greg/pcompress
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

Switch to multiplicative rolling hash for good distribution properties.

Browse source
This commit is contained in:
Moinak Ghosh 2012-09-29 00:09:49 +05:30
parent d94be4e314
commit 24e6f4e629
2 changed files with 22 additions and 18 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

36
rabin/rabin_dedup.c
View file

@ -84,7 +84,7 @@ extern int bspatch(u_char *pbuf, u_char *old, bsize_t oldsize, u_char *new,
bsize_t *_newsize);
static pthread_mutex_t init_lock = PTHREAD_MUTEX_INITIALIZER;
uint64_t ir[256];
uint64_t ir[256], out[256];
static int inited = 0;
static uint32_t
@ -130,16 +130,22 @@ create_dedupe_context(uint64_t chunksize, uint64_t real_chunksize, int rab_blk_s
pthread_mutex_lock(&init_lock);
if (!inited) {
int term, j;
uint64_t val;
uint64_t val, poly_pow;
poly_pow = 1;
for (j = 0; j < RAB_POLYNOMIAL_WIN_SIZE; j++) {
poly_pow = (poly_pow * RAB_POLYNOMIAL_CONST) & POLY_MASK;
}
for (j = 0; j < 256; j++) {
term = 1;
val = 0;
val = 1;
out[j] = (j * poly_pow) & POLY_MASK;
for (i=0; i<RAB_POLYNOMIAL_WIN_SIZE; i++) {
if (term & FP_POLY) {
val += term * j;
val += ((term * j) & POLY_MASK);
}
term <<= 1;
term = (term * RAB_POLYNOMIAL_CONST) & POLY_MASK;
}
ir[j] = val;
}
@ -346,10 +352,11 @@ dedupe_compress(dedupe_context_t *ctx, uchar_t *buf, ssize_t *size, ssize_t offs
for (i=offset; i<*size; i++) {
uchar_t cur_byte = buf1[i];
uint64_t pushed_out = ctx->current_window_data[ctx->window_pos];
ctx->current_window_data[ctx->window_pos] = cur_byte;
cur_roll_checksum = (cur_roll_checksum << 1) + cur_byte;
cur_roll_checksum -= (pushed_out << RAB_POLYNOMIAL_WIN_SIZE);
cur_roll_checksum = (cur_roll_checksum * RAB_POLYNOMIAL_CONST) & POLY_MASK;
cur_roll_checksum += cur_byte;
cur_roll_checksum -= out[pushed_out];
cur_pos_checksum = cur_roll_checksum ^ ir[pushed_out];
ctx->window_pos = (ctx->window_pos + 1) & (RAB_POLYNOMIAL_WIN_SIZE-1);
@ -377,15 +384,10 @@ dedupe_compress(dedupe_context_t *ctx, uchar_t *buf, ssize_t *size, ssize_t offs
uchar_t cur_byte = buf1[i];
uint64_t pushed_out = ctx->current_window_data[ctx->window_pos];
ctx->current_window_data[ctx->window_pos] = cur_byte;
/*
* We want to do:
* cur_roll_checksum = cur_roll_checksum * RAB_POLYNOMIAL_CONST + cur_byte;
* cur_roll_checksum -= pushed_out * polynomial_pow;
*
* However since RAB_POLYNOMIAL_CONST == 2, we use shifts.
*/
cur_roll_checksum = (cur_roll_checksum << 1) + cur_byte;
cur_roll_checksum -= (pushed_out << RAB_POLYNOMIAL_WIN_SIZE);
cur_roll_checksum = (cur_roll_checksum * RAB_POLYNOMIAL_CONST) & POLY_MASK;
cur_roll_checksum += cur_byte;
cur_roll_checksum -= out[pushed_out];
cur_pos_checksum = cur_roll_checksum ^ ir[pushed_out];
/*

4
rabin/rabin_dedup.h
View file

@ -63,7 +63,9 @@
//List of constants, mostly constraints and defaults for various parameters
//to the Rabin Fingerprinting algorithm
#define RAB_POLYNOMIAL_CONST 2
//Use prime constant from Bulat Ziganshin's REP. Seems to work best across wide range of data.
#define RAB_POLYNOMIAL_CONST 153191
#define POLY_MASK (0xffffffffffULL)
#define RAB_BLK_DEFAULT 1
#define RAB_BLK_MIN_BITS 11
#define LZMA_WINDOW_MAX (128L * 1024L * 1024L)