compress:check more bytes to reduce ZSTD_count call

Comparing 4B instead of comparing 1B in ZSTD_noDict
mode, thus it can avoid cases like match in match[ml]
but mismatch in match[ml-3]..match[ml-1]. So the call
count of ZSTD_count can be reduced.

Signed-off-by: Jun He <jun.he@arm.com>
Change-Id: I3449ea423d5c8e8344f75341f19a2d1643c703f6

diff --git a/lib/compress/zstd_lazy.c b/lib/compress/zstd_lazy.c
index e54b43c0cd40e765dd2cb8fc5f238ac3914c8472..3e2ee1dda01e0aa796371aac07f9208701cb23f0 100644 (file)
--- a/lib/compress/zstd_lazy.c
+++ b/lib/compress/zstd_lazy.c
@@ -692,7 +692,8 @@ size_t ZSTD_HcFindBestMatch(
         if ((dictMode != ZSTD_extDict) || matchIndex >= dictLimit) {
             const BYTE* const match = base + matchIndex;
             assert(matchIndex >= dictLimit);   /* ensures this is true if dictMode != ZSTD_extDict */
-            if (match[ml] == ip[ml])   /* potentially better */
+            /* read 4B starting from (match + ml + 1 - sizeof(U32)) */
+            if (MEM_read32(match + ml - 3) == MEM_read32(ip + ml - 3))   /* potentially better */
                 currentMl = ZSTD_count(ip, match, iLimit);
         } else {
             const BYTE* const match = dictBase + matchIndex;
@@ -1244,7 +1245,8 @@ size_t ZSTD_RowFindBestMatch(
             if ((dictMode != ZSTD_extDict) || matchIndex >= dictLimit) {
                 const BYTE* const match = base + matchIndex;
                 assert(matchIndex >= dictLimit);   /* ensures this is true if dictMode != ZSTD_extDict */
-                if (match[ml] == ip[ml])   /* potentially better */
+                /* read 4B starting from (match + ml + 1 - sizeof(U32)) */
+                if (MEM_read32(match + ml - 3) == MEM_read32(ip + ml - 3))   /* potentially better */
                     currentMl = ZSTD_count(ip, match, iLimit);
             } else {
                 const BYTE* const match = dictBase + matchIndex;