xref: /openbmc/u-boot/fs/jffs2/mini_inflate.c (revision 415a613b)
1 /*-------------------------------------------------------------------------
2  * Filename:      mini_inflate.c
3  * Version:       $Id: mini_inflate.c,v 1.3 2002/01/24 22:58:42 rfeany Exp $
4  * Copyright:     Copyright (C) 2001, Russ Dill
5  * Author:        Russ Dill <Russ.Dill@asu.edu>
6  * Description:   Mini inflate implementation (RFC 1951)
7  *-----------------------------------------------------------------------*/
8 /*
9  *
10  * This program is free software; you can redistribute it and/or modify
11  * it under the terms of the GNU General Public License as published by
12  * the Free Software Foundation; either version 2 of the License, or
13  * (at your option) any later version.
14  *
15  * This program is distributed in the hope that it will be useful,
16  * but WITHOUT ANY WARRANTY; without even the implied warranty of
17  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
18  * GNU General Public License for more details.
19  *
20  * You should have received a copy of the GNU General Public License
21  * along with this program; if not, write to the Free Software
22  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
23  *
24  */
25 
26 #include <config.h>
27 
28 #if defined(CONFIG_CMD_JFFS2)
29 
30 #include <jffs2/mini_inflate.h>
31 
32 /* The order that the code lengths in section 3.2.7 are in */
33 static unsigned char huffman_order[] = {16, 17, 18,  0,  8,  7,  9,  6, 10,  5,
34 					11,  4, 12,  3, 13,  2, 14,  1, 15};
35 
36 inline void cramfs_memset(int *s, const int c, size n)
37 {
38 	n--;
39 	for (;n > 0; n--) s[n] = c;
40 	s[0] = c;
41 }
42 
43 /* associate a stream with a block of data and reset the stream */
44 static void init_stream(struct bitstream *stream, unsigned char *data,
45 			void *(*inflate_memcpy)(void *, const void *, size))
46 {
47 	stream->error = NO_ERROR;
48 	stream->memcpy = inflate_memcpy;
49 	stream->decoded = 0;
50 	stream->data = data;
51 	stream->bit = 0;	/* The first bit of the stream is the lsb of the
52 				 * first byte */
53 
54 	/* really sorry about all this initialization, think of a better way,
55 	 * let me know and it will get cleaned up */
56 	stream->codes.bits = 8;
57 	stream->codes.num_symbols = 19;
58 	stream->codes.lengths = stream->code_lengths;
59 	stream->codes.symbols = stream->code_symbols;
60 	stream->codes.count = stream->code_count;
61 	stream->codes.first = stream->code_first;
62 	stream->codes.pos = stream->code_pos;
63 
64 	stream->lengths.bits = 16;
65 	stream->lengths.num_symbols = 288;
66 	stream->lengths.lengths = stream->length_lengths;
67 	stream->lengths.symbols = stream->length_symbols;
68 	stream->lengths.count = stream->length_count;
69 	stream->lengths.first = stream->length_first;
70 	stream->lengths.pos = stream->length_pos;
71 
72 	stream->distance.bits = 16;
73 	stream->distance.num_symbols = 32;
74 	stream->distance.lengths = stream->distance_lengths;
75 	stream->distance.symbols = stream->distance_symbols;
76 	stream->distance.count = stream->distance_count;
77 	stream->distance.first = stream->distance_first;
78 	stream->distance.pos = stream->distance_pos;
79 
80 }
81 
82 /* pull 'bits' bits out of the stream. The last bit pulled it returned as the
83  * msb. (section 3.1.1)
84  */
85 inline unsigned long pull_bits(struct bitstream *stream,
86 			       const unsigned int bits)
87 {
88 	unsigned long ret;
89 	int i;
90 
91 	ret = 0;
92 	for (i = 0; i < bits; i++) {
93 		ret += ((*(stream->data) >> stream->bit) & 1) << i;
94 
95 		/* if, before incrementing, we are on bit 7,
96 		 * go to the lsb of the next byte */
97 		if (stream->bit++ == 7) {
98 			stream->bit = 0;
99 			stream->data++;
100 		}
101 	}
102 	return ret;
103 }
104 
105 inline int pull_bit(struct bitstream *stream)
106 {
107 	int ret = ((*(stream->data) >> stream->bit) & 1);
108 	if (stream->bit++ == 7) {
109 		stream->bit = 0;
110 		stream->data++;
111 	}
112 	return ret;
113 }
114 
115 /* discard bits up to the next whole byte */
116 static void discard_bits(struct bitstream *stream)
117 {
118 	if (stream->bit != 0) {
119 		stream->bit = 0;
120 		stream->data++;
121 	}
122 }
123 
124 /* No decompression, the data is all literals (section 3.2.4) */
125 static void decompress_none(struct bitstream *stream, unsigned char *dest)
126 {
127 	unsigned int length;
128 
129 	discard_bits(stream);
130 	length = *(stream->data++);
131 	length += *(stream->data++) << 8;
132 	pull_bits(stream, 16);	/* throw away the inverse of the size */
133 
134 	stream->decoded += length;
135 	stream->memcpy(dest, stream->data, length);
136 	stream->data += length;
137 }
138 
139 /* Read in a symbol from the stream (section 3.2.2) */
140 static int read_symbol(struct bitstream *stream, struct huffman_set *set)
141 {
142 	int bits = 0;
143 	int code = 0;
144 	while (!(set->count[bits] && code < set->first[bits] +
145 					     set->count[bits])) {
146 		code = (code << 1) + pull_bit(stream);
147 		if (++bits > set->bits) {
148 			/* error decoding (corrupted data?) */
149 			stream->error = CODE_NOT_FOUND;
150 			return -1;
151 		}
152 	}
153 	return set->symbols[set->pos[bits] + code - set->first[bits]];
154 }
155 
156 /* decompress a stream of data encoded with the passed length and distance
157  * huffman codes */
158 static void decompress_huffman(struct bitstream *stream, unsigned char *dest)
159 {
160 	struct huffman_set *lengths = &(stream->lengths);
161 	struct huffman_set *distance = &(stream->distance);
162 
163 	int symbol, length, dist, i;
164 
165 	do {
166 		if ((symbol = read_symbol(stream, lengths)) < 0) return;
167 		if (symbol < 256) {
168 			*(dest++) = symbol; /* symbol is a literal */
169 			stream->decoded++;
170 		} else if (symbol > 256) {
171 			/* Determine the length of the repitition
172 			 * (section 3.2.5) */
173 			if (symbol < 265) length = symbol - 254;
174 			else if (symbol == 285) length = 258;
175 			else {
176 				length = pull_bits(stream, (symbol - 261) >> 2);
177 				length += (4 << ((symbol - 261) >> 2)) + 3;
178 				length += ((symbol - 1) % 4) <<
179 					  ((symbol - 261) >> 2);
180 			}
181 
182 			/* Determine how far back to go */
183 			if ((symbol = read_symbol(stream, distance)) < 0)
184 				return;
185 			if (symbol < 4) dist = symbol + 1;
186 			else {
187 				dist = pull_bits(stream, (symbol - 2) >> 1);
188 				dist += (2 << ((symbol - 2) >> 1)) + 1;
189 				dist += (symbol % 2) << ((symbol - 2) >> 1);
190 			}
191 			stream->decoded += length;
192 			for (i = 0; i < length; i++) {
193 				*dest = dest[-dist];
194 				dest++;
195 			}
196 		}
197 	} while (symbol != 256); /* 256 is the end of the data block */
198 }
199 
200 /* Fill the lookup tables (section 3.2.2) */
201 static void fill_code_tables(struct huffman_set *set)
202 {
203 	int code = 0, i, length;
204 
205 	/* fill in the first code of each bit length, and the pos pointer */
206 	set->pos[0] = 0;
207 	for (i = 1; i < set->bits; i++) {
208 		code = (code + set->count[i - 1]) << 1;
209 		set->first[i] = code;
210 		set->pos[i] = set->pos[i - 1] + set->count[i - 1];
211 	}
212 
213 	/* Fill in the table of symbols in order of their huffman code */
214 	for (i = 0; i < set->num_symbols; i++) {
215 		if ((length = set->lengths[i]))
216 			set->symbols[set->pos[length]++] = i;
217 	}
218 
219 	/* reset the pos pointer */
220 	for (i = 1; i < set->bits; i++) set->pos[i] -= set->count[i];
221 }
222 
223 static void init_code_tables(struct huffman_set *set)
224 {
225 	cramfs_memset(set->lengths, 0, set->num_symbols);
226 	cramfs_memset(set->count, 0, set->bits);
227 	cramfs_memset(set->first, 0, set->bits);
228 }
229 
230 /* read in the huffman codes for dynamic decoding (section 3.2.7) */
231 static void decompress_dynamic(struct bitstream *stream, unsigned char *dest)
232 {
233 	/* I tried my best to minimize the memory footprint here, while still
234 	 * keeping up performance. I really dislike the _lengths[] tables, but
235 	 * I see no way of eliminating them without a sizable performance
236 	 * impact. The first struct table keeps track of stats on each bit
237 	 * length. The _length table keeps a record of the bit length of each
238 	 * symbol. The _symbols table is for looking up symbols by the huffman
239 	 * code (the pos element points to the first place in the symbol table
240 	 * where that bit length occurs). I also hate the initization of these
241 	 * structs, if someone knows how to compact these, lemme know. */
242 
243 	struct huffman_set *codes = &(stream->codes);
244 	struct huffman_set *lengths = &(stream->lengths);
245 	struct huffman_set *distance = &(stream->distance);
246 
247 	int hlit = pull_bits(stream, 5) + 257;
248 	int hdist = pull_bits(stream, 5) + 1;
249 	int hclen = pull_bits(stream, 4) + 4;
250 	int length, curr_code, symbol, i, last_code;
251 
252 	last_code = 0;
253 
254 	init_code_tables(codes);
255 	init_code_tables(lengths);
256 	init_code_tables(distance);
257 
258 	/* fill in the count of each bit length' as well as the lengths
259 	 * table */
260 	for (i = 0; i < hclen; i++) {
261 		length = pull_bits(stream, 3);
262 		codes->lengths[huffman_order[i]] = length;
263 		if (length) codes->count[length]++;
264 
265 	}
266 	fill_code_tables(codes);
267 
268 	/* Do the same for the length codes, being carefull of wrap through
269 	 * to the distance table */
270 	curr_code = 0;
271 	while (curr_code < hlit) {
272 		if ((symbol = read_symbol(stream, codes)) < 0) return;
273 		if (symbol == 0) {
274 			curr_code++;
275 			last_code = 0;
276 		} else if (symbol < 16) { /* Literal length */
277 			lengths->lengths[curr_code] =  last_code = symbol;
278 			lengths->count[symbol]++;
279 			curr_code++;
280 		} else if (symbol == 16) { /* repeat the last symbol 3 - 6
281 					    * times */
282 			length = 3 + pull_bits(stream, 2);
283 			for (;length; length--, curr_code++)
284 				if (curr_code < hlit) {
285 					lengths->lengths[curr_code] =
286 						last_code;
287 					lengths->count[last_code]++;
288 				} else { /* wrap to the distance table */
289 					distance->lengths[curr_code - hlit] =
290 						last_code;
291 					distance->count[last_code]++;
292 				}
293 		} else if (symbol == 17) { /* repeat a bit length 0 */
294 			curr_code += 3 + pull_bits(stream, 3);
295 			last_code = 0;
296 		} else { /* same, but more times */
297 			curr_code += 11 + pull_bits(stream, 7);
298 			last_code = 0;
299 		}
300 	}
301 	fill_code_tables(lengths);
302 
303 	/* Fill the distance table, don't need to worry about wrapthrough
304 	 * here */
305 	curr_code -= hlit;
306 	while (curr_code < hdist) {
307 		if ((symbol = read_symbol(stream, codes)) < 0) return;
308 		if (symbol == 0) {
309 			curr_code++;
310 			last_code = 0;
311 		} else if (symbol < 16) {
312 			distance->lengths[curr_code] = last_code = symbol;
313 			distance->count[symbol]++;
314 			curr_code++;
315 		} else if (symbol == 16) {
316 			length = 3 + pull_bits(stream, 2);
317 			for (;length; length--, curr_code++) {
318 				distance->lengths[curr_code] =
319 					last_code;
320 				distance->count[last_code]++;
321 			}
322 		} else if (symbol == 17) {
323 			curr_code += 3 + pull_bits(stream, 3);
324 			last_code = 0;
325 		} else {
326 			curr_code += 11 + pull_bits(stream, 7);
327 			last_code = 0;
328 		}
329 	}
330 	fill_code_tables(distance);
331 
332 	decompress_huffman(stream, dest);
333 }
334 
335 /* fill in the length and distance huffman codes for fixed encoding
336  * (section 3.2.6) */
337 static void decompress_fixed(struct bitstream *stream, unsigned char *dest)
338 {
339 	/* let gcc fill in the initial values */
340 	struct huffman_set *lengths = &(stream->lengths);
341 	struct huffman_set *distance = &(stream->distance);
342 
343 	cramfs_memset(lengths->count, 0, 16);
344 	cramfs_memset(lengths->first, 0, 16);
345 	cramfs_memset(lengths->lengths, 8, 144);
346 	cramfs_memset(lengths->lengths + 144, 9, 112);
347 	cramfs_memset(lengths->lengths + 256, 7, 24);
348 	cramfs_memset(lengths->lengths + 280, 8, 8);
349 	lengths->count[7] = 24;
350 	lengths->count[8] = 152;
351 	lengths->count[9] = 112;
352 
353 	cramfs_memset(distance->count, 0, 16);
354 	cramfs_memset(distance->first, 0, 16);
355 	cramfs_memset(distance->lengths, 5, 32);
356 	distance->count[5] = 32;
357 
358 
359 	fill_code_tables(lengths);
360 	fill_code_tables(distance);
361 
362 
363 	decompress_huffman(stream, dest);
364 }
365 
366 /* returns the number of bytes decoded, < 0 if there was an error. Note that
367  * this function assumes that the block starts on a byte boundry
368  * (non-compliant, but I don't see where this would happen). section 3.2.3 */
369 long decompress_block(unsigned char *dest, unsigned char *source,
370 		      void *(*inflate_memcpy)(void *, const void *, size))
371 {
372 	int bfinal, btype;
373 	struct bitstream stream;
374 
375 	init_stream(&stream, source, inflate_memcpy);
376 	do {
377 		bfinal = pull_bit(&stream);
378 		btype = pull_bits(&stream, 2);
379 		if (btype == NO_COMP) decompress_none(&stream, dest + stream.decoded);
380 		else if (btype == DYNAMIC_COMP)
381 			decompress_dynamic(&stream, dest + stream.decoded);
382 		else if (btype == FIXED_COMP) decompress_fixed(&stream, dest + stream.decoded);
383 		else stream.error = COMP_UNKNOWN;
384 	} while (!bfinal && !stream.error);
385 
386 #if 0
387 	putstr("decompress_block start\r\n");
388 	putLabeledWord("stream.error = ",stream.error);
389 	putLabeledWord("stream.decoded = ",stream.decoded);
390 	putLabeledWord("dest = ",dest);
391 	putstr("decompress_block end\r\n");
392 #endif
393 	return stream.error ? -stream.error : stream.decoded;
394 }
395 
396 #endif
397