[Bug target/106601] New: __builtin_bswap16 code gen could be improved with ZBB enabled

[Bug target/106601] New: __builtin_bswap16 code gen could be improved with ZBB enabled

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106601

            Bug ID: 106601
           Summary: __builtin_bswap16 code gen could be improved with ZBB
                    enabled
           Product: gcc
           Version: 12.0
            Status: UNCONFIRMED
          Keywords: missed-optimization
          Severity: normal
          Priority: P3
         Component: target
          Assignee: pinskia at gcc dot gnu.org
          Reporter: pinskia at gcc dot gnu.org
  Target Milestone: ---
            Target: riscv

unsigned short t(unsigned short a)
{
  return __builtin_bswap16(a);
}

This could just be rev8 followed by a srli (I think).

[Bug target/106601] __builtin_bswap16 code gen could be improved with ZBB enabled

[andrew at sifive dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106601

Andrew Waterman <andrew at sifive dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |andrew at sifive dot com

--- Comment #1 from Andrew Waterman <andrew at sifive dot com> ---
Yep.

(I don't know whether bswap32 has already been similarly optimized on RV64, but
it needs to use srai so that the 32-bit result is canonicalized properly.)

[Bug target/106601] __builtin_bswap16 code gen could be improved with ZBB enabled

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106601

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |ASSIGNED
     Ever confirmed|0                           |1
   Last reconfirmed|                            |2022-08-12

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to Andrew Waterman from comment #1)

> (I don't know whether bswap32 has already been similarly optimized on RV64,
> but it needs to use srai so that the 32-bit result is canonicalized
> properly.)

It is:
t:
        rev8    a0,a0
        srai    a0,a0,32
        ret
For:
unsigned int t(unsigned int a)
{
  return __builtin_bswap32(a);
}

[Bug target/106601] __builtin_bswap16 code gen could be improved with ZBB enabled

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106601

--- Comment #4 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The trunk branch has been updated by Andrew Pinski <pinskia@gcc.gnu.org>:

https://gcc.gnu.org/g:e5e6983c3da53729e58a32af1d531ea74b3dbf5d

commit r13-2179-ge5e6983c3da53729e58a32af1d531ea74b3dbf5d
Author: Andrew Pinski <apinski@marvell.com>
Date:   Fri Aug 19 17:46:40 2022 +0000

    Fix PR 106601: __builtin_bswap16 code gen could be improved with ZBB
enabled

    The default expansion for bswap16 is two extractions (shift/and)
    followed by an insertation (ior) and then a zero extend. This can be
improved
    with ZBB enabled to just full byteswap followed by a (logical) shift right.
    This patch adds a new pattern for this which does that.

    OK? Built and tested on riscv32-linux-gnu and riscv64-linux-gnu.

    gcc/ChangeLog:

            PR target/106601
            * config/riscv/bitmanip.md (bswaphi2): New pattern.

    gcc/testsuite/ChangeLog:

            PR target/106601
            * gcc.target/riscv/zbb_32_bswap-2.c: New test.
            * gcc.target/riscv/zbb_bswap-2.c: New test.

[Bug target/106601] __builtin_bswap16 code gen could be improved with ZBB enabled

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106601

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |FIXED
             Status|ASSIGNED                    |RESOLVED
   Target Milestone|---                         |13.0

--- Comment #5 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Fixed.