[Bug rtl-optimization/64960] New: Inefficient address pre-computation in PIC mode

[Bug rtl-optimization/64960] New: Inefficient address pre-computation in PIC mode

[enkovich.gnu at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64960

            Bug ID: 64960
           Summary: Inefficient address pre-computation in PIC mode
           Product: gcc
           Version: 5.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: rtl-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: enkovich.gnu at gmail dot com

After EBX was unfixed in i386 PIC target, we may see addresses of static
objects are loaded from GOT and placed to the stack for later usage.  It allows
to reuse PIC register for other purposes.  But in cases when PIC register is
still used (e.g. for calls) it may cause inefficiency in produced code. Here is
an example:

>cat test.c
void f (int);
int val1, *val2, val3;
int test (int max)
{
  int i;
  for (i = 0; i < max; i++)
    {
      val1 += val2[i];
      f (val3);
    }
}
>gcc test.c -O2 -fPIE -S -m32 -ffixed-esi -ffixed-edi -ffixed-edx
>cat test.s
...
        movl    val1@GOT(%ebx), %eax  <-- may be removed
        xorl    %ebp, %ebp
        movl    %eax, 4(%esp)         <-- may be removed
        movl    val2@GOT(%ebx), %eax  <-- may be removed
        movl    %eax, 8(%esp)         <-- may be removed
        movl    val3@GOT(%ebx), %eax  <-- may be removed
        movl    %eax, 12(%esp)        <-- may be removed
.L3:
        movl    8(%esp), %eax         <-- equal to    movl  val2@GOT(%ebx),
%eax
        subl    $12, %esp
        movl    (%eax), %ecx
        movl    16(%esp), %eax        <-- equal to    movl  val1@GOT(%ebx),
%eax
        movl    (%ecx,%ebp,4), %ecx
        addl    %ecx, (%eax)
        addl    $1, %ebp
        movl    24(%esp), %eax        <-- equal to    movl  val3@GOT(%ebx),
%eax
        pushl   (%eax)
        call    f@PLT
        addl    $16, %esp
        cmpl    %ebp, 32(%esp)
        jne     .L3
...

Also storing value on the stack doesn't benefit on static objects optimization
performed by linker which transforms movl <symbol>@GOT<PIC> into lea
instruction.  It would be useful to avoid early address computation in case PIC
register is available at address usage.

Here is a code generated by GCC 4.9:

        xorl    %ebp, %ebp
.L2:
        movl    val2@GOT(%ebx), %eax
        subl    $12, %esp
        movl    (%eax), %ecx
        movl    val1@GOT(%ebx), %eax
        movl    (%ecx,%ebp,4), %ecx
        addl    %ecx, (%eax)
        addl    $1, %ebp
        movl    val3@GOT(%ebx), %eax
        pushl   (%eax)
        call    f@PLT
        addl    $16, %esp
        cmpl    16(%esp), %ebp
        jne     .L2

Used gcc (GCC) 5.0.0 20150205 (experimental).

[Bug target/64960] [5 Regression] Inefficient address pre-computation in PIC mode

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64960

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
            Summary|Inefficient address         |[5 Regression] Inefficient
                   |pre-computation in PIC mode |address pre-computation in
                   |                            |PIC mode
         Resolution|---                         |FIXED
   Target Milestone|---                         |5.0

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Was fixed during the developement of GCC 5.  That is GCC 5.1.0 release did not
have this issue.