On Sat, Jul 9, 2011 at 6:20 PM, H.J. Lu <hjl.to...@gmail.com> wrote:
> On Sat, Jul 9, 2011 at 4:12 PM, Richard Henderson <r...@redhat.com> wrote:
>> On 07/09/2011 04:05 PM, H.J. Lu wrote:
>>>> Is this change actually necessary? I would think that the
>>>> predicate has already been adjusted...
>>>>
>>>
>>> Since we always use short version for x32, there is no need to call.
>>> x86_64_zext_immediate_operand.
>>
>> Yes, but using the shorter condition, i.e. always calling
>> x86_64_zext_immediate_operand and letting it return true for
>> all x32 symbols, is easier to maintain.
>
> I always updated x86_64_zext_immediate_operand for x32.
> I will remove TARGET_X32.
>
> Thanks.
Here is the updated patch. OK for trunk?
Thanks.
--
H.J.
---
2011-07-10 H.J. Lu <hongjiu...@intel.com>
* config/i386/i386.c (ix86_trampoline_init): Use movl instead
of movabs for x32.
diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c
index 04cb07d..1b02312 100644
--- a/gcc/config/i386/i386.c
+++ b/gcc/config/i386/i386.c
@@ -22721,7 +23034,7 @@ ix86_trampoline_init (rtx m_tramp, tree
fndecl, rtx chain_value)
}
else
{
- int offset = 0;
+ int offset = 0, size;
/* Load the function address to r11. Try to load address using
the shorter movl instead of movabs. We may want to support
@@ -22750,11 +23063,21 @@ ix86_trampoline_init (rtx m_tramp, tree
fndecl, rtx chain_value)
/* Load static chain using movabs to r10. */
mem = adjust_address (m_tramp, HImode, offset);
- emit_move_insn (mem, gen_int_mode (0xba49, HImode));
+ /* Use the shorter movl instead of movabs for x32. */
+ if (TARGET_X32)
+ {
+ size = 6;
+ emit_move_insn (mem, gen_int_mode (0xba41, HImode));
+ }
+ else
+ {
+ size = 10;
+ emit_move_insn (mem, gen_int_mode (0xba49, HImode));
+ }
- mem = adjust_address (m_tramp, DImode, offset + 2);
+ mem = adjust_address (m_tramp, ptr_mode, offset + 2);
emit_move_insn (mem, chain_value);
- offset += 10;
+ offset += size;
/* Jump to r11; the last (unused) byte is a nop, only there to
pad the write out to a single 32-bit store. */
