Merge pull request #1022 from facebook/bmi2IntoC

Implemented BMI2 functions directly within huf_decompress.c
This commit is contained in:
Yann Collet
2018-02-22 14:30:43 -08:00
committed by GitHub
2 changed files with 376 additions and 394 deletions
+376 -29
View File
@@ -49,18 +49,19 @@
****************************************************************/
#define HUF_isError ERR_isError
#define HUF_STATIC_ASSERT(c) { enum { HUF_static_assert = 1/(int)(!!(c)) }; } /* use only *after* variable declarations */
#define CHECK_F(f) { size_t const err_ = (f); if (HUF_isError(err_)) return err_; }
/* **************************************************************
* Byte alignment for workSpace management
****************************************************************/
#define HUF_ALIGN(x, a) HUF_ALIGN_MASK((x), (a) - 1)
#define HUF_ALIGN(x, a) HUF_ALIGN_MASK((x), (a) - 1)
#define HUF_ALIGN_MASK(x, mask) (((x) + (mask)) & ~(mask))
/*-***************************/
/* generic DTableDesc */
/*-***************************/
typedef struct { BYTE maxTableLog; BYTE tableType; BYTE tableLog; BYTE reserved; } DTableDesc;
static DTableDesc HUF_getDTableDesc(const HUF_DTable* table)
@@ -74,7 +75,6 @@ static DTableDesc HUF_getDTableDesc(const HUF_DTable* table)
/*-***************************/
/* single-symbol decoding */
/*-***************************/
typedef struct { BYTE byte; BYTE nbBits; } HUF_DEltX2; /* single-symbol decoding */
size_t HUF_readDTableX2_wksp(HUF_DTable* DTable, const void* src, size_t srcSize, void* workSpace, size_t wkspSize)
@@ -146,44 +146,390 @@ size_t HUF_readDTableX2(HUF_DTable* DTable, const void* src, size_t srcSize)
typedef struct { U16 sequence; BYTE nbBits; BYTE length; } HUF_DEltX4; /* double-symbols decoding */
#define FUNCTION(fn) fn##_default
#define TARGET
#include "huf_decompress_impl.h"
#undef TARGET
#undef FUNCTION
static BYTE HUF_decodeSymbolX2(BIT_DStream_t* Dstream, const HUF_DEltX2* dt, const U32 dtLog)
{
size_t const val = BIT_lookBitsFast(Dstream, dtLog); /* note : dtLog >= 1 */
BYTE const c = dt[val].byte;
BIT_skipBits(Dstream, dt[val].nbBits);
return c;
}
#if DYNAMIC_BMI2
#define HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr) \
*ptr++ = HUF_decodeSymbolX2(DStreamPtr, dt, dtLog)
#define FUNCTION(fn) fn##_bmi2
#define TARGET TARGET_ATTRIBUTE("bmi2")
#include "huf_decompress_impl.h"
#undef TARGET
#undef FUNCTION
#define HUF_DECODE_SYMBOLX2_1(ptr, DStreamPtr) \
if (MEM_64bits() || (HUF_TABLELOG_MAX<=12)) \
HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr)
#define HUF_DECODE_SYMBOLX2_2(ptr, DStreamPtr) \
if (MEM_64bits()) \
HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr)
HINT_INLINE size_t
HUF_decodeStreamX2(BYTE* p, BIT_DStream_t* const bitDPtr, BYTE* const pEnd, const HUF_DEltX2* const dt, const U32 dtLog)
{
BYTE* const pStart = p;
/* up to 4 symbols at a time */
while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) && (p <= pEnd-4)) {
HUF_DECODE_SYMBOLX2_2(p, bitDPtr);
HUF_DECODE_SYMBOLX2_1(p, bitDPtr);
HUF_DECODE_SYMBOLX2_2(p, bitDPtr);
HUF_DECODE_SYMBOLX2_0(p, bitDPtr);
}
/* closer to the end */
while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) && (p < pEnd))
HUF_DECODE_SYMBOLX2_0(p, bitDPtr);
/* no more data to retrieve from bitstream, hence no need to reload */
while (p < pEnd)
HUF_DECODE_SYMBOLX2_0(p, bitDPtr);
return pEnd-pStart;
}
FORCE_INLINE_TEMPLATE size_t
HUF_decompress1X2_usingDTable_internal_body(
void* dst, size_t dstSize,
const void* cSrc, size_t cSrcSize,
const HUF_DTable* DTable)
{
BYTE* op = (BYTE*)dst;
BYTE* const oend = op + dstSize;
const void* dtPtr = DTable + 1;
const HUF_DEltX2* const dt = (const HUF_DEltX2*)dtPtr;
BIT_DStream_t bitD;
DTableDesc const dtd = HUF_getDTableDesc(DTable);
U32 const dtLog = dtd.tableLog;
CHECK_F( BIT_initDStream(&bitD, cSrc, cSrcSize) );
HUF_decodeStreamX2(op, &bitD, oend, dt, dtLog);
if (!BIT_endOfDStream(&bitD)) return ERROR(corruption_detected);
return dstSize;
}
FORCE_INLINE_TEMPLATE size_t
HUF_decompress4X2_usingDTable_internal_body(
void* dst, size_t dstSize,
const void* cSrc, size_t cSrcSize,
const HUF_DTable* DTable)
{
/* Check */
if (cSrcSize < 10) return ERROR(corruption_detected); /* strict minimum : jump table + 1 byte per stream */
{ const BYTE* const istart = (const BYTE*) cSrc;
BYTE* const ostart = (BYTE*) dst;
BYTE* const oend = ostart + dstSize;
const void* const dtPtr = DTable + 1;
const HUF_DEltX2* const dt = (const HUF_DEltX2*)dtPtr;
/* Init */
BIT_DStream_t bitD1;
BIT_DStream_t bitD2;
BIT_DStream_t bitD3;
BIT_DStream_t bitD4;
size_t const length1 = MEM_readLE16(istart);
size_t const length2 = MEM_readLE16(istart+2);
size_t const length3 = MEM_readLE16(istart+4);
size_t const length4 = cSrcSize - (length1 + length2 + length3 + 6);
const BYTE* const istart1 = istart + 6; /* jumpTable */
const BYTE* const istart2 = istart1 + length1;
const BYTE* const istart3 = istart2 + length2;
const BYTE* const istart4 = istart3 + length3;
const size_t segmentSize = (dstSize+3) / 4;
BYTE* const opStart2 = ostart + segmentSize;
BYTE* const opStart3 = opStart2 + segmentSize;
BYTE* const opStart4 = opStart3 + segmentSize;
BYTE* op1 = ostart;
BYTE* op2 = opStart2;
BYTE* op3 = opStart3;
BYTE* op4 = opStart4;
U32 endSignal = BIT_DStream_unfinished;
DTableDesc const dtd = HUF_getDTableDesc(DTable);
U32 const dtLog = dtd.tableLog;
if (length4 > cSrcSize) return ERROR(corruption_detected); /* overflow */
CHECK_F( BIT_initDStream(&bitD1, istart1, length1) );
CHECK_F( BIT_initDStream(&bitD2, istart2, length2) );
CHECK_F( BIT_initDStream(&bitD3, istart3, length3) );
CHECK_F( BIT_initDStream(&bitD4, istart4, length4) );
/* up to 16 symbols per loop (4 symbols per stream) in 64-bit mode */
endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
while ( (endSignal==BIT_DStream_unfinished) && (op4<(oend-3)) ) {
HUF_DECODE_SYMBOLX2_2(op1, &bitD1);
HUF_DECODE_SYMBOLX2_2(op2, &bitD2);
HUF_DECODE_SYMBOLX2_2(op3, &bitD3);
HUF_DECODE_SYMBOLX2_2(op4, &bitD4);
HUF_DECODE_SYMBOLX2_1(op1, &bitD1);
HUF_DECODE_SYMBOLX2_1(op2, &bitD2);
HUF_DECODE_SYMBOLX2_1(op3, &bitD3);
HUF_DECODE_SYMBOLX2_1(op4, &bitD4);
HUF_DECODE_SYMBOLX2_2(op1, &bitD1);
HUF_DECODE_SYMBOLX2_2(op2, &bitD2);
HUF_DECODE_SYMBOLX2_2(op3, &bitD3);
HUF_DECODE_SYMBOLX2_2(op4, &bitD4);
HUF_DECODE_SYMBOLX2_0(op1, &bitD1);
HUF_DECODE_SYMBOLX2_0(op2, &bitD2);
HUF_DECODE_SYMBOLX2_0(op3, &bitD3);
HUF_DECODE_SYMBOLX2_0(op4, &bitD4);
BIT_reloadDStream(&bitD1);
BIT_reloadDStream(&bitD2);
BIT_reloadDStream(&bitD3);
BIT_reloadDStream(&bitD4);
}
/* check corruption */
/* note : should not be necessary : op# advance in lock step, and we control op4.
* but curiously, binary generated by gcc 7.2 with -mbmi2 runs faster when they are present */
if (op1 > opStart2) return ERROR(corruption_detected);
if (op2 > opStart3) return ERROR(corruption_detected);
if (op3 > opStart4) return ERROR(corruption_detected);
/* note : op4 supposed already verified within main loop */
/* finish bitStreams one by one */
HUF_decodeStreamX2(op1, &bitD1, opStart2, dt, dtLog);
HUF_decodeStreamX2(op2, &bitD2, opStart3, dt, dtLog);
HUF_decodeStreamX2(op3, &bitD3, opStart4, dt, dtLog);
HUF_decodeStreamX2(op4, &bitD4, oend, dt, dtLog);
/* check */
{ U32 const endCheck = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4);
if (!endCheck) return ERROR(corruption_detected); }
/* decoded size */
return dstSize;
}
}
static U32 HUF_decodeSymbolX4(void* op, BIT_DStream_t* DStream, const HUF_DEltX4* dt, const U32 dtLog)
{
size_t const val = BIT_lookBitsFast(DStream, dtLog); /* note : dtLog >= 1 */
memcpy(op, dt+val, 2);
BIT_skipBits(DStream, dt[val].nbBits);
return dt[val].length;
}
FORCE_INLINE_TEMPLATE U32
HUF_decodeLastSymbolX4(void* op, BIT_DStream_t* DStream, const HUF_DEltX4* dt, const U32 dtLog)
{
size_t const val = BIT_lookBitsFast(DStream, dtLog); /* note : dtLog >= 1 */
memcpy(op, dt+val, 1);
if (dt[val].length==1) BIT_skipBits(DStream, dt[val].nbBits);
else {
if (DStream->bitsConsumed < (sizeof(DStream->bitContainer)*8)) {
BIT_skipBits(DStream, dt[val].nbBits);
if (DStream->bitsConsumed > (sizeof(DStream->bitContainer)*8))
/* ugly hack; works only because it's the last symbol. Note : can't easily extract nbBits from just this symbol */
DStream->bitsConsumed = (sizeof(DStream->bitContainer)*8);
} }
return 1;
}
#define HUF_DECODE_SYMBOLX4_0(ptr, DStreamPtr) \
ptr += HUF_decodeSymbolX4(ptr, DStreamPtr, dt, dtLog)
#define HUF_DECODE_SYMBOLX4_1(ptr, DStreamPtr) \
if (MEM_64bits() || (HUF_TABLELOG_MAX<=12)) \
ptr += HUF_decodeSymbolX4(ptr, DStreamPtr, dt, dtLog)
#define HUF_DECODE_SYMBOLX4_2(ptr, DStreamPtr) \
if (MEM_64bits()) \
ptr += HUF_decodeSymbolX4(ptr, DStreamPtr, dt, dtLog)
HINT_INLINE size_t
HUF_decodeStreamX4(BYTE* p, BIT_DStream_t* bitDPtr, BYTE* const pEnd,
const HUF_DEltX4* const dt, const U32 dtLog)
{
BYTE* const pStart = p;
/* up to 8 symbols at a time */
while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p < pEnd-(sizeof(bitDPtr->bitContainer)-1))) {
HUF_DECODE_SYMBOLX4_2(p, bitDPtr);
HUF_DECODE_SYMBOLX4_1(p, bitDPtr);
HUF_DECODE_SYMBOLX4_2(p, bitDPtr);
HUF_DECODE_SYMBOLX4_0(p, bitDPtr);
}
/* closer to end : up to 2 symbols at a time */
while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p <= pEnd-2))
HUF_DECODE_SYMBOLX4_0(p, bitDPtr);
while (p <= pEnd-2)
HUF_DECODE_SYMBOLX4_0(p, bitDPtr); /* no need to reload : reached the end of DStream */
if (p < pEnd)
p += HUF_decodeLastSymbolX4(p, bitDPtr, dt, dtLog);
return p-pStart;
}
FORCE_INLINE_TEMPLATE size_t
HUF_decompress1X4_usingDTable_internal_body(
void* dst, size_t dstSize,
const void* cSrc, size_t cSrcSize,
const HUF_DTable* DTable)
{
BIT_DStream_t bitD;
/* Init */
CHECK_F( BIT_initDStream(&bitD, cSrc, cSrcSize) );
/* decode */
{ BYTE* const ostart = (BYTE*) dst;
BYTE* const oend = ostart + dstSize;
const void* const dtPtr = DTable+1; /* force compiler to not use strict-aliasing */
const HUF_DEltX4* const dt = (const HUF_DEltX4*)dtPtr;
DTableDesc const dtd = HUF_getDTableDesc(DTable);
HUF_decodeStreamX4(ostart, &bitD, oend, dt, dtd.tableLog);
}
/* check */
if (!BIT_endOfDStream(&bitD)) return ERROR(corruption_detected);
/* decoded size */
return dstSize;
}
FORCE_INLINE_TEMPLATE size_t
HUF_decompress4X4_usingDTable_internal_body(
void* dst, size_t dstSize,
const void* cSrc, size_t cSrcSize,
const HUF_DTable* DTable)
{
if (cSrcSize < 10) return ERROR(corruption_detected); /* strict minimum : jump table + 1 byte per stream */
{ const BYTE* const istart = (const BYTE*) cSrc;
BYTE* const ostart = (BYTE*) dst;
BYTE* const oend = ostart + dstSize;
const void* const dtPtr = DTable+1;
const HUF_DEltX4* const dt = (const HUF_DEltX4*)dtPtr;
/* Init */
BIT_DStream_t bitD1;
BIT_DStream_t bitD2;
BIT_DStream_t bitD3;
BIT_DStream_t bitD4;
size_t const length1 = MEM_readLE16(istart);
size_t const length2 = MEM_readLE16(istart+2);
size_t const length3 = MEM_readLE16(istart+4);
size_t const length4 = cSrcSize - (length1 + length2 + length3 + 6);
const BYTE* const istart1 = istart + 6; /* jumpTable */
const BYTE* const istart2 = istart1 + length1;
const BYTE* const istart3 = istart2 + length2;
const BYTE* const istart4 = istart3 + length3;
size_t const segmentSize = (dstSize+3) / 4;
BYTE* const opStart2 = ostart + segmentSize;
BYTE* const opStart3 = opStart2 + segmentSize;
BYTE* const opStart4 = opStart3 + segmentSize;
BYTE* op1 = ostart;
BYTE* op2 = opStart2;
BYTE* op3 = opStart3;
BYTE* op4 = opStart4;
U32 endSignal;
DTableDesc const dtd = HUF_getDTableDesc(DTable);
U32 const dtLog = dtd.tableLog;
if (length4 > cSrcSize) return ERROR(corruption_detected); /* overflow */
CHECK_F( BIT_initDStream(&bitD1, istart1, length1) );
CHECK_F( BIT_initDStream(&bitD2, istart2, length2) );
CHECK_F( BIT_initDStream(&bitD3, istart3, length3) );
CHECK_F( BIT_initDStream(&bitD4, istart4, length4) );
/* 16-32 symbols per loop (4-8 symbols per stream) */
endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
for ( ; (endSignal==BIT_DStream_unfinished) & (op4<(oend-(sizeof(bitD4.bitContainer)-1))) ; ) {
HUF_DECODE_SYMBOLX4_2(op1, &bitD1);
HUF_DECODE_SYMBOLX4_2(op2, &bitD2);
HUF_DECODE_SYMBOLX4_2(op3, &bitD3);
HUF_DECODE_SYMBOLX4_2(op4, &bitD4);
HUF_DECODE_SYMBOLX4_1(op1, &bitD1);
HUF_DECODE_SYMBOLX4_1(op2, &bitD2);
HUF_DECODE_SYMBOLX4_1(op3, &bitD3);
HUF_DECODE_SYMBOLX4_1(op4, &bitD4);
HUF_DECODE_SYMBOLX4_2(op1, &bitD1);
HUF_DECODE_SYMBOLX4_2(op2, &bitD2);
HUF_DECODE_SYMBOLX4_2(op3, &bitD3);
HUF_DECODE_SYMBOLX4_2(op4, &bitD4);
HUF_DECODE_SYMBOLX4_0(op1, &bitD1);
HUF_DECODE_SYMBOLX4_0(op2, &bitD2);
HUF_DECODE_SYMBOLX4_0(op3, &bitD3);
HUF_DECODE_SYMBOLX4_0(op4, &bitD4);
endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
}
/* check corruption */
if (op1 > opStart2) return ERROR(corruption_detected);
if (op2 > opStart3) return ERROR(corruption_detected);
if (op3 > opStart4) return ERROR(corruption_detected);
/* note : op4 already verified within main loop */
/* finish bitStreams one by one */
HUF_decodeStreamX4(op1, &bitD1, opStart2, dt, dtLog);
HUF_decodeStreamX4(op2, &bitD2, opStart3, dt, dtLog);
HUF_decodeStreamX4(op3, &bitD3, opStart4, dt, dtLog);
HUF_decodeStreamX4(op4, &bitD4, oend, dt, dtLog);
/* check */
{ U32 const endCheck = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4);
if (!endCheck) return ERROR(corruption_detected); }
/* decoded size */
return dstSize;
}
}
#endif
typedef size_t (*HUF_decompress_usingDTable_t)(void *dst, size_t dstSize,
const void *cSrc,
size_t cSrcSize,
const HUF_DTable *DTable);
#if DYNAMIC_BMI2
#define X(fn) \
static size_t fn(void* dst, size_t dstSize, void const* cSrc, \
size_t cSrcSize, HUF_DTable const* DTable, int bmi2) \
{ \
if (bmi2) { \
return fn##_bmi2(dst, dstSize, cSrc, cSrcSize, DTable); \
} \
return fn##_default(dst, dstSize, cSrc, cSrcSize, DTable); \
#define X(fn) \
\
static size_t fn##_default( \
void* dst, size_t dstSize, \
const void* cSrc, size_t cSrcSize, \
const HUF_DTable* DTable) \
{ \
return fn##_body(dst, dstSize, cSrc, cSrcSize, DTable); \
} \
\
static TARGET_ATTRIBUTE("bmi2") size_t fn##_bmi2( \
void* dst, size_t dstSize, \
const void* cSrc, size_t cSrcSize, \
const HUF_DTable* DTable) \
{ \
return fn##_body(dst, dstSize, cSrc, cSrcSize, DTable); \
} \
\
static size_t fn(void* dst, size_t dstSize, void const* cSrc, \
size_t cSrcSize, HUF_DTable const* DTable, int bmi2) \
{ \
if (bmi2) { \
return fn##_bmi2(dst, dstSize, cSrc, cSrcSize, DTable); \
} \
return fn##_default(dst, dstSize, cSrc, cSrcSize, DTable); \
}
#else
#define X(fn) \
static size_t fn(void* dst, size_t dstSize, void const* cSrc, \
size_t cSrcSize, HUF_DTable const* DTable, int bmi2) \
{ \
(void)bmi2; \
return fn##_default(dst, dstSize, cSrc, cSrcSize, DTable); \
#define X(fn) \
static size_t fn(void* dst, size_t dstSize, void const* cSrc, \
size_t cSrcSize, HUF_DTable const* DTable, int bmi2) \
{ \
(void)bmi2; \
return fn##_body(dst, dstSize, cSrc, cSrcSize, DTable); \
}
#endif
X(HUF_decompress1X2_usingDTable_internal)
@@ -193,6 +539,7 @@ X(HUF_decompress4X4_usingDTable_internal)
#undef X
size_t HUF_decompress1X2_usingDTable(
void* dst, size_t dstSize,
const void* cSrc, size_t cSrcSize,
-365
View File
@@ -1,365 +0,0 @@
/*
* Copyright (c) 2018-present, Facebook, Inc.
* All rights reserved.
*
* This source code is licensed under both the BSD-style license (found in the
* LICENSE file in the root directory of this source tree) and the GPLv2 (found
* in the COPYING file in the root directory of this source tree).
* You may select, at your option, one of the above-listed licenses.
*/
#ifndef FUNCTION
# error "FUNCTION(name) must be defined"
#endif
#ifndef TARGET
# error "TARGET must be defined"
#endif
static TARGET BYTE FUNCTION(HUF_decodeSymbolX2)(BIT_DStream_t* Dstream, const HUF_DEltX2* dt, const U32 dtLog)
{
size_t const val = BIT_lookBitsFast(Dstream, dtLog); /* note : dtLog >= 1 */
BYTE const c = dt[val].byte;
BIT_skipBits(Dstream, dt[val].nbBits);
return c;
}
#define HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr) \
*ptr++ = FUNCTION(HUF_decodeSymbolX2)(DStreamPtr, dt, dtLog)
#define HUF_DECODE_SYMBOLX2_1(ptr, DStreamPtr) \
if (MEM_64bits() || (HUF_TABLELOG_MAX<=12)) \
HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr)
#define HUF_DECODE_SYMBOLX2_2(ptr, DStreamPtr) \
if (MEM_64bits()) \
HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr)
HINT_INLINE TARGET size_t FUNCTION(HUF_decodeStreamX2)(BYTE* p, BIT_DStream_t* const bitDPtr, BYTE* const pEnd, const HUF_DEltX2* const dt, const U32 dtLog)
{
BYTE* const pStart = p;
/* up to 4 symbols at a time */
while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) && (p <= pEnd-4)) {
HUF_DECODE_SYMBOLX2_2(p, bitDPtr);
HUF_DECODE_SYMBOLX2_1(p, bitDPtr);
HUF_DECODE_SYMBOLX2_2(p, bitDPtr);
HUF_DECODE_SYMBOLX2_0(p, bitDPtr);
}
/* closer to the end */
while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) && (p < pEnd))
HUF_DECODE_SYMBOLX2_0(p, bitDPtr);
/* no more data to retrieve from bitstream, hence no need to reload */
while (p < pEnd)
HUF_DECODE_SYMBOLX2_0(p, bitDPtr);
return pEnd-pStart;
}
static TARGET size_t FUNCTION(HUF_decompress1X2_usingDTable_internal)(
void* dst, size_t dstSize,
const void* cSrc, size_t cSrcSize,
const HUF_DTable* DTable)
{
BYTE* op = (BYTE*)dst;
BYTE* const oend = op + dstSize;
const void* dtPtr = DTable + 1;
const HUF_DEltX2* const dt = (const HUF_DEltX2*)dtPtr;
BIT_DStream_t bitD;
DTableDesc const dtd = HUF_getDTableDesc(DTable);
U32 const dtLog = dtd.tableLog;
{ size_t const errorCode = BIT_initDStream(&bitD, cSrc, cSrcSize);
if (HUF_isError(errorCode)) return errorCode; }
FUNCTION(HUF_decodeStreamX2)(op, &bitD, oend, dt, dtLog);
/* check */
if (!BIT_endOfDStream(&bitD)) return ERROR(corruption_detected);
return dstSize;
}
static TARGET size_t
FUNCTION(HUF_decompress4X2_usingDTable_internal)(
void* dst, size_t dstSize,
const void* cSrc, size_t cSrcSize,
const HUF_DTable* DTable)
{
/* Check */
if (cSrcSize < 10) return ERROR(corruption_detected); /* strict minimum : jump table + 1 byte per stream */
{ const BYTE* const istart = (const BYTE*) cSrc;
BYTE* const ostart = (BYTE*) dst;
BYTE* const oend = ostart + dstSize;
const void* const dtPtr = DTable + 1;
const HUF_DEltX2* const dt = (const HUF_DEltX2*)dtPtr;
/* Init */
BIT_DStream_t bitD1;
BIT_DStream_t bitD2;
BIT_DStream_t bitD3;
BIT_DStream_t bitD4;
size_t const length1 = MEM_readLE16(istart);
size_t const length2 = MEM_readLE16(istart+2);
size_t const length3 = MEM_readLE16(istart+4);
size_t const length4 = cSrcSize - (length1 + length2 + length3 + 6);
const BYTE* const istart1 = istart + 6; /* jumpTable */
const BYTE* const istart2 = istart1 + length1;
const BYTE* const istart3 = istart2 + length2;
const BYTE* const istart4 = istart3 + length3;
const size_t segmentSize = (dstSize+3) / 4;
BYTE* const opStart2 = ostart + segmentSize;
BYTE* const opStart3 = opStart2 + segmentSize;
BYTE* const opStart4 = opStart3 + segmentSize;
BYTE* op1 = ostart;
BYTE* op2 = opStart2;
BYTE* op3 = opStart3;
BYTE* op4 = opStart4;
U32 endSignal;
DTableDesc const dtd = HUF_getDTableDesc(DTable);
U32 const dtLog = dtd.tableLog;
if (length4 > cSrcSize) return ERROR(corruption_detected); /* overflow */
{ size_t const errorCode = BIT_initDStream(&bitD1, istart1, length1);
if (HUF_isError(errorCode)) return errorCode; }
{ size_t const errorCode = BIT_initDStream(&bitD2, istart2, length2);
if (HUF_isError(errorCode)) return errorCode; }
{ size_t const errorCode = BIT_initDStream(&bitD3, istart3, length3);
if (HUF_isError(errorCode)) return errorCode; }
{ size_t const errorCode = BIT_initDStream(&bitD4, istart4, length4);
if (HUF_isError(errorCode)) return errorCode; }
/* 16-32 symbols per loop (4-8 symbols per stream) */
endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
for ( ; (endSignal==BIT_DStream_unfinished) && (op4<(oend-7)) ; ) {
HUF_DECODE_SYMBOLX2_2(op1, &bitD1);
HUF_DECODE_SYMBOLX2_2(op2, &bitD2);
HUF_DECODE_SYMBOLX2_2(op3, &bitD3);
HUF_DECODE_SYMBOLX2_2(op4, &bitD4);
HUF_DECODE_SYMBOLX2_1(op1, &bitD1);
HUF_DECODE_SYMBOLX2_1(op2, &bitD2);
HUF_DECODE_SYMBOLX2_1(op3, &bitD3);
HUF_DECODE_SYMBOLX2_1(op4, &bitD4);
HUF_DECODE_SYMBOLX2_2(op1, &bitD1);
HUF_DECODE_SYMBOLX2_2(op2, &bitD2);
HUF_DECODE_SYMBOLX2_2(op3, &bitD3);
HUF_DECODE_SYMBOLX2_2(op4, &bitD4);
HUF_DECODE_SYMBOLX2_0(op1, &bitD1);
HUF_DECODE_SYMBOLX2_0(op2, &bitD2);
HUF_DECODE_SYMBOLX2_0(op3, &bitD3);
HUF_DECODE_SYMBOLX2_0(op4, &bitD4);
endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
}
/* check corruption */
if (op1 > opStart2) return ERROR(corruption_detected);
if (op2 > opStart3) return ERROR(corruption_detected);
if (op3 > opStart4) return ERROR(corruption_detected);
/* note : op4 supposed already verified within main loop */
/* finish bitStreams one by one */
FUNCTION(HUF_decodeStreamX2)(op1, &bitD1, opStart2, dt, dtLog);
FUNCTION(HUF_decodeStreamX2)(op2, &bitD2, opStart3, dt, dtLog);
FUNCTION(HUF_decodeStreamX2)(op3, &bitD3, opStart4, dt, dtLog);
FUNCTION(HUF_decodeStreamX2)(op4, &bitD4, oend, dt, dtLog);
/* check */
endSignal = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4);
if (!endSignal) return ERROR(corruption_detected);
/* decoded size */
return dstSize;
}
}
#undef HUF_DECODE_SYMBOLX2_0
#undef HUF_DECODE_SYMBOLX2_1
#undef HUF_DECODE_SYMBOLX2_2
static TARGET U32 FUNCTION(HUF_decodeSymbolX4)(void* op, BIT_DStream_t* DStream, const HUF_DEltX4* dt, const U32 dtLog)
{
size_t const val = BIT_lookBitsFast(DStream, dtLog); /* note : dtLog >= 1 */
memcpy(op, dt+val, 2);
BIT_skipBits(DStream, dt[val].nbBits);
return dt[val].length;
}
static TARGET U32 FUNCTION(HUF_decodeLastSymbolX4)(void* op, BIT_DStream_t* DStream, const HUF_DEltX4* dt, const U32 dtLog)
{
size_t const val = BIT_lookBitsFast(DStream, dtLog); /* note : dtLog >= 1 */
memcpy(op, dt+val, 1);
if (dt[val].length==1) BIT_skipBits(DStream, dt[val].nbBits);
else {
if (DStream->bitsConsumed < (sizeof(DStream->bitContainer)*8)) {
BIT_skipBits(DStream, dt[val].nbBits);
if (DStream->bitsConsumed > (sizeof(DStream->bitContainer)*8))
/* ugly hack; works only because it's the last symbol. Note : can't easily extract nbBits from just this symbol */
DStream->bitsConsumed = (sizeof(DStream->bitContainer)*8);
} }
return 1;
}
#define HUF_DECODE_SYMBOLX4_0(ptr, DStreamPtr) \
ptr += FUNCTION(HUF_decodeSymbolX4)(ptr, DStreamPtr, dt, dtLog)
#define HUF_DECODE_SYMBOLX4_1(ptr, DStreamPtr) \
if (MEM_64bits() || (HUF_TABLELOG_MAX<=12)) \
ptr += FUNCTION(HUF_decodeSymbolX4)(ptr, DStreamPtr, dt, dtLog)
#define HUF_DECODE_SYMBOLX4_2(ptr, DStreamPtr) \
if (MEM_64bits()) \
ptr += FUNCTION(HUF_decodeSymbolX4)(ptr, DStreamPtr, dt, dtLog)
HINT_INLINE TARGET size_t FUNCTION(HUF_decodeStreamX4)(BYTE* p, BIT_DStream_t* bitDPtr, BYTE* const pEnd, const HUF_DEltX4* const dt, const U32 dtLog)
{
BYTE* const pStart = p;
/* up to 8 symbols at a time */
while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p < pEnd-(sizeof(bitDPtr->bitContainer)-1))) {
HUF_DECODE_SYMBOLX4_2(p, bitDPtr);
HUF_DECODE_SYMBOLX4_1(p, bitDPtr);
HUF_DECODE_SYMBOLX4_2(p, bitDPtr);
HUF_DECODE_SYMBOLX4_0(p, bitDPtr);
}
/* closer to end : up to 2 symbols at a time */
while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p <= pEnd-2))
HUF_DECODE_SYMBOLX4_0(p, bitDPtr);
while (p <= pEnd-2)
HUF_DECODE_SYMBOLX4_0(p, bitDPtr); /* no need to reload : reached the end of DStream */
if (p < pEnd)
p += FUNCTION(HUF_decodeLastSymbolX4)(p, bitDPtr, dt, dtLog);
return p-pStart;
}
static TARGET size_t FUNCTION(HUF_decompress1X4_usingDTable_internal)(
void* dst, size_t dstSize,
const void* cSrc, size_t cSrcSize,
const HUF_DTable* DTable)
{
BIT_DStream_t bitD;
/* Init */
{ size_t const errorCode = BIT_initDStream(&bitD, cSrc, cSrcSize);
if (HUF_isError(errorCode)) return errorCode;
}
/* decode */
{ BYTE* const ostart = (BYTE*) dst;
BYTE* const oend = ostart + dstSize;
const void* const dtPtr = DTable+1; /* force compiler to not use strict-aliasing */
const HUF_DEltX4* const dt = (const HUF_DEltX4*)dtPtr;
DTableDesc const dtd = HUF_getDTableDesc(DTable);
FUNCTION(HUF_decodeStreamX4)(ostart, &bitD, oend, dt, dtd.tableLog);
}
/* check */
if (!BIT_endOfDStream(&bitD)) return ERROR(corruption_detected);
/* decoded size */
return dstSize;
}
static TARGET size_t FUNCTION(HUF_decompress4X4_usingDTable_internal)(
void* dst, size_t dstSize,
const void* cSrc, size_t cSrcSize,
const HUF_DTable* DTable)
{
if (cSrcSize < 10) return ERROR(corruption_detected); /* strict minimum : jump table + 1 byte per stream */
{ const BYTE* const istart = (const BYTE*) cSrc;
BYTE* const ostart = (BYTE*) dst;
BYTE* const oend = ostart + dstSize;
const void* const dtPtr = DTable+1;
const HUF_DEltX4* const dt = (const HUF_DEltX4*)dtPtr;
/* Init */
BIT_DStream_t bitD1;
BIT_DStream_t bitD2;
BIT_DStream_t bitD3;
BIT_DStream_t bitD4;
size_t const length1 = MEM_readLE16(istart);
size_t const length2 = MEM_readLE16(istart+2);
size_t const length3 = MEM_readLE16(istart+4);
size_t const length4 = cSrcSize - (length1 + length2 + length3 + 6);
const BYTE* const istart1 = istart + 6; /* jumpTable */
const BYTE* const istart2 = istart1 + length1;
const BYTE* const istart3 = istart2 + length2;
const BYTE* const istart4 = istart3 + length3;
size_t const segmentSize = (dstSize+3) / 4;
BYTE* const opStart2 = ostart + segmentSize;
BYTE* const opStart3 = opStart2 + segmentSize;
BYTE* const opStart4 = opStart3 + segmentSize;
BYTE* op1 = ostart;
BYTE* op2 = opStart2;
BYTE* op3 = opStart3;
BYTE* op4 = opStart4;
U32 endSignal;
DTableDesc const dtd = HUF_getDTableDesc(DTable);
U32 const dtLog = dtd.tableLog;
if (length4 > cSrcSize) return ERROR(corruption_detected); /* overflow */
{ size_t const errorCode = BIT_initDStream(&bitD1, istart1, length1);
if (HUF_isError(errorCode)) return errorCode; }
{ size_t const errorCode = BIT_initDStream(&bitD2, istart2, length2);
if (HUF_isError(errorCode)) return errorCode; }
{ size_t const errorCode = BIT_initDStream(&bitD3, istart3, length3);
if (HUF_isError(errorCode)) return errorCode; }
{ size_t const errorCode = BIT_initDStream(&bitD4, istart4, length4);
if (HUF_isError(errorCode)) return errorCode; }
/* 16-32 symbols per loop (4-8 symbols per stream) */
endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
for ( ; (endSignal==BIT_DStream_unfinished) & (op4<(oend-(sizeof(bitD4.bitContainer)-1))) ; ) {
HUF_DECODE_SYMBOLX4_2(op1, &bitD1);
HUF_DECODE_SYMBOLX4_2(op2, &bitD2);
HUF_DECODE_SYMBOLX4_2(op3, &bitD3);
HUF_DECODE_SYMBOLX4_2(op4, &bitD4);
HUF_DECODE_SYMBOLX4_1(op1, &bitD1);
HUF_DECODE_SYMBOLX4_1(op2, &bitD2);
HUF_DECODE_SYMBOLX4_1(op3, &bitD3);
HUF_DECODE_SYMBOLX4_1(op4, &bitD4);
HUF_DECODE_SYMBOLX4_2(op1, &bitD1);
HUF_DECODE_SYMBOLX4_2(op2, &bitD2);
HUF_DECODE_SYMBOLX4_2(op3, &bitD3);
HUF_DECODE_SYMBOLX4_2(op4, &bitD4);
HUF_DECODE_SYMBOLX4_0(op1, &bitD1);
HUF_DECODE_SYMBOLX4_0(op2, &bitD2);
HUF_DECODE_SYMBOLX4_0(op3, &bitD3);
HUF_DECODE_SYMBOLX4_0(op4, &bitD4);
endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
}
/* check corruption */
if (op1 > opStart2) return ERROR(corruption_detected);
if (op2 > opStart3) return ERROR(corruption_detected);
if (op3 > opStart4) return ERROR(corruption_detected);
/* note : op4 already verified within main loop */
/* finish bitStreams one by one */
FUNCTION(HUF_decodeStreamX4)(op1, &bitD1, opStart2, dt, dtLog);
FUNCTION(HUF_decodeStreamX4)(op2, &bitD2, opStart3, dt, dtLog);
FUNCTION(HUF_decodeStreamX4)(op3, &bitD3, opStart4, dt, dtLog);
FUNCTION(HUF_decodeStreamX4)(op4, &bitD4, oend, dt, dtLog);
/* check */
{ U32 const endCheck = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4);
if (!endCheck) return ERROR(corruption_detected); }
/* decoded size */
return dstSize;
}
}
#undef HUF_DECODE_SYMBOLX4_0
#undef HUF_DECODE_SYMBOLX4_1
#undef HUF_DECODE_SYMBOLX4_2