On Nov 25, 2013, at 12:03 PM, David Edelsohn <dje....@gmail.com> wrote:
> 4) easy_altivec_constant, the comment about 32 bit should be removed
> because wide-int should remove the dependency on 32 bit vs 64 bit host
> wide int.
Index: gcc/ChangeLog.wide-int
===================================================================
--- gcc/ChangeLog.wide-int (revision 205366)
+++ gcc/ChangeLog.wide-int (working copy)
@@ -73,7 +73,7 @@
(splat_input_operand): Likewise.
* config/rs6000/rs6000.c
(num_insns_constant): Handle CONST_WIDE_INT.
- (easy_altivec_constant): Update comment.
+ (easy_altivec_constant): Remove comment.
(paired_expand_vector_init): Use CONSTANT_P.
(rs6000_legitimize_address): Handle CONST_WIDE_INT.
(rs6000_emit_move): Update checks.
Index: gcc/config/rs6000/rs6000.c
===================================================================
--- gcc/config/rs6000/rs6000.c (revision 205364)
+++ gcc/config/rs6000/rs6000.c (working copy)
@@ -5054,8 +5054,6 @@ easy_altivec_constant (rtx op, enum mach
if (mode == V2DImode)
{
- /* In case the compiler is built 32-bit, CONST_WIDE_INT
- constants are not easy. */
if (GET_CODE (CONST_VECTOR_ELT (op, 0)) != CONST_INT
|| GET_CODE (CONST_VECTOR_ELT (op, 1)) != CONST_INT)
return false;
> The rest looks good. I'd like to see the revised patch before approving.
I will send out an updated patch once all the issues you raise are resolved. I
wanted to send out them each as developed so you can see them in isolation.
