[Bug target/44532] New: x86-64 unnecessary parameter extension

[Bug target/44532] New: x86-64 unnecessary parameter extension

[astrange at ithinksw dot com]

Source:
int f1(short a, int b)
{
    return a * b;
}

int f2(unsigned short a, int b)
{
    return a * b;
}

> gcc -O3 -fomit-frame-pointer -S paramext.c

_f1:
LFB0:
        movl    %esi, %eax
        movswl  %di, %edi <-
        imull   %edi, %eax
        ret
...
_f2:
LFB1:
        movl    %esi, %eax
        movzwl  %di, %edi <-
        imull   %edi, %eax
        ret

AFAIK integer parameters should already be extended to int, so those
instructions are redundant. llvm doesn't generate them.


-- 
           Summary: x86-64 unnecessary parameter extension
           Product: gcc
           Version: 4.6.0
            Status: UNCONFIRMED
          Severity: enhancement
          Priority: P3
         Component: target
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: astrange at ithinksw dot com
  GCC host triplet: x86_64-*-*


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44532

[Bug target/44532] x86-64 unnecessary parameter extension

[jakub at gcc dot gnu dot org]

------- Comment #1 from jakub at gcc dot gnu dot org  2010-06-14 08:01 -------
      tempreg = convert_to_mode (data->nominal_mode, tempreg, unsignedp);

      if (GET_CODE (tempreg) == SUBREG
          && GET_MODE (tempreg) == data->nominal_mode
          && REG_P (SUBREG_REG (tempreg))
          && data->nominal_mode == data->passed_mode
          && GET_MODE (SUBREG_REG (tempreg)) == GET_MODE (data->entry_parm)
          && GET_MODE_SIZE (GET_MODE (tempreg))
             < GET_MODE_SIZE (GET_MODE (data->entry_parm)))
        {
          /* The argument is already sign/zero extended, so note it
             into the subreg.  */
          SUBREG_PROMOTED_VAR_P (tempreg) = 1;
          SUBREG_PROMOTED_UNSIGNED_SET (tempreg, unsignedp);
        }

For x86_64 and short/char, all the conditions above are true except that
nominal_mode (HImode) != passed_mode (SImode).


-- 

jakub at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |hubicka at gcc dot gnu dot
                   |                            |org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44532

[Bug target/44532] x86-64 unnecessary parameter extension

[rguenth at gcc dot gnu dot org]

------- Comment #2 from rguenth at gcc dot gnu dot org  2010-06-14 11:30 -------
Confirmed.


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
 GCC target triplet|                            |x86-64-*-*, i?86-*-*
           Keywords|                            |missed-optimization
   Last reconfirmed|0000-00-00 00:00:00         |2010-06-14 11:30:39
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44532

[Bug target/44532] x86-64 unnecessary parameter extension

[hjl dot tools at gmail dot com]

------- Comment #3 from hjl dot tools at gmail dot com  2010-06-15 00:57 -------
We should consider:

1. The x86-64 psABI doesn't say how char/short should be extended
as function parameters.
2. Gcc may not touch upper bits, PR 42324.
3. Gcc never depends on the upper bits which are properly extended
in function parameters.
4. Icc may not touch the upper bits in function parameters on stack.
5. For a function without proper prototype, caller may not properly
extend function parameters.

I think the current way is safer.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44532