Speed up software CRC-32 computation by a factor of 1.5 to 3.

Use the interleaved method of Kadatch and Jenkins in order to make
use of pipelined instructions through multiple ALUs in a single
core. This also speeds up and simplifies the combination of CRCs,
and updates the functions to pre-calculate and use an operator for
CRC combination.

configure

@@ -367,8 +367,11 @@ else
   try()
   {
     show $*
-    ( $* ) >> configure.log 2>&1
+    got=`( $* ) 2>&1`
     ret=$?
+    if test "$got" != ""; then
+      printf "%s\n" "$got" >> configure.log
+    fi
     if test $ret -ne 0; then
       echo "(exit code "$ret")" >> configure.log
     fi
@@ -381,8 +384,11 @@ tryboth()
   show $*
   got=`( $* ) 2>&1`
   ret=$?
-  printf %s "$got" >> configure.log
+  if test "$got" != ""; then
+    printf "%s\n" "$got" >> configure.log
+  fi
   if test $ret -ne 0; then
+    echo "(exit code "$ret")" >> configure.log
     return $ret
   fi
   test "$got" = ""
@@ -457,17 +463,11 @@ size_t dummy = 0;
 EOF
 if try $CC -c $CFLAGS $test.c; then
   echo "Checking for size_t... Yes." | tee -a configure.log
-  need_sizet=0
 else
   echo "Checking for size_t... No." | tee -a configure.log
-  need_sizet=1
+  # find a size_t integer type
-fi
+  # check for long long
-
+  cat > $test.c << EOF
-echo >> configure.log
-
-# find the size_t integer type, if needed
-if test $need_sizet -eq 1; then
-  cat > $test.c <<EOF
 long long dummy = 0;
 EOF
   if try $CC -c $CFLAGS $test.c; then
@@ -495,17 +495,13 @@ EOF
   if try $CC $CFLAGS -o $test $test.c; then
     sizet=`./$test`
     echo "Checking for a pointer-size integer type..." $sizet"." | tee -a configure.log
+    CFLAGS="${CFLAGS} -DNO_SIZE_T=${sizet}"
+    SFLAGS="${SFLAGS} -DNO_SIZE_T=${sizet}"
   else
-    echo "Failed to find a pointer-size integer type." | tee -a configure.log
+    echo "Checking for a pointer-size integer type... not found." | tee -a configure.log
-    leave 1
   fi
 fi
 
-if test $need_sizet -eq 1; then
-  CFLAGS="${CFLAGS} -DNO_SIZE_T=${sizet}"
-  SFLAGS="${SFLAGS} -DNO_SIZE_T=${sizet}"
-fi
-
 echo >> configure.log
 
 # check for large file support, and if none, check for fseeko()
@@ -849,7 +845,6 @@ echo SHAREDLIBV = $SHAREDLIBV >> configure.log
 echo STATICLIB = $STATICLIB >> configure.log
 echo TEST = $TEST >> configure.log
 echo VER = $VER >> configure.log
-echo Z_U4 = $Z_U4 >> configure.log
 echo SRCDIR = $SRCDIR >> configure.log
 echo exec_prefix = $exec_prefix >> configure.log
 echo includedir = $includedir >> configure.log

zlib.h

@@ -1724,7 +1724,7 @@ ZEXTERN uLong ZEXPORT adler32_combine OF((uLong adler1, uLong adler2,
    negative, the result has no meaning or utility.
 */
 
-ZEXTERN uLong ZEXPORT crc32   OF((uLong crc, const Bytef *buf, uInt len));
+ZEXTERN uLong ZEXPORT crc32 OF((uLong crc, const Bytef *buf, uInt len));
 /*
      Update a running CRC-32 with the bytes buf[0..len-1] and return the
    updated CRC-32.  If buf is Z_NULL, this function returns the required
@@ -1758,19 +1758,17 @@ ZEXTERN uLong ZEXPORT crc32_combine OF((uLong crc1, uLong crc2, z_off_t len2));
 */
 
 /*
-ZEXTERN void ZEXPORT crc32_combine_gen OF((z_crc_t op[32], z_off_t len2));
+ZEXTERN uLong ZEXPORT crc32_combine_gen OF((z_off_t len2));
 
-     Generate the operator op corresponding to length len2, to be used with
+     Return the operator corresponding to length len2, to be used with
-   crc32_combine_op(). op must have room for 32 z_crc_t values. (32 is the
+   crc32_combine_op().
-   number of bits in the CRC.)
 */
 
-ZEXTERN uLong ZEXPORT crc32_combine_op OF((uLong crc1, uLong crc2,
+ZEXTERN uLong ZEXPORT crc32_combine_op OF((uLong crc1, uLong crc2, uLong op));
-                                           const z_crc_t *op));
 /*
      Give the same result as crc32_combine(), using op in place of len2. op is
    is generated from len2 by crc32_combine_gen(). This will be faster than
-   crc32_combine() if the generated op is used many times.
+   crc32_combine() if the generated op is used more than once.
 */
 
 
@@ -1860,7 +1858,7 @@ ZEXTERN int ZEXPORT gzgetc_ OF((gzFile file));  /* backward compatibility */
    ZEXTERN z_off64_t ZEXPORT gzoffset64 OF((gzFile));
    ZEXTERN uLong ZEXPORT adler32_combine64 OF((uLong, uLong, z_off64_t));
    ZEXTERN uLong ZEXPORT crc32_combine64 OF((uLong, uLong, z_off64_t));
-   ZEXTERN void ZEXPORT crc32_combine_gen64 OF((z_crc_t *op, z_off64_t));
+   ZEXTERN uLong ZEXPORT crc32_combine_gen64 OF((z_off64_t));
 #endif
 
 #if !defined(ZLIB_INTERNAL) && defined(Z_WANT64)
@@ -1888,7 +1886,7 @@ ZEXTERN int ZEXPORT gzgetc_ OF((gzFile file));  /* backward compatibility */
      ZEXTERN z_off_t ZEXPORT gzoffset64 OF((gzFile));
      ZEXTERN uLong ZEXPORT adler32_combine64 OF((uLong, uLong, z_off_t));
      ZEXTERN uLong ZEXPORT crc32_combine64 OF((uLong, uLong, z_off_t));
-     ZEXTERN void ZEXPORT crc32_combine_gen64 OF((z_crc_t *op, z_off_t));
+     ZEXTERN uLong ZEXPORT crc32_combine_gen64 OF((z_off_t));
 #  endif
 #else
    ZEXTERN gzFile ZEXPORT gzopen OF((const char *, const char *));
@@ -1897,14 +1895,14 @@ ZEXTERN int ZEXPORT gzgetc_ OF((gzFile file));  /* backward compatibility */
    ZEXTERN z_off_t ZEXPORT gzoffset OF((gzFile));
    ZEXTERN uLong ZEXPORT adler32_combine OF((uLong, uLong, z_off_t));
    ZEXTERN uLong ZEXPORT crc32_combine OF((uLong, uLong, z_off_t));
-   ZEXTERN void ZEXPORT crc32_combine_gen OF((z_crc_t *op, z_off_t));
+   ZEXTERN uLong ZEXPORT crc32_combine_gen OF((z_off_t));
 #endif
 
 #else /* Z_SOLO */
 
    ZEXTERN uLong ZEXPORT adler32_combine OF((uLong, uLong, z_off_t));
    ZEXTERN uLong ZEXPORT crc32_combine OF((uLong, uLong, z_off_t));
-   ZEXTERN void ZEXPORT crc32_combine_gen OF((z_crc_t *op, z_off_t));
+   ZEXTERN uLong ZEXPORT crc32_combine_gen OF((z_off_t));
 
 #endif /* !Z_SOLO */
 

zutil.h

@@ -42,6 +42,17 @@ typedef unsigned short ush;
 typedef ush FAR ushf;
 typedef unsigned long  ulg;
 
+#if !defined(Z_U8) && !defined(Z_SOLO) && defined(STDC)
+#  include <limits.h>
+#  if (ULONG_MAX == 0xffffffffffffffffULL)
+#    define Z_U8 unsigned long
+#  elif (ULLONG_MAX == 0xffffffffffffffffULL)
+#    define Z_U8 unsigned long long
+#  elif (UINT_MAX == 0xffffffffffffffffULL)
+#    define Z_U8 unsigned
+#  endif
+#endif
+
 extern z_const char * const z_errmsg[10]; /* indexed by 2-zlib_error */
 /* (size given to avoid silly warnings with Visual C++) */