[Bug c/32680] New: unneeded movsbl

[Bug c/32680] New: unneeded movsbl

[romikb at mail dot ru]

Code compiled with -O2 -fomit-frame-pointer

char function1(){ return 10; }
char (*pfunc)() = function1;
char function2(){ return pfunc(); }

produce unneeded movsbl instruction:

_function2:
        subl    $12, %esp << unneeded
        call    *_pfunc
        addl    $12, %esp << unneeded
        movsbl  %al,%eax << unneeded
        ret

eax register allready contains char and don't need to be converted.

Code for

long function1(){ return 10; }
long (*pfunc)() = function1;
lonf function2(){ return pfunc(); }

looks fine.


-- 
           Summary: unneeded movsbl
           Product: gcc
           Version: 4.1.2
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: romikb at mail dot ru


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32680

[Bug middle-end/32680] unneeded movsbl

[pinskia at gcc dot gnu dot org]

------- Comment #1 from pinskia at gcc dot gnu dot org  2007-07-08 20:29 -------
> eax register allready contains char 
not true but oh well, it is called promting for abi reasons.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|c                           |middle-end


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32680

[Bug middle-end/32680] unneeded movsbl

[romikb at mail dot ru]

------- Comment #2 from romikb at mail dot ru  2007-07-09 05:44 -------
(In reply to comment #1)
>> eax register allready contains char 
> not true

Can you explain, please, why it is not true?

Direct call of "function1" not produce "unneed movsb", so why indirect call
produce it?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32680

[Bug middle-end/32680] unneeded movsbl

[rguenth at gcc dot gnu dot org]

------- Comment #3 from rguenth at gcc dot gnu dot org  2007-07-09 09:50 -------
It does:

function2:
.LFB3:
        call    function1
        movsbl  %al,%eax
        ret

because this is what the C ABI specifies.  Now, the stack adjustments are
indeed superfluous.  And we fail to tailcall as well.  The latter is 
fixed with a patch of mine.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32680

[Bug middle-end/32680] fails to tailcall, useless stack adjustment

[rguenth at gcc dot gnu dot org]

------- Comment #4 from rguenth at gcc dot gnu dot org  2007-07-09 09:54 -------
So I just applied it and now we get

function1:
.LFB2:
        movl    $10, %eax
        ret

function2:
.LFB3:
        movq    pfunc(%rip), %r11
        jmp     *%r11

for the indirect call.  Fixed.


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |enhancement
             Status|UNCONFIRMED                 |RESOLVED
 GCC target triplet|                            |x86_64-*-*, i?86-*-*
           Keywords|                            |missed-optimization
         Resolution|                            |FIXED
            Summary|unneeded movsbl             |fails to tailcall, useless
                   |                            |stack adjustment
   Target Milestone|---                         |4.3.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32680