X-Git-Url: https://wimlib.net/git/?a=blobdiff_plain;f=include%2Fwimlib%2Flz_extend.h;h=26f0ce5cd11be42f76bbee3508d16e6c119d9772;hb=a60aa6bebc8cc1d72e89be6b9cfb7338f092158d;hp=d47e0976d5836ff0214dcfc6563bec63911a559d;hpb=bffa1310d24ef2e9330dd5ea3b5aef1ff0ca380b;p=wimlib

diff --git a/include/wimlib/lz_extend.h b/include/wimlib/lz_extend.h
index d47e0976..26f0ce5c 100644
--- a/include/wimlib/lz_extend.h
+++ b/include/wimlib/lz_extend.h
@@ -1,10 +1,21 @@
 /*
- * lz_extend.h
+ * lz_extend.h - fast match extension for Lempel-Ziv matchfinding
  *
- * Fast match extension for Lempel-Ziv matchfinding.
+ * The following copying information applies to this specific source code file:
  *
- * The author dedicates this file to the public domain.
- * You can do whatever you want with this file.
+ * Written in 2014-2016 by Eric Biggers <ebiggers3@gmail.com>
+ *
+ * To the extent possible under law, the author(s) have dedicated all copyright
+ * and related and neighboring rights to this software to the public domain
+ * worldwide via the Creative Commons Zero 1.0 Universal Public Domain
+ * Dedication (the "CC0").
+ *
+ * This software 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 CC0 for more details.
+ *
+ * You should have received a copy of the CC0 along with this software; if not
+ * see <http://creativecommons.org/publicdomain/zero/1.0/>.
  */
 
 #ifndef _WIMLIB_LZ_EXTEND_H
@@ -13,53 +24,30 @@
 #include "wimlib/bitops.h"
 #include "wimlib/unaligned.h"
 
-/* Return the number of bytes at @matchptr that match the bytes at @strptr, up
- * to a maximum of @max_len.  Initially, @start_len bytes are matched.  */
-static inline u32
+/*
+ * Return the number of bytes at @matchptr that match the bytes at @strptr, up
+ * to a maximum of @max_len.  Initially, @len bytes are matched.
+ */
+static forceinline u32
 lz_extend(const u8 * const strptr, const u8 * const matchptr,
-	  const u32 start_len, const u32 max_len)
+	  u32 len, const u32 max_len)
 {
-	u32 len = start_len;
-
-	if (UNALIGNED_ACCESS_IS_FAST && CPU_IS_LITTLE_ENDIAN) {
-
-		machine_word_t v_word;
-
-		if (likely(max_len - len >= 4 * WORDSIZE)) {
-
-		#define COMPARE_WORD_STEP					\
-			v_word = load_word_unaligned(&matchptr[len]) ^		\
-				 load_word_unaligned(&strptr[len]);		\
-			if (v_word != 0)					\
-				goto word_differs;				\
-			len += WORDSIZE;					\
-
-			COMPARE_WORD_STEP
-			COMPARE_WORD_STEP
-			COMPARE_WORD_STEP
-			COMPARE_WORD_STEP
-		#undef COMPARE_WORD_STEP
-		}
-
-		while (len + WORDSIZE <= max_len) {
-			v_word = load_word_unaligned(&matchptr[len]) ^
-				 load_word_unaligned(&strptr[len]);
-			if (v_word != 0)
-				goto word_differs;
-			len += WORDSIZE;
+	while (UNALIGNED_ACCESS_IS_FAST && len + WORDBYTES <= max_len) {
+		machine_word_t v = load_word_unaligned(matchptr + len) ^
+				   load_word_unaligned(strptr + len);
+		if (v != 0) {
+			if (CPU_IS_LITTLE_ENDIAN)
+				len += bsfw(v) >> 3;
+			else
+				len += (WORDBITS - 1 - bsrw(v)) >> 3;
+			return len;
 		}
-
-		while (len < max_len && matchptr[len] == strptr[len])
-			len++;
-		return len;
-
-	word_differs:
-		return len + (ffsw(v_word) >> 3);
-	} else {
-		while (len < max_len && matchptr[len] == strptr[len])
-			len++;
-		return len;
+		len += WORDBYTES;
 	}
+
+	while (len < max_len && matchptr[len] == strptr[len])
+		len++;
+	return len;
 }
 
 #endif /* _WIMLIB_LZ_EXTEND_H */