mirror of
https://github.com/AsahiLinux/u-boot
synced 2024-11-16 17:58:23 +00:00
dd60d1223b
Signed-off-by: Jon Loeliger <jdl@freescale.com>
262 lines
6.9 KiB
C
262 lines
6.9 KiB
C
/*
|
|
* JFFS2 -- Journalling Flash File System, Version 2.
|
|
*
|
|
* Copyright (C) 2004 Patrik Kluba,
|
|
* University of Szeged, Hungary
|
|
*
|
|
* For licensing information, see the file 'LICENCE' in the
|
|
* jffs2 directory.
|
|
*
|
|
* $Id: compr_lzari.c,v 1.3 2004/06/23 16:34:39 havasi Exp $
|
|
*
|
|
*/
|
|
|
|
/*
|
|
Lempel-Ziv-Arithmetic coding compression module for jffs2
|
|
Based on the LZARI source included in LDS (lossless datacompression sources)
|
|
*/
|
|
|
|
/* -*- Mode: C; indent-tabs-mode: t; c-basic-offset: 4; tab-width: 4 -*- */
|
|
|
|
/*
|
|
Original copyright follows:
|
|
|
|
**************************************************************
|
|
LZARI.C -- A Data Compression Program
|
|
(tab = 4 spaces)
|
|
**************************************************************
|
|
4/7/1989 Haruhiko Okumura
|
|
Use, distribute, and modify this program freely.
|
|
Please send me your improved versions.
|
|
PC-VAN SCIENCE
|
|
NIFTY-Serve PAF01022
|
|
CompuServe 74050,1022
|
|
**************************************************************
|
|
|
|
LZARI.C (c)1989 by Haruyasu Yoshizaki, Haruhiko Okumura, and Kenji Rikitake.
|
|
All rights reserved. Permission granted for non-commercial use.
|
|
|
|
*/
|
|
|
|
/*
|
|
|
|
2004-02-18 pajko <pajko(AT)halom(DOT)u-szeged(DOT)hu>
|
|
Removed unused variables and fixed no return value
|
|
|
|
2004-02-16 pajko <pajko(AT)halom(DOT)u-szeged(DOT)hu>
|
|
Initial release
|
|
|
|
*/
|
|
|
|
|
|
#include <config.h>
|
|
#if defined(CONFIG_CMD_JFFS2) && defined(CONFIG_JFFS2_LZO_LZARI)
|
|
|
|
#include <linux/stddef.h>
|
|
#include <jffs2/jffs2.h>
|
|
|
|
|
|
#define N 4096 /* size of ring buffer */
|
|
#define F 60 /* upper limit for match_length */
|
|
#define THRESHOLD 2 /* encode string into position and length
|
|
if match_length is greater than this */
|
|
#define NIL N /* index for root of binary search trees */
|
|
|
|
static unsigned char
|
|
text_buf[N + F - 1]; /* ring buffer of size N,
|
|
with extra F-1 bytes to facilitate string comparison */
|
|
|
|
/********** Arithmetic Compression **********/
|
|
|
|
/* If you are not familiar with arithmetic compression, you should read
|
|
I. E. Witten, R. M. Neal, and J. G. Cleary,
|
|
Communications of the ACM, Vol. 30, pp. 520-540 (1987),
|
|
from which much have been borrowed. */
|
|
|
|
#define M 15
|
|
|
|
/* Q1 (= 2 to the M) must be sufficiently large, but not so
|
|
large as the unsigned long 4 * Q1 * (Q1 - 1) overflows. */
|
|
|
|
#define Q1 (1UL << M)
|
|
#define Q2 (2 * Q1)
|
|
#define Q3 (3 * Q1)
|
|
#define Q4 (4 * Q1)
|
|
#define MAX_CUM (Q1 - 1)
|
|
|
|
#define N_CHAR (256 - THRESHOLD + F)
|
|
/* character code = 0, 1, ..., N_CHAR - 1 */
|
|
|
|
static unsigned long char_to_sym[N_CHAR], sym_to_char[N_CHAR + 1];
|
|
static unsigned long
|
|
sym_freq[N_CHAR + 1], /* frequency for symbols */
|
|
sym_cum[N_CHAR + 1], /* cumulative freq for symbols */
|
|
position_cum[N + 1]; /* cumulative freq for positions */
|
|
|
|
static void StartModel(void) /* Initialize model */
|
|
{
|
|
unsigned long ch, sym, i;
|
|
|
|
sym_cum[N_CHAR] = 0;
|
|
for (sym = N_CHAR; sym >= 1; sym--) {
|
|
ch = sym - 1;
|
|
char_to_sym[ch] = sym; sym_to_char[sym] = ch;
|
|
sym_freq[sym] = 1;
|
|
sym_cum[sym - 1] = sym_cum[sym] + sym_freq[sym];
|
|
}
|
|
sym_freq[0] = 0; /* sentinel (!= sym_freq[1]) */
|
|
position_cum[N] = 0;
|
|
for (i = N; i >= 1; i--)
|
|
position_cum[i - 1] = position_cum[i] + 10000 / (i + 200);
|
|
/* empirical distribution function (quite tentative) */
|
|
/* Please devise a better mechanism! */
|
|
}
|
|
|
|
static void UpdateModel(unsigned long sym)
|
|
{
|
|
unsigned long c, ch_i, ch_sym;
|
|
unsigned long i;
|
|
if (sym_cum[0] >= MAX_CUM) {
|
|
c = 0;
|
|
for (i = N_CHAR; i > 0; i--) {
|
|
sym_cum[i] = c;
|
|
c += (sym_freq[i] = (sym_freq[i] + 1) >> 1);
|
|
}
|
|
sym_cum[0] = c;
|
|
}
|
|
for (i = sym; sym_freq[i] == sym_freq[i - 1]; i--) ;
|
|
if (i < sym) {
|
|
ch_i = sym_to_char[i]; ch_sym = sym_to_char[sym];
|
|
sym_to_char[i] = ch_sym; sym_to_char[sym] = ch_i;
|
|
char_to_sym[ch_i] = sym; char_to_sym[ch_sym] = i;
|
|
}
|
|
sym_freq[i]++;
|
|
while (--i > 0) sym_cum[i]++;
|
|
sym_cum[0]++;
|
|
}
|
|
|
|
static unsigned long BinarySearchSym(unsigned long x)
|
|
/* 1 if x >= sym_cum[1],
|
|
N_CHAR if sym_cum[N_CHAR] > x,
|
|
i such that sym_cum[i - 1] > x >= sym_cum[i] otherwise */
|
|
{
|
|
unsigned long i, j, k;
|
|
|
|
i = 1; j = N_CHAR;
|
|
while (i < j) {
|
|
k = (i + j) / 2;
|
|
if (sym_cum[k] > x) i = k + 1; else j = k;
|
|
}
|
|
return i;
|
|
}
|
|
|
|
unsigned long BinarySearchPos(unsigned long x)
|
|
/* 0 if x >= position_cum[1],
|
|
N - 1 if position_cum[N] > x,
|
|
i such that position_cum[i] > x >= position_cum[i + 1] otherwise */
|
|
{
|
|
unsigned long i, j, k;
|
|
|
|
i = 1; j = N;
|
|
while (i < j) {
|
|
k = (i + j) / 2;
|
|
if (position_cum[k] > x) i = k + 1; else j = k;
|
|
}
|
|
return i - 1;
|
|
}
|
|
|
|
static int Decode(unsigned char *srcbuf, unsigned char *dstbuf, unsigned long srclen,
|
|
unsigned long dstlen) /* Just the reverse of Encode(). */
|
|
{
|
|
unsigned long i, r, j, k, c, range, sym;
|
|
unsigned char *ip, *op;
|
|
unsigned char *srcend = srcbuf + srclen;
|
|
unsigned char *dstend = dstbuf + dstlen;
|
|
unsigned char buffer = 0;
|
|
unsigned char mask = 0;
|
|
unsigned long low = 0;
|
|
unsigned long high = Q4;
|
|
unsigned long value = 0;
|
|
|
|
ip = srcbuf;
|
|
op = dstbuf;
|
|
for (i = 0; i < M + 2; i++) {
|
|
value *= 2;
|
|
if ((mask >>= 1) == 0) {
|
|
buffer = (ip >= srcend) ? 0 : *(ip++);
|
|
mask = 128;
|
|
}
|
|
value += ((buffer & mask) != 0);
|
|
}
|
|
|
|
StartModel();
|
|
for (i = 0; i < N - F; i++) text_buf[i] = ' ';
|
|
r = N - F;
|
|
|
|
while (op < dstend) {
|
|
range = high - low;
|
|
sym = BinarySearchSym((unsigned long)
|
|
(((value - low + 1) * sym_cum[0] - 1) / range));
|
|
high = low + (range * sym_cum[sym - 1]) / sym_cum[0];
|
|
low += (range * sym_cum[sym ]) / sym_cum[0];
|
|
for ( ; ; ) {
|
|
if (low >= Q2) {
|
|
value -= Q2; low -= Q2; high -= Q2;
|
|
} else if (low >= Q1 && high <= Q3) {
|
|
value -= Q1; low -= Q1; high -= Q1;
|
|
} else if (high > Q2) break;
|
|
low += low; high += high;
|
|
value *= 2;
|
|
if ((mask >>= 1) == 0) {
|
|
buffer = (ip >= srcend) ? 0 : *(ip++);
|
|
mask = 128;
|
|
}
|
|
value += ((buffer & mask) != 0);
|
|
}
|
|
c = sym_to_char[sym];
|
|
UpdateModel(sym);
|
|
if (c < 256) {
|
|
if (op >= dstend) return -1;
|
|
*(op++) = c;
|
|
text_buf[r++] = c;
|
|
r &= (N - 1);
|
|
} else {
|
|
j = c - 255 + THRESHOLD;
|
|
range = high - low;
|
|
i = BinarySearchPos((unsigned long)
|
|
(((value - low + 1) * position_cum[0] - 1) / range));
|
|
high = low + (range * position_cum[i ]) / position_cum[0];
|
|
low += (range * position_cum[i + 1]) / position_cum[0];
|
|
for ( ; ; ) {
|
|
if (low >= Q2) {
|
|
value -= Q2; low -= Q2; high -= Q2;
|
|
} else if (low >= Q1 && high <= Q3) {
|
|
value -= Q1; low -= Q1; high -= Q1;
|
|
} else if (high > Q2) break;
|
|
low += low; high += high;
|
|
value *= 2;
|
|
if ((mask >>= 1) == 0) {
|
|
buffer = (ip >= srcend) ? 0 : *(ip++);
|
|
mask = 128;
|
|
}
|
|
value += ((buffer & mask) != 0);
|
|
}
|
|
i = (r - i - 1) & (N - 1);
|
|
for (k = 0; k < j; k++) {
|
|
c = text_buf[(i + k) & (N - 1)];
|
|
if (op >= dstend) return -1;
|
|
*(op++) = c;
|
|
text_buf[r++] = c;
|
|
r &= (N - 1);
|
|
}
|
|
}
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
int lzari_decompress(unsigned char *data_in, unsigned char *cpage_out,
|
|
u32 srclen, u32 destlen)
|
|
{
|
|
return Decode(data_in, cpage_out, srclen, destlen);
|
|
}
|
|
#endif
|