disassemble_result.txt

(gdb) disas set_optimize_info_from_tree

(snip)

   0x6d2e8d67 <+1207>:  pop    %ebp
   0x6d2e8d68 <+1208>:  ret
   0x6d2e8d69 <+1209>:  lea    0x0(%esi,%eiz,1),%esi
   0x6d2e8d70 <+1216>:  movl   $0xffffffff,0x2c(%ebx)
   0x6d2e8d77 <+1223>:  jmp    0x6d2e8afb <onig_compile+587>
   0x6d2e8d7c <+1228>:  mov    $0xfffffffb,%eax
   0x6d2e8d81 <+1233>:  jmp    0x6d2e8d57 <onig_compile+1191>
   0x6d2e8d83 <+1235>:  mov    -0x12c(%ebp),%edx
   0x6d2e8d89 <+1241>:  mov    -0x128(%ebp),%eax
   0x6d2e8d8f <+1247>:  cmp    $0xffffffff,%edx
=> 0x6d2e8d92 <+1250>:  movdqa -0x118(%ebp),%xmm0
   0x6d2e8d9a <+1258>:  movdqu %xmm0,0x70(%ebx)
   0x6d2e8d9f <+1263>:  movl   $0x5,0x50(%ebx)
   0x6d2e8da6 <+1270>:  movdqa -0x108(%ebp),%xmm0
   0x6d2e8dae <+1278>:  movdqu %xmm0,0x80(%ebx)
   0x6d2e8db6 <+1286>:  movdqa -0xf8(%ebp),%xmm0
   0x6d2e8dbe <+1294>:  movdqu %xmm0,0x90(%ebx)
   0x6d2e8dc6 <+1302>:  movdqa -0xe8(%ebp),%xmm0
   0x6d2e8dce <+1310>:  movdqu %xmm0,0xa0(%ebx)
   0x6d2e8dd6 <+1318>:  movdqa -0xd8(%ebp),%xmm0
   0x6d2e8dde <+1326>:  movdqu %xmm0,0xb0(%ebx)
   0x6d2e8de6 <+1334>:  movdqa -0xc8(%ebp),%xmm0

(snip)

(gdb) info register ebp
ebp            0x376ea64        0x376ea64

--------------

0x376ea64 - 0x118

not 16 byte aligned


--------------

With volatile, movdqa(require 16byte aligned) is not used.


   0x6d2e8d61 <+1201>:  pop    %ebp
   0x6d2e8d62 <+1202>:  ret
   0x6d2e8d63 <+1203>:  movl   $0xffffffff,0x2c(%ebx)
   0x6d2e8d6a <+1210>:  jmp    0x6d2e8afb <onig_compile+587>
   0x6d2e8d6f <+1215>:  mov    $0xfffffffb,%eax
   0x6d2e8d74 <+1220>:  jmp    0x6d2e8d51 <onig_compile+1185>
   0x6d2e8d76 <+1222>:  mov    -0x12c(%ebp),%ecx
   0x6d2e8d7c <+1228>:  mov    -0x128(%ebp),%eax
   0x6d2e8d82 <+1234>:  movl   $0x0,-0x28c(%ebp)
   0x6d2e8d8c <+1244>:  mov    -0x28c(%ebp),%edx
   0x6d2e8d92 <+1250>:  cmp    $0xff,%edx
   0x6d2e8d98 <+1256>:  jg     0x6d2e8dd5 <onig_compile+1317>
   0x6d2e8d9a <+1258>:  lea    0x0(%esi),%esi
=> 0x6d2e8da0 <+1264>:  mov    -0x28c(%ebp),%esi
   0x6d2e8da6 <+1270>:  mov    -0x28c(%ebp),%edx
   0x6d2e8dac <+1276>:  movzbl -0x118(%ebp,%edx,1),%edx
   0x6d2e8db4 <+1284>:  mov    %dl,0x70(%ebx,%esi,1)
   0x6d2e8db8 <+1288>:  mov    -0x28c(%ebp),%edx
   0x6d2e8dbe <+1294>:  add    $0x1,%edx
   0x6d2e8dc1 <+1297>:  mov    %edx,-0x28c(%ebp)
   0x6d2e8dc7 <+1303>:  mov    -0x28c(%ebp),%edx
   0x6d2e8dcd <+1309>:  cmp    $0xff,%edx
   0x6d2e8dd3 <+1315>:  jle    0x6d2e8da0 <onig_compile+1264>
   0x6d2e8dd5 <+1317>:  cmp    $0xffffffff,%ecx
   0x6d2e8dd8 <+1320>:  movl   $0x5,0x50(%ebx)
   0x6d2e8ddf <+1327>:  mov    %ecx,0x178(%ebx)


--------------

With -mstackrealign

   0x6d2e8bb4 <+1172>:  pop    %ebp
   0x6d2e8bb5 <+1173>:  ret
   0x6d2e8bb6 <+1174>:  movl   $0xffffffff,0x2c(%ebx)
   0x6d2e8bbd <+1181>:  jmp    0x6d2e8959 <onig_compile+569>
   0x6d2e8bc2 <+1186>:  mov    $0xfffffffb,%eax
   0x6d2e8bc7 <+1191>:  jmp    0x6d2e8ba7 <onig_compile+1159>
   0x6d2e8bc9 <+1193>:  mov    0x19c(%esp),%edx
   0x6d2e8bd0 <+1200>:  mov    0x1a0(%esp),%eax
   0x6d2e8bd7 <+1207>:  cmp    $0xffffffff,%edx
=> 0x6d2e8bda <+1210>:  movdqa 0x1b0(%esp),%xmm0
   0x6d2e8be3 <+1219>:  movdqu %xmm0,0x70(%ebx)
   0x6d2e8be8 <+1224>:  movl   $0x5,0x50(%ebx)
   0x6d2e8bef <+1231>:  movdqa 0x1c0(%esp),%xmm0
   0x6d2e8bf8 <+1240>:  movdqu %xmm0,0x80(%ebx)
   0x6d2e8c00 <+1248>:  movdqa 0x1d0(%esp),%xmm0
   0x6d2e8c09 <+1257>:  movdqu %xmm0,0x90(%ebx)
   0x6d2e8c11 <+1265>:  movdqa 0x1e0(%esp),%xmm0
   0x6d2e8c1a <+1274>:  movdqu %xmm0,0xa0(%ebx)
   0x6d2e8c22 <+1282>:  movdqa 0x1f0(%esp),%xmm0
   0x6d2e8c2b <+1291>:  movdqu %xmm0,0xb0(%ebx)
   0x6d2e8c33 <+1299>:  movdqa 0x200(%esp),%xmm0
   0x6d2e8c3c <+1308>:  movdqu %xmm0,0xc0(%ebx)
   0x6d2e8c44 <+1316>:  movdqa 0x210(%esp),%xmm0
   0x6d2e8c4d <+1325>:  movdqu %xmm0,0xd0(%ebx)

(snip)

(gdb) info register esp
esp            0x28e430 0x28e430

16 byte aligned