kern_lock.c   [plain text]

/*
 * Copyright (c) 2000-2001 Apple Computer, Inc. All rights reserved.
 *
 * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
 * 
 * This file contains Original Code and/or Modifications of Original Code
 * as defined in and that are subject to the Apple Public Source License
 * Version 2.0 (the 'License'). You may not use this file except in
 * compliance with the License. The rights granted to you under the License
 * may not be used to create, or enable the creation or redistribution of,
 * unlawful or unlicensed copies of an Apple operating system, or to
 * circumvent, violate, or enable the circumvention or violation of, any
 * terms of an Apple operating system software license agreement.
 * 
 * Please obtain a copy of the License at
 * http://www.opensource.apple.com/apsl/ and read it before using this file.
 * 
 * The Original Code and all software distributed under the License are
 * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
 * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
 * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
 * Please see the License for the specific language governing rights and
 * limitations under the License.
 * 
 * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
 */
/* Copyright (c) 1995, 1997 Apple Computer, Inc. All Rights Reserved */
/* 
 * Copyright (c) 1995
 *	The Regents of the University of California.  All rights reserved.
 *
 * This code contains ideas from software contributed to Berkeley by
 * Avadis Tevanian, Jr., Michael Wayne Young, and the Mach Operating
 * System project at Carnegie-Mellon University.
 *
 * 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.
 *
 *	@(#)kern_lock.c	8.18 (Berkeley) 5/21/95
 */

#include <sys/param.h>
#include <sys/proc_internal.h>
#include <sys/lock.h>
#include <kern/cpu_number.h>
#include <kern/thread.h>

#include <mach/mach_types.h>

/*
 * Locking primitives implementation.
 * Locks provide shared/exclusive sychronization.
 */

#if 0
#define COUNT(p, x) if (p) (p)->p_locks += (x)
#else
#define COUNT(p, x)
#endif

#if NCPUS > 1

/*
 * For multiprocessor system, try spin lock first.
 *
 * This should be inline expanded below, but we cannot have #if
 * inside a multiline define.
 */
int lock_wait_time = 100;
#define PAUSE(lkp, wanted)						\
		if (lock_wait_time > 0) {				\
			int i;						\
									\
			for (i = lock_wait_time; i > 0; i--)		\
				if (!(wanted))				\
					break;				\
		}							\
		if (!(wanted))						\
			break;

#else /* NCPUS == 1 */

/*
 * It is an error to spin on a uniprocessor as nothing will ever cause
 * the simple lock to clear while we are executing.
 */
#define PAUSE(lkp, wanted)

#endif /* NCPUS == 1 */

/*
 * Acquire a resource.
 */
#define ACQUIRE(lkp, error, extflags, wanted)				\
	PAUSE(lkp, wanted);						\
	for (error = 0; wanted; ) {					\
		(lkp)->lk_waitcount++;					\
		error = tsleep((void *)lkp, (lkp)->lk_prio,		\
		    (lkp)->lk_wmesg, (lkp)->lk_timo);			\
		(lkp)->lk_waitcount--;					\
		if (error)						\
			break;						\
		if ((extflags) & LK_SLEEPFAIL) {			\
			error = ENOLCK;					\
			break;						\
		}							\
	}

/*
 * Initialize a lock; required before use.
 */
void
lockinit(lkp, prio, wmesg, timo, flags)
	struct lock__bsd__ *lkp;
	int prio;
	const char *wmesg;
	int timo;
	int flags;
{

	bzero(lkp, sizeof(struct lock__bsd__));
	lkp->lk_flags = flags & LK_EXTFLG_MASK;
	lkp->lk_prio = prio;
	lkp->lk_timo = timo;
	lkp->lk_wmesg = wmesg;
	lkp->lk_lockholder = LK_NOPROC;
	lkp->lk_lockthread = 0;
}

/*
 * Determine the status of a lock.
 */
int
lockstatus(lkp)
	struct lock__bsd__ *lkp;
{
	int lock_type = 0;

	if (lkp->lk_exclusivecount != 0)
		lock_type = LK_EXCLUSIVE;
	else if (lkp->lk_sharecount != 0)
		lock_type = LK_SHARED;
	return (lock_type);
}

/*
 * Set, change, or release a lock.
 *
 * Shared requests increment the shared count. Exclusive requests set the
 * LK_WANT_EXCL flag (preventing further shared locks), and wait for already
 * accepted shared locks and shared-to-exclusive upgrades to go away.
 */
int
lockmgr(lkp, flags, interlkp, p)
	struct lock__bsd__ *lkp;
	u_int flags;
	void * interlkp;
	struct proc *p;
{
	int error;
	pid_t pid;
	int extflags;
	void *self;

	error = 0; self = current_thread();
	if (p)
		pid = p->p_pid;
	else
		pid = LK_KERNPROC;
	extflags = (flags | lkp->lk_flags) & LK_EXTFLG_MASK;
#if 0
	/*
	 * Once a lock has drained, the LK_DRAINING flag is set and an
	 * exclusive lock is returned. The only valid operation thereafter
	 * is a single release of that exclusive lock. This final release
	 * clears the LK_DRAINING flag and sets the LK_DRAINED flag. Any
	 * further requests of any sort will result in a panic. The bits
	 * selected for these two flags are chosen so that they will be set
	 * in memory that is freed (freed memory is filled with 0xdeadbeef).
	 * The final release is permitted to give a new lease on life to
	 * the lock by specifying LK_REENABLE.
	 */
	if (lkp->lk_flags & (LK_DRAINING|LK_DRAINED)) {
		if (lkp->lk_flags & LK_DRAINED)
			panic("lockmgr: using decommissioned lock");
		if ((flags & LK_TYPE_MASK) != LK_RELEASE ||
		    (lkp->lk_lockholder != pid && lkp->lk_lockthread != self)
			panic("lockmgr: non-release on draining lock: %d\n",
			    flags & LK_TYPE_MASK);
		lkp->lk_flags &= ~LK_DRAINING;
		if ((flags & LK_REENABLE) == 0)
			lkp->lk_flags |= LK_DRAINED;
	}
#endif

	switch (flags & LK_TYPE_MASK) {

	case LK_SHARED:
		if (lkp->lk_lockholder != pid || lkp->lk_lockthread != self) {
			/*
			 * If just polling, check to see if we will block.
			 */
			if ((extflags & LK_NOWAIT) && (lkp->lk_flags &
			    (LK_HAVE_EXCL | LK_WANT_EXCL | LK_WANT_UPGRADE))) {
				error = EBUSY;
				break;
			}
			/*
			 * Wait for exclusive locks and upgrades to clear.
			 */
			ACQUIRE(lkp, error, extflags, lkp->lk_flags &
			    (LK_HAVE_EXCL | LK_WANT_EXCL | LK_WANT_UPGRADE));
			if (error)
				break;
			lkp->lk_sharecount++;
			COUNT(p, 1);
			break;
		}
		/*
		 * We hold an exclusive lock, so downgrade it to shared.
		 * An alternative would be to fail with EDEADLK.
		 */
		lkp->lk_sharecount++;
		COUNT(p, 1);
		/* fall into downgrade */

	case LK_DOWNGRADE:
		if (lkp->lk_lockholder != pid ||
				lkp->lk_lockthread != self ||
					lkp->lk_exclusivecount == 0)
			panic("lockmgr: not holding exclusive lock");
		lkp->lk_sharecount += lkp->lk_exclusivecount;
		lkp->lk_exclusivecount = 0;
		lkp->lk_flags &= ~LK_HAVE_EXCL;
		lkp->lk_lockholder = LK_NOPROC;
		lkp->lk_lockthread = 0;
		if (lkp->lk_waitcount)
			wakeup((void *)lkp);
		break;

	case LK_EXCLUPGRADE:
		/*
		 * If another process is ahead of us to get an upgrade,
		 * then we want to fail rather than have an intervening
		 * exclusive access.
		 */
		if (lkp->lk_flags & LK_WANT_UPGRADE) {
			lkp->lk_sharecount--;
			COUNT(p, -1);
			error = EBUSY;
			break;
		}
		/* fall into normal upgrade */

	case LK_UPGRADE:
		/*
		 * Upgrade a shared lock to an exclusive one. If another
		 * shared lock has already requested an upgrade to an
		 * exclusive lock, our shared lock is released and an
		 * exclusive lock is requested (which will be granted
		 * after the upgrade). If we return an error, the file
		 * will always be unlocked.
		 */
		if ((lkp->lk_lockholder == pid &&
				lkp->lk_lockthread == self) ||
						lkp->lk_sharecount <= 0)
			panic("lockmgr: upgrade exclusive lock");
		lkp->lk_sharecount--;
		COUNT(p, -1);
		/*
		 * If we are just polling, check to see if we will block.
		 */
		if ((extflags & LK_NOWAIT) &&
		    ((lkp->lk_flags & LK_WANT_UPGRADE) ||
		     lkp->lk_sharecount > 1)) {
			error = EBUSY;
			break;
		}
		if ((lkp->lk_flags & LK_WANT_UPGRADE) == 0) {
			/*
			 * We are first shared lock to request an upgrade, so
			 * request upgrade and wait for the shared count to
			 * drop to zero, then take exclusive lock.
			 */
			lkp->lk_flags |= LK_WANT_UPGRADE;
			ACQUIRE(lkp, error, extflags, lkp->lk_sharecount);
			lkp->lk_flags &= ~LK_WANT_UPGRADE;
			if (error)
				break;
			lkp->lk_flags |= LK_HAVE_EXCL;
			lkp->lk_lockholder = pid;
			lkp->lk_lockthread = self;
			if (lkp->lk_exclusivecount != 0)
				panic("lockmgr: non-zero exclusive count");
			lkp->lk_exclusivecount = 1;
			COUNT(p, 1);
			break;
		}
		/*
		 * Someone else has requested upgrade. Release our shared
		 * lock, awaken upgrade requestor if we are the last shared
		 * lock, then request an exclusive lock.
		 */
		if (lkp->lk_sharecount == 0 && lkp->lk_waitcount)
			wakeup((void *)lkp);
		/* fall into exclusive request */

	case LK_EXCLUSIVE:
		if (lkp->lk_lockholder == pid && lkp->lk_lockthread == self) {
			/*
			 *	Recursive lock.
			 */
			if ((extflags & LK_CANRECURSE) == 0)
				panic("lockmgr: locking against myself");
			lkp->lk_exclusivecount++;
			COUNT(p, 1);
			break;
		}
		/*
		 * If we are just polling, check to see if we will sleep.
		 */
		if ((extflags & LK_NOWAIT) && ((lkp->lk_flags &
		     (LK_HAVE_EXCL | LK_WANT_EXCL | LK_WANT_UPGRADE)) ||
		     lkp->lk_sharecount != 0)) {
			error = EBUSY;
			break;
		}
		/*
		 * Try to acquire the want_exclusive flag.
		 */
		ACQUIRE(lkp, error, extflags, lkp->lk_flags &
		    (LK_HAVE_EXCL | LK_WANT_EXCL));
		if (error)
			break;
		lkp->lk_flags |= LK_WANT_EXCL;
		/*
		 * Wait for shared locks and upgrades to finish.
		 */
		ACQUIRE(lkp, error, extflags, lkp->lk_sharecount != 0 ||
		       (lkp->lk_flags & LK_WANT_UPGRADE));
		lkp->lk_flags &= ~LK_WANT_EXCL;
		if (error)
			break;
		lkp->lk_flags |= LK_HAVE_EXCL;
		lkp->lk_lockholder = pid;
		lkp->lk_lockthread = self;
		if (lkp->lk_exclusivecount != 0)
			panic("lockmgr: non-zero exclusive count");
		lkp->lk_exclusivecount = 1;
		COUNT(p, 1);
		break;

	case LK_RELEASE:
		if (lkp->lk_exclusivecount != 0) {
			if (pid != lkp->lk_lockholder ||
					lkp->lk_lockthread != self)
				panic("lockmgr: pid %d, thread 0x%8x,"
					" not exclusive lock holder pid %d"
					" thread 0x%8x unlocking, exclusive count %d",
				    pid, self, lkp->lk_lockholder,
					lkp->lk_lockthread, lkp->lk_exclusivecount);
			lkp->lk_exclusivecount--;
			COUNT(p, -1);
			if (lkp->lk_exclusivecount == 0) {
				lkp->lk_flags &= ~LK_HAVE_EXCL;
				lkp->lk_lockholder = LK_NOPROC;
				lkp->lk_lockthread = 0;
			}
		} else if (lkp->lk_sharecount != 0) {
			lkp->lk_sharecount--;
			COUNT(p, -1);
		}
		if (lkp->lk_waitcount)
			wakeup((void *)lkp);
		break;

	case LK_DRAIN:
		/*
		 * Check that we do not already hold the lock, as it can 
		 * never drain if we do. Unfortunately, we have no way to
		 * check for holding a shared lock, but at least we can
		 * check for an exclusive one.
		 */
		if (lkp->lk_lockholder == pid && lkp->lk_lockthread == self)
			panic("lockmgr: draining against myself");
		/*
		 * If we are just polling, check to see if we will sleep.
		 */
		if ((extflags & LK_NOWAIT) && ((lkp->lk_flags &
		     (LK_HAVE_EXCL | LK_WANT_EXCL | LK_WANT_UPGRADE)) ||
		     lkp->lk_sharecount != 0 || lkp->lk_waitcount != 0)) {
			error = EBUSY;
			break;
		}
		PAUSE(lkp, ((lkp->lk_flags &
		     (LK_HAVE_EXCL | LK_WANT_EXCL | LK_WANT_UPGRADE)) ||
		     lkp->lk_sharecount != 0 || lkp->lk_waitcount != 0));
		for (error = 0; ((lkp->lk_flags &
		     (LK_HAVE_EXCL | LK_WANT_EXCL | LK_WANT_UPGRADE)) ||
		     lkp->lk_sharecount != 0 || lkp->lk_waitcount != 0); ) {
			lkp->lk_flags |= LK_WAITDRAIN;
			if (error = tsleep((void *)&lkp->lk_flags, lkp->lk_prio,
			    lkp->lk_wmesg, lkp->lk_timo))
				return (error);
			if ((extflags) & LK_SLEEPFAIL)
				return (ENOLCK);
		}
		lkp->lk_flags |= LK_DRAINING | LK_HAVE_EXCL;
		lkp->lk_lockholder = pid;
		lkp->lk_lockthread = self;
		lkp->lk_exclusivecount = 1;
		COUNT(p, 1);
		break;

	default:
		panic("lockmgr: unknown locktype request %d",
		    flags & LK_TYPE_MASK);
		/* NOTREACHED */
	}
	if ((lkp->lk_flags & LK_WAITDRAIN) && ((lkp->lk_flags &
	     (LK_HAVE_EXCL | LK_WANT_EXCL | LK_WANT_UPGRADE)) == 0 &&
	     lkp->lk_sharecount == 0 && lkp->lk_waitcount == 0)) {
		lkp->lk_flags &= ~LK_WAITDRAIN;
		wakeup((void *)&lkp->lk_flags);
	}
	return (error);
}

/*
 * Print out information about state of a lock. Used by VOP_PRINT
 * routines to display ststus about contained locks.
 */
void
lockmgr_printinfo(lkp)
	struct lock__bsd__ *lkp;
{

	if (lkp->lk_sharecount)
		printf(" lock type %s: SHARED (count %d)", lkp->lk_wmesg,
		    lkp->lk_sharecount);
	else if (lkp->lk_flags & LK_HAVE_EXCL)
		printf(" lock type %s: EXCL (count %d) by pid %d",
		    lkp->lk_wmesg, lkp->lk_exclusivecount, lkp->lk_lockholder);
	if (lkp->lk_waitcount > 0)
		printf(" with %d pending", lkp->lk_waitcount);
}