[Bug target/95535] New: Failure to optimize out cdqe after __bultin_ctz

[Bug target/95535] New: Failure to optimize out cdqe after __bultin_ctz

[gabravier at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95535

            Bug ID: 95535
           Summary: Failure to optimize out cdqe after __bultin_ctz
           Product: gcc
           Version: 11.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: gabravier at gmail dot com
  Target Milestone: ---

long long f(int n)
{
    return __builtin_ctz(n);
}

With -O3, GCC outputs this :

f(int):
  xor eax, eax
  tzcnt eax, edi
  cdqe
  ret

LLVM outputs this :

f(int): # @f(int)
  tzcnt eax, edi
  ret

Looks like the cdqe can be omitted.

[Bug target/95535] Failure to optimize out cdqe after __bultin_ctz

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95535

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2020-06-04
           Assignee|unassigned at gcc dot gnu.org      |jakub at gcc dot gnu.org
             Status|UNCONFIRMED                 |ASSIGNED
                 CC|                            |jakub at gcc dot gnu.org
     Ever confirmed|0                           |1

--- Comment #1 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
I've done this for popcount in January, so need to do the same for tzcntl and
lzcntl.

[Bug target/95535] Failure to optimize out cdqe after __bultin_ctz

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95535

--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Created attachment 48676
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=48676&action=edit
gcc11-pr95535.patch

Untested version of the PR91824 popcnt fix.

[Bug target/95535] Failure to optimize out cdqe after __bultin_ctz

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95535

--- Comment #3 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Jakub Jelinek <jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:892b51cb73f24157391cd18c2215e8d703af97e7

commit r11-965-g892b51cb73f24157391cd18c2215e8d703af97e7
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Fri Jun 5 10:44:23 2020 +0200

    ix86: Improve __builtin_c[lt]z followed by extension [PR95535]

    In January I've added patterns to optimize SImode -> DImode sign or zero
    extension of __builtin_popcount, this patch does the same for
    __builtin_c[lt]z.  Like most other instructions, the [tl]zcntl instructions
    clear the upper 32 bits of the destination register and as the instructions
    only result in values 0 to 32 inclusive, both sign and zero extensions
    behave the same.

    2020-06-05  Jakub Jelinek  <jakub@redhat.com>

            PR target/95535
            * config/i386/i386.md (*ctzsi2_zext, *clzsi2_lzcnt_zext): New
            define_insn_and_split patterns.
            (*ctzsi2_zext_falsedep, *clzsi2_lzcnt_zext_falsedep): New
            define_insn patterns.

            * gcc.target/i386/pr95535-1.c: New test.
            * gcc.target/i386/pr95535-2.c: New test.

[Bug target/95535] Failure to optimize out cdqe after __bultin_ctz

[gabravier at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95535

Gabriel Ravier <gabravier at gmail dot com> changed:

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

--- Comment #4 from Gabriel Ravier <gabravier at gmail dot com> ---
Looks like this is fixed but not marked as such, so I'll make it so.