public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
From: Hongtao Liu <crazylht@gmail.com>
To: liuhongt <hongtao.liu@intel.com>
Cc: gcc-patches@gcc.gnu.org, hjl.tools@gmail.com
Subject: Re: [PATCH] Issue a warning for conversion between short and __bf16 under TARGET_AVX512BF16.
Date: Tue, 27 Jun 2023 13:13:31 +0800	[thread overview]
Message-ID: <CAMZc-bykONXMk=9bGe73sFFHC5LsMOEV0_GkH=aT9Z60tA3OmQ@mail.gmail.com> (raw)
In-Reply-To: <20230626085417.1357574-1-hongtao.liu@intel.com>

On Mon, Jun 26, 2023 at 4:54 PM liuhongt <hongtao.liu@intel.com> wrote:
>
> __bfloat16 is redefined from typedef short to real __bf16 since GCC
> V13. The patch issues an warning for potential silent implicit
> conversion between __bf16 and short where users may only expect a
> data movement.
>
> To avoid too many false positive, warning is only under
> TARGET_AVX512BF16.
>
> Bootstrapped and regtested on x86_64-pc-linux-gnu{-m32,}
> Ok for trunk?
Ready to install.
>
> gcc/ChangeLog:
>
>         * config/i386/i386.cc (ix86_invalid_conversion): New function.
>         (TARGET_INVALID_CONVERSION): Define as
>         ix86_invalid_conversion.
>
> gcc/testsuite/ChangeLog:
>
>         * gcc.target/i386/bf16_short_warn.c: New test.
> ---
>  gcc/config/i386/i386.cc                       | 32 +++++++++++++++++++
>  .../gcc.target/i386/bf16_short_warn.c         | 17 ++++++++++
>  2 files changed, 49 insertions(+)
>  create mode 100644 gcc/testsuite/gcc.target/i386/bf16_short_warn.c
>
> diff --git a/gcc/config/i386/i386.cc b/gcc/config/i386/i386.cc
> index 0761965344b..dc02eac6203 100644
> --- a/gcc/config/i386/i386.cc
> +++ b/gcc/config/i386/i386.cc
> @@ -22718,6 +22718,35 @@ x86_emit_floatuns (rtx operands[2])
>
>    emit_label (donelab);
>  }
> +
> +/* Return the diagnostic message string if conversion from FROMTYPE to
> +   TOTYPE is not allowed, NULL otherwise.
> +   Currently it's used to warn for silent implicit conversion between __bf16
> +   and short, since __bfloat16 is refined as real __bf16 instead of short
> +   since GCC13.  */
> +
> +static const char *
> +ix86_invalid_conversion (const_tree fromtype, const_tree totype)
> +{
> +  if (element_mode (fromtype) != element_mode (totype)
> +      && (TARGET_AVX512BF16 || TARGET_AVXNECONVERT))
> +    {
> +      /* Warn for silent implicit conversion where user may expect
> +        a bitcast.  */
> +      if ((TYPE_MODE (fromtype) == BFmode
> +          && TYPE_MODE (totype) == HImode)
> +         || (TYPE_MODE (totype) == BFmode
> +             && TYPE_MODE (fromtype) == HImode))
> +       warning (0, "%<__bfloat16%> is redefined from typedef %<short%> "
> +               "to real %<__bf16%> since GCC V13, be careful of "
> +                "implicit conversion between %<__bf16%> and %<short%>; "
> +                "a explicit bitcast may be needed here");
> +    }
> +
> +  /* Conversion allowed.  */
> +  return NULL;
> +}
> +
>
>  /* Target hook for scalar_mode_supported_p.  */
>  static bool
> @@ -25009,6 +25038,9 @@ ix86_run_selftests (void)
>  #  define TARGET_MERGE_DECL_ATTRIBUTES merge_dllimport_decl_attributes
>  #endif
>
> +#undef TARGET_INVALID_CONVERSION
> +#define TARGET_INVALID_CONVERSION ix86_invalid_conversion
> +
>  #undef TARGET_COMP_TYPE_ATTRIBUTES
>  #define TARGET_COMP_TYPE_ATTRIBUTES ix86_comp_type_attributes
>
> diff --git a/gcc/testsuite/gcc.target/i386/bf16_short_warn.c b/gcc/testsuite/gcc.target/i386/bf16_short_warn.c
> new file mode 100644
> index 00000000000..3e47a815200
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/i386/bf16_short_warn.c
> @@ -0,0 +1,17 @@
> +/* { dg-do compile } */
> +/* { dg-options "-O2" } */
> +
> +#include<immintrin.h>
> +typedef struct {
> +short payload;
> +} BFloat16;
> +
> +__attribute__((target("avx512vl,avx512bf16")))
> +BFloat16 tobf16_avx512(float f)
> +{
> +    BFloat16 r;
> +    __m128bh m = _mm_cvtneps_pbh(_mm_set_ss(f));
> +    r.payload = m[0]; /* { dg-warning " be careful of implicit conversion between '__bf16' and 'short'" } */
> +    return r;
> +}
> +
> --
> 2.39.1.388.g2fc9e9ca3c
>


-- 
BR,
Hongtao

      reply	other threads:[~2023-06-27  5:13 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-06-26  8:54 liuhongt
2023-06-27  5:13 ` Hongtao Liu [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to='CAMZc-bykONXMk=9bGe73sFFHC5LsMOEV0_GkH=aT9Z60tA3OmQ@mail.gmail.com' \
    --to=crazylht@gmail.com \
    --cc=gcc-patches@gcc.gnu.org \
    --cc=hjl.tools@gmail.com \
    --cc=hongtao.liu@intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).