXITCODE(error);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
return (error);
}
/*
* Remove a user/group name element.
*/
static void
nfsrv_removeuser(struct nfsusrgrp *usrp)
{
NFSNAMEIDREQUIRED();
LIST_REMOVE(usrp, lug_numhash);
LIST_REMOVE(usrp, lug_namehash);
TAILQ_REMOVE(&nfsuserlruhead, usrp, lug_lru);
nfsrv_usercnt--;
FREE((caddr_t)usrp, M_NFSUSERGROUP);
}
/*
* This function scans a byte string and checks for UTF-8 compliance.
* It returns 0 if it conforms and NFSERR_INVAL if not.
*/
APPLESTATIC int
nfsrv_checkutf8(u_int8_t *cp, int len)
{
u_int32_t val = 0x0;
int cnt = 0, gotd = 0, shift = 0;
u_int8_t byte;
static int utf8_shift[5] = { 7, 11, 16, 21, 26 };
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
int error = 0;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
/*
* Here are what the variables are used for:
* val - the calculated value of a multibyte char, used to check
* that it was coded with the correct range
* cnt - the number of 10xxxxxx bytes to follow
* gotd - set for a char of Dxxx, so D800<->DFFF can be checked for
* shift - lower order bits of range (ie. "val >> shift" should
* not be 0, in other words, dividing by the lower bound
* of the range should get a non-zero value)
* byte - used to calculate cnt
*/
while (len > 0) {
if (cnt > 0) {
/* This handles the 10xxxxxx bytes */
if ((*cp & 0xc0) != 0x80 ||
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
(gotd && (*cp & 0x20))) {
error = NFSERR_INVAL;
goto out;
}
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
gotd = 0;
val <<= 6;
val |= (*cp & 0x3f);
cnt--;
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
if (cnt == 0 && (val >> shift) == 0x0) {
error = NFSERR_INVAL;
goto out;
}
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
} else if (*cp & 0x80) {
/* first byte of multi byte char */
byte = *cp;
while ((byte & 0x40) && cnt < 6) {
cnt++;
byte <<= 1;
}
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
if (cnt == 0 || cnt == 6) {
error = NFSERR_INVAL;
goto out;
}
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
val = (*cp & (0x3f >> cnt));
shift = utf8_shift[cnt - 1];
if (cnt == 2 && val == 0xd)
/* Check for the 0xd800-0xdfff case */
gotd = 1;
}
cp++;
len--;
}
if (cnt > 0)
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
error = NFSERR_INVAL;
out:
NFSEXITCODE(error);
return (error);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
}
/*
* Parse the xdr for an NFSv4 FsLocations attribute. Return two malloc'd
* strings, one with the root path in it and the other with the list of
* locations. The list is in the same format as is found in nfr_refs.
* It is a "," separated list of entries, where each of them is of the
* form <server>:<rootpath>. For example
* "nfsv4-test:/sub2,nfsv4-test2:/user/mnt,nfsv4-test2:/user/mnt2"
* The nilp argument is set to 1 for the special case of a null fs_root
* and an empty server list.
* It returns NFSERR_BADXDR, if the xdr can't be parsed and returns the
* number of xdr bytes parsed in sump.
*/
static int
nfsrv_getrefstr(struct nfsrv_descript *nd, u_char **fsrootp, u_char **srvp,
int *sump, int *nilp)
{
u_int32_t *tl;
u_char *cp = NULL, *cp2 = NULL, *cp3, *str;
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
int i, j, len, stringlen, cnt, slen, siz, xdrsum, error = 0, nsrv;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
struct list {
SLIST_ENTRY(list) next;
int len;
u_char host[1];
} *lsp, *nlsp;
SLIST_HEAD(, list) head;
*fsrootp = NULL;
*srvp = NULL;
*nilp = 0;
/*
* Get the fs_root path and check for the special case of null path
* and 0 length server list.
*/
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
len = fxdr_unsigned(int, *tl);
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
if (len < 0 || len > 10240) {
error = NFSERR_BADXDR;
goto nfsmout;
}
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
if (len == 0) {
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
if (*tl != 0) {
error = NFSERR_BADXDR;
goto nfsmout;
}
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
*nilp = 1;
*sump = 2 * NFSX_UNSIGNED;
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
error = 0;
goto nfsmout;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
}
cp = malloc(len + 1, M_NFSSTRING, M_WAITOK);
error = nfsrv_mtostr(nd, cp, len);
if (!error) {
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
cnt = fxdr_unsigned(int, *tl);
if (cnt <= 0)
error = NFSERR_BADXDR;
}
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
if (error)
goto nfsmout;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
/*
* Now, loop through the location list and make up the srvlist.
*/
xdrsum = (2 * NFSX_UNSIGNED) + NFSM_RNDUP(len);
cp2 = cp3 = malloc(1024, M_NFSSTRING, M_WAITOK);
slen = 1024;
siz = 0;
for (i = 0; i < cnt; i++) {
SLIST_INIT(&head);
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
nsrv = fxdr_unsigned(int, *tl);
if (nsrv <= 0) {
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
error = NFSERR_BADXDR;
goto nfsmout;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
}
/*
* Handle the first server by putting it in the srvstr.
*/
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
len = fxdr_unsigned(int, *tl);
if (len <= 0 || len > 1024) {
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
error = NFSERR_BADXDR;
goto nfsmout;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
}
nfsrv_refstrbigenough(siz + len + 3, &cp2, &cp3, &slen);
if (cp3 != cp2) {
*cp3++ = ',';
siz++;
}
error = nfsrv_mtostr(nd, cp3, len);
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
if (error)
goto nfsmout;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
cp3 += len;
*cp3++ = ':';
siz += (len + 1);
xdrsum += (2 * NFSX_UNSIGNED) + NFSM_RNDUP(len);
for (j = 1; j < nsrv; j++) {
/*
* Yuck, put them in an slist and process them later.
*/
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
len = fxdr_unsigned(int, *tl);
if (len <= 0 || len > 1024) {
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
error = NFSERR_BADXDR;
goto nfsmout;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
}
lsp = (struct list *)malloc(sizeof (struct list)
+ len, M_TEMP, M_WAITOK);
error = nfsrv_mtostr(nd, lsp->host, len);
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
if (error)
goto nfsmout;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
xdrsum += NFSX_UNSIGNED + NFSM_RNDUP(len);
lsp->len = len;
SLIST_INSERT_HEAD(&head, lsp, next);
}
/*
* Finally, we can get the path.
*/
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
len = fxdr_unsigned(int, *tl);
if (len <= 0 || len > 1024) {
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
error = NFSERR_BADXDR;
goto nfsmout;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
}
nfsrv_refstrbigenough(siz + len + 1, &cp2, &cp3, &slen);
error = nfsrv_mtostr(nd, cp3, len);
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
if (error)
goto nfsmout;
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
xdrsum += NFSX_UNSIGNED + NFSM_RNDUP(len);
str = cp3;
stringlen = len;
cp3 += len;
siz += len;
SLIST_FOREACH_SAFE(lsp, &head, next, nlsp) {
nfsrv_refstrbigenough(siz + lsp->len + stringlen + 3,
&cp2, &cp3, &slen);
*cp3++ = ',';
NFSBCOPY(lsp->host, cp3, lsp->len);
cp3 += lsp->len;
*cp3++ = ':';
NFSBCOPY(str, cp3, stringlen);
cp3 += stringlen;
*cp3 = '\0';
siz += (lsp->len + stringlen + 2);
free((caddr_t)lsp, M_TEMP);
}
}
*fsrootp = cp;
*srvp = cp2;
*sump = xdrsum;
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
NFSEXITCODE2(0, nd);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
return (0);
nfsmout:
if (cp != NULL)
free(cp, M_NFSSTRING);
if (cp2 != NULL)
free(cp2, M_NFSSTRING);
Add DEXITCODE plumbing to NFS. Isilon has the concept of an in-memory exit-code ring that saves the last exit code of a function and allows for stack tracing. This is very helpful when debugging tough issues. This patch is essentially a no-op for BSD at this point, until we upstream the dexitcode logic itself. The patch adds DEXITCODE calls to every NFS function that returns an errno error code. A number of code paths were also reorganized to have single exit paths, to reduce code duplication. Submitted by: David Kwan <dkwan@isilon.com> Reviewed by: rmacklem Approved by: zml (mentor) MFC after: 2 weeks
2011-07-16 08:51:09 +00:00
NFSEXITCODE2(error, nd);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
return (error);
}
/*
* Make the malloc'd space large enough. This is a pain, but the xdr
* doesn't set an upper bound on the side, so...
*/
static void
nfsrv_refstrbigenough(int siz, u_char **cpp, u_char **cpp2, int *slenp)
{
u_char *cp;
int i;
if (siz <= *slenp)
return;
cp = malloc(siz + 1024, M_NFSSTRING, M_WAITOK);
NFSBCOPY(*cpp, cp, *slenp);
free(*cpp, M_NFSSTRING);
i = *cpp2 - *cpp;
*cpp = cp;
*cpp2 = cp + i;
*slenp = siz + 1024;
}
/*
* Initialize the reply header data structures.
*/
APPLESTATIC void
nfsrvd_rephead(struct nfsrv_descript *nd)
{
mbuf_t mreq;
/*
* If this is a big reply, use a cluster.
*/
if ((nd->nd_flag & ND_GSSINITREPLY) == 0 &&
nfs_bigreply[nd->nd_procnum]) {
Mechanically substitute flags from historic mbuf allocator with malloc(9) flags within sys. Exceptions: - sys/contrib not touched - sys/mbuf.h edited manually
2012-12-05 08:04:20 +00:00
NFSMCLGET(mreq, M_WAITOK);
Add the experimental nfs subtree to the kernel, that includes support for NFSv4 as well as NFSv2 and 3. It lives in 3 subdirs under sys/fs: nfs - functions that are common to the client and server nfsclient - a mutation of sys/nfsclient that call generic functions to do RPCs and handle state. As such, it retains the buffer cache handling characteristics and vnode semantics that are found in sys/nfsclient, for the most part. nfsserver - the server. It includes a DRC designed specifically for NFSv4, that is used instead of the generic DRC in sys/rpc. The build glue will be checked in later, so at this point, it consists of 3 new subdirs that should not affect kernel building. Approved by: kib (mentor)
2009-05-04 15:23:58 +00:00
nd->nd_mreq = mreq;
nd->nd_mb = mreq;
} else {
NFSMGET(mreq);
nd->nd_mreq = mreq;
nd->nd_mb = mreq;
}
nd->nd_bpos = NFSMTOD(mreq, caddr_t);
mbuf_setlen(mreq, 0);
if ((nd->nd_flag & ND_GSSINITREPLY) == 0)
NFSM_BUILD(nd->nd_errp, int *, NFSX_UNSIGNED);
}
/*
* Lock a socket against others.
* Currently used to serialize connect/disconnect attempts.
*/
int
newnfs_sndlock(int *flagp)
{
struct timespec ts;
NFSLOCKSOCK();
while (*flagp & NFSR_SNDLOCK) {
*flagp |= NFSR_WANTSND;
ts.tv_sec = 0;
ts.tv_nsec = 0;
(void) nfsmsleep((caddr_t)flagp, NFSSOCKMUTEXPTR,
PZERO - 1, "nfsndlck", &ts);
}
*flagp |= NFSR_SNDLOCK;
NFSUNLOCKSOCK();
return (0);
}
/*
* Unlock the stream socket for others.
*/
void
newnfs_sndunlock(int *flagp)
{
NFSLOCKSOCK();
if ((*flagp & NFSR_SNDLOCK) == 0)
panic("nfs sndunlock");
*flagp &= ~NFSR_SNDLOCK;
if (*flagp & NFSR_WANTSND) {
*flagp &= ~NFSR_WANTSND;
wakeup((caddr_t)flagp);
}
NFSUNLOCKSOCK();
}
Move the NFSv4.1 client patches over from projects/nfsv4.1-client to head. I don't think the NFS client behaviour will change unless the new "minorversion=1" mount option is used. It includes basic NFSv4.1 support plus support for pNFS using the Files Layout only. All problems detecting during an NFSv4.1 Bakeathon testing event in June 2012 have been resolved in this code and it has been tested against the NFSv4.1 server available to me. Although not reviewed, I believe that kib@ has looked at it.
2012-12-08 22:52:39 +00:00
APPLESTATIC int
nfsv4_getipaddr(struct nfsrv_descript *nd, struct sockaddr_storage *sa,
int *isudp)
{
struct sockaddr_in *sad;
struct sockaddr_in6 *sad6;
struct in_addr saddr;
uint32_t portnum, *tl;
int af = 0, i, j, k;
char addr[64], protocol[5], *cp;
int cantparse = 0, error = 0;
uint16_t portv;
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
i = fxdr_unsigned(int, *tl);
if (i >= 3 && i <= 4) {
error = nfsrv_mtostr(nd, protocol, i);
if (error)
goto nfsmout;
if (strcmp(protocol, "tcp") == 0) {
af = AF_INET;
*isudp = 0;
} else if (strcmp(protocol, "udp") == 0) {
af = AF_INET;
*isudp = 1;
} else if (strcmp(protocol, "tcp6") == 0) {
af = AF_INET6;
*isudp = 0;
} else if (strcmp(protocol, "udp6") == 0) {
af = AF_INET6;
*isudp = 1;
} else
cantparse = 1;
} else {
cantparse = 1;
if (i > 0) {
error = nfsm_advance(nd, NFSM_RNDUP(i), -1);
if (error)
goto nfsmout;
}
}
NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED);
i = fxdr_unsigned(int, *tl);
if (i < 0) {
error = NFSERR_BADXDR;
goto nfsmout;
} else if (cantparse == 0 && i >= 11 && i < 64) {
/*
* The shortest address is 11chars and the longest is < 64.
*/
error = nfsrv_mtostr(nd, addr, i);
if (error)
goto nfsmout;
/* Find the port# at the end and extract that. */
i = strlen(addr);
k = 0;
cp = &addr[i - 1];
/* Count back two '.'s from end to get port# field. */
for (j = 0; j < i; j++) {
if (*cp == '.') {
k++;
if (k == 2)
break;
}
cp--;
}
if (k == 2) {
/*
* The NFSv4 port# is appended as .N.N, where N is
* a decimal # in the range 0-255, just like an inet4
* address. Cheat and use inet_aton(), which will
* return a Class A address and then shift the high
* order 8bits over to convert it to the port#.
*/
*cp++ = '\0';
if (inet_aton(cp, &saddr) == 1) {
portnum = ntohl(saddr.s_addr);
portv = (uint16_t)((portnum >> 16) |
(portnum & 0xff));
} else
cantparse = 1;
} else
cantparse = 1;
if (cantparse == 0) {
if (af == AF_INET) {
sad = (struct sockaddr_in *)sa;
if (inet_pton(af, addr, &sad->sin_addr) == 1) {
sad->sin_len = sizeof(*sad);
sad->sin_family = AF_INET;
sad->sin_port = htons(portv);
return (0);
}
} else {
sad6 = (struct sockaddr_in6 *)sa;
if (inet_pton(af, addr, &sad6->sin6_addr)
== 1) {
sad6->sin6_len = sizeof(*sad6);
sad6->sin6_family = AF_INET6;
sad6->sin6_port = htons(portv);
return (0);
}
}
}
} else {
if (i > 0) {
error = nfsm_advance(nd, NFSM_RNDUP(i), -1);
if (error)
goto nfsmout;
}
}
error = EPERM;
nfsmout:
return (error);
}
/*
* Handle an NFSv4.1 Sequence request for the session.
Merge the NFSv4.1 server code in projects/nfsv4.1-server over into head. The code is not believed to have any effect on the semantics of non-NFSv4.1 server behaviour. It is a rather large merge, but I am hoping that there will not be any regressions for the NFS server. MFC after: 1 month
2014-07-01 20:47:16 +00:00
* If reply != NULL, use it to return the cached reply, as required.
* The client gets a cached reply via this call for callbacks, however the
* server gets a cached reply via the nfsv4_seqsess_cachereply() call.
Move the NFSv4.1 client patches over from projects/nfsv4.1-client to head. I don't think the NFS client behaviour will change unless the new "minorversion=1" mount option is used. It includes basic NFSv4.1 support plus support for pNFS using the Files Layout only. All problems detecting during an NFSv4.1 Bakeathon testing event in June 2012 have been resolved in this code and it has been tested against the NFSv4.1 server available to me. Although not reviewed, I believe that kib@ has looked at it.
2012-12-08 22:52:39 +00:00
*/
int
nfsv4_seqsession(uint32_t seqid, uint32_t slotid, uint32_t highslot,
struct nfsslot *slots, struct mbuf **reply, uint16_t maxslot)
{
int error;
error = 0;
Merge the NFSv4.1 server code in projects/nfsv4.1-server over into head. The code is not believed to have any effect on the semantics of non-NFSv4.1 server behaviour. It is a rather large merge, but I am hoping that there will not be any regressions for the NFS server. MFC after: 1 month
2014-07-01 20:47:16 +00:00
if (reply != NULL)
*reply = NULL;
Move the NFSv4.1 client patches over from projects/nfsv4.1-client to head. I don't think the NFS client behaviour will change unless the new "minorversion=1" mount option is used. It includes basic NFSv4.1 support plus support for pNFS using the Files Layout only. All problems detecting during an NFSv4.1 Bakeathon testing event in June 2012 have been resolved in this code and it has been tested against the NFSv4.1 server available to me. Although not reviewed, I believe that kib@ has looked at it.
2012-12-08 22:52:39 +00:00
if (slotid > maxslot)
return (NFSERR_BADSLOT);
if (seqid == slots[slotid].nfssl_seq) {
/* A retry. */
if (slots[slotid].nfssl_inprog != 0)
error = NFSERR_DELAY;
else if (slots[slotid].nfssl_reply != NULL) {
Merge the NFSv4.1 server code in projects/nfsv4.1-server over into head. The code is not believed to have any effect on the semantics of non-NFSv4.1 server behaviour. It is a rather large merge, but I am hoping that there will not be any regressions for the NFS server. MFC after: 1 month
2014-07-01 20:47:16 +00:00
if (reply != NULL) {
*reply = slots[slotid].nfssl_reply;
slots[slotid].nfssl_reply = NULL;
}
Move the NFSv4.1 client patches over from projects/nfsv4.1-client to head. I don't think the NFS client behaviour will change unless the new "minorversion=1" mount option is used. It includes basic NFSv4.1 support plus support for pNFS using the Files Layout only. All problems detecting during an NFSv4.1 Bakeathon testing event in June 2012 have been resolved in this code and it has been tested against the NFSv4.1 server available to me. Although not reviewed, I believe that kib@ has looked at it.
2012-12-08 22:52:39 +00:00
slots[slotid].nfssl_inprog = 1;
Merge the NFSv4.1 server code in projects/nfsv4.1-server over into head. The code is not believed to have any effect on the semantics of non-NFSv4.1 server behaviour. It is a rather large merge, but I am hoping that there will not be any regressions for the NFS server. MFC after: 1 month
2014-07-01 20:47:16 +00:00
error = NFSERR_REPLYFROMCACHE;
Move the NFSv4.1 client patches over from projects/nfsv4.1-client to head. I don't think the NFS client behaviour will change unless the new "minorversion=1" mount option is used. It includes basic NFSv4.1 support plus support for pNFS using the Files Layout only. All problems detecting during an NFSv4.1 Bakeathon testing event in June 2012 have been resolved in this code and it has been tested against the NFSv4.1 server available to me. Although not reviewed, I believe that kib@ has looked at it.
2012-12-08 22:52:39 +00:00
} else
Merge the NFSv4.1 server code in projects/nfsv4.1-server over into head. The code is not believed to have any effect on the semantics of non-NFSv4.1 server behaviour. It is a rather large merge, but I am hoping that there will not be any regressions for the NFS server. MFC after: 1 month
2014-07-01 20:47:16 +00:00
/* No reply cached, so just do it. */
slots[slotid].nfssl_inprog = 1;
Move the NFSv4.1 client patches over from projects/nfsv4.1-client to head. I don't think the NFS client behaviour will change unless the new "minorversion=1" mount option is used. It includes basic NFSv4.1 support plus support for pNFS using the Files Layout only. All problems detecting during an NFSv4.1 Bakeathon testing event in June 2012 have been resolved in this code and it has been tested against the NFSv4.1 server available to me. Although not reviewed, I believe that kib@ has looked at it.
2012-12-08 22:52:39 +00:00
} else if ((slots[slotid].nfssl_seq + 1) == seqid) {
Merge the NFSv4.1 server code in projects/nfsv4.1-server over into head. The code is not believed to have any effect on the semantics of non-NFSv4.1 server behaviour. It is a rather large merge, but I am hoping that there will not be any regressions for the NFS server. MFC after: 1 month
2014-07-01 20:47:16 +00:00
if (slots[slotid].nfssl_reply != NULL)
m_freem(slots[slotid].nfssl_reply);
Move the NFSv4.1 client patches over from projects/nfsv4.1-client to head. I don't think the NFS client behaviour will change unless the new "minorversion=1" mount option is used. It includes basic NFSv4.1 support plus support for pNFS using the Files Layout only. All problems detecting during an NFSv4.1 Bakeathon testing event in June 2012 have been resolved in this code and it has been tested against the NFSv4.1 server available to me. Although not reviewed, I believe that kib@ has looked at it.
2012-12-08 22:52:39 +00:00
slots[slotid].nfssl_reply = NULL;
slots[slotid].nfssl_inprog = 1;
slots[slotid].nfssl_seq++;
} else
error = NFSERR_SEQMISORDERED;
return (error);
}
/*
* Cache this reply for the slot.
Merge the NFSv4.1 server code in projects/nfsv4.1-server over into head. The code is not believed to have any effect on the semantics of non-NFSv4.1 server behaviour. It is a rather large merge, but I am hoping that there will not be any regressions for the NFS server. MFC after: 1 month
2014-07-01 20:47:16 +00:00
* Use the "rep" argument to return the cached reply if repstat is set to
* NFSERR_REPLYFROMCACHE. The client never sets repstat to this value.
Move the NFSv4.1 client patches over from projects/nfsv4.1-client to head. I don't think the NFS client behaviour will change unless the new "minorversion=1" mount option is used. It includes basic NFSv4.1 support plus support for pNFS using the Files Layout only. All problems detecting during an NFSv4.1 Bakeathon testing event in June 2012 have been resolved in this code and it has been tested against the NFSv4.1 server available to me. Although not reviewed, I believe that kib@ has looked at it.
2012-12-08 22:52:39 +00:00
*/
void
Merge the NFSv4.1 server code in projects/nfsv4.1-server over into head. The code is not believed to have any effect on the semantics of non-NFSv4.1 server behaviour. It is a rather large merge, but I am hoping that there will not be any regressions for the NFS server. MFC after: 1 month
2014-07-01 20:47:16 +00:00
nfsv4_seqsess_cacherep(uint32_t slotid, struct nfsslot *slots, int repstat,
struct mbuf **rep)
Move the NFSv4.1 client patches over from projects/nfsv4.1-client to head. I don't think the NFS client behaviour will change unless the new "minorversion=1" mount option is used. It includes basic NFSv4.1 support plus support for pNFS using the Files Layout only. All problems detecting during an NFSv4.1 Bakeathon testing event in June 2012 have been resolved in this code and it has been tested against the NFSv4.1 server available to me. Although not reviewed, I believe that kib@ has looked at it.
2012-12-08 22:52:39 +00:00
{
Merge the NFSv4.1 server code in projects/nfsv4.1-server over into head. The code is not believed to have any effect on the semantics of non-NFSv4.1 server behaviour. It is a rather large merge, but I am hoping that there will not be any regressions for the NFS server. MFC after: 1 month
2014-07-01 20:47:16 +00:00
if (repstat == NFSERR_REPLYFROMCACHE) {
*rep = slots[slotid].nfssl_reply;
slots[slotid].nfssl_reply = NULL;
} else {
if (slots[slotid].nfssl_reply != NULL)
m_freem(slots[slotid].nfssl_reply);
slots[slotid].nfssl_reply = *rep;
}
Move the NFSv4.1 client patches over from projects/nfsv4.1-client to head. I don't think the NFS client behaviour will change unless the new "minorversion=1" mount option is used. It includes basic NFSv4.1 support plus support for pNFS using the Files Layout only. All problems detecting during an NFSv4.1 Bakeathon testing event in June 2012 have been resolved in this code and it has been tested against the NFSv4.1 server available to me. Although not reviewed, I believe that kib@ has looked at it.
2012-12-08 22:52:39 +00:00
slots[slotid].nfssl_inprog = 0;
}
/*
* Generate the xdr for an NFSv4.1 Sequence Operation.
*/
APPLESTATIC void
Fix an NFSv4.1 client specific case where a forced dismount would hang. The hang occurred in nfsv4_setsequence() when it couldn't find an available session slot and is fixed by checking for a forced dismount in progress and just returning for this case. MFC after: 1 month
2013-11-09 21:24:56 +00:00
nfsv4_setsequence(struct nfsmount *nmp, struct nfsrv_descript *nd,
struct nfsclsession *sep, int dont_replycache)
Move the NFSv4.1 client patches over from projects/nfsv4.1-client to head. I don't think the NFS client behaviour will change unless the new "minorversion=1" mount option is used. It includes basic NFSv4.1 support plus support for pNFS using the Files Layout only. All problems detecting during an NFSv4.1 Bakeathon testing event in June 2012 have been resolved in this code and it has been tested against the NFSv4.1 server available to me. Although not reviewed, I believe that kib@ has looked at it.
2012-12-08 22:52:39 +00:00
{
uint32_t *tl, slotseq = 0;
Merge the NFSv4.1 server code in projects/nfsv4.1-server over into head. The code is not believed to have any effect on the semantics of non-NFSv4.1 server behaviour. It is a rather large merge, but I am hoping that there will not be any regressions for the NFS server. MFC after: 1 month
2014-07-01 20:47:16 +00:00
int error, maxslot, slotpos;
uint8_t sessionid[NFSX_V4SESSIONID];
error = nfsv4_sequencelookup(nmp, sep, &slotpos, &maxslot, &slotseq,
sessionid);
if (error != 0)
return;
KASSERT(maxslot >= 0, ("nfscl_setsequence neg maxslot"));
/* Build the Sequence arguments. */
NFSM_BUILD(tl, uint32_t *, NFSX_V4SESSIONID + 4 * NFSX_UNSIGNED);
bcopy(sessionid, tl, NFSX_V4SESSIONID);
tl += NFSX_V4SESSIONID / NFSX_UNSIGNED;
nd->nd_slotseq = tl;
*tl++ = txdr_unsigned(slotseq);
*tl++ = txdr_unsigned(slotpos);
*tl++ = txdr_unsigned(maxslot);
if (dont_replycache == 0)
*tl = newnfs_true;
else
*tl = newnfs_false;
nd->nd_flag |= ND_HASSEQUENCE;
}
int
nfsv4_sequencelookup(struct nfsmount *nmp, struct nfsclsession *sep,
int *slotposp, int *maxslotp, uint32_t *slotseqp, uint8_t *sessionid)
{
Move the NFSv4.1 client patches over from projects/nfsv4.1-client to head. I don't think the NFS client behaviour will change unless the new "minorversion=1" mount option is used. It includes basic NFSv4.1 support plus support for pNFS using the Files Layout only. All problems detecting during an NFSv4.1 Bakeathon testing event in June 2012 have been resolved in this code and it has been tested against the NFSv4.1 server available to me. Although not reviewed, I believe that kib@ has looked at it.
2012-12-08 22:52:39 +00:00
int i, maxslot, slotpos;
uint64_t bitval;
/* Find an unused slot. */
slotpos = -1;
maxslot = -1;
mtx_lock(&sep->nfsess_mtx);
do {
bitval = 1;
for (i = 0; i < sep->nfsess_foreslots; i++) {
if ((bitval & sep->nfsess_slots) == 0) {
slotpos = i;
sep->nfsess_slots |= bitval;
sep->nfsess_slotseq[i]++;
Merge the NFSv4.1 server code in projects/nfsv4.1-server over into head. The code is not believed to have any effect on the semantics of non-NFSv4.1 server behaviour. It is a rather large merge, but I am hoping that there will not be any regressions for the NFS server. MFC after: 1 month
2014-07-01 20:47:16 +00:00
*slotseqp = sep->nfsess_slotseq[i];
Move the NFSv4.1 client patches over from projects/nfsv4.1-client to head. I don't think the NFS client behaviour will change unless the new "minorversion=1" mount option is used. It includes basic NFSv4.1 support plus support for pNFS using the Files Layout only. All problems detecting during an NFSv4.1 Bakeathon testing event in June 2012 have been resolved in this code and it has been tested against the NFSv4.1 server available to me. Although not reviewed, I believe that kib@ has looked at it.
2012-12-08 22:52:39 +00:00
break;
}
bitval <<= 1;
}
Fix an NFSv4.1 client specific case where a forced dismount would hang. The hang occurred in nfsv4_setsequence() when it couldn't find an available session slot and is fixed by checking for a forced dismount in progress and just returning for this case. MFC after: 1 month
2013-11-09 21:24:56 +00:00
if (slotpos == -1) {
/*
* If a forced dismount is in progress, just return.
* This RPC attempt will fail when it calls
* newnfs_request().
*/
Merge the NFSv4.1 server code in projects/nfsv4.1-server over into head. The code is not believed to have any effect on the semantics of non-NFSv4.1 server behaviour. It is a rather large merge, but I am hoping that there will not be any regressions for the NFS server. MFC after: 1 month
2014-07-01 20:47:16 +00:00
if (nmp != NULL &&
(nmp->nm_mountp->mnt_kern_flag & MNTK_UNMOUNTF)
Fix an NFSv4.1 client specific case where a forced dismount would hang. The hang occurred in nfsv4_setsequence() when it couldn't find an available session slot and is fixed by checking for a forced dismount in progress and just returning for this case. MFC after: 1 month
2013-11-09 21:24:56 +00:00
!= 0) {
mtx_unlock(&sep->nfsess_mtx);
Merge the NFSv4.1 server code in projects/nfsv4.1-server over into head. The code is not believed to have any effect on the semantics of non-NFSv4.1 server behaviour. It is a rather large merge, but I am hoping that there will not be any regressions for the NFS server. MFC after: 1 month
2014-07-01 20:47:16 +00:00
return (ESTALE);
Fix an NFSv4.1 client specific case where a forced dismount would hang. The hang occurred in nfsv4_setsequence() when it couldn't find an available session slot and is fixed by checking for a forced dismount in progress and just returning for this case. MFC after: 1 month
2013-11-09 21:24:56 +00:00
}
/* Wake up once/sec, to check for a forced dismount. */
Move the NFSv4.1 client patches over from projects/nfsv4.1-client to head. I don't think the NFS client behaviour will change unless the new "minorversion=1" mount option is used. It includes basic NFSv4.1 support plus support for pNFS using the Files Layout only. All problems detecting during an NFSv4.1 Bakeathon testing event in June 2012 have been resolved in this code and it has been tested against the NFSv4.1 server available to me. Although not reviewed, I believe that kib@ has looked at it.
2012-12-08 22:52:39 +00:00
(void)mtx_sleep(&sep->nfsess_slots, &sep->nfsess_mtx,
Fix an NFSv4.1 client specific case where a forced dismount would hang. The hang occurred in nfsv4_setsequence() when it couldn't find an available session slot and is fixed by checking for a forced dismount in progress and just returning for this case. MFC after: 1 month
2013-11-09 21:24:56 +00:00
PZERO, "nfsclseq", hz);
}
Move the NFSv4.1 client patches over from projects/nfsv4.1-client to head. I don't think the NFS client behaviour will change unless the new "minorversion=1" mount option is used. It includes basic NFSv4.1 support plus support for pNFS using the Files Layout only. All problems detecting during an NFSv4.1 Bakeathon testing event in June 2012 have been resolved in this code and it has been tested against the NFSv4.1 server available to me. Although not reviewed, I believe that kib@ has looked at it.
2012-12-08 22:52:39 +00:00
} while (slotpos == -1);
/* Now, find the highest slot in use. (nfsc_slots is 64bits) */
bitval = 1;
for (i = 0; i < 64; i++) {
if ((bitval & sep->nfsess_slots) != 0)
maxslot = i;
bitval <<= 1;
}
bcopy(sep->nfsess_sessionid, sessionid, NFSX_V4SESSIONID);
mtx_unlock(&sep->nfsess_mtx);
Merge the NFSv4.1 server code in projects/nfsv4.1-server over into head. The code is not believed to have any effect on the semantics of non-NFSv4.1 server behaviour. It is a rather large merge, but I am hoping that there will not be any regressions for the NFS server. MFC after: 1 month
2014-07-01 20:47:16 +00:00
*slotposp = slotpos;
*maxslotp = maxslot;
return (0);
Move the NFSv4.1 client patches over from projects/nfsv4.1-client to head. I don't think the NFS client behaviour will change unless the new "minorversion=1" mount option is used. It includes basic NFSv4.1 support plus support for pNFS using the Files Layout only. All problems detecting during an NFSv4.1 Bakeathon testing event in June 2012 have been resolved in this code and it has been tested against the NFSv4.1 server available to me. Although not reviewed, I believe that kib@ has looked at it.
2012-12-08 22:52:39 +00:00
}
/*
* Free a session slot.
*/
APPLESTATIC void
nfsv4_freeslot(struct nfsclsession *sep, int slot)
{
uint64_t bitval;
bitval = 1;
if (slot > 0)
bitval <<= slot;
mtx_lock(&sep->nfsess_mtx);
if ((bitval & sep->nfsess_slots) == 0)
printf("freeing free slot!!\n");
sep->nfsess_slots &= ~bitval;
wakeup(&sep->nfsess_slots);
mtx_unlock(&sep->nfsess_mtx);
}
Reference in New Issue Copy Permalink