Large decompression speed improvement for GCC 32-bits. Thanks to Valery Croizier !

LZ4HC : Compression Level is now a programmable parameter (CLI from 4 to 9)
Separated IO routines from command line (lz4io.c)
Version number into lz4.h (suggested by Francesc Alted)


git-svn-id: https://lz4.googlecode.com/svn/trunk@113 650e7d94-2a16-8b24-b05c-7c0b3f6821cd

1 files changed, 170 insertions, 136 deletions

diff --git a/lz4hc.c b/lz4hc.c
index f28283f..e84de2b 100644
--- a/lz4hc.c
+++ b/lz4hc.c
@@ -1,6 +1,6 @@
 /*
    LZ4 HC - High Compression Mode of LZ4
-   Copyright (C) 2011-2013, Yann Collet.
+   Copyright (C) 2011-2014, Yann Collet.
    BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
 
    Redistribution and use in source and binary forms, with or without
@@ -31,31 +31,41 @@
    - LZ4 source repository : http://code.google.com/p/lz4/
 */
 
-//**************************************
-// Memory routines
-//**************************************
-#include <stdlib.h>   // calloc, free
+
+
+/**************************************
+   Tuning Parameter
+**************************************/
+#define LZ4HC_DEFAULT_COMPRESSIONLEVEL 8
+
+
+/**************************************
+   Memory routines
+**************************************/
+#include <stdlib.h>   /* calloc, free */
 #define ALLOCATOR(s)  calloc(1,s)
 #define FREEMEM       free
-#include <string.h>   // memset, memcpy
+#include <string.h>   /* memset, memcpy */
 #define MEM_INIT      memset
 
 
-//**************************************
-// CPU Feature Detection
-//**************************************
-// 32 or 64 bits ?
+/**************************************
+   CPU Feature Detection
+**************************************/
+/* 32 or 64 bits ? */
 #if (defined(__x86_64__) || defined(_M_X64) || defined(_WIN64) \
   || defined(__powerpc64__) || defined(__ppc64__) || defined(__PPC64__) \
   || defined(__64BIT__) || defined(_LP64) || defined(__LP64__) \
-  || defined(__ia64) || defined(__itanium__) || defined(_M_IA64) )   // Detects 64 bits mode
+  || defined(__ia64) || defined(__itanium__) || defined(_M_IA64) )   /* Detects 64 bits mode */
 #  define LZ4_ARCH64 1
 #else
 #  define LZ4_ARCH64 0
 #endif
 
-// Little Endian or Big Endian ?
-// Overwrite the #define below if you know your architecture endianess
+/*
+ * Little Endian or Big Endian ?
+ * Overwrite the #define below if you know your architecture endianess
+ */
 #if defined (__GLIBC__)
 #  include <endian.h>
 #  if (__BYTE_ORDER == __BIG_ENDIAN)
@@ -69,43 +79,45 @@
    || defined(_MIPSEB) || defined(__s390__)
 #  define LZ4_BIG_ENDIAN 1
 #else
-// Little Endian assumed. PDP Endian and other very rare endian format are unsupported.
+/* Little Endian assumed. PDP Endian and other very rare endian format are unsupported. */
 #endif
 
-// Unaligned memory access is automatically enabled for "common" CPU, such as x86.
-// For others CPU, the compiler will be more cautious, and insert extra code to ensure aligned access is respected
-// If you know your target CPU supports unaligned memory access, you want to force this option manually to improve performance
+/*
+ * Unaligned memory access is automatically enabled for "common" CPU, such as x86.
+ * For others CPU, the compiler will be more cautious, and insert extra code to ensure aligned access is respected
+ * If you know your target CPU supports unaligned memory access, you want to force this option manually to improve performance
+ */
 #if defined(__ARM_FEATURE_UNALIGNED)
 #  define LZ4_FORCE_UNALIGNED_ACCESS 1
 #endif
 
-// Define this parameter if your target system or compiler does not support hardware bit count
-#if defined(_MSC_VER) && defined(_WIN32_WCE)            // Visual Studio for Windows CE does not support Hardware bit count
+/* Define this parameter if your target system or compiler does not support hardware bit count */
+#if defined(_MSC_VER) && defined(_WIN32_WCE)            /* Visual Studio for Windows CE does not support Hardware bit count */
 #  define LZ4_FORCE_SW_BITCOUNT
 #endif
 
 
-//**************************************
-// Compiler Options
-//**************************************
-#if defined (__STDC_VERSION__) && __STDC_VERSION__ >= 199901L   // C99
-  /* "restrict" is a known keyword */
+/**************************************
+ Compiler Options
+**************************************/
+#if defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L)   /* C99 */
+/* "restrict" is a known keyword */
 #else
-#  define restrict  // Disable restrict
+#  define restrict /* Disable restrict */
 #endif
 
-#ifdef _MSC_VER    // Visual Studio
+#ifdef _MSC_VER    /* Visual Studio */
 #  define FORCE_INLINE static __forceinline
-#  include <intrin.h>                    // For Visual 2005
-#  if LZ4_ARCH64   // 64-bits
-#    pragma intrinsic(_BitScanForward64) // For Visual 2005
-#    pragma intrinsic(_BitScanReverse64) // For Visual 2005
-#  else            // 32-bits
-#    pragma intrinsic(_BitScanForward)   // For Visual 2005
-#    pragma intrinsic(_BitScanReverse)   // For Visual 2005
+#  include <intrin.h>                    /* For Visual 2005 */
+#  if LZ4_ARCH64   /* 64-bits */
+#    pragma intrinsic(_BitScanForward64) /* For Visual 2005 */
+#    pragma intrinsic(_BitScanReverse64) /* For Visual 2005 */
+#  else            /* 32-bits */
+#    pragma intrinsic(_BitScanForward)   /* For Visual 2005 */
+#    pragma intrinsic(_BitScanReverse)   /* For Visual 2005 */
 #  endif
-#  pragma warning(disable : 4127)        // disable: C4127: conditional expression is constant
-#  pragma warning(disable : 4701)        // disable: C4701: potentially uninitialized local variable used
+#  pragma warning(disable : 4127)        /* disable: C4127: conditional expression is constant */
+#  pragma warning(disable : 4701)        /* disable: C4701: potentially uninitialized local variable used */
 #else
 #  ifdef __GNUC__
 #    define FORCE_INLINE static inline __attribute__((always_inline))
@@ -114,24 +126,24 @@
 #  endif
 #endif
 
-#ifdef _MSC_VER  // Visual Studio
+#ifdef _MSC_VER  /* Visual Studio */
 #  define lz4_bswap16(x) _byteswap_ushort(x)
 #else
 #  define lz4_bswap16(x)  ((unsigned short int) ((((x) >> 8) & 0xffu) | (((x) & 0xffu) << 8)))
 #endif
 
 
-//**************************************
-// Includes
-//**************************************
+/**************************************
+   Includes
+**************************************/
 #include "lz4hc.h"
 #include "lz4.h"
 
 
-//**************************************
-// Basic Types
-//**************************************
-#if defined (__STDC_VERSION__) && __STDC_VERSION__ >= 199901L   // C99
+/**************************************
+   Basic Types
+**************************************/
+#if defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L)   /* C99 */
 # include <stdint.h>
   typedef uint8_t  BYTE;
   typedef uint16_t U16;
@@ -173,9 +185,9 @@ typedef struct _U64_S { U64 v; } _PACKED U64_S;
 #define A16(x) (((U16_S *)(x))->v)
 
 
-//**************************************
-// Constants
-//**************************************
+/**************************************
+   Constants
+**************************************/
 #define MINMATCH 4
 
 #define DICTIONARY_LOGSIZE 16
@@ -187,8 +199,6 @@ typedef struct _U64_S { U64 v; } _PACKED U64_S;
 #define HASHTABLESIZE (1 << HASH_LOG)
 #define HASH_MASK (HASHTABLESIZE - 1)
 
-#define MAX_NB_ATTEMPTS 256
-
 #define ML_BITS  4
 #define ML_MASK  (size_t)((1U<<ML_BITS)-1)
 #define RUN_BITS (8-ML_BITS)
@@ -205,25 +215,21 @@ typedef struct _U64_S { U64 v; } _PACKED U64_S;
 #define GB *(1U<<30)
 
 
-//**************************************
-// Architecture-specific macros
-//**************************************
-#if LZ4_ARCH64   // 64-bit
+/**************************************
+   Architecture-specific macros
+**************************************/
+#if LZ4_ARCH64   /* 64-bit */
 #  define STEPSIZE 8
 #  define LZ4_COPYSTEP(s,d)     A64(d) = A64(s); d+=8; s+=8;
 #  define LZ4_COPYPACKET(s,d)   LZ4_COPYSTEP(s,d)
-#  define UARCH U64
 #  define AARCH A64
 #  define HTYPE                 U32
 #  define INITBASE(b,s)         const BYTE* const b = s
-#else   // 32-bit
+#else            /* 32-bit */
 #  define STEPSIZE 4
 #  define LZ4_COPYSTEP(s,d)     A32(d) = A32(s); d+=4; s+=4;
 #  define LZ4_COPYPACKET(s,d)   LZ4_COPYSTEP(s,d); LZ4_COPYSTEP(s,d);
-#  define UARCH U32
 #  define AARCH A32
-//#  define HTYPE                 const BYTE*
-//#  define INITBASE(b,s)         const int b = 0
 #  define HTYPE                 U32
 #  define INITBASE(b,s)         const BYTE* const b = s
 #endif
@@ -231,15 +237,15 @@ typedef struct _U64_S { U64 v; } _PACKED U64_S;
 #if defined(LZ4_BIG_ENDIAN)
 #  define LZ4_READ_LITTLEENDIAN_16(d,s,p) { U16 v = A16(p); v = lz4_bswap16(v); d = (s) - v; }
 #  define LZ4_WRITE_LITTLEENDIAN_16(p,i)  { U16 v = (U16)(i); v = lz4_bswap16(v); A16(p) = v; p+=2; }
-#else   // Little Endian
+#else      /* Little Endian */
 #  define LZ4_READ_LITTLEENDIAN_16(d,s,p) { d = (s) - A16(p); }
 #  define LZ4_WRITE_LITTLEENDIAN_16(p,v)  { A16(p) = v; p+=2; }
 #endif
 
 
-//************************************************************
-// Local Types
-//************************************************************
+/**************************************
+   Local Types
+**************************************/
 typedef struct
 {
     const BYTE* inputBuffer;
@@ -251,9 +257,9 @@ typedef struct
 } LZ4HC_Data_Structure;
 
 
-//**************************************
-// Macros
-//**************************************
+/**************************************
+   Macros
+**************************************/
 #define LZ4_WILDCOPY(s,d,e)    do { LZ4_COPYPACKET(s,d) } while (d<e);
 #define LZ4_BLINDCOPY(s,d,l)   { BYTE* e=d+l; LZ4_WILDCOPY(s,d,e); d=e; }
 #define HASH_FUNCTION(i)       (((i) * 2654435761U) >> ((MINMATCH*8)-HASH_LOG))
@@ -263,9 +269,9 @@ typedef struct
 #define GETNEXT(p)             ((p) - (size_t)DELTANEXT(p))
 
 
-//**************************************
-// Private functions
-//**************************************
+/**************************************
+ Private functions
+**************************************/
 #if LZ4_ARCH64
 
 FORCE_INLINE int LZ4_NbCommonBytes (register U64 val)
@@ -349,7 +355,7 @@ FORCE_INLINE void LZ4_initHC (LZ4HC_Data_Structure* hc4, const BYTE* base)
 
 int LZ4_resetStreamStateHC(void* state, const char* inputBuffer)
 {
-    if ((((size_t)state) & (sizeof(void*)-1)) != 0) return 1;   // Error : pointer is not aligned for pointer (32 or 64 bits)
+    if ((((size_t)state) & (sizeof(void*)-1)) != 0) return 1;   /* Error : pointer is not aligned for pointer (32 or 64 bits) */
     LZ4_initHC((LZ4HC_Data_Structure*)state, (const BYTE*)inputBuffer);
     return 0;
 }
@@ -370,7 +376,7 @@ int LZ4_freeHC (void* LZ4HC_Data)
 }
 
 
-// Update chains up to ip (excluded)
+/* Update chains up to ip (excluded) */
 FORCE_INLINE void LZ4HC_Insert (LZ4HC_Data_Structure* hc4, const BYTE* ip)
 {
     U16*   chainTable = hc4->chainTable;
@@ -393,12 +399,12 @@ char* LZ4_slideInputBufferHC(void* LZ4HC_Data)
 {
     LZ4HC_Data_Structure* hc4 = (LZ4HC_Data_Structure*)LZ4HC_Data;
     U32 distance = (U32)(hc4->end - hc4->inputBuffer) - 64 KB;
-    distance = (distance >> 16) << 16;   // Must be a multiple of 64 KB
+    distance = (distance >> 16) << 16;   /* Must be a multiple of 64 KB */
     LZ4HC_Insert(hc4, hc4->end - MINMATCH);
     memcpy((void*)(hc4->end - 64 KB - distance), (const void*)(hc4->end - 64 KB), 64 KB);
     hc4->nextToUpdate -= distance;
     hc4->base -= distance;
-    if ((U32)(hc4->inputBuffer - hc4->base) > 1 GB + 64 KB)   // Avoid overflow
+    if ((U32)(hc4->inputBuffer - hc4->base) > 1 GB + 64 KB)   /* Avoid overflow */
     {
         int i;
         hc4->base += 1 GB;
@@ -415,7 +421,7 @@ FORCE_INLINE size_t LZ4HC_CommonLength (const BYTE* p1, const BYTE* p2, const BY
 
     while (p1t<matchlimit-(STEPSIZE-1))
     {
-        UARCH diff = AARCH(p2) ^ AARCH(p1t);
+        size_t diff = AARCH(p2) ^ AARCH(p1t);
         if (!diff) { p1t+=STEPSIZE; p2+=STEPSIZE; continue; }
         p1t += LZ4_NbCommonBytes(diff);
         return (p1t - p1);
@@ -427,26 +433,26 @@ FORCE_INLINE size_t LZ4HC_CommonLength (const BYTE* p1, const BYTE* p2, const BY
 }
 
 
-FORCE_INLINE int LZ4HC_InsertAndFindBestMatch (LZ4HC_Data_Structure* hc4, const BYTE* ip, const BYTE* const matchlimit, const BYTE** matchpos)
+FORCE_INLINE int LZ4HC_InsertAndFindBestMatch (LZ4HC_Data_Structure* hc4, const BYTE* ip, const BYTE* const matchlimit, const BYTE** matchpos, const int maxNbAttempts)
 {
     U16* const chainTable = hc4->chainTable;
     HTYPE* const HashTable = hc4->hashTable;
     const BYTE* ref;
     INITBASE(base,hc4->base);
-    int nbAttempts=MAX_NB_ATTEMPTS;
+    int nbAttempts=maxNbAttempts;
     size_t repl=0, ml=0;
-    U16 delta=0;  // useless assignment, to remove an uninitialization warning
+    U16 delta=0;  /* useless assignment, to remove an uninitialization warning */
 
-    // HC4 match finder
+    /* HC4 match finder */
     LZ4HC_Insert(hc4, ip);
     ref = HASH_POINTER(ip);
 
 #define REPEAT_OPTIMIZATION
 #ifdef REPEAT_OPTIMIZATION
-    // Detect repetitive sequences of length <= 4
-    if ((U32)(ip-ref) <= 4)        // potential repetition
+    /* Detect repetitive sequences of length <= 4 */
+    if ((U32)(ip-ref) <= 4)        /* potential repetition */
     {
-        if (A32(ref) == A32(ip))   // confirmed
+        if (A32(ref) == A32(ip))   /* confirmed */
         {
             delta = (U16)(ip-ref);
             repl = ml  = LZ4HC_CommonLength(ip+MINMATCH, ref+MINMATCH, matchlimit) + MINMATCH;
@@ -469,7 +475,7 @@ FORCE_INLINE int LZ4HC_InsertAndFindBestMatch (LZ4HC_Data_Structure* hc4, const
     }
 
 #ifdef REPEAT_OPTIMIZATION
-    // Complete table
+    /* Complete table */
     if (repl)
     {
         const BYTE* ptr = ip;
@@ -478,13 +484,13 @@ FORCE_INLINE int LZ4HC_InsertAndFindBestMatch (LZ4HC_Data_Structure* hc4, const
         end = ip + repl - (MINMATCH-1);
         while(ptr < end-delta)
         {
-            DELTANEXT(ptr) = delta;    // Pre-Load
+            DELTANEXT(ptr) = delta;    /* Pre-Load */
             ptr++;
         }
         do
         {
             DELTANEXT(ptr) = delta;
-            HashTable[HASH_VALUE(ptr)] = (HTYPE)((ptr) - base);     // Head of chain
+            HashTable[HASH_VALUE(ptr)] = (HTYPE)((ptr) - base);     /* Head of chain */
             ptr++;
         } while(ptr < end);
         hc4->nextToUpdate = end;
@@ -495,16 +501,16 @@ FORCE_INLINE int LZ4HC_InsertAndFindBestMatch (LZ4HC_Data_Structure* hc4, const
 }
 
 
-FORCE_INLINE int LZ4HC_InsertAndGetWiderMatch (LZ4HC_Data_Structure* hc4, const BYTE* ip, const BYTE* startLimit, const BYTE* matchlimit, int longest, const BYTE** matchpos, const BYTE** startpos)
+FORCE_INLINE int LZ4HC_InsertAndGetWiderMatch (LZ4HC_Data_Structure* hc4, const BYTE* ip, const BYTE* startLimit, const BYTE* matchlimit, int longest, const BYTE** matchpos, const BYTE** startpos, const int maxNbAttempts)
 {
     U16* const  chainTable = hc4->chainTable;
     HTYPE* const HashTable = hc4->hashTable;
     INITBASE(base,hc4->base);
     const BYTE*  ref;
-    int nbAttempts = MAX_NB_ATTEMPTS;
+    int nbAttempts = maxNbAttempts;
     int delta = (int)(ip-startLimit);
 
-    // First Match
+    /* First Match */
     LZ4HC_Insert(hc4, ip);
     ref = HASH_POINTER(ip);
 
@@ -521,7 +527,7 @@ FORCE_INLINE int LZ4HC_InsertAndGetWiderMatch (LZ4HC_Data_Structure* hc4, const
 
             while (ipt<matchlimit-(STEPSIZE-1))
             {
-                UARCH diff = AARCH(reft) ^ AARCH(ipt);
+                size_t diff = AARCH(reft) ^ AARCH(ipt);
                 if (!diff) { ipt+=STEPSIZE; reft+=STEPSIZE; continue; }
                 ipt += LZ4_NbCommonBytes(diff);
                 goto _endCount;
@@ -532,7 +538,7 @@ FORCE_INLINE int LZ4HC_InsertAndGetWiderMatch (LZ4HC_Data_Structure* hc4, const
 _endCount:
             reft = ref;
 #else
-            // Easier for code maintenance, but unfortunately slower too
+            /* Easier for code maintenance, but unfortunately slower too */
             const BYTE* startt = ip;
             const BYTE* reft = ref;
             const BYTE* ipt = ip + MINMATCH + LZ4HC_CommonLength(ip+MINMATCH, ref+MINMATCH, matchlimit);
@@ -568,26 +574,26 @@ FORCE_INLINE int LZ4HC_encodeSequence (
     int length;
     BYTE* token;
 
-    // Encode Literal length
+    /* Encode Literal length */
     length = (int)(*ip - *anchor);
     token = (*op)++;
-    if ((limitedOutputBuffer) && ((*op + length + (2 + 1 + LASTLITERALS) + (length>>8)) > oend)) return 1;   // Check output limit
+    if ((limitedOutputBuffer) && ((*op + length + (2 + 1 + LASTLITERALS) + (length>>8)) > oend)) return 1;   /* Check output limit */
     if (length>=(int)RUN_MASK) { int len; *token=(RUN_MASK<<ML_BITS); len = length-RUN_MASK; for(; len > 254 ; len-=255) *(*op)++ = 255;  *(*op)++ = (BYTE)len; }
     else *token = (BYTE)(length<<ML_BITS);
 
-    // Copy Literals
+    /* Copy Literals */
     LZ4_BLINDCOPY(*anchor, *op, length);
 
-    // Encode Offset
+    /* Encode Offset */
     LZ4_WRITE_LITTLEENDIAN_16(*op,(U16)(*ip-ref));
 
-    // Encode MatchLength
+    /* Encode MatchLength */
     length = (int)(matchLength-MINMATCH);
-    if ((limitedOutputBuffer) && (*op + (1 + LASTLITERALS) + (length>>8) > oend)) return 1;   // Check output limit
+    if ((limitedOutputBuffer) && (*op + (1 + LASTLITERALS) + (length>>8) > oend)) return 1;   /* Check output limit */
     if (length>=(int)ML_MASK) { *token+=ML_MASK; length-=ML_MASK; for(; length > 509 ; length-=510) { *(*op)++ = 255; *(*op)++ = 255; } if (length > 254) { length-=255; *(*op)++ = 255; } *(*op)++ = (BYTE)length; }
     else *token += (BYTE)(length);
 
-    // Prepare next loop
+    /* Prepare next loop */
     *ip += matchLength;
     *anchor = *ip;
 
@@ -595,12 +601,14 @@ FORCE_INLINE int LZ4HC_encodeSequence (
 }
 
 
+#define MAX_COMPRESSION_LEVEL 16
 static int LZ4HC_compress_generic (
                  void* ctxvoid,
                  const char* source,
                  char* dest,
                  int inputSize,
                  int maxOutputSize,
+                 int compressionLevel,
                  limitedOutput_directive limit
                 )
 {
@@ -614,6 +622,7 @@ static int LZ4HC_compress_generic (
     BYTE* op = (BYTE*) dest;
     BYTE* const oend = op + maxOutputSize;
 
+    const int maxNbAttempts = compressionLevel > MAX_COMPRESSION_LEVEL ? 1 << MAX_COMPRESSION_LEVEL : compressionLevel ? 1<<(compressionLevel-1) : 1<<LZ4HC_DEFAULT_COMPRESSIONLEVEL;
     int   ml, ml2, ml3, ml0;
     const BYTE* ref=NULL;
     const BYTE* start2=NULL;
@@ -624,29 +633,29 @@ static int LZ4HC_compress_generic (
     const BYTE* ref0;
 
 
-    // Ensure blocks follow each other
+    /* Ensure blocks follow each other */
     if (ip != ctx->end) return 0;
     ctx->end += inputSize;
 
     ip++;
 
-    // Main Loop
+    /* Main Loop */
     while (ip < mflimit)
     {
-        ml = LZ4HC_InsertAndFindBestMatch (ctx, ip, matchlimit, (&ref));
+        ml = LZ4HC_InsertAndFindBestMatch (ctx, ip, matchlimit, (&ref), maxNbAttempts);
         if (!ml) { ip++; continue; }
 
-        // saved, in case we would skip too much
+        /* saved, in case we would skip too much */
         start0 = ip;
         ref0 = ref;
         ml0 = ml;
 
 _Search2:
         if (ip+ml < mflimit)
-            ml2 = LZ4HC_InsertAndGetWiderMatch(ctx, ip + ml - 2, ip + 1, matchlimit, ml, &ref2, &start2);
+            ml2 = LZ4HC_InsertAndGetWiderMatch(ctx, ip + ml - 2, ip + 1, matchlimit, ml, &ref2, &start2, maxNbAttempts);
         else ml2 = ml;
 
-        if (ml2 == ml)  // No better match
+        if (ml2 == ml)  /* No better match */
         {
             if (LZ4HC_encodeSequence(&ip, &op, &anchor, ml, ref, limit, oend)) return 0;
             continue;
@@ -654,7 +663,7 @@ _Search2:
 
         if (start0 < ip)
         {
-            if (start2 < ip + ml0)   // empirical
+            if (start2 < ip + ml0)   /* empirical */
             {
                 ip = start0;
                 ref = ref0;
@@ -662,8 +671,8 @@ _Search2:
             }
         }
 
-        // Here, start0==ip
-        if ((start2 - ip) < 3)   // First Match too small : removed
+        /* Here, start0==ip */
+        if ((start2 - ip) < 3)   /* First Match too small : removed */
         {
             ml = ml2;
             ip = start2;
@@ -672,9 +681,11 @@ _Search2:
         }
 
 _Search3:
-        // Currently we have :
-        // ml2 > ml1, and
-        // ip1+3 <= ip2 (usually < ip1+ml1)
+        /*
+         * Currently we have :
+         * ml2 > ml1, and
+         * ip1+3 <= ip2 (usually < ip1+ml1)
+         */
         if ((start2 - ip) < OPTIMAL_ML)
         {
             int correction;
@@ -689,26 +700,26 @@ _Search3:
                 ml2 -= correction;
             }
         }
-        // Now, we have start2 = ip+new_ml, with new_ml = min(ml, OPTIMAL_ML=18)
+        /* Now, we have start2 = ip+new_ml, with new_ml = min(ml, OPTIMAL_ML=18) */
 
         if (start2 + ml2 < mflimit)
-            ml3 = LZ4HC_InsertAndGetWiderMatch(ctx, start2 + ml2 - 3, start2, matchlimit, ml2, &ref3, &start3);
+            ml3 = LZ4HC_InsertAndGetWiderMatch(ctx, start2 + ml2 - 3, start2, matchlimit, ml2, &ref3, &start3, maxNbAttempts);
         else ml3 = ml2;
 
-        if (ml3 == ml2) // No better match : 2 sequences to encode
+        if (ml3 == ml2) /* No better match : 2 sequences to encode */
         {
-            // ip & ref are known; Now for ml
+            /* ip & ref are known; Now for ml */
             if (start2 < ip+ml)  ml = (int)(start2 - ip);
-            // Now, encode 2 sequences
+            /* Now, encode 2 sequences */
             if (LZ4HC_encodeSequence(&ip, &op, &anchor, ml, ref, limit, oend)) return 0;
             ip = start2;
             if (LZ4HC_encodeSequence(&ip, &op, &anchor, ml2, ref2, limit, oend)) return 0;
             continue;
         }
 
-        if (start3 < ip+ml+3) // Not enough space for match 2 : remove it
+        if (start3 < ip+ml+3) /* Not enough space for match 2 : remove it */
         {
-            if (start3 >= (ip+ml)) // can write Seq1 immediately ==> Seq2 is removed, so Seq3 becomes Seq1
+            if (start3 >= (ip+ml)) /* can write Seq1 immediately ==> Seq2 is removed, so Seq3 becomes Seq1 */
             {
                 if (start2 < ip+ml)
                 {
@@ -741,8 +752,10 @@ _Search3:
             goto _Search3;
         }
 
-        // OK, now we have 3 ascending matches; let's write at least the first one
-        // ip & ref are known; Now for ml
+        /*
+         * OK, now we have 3 ascending matches; let's write at least the first one
+         * ip & ref are known; Now for ml
+         */
         if (start2 < ip+ml)
         {
             if ((start2 - ip) < (int)ML_MASK)
@@ -777,80 +790,101 @@ _Search3:
 
     }
 
-    // Encode Last Literals
+    /* Encode Last Literals */
     {
         int lastRun = (int)(iend - anchor);
-        if ((limit) && (((char*)op - dest) + lastRun + 1 + ((lastRun+255-RUN_MASK)/255) > (U32)maxOutputSize)) return 0;  // Check output limit
+        if ((limit) && (((char*)op - dest) + lastRun + 1 + ((lastRun+255-RUN_MASK)/255) > (U32)maxOutputSize)) return 0;  /* Check output limit */
         if (lastRun>=(int)RUN_MASK) { *op++=(RUN_MASK<<ML_BITS); lastRun-=RUN_MASK; for(; lastRun > 254 ; lastRun-=255) *op++ = 255; *op++ = (BYTE) lastRun; }
         else *op++ = (BYTE)(lastRun<<ML_BITS);
         memcpy(op, anchor, iend - anchor);
         op += iend-anchor;
     }
 
-    // End
+    /* End */
     return (int) (((char*)op)-dest);
 }
 
 
-int LZ4_compressHC(const char* source, char* dest, int inputSize)
+int LZ4_compressHC2(const char* source, char* dest, int inputSize, int compressionLevel)
 {
     void* ctx = LZ4_createHC(source);
     int result;
     if (ctx==NULL) return 0;
 
-    result = LZ4HC_compress_generic (ctx, source, dest, inputSize, 0, noLimit);
+    result = LZ4HC_compress_generic (ctx, source, dest, inputSize, 0, compressionLevel, noLimit);
 
     LZ4_freeHC(ctx);
     return result;
 }
 
-int LZ4_compressHC_limitedOutput(const char* source, char* dest, int inputSize, int maxOutputSize)
+int LZ4_compressHC(const char* source, char* dest, int inputSize) { return LZ4_compressHC2(source, dest, inputSize, 0); }
+
+int LZ4_compressHC2_limitedOutput(const char* source, char* dest, int inputSize, int maxOutputSize, int compressionLevel)
 {
     void* ctx = LZ4_createHC(source);
     int result;
     if (ctx==NULL) return 0;
 
-    result = LZ4HC_compress_generic (ctx, source, dest, inputSize, maxOutputSize, limitedOutput);
+    result = LZ4HC_compress_generic (ctx, source, dest, inputSize, maxOutputSize, compressionLevel, limitedOutput);
 
     LZ4_freeHC(ctx);
     return result;
 }
 
+int LZ4_compressHC_limitedOutput(const char* source, char* dest, int inputSize, int maxOutputSize)
+{
+    return LZ4_compressHC2_limitedOutput(source, dest, inputSize, maxOutputSize, 0);
+}
 
-//*****************************
-// Using an external allocation
-//*****************************
 
+/*****************************
+   Using external allocation
+*****************************/
 int LZ4_sizeofStateHC() { return sizeof(LZ4HC_Data_Structure); }
 
 
-int LZ4_compressHC_withStateHC (void* state, const char* source, char* dest, int inputSize)
+int LZ4_compressHC2_withStateHC (void* state, const char* source, char* dest, int inputSize, int compressionLevel)
 {
-    if (((size_t)(state)&(sizeof(void*)-1)) != 0) return 0;   // Error : state is not aligned for pointers (32 or 64 bits)
+    if (((size_t)(state)&(sizeof(void*)-1)) != 0) return 0;   /* Error : state is not aligned for pointers (32 or 64 bits) */
     LZ4_initHC ((LZ4HC_Data_Structure*)state, (const BYTE*)source);
-    return LZ4HC_compress_generic (state, source, dest, inputSize, 0, noLimit);
+    return LZ4HC_compress_generic (state, source, dest, inputSize, 0, compressionLevel, noLimit);
 }
 
+int LZ4_compressHC_withStateHC (void* state, const char* source, char* dest, int inputSize)
+{ return LZ4_compressHC2_withStateHC (state, source, dest, inputSize, 0); }
 
-int LZ4_compressHC_limitedOutput_withStateHC (void* state, const char* source, char* dest, int inputSize, int maxOutputSize)
+
+int LZ4_compressHC2_limitedOutput_withStateHC (void* state, const char* source, char* dest, int inputSize, int maxOutputSize, int compressionLevel)
 {
-    if (((size_t)(state)&(sizeof(void*)-1)) != 0) return 0;   // Error : state is not aligned for pointers (32 or 64 bits)
+    if (((size_t)(state)&(sizeof(void*)-1)) != 0) return 0;   /* Error : state is not aligned for pointers (32 or 64 bits) */
     LZ4_initHC ((LZ4HC_Data_Structure*)state, (const BYTE*)source);
-    return LZ4HC_compress_generic (state, source, dest, inputSize, maxOutputSize, limitedOutput);
+    return LZ4HC_compress_generic (state, source, dest, inputSize, maxOutputSize, compressionLevel, limitedOutput);
 }
 
+int LZ4_compressHC_limitedOutput_withStateHC (void* state, const char* source, char* dest, int inputSize, int maxOutputSize)
+{ return LZ4_compressHC2_limitedOutput_withStateHC (state, source, dest, inputSize, maxOutputSize, 0); }
+
 
-//****************************
-// Stream functions
-//****************************
+/****************************
+   Stream functions
+****************************/
 
 int LZ4_compressHC_continue (void* LZ4HC_Data, const char* source, char* dest, int inputSize)
 {
-    return LZ4HC_compress_generic (LZ4HC_Data, source, dest, inputSize, 0, noLimit);
+    return LZ4HC_compress_generic (LZ4HC_Data, source, dest, inputSize, 0, 0, noLimit);
+}
+
+int LZ4_compressHC2_continue (void* LZ4HC_Data, const char* source, char* dest, int inputSize, int compressionLevel)
+{
+    return LZ4HC_compress_generic (LZ4HC_Data, source, dest, inputSize, 0, compressionLevel, noLimit);
 }
 
 int LZ4_compressHC_limitedOutput_continue (void* LZ4HC_Data, const char* source, char* dest, int inputSize, int maxOutputSize)
 {
-    return LZ4HC_compress_generic (LZ4HC_Data, source, dest, inputSize, maxOutputSize, limitedOutput);
+    return LZ4HC_compress_generic (LZ4HC_Data, source, dest, inputSize, maxOutputSize, 0, limitedOutput);
 }
 
+int LZ4_compressHC2_limitedOutput_continue (void* LZ4HC_Data, const char* source, char* dest, int inputSize, int maxOutputSize, int compressionLevel)
+{
+    return LZ4HC_compress_generic (LZ4HC_Data, source, dest, inputSize, maxOutputSize, compressionLevel, limitedOutput);
+}