from Rick Macklem
authorKirk McKusick <mckusick@ucbvax.Berkeley.EDU>
Tue, 7 Jan 1992 08:31:46 +0000 (00:31 -0800)
committerKirk McKusick <mckusick@ucbvax.Berkeley.EDU>
Tue, 7 Jan 1992 08:31:46 +0000 (00:31 -0800)
SCCS-vsn: sbin/mount_nfs/Makefile 5.1
SCCS-vsn: sbin/mount_nfs/mount_nfs.c 5.1
SCCS-vsn: sbin/mount_nfs/mount_nfs.8 5.1

usr/src/sbin/mount_nfs/Makefile [new file with mode: 0644]
usr/src/sbin/mount_nfs/mount_nfs.8 [new file with mode: 0644]
usr/src/sbin/mount_nfs/mount_nfs.c [new file with mode: 0644]

diff --git a/usr/src/sbin/mount_nfs/Makefile b/usr/src/sbin/mount_nfs/Makefile
new file mode 100644 (file)
index 0000000..225c058
--- /dev/null
@@ -0,0 +1,9 @@
+#      @(#)Makefile    5.1 (Berkeley) %G%
+
+PROG=  mount_nfs
+CFLAGS+=-DNFS
+MAN8=  mount_nfs.0
+DPADD= ${LIBRPC}
+LDADD= -lrpc
+
+.include <bsd.prog.mk>
diff --git a/usr/src/sbin/mount_nfs/mount_nfs.8 b/usr/src/sbin/mount_nfs/mount_nfs.8
new file mode 100644 (file)
index 0000000..d9c03af
--- /dev/null
@@ -0,0 +1,181 @@
+.\" Copyright (c) 1992 The Regents of the University of California.
+.\" All rights reserved.
+.\"
+.\" %sccs.include.redist.roff%
+.\"
+.\"    @(#)mount_nfs.8 5.1 (Berkeley) %G%
+.\"
+.Dd 
+.Dt MOUNT_NFS 8
+.Os BSD 4.4
+.Sh NAME
+.Nm mount_nfs
+.Nd mount nfs file systems
+.Sh SYNOPSIS
+.Nm mount_nfs
+.Op Fl bsiTMlqdckK
+.Op Fl R Ar retrycnt
+.Op Fl r Ar readsize
+.Op Fl w Ar writesize
+.Op Fl t Ar timeout
+.Op Fl x Ar retrans
+.Op Fl g Ar maxgroups
+.Op Fl a Ar maxreadahead
+.Op Fl L Ar leaseterm
+.Op Fl D Ar deadthresh
+.Op Fl m Ar realm
+.Ar rhost:path node
+.Sh DESCRIPTION
+The
+.Nm mount_nfs
+command
+calls the
+.Xr mount 2
+system call to prepare and graft a
+remote nfs file system
+(rhost:path)
+on to the file system tree at the point
+.Ar node.
+This command is normally executed by
+.Xr mount 8
+setting the options with the
+.Fl o
+flag.
+It implements the mount protocol as described in RFC 1094, Appendix A.
+.Pp
+The options are:
+.Bl -tag -width indent
+.It Fl b
+If an initial attempt to contact the server fails, fork off a child to keep
+trying the mount in the background. Useful for
+.Xr fstab 5
+where the filesystem mount is not critical to multiuser operation.
+.It Fl s
+A soft mount, which implies that file system calls will fail
+after \fBRetry\fR round trip timeout intervals.
+.It Fl i
+An interruptible mount, which implies that file system calls that are delayed
+due to an unresponsive server will fail with EINTR when a termination signal
+is posted for the process.
+.It Fl T
+Use TCP transport instead of UDP.
+This is recommended for servers that are not on the
+same LAN cable as the client.
+(NB: This is NOT supported by most non-BSD servers.)
+.It Fl M
+Assume that other clients are not writing a file concurrently with this client.
+This implements a slightly less strict
+consistency criteria than 4.3BSD Reno did,
+that is more in line with most commercial client implementations.
+This is recommended for servers that do not support leasing.
+.It Fl d
+Do not estimate retransmit timeout dynamically.
+This may be useful for UDP
+mounts that exhibit high retry rates.
+.It Fl c
+For UDP mount points, do not do a
+.Xr connect 2.
+This must be used for servers that do not reply to requests from the standard
+port number.
+.It Fl K
+Pass Kerberos authenticators to the server for
+client-to-server user-credential mapping.
+This may only be used over TCP mounts between 4.4BSD clients and
+servers.
+.It Fl q
+Use the leasing extensions to the protocol to maintain cache consistency.
+This protocol, referred to as Not Quite Nfs (NQNFS),
+is only supported by 4.4BSD servers.
+.It Fl l
+Used with NQNFS to specify that the
+\fBReaddir_and_Lookup\fR RPC should be used.
+This option reduces RPC traffic for cases such as ``ls -l'',
+but increases the lease load on the server.
+This is recommended unless the server is complaining
+about excessive lease load.
+.It Fl k
+Used with NQNFS to specify ``get a lease'' for the file name being looked up.
+This is recommended unless the server is complaining about excessive lease
+load.
+.El
+.Pp
+The following arguments take a value parameter that is either a decimal
+number or a character string specified as =<value> after the option flag.
+.Bl -tag -width indent
+.It Fl R
+Set the retry count for doing the mount to <value>.
+.It Fl x
+Set the retransmit timeout count for soft mounts to <value>.
+.It Fl r
+Set the read data size to <value>.
+It should be a power of 2 greater than 512.
+This should be used for UDP mounts when the
+``fragments dropped due to timeout''
+value is getting large while actively using a mount point.
+(Use
+.Xr netstat 1
+with the
+.FL s
+option to see what the ``fragments dropped due to timeout'' value is.)
+.It Fl w
+Set the write data size to <value>.
+Ditto the comments w.r.t. the
+.Fl r
+option, but using the ``fragments dropped due to timeout'' value on the
+server instead of the client.
+Note that both the
+.Fl r
+and
+.Fl w
+options should only be used as a last ditch effort at improving performance
+when mounting servers that do not support TCP mounts.
+.It Fl t
+Set the initial retransmit timeout to <value>.
+May be useful for fine tuning UDP mounts over internetworks
+with high packet loss rates or an overloaded server.
+Try increasing the interval if
+.Xr nfsstat 1
+shows high retransmit rates while the file system is active or reducing the
+value if there is a low retransmit rate but long response delay observed.
+.It Fl g
+Set the maximum size of the group list for the credentials to <value>.
+This should be used for mounts on old servers that cannot handle a group list
+size of 16, as specified in RFC 1057.
+Try 8, if users in a lot of groups
+cannot get response from the mount point.
+.It Fl a
+Set the read-ahead count to <value>.
+This may be in the range of 0 - 4 and determines how many blocks
+will be read ahead when a large file is being read sequentially.
+This is recommended for mounts with a large
+bandwidth * delay product.
+.It Fl L
+Used with NQNFS to set the lease term to <value> seconds.
+Only use this argument for mounts with a large round trip delay.
+Values are normally in the 10-30 second range.
+.It Fl D
+Used with NQNFS to set the ``dead server threshold'' to <value> round trip
+timeout intervals.
+After a ``dead server threshold'' of retransmit timeouts,
+cached data for the unresponsive server is assumed to still be valid.
+Values may be set in the range of 1 - 9, with 9 referring to an ``infinite
+dead threshold'' (i.e. never assume cached data still valid).
+This option is not generally recommended and
+is really an experimental feature.
+.It Fl m
+Set the Kerberos realm to the string argument. Used with the
+.Fl K
+option for mounts to other realms.
+.El
+.Sh SEE ALSO
+.Xr mount 2 ,
+.Xr unmount 2 ,
+.Xr fstab 5
+.Sh BUGS
+Due to the way that Sun RPC is implemented on top of UDP (unreliable datagram)
+transport, tuning such mounts is really a black art that can only be expected
+to have limited success.
+For clients mounting servers that are not on the same
+LAN cable or that tend to be overloaded,
+TCP transport is strongly recommended,
+but unfortunately this is restricted to mostly 4.4BSD servers.
diff --git a/usr/src/sbin/mount_nfs/mount_nfs.c b/usr/src/sbin/mount_nfs/mount_nfs.c
new file mode 100644 (file)
index 0000000..5f6c1ab
--- /dev/null
@@ -0,0 +1,480 @@
+/*
+ * Copyright (c) 1992 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.
+ *
+ * %sccs.include.redist.c%
+ */
+
+#ifndef lint
+char copyright[] =
+"@(#) Copyright (c) 1992 The Regents of the University of California.\n\
+ All rights reserved.\n";
+#endif /* not lint */
+
+#ifndef lint
+static char sccsid[] = "@(#)mount_nfs.c        5.1 (Berkeley) %G%";
+#endif /* not lint */
+
+#include <stdio.h>
+#include <signal.h>
+#include <strings.h>
+#include <sys/syslog.h>
+#include <sys/param.h>
+#include <sys/file.h>
+#include <sys/errno.h>
+#include <sys/ucred.h>
+#include <sys/mount.h>
+#include <sys/socket.h>
+#include <sys/socketvar.h>
+#include <netdb.h>
+#include <rpc/rpc.h>
+#include <rpc/pmap_clnt.h>
+#include <rpc/pmap_prot.h>
+#ifdef ISO
+#include <netiso/iso.h>
+#endif
+#ifdef KERBEROS
+#include <kerberosIV/krb.h>
+#endif
+#include <nfs/rpcv2.h>
+#include <nfs/nfsv2.h>
+#include <nfs/nfs.h>
+#include <nfs/nqnfs.h>
+
+int xdr_dir(), xdr_fh();
+struct nfs_args nfsdefargs = {
+       (struct sockaddr *)0,
+       sizeof (struct sockaddr_in),
+       SOCK_DGRAM,
+       0,
+       (nfsv2fh_t *)0,
+       0,
+       NFS_WSIZE,
+       NFS_RSIZE,
+       NFS_TIMEO,
+       NFS_RETRANS,
+       NFS_MAXGRPS,
+       NFS_DEFRAHEAD,
+       NQ_DEFLEASE,
+       NQ_DEADTHRESH,
+       (char *)0,
+};
+
+struct nfhret {
+       u_long  stat;
+       nfsv2fh_t nfh;
+};
+#define        DEF_RETRY       10000
+#define        BGRND   1
+#define        ISBGRND 2
+int retrycnt = DEF_RETRY;
+int opflags = 0;
+extern int errno;
+
+#ifdef ISO
+struct iso_addr *iso_addr();
+#endif
+
+#ifdef KERBEROS
+char inst[INST_SZ];
+char realm[REALM_SZ];
+KTEXT_ST kt;
+#endif
+
+main(argc, argv, arge)
+       int argc;
+       char **argv;
+       char **arge;
+{
+       struct nfs_args nfsargs;
+       register int c;
+       register struct nfs_args *nfsargsp = &nfsargs;
+       struct nfsd_cargs ncd;
+       int num, flags = 0, match = 1, i, nfssvc_flag;
+       char *spec, *name;
+       uid_t last_ruid = -1;
+       extern int optind;
+       extern char *optarg;
+
+#ifdef KERBEROS
+       strcpy(realm, KRB_REALM);
+#endif
+       nfsargs = nfsdefargs;
+       retrycnt = DEF_RETRY;
+       if (argc <= 1)
+               Usage(argc, argv);
+       while ((c = getopt(argc, argv, "bsiTpMlqdckF:R:r:w:t:x:g:a:L:D:Km:"))
+               != EOF)
+               switch (c) {
+               case 'b':
+                       opflags |= BGRND;
+                       break;
+               case 's':
+                       nfsargsp->flags |= NFSMNT_SOFT;
+                       break;
+               case 'i':
+                       nfsargsp->flags |= NFSMNT_INT;
+                       break;
+               case 'T':
+                       nfsargsp->sotype = SOCK_STREAM;
+                       break;
+#ifdef ISO
+               case 'p':
+                       nfsargsp->sotype = SOCK_SEQPACKET;
+                       break;
+#endif
+               case 'M':
+                       nfsargsp->flags |= NFSMNT_MYWRITE;
+                       break;
+               case 'l':
+                       nfsargsp->flags |= NFSMNT_RDIRALOOK;
+                       break;
+               case 'q':
+                       nfsargsp->flags |= NFSMNT_NQNFS;
+                       break;
+               case 'd':
+                       nfsargsp->flags |= NFSMNT_DUMBTIMR;
+                       break;
+               case 'c':
+                       nfsargsp->flags |= NFSMNT_NOCONN;
+                       break;
+               case 'k':
+                       nfsargsp->flags |= NFSMNT_NQLOOKLEASE;
+                       break;
+               case 'F':
+                       if ((num = atoi(optarg)) != 0)
+                               flags = num;
+                       break;
+               case 'R':
+                       if ((num = atoi(optarg)) > 0)
+                               retrycnt = num;
+                       break;
+               case 'r':
+                       if ((num = atoi(optarg)) > 0) {
+                               nfsargsp->rsize = num;
+                               nfsargsp->flags |= NFSMNT_RSIZE;
+                       }
+                       break;
+               case 'w':
+                       if ((num = atoi(optarg)) > 0) {
+                               nfsargsp->wsize = num;
+                               nfsargsp->flags |= NFSMNT_WSIZE;
+                       }
+                       break;
+               case 't':
+                       if ((num = atoi(optarg)) > 0) {
+                               nfsargsp->timeo = num;
+                               nfsargsp->flags |= NFSMNT_TIMEO;
+                       }
+                       break;
+               case 'x':
+                       if ((num = atoi(optarg)) > 0) {
+                               nfsargsp->retrans = num;
+                               nfsargsp->flags |= NFSMNT_RETRANS;
+                       }
+                       break;
+               case 'g':
+                       if ((num = atoi(optarg)) > 0) {
+                               nfsargsp->maxgrouplist = num;
+                               nfsargsp->flags |= NFSMNT_MAXGRPS;
+                       }
+                       break;
+               case 'a':
+                       if ((num = atoi(optarg)) >= 0) {
+                               nfsargsp->readahead = num;
+                               nfsargsp->flags |= NFSMNT_READAHEAD;
+                       }
+                       break;
+               case 'L':
+                       if ((num = atoi(optarg)) >= 2) {
+                               nfsargsp->leaseterm = num;
+                               nfsargsp->flags |= NFSMNT_LEASETERM;
+                       }
+                       break;
+               case 'D':
+                       if ((num = atoi(optarg)) > 0) {
+                               nfsargsp->deadthresh = num;
+                               nfsargsp->flags |= NFSMNT_DEADTHRESH;
+                       }
+                       break;
+#ifdef KERBEROS
+               case 'K':
+                       nfsargsp->flags |= NFSMNT_KERB;
+                       break;
+               case 'm':
+                       strncpy(realm, optarg, REALM_SZ - 1);
+                       realm[REALM_SZ - 1] = '\0';
+                       break;
+#endif /* KERBEROS */
+               default:
+                       Usage(argc, argv);
+               };
+       if ((argc - optind) == 2) {
+               spec = argv[optind];
+               name = argv[optind + 1];
+       } else
+               Usage(argc, argv);
+       if (getnfsargs(spec, nfsargsp)) {
+               if (mount(MOUNT_NFS, name, flags, nfsargsp))
+                       exit(1);
+               if (nfsargsp->flags & (NFSMNT_NQNFS | NFSMNT_KERB)) {
+                       if ((opflags & ISBGRND) == 0) {
+                               if (i = fork()) {
+                                       if (i == -1) {
+                                               perror("nqnfs");
+                                               exit(1);
+                                       }
+                                       exit();
+                               }
+                               (void) setsid();
+                               (void) close(0);
+                               (void) close(1);
+                               (void) close(2);
+                               (void) chdir("/");
+                       }
+                       openlog("mount_nfs:", LOG_PID, LOG_DAEMON);
+                       nfssvc_flag = NFSSVC_MNTD;
+                       ncd.ncd_dirp = name;
+                       while (nfssvc(nfssvc_flag, (caddr_t)&ncd) < 0) {
+                           if (errno == ENEEDAUTH) {
+syslog(LOG_ERR, "in eacces");
+                               nfssvc_flag = NFSSVC_MNTD | NFSSVC_GOTAUTH |
+                                       NFSSVC_AUTHINFAIL;
+#ifdef KERBEROS
+syslog(LOG_ERR,"Callin krb uid=%d inst=%s realm=%s",ncd.ncd_authuid,inst,realm);
+                               /*
+                                * Set up as ncd_authuid for the kerberos call.
+                                * Must set ruid to ncd_authuid and reset the
+                                * ticket name iff ncd_authuid is not the same
+                                * as last time, so that the right ticket file
+                                * is found.
+                                */
+                               if (ncd.ncd_authuid != last_ruid) {
+                                       krb_set_tkt_string("");
+                                       last_ruid = ncd.ncd_authuid;
+                               }
+                               setreuid(ncd.ncd_authuid, 0);
+                               if (krb_mk_req(&kt, "rcmd", inst, realm, 0) ==
+                                   KSUCCESS &&
+                                   kt.length <= (RPCAUTH_MAXSIZ - 2*NFSX_UNSIGNED)) {
+syslog(LOG_ERR,"Got it\n");
+                                   ncd.ncd_authtype = RPCAUTH_NQNFS;
+                                   ncd.ncd_authlen = kt.length;
+                                   ncd.ncd_authstr = (char *)kt.dat;
+                                   nfssvc_flag = NFSSVC_MNTD | NFSSVC_GOTAUTH;
+                               }
+                               setreuid(0, 0);
+syslog(LOG_ERR,"ktlen=%d\n", kt.length);
+#endif /* KERBEROS */
+                           } else
+                               syslog(LOG_ERR, "nfssvc err %m");
+                       }
+               }
+               exit();
+       } else
+               exit(1);
+}
+
+getnfsargs(spec, nfsargsp)
+       char *spec;
+       struct nfs_args *nfsargsp;
+{
+       register CLIENT *clp;
+       struct hostent *hp;
+       static struct sockaddr_in saddr;
+#ifdef ISO
+       static struct sockaddr_iso isoaddr;
+       struct iso_addr *isop;
+#endif
+       struct timeval pertry, try;
+       enum clnt_stat clnt_stat;
+       int so = RPC_ANYSOCK, isoflag = 0, i;
+       char *hostp, *delimp, *cp;
+       u_short tport;
+       static struct nfhret nfhret;
+       static char nam[MNAMELEN + 1];
+
+       strncpy(nam, spec, MNAMELEN);
+       nam[MNAMELEN] = '\0';
+       if ((delimp = index(spec, '@')) != NULL) {
+               hostp = delimp + 1;
+       } else if ((delimp = index(spec, ':')) != NULL) {
+               hostp = spec;
+               spec = delimp + 1;
+       } else {
+               fprintf(stderr,
+                   "No <host>:<dirpath> or <dirpath>@<host> spec\n");
+               return (0);
+       }
+       *delimp = '\0';
+       /*
+        * DUMB!! Until the mount protocol works on iso transport, we must
+        * supply both an iso and an inet address for the host.
+        */
+#ifdef ISO
+       if (!strncmp(hostp, "iso=", 4)) {
+               u_short isoport;
+
+               hostp += 4;
+               isoflag++;
+               if ((delimp = index(hostp, '+')) == NULL) {
+                       fprintf(stderr, "No iso+inet address\n");
+                       return (0);
+               }
+               *delimp = '\0';
+               if ((isop = iso_addr(hostp)) == NULL) {
+                       fprintf(stderr, "Bad iso address\n");
+                       return (0);
+               }
+               bzero((caddr_t)&isoaddr, sizeof (isoaddr));
+               bcopy((caddr_t)isop, (caddr_t)&isoaddr.siso_addr,
+                       sizeof (struct iso_addr));
+               isoaddr.siso_len = sizeof (isoaddr);
+               isoaddr.siso_family = AF_ISO;
+               isoaddr.siso_tlen = 2;
+               isoport = htons(NFS_PORT);
+               bcopy((caddr_t)&isoport, TSEL(&isoaddr), isoaddr.siso_tlen);
+               hostp = delimp + 1;
+       }
+#endif /* ISO */
+
+       /*
+        * Handle an internet host address and reverse resolve it if
+        * doing Kerberos.
+        */
+       if (isdigit(*hostp)) {
+               if ((saddr.sin_addr.s_addr = inet_addr(hostp)) == -1) {
+                       fprintf(stderr, "Bad net addr %s\n", hostp);
+                       return (0);
+               }
+               if ((nfsargsp->flags & NFSMNT_KERB) &&
+                   (hp = gethostbyaddr((char *)&saddr.sin_addr.s_addr,
+                       sizeof (u_long), AF_INET)) == (struct hostent *)0) {
+                       fprintf(stderr, "Can't reverse resolve net addr\n");
+                       return (0);
+               }
+       } else if ((hp = gethostbyname(hostp)) == NULL) {
+               fprintf(stderr, "Can't get net id for host\n");
+               return (0);
+       }
+#ifdef KERBEROS
+       if (nfsargsp->flags & NFSMNT_KERB) {
+               strncpy(inst, hp->h_name, INST_SZ);
+               inst[INST_SZ - 1] = '\0';
+               if (cp = index(inst, '.'))
+                       *cp = '\0';
+       }
+#endif /* KERBEROS */
+
+       bcopy(hp->h_addr, (caddr_t)&saddr.sin_addr, hp->h_length);
+       nfhret.stat = EACCES;   /* Mark not yet successful */
+       while (retrycnt > 0) {
+               saddr.sin_family = AF_INET;
+               saddr.sin_port = htons(PMAPPORT);
+               if ((tport = pmap_getport(&saddr, RPCPROG_NFS,
+                   NFS_VER2, IPPROTO_UDP)) == 0) {
+                       if ((opflags & ISBGRND) == 0)
+                               clnt_pcreateerror("NFS Portmap");
+               } else {
+                       saddr.sin_port = 0;
+                       pertry.tv_sec = 10;
+                       pertry.tv_usec = 0;
+                       if ((clp = clntudp_create(&saddr, RPCPROG_MNT,
+                           RPCMNT_VER1, pertry, &so)) == NULL) {
+                               if ((opflags & ISBGRND) == 0)
+                                       clnt_pcreateerror("Cannot MNT PRC");
+                       } else {
+                               clp->cl_auth = authunix_create_default();
+                               try.tv_sec = 10;
+                               try.tv_usec = 0;
+                               clnt_stat = clnt_call(clp, RPCMNT_MOUNT,
+                                   xdr_dir, spec, xdr_fh, &nfhret, try);
+                               if (clnt_stat != RPC_SUCCESS) {
+                                       if ((opflags & ISBGRND) == 0)
+                                               clnt_perror(clp, "Bad MNT RPC");
+                               } else {
+                                       auth_destroy(clp->cl_auth);
+                                       clnt_destroy(clp);
+                                       retrycnt = 0;
+                               }
+                       }
+               }
+               if (--retrycnt > 0) {
+                       if (opflags & BGRND) {
+                               opflags &= ~BGRND;
+                               if (i = fork()) {
+                                       if (i == -1) {
+                                               perror("nqnfs");
+                                               exit(1);
+                                       }
+                                       exit();
+                               }
+                               (void) setsid();
+                               (void) close(0);
+                               (void) close(1);
+                               (void) close(2);
+                               (void) chdir("/");
+                               opflags |= ISBGRND;
+                       } 
+                       sleep(60);
+               }
+       }
+       if (nfhret.stat) {
+               if (opflags & ISBGRND)
+                       exit(1);
+               fprintf(stderr, "Can't access %s: ", spec);
+               errno = nfhret.stat;
+               perror(NULL);
+               return (0);
+       }
+       saddr.sin_port = htons(tport);
+#ifdef ISO
+       if (isoflag) {
+               nfsargsp->addr = (struct sockaddr *) &isoaddr;
+               nfsargsp->addrlen = sizeof (isoaddr);
+       } else
+#endif /* ISO */
+       {
+               nfsargsp->addr = (struct sockaddr *) &saddr;
+               nfsargsp->addrlen = sizeof (saddr);
+       }
+       nfsargsp->fh = &nfhret.nfh;
+       nfsargsp->hostname = nam;
+       return (1);
+}
+
+/*
+ * xdr routines for mount rpc's
+ */
+xdr_dir(xdrsp, dirp)
+       XDR *xdrsp;
+       char *dirp;
+{
+       return (xdr_string(xdrsp, &dirp, RPCMNT_PATHLEN));
+}
+
+xdr_fh(xdrsp, np)
+       XDR *xdrsp;
+       struct nfhret *np;
+{
+       if (!xdr_u_long(xdrsp, &(np->stat)))
+               return (0);
+       if (np->stat)
+               return (1);
+       return (xdr_opaque(xdrsp, (caddr_t)&(np->nfh), NFSX_FH));
+}
+
+Usage(argc, argv)
+       int argc;
+       char *argv[];
+{
+       register int i;
+
+       for (i = 0; i < argc; i++)
+               fprintf(stderr, "%s ", argv[i]);
+       fprintf(stderr, "\nBad mount_nfs arg\n");
+       exit(1);
+}