Re: [PATCH v11 09/29] string: Improve generic strncmp

[Noah Goldstein <goldstein.w.n@gmail.com>]

On Wed, Feb 1, 2023 at 11:04 AM Adhemerval Zanella
<adhemerval.zanella@linaro.org> wrote:
>
> It follows the strategy:
>
>    - Align the first input to word boundary using byte operations.
>
>    - If second input is also word aligned, read a word per time, check
>      for  null (using has_zero), and check final words using byte
>      operation.
>
>    - If second input is not word aligned, loop by aligning the source,
>      and merge the result of two reads.  Similar to aligned case, check
>      for null with has_zero, and check final words using byte operation.
>
> Checked on x86_64-linux-gnu, i686-linux-gnu, powerpc64-linux-gnu,
> and powerpc-linux-gnu by removing the arch-specific assembly
> implementation and disabling multi-arch (it covers both LE and BE
> for 64 and 32 bits).
> ---
>  string/strcmp-impl.h |  41 ++++++++++++++
>  string/strcmp.c      |  23 ++------
>  string/strncmp.c     | 130 +++++++++++++++++++++++++++++++------------
>  3 files changed, 138 insertions(+), 56 deletions(-)
>  create mode 100644 string/strcmp-impl.h
>
> diff --git a/string/strcmp-impl.h b/string/strcmp-impl.h
> new file mode 100644
> index 0000000000..618240368a
> --- /dev/null
> +++ b/string/strcmp-impl.h
> @@ -0,0 +1,41 @@
> +/* Common definition for string compare implementations.
> +   Copyright (C) 2023 Free Software Foundation, Inc.
> +   This file is part of the GNU C Library.
> +
> +   The GNU C Library is free software; you can redistribute it and/or
> +   modify it under the terms of the GNU Lesser General Public
> +   License as published by the Free Software Foundation; either
> +   version 2.1 of the License, or (at your option) any later version.
> +
> +   The GNU C Library is distributed in the hope that it will be useful,
> +   but WITHOUT ANY WARRANTY; without even the implied warranty of
> +   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> +   Lesser General Public License for more details.
> +
> +   You should have received a copy of the GNU Lesser General Public
> +   License along with the GNU C Library; if not, see
> +   <https://www.gnu.org/licenses/>.  */
> +
> +#ifndef _STRCMP_IMPL_H
> +#define _STRCMP_IMPL_H
> +
> +#include <stddef.h>
> +#include <string-misc.h>
> +
> +static inline int
> +final_cmp (const op_t w1, const op_t w2, size_t n)
> +{
> +  /* It can not use index_first_zero_ne because it must not compare past the
> +     final '\0' is present (and final_cmp is called before has_zero check).
> +   */
> +  for (size_t i = 0; i < n; i++)
> +    {
> +      unsigned char c1 = extractbyte (w1, i);
> +      unsigned char c2 = extractbyte (w2, i);
> +      if (c1 == '\0' || c1 != c2)
> +        return c1 - c2;
> +    }
> +  return 0;
> +}
> +
> +#endif
> diff --git a/string/strcmp.c b/string/strcmp.c
> index 42e24242b6..8e7b3310db 100644
> --- a/string/strcmp.c
> +++ b/string/strcmp.c
> @@ -16,6 +16,7 @@
>     <https://www.gnu.org/licenses/>.  */
>
>  #include <stdint.h>
> +#include <strcmp-impl.h>
>  #include <string-fzb.h>
>  #include <string-fzi.h>
>  #include <string.h>
> @@ -25,22 +26,6 @@
>  # define strcmp STRCMP
>  #endif
>
> -static inline int
> -final_cmp (const op_t w1, const op_t w2)
> -{
> -  /* It can not use index_first_zero_ne because it must not compare past the
> -     final '\0' is present (and final_cmp is called before has_zero check).
> -   */
> -  for (size_t i = 0; i < sizeof (op_t); i++)
> -    {
> -      unsigned char c1 = extractbyte (w1, i);
> -      unsigned char c2 = extractbyte (w2, i);
> -      if (c1 == '\0' || c1 != c2)
> -        return c1 - c2;
> -    }
> -  return 0;
> -}
> -
>  /* Aligned loop: if a difference is found, exit to compare the bytes.  Else
>     if a zero is found we have equal strings.  */
>  static inline int
> @@ -56,7 +41,7 @@ strcmp_aligned_loop (const op_t *x1, const op_t *x2, op_t w1)
>        w2 = *x2++;
>      }
>
> -  return final_cmp (w1, w2);
> +  return final_cmp (w1, w2, sizeof (op_t));
>  }
>
>  /* Unaligned loop: align the first partial of P2, with 0xff for the rest of
> @@ -83,7 +68,7 @@ strcmp_unaligned_loop (const op_t *x1, const op_t *x2, op_t w1, uintptr_t ofs)
>           w2b = *x2++;
>           w2 = MERGE (w2a, sh_1, w2b, sh_2);
>           if (w1 != w2)
> -           return final_cmp (w1, w2);
> +           return final_cmp (w1, w2, sizeof (op_t));
>           if (has_zero (w2b))
>             break;
>           w1 = *x1++;
> @@ -100,7 +85,7 @@ strcmp_unaligned_loop (const op_t *x1, const op_t *x2, op_t w1, uintptr_t ofs)
>        w2 = MERGE (w2b, sh_1, 0, sh_2);
>      }
>
> -  return final_cmp (w1, w2);
> +  return final_cmp (w1, w2, sizeof (op_t));
>  }
>
>  /* Compare S1 and S2, returning less than, equal to or
> diff --git a/string/strncmp.c b/string/strncmp.c
> index fd7cee09b6..3e6040df09 100644
> --- a/string/strncmp.c
> +++ b/string/strncmp.c
> @@ -15,7 +15,13 @@
>     License along with the GNU C Library; if not, see
>     <https://www.gnu.org/licenses/>.  */
>
> +#include <stdint.h>
> +#include <strcmp-impl.h>
> +#include <string-fza.h>
> +#include <string-fzb.h>
> +#include <string-fzi.h>
>  #include <string.h>
> +#include <sys/param.h>
>  #include <memcopy.h>
>
>  #undef strncmp
> @@ -24,51 +30,101 @@
>  #define STRNCMP strncmp
>  #endif
>
> -/* Compare no more than N characters of S1 and S2,
> -   returning less than, equal to or greater than zero
> -   if S1 is lexicographically less than, equal to or
> -   greater than S2.  */
> -int
> -STRNCMP (const char *s1, const char *s2, size_t n)
> +/* Aligned loop: if a difference is found, exit to compare the bytes.  Else
> +   if a zero is found we have equal strings.  */
> +static inline int
> +strncmp_aligned_loop (const op_t *x1, const op_t *x2, op_t w1, size_t n)
>  {
> -  unsigned char c1 = '\0';
> -  unsigned char c2 = '\0';
> +  op_t w2 = *x2++;
>
> -  if (n >= 4)
> +  while (w1 == w2)
>      {
> -      size_t n4 = n >> 2;
> -      do
> -       {
> -         c1 = (unsigned char) *s1++;
> -         c2 = (unsigned char) *s2++;
> -         if (c1 == '\0' || c1 != c2)
> -           return c1 - c2;
> -         c1 = (unsigned char) *s1++;
> -         c2 = (unsigned char) *s2++;
> -         if (c1 == '\0' || c1 != c2)
> -           return c1 - c2;
> -         c1 = (unsigned char) *s1++;
> -         c2 = (unsigned char) *s2++;
> -         if (c1 == '\0' || c1 != c2)
> -           return c1 - c2;
> -         c1 = (unsigned char) *s1++;
> -         c2 = (unsigned char) *s2++;
> -         if (c1 == '\0' || c1 != c2)
> -           return c1 - c2;
> -       } while (--n4 > 0);
> -      n &= 3;
> +      if (n <= sizeof (op_t))
> +       break;
> +      n -= sizeof (op_t);
> +
> +      if (has_zero (w1))
> +       return 0;
> +      w1 = *x1++;
> +      w2 = *x2++;
>      }
>
> -  while (n > 0)
> +  return final_cmp (w1, w2, n);
> +}
> +
> +/* Unaligned loop: align the first partial of P2, with 0xff for the rest of
> +   the bytes so that we can also apply the has_zero test to see if we have
> +   already reached EOS.  If we have, then we can simply fall through to the
> +   final comparison.  */
> +static inline int
> +strncmp_unaligned_loop (const op_t *x1, const op_t *x2, op_t w1, uintptr_t ofs,
> +                       size_t n)
> +{
> +  op_t w2a = *x2++;
> +  uintptr_t sh_1 = ofs * CHAR_BIT;
> +  uintptr_t sh_2 = sizeof(op_t) * CHAR_BIT - sh_1;
> +
> +  op_t w2 = MERGE (w2a, sh_1, (op_t)-1, sh_2);
> +  if (!has_zero (w2) && n > (sizeof (op_t) - ofs))
>      {
> -      c1 = (unsigned char) *s1++;
> -      c2 = (unsigned char) *s2++;
> -      if (c1 == '\0' || c1 != c2)
> -       return c1 - c2;
> -      n--;
> +      op_t w2b;
> +
> +      /* Unaligned loop.  The invariant is that W2B, which is "ahead" of W1,
> +        does not contain end-of-string.  Therefore it is safe (and necessary)
> +        to read another word from each while we do not have a difference.  */
> +      while (1)
> +       {
> +         w2b = *x2++;
> +         w2 = MERGE (w2a, sh_1, w2b, sh_2);
> +         if (n <= sizeof (op_t) || w1 != w2)
> +           return final_cmp (w1, w2, n);
> +         n -= sizeof(op_t);
> +         if (has_zero (w2b) || n <= (sizeof (op_t) - ofs))
> +           break;
> +         w1 = *x1++;
> +         w2a = w2b;
> +       }
> +
> +      /* Zero found in the second partial of P2.  If we had EOS in the aligned
> +        word, we have equality.  */
> +      if (has_zero (w1))
> +       return 0;
> +
> +      /* Load the final word of P1 and align the final partial of P2.  */
> +      w1 = *x1++;
> +      w2 = MERGE (w2b, sh_1, 0, sh_2);
>      }
>
> -  return c1 - c2;
> +  return final_cmp (w1, w2, n);
>  }
>
> +/* Compare no more than N characters of S1 and S2,
> +   returning less than, equal to or greater than zero
> +   if S1 is lexicographically less than, equal to or
> +   greater than S2.  */
> +int
> +STRNCMP (const char *p1, const char *p2, size_t n)
> +{
> +  /* Handle the unaligned bytes of p1 first.  */
> +  uintptr_t a = MIN (-(uintptr_t)p1 % sizeof(op_t), n);
> +  int diff = 0;
> +  for (int i = 0; i < a; ++i)
> +    {
> +      unsigned char c1 = *p1++;
> +      unsigned char c2 = *p2++;
> +      diff = c1 - c2;
> +      if (c1 == '\0' || diff != 0)
> +       return diff;
> +    }
> +  if (a == n)
> +    return diff;

NIT: return 0 IMO
> +
> +  /* P1 is now aligned to op_t.  P2 may or may not be.  */
> +  const op_t *x1 = (const op_t *) p1;
> +  op_t w1 = *x1++;
> +  uintptr_t ofs = (uintptr_t) p2 % sizeof(op_t);
> +  return ofs == 0
> +    ? strncmp_aligned_loop (x1, (const op_t *) p2, w1, n - a)
> +    : strncmp_unaligned_loop (x1, (const op_t *) (p2 - ofs), w1, ofs, n - a);
> +}
>  libc_hidden_builtin_def (STRNCMP)
> --
> 2.34.1
>