Simplify and speedup strstr/strcasestr first match

Looking at the benchtests, both strstr and strcasestr spend a lot of time
in a slow initialization loop handling one character per iteration.
This can be simplified and use the much faster strlen/strnlen/strchr/memcmp.
Read ahead a few cachelines to reduce the number of strnlen calls, which
improves performance by ~3-4%.  This patch improves the time taken for the
full strstr benchtest by >40%.

	* string/strcasestr.c (STRCASESTR): Simplify and speedup first match.
	* string/strstr.c (AVAILABLE): Likewise.

(cherry picked from commit 284f42bc778e487dfd5dff5c01959f93b9e0c4f5)

3 files changed, 40 insertions, 45 deletions

diff --git a/ChangeLog b/ChangeLog
index d21bdbcce8..22012fb458 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,5 +1,10 @@
 2019-09-13  Wilco Dijkstra  <wdijkstr@arm.com>
 
+	* string/strcasestr.c (STRCASESTR): Simplify and speedup first match.
+	* string/strstr.c (AVAILABLE): Likewise.
+
+2019-09-13  Wilco Dijkstra  <wdijkstr@arm.com>
+
 	* benchtests/bench-strcasestr.c: Rename __strnlen to strnlen.
 	* benchtests/bench-strstr.c: Likewise.
 	* string/memmem.c (FASTSEARCH): Define.
diff --git a/string/strcasestr.c b/string/strcasestr.c
index 7caaade5d2..772e513626 100644
--- a/string/strcasestr.c
+++ b/string/strcasestr.c
@@ -58,31 +58,22 @@
    case-insensitive comparison.  This function gives unspecified
    results in multibyte locales.  */
 char *
-STRCASESTR (const char *haystack_start, const char *needle_start)
+STRCASESTR (const char *haystack, const char *needle)
 {
-  const char *haystack = haystack_start;
-  const char *needle = needle_start;
   size_t needle_len; /* Length of NEEDLE.  */
   size_t haystack_len; /* Known minimum length of HAYSTACK.  */
-  bool ok = true; /* True if NEEDLE is prefix of HAYSTACK.  */
-
-  /* Determine length of NEEDLE, and in the process, make sure
-     HAYSTACK is at least as long (no point processing all of a long
-     NEEDLE if HAYSTACK is too short).  */
-  while (*haystack && *needle)
-    {
-      ok &= (TOLOWER ((unsigned char) *haystack)
-	     == TOLOWER ((unsigned char) *needle));
-      haystack++;
-      needle++;
-    }
-  if (*needle)
+
+  /* Handle empty NEEDLE special case.  */
+  if (needle[0] == '\0')
+    return (char *) haystack;
+
+  /* Ensure HAYSTACK length is at least as long as NEEDLE length.
+     Since a match may occur early on in a huge HAYSTACK, use strnlen
+     and read ahead a few cachelines for improved performance.  */
+  needle_len = strlen (needle);
+  haystack_len = __strnlen (haystack, needle_len + 256);
+  if (haystack_len < needle_len)
     return NULL;
-  if (ok)
-    return (char *) haystack_start;
-  needle_len = needle - needle_start;
-  haystack = haystack_start + 1;
-  haystack_len = needle_len - 1;
 
   /* Perform the search.  Abstract memory is considered to be an array
      of 'unsigned char' values, not an array of 'char' values.  See
@@ -90,10 +81,10 @@ STRCASESTR (const char *haystack_start, const char *needle_start)
   if (needle_len < LONG_NEEDLE_THRESHOLD)
     return two_way_short_needle ((const unsigned char *) haystack,
 				 haystack_len,
-				 (const unsigned char *) needle_start,
+				 (const unsigned char *) needle,
 				 needle_len);
   return two_way_long_needle ((const unsigned char *) haystack, haystack_len,
-			      (const unsigned char *) needle_start,
+			      (const unsigned char *) needle,
 			      needle_len);
 }
 
diff --git a/string/strstr.c b/string/strstr.c
index 63facae4a5..1eb4dc11db 100644
--- a/string/strstr.c
+++ b/string/strstr.c
@@ -50,33 +50,32 @@
    if NEEDLE is empty, otherwise NULL if NEEDLE is not found in
    HAYSTACK.  */
 char *
-STRSTR (const char *haystack_start, const char *needle_start)
+STRSTR (const char *haystack, const char *needle)
 {
-  const char *haystack = haystack_start;
-  const char *needle = needle_start;
   size_t needle_len; /* Length of NEEDLE.  */
   size_t haystack_len; /* Known minimum length of HAYSTACK.  */
-  bool ok = true; /* True if NEEDLE is prefix of HAYSTACK.  */
-
-  /* Determine length of NEEDLE, and in the process, make sure
-     HAYSTACK is at least as long (no point processing all of a long
-     NEEDLE if HAYSTACK is too short).  */
-  while (*haystack && *needle)
-    ok &= *haystack++ == *needle++;
-  if (*needle)
+
+  /* Handle empty NEEDLE special case.  */
+  if (needle[0] == '\0')
+    return (char *) haystack;
+
+  /* Skip until we find the first matching char from NEEDLE.  */
+  haystack = strchr (haystack, needle[0]);
+  if (haystack == NULL || needle[1] == '\0')
+    return (char *) haystack;
+
+  /* Ensure HAYSTACK length is at least as long as NEEDLE length.
+     Since a match may occur early on in a huge HAYSTACK, use strnlen
+     and read ahead a few cachelines for improved performance.  */
+  needle_len = strlen (needle);
+  haystack_len = __strnlen (haystack, needle_len + 256);
+  if (haystack_len < needle_len)
     return NULL;
-  if (ok)
-    return (char *) haystack_start;
-
-  /* Reduce the size of haystack using strchr, since it has a smaller
-     linear coefficient than the Two-Way algorithm.  */
-  needle_len = needle - needle_start;
-  haystack = strchr (haystack_start + 1, *needle_start);
-  if (!haystack || __builtin_expect (needle_len == 1, 0))
+
+  /* Check whether we have a match.  This improves performance since we avoid
+     the initialization overhead of the two-way algorithm.  */
+  if (memcmp (haystack, needle, needle_len) == 0)
     return (char *) haystack;
-  needle -= needle_len;
-  haystack_len = (haystack > haystack_start + needle_len ? 1
-		  : needle_len + haystack_start - haystack);
 
   /* Perform the search.  Abstract memory is considered to be an array
      of 'unsigned char' values, not an array of 'char' values.  See