On Tue, Oct 27, 2020 at 11:49 PM Jan Hubicka <hubi...@ucw.cz> wrote:
>
> Hi,
> this patch avoid typeless storage in trailing_wide_ints.  This improves
> TBAA disambiguation rate on cc1plus by 7%.
>
> Bootstrapped/regtested x86_64-linux, OK?

OK.

I guess if uint8_t would be a separate type from unsigned char that
would work as well but it is not.

Thanks,
Richard.

>         * wide-int.h (trailing_wide_ints <N>): Turn len to array of structures
>         so it does not imply typeless storage.
>         (trailing_wide_ints <N>::operator): update
>         (trailing_wide_ints <N>::operator []): Update.
> diff --git a/gcc/wide-int.h b/gcc/wide-int.h
> index 39cd5b9bd17..6eae46da12e 100644
> --- a/gcc/wide-int.h
> +++ b/gcc/wide-int.h
> @@ -1387,8 +1387,10 @@ private:
>    /* The shared maximum length of each number.  */
>    unsigned char m_max_len;
>
> -  /* The current length of each number.  */
> -  unsigned char m_len[N];
> +  /* The current length of each number.
> +     Avoid char array so the whole structure is not a typeless storage
> +     that will, in turn, turn off TBAA on gimple, trees and RTL.  */
> +  struct {unsigned char len;} m_len[N];
>
>    /* The variable-length part of the structure, which always contains
>       at least one HWI.  Element I starts at index I * M_MAX_LEN.  */
> @@ -1470,7 +1472,7 @@ template <int N>
>  inline trailing_wide_int
>  trailing_wide_ints <N>::operator [] (unsigned int index)
>  {
> -  return trailing_wide_int_storage (m_precision, &m_len[index],
> +  return trailing_wide_int_storage (m_precision, &m_len[index].len,
>                                     &m_val[index * m_max_len]);
>  }
>
> @@ -1479,7 +1481,7 @@ inline typename trailing_wide_ints <N>::const_reference
>  trailing_wide_ints <N>::operator [] (unsigned int index) const
>  {
>    return wi::storage_ref (&m_val[index * m_max_len],
> -                         m_len[index], m_precision);
> +                         m_len[index].len, m_precision);
>  }
>
>  /* Return how many extra bytes need to be added to the end of the structure

Reply via email to