PING: PATCH [4/n]: Prepare x32: Permute the conversion and addition if one operand is a constant

H.J. Lu hjl.tools@gmail.com
Sat Jul 9 21:41:00 GMT 2011


On Sat, Jul 9, 2011 at 2:18 PM, Paolo Bonzini <bonzini@gnu.org> wrote:
> On 07/05/2011 04:27 PM, H.J. Lu wrote:
>>>
>>>  diff --git a/gcc/explow.c b/gcc/explow.c
>>>  index 7387dad..b343bf8 100644
>>>  --- a/gcc/explow.c
>>>  +++ b/gcc/explow.c
>>>  @@ -383,18 +383,13 @@ convert_memory_address_addr_space (enum
>>> machine_mode to_mode ATTRIBUTE_UNUSED,
>>>
>>>      case PLUS:
>>>      case MULT:
>>>  -      /* For addition we can safely permute the conversion and addition
>>>  -        operation if one operand is a constant and converting the
>>> constant
>>>  -        does not change it or if one operand is a constant and we are
>>>  -        using a ptr_extend instruction  (POINTERS_EXTEND_UNSIGNED<  0).
>>>  -        We can always safely permute them if we are making the address
>>>  -        narrower.  */
>>>  +      /* For addition we safely permute the conversion and addition
>>>  +        operation if one operand is a constant since we can't generate
>>>  +        new instructions.  We can always safely permute them if we are
>>>  +        making the address narrower.  */
>>>        if (GET_MODE_SIZE (to_mode)<  GET_MODE_SIZE (from_mode)
>>>           || (GET_CODE (x) == PLUS
>>>  -&&  CONST_INT_P (XEXP (x, 1))
>>>  -&&  (XEXP (x, 1) == convert_memory_address_addr_space
>>>  -                                  (to_mode, XEXP (x, 1), as)
>>>  -                 || POINTERS_EXTEND_UNSIGNED<  0)))
>>>  +&&  CONST_INT_P (XEXP (x, 1))))
>>>         return gen_rtx_fmt_ee (GET_CODE (x), to_mode,
>>>                                convert_memory_address_addr_space
>>>                                  (to_mode, XEXP (x, 0), as),
>
> This does not seem safe to me.
>
> Paolo
>

The current code is broken for x32.  See:

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47727

We can't generate any new instructions.  Do you have any suggestions.

Thanks.

-- 
H.J.



More information about the Gcc-patches mailing list