xref: /linux/lib/lz4/lz4_decompress.c (revision e2be04c7f9958dde770eeb8b30e829ca969b37bb)
1 /*
2  * LZ4 - Fast LZ compression algorithm
3  * Copyright (C) 2011 - 2016, Yann Collet.
4  * BSD 2 - Clause License (http://www.opensource.org/licenses/bsd - license.php)
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions are
7  * met:
8  *	* Redistributions of source code must retain the above copyright
9  *	  notice, this list of conditions and the following disclaimer.
10  *	* Redistributions in binary form must reproduce the above
11  * copyright notice, this list of conditions and the following disclaimer
12  * in the documentation and/or other materials provided with the
13  * distribution.
14  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
15  * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
16  * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
17  * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
18  * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
19  * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
20  * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
21  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
22  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
23  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
24  * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
25  * You can contact the author at :
26  *	- LZ4 homepage : http://www.lz4.org
27  *	- LZ4 source repository : https://github.com/lz4/lz4
28  *
29  *	Changed for kernel usage by:
30  *	Sven Schmidt <4sschmid@informatik.uni-hamburg.de>
31  */
32 
33 /*-************************************
34  *	Dependencies
35  **************************************/
36 #include <linux/lz4.h>
37 #include "lz4defs.h"
38 #include <linux/init.h>
39 #include <linux/module.h>
40 #include <linux/kernel.h>
41 #include <asm/unaligned.h>
42 
43 /*-*****************************
44  *	Decompression functions
45  *******************************/
46 /* LZ4_decompress_generic() :
47  * This generic decompression function cover all use cases.
48  * It shall be instantiated several times, using different sets of directives
49  * Note that it is important this generic function is really inlined,
50  * in order to remove useless branches during compilation optimization.
51  */
52 static FORCE_INLINE int LZ4_decompress_generic(
53 	 const char * const source,
54 	 char * const dest,
55 	 int inputSize,
56 		/*
57 		 * If endOnInput == endOnInputSize,
58 		 * this value is the max size of Output Buffer.
59 		 */
60 	 int outputSize,
61 	 /* endOnOutputSize, endOnInputSize */
62 	 int endOnInput,
63 	 /* full, partial */
64 	 int partialDecoding,
65 	 /* only used if partialDecoding == partial */
66 	 int targetOutputSize,
67 	 /* noDict, withPrefix64k, usingExtDict */
68 	 int dict,
69 	 /* == dest when no prefix */
70 	 const BYTE * const lowPrefix,
71 	 /* only if dict == usingExtDict */
72 	 const BYTE * const dictStart,
73 	 /* note : = 0 if noDict */
74 	 const size_t dictSize
75 	 )
76 {
77 	/* Local Variables */
78 	const BYTE *ip = (const BYTE *) source;
79 	const BYTE * const iend = ip + inputSize;
80 
81 	BYTE *op = (BYTE *) dest;
82 	BYTE * const oend = op + outputSize;
83 	BYTE *cpy;
84 	BYTE *oexit = op + targetOutputSize;
85 	const BYTE * const lowLimit = lowPrefix - dictSize;
86 
87 	const BYTE * const dictEnd = (const BYTE *)dictStart + dictSize;
88 	static const unsigned int dec32table[] = { 0, 1, 2, 1, 4, 4, 4, 4 };
89 	static const int dec64table[] = { 0, 0, 0, -1, 0, 1, 2, 3 };
90 
91 	const int safeDecode = (endOnInput == endOnInputSize);
92 	const int checkOffset = ((safeDecode) && (dictSize < (int)(64 * KB)));
93 
94 	/* Special cases */
95 	/* targetOutputSize too high => decode everything */
96 	if ((partialDecoding) && (oexit > oend - MFLIMIT))
97 		oexit = oend - MFLIMIT;
98 
99 	/* Empty output buffer */
100 	if ((endOnInput) && (unlikely(outputSize == 0)))
101 		return ((inputSize == 1) && (*ip == 0)) ? 0 : -1;
102 
103 	if ((!endOnInput) && (unlikely(outputSize == 0)))
104 		return (*ip == 0 ? 1 : -1);
105 
106 	/* Main Loop : decode sequences */
107 	while (1) {
108 		size_t length;
109 		const BYTE *match;
110 		size_t offset;
111 
112 		/* get literal length */
113 		unsigned int const token = *ip++;
114 
115 		length = token>>ML_BITS;
116 
117 		if (length == RUN_MASK) {
118 			unsigned int s;
119 
120 			do {
121 				s = *ip++;
122 				length += s;
123 			} while (likely(endOnInput
124 				? ip < iend - RUN_MASK
125 				: 1) & (s == 255));
126 
127 			if ((safeDecode)
128 				&& unlikely(
129 					(size_t)(op + length) < (size_t)(op))) {
130 				/* overflow detection */
131 				goto _output_error;
132 			}
133 			if ((safeDecode)
134 				&& unlikely(
135 					(size_t)(ip + length) < (size_t)(ip))) {
136 				/* overflow detection */
137 				goto _output_error;
138 			}
139 		}
140 
141 		/* copy literals */
142 		cpy = op + length;
143 		if (((endOnInput) && ((cpy > (partialDecoding ? oexit : oend - MFLIMIT))
144 			|| (ip + length > iend - (2 + 1 + LASTLITERALS))))
145 			|| ((!endOnInput) && (cpy > oend - WILDCOPYLENGTH))) {
146 			if (partialDecoding) {
147 				if (cpy > oend) {
148 					/*
149 					 * Error :
150 					 * write attempt beyond end of output buffer
151 					 */
152 					goto _output_error;
153 				}
154 				if ((endOnInput)
155 					&& (ip + length > iend)) {
156 					/*
157 					 * Error :
158 					 * read attempt beyond
159 					 * end of input buffer
160 					 */
161 					goto _output_error;
162 				}
163 			} else {
164 				if ((!endOnInput)
165 					&& (cpy != oend)) {
166 					/*
167 					 * Error :
168 					 * block decoding must
169 					 * stop exactly there
170 					 */
171 					goto _output_error;
172 				}
173 				if ((endOnInput)
174 					&& ((ip + length != iend)
175 					|| (cpy > oend))) {
176 					/*
177 					 * Error :
178 					 * input must be consumed
179 					 */
180 					goto _output_error;
181 				}
182 			}
183 
184 			memcpy(op, ip, length);
185 			ip += length;
186 			op += length;
187 			/* Necessarily EOF, due to parsing restrictions */
188 			break;
189 		}
190 
191 		LZ4_wildCopy(op, ip, cpy);
192 		ip += length;
193 		op = cpy;
194 
195 		/* get offset */
196 		offset = LZ4_readLE16(ip);
197 		ip += 2;
198 		match = op - offset;
199 
200 		if ((checkOffset) && (unlikely(match < lowLimit))) {
201 			/* Error : offset outside buffers */
202 			goto _output_error;
203 		}
204 
205 		/* costs ~1%; silence an msan warning when offset == 0 */
206 		LZ4_write32(op, (U32)offset);
207 
208 		/* get matchlength */
209 		length = token & ML_MASK;
210 		if (length == ML_MASK) {
211 			unsigned int s;
212 
213 			do {
214 				s = *ip++;
215 
216 				if ((endOnInput) && (ip > iend - LASTLITERALS))
217 					goto _output_error;
218 
219 				length += s;
220 			} while (s == 255);
221 
222 			if ((safeDecode)
223 				&& unlikely(
224 					(size_t)(op + length) < (size_t)op)) {
225 				/* overflow detection */
226 				goto _output_error;
227 			}
228 		}
229 
230 		length += MINMATCH;
231 
232 		/* check external dictionary */
233 		if ((dict == usingExtDict) && (match < lowPrefix)) {
234 			if (unlikely(op + length > oend - LASTLITERALS)) {
235 				/* doesn't respect parsing restriction */
236 				goto _output_error;
237 			}
238 
239 			if (length <= (size_t)(lowPrefix - match)) {
240 				/*
241 				 * match can be copied as a single segment
242 				 * from external dictionary
243 				 */
244 				memmove(op, dictEnd - (lowPrefix - match),
245 					length);
246 				op += length;
247 			} else {
248 				/*
249 				 * match encompass external
250 				 * dictionary and current block
251 				 */
252 				size_t const copySize = (size_t)(lowPrefix - match);
253 				size_t const restSize = length - copySize;
254 
255 				memcpy(op, dictEnd - copySize, copySize);
256 				op += copySize;
257 
258 				if (restSize > (size_t)(op - lowPrefix)) {
259 					/* overlap copy */
260 					BYTE * const endOfMatch = op + restSize;
261 					const BYTE *copyFrom = lowPrefix;
262 
263 					while (op < endOfMatch)
264 						*op++ = *copyFrom++;
265 				} else {
266 					memcpy(op, lowPrefix, restSize);
267 					op += restSize;
268 				}
269 			}
270 
271 			continue;
272 		}
273 
274 		/* copy match within block */
275 		cpy = op + length;
276 
277 		if (unlikely(offset < 8)) {
278 			const int dec64 = dec64table[offset];
279 
280 			op[0] = match[0];
281 			op[1] = match[1];
282 			op[2] = match[2];
283 			op[3] = match[3];
284 			match += dec32table[offset];
285 			memcpy(op + 4, match, 4);
286 			match -= dec64;
287 		} else {
288 			LZ4_copy8(op, match);
289 			match += 8;
290 		}
291 
292 		op += 8;
293 
294 		if (unlikely(cpy > oend - 12)) {
295 			BYTE * const oCopyLimit = oend - (WILDCOPYLENGTH - 1);
296 
297 			if (cpy > oend - LASTLITERALS) {
298 				/*
299 				 * Error : last LASTLITERALS bytes
300 				 * must be literals (uncompressed)
301 				 */
302 				goto _output_error;
303 			}
304 
305 			if (op < oCopyLimit) {
306 				LZ4_wildCopy(op, match, oCopyLimit);
307 				match += oCopyLimit - op;
308 				op = oCopyLimit;
309 			}
310 
311 			while (op < cpy)
312 				*op++ = *match++;
313 		} else {
314 			LZ4_copy8(op, match);
315 
316 			if (length > 16)
317 				LZ4_wildCopy(op + 8, match + 8, cpy);
318 		}
319 
320 		op = cpy; /* correction */
321 	}
322 
323 	/* end of decoding */
324 	if (endOnInput) {
325 		/* Nb of output bytes decoded */
326 		return (int) (((char *)op) - dest);
327 	} else {
328 		/* Nb of input bytes read */
329 		return (int) (((const char *)ip) - source);
330 	}
331 
332 	/* Overflow error detected */
333 _output_error:
334 	return -1;
335 }
336 
337 int LZ4_decompress_safe(const char *source, char *dest,
338 	int compressedSize, int maxDecompressedSize)
339 {
340 	return LZ4_decompress_generic(source, dest, compressedSize,
341 		maxDecompressedSize, endOnInputSize, full, 0,
342 		noDict, (BYTE *)dest, NULL, 0);
343 }
344 
345 int LZ4_decompress_safe_partial(const char *source, char *dest,
346 	int compressedSize, int targetOutputSize, int maxDecompressedSize)
347 {
348 	return LZ4_decompress_generic(source, dest, compressedSize,
349 		maxDecompressedSize, endOnInputSize, partial,
350 		targetOutputSize, noDict, (BYTE *)dest, NULL, 0);
351 }
352 
353 int LZ4_decompress_fast(const char *source, char *dest, int originalSize)
354 {
355 	return LZ4_decompress_generic(source, dest, 0, originalSize,
356 		endOnOutputSize, full, 0, withPrefix64k,
357 		(BYTE *)(dest - 64 * KB), NULL, 64 * KB);
358 }
359 
360 int LZ4_setStreamDecode(LZ4_streamDecode_t *LZ4_streamDecode,
361 	const char *dictionary, int dictSize)
362 {
363 	LZ4_streamDecode_t_internal *lz4sd = (LZ4_streamDecode_t_internal *) LZ4_streamDecode;
364 
365 	lz4sd->prefixSize = (size_t) dictSize;
366 	lz4sd->prefixEnd = (const BYTE *) dictionary + dictSize;
367 	lz4sd->externalDict = NULL;
368 	lz4sd->extDictSize	= 0;
369 	return 1;
370 }
371 
372 /*
373  * *_continue() :
374  * These decoding functions allow decompression of multiple blocks
375  * in "streaming" mode.
376  * Previously decoded blocks must still be available at the memory
377  * position where they were decoded.
378  * If it's not possible, save the relevant part of
379  * decoded data into a safe buffer,
380  * and indicate where it stands using LZ4_setStreamDecode()
381  */
382 int LZ4_decompress_safe_continue(LZ4_streamDecode_t *LZ4_streamDecode,
383 	const char *source, char *dest, int compressedSize, int maxOutputSize)
384 {
385 	LZ4_streamDecode_t_internal *lz4sd = &LZ4_streamDecode->internal_donotuse;
386 	int result;
387 
388 	if (lz4sd->prefixEnd == (BYTE *)dest) {
389 		result = LZ4_decompress_generic(source, dest,
390 			compressedSize,
391 			maxOutputSize,
392 			endOnInputSize, full, 0,
393 			usingExtDict, lz4sd->prefixEnd - lz4sd->prefixSize,
394 			lz4sd->externalDict,
395 			lz4sd->extDictSize);
396 
397 		if (result <= 0)
398 			return result;
399 
400 		lz4sd->prefixSize += result;
401 		lz4sd->prefixEnd	+= result;
402 	} else {
403 		lz4sd->extDictSize = lz4sd->prefixSize;
404 		lz4sd->externalDict = lz4sd->prefixEnd - lz4sd->extDictSize;
405 		result = LZ4_decompress_generic(source, dest,
406 			compressedSize, maxOutputSize,
407 			endOnInputSize, full, 0,
408 			usingExtDict, (BYTE *)dest,
409 			lz4sd->externalDict, lz4sd->extDictSize);
410 		if (result <= 0)
411 			return result;
412 		lz4sd->prefixSize = result;
413 		lz4sd->prefixEnd	= (BYTE *)dest + result;
414 	}
415 
416 	return result;
417 }
418 
419 int LZ4_decompress_fast_continue(LZ4_streamDecode_t *LZ4_streamDecode,
420 	const char *source, char *dest, int originalSize)
421 {
422 	LZ4_streamDecode_t_internal *lz4sd = &LZ4_streamDecode->internal_donotuse;
423 	int result;
424 
425 	if (lz4sd->prefixEnd == (BYTE *)dest) {
426 		result = LZ4_decompress_generic(source, dest, 0, originalSize,
427 			endOnOutputSize, full, 0,
428 			usingExtDict,
429 			lz4sd->prefixEnd - lz4sd->prefixSize,
430 			lz4sd->externalDict, lz4sd->extDictSize);
431 
432 		if (result <= 0)
433 			return result;
434 
435 		lz4sd->prefixSize += originalSize;
436 		lz4sd->prefixEnd	+= originalSize;
437 	} else {
438 		lz4sd->extDictSize = lz4sd->prefixSize;
439 		lz4sd->externalDict = lz4sd->prefixEnd - lz4sd->extDictSize;
440 		result = LZ4_decompress_generic(source, dest, 0, originalSize,
441 			endOnOutputSize, full, 0,
442 			usingExtDict, (BYTE *)dest,
443 			lz4sd->externalDict, lz4sd->extDictSize);
444 		if (result <= 0)
445 			return result;
446 		lz4sd->prefixSize = originalSize;
447 		lz4sd->prefixEnd	= (BYTE *)dest + originalSize;
448 	}
449 
450 	return result;
451 }
452 
453 /*
454  * Advanced decoding functions :
455  * *_usingDict() :
456  * These decoding functions work the same as "_continue" ones,
457  * the dictionary must be explicitly provided within parameters
458  */
459 static FORCE_INLINE int LZ4_decompress_usingDict_generic(const char *source,
460 	char *dest, int compressedSize, int maxOutputSize, int safe,
461 	const char *dictStart, int dictSize)
462 {
463 	if (dictSize == 0)
464 		return LZ4_decompress_generic(source, dest,
465 			compressedSize, maxOutputSize, safe, full, 0,
466 			noDict, (BYTE *)dest, NULL, 0);
467 	if (dictStart + dictSize == dest) {
468 		if (dictSize >= (int)(64 * KB - 1))
469 			return LZ4_decompress_generic(source, dest,
470 				compressedSize, maxOutputSize, safe, full, 0,
471 				withPrefix64k, (BYTE *)dest - 64 * KB, NULL, 0);
472 		return LZ4_decompress_generic(source, dest, compressedSize,
473 			maxOutputSize, safe, full, 0, noDict,
474 			(BYTE *)dest - dictSize, NULL, 0);
475 	}
476 	return LZ4_decompress_generic(source, dest, compressedSize,
477 		maxOutputSize, safe, full, 0, usingExtDict,
478 		(BYTE *)dest, (const BYTE *)dictStart, dictSize);
479 }
480 
481 int LZ4_decompress_safe_usingDict(const char *source, char *dest,
482 	int compressedSize, int maxOutputSize,
483 	const char *dictStart, int dictSize)
484 {
485 	return LZ4_decompress_usingDict_generic(source, dest,
486 		compressedSize, maxOutputSize, 1, dictStart, dictSize);
487 }
488 
489 int LZ4_decompress_fast_usingDict(const char *source, char *dest,
490 	int originalSize, const char *dictStart, int dictSize)
491 {
492 	return LZ4_decompress_usingDict_generic(source, dest, 0,
493 		originalSize, 0, dictStart, dictSize);
494 }
495 
496 #ifndef STATIC
497 EXPORT_SYMBOL(LZ4_decompress_safe);
498 EXPORT_SYMBOL(LZ4_decompress_safe_partial);
499 EXPORT_SYMBOL(LZ4_decompress_fast);
500 EXPORT_SYMBOL(LZ4_setStreamDecode);
501 EXPORT_SYMBOL(LZ4_decompress_safe_continue);
502 EXPORT_SYMBOL(LZ4_decompress_fast_continue);
503 EXPORT_SYMBOL(LZ4_decompress_safe_usingDict);
504 EXPORT_SYMBOL(LZ4_decompress_fast_usingDict);
505 
506 MODULE_LICENSE("Dual BSD/GPL");
507 MODULE_DESCRIPTION("LZ4 decompressor");
508 #endif
509