Index: libexec/rtld-elf/amd64/reloc.c
===================================================================
--- libexec/rtld-elf/amd64/reloc.c
+++ libexec/rtld-elf/amd64/reloc.c
@@ -547,30 +547,23 @@
 	return (tls_get_addr_common(dtvp, ti->ti_module, ti->ti_offset));
 }
 
-size_t
-calculate_first_tls_offset(size_t size, size_t align, size_t offset)
-{
-	size_t res;
-
-	res = roundup(size, align);
-	offset &= align - 1;
-	if (offset != 0)
-		res += align - offset;
-	return (res);
-}
-
 size_t
 calculate_tls_offset(size_t prev_offset, size_t prev_size __unused, size_t size,
     size_t align, size_t offset)
 {
-	size_t res;
-
-	res = roundup(prev_offset + size, align);
-	offset &= align - 1;
-	if (offset != 0)
-		res += align - offset;
+	// res is the smallest integer satisfying res-prev_offset >= size and (-res)%p_align =
+	// p_vaddr%p_align (= p_offset%p_align).
+	size_t res = prev_offset + size + align - 1;
+	res -= (res + offset) & (align - 1);
 	return (res);
 }
+
+size_t
+calculate_first_tls_offset(size_t size, size_t align, size_t offset)
+{
+	return calculate_tls_offset(0, 0, size, align, offset);
+}
+
 size_t
 calculate_tls_end(size_t off, size_t size __unused)
 {
Index: libexec/rtld-elf/i386/reloc.c
===================================================================
--- libexec/rtld-elf/i386/reloc.c
+++ libexec/rtld-elf/i386/reloc.c
@@ -538,30 +538,23 @@
 	return (tls_get_addr_common(dtvp, ti->ti_module, ti->ti_offset));
 }
 
-size_t
-calculate_first_tls_offset(size_t size, size_t align, size_t offset)
-{
-	size_t res;
-
-	res = roundup(size, align);
-	offset &= align - 1;
-	if (offset != 0)
-		res += align - offset;
-	return (res);
-}
-
 size_t
 calculate_tls_offset(size_t prev_offset, size_t prev_size __unused, size_t size,
     size_t align, size_t offset)
 {
-	size_t res;
-
-	res = roundup(prev_offset + size, align);
-	offset &= align - 1;
-	if (offset != 0)
-		res += align - offset;
+	// res is the smallest integer satisfying res-prev_offset >= size and (-res)%p_align =
+	// p_vaddr%p_align (= p_offset%p_align).
+	size_t res = prev_offset + size + align - 1;
+	res -= (res + offset) & (align - 1);
 	return (res);
 }
+
+size_t
+calculate_first_tls_offset(size_t size, size_t align, size_t offset)
+{
+	return calculate_tls_offset(0, 0, size, align, offset);
+}
+
 size_t
 calculate_tls_end(size_t off, size_t size __unused)
 {