mirror of
https://github.com/facebook/zstd.git
synced 2024-11-24 03:27:31 +08:00
Allow parameters to be modified from a separate file
This commit is contained in:
Stella Lau
parent
629c300118
commit
e9161637b2
@ -25,16 +25,16 @@ LDFLAGS += -lzstd
|
||||
|
||||
default: all
|
||||
|
||||
all: main-hash32 main-hash64
|
||||
all: main-64 main-integrated
|
||||
|
||||
main-hash64: ldm_common.c ldm_hash64.c main.c
|
||||
main-64: ldm_common.c ldm_hash64.c main.c
|
||||
$(CC) $(CPPFLAGS) $(CFLAGS) $^ $(LDFLAGS) -o $@
|
||||
|
||||
main-hash32: ldm_common.c ldm_hash32.c main.c
|
||||
main-integrated: ldm_common.c ldm_hash32.c main.c
|
||||
$(CC) $(CPPFLAGS) $(CFLAGS) $^ $(LDFLAGS) -o $@
|
||||
|
||||
clean:
|
||||
@rm -f core *.o tmp* result* *.ldm *.ldm.dec \
|
||||
main-hash64 main-hash32
|
||||
main-hash64 main-hash32 main-64 main-integrated
|
||||
@echo Cleaning completed
|
||||
|
||||
|
||||
@ -1,539 +0,0 @@
|
||||
#include <limits.h>
|
||||
#include <stdint.h>
|
||||
#include <stdio.h>
|
||||
#include <stdlib.h>
|
||||
#include <string.h>
|
||||
|
||||
#include "ldm.h"
|
||||
#include "ldm_hashtable.h"
|
||||
|
||||
#define LDM_HASHTABLESIZE (1 << (LDM_MEMORY_USAGE))
|
||||
#define LDM_HASHTABLESIZE_U64 ((LDM_HASHTABLESIZE) >> 3)
|
||||
|
||||
#define LDM_HASH_ENTRY_SIZE_LOG 3
|
||||
|
||||
//#define HASH_ONLY_EVERY_LOG 7
|
||||
#define HASH_ONLY_EVERY_LOG (LDM_WINDOW_SIZE_LOG-((LDM_MEMORY_USAGE)-(LDM_HASH_ENTRY_SIZE_LOG)))
|
||||
|
||||
#define HASH_ONLY_EVERY ((1 << HASH_ONLY_EVERY_LOG) - 1)
|
||||
|
||||
|
||||
#define COMPUTE_STATS
|
||||
#define OUTPUT_CONFIGURATION
|
||||
#define CHECKSUM_CHAR_OFFSET 10
|
||||
|
||||
//#define RUN_CHECKS
|
||||
|
||||
typedef U32 checksum_t;
|
||||
|
||||
struct LDM_compressStats {
|
||||
U32 windowSizeLog, hashTableSizeLog;
|
||||
U32 numMatches;
|
||||
U64 totalMatchLength;
|
||||
U64 totalLiteralLength;
|
||||
U64 totalOffset;
|
||||
|
||||
U32 minOffset, maxOffset;
|
||||
|
||||
U32 offsetHistogram[32];
|
||||
};
|
||||
|
||||
struct LDM_CCtx {
|
||||
U64 isize; /* Input size */
|
||||
U64 maxOSize; /* Maximum output size */
|
||||
|
||||
const BYTE *ibase; /* Base of input */
|
||||
const BYTE *ip; /* Current input position */
|
||||
const BYTE *iend; /* End of input */
|
||||
|
||||
// Maximum input position such that hashing at the position does not exceed
|
||||
// end of input.
|
||||
const BYTE *ihashLimit;
|
||||
|
||||
// Maximum input position such that finding a match of at least the minimum
|
||||
// match length does not exceed end of input.
|
||||
const BYTE *imatchLimit;
|
||||
|
||||
const BYTE *obase; /* Base of output */
|
||||
BYTE *op; /* Output */
|
||||
|
||||
const BYTE *anchor; /* Anchor to start of current (match) block */
|
||||
|
||||
LDM_compressStats stats; /* Compression statistics */
|
||||
|
||||
LDM_hashTable *hashTable;
|
||||
|
||||
const BYTE *lastPosHashed; /* Last position hashed */
|
||||
hash_t lastHash; /* Hash corresponding to lastPosHashed */
|
||||
checksum_t lastSum;
|
||||
|
||||
const BYTE *nextIp; // TODO: this is redundant (ip + step)
|
||||
const BYTE *nextPosHashed;
|
||||
hash_t nextHash; /* Hash corresponding to nextPosHashed */
|
||||
checksum_t nextSum;
|
||||
|
||||
unsigned step; // ip step, should be 1.
|
||||
|
||||
const BYTE *lagIp;
|
||||
hash_t lagHash;
|
||||
checksum_t lagSum;
|
||||
|
||||
// DEBUG
|
||||
const BYTE *DEBUG_setNextHash;
|
||||
};
|
||||
|
||||
// TODO: This can be done more efficiently (but it is not that important as it
|
||||
// is only used for computing stats).
|
||||
static int intLog2(U32 x) {
|
||||
int ret = 0;
|
||||
while (x >>= 1) {
|
||||
ret++;
|
||||
}
|
||||
return ret;
|
||||
}
|
||||
|
||||
void LDM_printCompressStats(const LDM_compressStats *stats) {
|
||||
int i = 0;
|
||||
printf("=====================\n");
|
||||
printf("Compression statistics\n");
|
||||
printf("Window size, hash table size (bytes): 2^%u, 2^%u\n",
|
||||
stats->windowSizeLog, stats->hashTableSizeLog);
|
||||
printf("num matches, total match length, %% matched: %u, %llu, %.3f\n",
|
||||
stats->numMatches,
|
||||
stats->totalMatchLength,
|
||||
100.0 * (double)stats->totalMatchLength /
|
||||
(double)(stats->totalMatchLength + stats->totalLiteralLength));
|
||||
printf("avg match length: %.1f\n", ((double)stats->totalMatchLength) /
|
||||
(double)stats->numMatches);
|
||||
printf("avg literal length, total literalLength: %.1f, %llu\n",
|
||||
((double)stats->totalLiteralLength) / (double)stats->numMatches,
|
||||
stats->totalLiteralLength);
|
||||
printf("avg offset length: %.1f\n",
|
||||
((double)stats->totalOffset) / (double)stats->numMatches);
|
||||
printf("min offset, max offset: %u, %u\n",
|
||||
stats->minOffset, stats->maxOffset);
|
||||
|
||||
printf("\n");
|
||||
printf("offset histogram: offset, num matches, %% of matches\n");
|
||||
|
||||
for (; i <= intLog2(stats->maxOffset); i++) {
|
||||
printf("2^%*d: %10u %6.3f%%\n", 2, i,
|
||||
stats->offsetHistogram[i],
|
||||
100.0 * (double) stats->offsetHistogram[i] /
|
||||
(double) stats->numMatches);
|
||||
}
|
||||
printf("\n");
|
||||
printf("=====================\n");
|
||||
}
|
||||
|
||||
/**
|
||||
* Convert a sum computed from getChecksum to a hash value in the range
|
||||
* of the hash table.
|
||||
*/
|
||||
static hash_t checksumToHash(U32 sum) {
|
||||
return HASH_hashU32(sum);
|
||||
}
|
||||
|
||||
/**
|
||||
* Computes a 32-bit checksum based on rsync's checksum.
|
||||
*
|
||||
* a(k,l) = \sum_{i = k}^l x_i (mod M)
|
||||
* b(k,l) = \sum_{i = k}^l ((l - i + 1) * x_i) (mod M)
|
||||
* checksum(k,l) = a(k,l) + 2^{16} * b(k,l)
|
||||
*/
|
||||
static checksum_t getChecksum(const BYTE *buf, U32 len) {
|
||||
U32 i;
|
||||
checksum_t s1, s2;
|
||||
|
||||
s1 = s2 = 0;
|
||||
for (i = 0; i < (len - 4); i += 4) {
|
||||
s2 += (4 * (s1 + buf[i])) + (3 * buf[i + 1]) +
|
||||
(2 * buf[i + 2]) + (buf[i + 3]) +
|
||||
(10 * CHECKSUM_CHAR_OFFSET);
|
||||
s1 += buf[i] + buf[i + 1] + buf[i + 2] + buf[i + 3] +
|
||||
+ (4 * CHECKSUM_CHAR_OFFSET);
|
||||
|
||||
}
|
||||
for(; i < len; i++) {
|
||||
s1 += buf[i] + CHECKSUM_CHAR_OFFSET;
|
||||
s2 += s1;
|
||||
}
|
||||
return (s1 & 0xffff) + (s2 << 16);
|
||||
}
|
||||
|
||||
/**
|
||||
* Update a checksum computed from getChecksum(data, len).
|
||||
*
|
||||
* The checksum can be updated along its ends as follows:
|
||||
* a(k+1, l+1) = (a(k,l) - x_k + x_{l+1}) (mod M)
|
||||
* b(k+1, l+1) = (b(k,l) - (l-k+1)*x_k + (a(k+1,l+1)) (mod M)
|
||||
*
|
||||
* Thus toRemove should correspond to data[0].
|
||||
*/
|
||||
static checksum_t updateChecksum(checksum_t sum, U32 len,
|
||||
BYTE toRemove, BYTE toAdd) {
|
||||
U32 s1 = (sum & 0xffff) - toRemove + toAdd;
|
||||
U32 s2 = (sum >> 16) - ((toRemove + CHECKSUM_CHAR_OFFSET) * len) + s1;
|
||||
|
||||
return (s1 & 0xffff) + (s2 << 16);
|
||||
}
|
||||
|
||||
/**
|
||||
* Update cctx->nextSum, cctx->nextHash, and cctx->nextPosHashed
|
||||
* based on cctx->lastSum and cctx->lastPosHashed.
|
||||
*
|
||||
* This uses a rolling hash and requires that the last position hashed
|
||||
* corresponds to cctx->nextIp - step.
|
||||
*/
|
||||
static void setNextHash(LDM_CCtx *cctx) {
|
||||
#ifdef RUN_CHECKS
|
||||
U32 check;
|
||||
if ((cctx->nextIp - cctx->ibase != 1) &&
|
||||
(cctx->nextIp - cctx->DEBUG_setNextHash != 1)) {
|
||||
printf("CHECK debug fail: %zu %zu\n", cctx->nextIp - cctx->ibase,
|
||||
cctx->DEBUG_setNextHash - cctx->ibase);
|
||||
}
|
||||
|
||||
cctx->DEBUG_setNextHash = cctx->nextIp;
|
||||
#endif
|
||||
|
||||
cctx->nextSum = updateChecksum(
|
||||
cctx->lastSum, LDM_HASH_LENGTH,
|
||||
cctx->lastPosHashed[0],
|
||||
cctx->lastPosHashed[LDM_HASH_LENGTH]);
|
||||
cctx->nextPosHashed = cctx->nextIp;
|
||||
cctx->nextHash = checksumToHash(cctx->nextSum);
|
||||
|
||||
#if LDM_LAG
|
||||
if (cctx->ip - cctx->ibase > LDM_LAG) {
|
||||
cctx->lagSum = updateChecksum(
|
||||
cctx->lagSum, LDM_HASH_LENGTH,
|
||||
cctx->lagIp[0], cctx->lagIp[LDM_HASH_LENGTH]);
|
||||
cctx->lagIp++;
|
||||
cctx->lagHash = checksumToHash(cctx->lagSum);
|
||||
}
|
||||
#endif
|
||||
|
||||
#ifdef RUN_CHECKS
|
||||
check = getChecksum(cctx->nextIp, LDM_HASH_LENGTH);
|
||||
|
||||
if (check != cctx->nextSum) {
|
||||
printf("CHECK: setNextHash failed %u %u\n", check, cctx->nextSum);
|
||||
}
|
||||
|
||||
if ((cctx->nextIp - cctx->lastPosHashed) != 1) {
|
||||
printf("setNextHash: nextIp != lastPosHashed + 1. %zu %zu %zu\n",
|
||||
cctx->nextIp - cctx->ibase, cctx->lastPosHashed - cctx->ibase,
|
||||
cctx->ip - cctx->ibase);
|
||||
}
|
||||
#endif
|
||||
}
|
||||
|
||||
static void putHashOfCurrentPositionFromHash(
|
||||
LDM_CCtx *cctx, hash_t hash, U32 checksum) {
|
||||
// Hash only every HASH_ONLY_EVERY times, based on cctx->ip.
|
||||
// Note: this works only when cctx->step is 1.
|
||||
if (((cctx->ip - cctx->ibase) & HASH_ONLY_EVERY) == HASH_ONLY_EVERY) {
|
||||
#if LDM_LAG
|
||||
// Off by 1, but whatever
|
||||
if (cctx->lagIp - cctx->ibase > 0) {
|
||||
const LDM_hashEntry entry = { cctx->lagIp - cctx->ibase, cctx->lagSum };
|
||||
HASH_insert(cctx->hashTable, cctx->lagHash, entry);
|
||||
} else {
|
||||
const LDM_hashEntry entry = { cctx->ip - cctx->ibase, checksum };
|
||||
HASH_insert(cctx->hashTable, hash, entry);
|
||||
}
|
||||
#else
|
||||
const LDM_hashEntry entry = { cctx->ip - cctx->ibase, checksum };
|
||||
HASH_insert(cctx->hashTable, hash, entry);
|
||||
#endif
|
||||
}
|
||||
|
||||
cctx->lastPosHashed = cctx->ip;
|
||||
cctx->lastHash = hash;
|
||||
cctx->lastSum = checksum;
|
||||
}
|
||||
|
||||
/**
|
||||
* Copy over the cctx->lastHash, cctx->lastSum, and cctx->lastPosHashed
|
||||
* fields from the "next" fields.
|
||||
*
|
||||
* This requires that cctx->ip == cctx->nextPosHashed.
|
||||
*/
|
||||
static void LDM_updateLastHashFromNextHash(LDM_CCtx *cctx) {
|
||||
#ifdef RUN_CHECKS
|
||||
if (cctx->ip != cctx->nextPosHashed) {
|
||||
printf("CHECK failed: updateLastHashFromNextHash %zu\n",
|
||||
cctx->ip - cctx->ibase);
|
||||
}
|
||||
#endif
|
||||
putHashOfCurrentPositionFromHash(cctx, cctx->nextHash, cctx->nextSum);
|
||||
}
|
||||
|
||||
/**
|
||||
* Insert hash of the current position into the hash table.
|
||||
*/
|
||||
static void LDM_putHashOfCurrentPosition(LDM_CCtx *cctx) {
|
||||
checksum_t sum = getChecksum(cctx->ip, LDM_HASH_LENGTH);
|
||||
hash_t hash = checksumToHash(sum);
|
||||
|
||||
#ifdef RUN_CHECKS
|
||||
if (cctx->nextPosHashed != cctx->ip && (cctx->ip != cctx->ibase)) {
|
||||
printf("CHECK failed: putHashOfCurrentPosition %zu\n",
|
||||
cctx->ip - cctx->ibase);
|
||||
}
|
||||
#endif
|
||||
|
||||
putHashOfCurrentPositionFromHash(cctx, hash, sum);
|
||||
}
|
||||
|
||||
void LDM_initializeCCtx(LDM_CCtx *cctx,
|
||||
const void *src, size_t srcSize,
|
||||
void *dst, size_t maxDstSize) {
|
||||
cctx->isize = srcSize;
|
||||
cctx->maxOSize = maxDstSize;
|
||||
|
||||
cctx->ibase = (const BYTE *)src;
|
||||
cctx->ip = cctx->ibase;
|
||||
cctx->iend = cctx->ibase + srcSize;
|
||||
|
||||
cctx->ihashLimit = cctx->iend - LDM_HASH_LENGTH;
|
||||
cctx->imatchLimit = cctx->iend - LDM_MIN_MATCH_LENGTH;
|
||||
|
||||
cctx->obase = (BYTE *)dst;
|
||||
cctx->op = (BYTE *)dst;
|
||||
|
||||
cctx->anchor = cctx->ibase;
|
||||
|
||||
memset(&(cctx->stats), 0, sizeof(cctx->stats));
|
||||
cctx->hashTable = HASH_createTable(LDM_HASHTABLESIZE_U64, cctx->ibase,
|
||||
LDM_MIN_MATCH_LENGTH, LDM_WINDOW_SIZE);
|
||||
|
||||
cctx->stats.minOffset = UINT_MAX;
|
||||
cctx->stats.windowSizeLog = LDM_WINDOW_SIZE_LOG;
|
||||
cctx->stats.hashTableSizeLog = LDM_MEMORY_USAGE;
|
||||
|
||||
|
||||
cctx->lastPosHashed = NULL;
|
||||
|
||||
cctx->step = 1; // Fixed to be 1 for now. Changing may break things.
|
||||
cctx->nextIp = cctx->ip + cctx->step;
|
||||
cctx->nextPosHashed = 0;
|
||||
|
||||
cctx->DEBUG_setNextHash = 0;
|
||||
}
|
||||
|
||||
void LDM_destroyCCtx(LDM_CCtx *cctx) {
|
||||
HASH_destroyTable(cctx->hashTable);
|
||||
}
|
||||
|
||||
/**
|
||||
* Finds the "best" match.
|
||||
*
|
||||
* Returns 0 if successful and 1 otherwise (i.e. no match can be found
|
||||
* in the remaining input that is long enough).
|
||||
*
|
||||
* matchLength contains the forward length of the match.
|
||||
*/
|
||||
static int LDM_findBestMatch(LDM_CCtx *cctx, const BYTE **match,
|
||||
U64 *matchLength, U64 *backwardMatchLength) {
|
||||
|
||||
LDM_hashEntry *entry = NULL;
|
||||
cctx->nextIp = cctx->ip + cctx->step;
|
||||
|
||||
while (entry == NULL) {
|
||||
hash_t h;
|
||||
checksum_t sum;
|
||||
setNextHash(cctx);
|
||||
h = cctx->nextHash;
|
||||
sum = cctx->nextSum;
|
||||
cctx->ip = cctx->nextIp;
|
||||
cctx->nextIp += cctx->step;
|
||||
|
||||
if (cctx->ip > cctx->imatchLimit) {
|
||||
return 1;
|
||||
}
|
||||
|
||||
entry = HASH_getBestEntry(cctx->hashTable, h, sum,
|
||||
cctx->ip, cctx->iend,
|
||||
cctx->anchor,
|
||||
matchLength, backwardMatchLength);
|
||||
|
||||
if (entry != NULL) {
|
||||
*match = entry->offset + cctx->ibase;
|
||||
}
|
||||
putHashOfCurrentPositionFromHash(cctx, h, sum);
|
||||
}
|
||||
setNextHash(cctx);
|
||||
return 0;
|
||||
}
|
||||
|
||||
void LDM_encodeLiteralLengthAndLiterals(
|
||||
LDM_CCtx *cctx, BYTE *pToken, const U64 literalLength) {
|
||||
/* Encode the literal length. */
|
||||
if (literalLength >= RUN_MASK) {
|
||||
int len = (int)literalLength - RUN_MASK;
|
||||
*pToken = (RUN_MASK << ML_BITS);
|
||||
for (; len >= 255; len -= 255) {
|
||||
*(cctx->op)++ = 255;
|
||||
}
|
||||
*(cctx->op)++ = (BYTE)len;
|
||||
} else {
|
||||
*pToken = (BYTE)(literalLength << ML_BITS);
|
||||
}
|
||||
|
||||
/* Encode the literals. */
|
||||
memcpy(cctx->op, cctx->anchor, literalLength);
|
||||
cctx->op += literalLength;
|
||||
}
|
||||
|
||||
void LDM_outputBlock(LDM_CCtx *cctx,
|
||||
const U64 literalLength,
|
||||
const U32 offset,
|
||||
const U64 matchLength) {
|
||||
BYTE *pToken = cctx->op++;
|
||||
|
||||
/* Encode the literal length and literals. */
|
||||
LDM_encodeLiteralLengthAndLiterals(cctx, pToken, literalLength);
|
||||
|
||||
/* Encode the offset. */
|
||||
MEM_write32(cctx->op, offset);
|
||||
cctx->op += LDM_OFFSET_SIZE;
|
||||
|
||||
/* Encode the match length. */
|
||||
if (matchLength >= ML_MASK) {
|
||||
U64 matchLengthRemaining = matchLength;
|
||||
*pToken += ML_MASK;
|
||||
matchLengthRemaining -= ML_MASK;
|
||||
MEM_write32(cctx->op, 0xFFFFFFFF);
|
||||
while (matchLengthRemaining >= 4*0xFF) {
|
||||
cctx->op += 4;
|
||||
MEM_write32(cctx->op, 0xffffffff);
|
||||
matchLengthRemaining -= 4*0xFF;
|
||||
}
|
||||
cctx->op += matchLengthRemaining / 255;
|
||||
*(cctx->op)++ = (BYTE)(matchLengthRemaining % 255);
|
||||
} else {
|
||||
*pToken += (BYTE)(matchLength);
|
||||
}
|
||||
}
|
||||
|
||||
// TODO: maxDstSize is unused. This function may seg fault when writing
|
||||
// beyond the size of dst, as it does not check maxDstSize. Writing to
|
||||
// a buffer and performing checks is a possible solution.
|
||||
//
|
||||
// This is based upon lz4.
|
||||
size_t LDM_compress(const void *src, size_t srcSize,
|
||||
void *dst, size_t maxDstSize) {
|
||||
LDM_CCtx cctx;
|
||||
const BYTE *match = NULL;
|
||||
U64 forwardMatchLength = 0;
|
||||
U64 backwardsMatchLength = 0;
|
||||
|
||||
LDM_initializeCCtx(&cctx, src, srcSize, dst, maxDstSize);
|
||||
LDM_outputConfiguration();
|
||||
|
||||
/* Hash the first position and put it into the hash table. */
|
||||
LDM_putHashOfCurrentPosition(&cctx);
|
||||
|
||||
#if LDM_LAG
|
||||
cctx.lagIp = cctx.ip;
|
||||
cctx.lagHash = cctx.lastHash;
|
||||
cctx.lagSum = cctx.lastSum;
|
||||
#endif
|
||||
/**
|
||||
* Find a match.
|
||||
* If no more matches can be found (i.e. the length of the remaining input
|
||||
* is less than the minimum match length), then stop searching for matches
|
||||
* and encode the final literals.
|
||||
*/
|
||||
while (LDM_findBestMatch(&cctx, &match, &forwardMatchLength,
|
||||
&backwardsMatchLength) == 0) {
|
||||
#ifdef COMPUTE_STATS
|
||||
cctx.stats.numMatches++;
|
||||
#endif
|
||||
|
||||
cctx.ip -= backwardsMatchLength;
|
||||
match -= backwardsMatchLength;
|
||||
|
||||
/**
|
||||
* Write current block (literals, literal length, match offset, match
|
||||
* length) and update pointers and hashes.
|
||||
*/
|
||||
{
|
||||
const U32 literalLength = cctx.ip - cctx.anchor;
|
||||
const U32 offset = cctx.ip - match;
|
||||
const U32 matchLength = forwardMatchLength +
|
||||
backwardsMatchLength -
|
||||
LDM_MIN_MATCH_LENGTH;
|
||||
|
||||
LDM_outputBlock(&cctx, literalLength, offset, matchLength);
|
||||
|
||||
#ifdef COMPUTE_STATS
|
||||
cctx.stats.totalLiteralLength += literalLength;
|
||||
cctx.stats.totalOffset += offset;
|
||||
cctx.stats.totalMatchLength += matchLength + LDM_MIN_MATCH_LENGTH;
|
||||
cctx.stats.minOffset =
|
||||
offset < cctx.stats.minOffset ? offset : cctx.stats.minOffset;
|
||||
cctx.stats.maxOffset =
|
||||
offset > cctx.stats.maxOffset ? offset : cctx.stats.maxOffset;
|
||||
cctx.stats.offsetHistogram[(U32)intLog2(offset)]++;
|
||||
#endif
|
||||
|
||||
// Move ip to end of block, inserting hashes at each position.
|
||||
cctx.nextIp = cctx.ip + cctx.step;
|
||||
while (cctx.ip < cctx.anchor + LDM_MIN_MATCH_LENGTH +
|
||||
matchLength + literalLength) {
|
||||
if (cctx.ip > cctx.lastPosHashed) {
|
||||
// TODO: Simplify.
|
||||
LDM_updateLastHashFromNextHash(&cctx);
|
||||
setNextHash(&cctx);
|
||||
}
|
||||
cctx.ip++;
|
||||
cctx.nextIp++;
|
||||
}
|
||||
}
|
||||
|
||||
// Set start of next block to current input pointer.
|
||||
cctx.anchor = cctx.ip;
|
||||
LDM_updateLastHashFromNextHash(&cctx);
|
||||
}
|
||||
|
||||
/* Encode the last literals (no more matches). */
|
||||
{
|
||||
const U32 lastRun = cctx.iend - cctx.anchor;
|
||||
BYTE *pToken = cctx.op++;
|
||||
LDM_encodeLiteralLengthAndLiterals(&cctx, pToken, lastRun);
|
||||
}
|
||||
|
||||
#ifdef COMPUTE_STATS
|
||||
LDM_printCompressStats(&cctx.stats);
|
||||
HASH_outputTableOccupancy(cctx.hashTable);
|
||||
#endif
|
||||
|
||||
{
|
||||
const size_t ret = cctx.op - cctx.obase;
|
||||
LDM_destroyCCtx(&cctx);
|
||||
return ret;
|
||||
}
|
||||
}
|
||||
|
||||
void LDM_outputConfiguration(void) {
|
||||
printf("=====================\n");
|
||||
printf("Configuration\n");
|
||||
printf("LDM_WINDOW_SIZE_LOG: %d\n", LDM_WINDOW_SIZE_LOG);
|
||||
printf("LDM_MIN_MATCH_LENGTH, LDM_HASH_LENGTH: %d, %d\n",
|
||||
LDM_MIN_MATCH_LENGTH, LDM_HASH_LENGTH);
|
||||
printf("LDM_MEMORY_USAGE: %d\n", LDM_MEMORY_USAGE);
|
||||
printf("HASH_ONLY_EVERY_LOG: %d\n", HASH_ONLY_EVERY_LOG);
|
||||
printf("HASH_BUCKET_SIZE_LOG: %d\n", HASH_BUCKET_SIZE_LOG);
|
||||
printf("LDM_LAG %d\n", LDM_LAG);
|
||||
printf("=====================\n");
|
||||
}
|
||||
|
||||
|
||||
|
||||
void LDM_test(const BYTE *src) {
|
||||
(void)src;
|
||||
}
|
||||
|
||||
@ -2,6 +2,7 @@
|
||||
#define LDM_H
|
||||
|
||||
#include "mem.h" // from /lib/common/mem.h
|
||||
#include "ldm_params.h"
|
||||
|
||||
// The number of bytes storing the compressed and decompressed size
|
||||
// in the header.
|
||||
@ -18,35 +19,38 @@
|
||||
#define LDM_OFFSET_SIZE 4
|
||||
|
||||
// =============================================================================
|
||||
// User parameters.
|
||||
// Modify parameters in ldm_params.h if "ldm_params.h" is included.
|
||||
// =============================================================================
|
||||
|
||||
#ifndef LDM_PARAMS_H
|
||||
// Defines the size of the hash table.
|
||||
// Note that this is not the number of buckets.
|
||||
// Currently this should be less than WINDOW_SIZE_LOG + 4?
|
||||
#define LDM_MEMORY_USAGE 25
|
||||
#define LDM_MEMORY_USAGE 25
|
||||
|
||||
// The number of entries in a hash bucket.
|
||||
#define HASH_BUCKET_SIZE_LOG 3 // The maximum is 4 for now.
|
||||
#define HASH_BUCKET_SIZE_LOG 3 // The maximum is 4 for now.
|
||||
|
||||
// Defines the lag in inserting elements into the hash table.
|
||||
#define LDM_LAG 0
|
||||
#define LDM_LAG 0
|
||||
|
||||
// The maximum window size.
|
||||
#define LDM_WINDOW_SIZE_LOG 28 // Max value is 30
|
||||
#define LDM_WINDOW_SIZE (1 << (LDM_WINDOW_SIZE_LOG))
|
||||
#define LDM_WINDOW_SIZE_LOG 28 // Max value is 30
|
||||
|
||||
//These should be multiples of four (and perhaps set to the same value?).
|
||||
#define LDM_MIN_MATCH_LENGTH 64
|
||||
#define LDM_HASH_LENGTH 64
|
||||
#define LDM_MIN_MATCH_LENGTH 64
|
||||
|
||||
// Experimental.
|
||||
//#define TMP_EVICTION // Experiment with eviction policies.
|
||||
#define INSERT_BY_TAG // Insertion policy based on hash.
|
||||
#define INSERT_BY_TAG 1 // Insertion policy based on hash.
|
||||
|
||||
#define USE_CHECKSUM 1
|
||||
#define USE_CHECKSUM 1
|
||||
#endif
|
||||
|
||||
// =============================================================================
|
||||
#define COMPUTE_STATS
|
||||
#define OUTPUT_CONFIGURATION
|
||||
|
||||
#define LDM_WINDOW_SIZE (1 << (LDM_WINDOW_SIZE_LOG))
|
||||
#define LDM_HASH_LENGTH LDM_MIN_MATCH_LENGTH
|
||||
|
||||
typedef struct LDM_compressStats LDM_compressStats;
|
||||
typedef struct LDM_CCtx LDM_CCtx;
|
||||
@ -164,6 +168,4 @@ void LDM_writeHeader(void *memPtr, U64 compressedSize,
|
||||
*/
|
||||
void LDM_outputConfiguration(void);
|
||||
|
||||
void LDM_test(const BYTE *src);
|
||||
|
||||
#endif /* LDM_H */
|
||||
|
||||
@ -70,7 +70,8 @@ size_t LDM_decompress(const void *src, size_t compressedSize,
|
||||
dctx.ip += length;
|
||||
dctx.op = cpy;
|
||||
|
||||
//TODO : dynamic offset size
|
||||
//TODO: dynamic offset size?
|
||||
/* Encode the offset. */
|
||||
offset = MEM_read32(dctx.ip);
|
||||
dctx.ip += LDM_OFFSET_SIZE;
|
||||
match = dctx.op - offset;
|
||||
@ -89,7 +90,7 @@ size_t LDM_decompress(const void *src, size_t compressedSize,
|
||||
/* Copy match. */
|
||||
cpy = dctx.op + length;
|
||||
|
||||
// Inefficient for now.
|
||||
// TODO: this can be made more efficient.
|
||||
while (match < cpy - offset && dctx.op < dctx.oend) {
|
||||
*(dctx.op)++ = *match++;
|
||||
}
|
||||
|
||||
@ -21,9 +21,7 @@
|
||||
#define HASH_BUCKET_SIZE (1 << (HASH_BUCKET_SIZE_LOG))
|
||||
#define LDM_HASHLOG ((LDM_MEMORY_USAGE)-(LDM_HASH_ENTRY_SIZE_LOG)-(HASH_BUCKET_SIZE_LOG))
|
||||
|
||||
#define COMPUTE_STATS
|
||||
#define OUTPUT_CONFIGURATION
|
||||
#define CHECKSUM_CHAR_OFFSET 1
|
||||
#define CHECKSUM_CHAR_OFFSET 10
|
||||
|
||||
// Take first match only.
|
||||
//#define ZSTD_SKIP
|
||||
@ -779,8 +777,6 @@ size_t LDM_compress(const void *src, size_t srcSize,
|
||||
LDM_updateLastHashFromNextHash(&cctx);
|
||||
}
|
||||
|
||||
// HASH_outputTableOffsetHistogram(&cctx);
|
||||
|
||||
/* Encode the last literals (no more matches). */
|
||||
{
|
||||
const U32 lastRun = cctx.iend - cctx.anchor;
|
||||
@ -815,9 +811,3 @@ void LDM_outputConfiguration(void) {
|
||||
|
||||
|
||||
|
||||
// TODO: implement and test hash function
|
||||
void LDM_test(const BYTE *src) {
|
||||
(void)src;
|
||||
}
|
||||
|
||||
|
||||
|
||||
@ -24,8 +24,6 @@
|
||||
#define HASH_BUCKET_SIZE (1 << (HASH_BUCKET_SIZE_LOG))
|
||||
#define LDM_HASHLOG ((LDM_MEMORY_USAGE)-(LDM_HASH_ENTRY_SIZE_LOG)-(HASH_BUCKET_SIZE_LOG))
|
||||
|
||||
#define COMPUTE_STATS
|
||||
#define OUTPUT_CONFIGURATION
|
||||
#define HASH_CHAR_OFFSET 10
|
||||
|
||||
// Take first match only.
|
||||
@ -63,11 +61,6 @@ struct LDM_compressStats {
|
||||
|
||||
U64 TMP_hashCount[1 << HASH_ONLY_EVERY_LOG];
|
||||
U64 TMP_totalHashCount;
|
||||
|
||||
U64 TMP_totalInWindow;
|
||||
U64 TMP_totalInserts;
|
||||
|
||||
U64 TMP_matchCount;
|
||||
};
|
||||
|
||||
typedef struct LDM_hashTable LDM_hashTable;
|
||||
@ -328,91 +321,12 @@ LDM_hashEntry *HASH_getBestEntry(const LDM_CCtx *cctx,
|
||||
return NULL;
|
||||
}
|
||||
|
||||
#ifdef TMP_EVICTION
|
||||
void HASH_insert(LDM_hashTable *table,
|
||||
const hash_t hash, const LDM_hashEntry entry,
|
||||
LDM_CCtx *cctx) {
|
||||
// Overwrite based on part of checksum.
|
||||
/*
|
||||
LDM_hashEntry *toOverwrite =
|
||||
getBucket(table, hash) + table->bucketOffsets[hash];
|
||||
const BYTE *pMatch = toOverwrite->offset + cctx->ibase;
|
||||
if (toOverwrite->offset != 0 &&
|
||||
cctx->ip - pMatch <= LDM_WINDOW_SIZE) {
|
||||
cctx->stats.TMP_totalInWindow++;
|
||||
}
|
||||
|
||||
cctx->stats.TMP_totalInserts++;
|
||||
*(toOverwrite) = entry;
|
||||
*/
|
||||
|
||||
/*
|
||||
int i;
|
||||
LDM_hashEntry *bucket = getBucket(table, hash);
|
||||
for (i = 0; i < HASH_BUCKET_SIZE; i++) {
|
||||
if (bucket[i].checksum == entry.checksum) {
|
||||
bucket[i] = entry;
|
||||
cctx->stats.TMP_matchCount++;
|
||||
return;
|
||||
}
|
||||
}
|
||||
*/
|
||||
|
||||
// Find entry beyond window size, replace. Else, random.
|
||||
int i;
|
||||
LDM_hashEntry *bucket = getBucket(table, hash);
|
||||
for (i = 0; i < HASH_BUCKET_SIZE; i++) {
|
||||
if (cctx->ip - cctx->ibase - bucket[i].offset > LDM_WINDOW_SIZE) {
|
||||
bucket[i] = entry;
|
||||
return;
|
||||
}
|
||||
}
|
||||
|
||||
i = rand() & (HASH_BUCKET_SIZE - 1);
|
||||
*(bucket + i) = entry;
|
||||
|
||||
|
||||
/**
|
||||
* Sliding buffer style pointer
|
||||
* Keep old entry as temporary. If the old entry is outside the window,
|
||||
* overwrite and we are done.
|
||||
*
|
||||
* Backwards (insert at x):
|
||||
* x, a, b b, c c c c, d d d d d d d d
|
||||
* x, d d d d d d d d, c c c c, b b, a
|
||||
*
|
||||
* Else, find something to evict.
|
||||
* If old entry has more ones, it takes
|
||||
* the next spot. <-- reversed order?
|
||||
*
|
||||
* If window size > LDM_WINDOW_SIZE,
|
||||
* overwrite,
|
||||
*
|
||||
* Insert forwards. If > tag, keep. Else evict.
|
||||
*
|
||||
*/
|
||||
|
||||
|
||||
/*
|
||||
*(getBucket(table, hash) + table->bucketOffsets[hash]) = entry;
|
||||
table->bucketOffsets[hash]++;
|
||||
table->bucketOffsets[hash] &= HASH_BUCKET_SIZE - 1;
|
||||
*/
|
||||
|
||||
// U16 mask = entry.checksum & (HASH_BUCKET_SIZE - 1);
|
||||
// *(getBucket(table, hash) + mask) = entry;
|
||||
}
|
||||
|
||||
#else
|
||||
|
||||
void HASH_insert(LDM_hashTable *table,
|
||||
const hash_t hash, const LDM_hashEntry entry) {
|
||||
*(getBucket(table, hash) + table->bucketOffsets[hash]) = entry;
|
||||
table->bucketOffsets[hash]++;
|
||||
table->bucketOffsets[hash] &= HASH_BUCKET_SIZE - 1;
|
||||
}
|
||||
#endif // TMP_EVICTION
|
||||
|
||||
|
||||
U32 HASH_getSize(const LDM_hashTable *table) {
|
||||
return table->numBuckets;
|
||||
@ -489,7 +403,7 @@ void LDM_printCompressStats(const LDM_compressStats *stats) {
|
||||
(double) stats->numMatches);
|
||||
}
|
||||
printf("\n");
|
||||
#ifdef INSERT_BY_TAG
|
||||
#if INSERT_BY_TAG
|
||||
/*
|
||||
printf("Lower bit distribution\n");
|
||||
for (i = 0; i < (1 << HASH_ONLY_EVERY_LOG); i++) {
|
||||
@ -500,13 +414,6 @@ void LDM_printCompressStats(const LDM_compressStats *stats) {
|
||||
*/
|
||||
#endif
|
||||
|
||||
#ifdef TMP_EVICTION
|
||||
printf("Evicted something in window: %llu %6.3f\n",
|
||||
stats->TMP_totalInWindow,
|
||||
100.0 * (double)stats->TMP_totalInWindow /
|
||||
(double)stats->TMP_totalInserts);
|
||||
printf("Match count: %llu\n", stats->TMP_matchCount);
|
||||
#endif
|
||||
printf("=====================\n");
|
||||
}
|
||||
|
||||
@ -524,7 +431,7 @@ static U32 getChecksum(U64 hash) {
|
||||
return (hash >> (64 - 32 - LDM_HASHLOG)) & 0xFFFFFFFF;
|
||||
}
|
||||
|
||||
#ifdef INSERT_BY_TAG
|
||||
#if INSERT_BY_TAG
|
||||
static U32 lowerBitsFromHfHash(U64 hash) {
|
||||
// The number of bits used so far is LDM_HASHLOG + 32.
|
||||
// So there are 32 - LDM_HASHLOG bits left.
|
||||
@ -611,7 +518,7 @@ static void setNextHash(LDM_CCtx *cctx) {
|
||||
cctx->lastPosHashed[LDM_HASH_LENGTH]);
|
||||
cctx->nextPosHashed = cctx->nextIp;
|
||||
|
||||
#ifdef INSERT_BY_TAG
|
||||
#if INSERT_BY_TAG
|
||||
{
|
||||
U32 hashEveryMask = lowerBitsFromHfHash(cctx->nextHash);
|
||||
cctx->stats.TMP_totalHashCount++;
|
||||
@ -648,7 +555,7 @@ static void putHashOfCurrentPositionFromHash(LDM_CCtx *cctx, U64 hash) {
|
||||
// Note: this works only when cctx->step is 1.
|
||||
#if LDM_LAG
|
||||
if (cctx -> lagIp - cctx->ibase > 0) {
|
||||
#ifdef INSERT_BY_TAG
|
||||
#if INSERT_BY_TAG
|
||||
U32 hashEveryMask = lowerBitsFromHfHash(cctx->lagHash);
|
||||
if (hashEveryMask == HASH_ONLY_EVERY) {
|
||||
#else
|
||||
@ -663,14 +570,11 @@ static void putHashOfCurrentPositionFromHash(LDM_CCtx *cctx, U64 hash) {
|
||||
const LDM_hashEntry entry = { cctx->lagIp - cctx->ibase };
|
||||
# endif
|
||||
|
||||
# ifdef TMP_EVICTION
|
||||
HASH_insert(cctx->hashTable, smallHash, entry, cctx);
|
||||
# else
|
||||
HASH_insert(cctx->hashTable, smallHash, entry);
|
||||
# endif
|
||||
}
|
||||
} else {
|
||||
#ifdef INSERT_BY_TAG
|
||||
#endif // LDM_LAG
|
||||
#if INSERT_BY_TAG
|
||||
U32 hashEveryMask = lowerBitsFromHfHash(hash);
|
||||
if (hashEveryMask == HASH_ONLY_EVERY) {
|
||||
#else
|
||||
@ -684,33 +588,9 @@ static void putHashOfCurrentPositionFromHash(LDM_CCtx *cctx, U64 hash) {
|
||||
#else
|
||||
const LDM_hashEntry entry = { cctx->ip - cctx->ibase };
|
||||
#endif
|
||||
|
||||
#ifdef TMP_EVICTION
|
||||
HASH_insert(cctx->hashTable, smallHash, entry, cctx);
|
||||
#else
|
||||
HASH_insert(cctx->hashTable, smallHash, entry);
|
||||
#endif
|
||||
}
|
||||
}
|
||||
#else
|
||||
#ifdef INSERT_BY_TAG
|
||||
U32 hashEveryMask = lowerBitsFromHfHash(hash);
|
||||
if (hashEveryMask == HASH_ONLY_EVERY) {
|
||||
#else
|
||||
if (((cctx->ip - cctx->ibase) & HASH_ONLY_EVERY) == HASH_ONLY_EVERY) {
|
||||
#endif
|
||||
U32 smallHash = getSmallHash(hash);
|
||||
#if USE_CHECKSUM
|
||||
U32 checksum = getChecksum(hash);
|
||||
const LDM_hashEntry entry = { cctx->ip - cctx->ibase, checksum };
|
||||
#else
|
||||
const LDM_hashEntry entry = { cctx->ip - cctx->ibase };
|
||||
#endif
|
||||
#ifdef TMP_EVICTION
|
||||
HASH_insert(cctx->hashTable, smallHash, entry, cctx);
|
||||
#else
|
||||
HASH_insert(cctx->hashTable, smallHash, entry);
|
||||
#endif
|
||||
#if LDM_LAG
|
||||
}
|
||||
#endif
|
||||
|
||||
@ -812,7 +692,7 @@ static int LDM_findBestMatch(LDM_CCtx *cctx, const BYTE **match,
|
||||
U64 hash;
|
||||
hash_t smallHash;
|
||||
U32 checksum;
|
||||
#ifdef INSERT_BY_TAG
|
||||
#if INSERT_BY_TAG
|
||||
U32 hashEveryMask;
|
||||
#endif
|
||||
setNextHash(cctx);
|
||||
@ -820,7 +700,7 @@ static int LDM_findBestMatch(LDM_CCtx *cctx, const BYTE **match,
|
||||
hash = cctx->nextHash;
|
||||
smallHash = getSmallHash(hash);
|
||||
checksum = getChecksum(hash);
|
||||
#ifdef INSERT_BY_TAG
|
||||
#if INSERT_BY_TAG
|
||||
hashEveryMask = lowerBitsFromHfHash(hash);
|
||||
#endif
|
||||
|
||||
@ -830,7 +710,7 @@ static int LDM_findBestMatch(LDM_CCtx *cctx, const BYTE **match,
|
||||
if (cctx->ip > cctx->imatchLimit) {
|
||||
return 1;
|
||||
}
|
||||
#ifdef INSERT_BY_TAG
|
||||
#if INSERT_BY_TAG
|
||||
if (hashEveryMask == HASH_ONLY_EVERY) {
|
||||
|
||||
entry = HASH_getBestEntry(cctx, smallHash, checksum,
|
||||
@ -923,10 +803,8 @@ size_t LDM_compress(const void *src, size_t srcSize,
|
||||
/* Hash the first position and put it into the hash table. */
|
||||
LDM_putHashOfCurrentPosition(&cctx);
|
||||
|
||||
#if LDM_LAG
|
||||
cctx.lagIp = cctx.ip;
|
||||
cctx.lagHash = cctx.lastHash;
|
||||
#endif
|
||||
/**
|
||||
* Find a match.
|
||||
* If no more matches can be found (i.e. the length of the remaining input
|
||||
@ -1018,28 +896,8 @@ void LDM_outputConfiguration(void) {
|
||||
printf("HASH_BUCKET_SIZE_LOG: %d\n", HASH_BUCKET_SIZE_LOG);
|
||||
printf("LDM_LAG: %d\n", LDM_LAG);
|
||||
printf("USE_CHECKSUM: %d\n", USE_CHECKSUM);
|
||||
#ifdef INSERT_BY_TAG
|
||||
printf("INSERT_BY_TAG: %d\n", 1);
|
||||
#else
|
||||
printf("INSERT_BY_TAG: %d\n", 0);
|
||||
#endif
|
||||
printf("INSERT_BY_TAG: %d\n", INSERT_BY_TAG);
|
||||
printf("HASH_CHAR_OFFSET: %d\n", HASH_CHAR_OFFSET);
|
||||
printf("=====================\n");
|
||||
}
|
||||
|
||||
// TODO: implement and test hash function
|
||||
void LDM_test(const BYTE *src) {
|
||||
const U32 diff = 100;
|
||||
const BYTE *pCur = src + diff;
|
||||
U64 hash = getHash(pCur, LDM_HASH_LENGTH);
|
||||
|
||||
for (; pCur < src + diff + 60; ++pCur) {
|
||||
U64 nextHash = getHash(pCur + 1, LDM_HASH_LENGTH);
|
||||
U64 updatedHash = updateHash(hash, LDM_HASH_LENGTH,
|
||||
pCur[0], pCur[LDM_HASH_LENGTH]);
|
||||
hash = nextHash;
|
||||
printf("%llu %llu\n", nextHash, updatedHash);
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
|
||||
10
contrib/long_distance_matching/ldm_params.h
Normal file
10
contrib/long_distance_matching/ldm_params.h
Normal file
@ -0,0 +1,10 @@
|
||||
#ifndef LDM_PARAMS_H
|
||||
#define LDM_PARAMS_H
|
||||
#define LDM_MEMORY_USAGE 23
|
||||
#define HASH_BUCKET_SIZE_LOG 3
|
||||
#define LDM_LAG 0
|
||||
#define LDM_WINDOW_SIZE_LOG 28
|
||||
#define LDM_MIN_MATCH_LENGTH 64
|
||||
#define INSERT_BY_TAG 1
|
||||
#define USE_CHECKSUM 1
|
||||
#endif
|
||||
@ -12,7 +12,7 @@
|
||||
#include "ldm.h"
|
||||
#include "zstd.h"
|
||||
|
||||
//#define DECOMPRESS_AND_VERIFY
|
||||
#define DECOMPRESS_AND_VERIFY
|
||||
|
||||
/* Compress file given by fname and output to oname.
|
||||
* Returns 0 if successful, error code otherwise.
|
||||
@ -71,10 +71,6 @@ static int compress(const char *fname, const char *oname) {
|
||||
return 1;
|
||||
}
|
||||
|
||||
#ifdef TEST
|
||||
LDM_test((const BYTE *)src);
|
||||
#endif
|
||||
|
||||
gettimeofday(&tv1, NULL);
|
||||
|
||||
compressedSize = LDM_HEADER_SIZE +
|
||||
@ -111,7 +107,7 @@ static int compress(const char *fname, const char *oname) {
|
||||
return 0;
|
||||
}
|
||||
|
||||
#ifdef DECOMPRESS
|
||||
#ifdef DECOMPRESS_AND_VERIFY
|
||||
/* Decompress file compressed using LDM_compress.
|
||||
* The input file should have the LDM_HEADER followed by payload.
|
||||
* Returns 0 if succesful, and an error code otherwise.
|
||||
|
||||
Loading…
Reference in New Issue
Block a user