int main()

{
	struct st {
		int aa:16;
		int bb:16;
	} s;
	s.aa = 1;
	s.bb = 2;
	return 0;
}
objdump -d:
elf64-x86-64
Disassembly of section .text:

0000000000000000 <main>:
   0:	55                   	push   %rbp
   1:	48 89 e5             	mov    %rsp,%rbp
   4:	48 81 ec 10 00 00 00 	sub    $0x10,%rsp
   b:	b8 01 00 00 00       	mov    $0x1,%eax
  10:	48 89 c1             	mov    %rax,%rcx
  13:	81 e1 ff ff 00 00    	and    $0xffff,%ecx
  19:	8b 55 fc             	mov    -0x4(%rbp),%edx
  1c:	81 e2 00 00 ff ff    	and    $0xffff0000,%edx
  22:	09 d1                	or     %edx,%ecx
  24:	89 4d fc             	mov    %ecx,-0x4(%rbp)
  27:	b8 02 00 00 00       	mov    $0x2,%eax
  2c:	48 89 c1             	mov    %rax,%rcx
  2f:	81 e1 ff ff 00 00    	and    $0xffff,%ecx
  35:	c1 e1 10             	shl    $0x10,%ecx
  38:	8b 55 fc             	mov    -0x4(%rbp),%edx
  3b:	81 e2 ff ff 00 00    	and    $0xffff,%edx
  41:	09 d1                	or     %edx,%ecx
  43:	89 4d fc             	mov    %ecx,-0x4(%rbp)
  46:	b8 00 00 00 00       	mov    $0x0,%eax
  4b:	e9 00 00 00 00       	jmpq   50 <main+0x50>
  50:	c9                   	leaveq
  51:	c3                   	retq

After the patch

Disassembly of section .text:

0000000000000000 <main>:
   0:	55                   	push   %rbp
   1:	48 89 e5             	mov    %rsp,%rbp
   4:	48 81 ec 10 00 00 00 	sub    $0x10,%rsp
   b:	8b 45 fc             	mov    -0x4(%rbp),%eax
   e:	81 e0 00 00 ff ff    	and    $0xffff0000,%eax
  14:	83 c8 01             	or     $0x1,%eax
  17:	89 45 fc             	mov    %eax,-0x4(%rbp)
  1a:	8b 45 fc             	mov    -0x4(%rbp),%eax
  1d:	81 e0 ff ff 00 00    	and    $0xffff,%eax
  23:	81 c8 00 00 02 00    	or     $0x20000,%eax
  29:	89 45 fc             	mov    %eax,-0x4(%rbp)
  2c:	b8 00 00 00 00       	mov    $0x0,%eax
  31:	e9 00 00 00 00       	jmpq   36 <main+0x36>
  36:	c9                   	leaveq
  37:	c3                   	retq

tccgen.c

@@ -1019,6 +1019,7 @@ ST_FUNC void lexpand_nr(void)
 }
 #endif
 
+#ifndef TCC_TARGET_X86_64
 /* build a long long from two ints */
 static void lbuild(int t)
 {
@@ -1027,6 +1028,7 @@ static void lbuild(int t)
     vtop[-1].type.t = t;
     vpop();
 }
+#endif
 
 /* rotate n first stack elements to the bottom 
    I1 ... In -> I2 ... In I1 [top is right]
@@ -1088,8 +1090,8 @@ static void gv_dup(void)
 {
     int rc, t, r, r1;
     SValue sv;
-
     t = vtop->type.t;
+#ifndef TCC_TARGET_X86_64
     if ((t & VT_BTYPE) == VT_LLONG) {
         lexpand();
         gv_dup();
@@ -1099,15 +1101,14 @@ static void gv_dup(void)
         vrotb(4);
         /* stack: H L L1 H1 */
         lbuild(t);
-        vrotb(3);
+        vrott(3);
-        vrotb(3);
         vswap();
         lbuild(t);
         vswap();
-    } else {
+    } else
+#endif
+    {
         /* duplicate value */
-        rc = RC_INT;
-        sv.type.t = VT_INT;
         if (is_float(t)) {
             rc = RC_FLOAT;
 #ifdef TCC_TARGET_X86_64
@@ -1115,8 +1116,9 @@ static void gv_dup(void)
                 rc = RC_ST0;
             }
 #endif
-            sv.type.t = t;
+        }else
-        }
+            rc = RC_INT;
+        sv.type.t = t;
         r = gv(rc);
         r1 = get_reg(rc);
         sv.r = r;
@@ -2549,11 +2551,6 @@ ST_FUNC void vstore(void)
         /* remove bit field info to avoid loops */
         vtop[-1].type.t = ft & ~(VT_BITFIELD | (-1 << VT_STRUCT_SHIFT));
 
-        /* duplicate source into other register */
-        gv_dup();
-        vswap();
-        vrott(3);
-
         if((ft & VT_BTYPE) == VT_BOOL) {
             gen_cast(&vtop[-1].type);
             vtop[-1].type.t = (vtop[-1].type.t & ~VT_BTYPE) | (VT_BYTE | VT_UNSIGNED);
@@ -2585,10 +2582,6 @@ ST_FUNC void vstore(void)
         gen_op('|');
         /* store result */
         vstore();
-
-        /* pop off shifted source from "duplicate source..." above */
-        vpop();
-
     } else {
 #ifdef CONFIG_TCC_BCHECK
         /* bound check case */