diff --git a/lib/libthr/thread/thr_malloc.c b/lib/libthr/thread/thr_malloc.c
--- a/lib/libthr/thread/thr_malloc.c
+++ b/lib/libthr/thread/thr_malloc.c
@@ -34,6 +34,7 @@
 
 #include <sys/types.h>
 #include <sys/mman.h>
+#include <string.h>
 #include <rtld_malloc.h>
 #include "thr_private.h"
 
@@ -136,6 +137,28 @@
 	return (res);
 }
 
+void *
+__thr_calloc_aligned_cacheline(size_t nitems, size_t nbytes)
+{
+	struct pthread *curthread;
+	void *res;
+	size_t cnt;
+
+	cnt = nitems * nbytes;
+	if (nbytes != 0 && cnt / nbytes != nitems) {
+		/* size_t overflow. */
+		return (NULL);
+	}
+
+	curthread = _get_curthread();
+	thr_malloc_lock(curthread);
+	res = __crt_aligned_alloc(CACHE_LINE_SIZE, cnt);
+	thr_malloc_unlock(curthread);
+	if (res != NULL)
+		memset(res, 0, cnt);
+	return (res);
+}
+
 void
 __thr_malloc_prefork(struct pthread *curthread)
 {
diff --git a/lib/libthr/thread/thr_mutex.c b/lib/libthr/thread/thr_mutex.c
--- a/lib/libthr/thread/thr_mutex.c
+++ b/lib/libthr/thread/thr_mutex.c
@@ -291,8 +291,8 @@
 		if (error != 0)
 			return (error);
 	}
-	if ((pmutex = (pthread_mutex_t)calloc_cb(1,
-	    sizeof(struct pthread_mutex))) == NULL)
+	pmutex = calloc_cb(1, sizeof(struct pthread_mutex));
+	if (pmutex == NULL)
 		return (ENOMEM);
 	mutex_init_body(pmutex, attr);
 	*mutex = pmutex;
@@ -308,10 +308,10 @@
 
 	if (*mutex == THR_MUTEX_INITIALIZER)
 		ret = mutex_init(mutex, &_pthread_mutexattr_default,
-		    __thr_calloc);
+		    __thr_calloc_aligned_cacheline);
 	else if (*mutex == THR_ADAPTIVE_MUTEX_INITIALIZER)
 		ret = mutex_init(mutex, &_pthread_mutexattr_adaptive_default,
-		    __thr_calloc);
+		    __thr_calloc_aligned_cacheline);
 	else
 		ret = 0;
 	THR_LOCK_RELEASE(thread, &_mutex_static_lock);
@@ -395,7 +395,7 @@
 	    (*mutex_attr)->m_pshared == PTHREAD_PROCESS_PRIVATE) {
 		__thr_malloc_init();
 		return (mutex_init(mutex, mutex_attr ? *mutex_attr : NULL,
-		    __thr_calloc));
+		    __thr_calloc_aligned_cacheline));
 	}
 	pmtx = __thr_pshared_offpage(__DECONST(void *, mutex), 1);
 	if (pmtx == NULL)
diff --git a/lib/libthr/thread/thr_private.h b/lib/libthr/thread/thr_private.h
--- a/lib/libthr/thread/thr_private.h
+++ b/lib/libthr/thread/thr_private.h
@@ -1022,6 +1022,7 @@
 void *__thr_malloc(size_t nbytes);
 void *__thr_realloc(void *cp, size_t nbytes);
 void __thr_malloc_init(void);
+void *__thr_calloc_aligned_cacheline(size_t nitmes, size_t nbytes);
 void __thr_malloc_prefork(struct pthread *curthread);
 void __thr_malloc_postfork(struct pthread *curthread);
 
diff --git a/libexec/rtld-elf/rtld_malloc.h b/libexec/rtld-elf/rtld_malloc.h
--- a/libexec/rtld-elf/rtld_malloc.h
+++ b/libexec/rtld-elf/rtld_malloc.h
@@ -34,6 +34,7 @@
 #ifndef RTLD_MALLOC_H
 #define	RTLD_MALLOC_H
 
+void *__crt_aligned_alloc(size_t align, size_t size);
 void *__crt_calloc(size_t num, size_t size);
 void __crt_free(void *cp);
 void *__crt_malloc(size_t nbytes);
diff --git a/libexec/rtld-elf/rtld_malloc.c b/libexec/rtld-elf/rtld_malloc.c
--- a/libexec/rtld-elf/rtld_malloc.c
+++ b/libexec/rtld-elf/rtld_malloc.c
@@ -75,8 +75,8 @@
 union	overhead {
 	union	overhead *ov_next;	/* when free */
 	struct {
-		u_char	ovu_magic;	/* magic number */
-		u_char	ovu_index;	/* bucket # */
+		uint16_t ovu_index;	/* bucket # */
+		uint8_t ovu_magic;	/* magic number */
 	} ovu;
 #define	ov_magic	ovu.ovu_magic
 #define	ov_index	ovu.ovu_index
@@ -86,13 +86,15 @@
 static int morepages(int n);
 
 #define	MAGIC		0xef		/* magic # on accounting info */
+#define	AMAGIC		0xdf		/* magic # for aligned alloc */
 
 /*
  * nextf[i] is the pointer to the next free block of size
  * (FIRST_BUCKET_SIZE << i).  The overhead information precedes the data
  * area returned to the user.
  */
-#define	FIRST_BUCKET_SIZE	8
+#define	LOW_BITS		3
+#define	FIRST_BUCKET_SIZE	(1U << LOW_BITS)
 #define	NBUCKETS 30
 static	union overhead *nextf[NBUCKETS];
 
@@ -169,6 +171,26 @@
 	return (ret);
 }
 
+void *
+__crt_aligned_alloc(size_t align, size_t size)
+{
+	void *mem;
+	union overhead *ov;
+	uintptr_t x;
+
+	if (align < FIRST_BUCKET_SIZE)
+		align = FIRST_BUCKET_SIZE;
+	mem = __crt_malloc(size + align + sizeof(union overhead));
+	if (mem == NULL)
+		return (NULL);
+	x = roundup((uintptr_t)mem + sizeof(union overhead), align);
+	ov = cp2op((void *)x);
+	ov->ov_magic = AMAGIC;
+	ov->ov_index = (x - (uintptr_t)mem - sizeof(union overhead)) >>
+	    LOW_BITS;
+	return ((void *)x);
+}
+
 /*
  * Allocate more memory to the indicated bucket.
  */
@@ -216,6 +238,8 @@
   	if (cp == NULL)
   		return;
 	op = cp2op(cp);
+	if (op->ov_magic == AMAGIC)
+		op = (void *)((caddr_t)cp - (op->ov_index << LOW_BITS));
 	if (op->ov_magic != MAGIC)
 		return;				/* sanity */
   	size = op->ov_index;