Index: sys/kern/subr_lock.c
===================================================================
--- sys/kern/subr_lock.c
+++ sys/kern/subr_lock.c
@@ -308,6 +308,32 @@
 		sched_relinquish(curthread);
 }
 
+static void
+lock_prof_reset_cpu_wait(int cpu)
+{
+	struct thread *td, *newtd;
+	struct pcpu *pcpu;
+
+	MPASS(td->td_critnest == 0);
+
+	/*
+	 * Observe all CPUs not executing in critical section.
+	 * We are not in one so the check for us is safe. If the found
+	 * thread changes to something else we know the section was
+	 * exited as well.
+	 */
+	pcpu = cpuid_to_pcpu[cpu];
+	td = pcpu->pc_curthread;
+	for (;;) {
+		if (td->td_critnest == 0)
+			break;
+		cpu_spinwait();
+		newtd = (struct thread *)atomic_load_acq_ptr((u_long *)pcpu->pc_curthread);
+		if (td != newtd)
+			break;
+	}
+}
+
 static void
 lock_prof_reset(void)
 {
@@ -324,7 +350,12 @@
 	atomic_store_rel_int(&lock_prof_resetting, 1);
 	enabled = lock_prof_enable;
 	lock_prof_enable = 0;
-	quiesce_all_cpus("profreset", 0);
+	/*
+	 * This both publishes lock_prof_enable as disabled and makes sure
+	 * everyone else reads it if they are not far enough. We wait for the
+	 * rest down below.
+	 */
+	cpus_fence_seq_cst();
 	/*
 	 * Some objects may have migrated between CPUs.  Clear all links
 	 * before we zero the structures.  Some items may still be linked
@@ -343,6 +374,9 @@
 		lock_prof_init_type(&lpc->lpc_types[0]);
 		lock_prof_init_type(&lpc->lpc_types[1]);
 	}
+	/*
+	 * Paired with the fence from cpus_fence_seq_cst()
+	 */
 	atomic_store_rel_int(&lock_prof_resetting, 0);
 	lock_prof_enable = enabled;
 }
@@ -433,12 +467,19 @@
 	    "max", "wait_max", "total", "wait_total", "count", "avg", "wait_avg", "cnt_hold", "cnt_lock", "name");
 	enabled = lock_prof_enable;
 	lock_prof_enable = 0;
-	quiesce_all_cpus("profstat", 0);
+	/*
+	 * See the comment in lock_prof_reset
+	 */
+	cpus_fence_seq_cst();
+	CPU_FOREACH(cpu) {
+		lock_prof_reset_cpu_wait(cpu);
+	}
 	t = ticks;
 	CPU_FOREACH(cpu) {
 		lock_prof_type_stats(&LP_CPU(cpu)->lpc_types[0], sb, 0, t);
 		lock_prof_type_stats(&LP_CPU(cpu)->lpc_types[1], sb, 1, t);
 	}
+	atomic_thread_fence_rel();
 	lock_prof_enable = enabled;
 
 	error = sbuf_finish(sb);
@@ -591,6 +632,10 @@
 	else
 		l->lpo_waittime = 0;
 out:
+	/*
+	 * Paired with cpus_fence_seq_cst().
+	 */
+	atomic_thread_fence_rel();
 	critical_exit();
 }
 
@@ -677,6 +722,10 @@
 	type = &LP_CPU_SELF->lpc_types[spin];
 	LIST_INSERT_HEAD(&type->lpt_lpoalloc, l, lpo_link);
 out:
+	/*
+	 * Paired with cpus_fence_seq_cst().
+	 */
+	atomic_thread_fence_rel();
 	critical_exit();
 }
 
Index: sys/kern/subr_smp.c
===================================================================
--- sys/kern/subr_smp.c
+++ sys/kern/subr_smp.c
@@ -929,6 +929,29 @@
 	return quiesce_cpus(all_cpus, wmesg, prio);
 }
 
+static void
+cpus_fence_seq_cst_issue(void *arg __unused)
+{
+
+	atomic_thread_fence_seq_cst();
+}
+
+void
+cpus_fence_seq_cst(void)
+{
+
+#ifdef SMP
+	smp_rendezvous(
+	    smp_no_rendezvous_barrier,
+	    cpus_fence_seq_cst_issue,
+	    smp_no_rendezvous_barrier,
+	    NULL
+	);
+#else
+	cpus_fence_seq_cst_issue(NULL);
+#endif
+}
+
 /* Extra care is taken with this sysctl because the data type is volatile */
 static int
 sysctl_kern_smp_active(SYSCTL_HANDLER_ARGS)
Index: sys/sys/smp.h
===================================================================
--- sys/sys/smp.h
+++ sys/sys/smp.h
@@ -264,6 +264,7 @@
 
 int	quiesce_all_cpus(const char *, int);
 int	quiesce_cpus(cpuset_t, const char *, int);
+void	cpus_fence_seq_cst(void);
 void	smp_no_rendezvous_barrier(void *);
 void	smp_rendezvous(void (*)(void *), 
 		       void (*)(void *),