nfs_nqlease.c

早期freebsd实现

/*
 * Copyright (c) 1992, 1993
 *	The Regents of the University of California.  All rights reserved.
 *
 * This code is derived from software contributed to Berkeley by
 * Rick Macklem at The University of Guelph.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 3. All advertising materials mentioning features or use of this software
 *    must display the following acknowledgement:
 *	This product includes software developed by the University of
 *	California, Berkeley and its contributors.
 * 4. Neither the name of the University nor the names of its contributors
 *    may be used to endorse or promote products derived from this software
 *    without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 *
 *	@(#)nfs_nqlease.c	8.3 (Berkeley) 1/4/94
 */

/*
 * References:
 *	Cary G. Gray and David R. Cheriton, "Leases: An Efficient Fault-Tolerant
 *		Mechanism for Distributed File Cache Consistency",
 *		In Proc. of the Twelfth ACM Symposium on Operating Systems
 *		Principals, pg. 202-210, Litchfield Park, AZ, Dec. 1989.
 *	Michael N. Nelson, Brent B. Welch and John K. Ousterhout, "Caching
 *		in the Sprite Network File System", ACM TOCS 6(1),
 *		pages 134-154, February 1988.
 *	V. Srinivasan and Jeffrey C. Mogul, "Spritely NFS: Implementation and
 *		Performance of Cache-Consistency Protocols", Digital
 *		Equipment Corporation WRL Research Report 89/5, May 1989.
 */
#include <sys/param.h>
#include <sys/vnode.h>
#include <sys/mount.h>
#include <sys/kernel.h>
#include <sys/proc.h>
#include <sys/systm.h>
#include <sys/mbuf.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/file.h>
#include <sys/buf.h>
#include <sys/stat.h>
#include <sys/protosw.h>

#include <netinet/in.h>
#include <nfs/rpcv2.h>
#include <nfs/nfsv2.h>
#include <nfs/nfs.h>
#include <nfs/nfsm_subs.h>
#include <nfs/xdr_subs.h>
#include <nfs/nqnfs.h>
#include <nfs/nfsnode.h>
#include <nfs/nfsmount.h>

/*
 * List head for the lease queue and other global data.
 * At any time a lease is linked into a list ordered by increasing expiry time.
 */
#define	NQFHHASH(f)	((*((u_long *)(f)))&nqfheadhash)

union nqsrvthead nqthead;
struct nqlease **nqfhead;
u_long nqfheadhash;
time_t nqnfsstarttime = (time_t)0;
u_long nqnfs_prog, nqnfs_vers;
int nqsrv_clockskew = NQ_CLOCKSKEW;
int nqsrv_writeslack = NQ_WRITESLACK;
int nqsrv_maxlease = NQ_MAXLEASE;
int nqsrv_maxnumlease = NQ_MAXNUMLEASE;
void nqsrv_instimeq(), nqsrv_send_eviction(), nfs_sndunlock();
void nqsrv_unlocklease(), nqsrv_waitfor_expiry(), nfsrv_slpderef();
void nqsrv_addhost(), nqsrv_locklease(), nqnfs_serverd();
void nqnfs_clientlease();
struct mbuf *nfsm_rpchead();

/*
 * Signifies which rpcs can have piggybacked lease requests
 */
int nqnfs_piggy[NFS_NPROCS] = {
	0,
	NQL_READ,
	NQL_WRITE,
	0,
	NQL_READ,
	NQL_READ,
	NQL_READ,
	0,
	NQL_WRITE,
	0,
	0,
	0,
	0,
	0,
	0,
	0,
	NQL_READ,
	0,
	NQL_READ,
	0,
	0,
	0,
	0,
};

int nnnnnn = sizeof (struct nqlease);
int oooooo = sizeof (struct nfsnode);
extern nfstype nfs_type[9];
extern struct nfssvc_sock *nfs_udpsock, *nfs_cltpsock;
extern struct nfsd nfsd_head;
extern int nfsd_waiting;
extern struct nfsreq nfsreqh;

#define TRUE	1
#define	FALSE	0

/*
 * Get or check for a lease for "vp", based on NQL_CHECK flag.
 * The rules are as follows:
 * - if a current non-caching lease, reply non-caching
 * - if a current lease for same host only, extend lease
 * - if a read cachable lease and a read lease request
 *	add host to list any reply cachable
 * - else { set non-cachable for read-write sharing }
 *	send eviction notice messages to all other hosts that have lease
 *	wait for lease termination { either by receiving vacated messages
 *					from all the other hosts or expiry
 *					via. timeout }
 *	modify lease to non-cachable
 * - else if no current lease, issue new one
 * - reply
 * - return boolean TRUE iff nam should be m_freem()'d
 * NB: Since nqnfs_serverd() is called from a timer, any potential tsleep()
 *     in here must be framed by nqsrv_locklease() and nqsrv_unlocklease().
 *     nqsrv_locklease() is coded such that at least one of LC_LOCKED and
 *     LC_WANTED is set whenever a process is tsleeping in it. The exception
 *     is when a new lease is being allocated, since it is not in the timer
 *     queue yet. (Ditto for the splsoftclock() and splx(s) calls)
 */
nqsrv_getlease(vp, duration, flags, nd, nam, cachablep, frev, cred)
	struct vnode *vp;
	u_long *duration;
	int flags;
	struct nfsd *nd;
	struct mbuf *nam;
	int *cachablep;
	u_quad_t *frev;
	struct ucred *cred;
{
	register struct nqlease *lp, *lq, **lpp;
	register struct nqhost *lph;
	struct nqlease *tlp;
	struct nqm **lphp;
	struct vattr vattr;
	fhandle_t fh;
	int i, ok, error, s;

	if (vp->v_type != VREG && vp->v_type != VDIR && vp->v_type != VLNK)
		return (0);
	if (*duration > nqsrv_maxlease)
		*duration = nqsrv_maxlease;
	if (error = VOP_GETATTR(vp, &vattr, cred, nd->nd_procp))
		return (error);
	*frev = vattr.va_filerev;
	s = splsoftclock();
	tlp = vp->v_lease;
	if ((flags & NQL_CHECK) == 0)
		nfsstats.srvnqnfs_getleases++;
	if (tlp == (struct nqlease *)0) {

		/*
		 * Find the lease by searching the hash list.
		 */
		fh.fh_fsid = vp->v_mount->mnt_stat.f_fsid;
		if (error = VFS_VPTOFH(vp, &fh.fh_fid)) {
			splx(s);
			return (error);
		}
		lpp = &nqfhead[NQFHHASH(fh.fh_fid.fid_data)];
		for (lp = *lpp; lp; lp = lp->lc_fhnext)
			if (fh.fh_fsid.val[0] == lp->lc_fsid.val[0] &&
			    fh.fh_fsid.val[1] == lp->lc_fsid.val[1] &&
			    !bcmp(fh.fh_fid.fid_data, lp->lc_fiddata,
				  fh.fh_fid.fid_len - sizeof (long))) {
				/* Found it */
				lp->lc_vp = vp;
				vp->v_lease = lp;
				tlp = lp;
				break;
			}
	}
	lp = tlp;
	if (lp) {
		if ((lp->lc_flag & LC_NONCACHABLE) ||
		    (lp->lc_morehosts == (struct nqm *)0 &&
		     nqsrv_cmpnam(nd->nd_slp, nam, &lp->lc_host)))
			goto doreply;
		if ((flags & NQL_READ) && (lp->lc_flag & LC_WRITE)==0) {
			if (flags & NQL_CHECK)
				goto doreply;
			if (nqsrv_cmpnam(nd->nd_slp, nam, &lp->lc_host))
				goto doreply;
			i = 0;
			if (lp->lc_morehosts) {
				lph = lp->lc_morehosts->lpm_hosts;
				lphp = &lp->lc_morehosts->lpm_next;
				ok = 1;
			} else {
				lphp = &lp->lc_morehosts;
				ok = 0;
			}
			while (ok && (lph->lph_flag & LC_VALID)) {
				if (nqsrv_cmpnam(nd->nd_slp, nam, lph))
					goto doreply;
				if (++i == LC_MOREHOSTSIZ) {
					i = 0;
					if (*lphp) {
						lph = (*lphp)->lpm_hosts;
						lphp = &((*lphp)->lpm_next);
					} else
						ok = 0;
				} else
					lph++;
			}
			nqsrv_locklease(lp);
			if (!ok) {
				*lphp = (struct nqm *)
					malloc(sizeof (struct nqm),
						M_NQMHOST, M_WAITOK);
				bzero((caddr_t)*lphp, sizeof (struct nqm));
				lph = (*lphp)->lpm_hosts;
			}
			nqsrv_addhost(lph, nd->nd_slp, nam);
			nqsrv_unlocklease(lp);
		} else {
			lp->lc_flag |= LC_NONCACHABLE;
			nqsrv_locklease(lp);
			nqsrv_send_eviction(vp, lp, nd->nd_slp, nam, cred);
			nqsrv_waitfor_expiry(lp);
			nqsrv_unlocklease(lp);
		}
doreply:
		/*
		 * Update the lease and return
		 */
		if ((flags & NQL_CHECK) == 0)
			nqsrv_instimeq(lp, *duration);
		if (lp->lc_flag & LC_NONCACHABLE)
			*cachablep = 0;
		else {
			*cachablep = 1;
			if (flags & NQL_WRITE)
				lp->lc_flag |= LC_WRITTEN;
		}
		splx(s);
		return (0);
	}
	splx(s);
	if (flags & NQL_CHECK)
		return (0);

	/*
	 * Allocate new lease
	 * The value of nqsrv_maxnumlease should be set generously, so that
	 * the following "printf" happens infrequently.
	 */
	if (nfsstats.srvnqnfs_leases > nqsrv_maxnumlease) {
		printf("Nqnfs server, too many leases\n");
		do {
			(void) tsleep((caddr_t)&lbolt, PSOCK,
					"nqsrvnuml", 0);
		} while (nfsstats.srvnqnfs_leases > nqsrv_maxnumlease);
	}
	MALLOC(lp, struct nqlease *, sizeof (struct nqlease), M_NQLEASE, M_WAITOK);
	bzero((caddr_t)lp, sizeof (struct nqlease));
	if (flags & NQL_WRITE)
		lp->lc_flag |= (LC_WRITE | LC_WRITTEN);
	nqsrv_addhost(&lp->lc_host, nd->nd_slp, nam);
	lp->lc_vp = vp;
	lp->lc_fsid = fh.fh_fsid;
	bcopy(fh.fh_fid.fid_data, lp->lc_fiddata, fh.fh_fid.fid_len - sizeof (long));
	if (lq = *lpp)
		lq->lc_fhprev = &lp->lc_fhnext;
	lp->lc_fhnext = lq;
	lp->lc_fhprev = lpp;
	*lpp = lp;
	vp->v_lease = lp;
	s = splsoftclock();
	nqsrv_instimeq(lp, *duration);
	splx(s);
	*cachablep = 1;
	if (++nfsstats.srvnqnfs_leases > nfsstats.srvnqnfs_maxleases)
		nfsstats.srvnqnfs_maxleases = nfsstats.srvnqnfs_leases;
	return (0);
}

/*
 * Local lease check for server syscalls.
 * Just set up args and let nqsrv_getlease() do the rest.
 */
void
lease_check(vp, p, cred, flag)
	struct vnode *vp;
	struct proc *p;
	struct ucred *cred;
	int flag;
{
	int duration = 0, cache;
	struct nfsd nfsd;
	u_quad_t frev;

	nfsd.nd_slp = NQLOCALSLP;
	nfsd.nd_procp = p;
	(void) nqsrv_getlease(vp, &duration, NQL_CHECK | flag, &nfsd,
		(struct mbuf *)0, &cache, &frev, cred);
}

/*
 * Add a host to an nqhost structure for a lease.
 */
void
nqsrv_addhost(lph, slp, nam)
	register struct nqhost *lph;
	struct nfssvc_sock *slp;
	struct mbuf *nam;
{
	register struct sockaddr_in *saddr;

	if (slp == NQLOCALSLP)
		lph->lph_flag |= (LC_VALID | LC_LOCAL);
	else if (slp == nfs_udpsock) {
		saddr = mtod(nam, struct sockaddr_in *);
		lph->lph_flag |= (LC_VALID | LC_UDP);
		lph->lph_inetaddr = saddr->sin_addr.s_addr;
		lph->lph_port = saddr->sin_port;
	} else if (slp == nfs_cltpsock) {
		lph->lph_nam = m_copym(nam, 0, M_COPYALL, M_WAIT);
		lph->lph_flag |= (LC_VALID | LC_CLTP);
	} else {
		lph->lph_flag |= (LC_VALID | LC_SREF);
		lph->lph_slp = slp;
		slp->ns_sref++;
	}
}

/*
 * Update the lease expiry time and position it in the timer queue correctly.
 */
void
nqsrv_instimeq(lp, duration)
	register struct nqlease *lp;
	u_long duration;
{
	register struct nqlease *tlp;
	time_t newexpiry;

	newexpiry = time.tv_sec + duration + nqsrv_clockskew;
	if (lp->lc_expiry == newexpiry)
		return;
	if (lp->lc_chain1[0])
		remque(lp);
	lp->lc_expiry = newexpiry;

	/*
	 * Find where in the queue it should be.
	 */
	tlp = nqthead.th_chain[1];
	while (tlp->lc_expiry > newexpiry && tlp != (struct nqlease *)&nqthead)
		tlp = tlp->lc_chain1[1];
	if (tlp == nqthead.th_chain[1])
		NQSTORENOVRAM(newexpiry);
	insque(lp, tlp);
}

/*
 * Compare the requesting host address with the lph entry in the lease.
 * Return true iff it is the same.
 * This is somewhat messy due to the union in the nqhost structure.
 * The local host is indicated by the special value of NQLOCALSLP for slp.
 */
nqsrv_cmpnam(slp, nam, lph)
	register struct nfssvc_sock *slp;
	struct mbuf *nam;
	register struct nqhost *lph;
{
	register struct sockaddr_in *saddr;
	struct mbuf *addr;
	union nethostaddr lhaddr;
	int ret;