[Bug middle-end/23669] New: fold does convert (-a)/10 into a/-10.

[Bug middle-end/23669] New: fold does convert (-a)/10 into a/-10.

[pinskia at gcc dot gnu dot org]

Another idea from LLVM.
The following two functions should be the same asm:
int f(int a)
{
  return (-a)/10;
}
int f1(int a)
{
  return a/-10;
}

----
We currently get on PPC:
_f:
        lis r0,0x6666
        neg r3,r3
        ori r0,r0,26215
        mulhw r0,r3,r0
        srawi r3,r3,31
        srawi r0,r0,2
        subf r3,r3,r0
        blr
_f1:
        lis r0,0x6666
        ori r0,r0,26215
        mulhw r0,r3,r0
        srawi r3,r3,31
        srawi r0,r0,2
        subf r3,r0,r3
        blr

As you can see f1 has one less instruction which should be better.

-- 
           Summary: fold does convert (-a)/10 into a/-10.
           Product: gcc
           Version: 4.1.0
            Status: UNCONFIRMED
          Keywords: missed-optimization
          Severity: normal
          Priority: P2
         Component: middle-end
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: pinskia at gcc dot gnu dot org
                CC: gcc-bugs at gcc dot gnu dot org
OtherBugsDependingO 19987
             nThis:


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23669

[Bug middle-end/23669] fold does convert (-a)/10 into a/-10.

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |enhancement


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23669

[Bug middle-end/23669] fold does convert (-a)/10 into a/-10.

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-09-17 02:12 -------
Confirmed.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
   Last reconfirmed|0000-00-00 00:00:00         |2005-09-17 02:12:00
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23669

[Bug middle-end/23669] fold does convert (-a)/10 into a/-10.

[mrnobo1024 at yahoo dot com]

------- Additional Comments From mrnobo1024 at yahoo dot com  2005-10-01 17:25 -------
Those expressions are only the same for floating point types.

f(-2147483648)  == -214748364
f1(-2147483648) == 214748364

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23669

[Bug middle-end/23669] fold does convert (-a)/10 into a/-10 with -fno-wrapv

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-10-01 17:34 -------
(In reply to comment #2)
> Those expressions are only the same for floating point types.

But -a for INT_MIN is undefined for signed type since it is an overflow.  So the conversion is only valid 
for -fno-wrapv which is default for C, C++, Ada, and fortran.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|fold does convert (-a)/10   |fold does convert (-a)/10
                   |into a/-10.                 |into a/-10 with -fno-wrapv


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23669

[Bug middle-end/23669] fold does convert (-a)/10 into a/-10 with -fno-wrapv

[pinskia at gcc dot gnu dot org]

-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.2.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23669

[Bug middle-end/23669] fold does convert (-a)/10 into a/-10 with -fno-wrapv

[pinskia at gcc dot gnu dot org]

------- Comment #7 from pinskia at gcc dot gnu dot org  2005-11-26 22:18 -------
Fixed.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23669

[Bug middle-end/23669] fold does convert (-a)/10 into a/-10 with -fno-wrapv

[pinskia at gcc dot gnu dot org]

------- Comment #6 from pinskia at gcc dot gnu dot org  2005-11-26 22:18 -------
Subject: Bug 23669

Author: pinskia
Date: Sat Nov 26 22:18:04 2005
New Revision: 107543

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=107543
Log:
2005-11-26  Andrew Pinski  <pinskia@physics.uc.edu>

        PR middle-end/23669
        * fold-const.c (fold_binary): Convert -A/-B to A/B for signed types
        when overflow is undefined.
2005-11-26  Andrew Pinski  <pinskia@physics.uc.edu>

        PR middle-end/23669
        * gcc.dg/tree-ssa/divide-1.c: New test.
        * gcc.dg/tree-ssa/divide-2.c: New test.



Added:
    trunk/gcc/testsuite/gcc.dg/tree-ssa/divide-1.c
    trunk/gcc/testsuite/gcc.dg/tree-ssa/divide-2.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/fold-const.c
    trunk/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23669

[Bug middle-end/23669] fold does convert (-a)/10 into a/-10 with -fno-wrapv

[pinskia at gcc dot gnu dot org]

------- Comment #5 from pinskia at gcc dot gnu dot org  2005-11-26 18:08 -------
Patch posted:
http://gcc.gnu.org/ml/gcc-patches/2005-11/msg01850.html


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                URL|                            |http://gcc.gnu.org/ml/gcc-
                   |                            |patches/2005-
                   |                            |11/msg01850.html
           Keywords|                            |patch


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23669

[Bug middle-end/23669] fold does convert (-a)/10 into a/-10 with -fno-wrapv

[pinskia at gcc dot gnu dot org]

------- Comment #4 from pinskia at gcc dot gnu dot org  2005-11-25 06:17 -------
Ok, I have a patch which copies the code for Real divides into the int divides,
though it adds !flag_wrapv.  Though I am worry about the non truncate divide, I
think it should be the same for them but for some reason I am worried.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |pinskia at gcc dot gnu dot
                   |dot org                     |org
             Status|NEW                         |ASSIGNED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23669