xref: /openbmc/u-boot/fs/jffs2/mini_inflate.c (revision fea25720)
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 #include <jffs2/mini_inflate.h>
28 
29 /* The order that the code lengths in section 3.2.7 are in */
30 static unsigned char huffman_order[] = {16, 17, 18,  0,  8,  7,  9,  6, 10,  5,
31 					11,  4, 12,  3, 13,  2, 14,  1, 15};
32 
33 inline void cramfs_memset(int *s, const int c, size n)
34 {
35 	n--;
36 	for (;n > 0; n--) s[n] = c;
37 	s[0] = c;
38 }
39 
40 /* associate a stream with a block of data and reset the stream */
41 static void init_stream(struct bitstream *stream, unsigned char *data,
42 			void *(*inflate_memcpy)(void *, const void *, size))
43 {
44 	stream->error = NO_ERROR;
45 	stream->memcpy = inflate_memcpy;
46 	stream->decoded = 0;
47 	stream->data = data;
48 	stream->bit = 0;	/* The first bit of the stream is the lsb of the
49 				 * first byte */
50 
51 	/* really sorry about all this initialization, think of a better way,
52 	 * let me know and it will get cleaned up */
53 	stream->codes.bits = 8;
54 	stream->codes.num_symbols = 19;
55 	stream->codes.lengths = stream->code_lengths;
56 	stream->codes.symbols = stream->code_symbols;
57 	stream->codes.count = stream->code_count;
58 	stream->codes.first = stream->code_first;
59 	stream->codes.pos = stream->code_pos;
60 
61 	stream->lengths.bits = 16;
62 	stream->lengths.num_symbols = 288;
63 	stream->lengths.lengths = stream->length_lengths;
64 	stream->lengths.symbols = stream->length_symbols;
65 	stream->lengths.count = stream->length_count;
66 	stream->lengths.first = stream->length_first;
67 	stream->lengths.pos = stream->length_pos;
68 
69 	stream->distance.bits = 16;
70 	stream->distance.num_symbols = 32;
71 	stream->distance.lengths = stream->distance_lengths;
72 	stream->distance.symbols = stream->distance_symbols;
73 	stream->distance.count = stream->distance_count;
74 	stream->distance.first = stream->distance_first;
75 	stream->distance.pos = stream->distance_pos;
76 
77 }
78 
79 /* pull 'bits' bits out of the stream. The last bit pulled it returned as the
80  * msb. (section 3.1.1)
81  */
82 inline unsigned long pull_bits(struct bitstream *stream,
83 			       const unsigned int bits)
84 {
85 	unsigned long ret;
86 	int i;
87 
88 	ret = 0;
89 	for (i = 0; i < bits; i++) {
90 		ret += ((*(stream->data) >> stream->bit) & 1) << i;
91 
92 		/* if, before incrementing, we are on bit 7,
93 		 * go to the lsb of the next byte */
94 		if (stream->bit++ == 7) {
95 			stream->bit = 0;
96 			stream->data++;
97 		}
98 	}
99 	return ret;
100 }
101 
102 inline int pull_bit(struct bitstream *stream)
103 {
104 	int ret = ((*(stream->data) >> stream->bit) & 1);
105 	if (stream->bit++ == 7) {
106 		stream->bit = 0;
107 		stream->data++;
108 	}
109 	return ret;
110 }
111 
112 /* discard bits up to the next whole byte */
113 static void discard_bits(struct bitstream *stream)
114 {
115 	if (stream->bit != 0) {
116 		stream->bit = 0;
117 		stream->data++;
118 	}
119 }
120 
121 /* No decompression, the data is all literals (section 3.2.4) */
122 static void decompress_none(struct bitstream *stream, unsigned char *dest)
123 {
124 	unsigned int length;
125 
126 	discard_bits(stream);
127 	length = *(stream->data++);
128 	length += *(stream->data++) << 8;
129 	pull_bits(stream, 16);	/* throw away the inverse of the size */
130 
131 	stream->decoded += length;
132 	stream->memcpy(dest, stream->data, length);
133 	stream->data += length;
134 }
135 
136 /* Read in a symbol from the stream (section 3.2.2) */
137 static int read_symbol(struct bitstream *stream, struct huffman_set *set)
138 {
139 	int bits = 0;
140 	int code = 0;
141 	while (!(set->count[bits] && code < set->first[bits] +
142 					     set->count[bits])) {
143 		code = (code << 1) + pull_bit(stream);
144 		if (++bits > set->bits) {
145 			/* error decoding (corrupted data?) */
146 			stream->error = CODE_NOT_FOUND;
147 			return -1;
148 		}
149 	}
150 	return set->symbols[set->pos[bits] + code - set->first[bits]];
151 }
152 
153 /* decompress a stream of data encoded with the passed length and distance
154  * huffman codes */
155 static void decompress_huffman(struct bitstream *stream, unsigned char *dest)
156 {
157 	struct huffman_set *lengths = &(stream->lengths);
158 	struct huffman_set *distance = &(stream->distance);
159 
160 	int symbol, length, dist, i;
161 
162 	do {
163 		if ((symbol = read_symbol(stream, lengths)) < 0) return;
164 		if (symbol < 256) {
165 			*(dest++) = symbol; /* symbol is a literal */
166 			stream->decoded++;
167 		} else if (symbol > 256) {
168 			/* Determine the length of the repitition
169 			 * (section 3.2.5) */
170 			if (symbol < 265) length = symbol - 254;
171 			else if (symbol == 285) length = 258;
172 			else {
173 				length = pull_bits(stream, (symbol - 261) >> 2);
174 				length += (4 << ((symbol - 261) >> 2)) + 3;
175 				length += ((symbol - 1) % 4) <<
176 					  ((symbol - 261) >> 2);
177 			}
178 
179 			/* Determine how far back to go */
180 			if ((symbol = read_symbol(stream, distance)) < 0)
181 				return;
182 			if (symbol < 4) dist = symbol + 1;
183 			else {
184 				dist = pull_bits(stream, (symbol - 2) >> 1);
185 				dist += (2 << ((symbol - 2) >> 1)) + 1;
186 				dist += (symbol % 2) << ((symbol - 2) >> 1);
187 			}
188 			stream->decoded += length;
189 			for (i = 0; i < length; i++) {
190 				*dest = dest[-dist];
191 				dest++;
192 			}
193 		}
194 	} while (symbol != 256); /* 256 is the end of the data block */
195 }
196 
197 /* Fill the lookup tables (section 3.2.2) */
198 static void fill_code_tables(struct huffman_set *set)
199 {
200 	int code = 0, i, length;
201 
202 	/* fill in the first code of each bit length, and the pos pointer */
203 	set->pos[0] = 0;
204 	for (i = 1; i < set->bits; i++) {
205 		code = (code + set->count[i - 1]) << 1;
206 		set->first[i] = code;
207 		set->pos[i] = set->pos[i - 1] + set->count[i - 1];
208 	}
209 
210 	/* Fill in the table of symbols in order of their huffman code */
211 	for (i = 0; i < set->num_symbols; i++) {
212 		if ((length = set->lengths[i]))
213 			set->symbols[set->pos[length]++] = i;
214 	}
215 
216 	/* reset the pos pointer */
217 	for (i = 1; i < set->bits; i++) set->pos[i] -= set->count[i];
218 }
219 
220 static void init_code_tables(struct huffman_set *set)
221 {
222 	cramfs_memset(set->lengths, 0, set->num_symbols);
223 	cramfs_memset(set->count, 0, set->bits);
224 	cramfs_memset(set->first, 0, set->bits);
225 }
226 
227 /* read in the huffman codes for dynamic decoding (section 3.2.7) */
228 static void decompress_dynamic(struct bitstream *stream, unsigned char *dest)
229 {
230 	/* I tried my best to minimize the memory footprint here, while still
231 	 * keeping up performance. I really dislike the _lengths[] tables, but
232 	 * I see no way of eliminating them without a sizable performance
233 	 * impact. The first struct table keeps track of stats on each bit
234 	 * length. The _length table keeps a record of the bit length of each
235 	 * symbol. The _symbols table is for looking up symbols by the huffman
236 	 * code (the pos element points to the first place in the symbol table
237 	 * where that bit length occurs). I also hate the initization of these
238 	 * structs, if someone knows how to compact these, lemme know. */
239 
240 	struct huffman_set *codes = &(stream->codes);
241 	struct huffman_set *lengths = &(stream->lengths);
242 	struct huffman_set *distance = &(stream->distance);
243 
244 	int hlit = pull_bits(stream, 5) + 257;
245 	int hdist = pull_bits(stream, 5) + 1;
246 	int hclen = pull_bits(stream, 4) + 4;
247 	int length, curr_code, symbol, i, last_code;
248 
249 	last_code = 0;
250 
251 	init_code_tables(codes);
252 	init_code_tables(lengths);
253 	init_code_tables(distance);
254 
255 	/* fill in the count of each bit length' as well as the lengths
256 	 * table */
257 	for (i = 0; i < hclen; i++) {
258 		length = pull_bits(stream, 3);
259 		codes->lengths[huffman_order[i]] = length;
260 		if (length) codes->count[length]++;
261 
262 	}
263 	fill_code_tables(codes);
264 
265 	/* Do the same for the length codes, being carefull of wrap through
266 	 * to the distance table */
267 	curr_code = 0;
268 	while (curr_code < hlit) {
269 		if ((symbol = read_symbol(stream, codes)) < 0) return;
270 		if (symbol == 0) {
271 			curr_code++;
272 			last_code = 0;
273 		} else if (symbol < 16) { /* Literal length */
274 			lengths->lengths[curr_code] =  last_code = symbol;
275 			lengths->count[symbol]++;
276 			curr_code++;
277 		} else if (symbol == 16) { /* repeat the last symbol 3 - 6
278 					    * times */
279 			length = 3 + pull_bits(stream, 2);
280 			for (;length; length--, curr_code++)
281 				if (curr_code < hlit) {
282 					lengths->lengths[curr_code] =
283 						last_code;
284 					lengths->count[last_code]++;
285 				} else { /* wrap to the distance table */
286 					distance->lengths[curr_code - hlit] =
287 						last_code;
288 					distance->count[last_code]++;
289 				}
290 		} else if (symbol == 17) { /* repeat a bit length 0 */
291 			curr_code += 3 + pull_bits(stream, 3);
292 			last_code = 0;
293 		} else { /* same, but more times */
294 			curr_code += 11 + pull_bits(stream, 7);
295 			last_code = 0;
296 		}
297 	}
298 	fill_code_tables(lengths);
299 
300 	/* Fill the distance table, don't need to worry about wrapthrough
301 	 * here */
302 	curr_code -= hlit;
303 	while (curr_code < hdist) {
304 		if ((symbol = read_symbol(stream, codes)) < 0) return;
305 		if (symbol == 0) {
306 			curr_code++;
307 			last_code = 0;
308 		} else if (symbol < 16) {
309 			distance->lengths[curr_code] = last_code = symbol;
310 			distance->count[symbol]++;
311 			curr_code++;
312 		} else if (symbol == 16) {
313 			length = 3 + pull_bits(stream, 2);
314 			for (;length; length--, curr_code++) {
315 				distance->lengths[curr_code] =
316 					last_code;
317 				distance->count[last_code]++;
318 			}
319 		} else if (symbol == 17) {
320 			curr_code += 3 + pull_bits(stream, 3);
321 			last_code = 0;
322 		} else {
323 			curr_code += 11 + pull_bits(stream, 7);
324 			last_code = 0;
325 		}
326 	}
327 	fill_code_tables(distance);
328 
329 	decompress_huffman(stream, dest);
330 }
331 
332 /* fill in the length and distance huffman codes for fixed encoding
333  * (section 3.2.6) */
334 static void decompress_fixed(struct bitstream *stream, unsigned char *dest)
335 {
336 	/* let gcc fill in the initial values */
337 	struct huffman_set *lengths = &(stream->lengths);
338 	struct huffman_set *distance = &(stream->distance);
339 
340 	cramfs_memset(lengths->count, 0, 16);
341 	cramfs_memset(lengths->first, 0, 16);
342 	cramfs_memset(lengths->lengths, 8, 144);
343 	cramfs_memset(lengths->lengths + 144, 9, 112);
344 	cramfs_memset(lengths->lengths + 256, 7, 24);
345 	cramfs_memset(lengths->lengths + 280, 8, 8);
346 	lengths->count[7] = 24;
347 	lengths->count[8] = 152;
348 	lengths->count[9] = 112;
349 
350 	cramfs_memset(distance->count, 0, 16);
351 	cramfs_memset(distance->first, 0, 16);
352 	cramfs_memset(distance->lengths, 5, 32);
353 	distance->count[5] = 32;
354 
355 
356 	fill_code_tables(lengths);
357 	fill_code_tables(distance);
358 
359 
360 	decompress_huffman(stream, dest);
361 }
362 
363 /* returns the number of bytes decoded, < 0 if there was an error. Note that
364  * this function assumes that the block starts on a byte boundry
365  * (non-compliant, but I don't see where this would happen). section 3.2.3 */
366 long decompress_block(unsigned char *dest, unsigned char *source,
367 		      void *(*inflate_memcpy)(void *, const void *, size))
368 {
369 	int bfinal, btype;
370 	struct bitstream stream;
371 
372 	init_stream(&stream, source, inflate_memcpy);
373 	do {
374 		bfinal = pull_bit(&stream);
375 		btype = pull_bits(&stream, 2);
376 		if (btype == NO_COMP) decompress_none(&stream, dest + stream.decoded);
377 		else if (btype == DYNAMIC_COMP)
378 			decompress_dynamic(&stream, dest + stream.decoded);
379 		else if (btype == FIXED_COMP) decompress_fixed(&stream, dest + stream.decoded);
380 		else stream.error = COMP_UNKNOWN;
381 	} while (!bfinal && !stream.error);
382 
383 #if 0
384 	putstr("decompress_block start\r\n");
385 	putLabeledWord("stream.error = ",stream.error);
386 	putLabeledWord("stream.decoded = ",stream.decoded);
387 	putLabeledWord("dest = ",dest);
388 	putstr("decompress_block end\r\n");
389 #endif
390 	return stream.error ? -stream.error : stream.decoded;
391 }
392