Index: amd64/amd64/cpu_switch.S
===================================================================
RCS file: /usr/home/ncvs/src/sys/amd64/amd64/cpu_switch.S,v
retrieving revision 1.158
diff -u -p -r1.158 cpu_switch.S
--- amd64/amd64/cpu_switch.S	6 Jun 2007 07:35:07 -0000	1.158
+++ amd64/amd64/cpu_switch.S	7 Jun 2007 03:06:05 -0000
@@ -69,6 +69,9 @@ ENTRY(cpu_throw)
 	movq	TD_PROC(%rdi), %rdx		/* oldtd->td_proc */
 	movq	P_VMSPACE(%rdx), %rdx		/* proc->p_vmspace */
 	LK btrl	%eax, VM_PMAP+PM_ACTIVE(%rdx)	/* clear old */
+	movq	TD_LOCK(%rdi), %rdx
+	movq	$MTX_UNOWNED, %rdi
+	xchgq	%rdi, MTX_LOCK(%rdx)		/* Release lock */
 1:
 	movq	TD_PCB(%rsi),%rdx		/* newtd->td_proc */
 	movq	PCB_CR3(%rdx),%rdx
Index: amd64/amd64/genassym.c
===================================================================
RCS file: /usr/home/ncvs/src/sys/amd64/amd64/genassym.c,v
retrieving revision 1.163
diff -u -p -r1.163 genassym.c
--- amd64/amd64/genassym.c	6 Jun 2007 07:35:07 -0000	1.163
+++ amd64/amd64/genassym.c	7 Jun 2007 03:06:05 -0000
@@ -210,6 +210,7 @@ ASSYM(KUDSEL, GSEL(GUDATA_SEL, SEL_UPL))
 ASSYM(KUC32SEL, GSEL(GUCODE32_SEL, SEL_UPL));
 ASSYM(SEL_RPL_MASK, SEL_RPL_MASK);
 
+ASSYM(MTX_UNOWNED, MTX_UNOWNED);
 ASSYM(MTX_LOCK, offsetof(struct mtx, mtx_lock));
 ASSYM(MTX_RECURSECNT, offsetof(struct mtx, mtx_recurse));
 
Index: conf/NOTES
===================================================================
RCS file: /usr/home/ncvs/src/sys/conf/NOTES,v
retrieving revision 1.1428
diff -u -p -r1.1428 NOTES
--- conf/NOTES	5 Jun 2007 00:12:36 -0000	1.1428
+++ conf/NOTES	7 Jun 2007 03:06:05 -0000
@@ -186,6 +186,7 @@ options 	ROOTDEVNAME=\"ufs:da0s2e\"
 #
 options 	SCHED_4BSD
 #options 	SCHED_ULE
+#options	SCHED_SMP
 
 #####################################################################
 # SMP OPTIONS:
Index: conf/files
===================================================================
RCS file: /usr/home/ncvs/src/sys/conf/files,v
retrieving revision 1.1214
diff -u -p -r1.1214 files
--- conf/files	5 Jun 2007 00:12:36 -0000	1.1214
+++ conf/files	7 Jun 2007 03:06:05 -0000
@@ -1428,6 +1428,7 @@ kern/md5c.c			standard
 kern/p1003_1b.c			standard
 kern/posix4_mib.c		standard
 kern/sched_4bsd.c		optional sched_4bsd
+kern/sched_smp.c		optional sched_smp
 kern/sched_ule.c		optional sched_ule
 kern/serdev_if.m		standard
 kern/subr_acl_posix1e.c		standard
Index: conf/options
===================================================================
RCS file: /usr/home/ncvs/src/sys/conf/options,v
retrieving revision 1.590
diff -u -p -r1.590 options
--- conf/options	5 Jun 2007 00:12:36 -0000	1.590
+++ conf/options	7 Jun 2007 03:06:05 -0000
@@ -135,6 +135,7 @@ PPS_SYNC	opt_ntp.h
 PREEMPTION	opt_sched.h
 QUOTA
 SCHED_4BSD	opt_sched.h
+SCHED_SMP	opt_sched.h
 SCHED_ULE	opt_sched.h
 SHOW_BUSYBUFS
 SLEEPQUEUE_PROFILING
Index: kern/kern_synch.c
===================================================================
RCS file: /usr/home/ncvs/src/sys/kern/kern_synch.c,v
retrieving revision 1.300
diff -u -p -r1.300 kern_synch.c
--- kern/kern_synch.c	4 Jun 2007 23:50:56 -0000	1.300
+++ kern/kern_synch.c	7 Jun 2007 03:06:05 -0000
@@ -414,8 +414,6 @@ mi_switch(int flags, struct thread *newt
 	td->td_generation++;	/* bump preempt-detect counter */
 	PCPU_INC(cnt.v_swtch);
 	PCPU_SET(switchticks, ticks);
-	CTR4(KTR_PROC, "mi_switch: old thread %ld (kse %p, pid %ld, %s)",
-	    td->td_tid, td->td_sched, p->p_pid, p->p_comm);
 #if (KTR_COMPILE & KTR_SCHED) != 0
 	if (TD_IS_IDLETHREAD(td))
 		CTR3(KTR_SCHED, "mi_switch: %p(%s) prio %d idle",
@@ -443,9 +441,6 @@ mi_switch(int flags, struct thread *newt
 	CTR3(KTR_SCHED, "mi_switch: running %p(%s) prio %d",
 	    td, td->td_proc->p_comm, td->td_priority);
 
-	CTR4(KTR_PROC, "mi_switch: new thread %ld (kse %p, pid %ld, %s)",
-	    td->td_tid, td->td_sched, p->p_pid, p->p_comm);
-
 	/* 
 	 * If the last thread was exiting, finish cleaning it up.
 	 */
Index: kern/kern_thread.c
===================================================================
RCS file: /usr/home/ncvs/src/sys/kern/kern_thread.c,v
retrieving revision 1.247
diff -u -p -r1.247 kern_thread.c
--- kern/kern_thread.c	4 Jun 2007 23:52:24 -0000	1.247
+++ kern/kern_thread.c	7 Jun 2007 03:06:05 -0000
@@ -348,6 +348,7 @@ thread_exit(void)
 	struct thread *td;
 	struct thread *td2;
 	struct proc *p;
+	struct mtx *m;
 
 	td = curthread;
 	p = td->td_proc;
@@ -464,7 +465,12 @@ thread_exit(void)
 	thread_lock(td);
 	/* Aggregate our tick statistics into our parents rux. */
 	ruxagg(&p->p_rux, td);
-	PROC_SUNLOCK(p);
+	m = &p->p_slock;
+	thread_lock_set(td, m);
+	KASSERT(m->mtx_recurse == 0, ("thread_exit: proc slock recursed"));
+	WITNESS_UNLOCK(&m->lock_object, LOP_EXCLUSIVE, __FILE__, __LINE__);
+	LOCK_LOG_LOCK("UNLOCK", &m->lock_object, 0, m->mtx_recurse,
+	    __FILE__, __LINE__);
 	td->td_state = TDS_INACTIVE;
 	CTR1(KTR_PROC, "thread_exit: cpu_throw() thread %p", td);
 	sched_throw(td);
Index: kern/sched_4bsd.c
===================================================================
RCS file: /usr/home/ncvs/src/sys/kern/sched_4bsd.c,v
retrieving revision 1.101
diff -u -p -r1.101 sched_4bsd.c
--- kern/sched_4bsd.c	6 Jun 2007 03:40:46 -0000	1.101
+++ kern/sched_4bsd.c	7 Jun 2007 03:06:05 -0000
@@ -1387,22 +1387,20 @@ sched_idletd(void *dummy)
 void
 sched_throw(struct thread *td)
 {
+
+	mtx_lock_spin(&sched_lock);
 	/*
-	 * Correct spinlock nesting.  The idle thread context that we are
-	 * borrowing was created so that it would start out with a single
-	 * spin lock (sched_lock) held in fork_trampoline().  Since we've
-	 * explicitly acquired locks in this function, the nesting count
-	 * is now 2 rather than 1.  Since we are nested, calling
-	 * spinlock_exit() will simply adjust the counts without allowing
-	 * spin lock using code to interrupt us.
+	 * Correct spinlock nesting.  Two cases:
+	 * 1) The idle thread context that we are borrowing was created so that it would
+	 * start out with a single spin lock held in fork_trampoline().
+	 * 2) The exiting thread holds the per-process spinlock and this is pointed at
+	 * by td_lock.
+	 *
+	 * Since we've explicitly acquired locks in this function, the nesting count
+	 * is now 2 rather than 1.  Since we are nested, calling spinlock_exit() will
+	 * simply adjust the counts without allowing interrupts.
 	 */
-	if (td == NULL) {
-		mtx_lock_spin(&sched_lock);
-		spinlock_exit();
-	} else {
-		MPASS(td->td_lock == &sched_lock);
-	}
-	mtx_assert(&sched_lock, MA_OWNED);
+	spinlock_exit();
 	KASSERT(curthread->td_md.md_spinlock_count == 1, ("invalid count"));
 	PCPU_SET(switchtime, cpu_ticks());
 	PCPU_SET(switchticks, ticks);
Index: kern/sched_ule.c
===================================================================
RCS file: /usr/home/ncvs/src/sys/kern/sched_ule.c,v
retrieving revision 1.197
diff -u -p -r1.197 sched_ule.c
--- kern/sched_ule.c	6 Jun 2007 03:40:46 -0000	1.197
+++ kern/sched_ule.c	7 Jun 2007 03:06:05 -0000
@@ -2122,22 +2122,20 @@ sched_idletd(void *dummy)
 void
 sched_throw(struct thread *td)
 {
+
+	mtx_lock_spin(&sched_lock);
 	/*
-	 * Correct spinlock nesting.  The idle thread context that we are
-	 * borrowing was created so that it would start out with a single
-	 * spin lock (sched_lock) held in fork_trampoline().  Since we've
-	 * explicitly acquired locks in this function, the nesting count
-	 * is now 2 rather than 1.  Since we are nested, calling
-	 * spinlock_exit() will simply adjust the counts without allowing
-	 * spin lock using code to interrupt us.
+	 * Correct spinlock nesting.  Two cases:
+	 * 1) The idle thread context that we are borrowing was created so that it would
+	 * start out with a single spin lock held in fork_trampoline().
+	 * 2) The exiting thread holds the per-process spinlock and this is pointed at
+	 * by td_lock.
+	 *
+	 * Since we've explicitly acquired locks in this function, the nesting count
+	 * is now 2 rather than 1.  Since we are nested, calling spinlock_exit() will
+	 * simply adjust the counts without allowing interrupts.
 	 */
-	if (td == NULL) {
-		mtx_lock_spin(&sched_lock);
-		spinlock_exit();
-	} else {
-		MPASS(td->td_lock == &sched_lock);
-	}
-	mtx_assert(&sched_lock, MA_OWNED);
+	spinlock_exit();
 	KASSERT(curthread->td_md.md_spinlock_count == 1, ("invalid count"));
 	PCPU_SET(switchtime, cpu_ticks());
 	PCPU_SET(switchticks, ticks);