NEON implementation for Adler32

third_party/zlib/adler32.c

 /* adler32.c -- compute the Adler-32 checksum of a data stream
  * Copyright (C) 1995-2011, 2016 Mark Adler
  * For conditions of distribution and use, see copyright notice in zlib.h
  */
 
 /* @(#) $Id$ */
 
 #include "zutil.h"
+#ifdef __ARM_NEON__
+#include "neon_adler32.h"
+#endif
 
 local uLong adler32_combine_ OF((uLong adler1, uLong adler2, z_off64_t len2));
 
 #define BASE 65521U     /* largest prime smaller than 65536 */
 #define NMAX 5552
 /* NMAX is the largest n such that 255n(n+1)/2 + (n+1)(BASE-1) <= 2^32-1 */
 
 #define DO1(buf,i)  {adler += (buf)[i]; sum2 += adler;}
 #define DO2(buf,i)  DO1(buf,i); DO1(buf,i+1);
 #define DO4(buf,i)  DO2(buf,i); DO2(buf,i+2);
@@ skipping 39 matching lines @@
 #  define MOD28(a) a %= BASE
 #  define MOD63(a) a %= BASE
 #endif
 
 /* ========================================================================= */
 uLong ZEXPORT adler32_z(adler, buf, len)
     uLong adler;
     const Bytef *buf;
     z_size_t len;
 {
+#ifdef __ARM_NEON__
+    if (len > 31) {

[f(malita), 2017/02/21 14:36:37]

nit: local code style doesn't seem to require braces for single-line conditional
blocks.

+      return NEON_adler32(adler, buf, len);

[f(malita), 2017/02/21 14:36:37]

general nit: local code style is 4x-space indentation.

+    }
+#endif
+
     unsigned long sum2;
     unsigned n;
 
     /* split Adler-32 into component sums */
     sum2 = (adler >> 16) & 0xffff;
     adler &= 0xffff;
 
     /* in case user likes doing a byte at a time, keep it fast */
     if (len == 1) {
         adler += buf[0];
@@ skipping 99 matching lines @@
     return adler32_combine_(adler1, adler2, len2);
 }
 
 uLong ZEXPORT adler32_combine64(adler1, adler2, len2)
     uLong adler1;
     uLong adler2;
     z_off64_t len2;
 {
     return adler32_combine_(adler1, adler2, len2);
 }