2 * Copyright (c) 1997 Berkeley Software Design, Inc. All rights reserved.
4 * Redistribution and use in source and binary forms, with or without
5 * modification, are permitted provided that the following conditions
7 * 1. Redistributions of source code must retain the above copyright
8 * notice, this list of conditions and the following disclaimer.
9 * 2. Redistributions in binary form must reproduce the above copyright
10 * notice, this list of conditions and the following disclaimer in the
11 * documentation and/or other materials provided with the distribution.
12 * 3. Berkeley Software Design Inc's name may not be used to endorse or
13 * promote products derived from this software without specific prior
16 * THIS SOFTWARE IS PROVIDED BY BERKELEY SOFTWARE DESIGN INC ``AS IS'' AND
17 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19 * ARE DISCLAIMED. IN NO EVENT SHALL BERKELEY SOFTWARE DESIGN INC BE LIABLE
20 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
21 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
22 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
23 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
24 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
25 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
28 * from BSDI nfs_lock.c,v 2.4 1998/12/14 23:49:56 jch Exp
32 #include <sys/param.h>
33 #include <sys/systm.h>
34 #include <sys/fcntl.h>
35 #include <sys/kernel.h> /* for hz */
37 #include <sys/malloc.h>
38 #include <sys/lockf.h> /* for hz */ /* Must come after sys/malloc.h */
40 #include <sys/mount.h>
41 #include <sys/namei.h>
43 #include <sys/resourcevar.h>
44 #include <sys/socket.h>
45 #include <sys/socket.h>
46 #include <sys/unistd.h>
47 #include <sys/vnode.h>
51 #include <nfs/rpcv2.h>
52 #include <nfs/nfsproto.h>
54 #include <nfs/nfsmount.h>
55 #include <nfs/nfsnode.h>
56 #include <nfs/nfs_lock.h>
57 #include <nfs/nlminfo.h>
59 #define NFSOWNER_1ST_LEVEL_START 1 /* initial entries */
60 #define NFSOWNER_2ND_LEVEL 256 /* some power of 2 */
62 #define NFSOWNER(tbl, i) \
63 (tbl)[(i) / NFSOWNER_2ND_LEVEL][(i) % NFSOWNER_2ND_LEVEL]
67 * We have to let the process know if the call succeeded. I'm using an extra
68 * field in the p_nlminfo field in the proc structure, as it is already for
74 * NFS advisory byte-level locks.
78 struct vop_advlock_args /* {
90 struct vnode *vp, *wvp;
100 * the NLM protocol doesn't allow the server to return an error
101 * on ranges, so we do it. Note that we should be returning
102 * EOVERFLOW in some cases, but we don't have it.
104 if (fl->l_start < 0 || fl->l_len < 0 ||
106 (fl->l_start + fl->l_len - 1) < 0)))
110 * Fill in the information structure.
112 msg.lm_version = LOCKD_MSG_VERSION;
113 msg.lm_msg_ident.pid = p->p_pid;
115 * if there is no nfsowner table yet, allocate one.
117 if (p->p_nlminfo == NULL) {
118 MALLOC(p->p_nlminfo, struct nlminfo *,
119 sizeof(struct nlminfo), M_LOCKF, M_WAITOK | M_ZERO);
120 p->p_nlminfo->pid_start = p->p_stats->p_start;
122 msg.lm_msg_ident.pid_start = p->p_nlminfo->pid_start;
123 msg.lm_msg_ident.msg_seq = ++(p->p_nlminfo->msg_seq);
126 msg.lm_wait = ap->a_flags & F_WAIT;
127 msg.lm_getlk = ap->a_op == F_GETLK;
129 * XXX -- I think this is wrong for anything other AF_INET.
131 msg.lm_addr = *(VFSTONFS(vp->v_mount)->nm_nam);
132 msg.lm_fh_len = NFS_ISV3(vp) ? VTONFS(vp)->n_fhsize : NFSX_V2FH;
133 bcopy(VTONFS(vp)->n_fhp, msg.lm_fh, msg.lm_fh_len);
134 msg.lm_nfsv3 = NFS_ISV3(vp);
135 msg.lm_cred = *(p->p_ucred);
138 * Open the lock fifo. If for any reason we don't find the fifo, it
139 * means that the lock daemon isn't running. Translate any missing
140 * file error message for the user, otherwise the application will
141 * complain that the user's file is missing, which isn't the case.
142 * Note that we use proc0's cred, so the fifo is opened as root.
144 NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, _PATH_LCKFIFO, p);
147 * XXX Hack to temporarily allow this process (regardless of it's creds)
148 * to open the fifo we need to write to. vn_open() really should
149 * take a ucred (and once it does, this code should be fixed to use
152 saved_uid = p->p_ucred->cr_uid;
153 p->p_ucred->cr_uid = 0; /* temporarly run the vn_open as root */
155 fmode = FFLAGS(O_WRONLY);
156 error = vn_open(&nd, &fmode, 0);
157 p->p_ucred->cr_uid = saved_uid;
159 return (error == ENOENT ? EOPNOTSUPP : error);
162 VOP_UNLOCK(wvp, 0, p); /* vn_open leaves it locked */
167 VOP_LEASE(wvp, p, proc0.p_ucred, LEASE_WRITE);
169 error = vn_rdwr(UIO_WRITE, wvp, (caddr_t)&msg, sizeof(msg), 0,
170 UIO_SYSSPACE, ioflg, proc0.p_ucred, NULL, p);
172 if (error && (((ioflg & IO_NDELAY) == 0) || error != EAGAIN)) {
176 * If we're locking a file, wait for an answer. Unlocks succeed
179 if (fl->l_type == F_UNLCK)
181 * XXX this isn't exactly correct. The client side
182 * needs to continue sending it's unlock until
183 * it gets a responce back.
188 * retry after 20 seconds if we haven't gotten a responce yet.
189 * This number was picked out of thin air... but is longer
190 * then even a reasonably loaded system should take (at least
191 * on a local network). XXX Probably should use a back-off
194 if ((error = tsleep((void *)p->p_nlminfo,
195 PCATCH | PUSER, "lockd", 20*hz)) != 0) {
196 if (error == EWOULDBLOCK) {
198 * We timed out, so we rewrite the request
199 * to the fifo, but only if it isn't already
209 if (msg.lm_getlk && p->p_nlminfo->retcode == 0) {
210 if (p->p_nlminfo->set_getlk_pid) {
211 fl->l_pid = p->p_nlminfo->getlk_pid;
213 fl->l_type = F_UNLCK;
216 error = p->p_nlminfo->retcode;
220 if ((error1 = vn_close(wvp, FWRITE, proc0.p_ucred, p)) && error == 0)
228 * NFS advisory byte-level locks answer from the lock daemon.
233 struct lockd_ans *ansp;
237 /* Let root, or someone who once was root (lockd generally
238 * switches to the daemon uid once it is done setting up) make
241 * XXX This authorization check is probably not right.
243 if ((error = suser(p)) != 0 && p->p_ucred->cr_svuid != 0)
246 /* the version should match, or we're out of sync */
247 if (ansp->la_vers != LOCKD_ANS_VERSION)
250 /* Find the process, set its return errno and wake it up. */
251 if ((p = pfind(ansp->la_msg_ident.pid)) == NULL)
254 /* verify the pid hasn't been reused (if we can), and it isn't waiting
255 * for an answer from a more recent request. We return an EPIPE if
256 * the match fails, because we've already used ESRCH above, and this
257 * is sort of like writing on a pipe after the reader has closed it.
259 if (p->p_nlminfo == NULL ||
260 ((ansp->la_msg_ident.msg_seq != -1) &&
261 (timevalcmp(&p->p_nlminfo->pid_start,
262 &ansp->la_msg_ident.pid_start, !=) ||
263 p->p_nlminfo->msg_seq != ansp->la_msg_ident.msg_seq))) {
268 p->p_nlminfo->retcode = ansp->la_errno;
269 p->p_nlminfo->set_getlk_pid = ansp->la_set_getlk_pid;
270 p->p_nlminfo->getlk_pid = ansp->la_getlk_pid;
272 (void)wakeup((void *)p->p_nlminfo);