[Bug c/66663] New: gcc misses optimization emits useless test of (a & 31) with 32

[Bug c/66663] New: gcc misses optimization emits useless test of (a & 31) with 32

[fuz at fuz dot su]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66663

            Bug ID: 66663
           Summary: gcc misses optimization emits useless test of (a & 31)
                    with 32
           Product: gcc
           Version: 5.0
            Status: UNCONFIRMED
          Severity: minor
          Priority: P3
         Component: c
          Assignee: unassigned at gcc dot gnu.org
          Reporter: fuz at fuz dot su
  Target Milestone: ---

The following code:

    unsigned long long foo(unsigned long long x, int a) {
        return x << (a & 31);
    }

compiled with a recent build of gcc for i386 Linux (with -O3) yields the
following assembly:

    foo:
        movl    12(%esp), %ecx
        movl    4(%esp), %eax
        movl    8(%esp), %edx
        andl    $31, %ecx
        shldl   %eax, %edx
        sall    %cl, %eax
        testb   $32, %cl
        je      .L2
        movl    %eax, %edx
        xorl    %eax, %eax
.L2:
        rep ret

        .ident  "GCC: (GNU) 5.0.0 20150314 (experimental)"

The testb instruction seems redundant as %cl has been masked with $31 a few
instructions before: The jump will never be taken.

[Bug c/66663] gcc misses optimization emits useless test of (a & 31) with 32

[fuz at fuz dot su]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66663

--- Comment #1 from Robert Clausecker <fuz at fuz dot su> ---
Sorry. I meant to say: The branch will always be taken, not never.

[Bug target/66663] gcc misses optimization emits useless test of (a & 31) with 32

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66663

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
   Target Milestone|---                         |9.0
         Resolution|---                         |FIXED

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
GCC 9+ is able to produce:
foo(unsigned long long, int):
.LFB0:
        .cfi_startproc
        movl    4(%esp), %eax
        movl    12(%esp), %ecx
        movl    8(%esp), %edx
        shldl   %eax, %edx
        sall    %cl, %eax
        ret

[Bug target/66663] gcc misses optimization emits useless test of (a & 31) with 32

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66663

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|FIXED                       |DUPLICATE

--- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
It was fixed by r9-62 which means this is a dup of bug 84431.

*** This bug has been marked as a duplicate of bug 84431 ***