[Bug other/27567] New: __builtin_memcpy generates redundant stores/moves.

[Bug other/27567] New: __builtin_memcpy generates redundant stores/moves.

[pluto at agmk dot net]

float convert( unsigned in )
{
        float f;
        __builtin_memcpy( &f, &in, sizeof( in ) );
        return f;
}

$ gcc bug.c -S -O2 -fverbose-asm -m64

convert:
        movl    %edi, -4(%rsp)  # in, f
        movl    %edi, -20(%rsp) # in, in      <=== what for this?
        movss   -4(%rsp), %xmm0 # f, <result>
        ret

$ gcc bug.c -S -O2 -fverbose-asm -m32

convert:
        pushl   %ebp    #
        movl    %esp, %ebp      #,
        subl    $16, %esp       #,
        movl    8(%ebp), %eax   # in, in   \
        movl    %eax, -4(%ebp)  # in, f    |--- why not flds 8(%ebp) ?
        flds    -4(%ebp)        # f        /
        leave
        ret


-- 
           Summary: __builtin_memcpy generates redundant stores/moves.
           Product: gcc
           Version: 4.1.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: other
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: pluto at agmk dot net
 GCC build triplet: x86-*-linux
  GCC host triplet: x86-*-linux
GCC target triplet: x86-*-linux


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27567

[Bug rtl-optimization/27567] [4.0/4.1/4.2 Regression] __builtin_memcpy generates redundant stores/moves.

[rguenth at gcc dot gnu dot org]

------- Comment #1 from rguenth at gcc dot gnu dot org  2006-05-12 08:27 -------
f is marked addressable and assigned a stack slot.  That the stores are not
optimized away later is probably due to aliasing issues - an open-coded memcpy
must behave like the -fno-strict-aliasing case, so I guess we applied measures
to assure this and this prevents this optimization.  3.3 and 3.3-hammer can do
the optimization.


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |rguenth at gcc dot gnu dot
                   |                            |org, hubicka at gcc dot gnu
                   |                            |dot org
             Status|UNCONFIRMED                 |NEW
          Component|other                       |rtl-optimization
     Ever Confirmed|0                           |1
           Keywords|                            |missed-optimization
      Known to fail|                            |3.4.6 4.0.3 4.1.0 4.2.0
      Known to work|                            |3.3.6
   Last reconfirmed|0000-00-00 00:00:00         |2006-05-12 08:27:14
               date|                            |
            Summary|__builtin_memcpy generates  |[4.0/4.1/4.2 Regression]
                   |redundant stores/moves.     |__builtin_memcpy generates
                   |                            |redundant stores/moves.
   Target Milestone|---                         |4.0.4


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27567

[Bug rtl-optimization/27567] [4.0/4.1/4.2 Regression] __builtin_memcpy generates redundant stores/moves.

[mmitchel at gcc dot gnu dot org]

-- 

mmitchel at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P2


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27567

[Bug rtl-optimization/27567] [4.0/4.1/4.2 Regression] __builtin_memcpy generates redundant stores/moves.

[pinskia at gcc dot gnu dot org]

------- Comment #2 from pinskia at gcc dot gnu dot org  2006-08-27 18:59 -------
Part of the problem here is that 4.0 removed addressof but that does not
explain why 3.4.0 could not do this optimization.
Maybe one of the problems is that we are no "lowering" memcpy at the tree level
into VCE (which should fix this one issue).


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27567

[Bug middle-end/27567] [4.0/4.1/4.2 Regression] __builtin_memcpy generates redundant stores/moves.

[pinskia at gcc dot gnu dot org]

------- Comment #3 from pinskia at gcc dot gnu dot org  2006-08-27 19:00 -------
For PPC, we get:
        stw 3,-16(1)
        stw 3,48(1)
        nop
        nop
        lfs 1,-16(1)
        blr

Notice the extra store, that does not need to be there.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|rtl-optimization            |middle-end
  GCC build triplet|x86-*-linux                 |
   GCC host triplet|x86-*-linux                 |
 GCC target triplet|x86-*-linux                 |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27567

[Bug middle-end/27567] [4.0/4.1/4.2 Regression] __builtin_memcpy generates redundant stores/moves.

[jakub at gcc dot gnu dot org]

------- Comment #4 from jakub at gcc dot gnu dot org  2006-08-29 14:35 -------
Playing with fold_builtin_mem*.


-- 

jakub at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |jakub at gcc dot gnu dot org
                   |dot org                     |
             Status|NEW                         |ASSIGNED
   Last reconfirmed|2006-05-12 08:27:14         |2006-08-29 14:35:31
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27567

[Bug middle-end/27567] [4.0/4.1/4.2 Regression] __builtin_memcpy generates redundant stores/moves.

[patchapp at dberlin dot org]

------- Comment #5 from patchapp at dberlin dot org  2006-08-30 15:50 -------
Subject: Bug number PR middle-end/27567

A patch for this bug has been added to the patch tracker.
The mailing list url for the patch is
http://gcc.gnu.org/ml/gcc-patches/2006-08/msg01134.html


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27567

[Bug middle-end/27567] [4.0/4.1/4.2 Regression] __builtin_memcpy generates redundant stores/moves.

[jakub at gcc dot gnu dot org]

------- Comment #6 from jakub at gcc dot gnu dot org  2006-09-02 06:55 -------
Subject: Bug 27567

Author: jakub
Date: Sat Sep  2 06:55:09 2006
New Revision: 116656

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=116656
Log:
        PR middle-end/27567
        * builtins.c (fold_builtin_memcpy, fold_builtin_memmove): Remove.
        (fold_builtin_mempcpy): Rename to...
        (fold_builtin_memory_op): ... this.  Optimize one element copy
        into an assignment.
        (fold_builtin_memset, fold_builtin_bzero, fold_builtin_bcopy): New
        functions.
        (expand_builtin_strcpy, expand_builtin_strncpy): Handle COMPOUND_EXPR.
        (expand_builtin_memcpy, expand_builtin_mempcpy,
        expand_builtin_memmove): Likewise.  Use fold_builtin_memory_op.
        (fold_builtin_1): Handle BUILT_IN_MEMSET, BUILT_IN_BZERO and
        BUILT_IN_BCOPY.  Use fold_builtin_memory_op for
        BUILT_IN_MEM{CPY,PCPY,MOVE}.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/builtins.c


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27567

[Bug middle-end/27567] [4.0/4.1 Regression] __builtin_memcpy generates redundant stores/moves.

[gdr at gcc dot gnu dot org]

------- Comment #7 from gdr at gcc dot gnu dot org  2007-02-03 17:01 -------
Won't fix in GCC-4.0.x.  Adjusting milestone.


-- 

gdr at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.0.4                       |4.1.3


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27567

[Bug middle-end/27567] [4.0/4.1 Regression] __builtin_memcpy generates redundant stores/moves.

[pinskia at gcc dot gnu dot org]

-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.1.3                       |4.1.2


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27567

[Bug middle-end/27567] [4.0/4.1 Regression] __builtin_memcpy generates redundant stores/moves.

[mmitchel at gcc dot gnu dot org]

-- 

mmitchel at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.1.2                       |4.1.3


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27567

[Bug middle-end/27567] [4.0/4.1 Regression] __builtin_memcpy generates redundant stores/moves.

[pluto at agmk dot net]

------- Comment #8 from pluto at agmk dot net  2007-02-14 22:24 -------
still bad :/

float convert( unsigned in )
{
        float f;
        __builtin_memcpy( &f, &in, sizeof( in ) );
        return f;
}

unsigned short load( void* p )
{
        unsigned short v;
        __builtin_memcpy( &v, p, sizeof( v ) );
        return v;
}

convert:
        movl    %edi, -4(%rsp)
        movss   -4(%rsp), %xmm1
        movaps  %xmm1, %xmm0
        ret

load:
        movzwl  (%rdi), %eax
        movw    %ax, -2(%rsp)
        movzwl  %ax, %eax
        ret

gcc-42 (20070207)


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27567

[Bug middle-end/27567] [4.0/4.1/4.2 Regression] __builtin_memcpy generates redundant stores/moves.

[pluto at agmk dot net]

------- Comment #9 from pluto at agmk dot net  2007-02-20 10:25 -------
adjust summary ( 4.2 regression status reopened ).


-- 

pluto at agmk dot net changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|[4.0/4.1 Regression]        |[4.0/4.1/4.2 Regression]
                   |__builtin_memcpy generates  |__builtin_memcpy generates
                   |redundant stores/moves.     |redundant stores/moves.


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27567

[Bug middle-end/27567] [4.0/4.1 Regression] __builtin_memcpy generates redundant stores/moves.

[pinskia at gcc dot gnu dot org]

------- Comment #10 from pinskia at gcc dot gnu dot org  2007-02-20 16:41 -------
(In reply to comment #9)
> adjust summary ( 4.2 regression status reopened ).
That is a target specific issue really and should be filed seperately.  


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to fail|3.4.6 4.0.3 4.1.2 4.2.0     |3.4.6 4.0.3 4.1.2
                   |4.3.0                       |
      Known to work|3.3.6 3.2.3 3.3.3           |3.3.6 3.2.3 3.3.3 4.2.0
                   |                            |4.3.0
            Summary|[4.0/4.1/4.2 Regression]    |[4.0/4.1 Regression]
                   |__builtin_memcpy generates  |__builtin_memcpy generates
                   |redundant stores/moves.     |redundant stores/moves.


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27567

[Bug middle-end/27567] [4.0/4.1 Regression] __builtin_memcpy generates redundant stores/moves.

[pluto at agmk dot net]

------- Comment #11 from pluto at agmk dot net  2007-02-25 18:27 -------
(In reply to comment #10)
> (In reply to comment #9)
> > adjust summary ( 4.2 regression status reopened ).
> That is a target specific issue really and should be filed seperately.  

filled as PR30961


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27567

[Bug middle-end/27567] [4.0/4.1 Regression] __builtin_memcpy generates redundant stores/moves.

[jakub at gcc dot gnu dot org]

------- Comment #12 from jakub at gcc dot gnu dot org  2007-06-26 11:45 -------
Subject: Bug 27567

Author: jakub
Date: Tue Jun 26 11:45:35 2007
New Revision: 126022

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=126022
Log:
        PR middle-end/27567
        * builtins.c (fold_builtin_memcpy, fold_builtin_memmove): Remove.
        (fold_builtin_mempcpy): Rename to...
        (fold_builtin_memory_op): ... this.  Optimize one element copy
        into an assignment.
        (fold_builtin_memset, fold_builtin_bzero, fold_builtin_bcopy): New
        functions.
        (expand_builtin_strcpy, expand_builtin_strncpy): Handle COMPOUND_EXPR.
        (expand_builtin_memcpy, expand_builtin_mempcpy,
        expand_builtin_memmove): Likewise.  Use fold_builtin_memory_op.
        (fold_builtin_1): Handle BUILT_IN_MEMSET, BUILT_IN_BZERO and
        BUILT_IN_BCOPY.  Use fold_builtin_memory_op for
        BUILT_IN_MEM{CPY,PCPY,MOVE}.

Modified:
    branches/redhat/gcc-4_1-branch/gcc/ChangeLog
    branches/redhat/gcc-4_1-branch/gcc/builtins.c


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27567

[Bug middle-end/27567] [4.0/4.1 Regression] __builtin_memcpy generates redundant stores/moves.

[rguenth at gcc dot gnu dot org]

------- Comment #13 from rguenth at gcc dot gnu dot org  2008-02-20 22:37 -------
Fixed since 4.2.0, wontfix on earlier branches.


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
      Known to fail|3.4.6 4.0.3 4.1.2           |3.4.6 4.0.3 4.1.2 4.1.3
         Resolution|                            |FIXED
   Target Milestone|4.1.3                       |4.2.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27567