diff --git a/sys/kern/kern_mutex.c b/sys/kern/kern_mutex.c
index 5d2146b7909c..b3e511611e20 100644
--- a/sys/kern/kern_mutex.c
+++ b/sys/kern/kern_mutex.c
@@ -289,6 +289,7 @@ __mtx_lock_spin_flags(volatile uintptr_t *c, int opts, const char *file,
     int line)
 {
 	struct mtx *m;
+	uintptr_t tid, v;
 
 	if (SCHEDULER_STOPPED())
 		return;
@@ -308,7 +309,14 @@ __mtx_lock_spin_flags(volatile uintptr_t *c, int opts, const char *file,
 	opts &= ~MTX_RECURSE;
 	WITNESS_CHECKORDER(&m->lock_object, opts | LOP_NEWORDER | LOP_EXCLUSIVE,
 	    file, line, NULL);
-	__mtx_lock_spin(m, curthread, opts, file, line);
+	spinlock_enter();
+	tid = (uintptr_t)curthread;
+	v = MTX_UNOWNED;
+	if (!_mtx_obtain_lock_fetch(m, &v, tid))
+		_mtx_lock_spin(m, v, opts, file, line);
+	else
+		LOCKSTAT_PROFILE_OBTAIN_LOCK_SUCCESS(spin__acquire,
+		    m, 0, 0, file, line);
 	LOCK_LOG_LOCK("LOCK", &m->lock_object, opts, m->mtx_recurse, file,
 	    line);
 	WITNESS_LOCK(&m->lock_object, opts | LOP_EXCLUSIVE, file, line);
@@ -682,12 +690,18 @@ _mtx_lock_spin_failed(struct mtx *m)
  * This is only called if we need to actually spin for the lock. Recursion
  * is handled inline.
  */
+#if LOCK_DEBUG > 0
 void
-_mtx_lock_spin_cookie(volatile uintptr_t *c, uintptr_t v, uintptr_t tid,
-    int opts, const char *file, int line)
+_mtx_lock_spin_cookie(volatile uintptr_t *c, uintptr_t v, int opts,
+    const char *file, int line)
+#else
+void
+_mtx_lock_spin_cookie(volatile uintptr_t *c, uintptr_t v)
+#endif
 {
 	struct mtx *m;
 	struct lock_delay_arg lda;
+	uintptr_t tid;
 #ifdef LOCK_PROFILING
 	int contested = 0;
 	uint64_t waittime = 0;
@@ -699,10 +713,7 @@ _mtx_lock_spin_cookie(volatile uintptr_t *c, uintptr_t v, uintptr_t tid,
 	int doing_lockprof;
 #endif
 
-	if (SCHEDULER_STOPPED())
-		return;
-
-	lock_delay_arg_init(&lda, &mtx_spin_delay);
+	tid = (uintptr_t)curthread;
 	m = mtxlock2mtx(c);
 
 	if (__predict_false(v == MTX_UNOWNED))
@@ -713,6 +724,11 @@ _mtx_lock_spin_cookie(volatile uintptr_t *c, uintptr_t v, uintptr_t tid,
 		return;
 	}
 
+	if (SCHEDULER_STOPPED())
+		return;
+
+	lock_delay_arg_init(&lda, &mtx_spin_delay);
+
 	if (LOCK_LOG_TEST(&m->lock_object, opts))
 		CTR1(KTR_LOCK, "_mtx_lock_spin: %p spinning", m);
 	KTR_STATE1(KTR_SCHED, "thread", sched_tdname((struct thread *)tid),
diff --git a/sys/sys/mutex.h b/sys/sys/mutex.h
index 359b1199714c..c37d4ebd2aaa 100644
--- a/sys/sys/mutex.h
+++ b/sys/sys/mutex.h
@@ -105,8 +105,12 @@ void	__mtx_unlock_sleep(volatile uintptr_t *c);
 #endif
 
 #ifdef SMP
-void	_mtx_lock_spin_cookie(volatile uintptr_t *c, uintptr_t v, uintptr_t tid,
-	    int opts, const char *file, int line);
+#if LOCK_DEBUG > 0
+void	_mtx_lock_spin_cookie(volatile uintptr_t *c, uintptr_t v, int opts,
+	    const char *file, int line);
+#else
+void	_mtx_lock_spin_cookie(volatile uintptr_t *c, uintptr_t v);
+#endif
 #endif
 void	__mtx_lock_flags(volatile uintptr_t *c, int opts, const char *file,
 	    int line);
@@ -154,8 +158,13 @@ void	thread_lock_flags_(struct thread *, int, const char *, int);
 	__mtx_unlock_sleep(&(m)->mtx_lock)
 #endif
 #ifdef SMP
-#define	_mtx_lock_spin(m, v, t, o, f, l)				\
-	_mtx_lock_spin_cookie(&(m)->mtx_lock, v, t, o, f, l)
+#if LOCK_DEBUG > 0
+#define	_mtx_lock_spin(m, v, o, f, l)					\
+	_mtx_lock_spin_cookie(&(m)->mtx_lock, v, o, f, l)
+#else
+#define	_mtx_lock_spin(m, v, o, f, l)					\
+	_mtx_lock_spin_cookie(&(m)->mtx_lock, v)
+#endif
 #endif
 #define	_mtx_lock_flags(m, o, f, l)					\
 	__mtx_lock_flags(&(m)->mtx_lock, o, f, l)
@@ -219,11 +228,9 @@ void	thread_lock_flags_(struct thread *, int, const char *, int);
 	uintptr_t _v = MTX_UNOWNED;					\
 									\
 	spinlock_enter();						\
-	if (!_mtx_obtain_lock_fetch((mp), &_v, _tid)) 			\
-		_mtx_lock_spin((mp), _v, _tid, (opts), (file), (line)); \
-	else 								\
-		LOCKSTAT_PROFILE_OBTAIN_LOCK_SUCCESS(spin__acquire,	\
-		    mp, 0, 0, file, line);				\
+	if (__predict_false(LOCKSTAT_PROFILE_ENABLED(spin__acquire) ||	\
+	    !_mtx_obtain_lock_fetch((mp), &_v, _tid))) 			\
+		_mtx_lock_spin((mp), _v, (opts), (file), (line)); 	\
 } while (0)
 #define __mtx_trylock_spin(mp, tid, opts, file, line) __extension__  ({	\
 	uintptr_t _tid = (uintptr_t)(tid);				\