1 /*
2 LZ4 - Fast LZ compression algorithm
3 Copyright (C) 2011-2020, Yann Collet.
4
5 BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
6
7 Redistribution and use in source and binary forms, with or without
8 modification, are permitted provided that the following conditions are
9 met:
10
11 * Redistributions of source code must retain the above copyright
12 notice, this list of conditions and the following disclaimer.
13 * Redistributions in binary form must reproduce the above
14 copyright notice, this list of conditions and the following disclaimer
15 in the documentation and/or other materials provided with the
16 distribution.
17
18 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
19 "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
20 LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
21 A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
22 OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
23 SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
24 LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
25 DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
26 THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
27 (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
28 OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
29
30 You can contact the author at :
31 - LZ4 homepage : http://www.lz4.org
32 - LZ4 source repository : https://github.com/lz4/lz4
33 */
34
35 /*-************************************
36 * Tuning parameters
37 **************************************/
38 /*
39 * LZ4_HEAPMODE :
40 * Select how default compression functions will allocate memory for their hash table,
41 * in memory stack (0:default, fastest), or in memory heap (1:requires malloc()).
42 */
43 #ifndef LZ4_HEAPMODE
44 # define LZ4_HEAPMODE 0
45 #endif
46
47 /*
48 * LZ4_ACCELERATION_DEFAULT :
49 * Select "acceleration" for LZ4_compress_fast() when parameter value <= 0
50 */
51 #define LZ4_ACCELERATION_DEFAULT 1
52 /*
53 * LZ4_ACCELERATION_MAX :
54 * Any "acceleration" value higher than this threshold
55 * get treated as LZ4_ACCELERATION_MAX instead (fix #876)
56 */
57 #define LZ4_ACCELERATION_MAX 65537
58
59
60 /*-************************************
61 * CPU Feature Detection
62 **************************************/
63 /* LZ4_FORCE_MEMORY_ACCESS
64 * By default, access to unaligned memory is controlled by `memcpy()`, which is safe and portable.
65 * Unfortunately, on some target/compiler combinations, the generated assembly is sub-optimal.
66 * The below switch allow to select different access method for improved performance.
67 * Method 0 (default) : use `memcpy()`. Safe and portable.
68 * Method 1 : `__packed` statement. It depends on compiler extension (ie, not portable).
69 * This method is safe if your compiler supports it, and *generally* as fast or faster than `memcpy`.
70 * Method 2 : direct access. This method is portable but violate C standard.
71 * It can generate buggy code on targets which assembly generation depends on alignment.
72 * But in some circumstances, it's the only known way to get the most performance (ie GCC + ARMv6)
73 * See https://fastcompression.blogspot.fr/2015/08/accessing-unaligned-memory.html for details.
74 * Prefer these methods in priority order (0 > 1 > 2)
75 */
76 #ifndef LZ4_FORCE_MEMORY_ACCESS /* can be defined externally */
77 # if defined(__GNUC__) && \
78 ( defined(__ARM_ARCH_6__) || defined(__ARM_ARCH_6J__) || defined(__ARM_ARCH_6K__) \
79 || defined(__ARM_ARCH_6Z__) || defined(__ARM_ARCH_6ZK__) || defined(__ARM_ARCH_6T2__) )
80 # define LZ4_FORCE_MEMORY_ACCESS 2
81 # elif (defined(__INTEL_COMPILER) && !defined(_WIN32)) || defined(__GNUC__)
82 # define LZ4_FORCE_MEMORY_ACCESS 1
83 # endif
84 #endif
85
86 /*
87 * LZ4_FORCE_SW_BITCOUNT
88 * Define this parameter if your target system or compiler does not support hardware bit count
89 */
90 #if defined(_MSC_VER) && defined(_WIN32_WCE) /* Visual Studio for WinCE doesn't support Hardware bit count */
91 # undef LZ4_FORCE_SW_BITCOUNT /* avoid double def */
92 # define LZ4_FORCE_SW_BITCOUNT
93 #endif
94
95
96
97 /*-************************************
98 * Dependency
99 **************************************/
100 /*
101 * LZ4_SRC_INCLUDED:
102 * Amalgamation flag, whether lz4.c is included
103 */
104 #ifndef LZ4_SRC_INCLUDED
105 # define LZ4_SRC_INCLUDED 1
106 #endif
107
108 #ifndef LZ4_STATIC_LINKING_ONLY
109 #define LZ4_STATIC_LINKING_ONLY
110 #endif
111
112 #ifndef LZ4_DISABLE_DEPRECATE_WARNINGS
113 #define LZ4_DISABLE_DEPRECATE_WARNINGS /* due to LZ4_decompress_safe_withPrefix64k */
114 #endif
115
116 #define LZ4_STATIC_LINKING_ONLY /* LZ4_DISTANCE_MAX */
117 #include "lz4.h"
118 /* see also "memory routines" below */
119
120
121 /*-************************************
122 * Compiler Options
123 **************************************/
124 #if defined(_MSC_VER) && (_MSC_VER >= 1400) /* Visual Studio 2005+ */
125 # include <intrin.h> /* only present in VS2005+ */
126 # pragma warning(disable : 4127) /* disable: C4127: conditional expression is constant */
127 # pragma warning(disable : 6237) /* disable: C6237: conditional expression is always 0 */
128 #endif /* _MSC_VER */
129
130 #ifndef LZ4_FORCE_INLINE
131 # ifdef _MSC_VER /* Visual Studio */
132 # define LZ4_FORCE_INLINE static __forceinline
133 # else
134 # if defined (__cplusplus) || defined (__STDC_VERSION__) && __STDC_VERSION__ >= 199901L /* C99 */
135 # ifdef __GNUC__
136 # define LZ4_FORCE_INLINE static inline __attribute__((always_inline))
137 # else
138 # define LZ4_FORCE_INLINE static inline
139 # endif
140 # else
141 # define LZ4_FORCE_INLINE static
142 # endif /* __STDC_VERSION__ */
143 # endif /* _MSC_VER */
144 #endif /* LZ4_FORCE_INLINE */
145
146 /* LZ4_FORCE_O2 and LZ4_FORCE_INLINE
147 * gcc on ppc64le generates an unrolled SIMDized loop for LZ4_wildCopy8,
148 * together with a simple 8-byte copy loop as a fall-back path.
149 * However, this optimization hurts the decompression speed by >30%,
150 * because the execution does not go to the optimized loop
151 * for typical compressible data, and all of the preamble checks
152 * before going to the fall-back path become useless overhead.
153 * This optimization happens only with the -O3 flag, and -O2 generates
154 * a simple 8-byte copy loop.
155 * With gcc on ppc64le, all of the LZ4_decompress_* and LZ4_wildCopy8
156 * functions are annotated with __attribute__((optimize("O2"))),
157 * and also LZ4_wildCopy8 is forcibly inlined, so that the O2 attribute
158 * of LZ4_wildCopy8 does not affect the compression speed.
159 */
160 #if defined(__PPC64__) && defined(__LITTLE_ENDIAN__) && defined(__GNUC__) && !defined(__clang__)
161 # define LZ4_FORCE_O2 __attribute__((optimize("O2")))
162 # undef LZ4_FORCE_INLINE
163 # define LZ4_FORCE_INLINE static __inline __attribute__((optimize("O2"),always_inline))
164 #else
165 # define LZ4_FORCE_O2
166 #endif
167
168 #if (defined(__GNUC__) && (__GNUC__ >= 3)) || (defined(__INTEL_COMPILER) && (__INTEL_COMPILER >= 800)) || defined(__clang__)
169 # define expect(expr,value) (__builtin_expect ((expr),(value)) )
170 #else
171 # define expect(expr,value) (expr)
172 #endif
173
174 #ifndef likely
175 #define likely(expr) expect((expr) != 0, 1)
176 #endif
177 #ifndef unlikely
178 #define unlikely(expr) expect((expr) != 0, 0)
179 #endif
180
181 /* Should the alignment test prove unreliable, for some reason,
182 * it can be disabled by setting LZ4_ALIGN_TEST to 0 */
183 #ifndef LZ4_ALIGN_TEST /* can be externally provided */
184 # define LZ4_ALIGN_TEST 1
185 #endif
186
187
188 /*-************************************
189 * Memory routines
190 **************************************/
191
192 /*! LZ4_STATIC_LINKING_ONLY_DISABLE_MEMORY_ALLOCATION :
193 * Disable relatively high-level LZ4/HC functions that use dynamic memory
194 * allocation functions (malloc(), calloc(), free()).
195 *
196 * Note that this is a compile-time switch. And since it disables
197 * public/stable LZ4 v1 API functions, we don't recommend using this
198 * symbol to generate a library for distribution.
199 *
200 * The following public functions are removed when this symbol is defined.
201 * - lz4 : LZ4_createStream, LZ4_freeStream,
202 * LZ4_createStreamDecode, LZ4_freeStreamDecode, LZ4_create (deprecated)
203 * - lz4hc : LZ4_createStreamHC, LZ4_freeStreamHC,
204 * LZ4_createHC (deprecated), LZ4_freeHC (deprecated)
205 * - lz4frame, lz4file : All LZ4F_* functions
206 */
207 #if defined(LZ4_STATIC_LINKING_ONLY_DISABLE_MEMORY_ALLOCATION)
208 # define ALLOC(s) lz4_error_memory_allocation_is_disabled
209 # define ALLOC_AND_ZERO(s) lz4_error_memory_allocation_is_disabled
210 # define FREEMEM(p) lz4_error_memory_allocation_is_disabled
211 #elif defined(LZ4_USER_MEMORY_FUNCTIONS)
212 /* memory management functions can be customized by user project.
213 * Below functions must exist somewhere in the Project
214 * and be available at link time */
215 void* LZ4_malloc(size_t s);
216 void* LZ4_calloc(size_t n, size_t s);
217 void LZ4_free(void* p);
218 # define ALLOC(s) LZ4_malloc(s)
219 # define ALLOC_AND_ZERO(s) LZ4_calloc(1,s)
220 # define FREEMEM(p) LZ4_free(p)
221 #else
222 # include <stdlib.h> /* malloc, calloc, free */
223 # define ALLOC(s) malloc(s)
224 # define ALLOC_AND_ZERO(s) calloc(1,s)
225 # define FREEMEM(p) free(p)
226 #endif
227
228 #if ! LZ4_FREESTANDING
229 # include <string.h> /* memset, memcpy */
230 #endif
231 #if !defined(LZ4_memset)
232 # define LZ4_memset(p,v,s) memset((p),(v),(s))
233 #endif
234 #define MEM_INIT(p,v,s) LZ4_memset((p),(v),(s))
235
236
237 /*-************************************
238 * Common Constants
239 **************************************/
240 #define MINMATCH 4
241
242 #define WILDCOPYLENGTH 8
243 #define LASTLITERALS 5 /* see ../doc/lz4_Block_format.md#parsing-restrictions */
244 #define MFLIMIT 12 /* see ../doc/lz4_Block_format.md#parsing-restrictions */
245 #define MATCH_SAFEGUARD_DISTANCE ((2*WILDCOPYLENGTH) - MINMATCH) /* ensure it's possible to write 2 x wildcopyLength without overflowing output buffer */
246 #define FASTLOOP_SAFE_DISTANCE 64
247 static const int LZ4_minLength = (MFLIMIT+1);
248
249 #define KB *(1 <<10)
250 #define MB *(1 <<20)
251 #define GB *(1U<<30)
252
253 #define LZ4_DISTANCE_ABSOLUTE_MAX 65535
254 #if (LZ4_DISTANCE_MAX > LZ4_DISTANCE_ABSOLUTE_MAX) /* max supported by LZ4 format */
255 # error "LZ4_DISTANCE_MAX is too big : must be <= 65535"
256 #endif
257
258 #define ML_BITS 4
259 #define ML_MASK ((1U<<ML_BITS)-1)
260 #define RUN_BITS (8-ML_BITS)
261 #define RUN_MASK ((1U<<RUN_BITS)-1)
262
263
264 /*-************************************
265 * Error detection
266 **************************************/
267 #if defined(LZ4_DEBUG) && (LZ4_DEBUG>=1)
268 # include <assert.h>
269 #else
270 # ifndef assert
271 # define assert(condition) ((void)0)
272 # endif
273 #endif
274
275 #define LZ4_STATIC_ASSERT(c) { enum { LZ4_static_assert = 1/(int)(!!(c)) }; } /* use after variable declarations */
276
277 #if defined(LZ4_DEBUG) && (LZ4_DEBUG>=2)
278 # include <stdio.h>
279 static int g_debuglog_enable = 1;
280 # define DEBUGLOG(l, ...) { \
281 if ((g_debuglog_enable) && (l<=LZ4_DEBUG)) { \
282 fprintf(stderr, __FILE__ ": "); \
283 fprintf(stderr, __VA_ARGS__); \
284 fprintf(stderr, " \n"); \
285 } }
286 #else
287 # define DEBUGLOG(l, ...) {} /* disabled */
288 #endif
289
LZ4_isAligned(const void * ptr,size_t alignment)290 static int LZ4_isAligned(const void* ptr, size_t alignment)
291 {
292 return ((size_t)ptr & (alignment -1)) == 0;
293 }
294
295
296 /*-************************************
297 * Types
298 **************************************/
299 #include <limits.h>
300 #if defined(__cplusplus) || (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */)
301 # include <stdint.h>
302 typedef uint8_t BYTE;
303 typedef uint16_t U16;
304 typedef uint32_t U32;
305 typedef int32_t S32;
306 typedef uint64_t U64;
307 typedef uintptr_t uptrval;
308 #else
309 # if UINT_MAX != 4294967295UL
310 # error "LZ4 code (when not C++ or C99) assumes that sizeof(int) == 4"
311 # endif
312 typedef unsigned char BYTE;
313 typedef unsigned short U16;
314 typedef unsigned int U32;
315 typedef signed int S32;
316 typedef unsigned long long U64;
317 typedef size_t uptrval; /* generally true, except OpenVMS-64 */
318 #endif
319
320 #if defined(__x86_64__)
321 typedef U64 reg_t; /* 64-bits in x32 mode */
322 #else
323 typedef size_t reg_t; /* 32-bits in x32 mode */
324 #endif
325
326 typedef enum {
327 notLimited = 0,
328 limitedOutput = 1,
329 fillOutput = 2
330 } limitedOutput_directive;
331
332
333 /*-************************************
334 * Reading and writing into memory
335 **************************************/
336
337 /**
338 * LZ4 relies on memcpy with a constant size being inlined. In freestanding
339 * environments, the compiler can't assume the implementation of memcpy() is
340 * standard compliant, so it can't apply its specialized memcpy() inlining
341 * logic. When possible, use __builtin_memcpy() to tell the compiler to analyze
342 * memcpy() as if it were standard compliant, so it can inline it in freestanding
343 * environments. This is needed when decompressing the Linux Kernel, for example.
344 */
345 #if !defined(LZ4_memcpy)
346 # if defined(__GNUC__) && (__GNUC__ >= 4)
347 # define LZ4_memcpy(dst, src, size) __builtin_memcpy(dst, src, size)
348 # else
349 # define LZ4_memcpy(dst, src, size) memcpy(dst, src, size)
350 # endif
351 #endif
352
353 #if !defined(LZ4_memmove)
354 # if defined(__GNUC__) && (__GNUC__ >= 4)
355 # define LZ4_memmove __builtin_memmove
356 # else
357 # define LZ4_memmove memmove
358 # endif
359 #endif
360
LZ4_isLittleEndian(void)361 static unsigned LZ4_isLittleEndian(void)
362 {
363 const union { U32 u; BYTE c[4]; } one = { 1 }; /* don't use static : performance detrimental */
364 return one.c[0];
365 }
366
367
368 #if defined(LZ4_FORCE_MEMORY_ACCESS) && (LZ4_FORCE_MEMORY_ACCESS==2)
369 /* lie to the compiler about data alignment; use with caution */
370
LZ4_read16(const void * memPtr)371 static U16 LZ4_read16(const void* memPtr) { return *(const U16*) memPtr; }
LZ4_read32(const void * memPtr)372 static U32 LZ4_read32(const void* memPtr) { return *(const U32*) memPtr; }
LZ4_read_ARCH(const void * memPtr)373 static reg_t LZ4_read_ARCH(const void* memPtr) { return *(const reg_t*) memPtr; }
374
LZ4_write16(void * memPtr,U16 value)375 static void LZ4_write16(void* memPtr, U16 value) { *(U16*)memPtr = value; }
LZ4_write32(void * memPtr,U32 value)376 static void LZ4_write32(void* memPtr, U32 value) { *(U32*)memPtr = value; }
377
378 #elif defined(LZ4_FORCE_MEMORY_ACCESS) && (LZ4_FORCE_MEMORY_ACCESS==1)
379
380 /* __pack instructions are safer, but compiler specific, hence potentially problematic for some compilers */
381 /* currently only defined for gcc and icc */
382 typedef union { U16 u16; U32 u32; reg_t uArch; } __attribute__((packed)) LZ4_unalign;
383
LZ4_read16(const void * ptr)384 static U16 LZ4_read16(const void* ptr) { return ((const LZ4_unalign*)ptr)->u16; }
LZ4_read32(const void * ptr)385 static U32 LZ4_read32(const void* ptr) { return ((const LZ4_unalign*)ptr)->u32; }
LZ4_read_ARCH(const void * ptr)386 static reg_t LZ4_read_ARCH(const void* ptr) { return ((const LZ4_unalign*)ptr)->uArch; }
387
LZ4_write16(void * memPtr,U16 value)388 static void LZ4_write16(void* memPtr, U16 value) { ((LZ4_unalign*)memPtr)->u16 = value; }
LZ4_write32(void * memPtr,U32 value)389 static void LZ4_write32(void* memPtr, U32 value) { ((LZ4_unalign*)memPtr)->u32 = value; }
390
391 #else /* safe and portable access using memcpy() */
392
LZ4_read16(const void * memPtr)393 static U16 LZ4_read16(const void* memPtr)
394 {
395 U16 val; LZ4_memcpy(&val, memPtr, sizeof(val)); return val;
396 }
397
LZ4_read32(const void * memPtr)398 static U32 LZ4_read32(const void* memPtr)
399 {
400 U32 val; LZ4_memcpy(&val, memPtr, sizeof(val)); return val;
401 }
402
LZ4_read_ARCH(const void * memPtr)403 static reg_t LZ4_read_ARCH(const void* memPtr)
404 {
405 reg_t val; LZ4_memcpy(&val, memPtr, sizeof(val)); return val;
406 }
407
LZ4_write16(void * memPtr,U16 value)408 static void LZ4_write16(void* memPtr, U16 value)
409 {
410 LZ4_memcpy(memPtr, &value, sizeof(value));
411 }
412
LZ4_write32(void * memPtr,U32 value)413 static void LZ4_write32(void* memPtr, U32 value)
414 {
415 LZ4_memcpy(memPtr, &value, sizeof(value));
416 }
417
418 #endif /* LZ4_FORCE_MEMORY_ACCESS */
419
420
LZ4_readLE16(const void * memPtr)421 static U16 LZ4_readLE16(const void* memPtr)
422 {
423 if (LZ4_isLittleEndian()) {
424 return LZ4_read16(memPtr);
425 } else {
426 const BYTE* p = (const BYTE*)memPtr;
427 return (U16)((U16)p[0] + (p[1]<<8));
428 }
429 }
430
LZ4_writeLE16(void * memPtr,U16 value)431 static void LZ4_writeLE16(void* memPtr, U16 value)
432 {
433 if (LZ4_isLittleEndian()) {
434 LZ4_write16(memPtr, value);
435 } else {
436 BYTE* p = (BYTE*)memPtr;
437 p[0] = (BYTE) value;
438 p[1] = (BYTE)(value>>8);
439 }
440 }
441
442 /* customized variant of memcpy, which can overwrite up to 8 bytes beyond dstEnd */
443 LZ4_FORCE_INLINE
LZ4_wildCopy8(void * dstPtr,const void * srcPtr,void * dstEnd)444 void LZ4_wildCopy8(void* dstPtr, const void* srcPtr, void* dstEnd)
445 {
446 BYTE* d = (BYTE*)dstPtr;
447 const BYTE* s = (const BYTE*)srcPtr;
448 BYTE* const e = (BYTE*)dstEnd;
449
450 do { LZ4_memcpy(d,s,8); d+=8; s+=8; } while (d<e);
451 }
452
453 static const unsigned inc32table[8] = {0, 1, 2, 1, 0, 4, 4, 4};
454 static const int dec64table[8] = {0, 0, 0, -1, -4, 1, 2, 3};
455
456
457 #ifndef LZ4_FAST_DEC_LOOP
458 # if defined __i386__ || defined _M_IX86 || defined __x86_64__ || defined _M_X64
459 # define LZ4_FAST_DEC_LOOP 1
460 # elif defined(__aarch64__) && defined(__APPLE__)
461 # define LZ4_FAST_DEC_LOOP 1
462 # elif defined(__aarch64__) && !defined(__clang__)
463 /* On non-Apple aarch64, we disable this optimization for clang because
464 * on certain mobile chipsets, performance is reduced with clang. For
465 * more information refer to https://github.com/lz4/lz4/pull/707 */
466 # define LZ4_FAST_DEC_LOOP 1
467 # else
468 # define LZ4_FAST_DEC_LOOP 0
469 # endif
470 #endif
471
472 #if LZ4_FAST_DEC_LOOP
473
474 LZ4_FORCE_INLINE void
LZ4_memcpy_using_offset_base(BYTE * dstPtr,const BYTE * srcPtr,BYTE * dstEnd,const size_t offset)475 LZ4_memcpy_using_offset_base(BYTE* dstPtr, const BYTE* srcPtr, BYTE* dstEnd, const size_t offset)
476 {
477 assert(srcPtr + offset == dstPtr);
478 if (offset < 8) {
479 LZ4_write32(dstPtr, 0); /* silence an msan warning when offset==0 */
480 dstPtr[0] = srcPtr[0];
481 dstPtr[1] = srcPtr[1];
482 dstPtr[2] = srcPtr[2];
483 dstPtr[3] = srcPtr[3];
484 srcPtr += inc32table[offset];
485 LZ4_memcpy(dstPtr+4, srcPtr, 4);
486 srcPtr -= dec64table[offset];
487 dstPtr += 8;
488 } else {
489 LZ4_memcpy(dstPtr, srcPtr, 8);
490 dstPtr += 8;
491 srcPtr += 8;
492 }
493
494 LZ4_wildCopy8(dstPtr, srcPtr, dstEnd);
495 }
496
497 /* customized variant of memcpy, which can overwrite up to 32 bytes beyond dstEnd
498 * this version copies two times 16 bytes (instead of one time 32 bytes)
499 * because it must be compatible with offsets >= 16. */
500 LZ4_FORCE_INLINE void
LZ4_wildCopy32(void * dstPtr,const void * srcPtr,void * dstEnd)501 LZ4_wildCopy32(void* dstPtr, const void* srcPtr, void* dstEnd)
502 {
503 BYTE* d = (BYTE*)dstPtr;
504 const BYTE* s = (const BYTE*)srcPtr;
505 BYTE* const e = (BYTE*)dstEnd;
506
507 do { LZ4_memcpy(d,s,16); LZ4_memcpy(d+16,s+16,16); d+=32; s+=32; } while (d<e);
508 }
509
510 /* LZ4_memcpy_using_offset() presumes :
511 * - dstEnd >= dstPtr + MINMATCH
512 * - there is at least 8 bytes available to write after dstEnd */
513 LZ4_FORCE_INLINE void
LZ4_memcpy_using_offset(BYTE * dstPtr,const BYTE * srcPtr,BYTE * dstEnd,const size_t offset)514 LZ4_memcpy_using_offset(BYTE* dstPtr, const BYTE* srcPtr, BYTE* dstEnd, const size_t offset)
515 {
516 BYTE v[8];
517
518 assert(dstEnd >= dstPtr + MINMATCH);
519
520 switch(offset) {
521 case 1:
522 MEM_INIT(v, *srcPtr, 8);
523 break;
524 case 2:
525 LZ4_memcpy(v, srcPtr, 2);
526 LZ4_memcpy(&v[2], srcPtr, 2);
527 #if defined(_MSC_VER) && (_MSC_VER <= 1933) /* MSVC 2022 ver 17.3 or earlier */
528 # pragma warning(push)
529 # pragma warning(disable : 6385) /* warning C6385: Reading invalid data from 'v'. */
530 #endif
531 LZ4_memcpy(&v[4], v, 4);
532 #if defined(_MSC_VER) && (_MSC_VER <= 1933) /* MSVC 2022 ver 17.3 or earlier */
533 # pragma warning(pop)
534 #endif
535 break;
536 case 4:
537 LZ4_memcpy(v, srcPtr, 4);
538 LZ4_memcpy(&v[4], srcPtr, 4);
539 break;
540 default:
541 LZ4_memcpy_using_offset_base(dstPtr, srcPtr, dstEnd, offset);
542 return;
543 }
544
545 LZ4_memcpy(dstPtr, v, 8);
546 dstPtr += 8;
547 while (dstPtr < dstEnd) {
548 LZ4_memcpy(dstPtr, v, 8);
549 dstPtr += 8;
550 }
551 }
552 #endif
553
554
555 /*-************************************
556 * Common functions
557 **************************************/
LZ4_NbCommonBytes(reg_t val)558 static unsigned LZ4_NbCommonBytes (reg_t val)
559 {
560 assert(val != 0);
561 if (LZ4_isLittleEndian()) {
562 if (sizeof(val) == 8) {
563 # if defined(_MSC_VER) && (_MSC_VER >= 1800) && (defined(_M_AMD64) && !defined(_M_ARM64EC)) && !defined(LZ4_FORCE_SW_BITCOUNT)
564 /*-*************************************************************************************************
565 * ARM64EC is a Microsoft-designed ARM64 ABI compatible with AMD64 applications on ARM64 Windows 11.
566 * The ARM64EC ABI does not support AVX/AVX2/AVX512 instructions, nor their relevant intrinsics
567 * including _tzcnt_u64. Therefore, we need to neuter the _tzcnt_u64 code path for ARM64EC.
568 ****************************************************************************************************/
569 # if defined(__clang__) && (__clang_major__ < 10)
570 /* Avoid undefined clang-cl intrinsics issue.
571 * See https://github.com/lz4/lz4/pull/1017 for details. */
572 return (unsigned)__builtin_ia32_tzcnt_u64(val) >> 3;
573 # else
574 /* x64 CPUS without BMI support interpret `TZCNT` as `REP BSF` */
575 return (unsigned)_tzcnt_u64(val) >> 3;
576 # endif
577 # elif defined(_MSC_VER) && defined(_WIN64) && !defined(LZ4_FORCE_SW_BITCOUNT)
578 unsigned long r = 0;
579 _BitScanForward64(&r, (U64)val);
580 return (unsigned)r >> 3;
581 # elif (defined(__clang__) || (defined(__GNUC__) && ((__GNUC__ > 3) || \
582 ((__GNUC__ == 3) && (__GNUC_MINOR__ >= 4))))) && \
583 !defined(LZ4_FORCE_SW_BITCOUNT)
584 return (unsigned)__builtin_ctzll((U64)val) >> 3;
585 # else
586 const U64 m = 0x0101010101010101ULL;
587 val ^= val - 1;
588 return (unsigned)(((U64)((val & (m - 1)) * m)) >> 56);
589 # endif
590 } else /* 32 bits */ {
591 # if defined(_MSC_VER) && (_MSC_VER >= 1400) && !defined(LZ4_FORCE_SW_BITCOUNT)
592 unsigned long r;
593 _BitScanForward(&r, (U32)val);
594 return (unsigned)r >> 3;
595 # elif (defined(__clang__) || (defined(__GNUC__) && ((__GNUC__ > 3) || \
596 ((__GNUC__ == 3) && (__GNUC_MINOR__ >= 4))))) && \
597 !defined(__TINYC__) && !defined(LZ4_FORCE_SW_BITCOUNT)
598 return (unsigned)__builtin_ctz((U32)val) >> 3;
599 # else
600 const U32 m = 0x01010101;
601 return (unsigned)((((val - 1) ^ val) & (m - 1)) * m) >> 24;
602 # endif
603 }
604 } else /* Big Endian CPU */ {
605 if (sizeof(val)==8) {
606 # if (defined(__clang__) || (defined(__GNUC__) && ((__GNUC__ > 3) || \
607 ((__GNUC__ == 3) && (__GNUC_MINOR__ >= 4))))) && \
608 !defined(__TINYC__) && !defined(LZ4_FORCE_SW_BITCOUNT)
609 return (unsigned)__builtin_clzll((U64)val) >> 3;
610 # else
611 #if 1
612 /* this method is probably faster,
613 * but adds a 128 bytes lookup table */
614 static const unsigned char ctz7_tab[128] = {
615 7, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0,
616 4, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0,
617 5, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0,
618 4, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0,
619 6, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0,
620 4, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0,
621 5, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0,
622 4, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0,
623 };
624 U64 const mask = 0x0101010101010101ULL;
625 U64 const t = (((val >> 8) - mask) | val) & mask;
626 return ctz7_tab[(t * 0x0080402010080402ULL) >> 57];
627 #else
628 /* this method doesn't consume memory space like the previous one,
629 * but it contains several branches,
630 * that may end up slowing execution */
631 static const U32 by32 = sizeof(val)*4; /* 32 on 64 bits (goal), 16 on 32 bits.
632 Just to avoid some static analyzer complaining about shift by 32 on 32-bits target.
633 Note that this code path is never triggered in 32-bits mode. */
634 unsigned r;
635 if (!(val>>by32)) { r=4; } else { r=0; val>>=by32; }
636 if (!(val>>16)) { r+=2; val>>=8; } else { val>>=24; }
637 r += (!val);
638 return r;
639 #endif
640 # endif
641 } else /* 32 bits */ {
642 # if (defined(__clang__) || (defined(__GNUC__) && ((__GNUC__ > 3) || \
643 ((__GNUC__ == 3) && (__GNUC_MINOR__ >= 4))))) && \
644 !defined(LZ4_FORCE_SW_BITCOUNT)
645 return (unsigned)__builtin_clz((U32)val) >> 3;
646 # else
647 val >>= 8;
648 val = ((((val + 0x00FFFF00) | 0x00FFFFFF) + val) |
649 (val + 0x00FF0000)) >> 24;
650 return (unsigned)val ^ 3;
651 # endif
652 }
653 }
654 }
655
656
657 #define STEPSIZE sizeof(reg_t)
658 LZ4_FORCE_INLINE
LZ4_count(const BYTE * pIn,const BYTE * pMatch,const BYTE * pInLimit)659 unsigned LZ4_count(const BYTE* pIn, const BYTE* pMatch, const BYTE* pInLimit)
660 {
661 const BYTE* const pStart = pIn;
662
663 if (likely(pIn < pInLimit-(STEPSIZE-1))) {
664 reg_t const diff = LZ4_read_ARCH(pMatch) ^ LZ4_read_ARCH(pIn);
665 if (!diff) {
666 pIn+=STEPSIZE; pMatch+=STEPSIZE;
667 } else {
668 return LZ4_NbCommonBytes(diff);
669 } }
670
671 while (likely(pIn < pInLimit-(STEPSIZE-1))) {
672 reg_t const diff = LZ4_read_ARCH(pMatch) ^ LZ4_read_ARCH(pIn);
673 if (!diff) { pIn+=STEPSIZE; pMatch+=STEPSIZE; continue; }
674 pIn += LZ4_NbCommonBytes(diff);
675 return (unsigned)(pIn - pStart);
676 }
677
678 if ((STEPSIZE==8) && (pIn<(pInLimit-3)) && (LZ4_read32(pMatch) == LZ4_read32(pIn))) { pIn+=4; pMatch+=4; }
679 if ((pIn<(pInLimit-1)) && (LZ4_read16(pMatch) == LZ4_read16(pIn))) { pIn+=2; pMatch+=2; }
680 if ((pIn<pInLimit) && (*pMatch == *pIn)) pIn++;
681 return (unsigned)(pIn - pStart);
682 }
683
684
685 #ifndef LZ4_COMMONDEFS_ONLY
686 /*-************************************
687 * Local Constants
688 **************************************/
689 static const int LZ4_64Klimit = ((64 KB) + (MFLIMIT-1));
690 static const U32 LZ4_skipTrigger = 6; /* Increase this value ==> compression run slower on incompressible data */
691
692
693 /*-************************************
694 * Local Structures and types
695 **************************************/
696 typedef enum { clearedTable = 0, byPtr, byU32, byU16 } tableType_t;
697
698 /**
699 * This enum distinguishes several different modes of accessing previous
700 * content in the stream.
701 *
702 * - noDict : There is no preceding content.
703 * - withPrefix64k : Table entries up to ctx->dictSize before the current blob
704 * blob being compressed are valid and refer to the preceding
705 * content (of length ctx->dictSize), which is available
706 * contiguously preceding in memory the content currently
707 * being compressed.
708 * - usingExtDict : Like withPrefix64k, but the preceding content is somewhere
709 * else in memory, starting at ctx->dictionary with length
710 * ctx->dictSize.
711 * - usingDictCtx : Everything concerning the preceding content is
712 * in a separate context, pointed to by ctx->dictCtx.
713 * ctx->dictionary, ctx->dictSize, and table entries
714 * in the current context that refer to positions
715 * preceding the beginning of the current compression are
716 * ignored. Instead, ctx->dictCtx->dictionary and ctx->dictCtx
717 * ->dictSize describe the location and size of the preceding
718 * content, and matches are found by looking in the ctx
719 * ->dictCtx->hashTable.
720 */
721 typedef enum { noDict = 0, withPrefix64k, usingExtDict, usingDictCtx } dict_directive;
722 typedef enum { noDictIssue = 0, dictSmall } dictIssue_directive;
723
724
725 /*-************************************
726 * Local Utils
727 **************************************/
LZ4_versionNumber(void)728 int LZ4_versionNumber (void) { return LZ4_VERSION_NUMBER; }
LZ4_versionString(void)729 const char* LZ4_versionString(void) { return LZ4_VERSION_STRING; }
LZ4_compressBound(int isize)730 int LZ4_compressBound(int isize) { return LZ4_COMPRESSBOUND(isize); }
LZ4_sizeofState(void)731 int LZ4_sizeofState(void) { return sizeof(LZ4_stream_t); }
732
733
734 /*-****************************************
735 * Internal Definitions, used only in Tests
736 *******************************************/
737 #if defined (__cplusplus)
738 extern "C" {
739 #endif
740
741 int LZ4_compress_forceExtDict (LZ4_stream_t* LZ4_dict, const char* source, char* dest, int srcSize);
742
743 int LZ4_decompress_safe_forceExtDict(const char* source, char* dest,
744 int compressedSize, int maxOutputSize,
745 const void* dictStart, size_t dictSize);
746 int LZ4_decompress_safe_partial_forceExtDict(const char* source, char* dest,
747 int compressedSize, int targetOutputSize, int dstCapacity,
748 const void* dictStart, size_t dictSize);
749 #if defined (__cplusplus)
750 }
751 #endif
752
753 /*-******************************
754 * Compression functions
755 ********************************/
LZ4_hash4(U32 sequence,tableType_t const tableType)756 LZ4_FORCE_INLINE U32 LZ4_hash4(U32 sequence, tableType_t const tableType)
757 {
758 if (tableType == byU16)
759 return ((sequence * 2654435761U) >> ((MINMATCH*8)-(LZ4_HASHLOG+1)));
760 else
761 return ((sequence * 2654435761U) >> ((MINMATCH*8)-LZ4_HASHLOG));
762 }
763
LZ4_hash5(U64 sequence,tableType_t const tableType)764 LZ4_FORCE_INLINE U32 LZ4_hash5(U64 sequence, tableType_t const tableType)
765 {
766 const U32 hashLog = (tableType == byU16) ? LZ4_HASHLOG+1 : LZ4_HASHLOG;
767 if (LZ4_isLittleEndian()) {
768 const U64 prime5bytes = 889523592379ULL;
769 return (U32)(((sequence << 24) * prime5bytes) >> (64 - hashLog));
770 } else {
771 const U64 prime8bytes = 11400714785074694791ULL;
772 return (U32)(((sequence >> 24) * prime8bytes) >> (64 - hashLog));
773 }
774 }
775
LZ4_hashPosition(const void * const p,tableType_t const tableType)776 LZ4_FORCE_INLINE U32 LZ4_hashPosition(const void* const p, tableType_t const tableType)
777 {
778 if ((sizeof(reg_t)==8) && (tableType != byU16)) return LZ4_hash5(LZ4_read_ARCH(p), tableType);
779 return LZ4_hash4(LZ4_read32(p), tableType);
780 }
781
LZ4_clearHash(U32 h,void * tableBase,tableType_t const tableType)782 LZ4_FORCE_INLINE void LZ4_clearHash(U32 h, void* tableBase, tableType_t const tableType)
783 {
784 switch (tableType)
785 {
786 default: /* fallthrough */
787 case clearedTable: { /* illegal! */ assert(0); return; }
788 case byPtr: { const BYTE** hashTable = (const BYTE**)tableBase; hashTable[h] = NULL; return; }
789 case byU32: { U32* hashTable = (U32*) tableBase; hashTable[h] = 0; return; }
790 case byU16: { U16* hashTable = (U16*) tableBase; hashTable[h] = 0; return; }
791 }
792 }
793
LZ4_putIndexOnHash(U32 idx,U32 h,void * tableBase,tableType_t const tableType)794 LZ4_FORCE_INLINE void LZ4_putIndexOnHash(U32 idx, U32 h, void* tableBase, tableType_t const tableType)
795 {
796 switch (tableType)
797 {
798 default: /* fallthrough */
799 case clearedTable: /* fallthrough */
800 case byPtr: { /* illegal! */ assert(0); return; }
801 case byU32: { U32* hashTable = (U32*) tableBase; hashTable[h] = idx; return; }
802 case byU16: { U16* hashTable = (U16*) tableBase; assert(idx < 65536); hashTable[h] = (U16)idx; return; }
803 }
804 }
805
LZ4_putPositionOnHash(const BYTE * p,U32 h,void * tableBase,tableType_t const tableType,const BYTE * srcBase)806 LZ4_FORCE_INLINE void LZ4_putPositionOnHash(const BYTE* p, U32 h,
807 void* tableBase, tableType_t const tableType,
808 const BYTE* srcBase)
809 {
810 switch (tableType)
811 {
812 case clearedTable: { /* illegal! */ assert(0); return; }
813 case byPtr: { const BYTE** hashTable = (const BYTE**)tableBase; hashTable[h] = p; return; }
814 case byU32: { U32* hashTable = (U32*) tableBase; hashTable[h] = (U32)(p-srcBase); return; }
815 case byU16: { U16* hashTable = (U16*) tableBase; hashTable[h] = (U16)(p-srcBase); return; }
816 }
817 }
818
LZ4_putPosition(const BYTE * p,void * tableBase,tableType_t tableType,const BYTE * srcBase)819 LZ4_FORCE_INLINE void LZ4_putPosition(const BYTE* p, void* tableBase, tableType_t tableType, const BYTE* srcBase)
820 {
821 U32 const h = LZ4_hashPosition(p, tableType);
822 LZ4_putPositionOnHash(p, h, tableBase, tableType, srcBase);
823 }
824
825 /* LZ4_getIndexOnHash() :
826 * Index of match position registered in hash table.
827 * hash position must be calculated by using base+index, or dictBase+index.
828 * Assumption 1 : only valid if tableType == byU32 or byU16.
829 * Assumption 2 : h is presumed valid (within limits of hash table)
830 */
LZ4_getIndexOnHash(U32 h,const void * tableBase,tableType_t tableType)831 LZ4_FORCE_INLINE U32 LZ4_getIndexOnHash(U32 h, const void* tableBase, tableType_t tableType)
832 {
833 LZ4_STATIC_ASSERT(LZ4_MEMORY_USAGE > 2);
834 if (tableType == byU32) {
835 const U32* const hashTable = (const U32*) tableBase;
836 assert(h < (1U << (LZ4_MEMORY_USAGE-2)));
837 return hashTable[h];
838 }
839 if (tableType == byU16) {
840 const U16* const hashTable = (const U16*) tableBase;
841 assert(h < (1U << (LZ4_MEMORY_USAGE-1)));
842 return hashTable[h];
843 }
844 assert(0); return 0; /* forbidden case */
845 }
846
LZ4_getPositionOnHash(U32 h,const void * tableBase,tableType_t tableType,const BYTE * srcBase)847 static const BYTE* LZ4_getPositionOnHash(U32 h, const void* tableBase, tableType_t tableType, const BYTE* srcBase)
848 {
849 if (tableType == byPtr) { const BYTE* const* hashTable = (const BYTE* const*) tableBase; return hashTable[h]; }
850 if (tableType == byU32) { const U32* const hashTable = (const U32*) tableBase; return hashTable[h] + srcBase; }
851 { const U16* const hashTable = (const U16*) tableBase; return hashTable[h] + srcBase; } /* default, to ensure a return */
852 }
853
854 LZ4_FORCE_INLINE const BYTE*
LZ4_getPosition(const BYTE * p,const void * tableBase,tableType_t tableType,const BYTE * srcBase)855 LZ4_getPosition(const BYTE* p,
856 const void* tableBase, tableType_t tableType,
857 const BYTE* srcBase)
858 {
859 U32 const h = LZ4_hashPosition(p, tableType);
860 return LZ4_getPositionOnHash(h, tableBase, tableType, srcBase);
861 }
862
863 LZ4_FORCE_INLINE void
LZ4_prepareTable(LZ4_stream_t_internal * const cctx,const int inputSize,const tableType_t tableType)864 LZ4_prepareTable(LZ4_stream_t_internal* const cctx,
865 const int inputSize,
866 const tableType_t tableType) {
867 /* If the table hasn't been used, it's guaranteed to be zeroed out, and is
868 * therefore safe to use no matter what mode we're in. Otherwise, we figure
869 * out if it's safe to leave as is or whether it needs to be reset.
870 */
871 if ((tableType_t)cctx->tableType != clearedTable) {
872 assert(inputSize >= 0);
873 if ((tableType_t)cctx->tableType != tableType
874 || ((tableType == byU16) && cctx->currentOffset + (unsigned)inputSize >= 0xFFFFU)
875 || ((tableType == byU32) && cctx->currentOffset > 1 GB)
876 || tableType == byPtr
877 || inputSize >= 4 KB)
878 {
879 DEBUGLOG(4, "LZ4_prepareTable: Resetting table in %p", cctx);
880 MEM_INIT(cctx->hashTable, 0, LZ4_HASHTABLESIZE);
881 cctx->currentOffset = 0;
882 cctx->tableType = (U32)clearedTable;
883 } else {
884 DEBUGLOG(4, "LZ4_prepareTable: Re-use hash table (no reset)");
885 }
886 }
887
888 /* Adding a gap, so all previous entries are > LZ4_DISTANCE_MAX back,
889 * is faster than compressing without a gap.
890 * However, compressing with currentOffset == 0 is faster still,
891 * so we preserve that case.
892 */
893 if (cctx->currentOffset != 0 && tableType == byU32) {
894 DEBUGLOG(5, "LZ4_prepareTable: adding 64KB to currentOffset");
895 cctx->currentOffset += 64 KB;
896 }
897
898 /* Finally, clear history */
899 cctx->dictCtx = NULL;
900 cctx->dictionary = NULL;
901 cctx->dictSize = 0;
902 }
903
904 /** LZ4_compress_generic() :
905 * inlined, to ensure branches are decided at compilation time.
906 * Presumed already validated at this stage:
907 * - source != NULL
908 * - inputSize > 0
909 */
LZ4_compress_generic_validated(LZ4_stream_t_internal * const cctx,const char * const source,char * const dest,const int inputSize,int * inputConsumed,const int maxOutputSize,const limitedOutput_directive outputDirective,const tableType_t tableType,const dict_directive dictDirective,const dictIssue_directive dictIssue,const int acceleration)910 LZ4_FORCE_INLINE int LZ4_compress_generic_validated(
911 LZ4_stream_t_internal* const cctx,
912 const char* const source,
913 char* const dest,
914 const int inputSize,
915 int* inputConsumed, /* only written when outputDirective == fillOutput */
916 const int maxOutputSize,
917 const limitedOutput_directive outputDirective,
918 const tableType_t tableType,
919 const dict_directive dictDirective,
920 const dictIssue_directive dictIssue,
921 const int acceleration)
922 {
923 int result;
924 const BYTE* ip = (const BYTE*) source;
925
926 U32 const startIndex = cctx->currentOffset;
927 const BYTE* base = (const BYTE*) source - startIndex;
928 const BYTE* lowLimit;
929
930 const LZ4_stream_t_internal* dictCtx = (const LZ4_stream_t_internal*) cctx->dictCtx;
931 const BYTE* const dictionary =
932 dictDirective == usingDictCtx ? dictCtx->dictionary : cctx->dictionary;
933 const U32 dictSize =
934 dictDirective == usingDictCtx ? dictCtx->dictSize : cctx->dictSize;
935 const U32 dictDelta = (dictDirective == usingDictCtx) ? startIndex - dictCtx->currentOffset : 0; /* make indexes in dictCtx comparable with index in current context */
936
937 int const maybe_extMem = (dictDirective == usingExtDict) || (dictDirective == usingDictCtx);
938 U32 const prefixIdxLimit = startIndex - dictSize; /* used when dictDirective == dictSmall */
939 const BYTE* const dictEnd = dictionary ? dictionary + dictSize : dictionary;
940 const BYTE* anchor = (const BYTE*) source;
941 const BYTE* const iend = ip + inputSize;
942 const BYTE* const mflimitPlusOne = iend - MFLIMIT + 1;
943 const BYTE* const matchlimit = iend - LASTLITERALS;
944
945 /* the dictCtx currentOffset is indexed on the start of the dictionary,
946 * while a dictionary in the current context precedes the currentOffset */
947 const BYTE* dictBase = (dictionary == NULL) ? NULL :
948 (dictDirective == usingDictCtx) ?
949 dictionary + dictSize - dictCtx->currentOffset :
950 dictionary + dictSize - startIndex;
951
952 BYTE* op = (BYTE*) dest;
953 BYTE* const olimit = op + maxOutputSize;
954
955 U32 offset = 0;
956 U32 forwardH;
957
958 DEBUGLOG(5, "LZ4_compress_generic_validated: srcSize=%i, tableType=%u", inputSize, tableType);
959 assert(ip != NULL);
960 /* If init conditions are not met, we don't have to mark stream
961 * as having dirty context, since no action was taken yet */
962 if (outputDirective == fillOutput && maxOutputSize < 1) { return 0; } /* Impossible to store anything */
963 if ((tableType == byU16) && (inputSize>=LZ4_64Klimit)) { return 0; } /* Size too large (not within 64K limit) */
964 if (tableType==byPtr) assert(dictDirective==noDict); /* only supported use case with byPtr */
965 assert(acceleration >= 1);
966
967 lowLimit = (const BYTE*)source - (dictDirective == withPrefix64k ? dictSize : 0);
968
969 /* Update context state */
970 if (dictDirective == usingDictCtx) {
971 /* Subsequent linked blocks can't use the dictionary. */
972 /* Instead, they use the block we just compressed. */
973 cctx->dictCtx = NULL;
974 cctx->dictSize = (U32)inputSize;
975 } else {
976 cctx->dictSize += (U32)inputSize;
977 }
978 cctx->currentOffset += (U32)inputSize;
979 cctx->tableType = (U32)tableType;
980
981 if (inputSize<LZ4_minLength) goto _last_literals; /* Input too small, no compression (all literals) */
982
983 /* First Byte */
984 LZ4_putPosition(ip, cctx->hashTable, tableType, base);
985 ip++; forwardH = LZ4_hashPosition(ip, tableType);
986
987 /* Main Loop */
988 for ( ; ; ) {
989 const BYTE* match;
990 BYTE* token;
991 const BYTE* filledIp;
992
993 /* Find a match */
994 if (tableType == byPtr) {
995 const BYTE* forwardIp = ip;
996 int step = 1;
997 int searchMatchNb = acceleration << LZ4_skipTrigger;
998 do {
999 U32 const h = forwardH;
1000 ip = forwardIp;
1001 forwardIp += step;
1002 step = (searchMatchNb++ >> LZ4_skipTrigger);
1003
1004 if (unlikely(forwardIp > mflimitPlusOne)) goto _last_literals;
1005 assert(ip < mflimitPlusOne);
1006
1007 match = LZ4_getPositionOnHash(h, cctx->hashTable, tableType, base);
1008 forwardH = LZ4_hashPosition(forwardIp, tableType);
1009 LZ4_putPositionOnHash(ip, h, cctx->hashTable, tableType, base);
1010
1011 } while ( (match+LZ4_DISTANCE_MAX < ip)
1012 || (LZ4_read32(match) != LZ4_read32(ip)) );
1013
1014 } else { /* byU32, byU16 */
1015
1016 const BYTE* forwardIp = ip;
1017 int step = 1;
1018 int searchMatchNb = acceleration << LZ4_skipTrigger;
1019 do {
1020 U32 const h = forwardH;
1021 U32 const current = (U32)(forwardIp - base);
1022 U32 matchIndex = LZ4_getIndexOnHash(h, cctx->hashTable, tableType);
1023 assert(matchIndex <= current);
1024 assert(forwardIp - base < (ptrdiff_t)(2 GB - 1));
1025 ip = forwardIp;
1026 forwardIp += step;
1027 step = (searchMatchNb++ >> LZ4_skipTrigger);
1028
1029 if (unlikely(forwardIp > mflimitPlusOne)) goto _last_literals;
1030 assert(ip < mflimitPlusOne);
1031
1032 if (dictDirective == usingDictCtx) {
1033 if (matchIndex < startIndex) {
1034 /* there was no match, try the dictionary */
1035 assert(tableType == byU32);
1036 matchIndex = LZ4_getIndexOnHash(h, dictCtx->hashTable, byU32);
1037 match = dictBase + matchIndex;
1038 matchIndex += dictDelta; /* make dictCtx index comparable with current context */
1039 lowLimit = dictionary;
1040 } else {
1041 match = base + matchIndex;
1042 lowLimit = (const BYTE*)source;
1043 }
1044 } else if (dictDirective == usingExtDict) {
1045 if (matchIndex < startIndex) {
1046 DEBUGLOG(7, "extDict candidate: matchIndex=%5u < startIndex=%5u", matchIndex, startIndex);
1047 assert(startIndex - matchIndex >= MINMATCH);
1048 assert(dictBase);
1049 match = dictBase + matchIndex;
1050 lowLimit = dictionary;
1051 } else {
1052 match = base + matchIndex;
1053 lowLimit = (const BYTE*)source;
1054 }
1055 } else { /* single continuous memory segment */
1056 match = base + matchIndex;
1057 }
1058 forwardH = LZ4_hashPosition(forwardIp, tableType);
1059 LZ4_putIndexOnHash(current, h, cctx->hashTable, tableType);
1060
1061 DEBUGLOG(7, "candidate at pos=%u (offset=%u \n", matchIndex, current - matchIndex);
1062 if ((dictIssue == dictSmall) && (matchIndex < prefixIdxLimit)) { continue; } /* match outside of valid area */
1063 assert(matchIndex < current);
1064 if ( ((tableType != byU16) || (LZ4_DISTANCE_MAX < LZ4_DISTANCE_ABSOLUTE_MAX))
1065 && (matchIndex+LZ4_DISTANCE_MAX < current)) {
1066 continue;
1067 } /* too far */
1068 assert((current - matchIndex) <= LZ4_DISTANCE_MAX); /* match now expected within distance */
1069
1070 if (LZ4_read32(match) == LZ4_read32(ip)) {
1071 if (maybe_extMem) offset = current - matchIndex;
1072 break; /* match found */
1073 }
1074
1075 } while(1);
1076 }
1077
1078 /* Catch up */
1079 filledIp = ip;
1080 while (((ip>anchor) & (match > lowLimit)) && (unlikely(ip[-1]==match[-1]))) { ip--; match--; }
1081
1082 /* Encode Literals */
1083 { unsigned const litLength = (unsigned)(ip - anchor);
1084 token = op++;
1085 if ((outputDirective == limitedOutput) && /* Check output buffer overflow */
1086 (unlikely(op + litLength + (2 + 1 + LASTLITERALS) + (litLength/255) > olimit)) ) {
1087 return 0; /* cannot compress within `dst` budget. Stored indexes in hash table are nonetheless fine */
1088 }
1089 if ((outputDirective == fillOutput) &&
1090 (unlikely(op + (litLength+240)/255 /* litlen */ + litLength /* literals */ + 2 /* offset */ + 1 /* token */ + MFLIMIT - MINMATCH /* min last literals so last match is <= end - MFLIMIT */ > olimit))) {
1091 op--;
1092 goto _last_literals;
1093 }
1094 if (litLength >= RUN_MASK) {
1095 int len = (int)(litLength - RUN_MASK);
1096 *token = (RUN_MASK<<ML_BITS);
1097 for(; len >= 255 ; len-=255) *op++ = 255;
1098 *op++ = (BYTE)len;
1099 }
1100 else *token = (BYTE)(litLength<<ML_BITS);
1101
1102 /* Copy Literals */
1103 LZ4_wildCopy8(op, anchor, op+litLength);
1104 op+=litLength;
1105 DEBUGLOG(6, "seq.start:%i, literals=%u, match.start:%i",
1106 (int)(anchor-(const BYTE*)source), litLength, (int)(ip-(const BYTE*)source));
1107 }
1108
1109 _next_match:
1110 /* at this stage, the following variables must be correctly set :
1111 * - ip : at start of LZ operation
1112 * - match : at start of previous pattern occurrence; can be within current prefix, or within extDict
1113 * - offset : if maybe_ext_memSegment==1 (constant)
1114 * - lowLimit : must be == dictionary to mean "match is within extDict"; must be == source otherwise
1115 * - token and *token : position to write 4-bits for match length; higher 4-bits for literal length supposed already written
1116 */
1117
1118 if ((outputDirective == fillOutput) &&
1119 (op + 2 /* offset */ + 1 /* token */ + MFLIMIT - MINMATCH /* min last literals so last match is <= end - MFLIMIT */ > olimit)) {
1120 /* the match was too close to the end, rewind and go to last literals */
1121 op = token;
1122 goto _last_literals;
1123 }
1124
1125 /* Encode Offset */
1126 if (maybe_extMem) { /* static test */
1127 DEBUGLOG(6, " with offset=%u (ext if > %i)", offset, (int)(ip - (const BYTE*)source));
1128 assert(offset <= LZ4_DISTANCE_MAX && offset > 0);
1129 LZ4_writeLE16(op, (U16)offset); op+=2;
1130 } else {
1131 DEBUGLOG(6, " with offset=%u (same segment)", (U32)(ip - match));
1132 assert(ip-match <= LZ4_DISTANCE_MAX);
1133 LZ4_writeLE16(op, (U16)(ip - match)); op+=2;
1134 }
1135
1136 /* Encode MatchLength */
1137 { unsigned matchCode;
1138
1139 if ( (dictDirective==usingExtDict || dictDirective==usingDictCtx)
1140 && (lowLimit==dictionary) /* match within extDict */ ) {
1141 const BYTE* limit = ip + (dictEnd-match);
1142 assert(dictEnd > match);
1143 if (limit > matchlimit) limit = matchlimit;
1144 matchCode = LZ4_count(ip+MINMATCH, match+MINMATCH, limit);
1145 ip += (size_t)matchCode + MINMATCH;
1146 if (ip==limit) {
1147 unsigned const more = LZ4_count(limit, (const BYTE*)source, matchlimit);
1148 matchCode += more;
1149 ip += more;
1150 }
1151 DEBUGLOG(6, " with matchLength=%u starting in extDict", matchCode+MINMATCH);
1152 } else {
1153 matchCode = LZ4_count(ip+MINMATCH, match+MINMATCH, matchlimit);
1154 ip += (size_t)matchCode + MINMATCH;
1155 DEBUGLOG(6, " with matchLength=%u", matchCode+MINMATCH);
1156 }
1157
1158 if ((outputDirective) && /* Check output buffer overflow */
1159 (unlikely(op + (1 + LASTLITERALS) + (matchCode+240)/255 > olimit)) ) {
1160 if (outputDirective == fillOutput) {
1161 /* Match description too long : reduce it */
1162 U32 newMatchCode = 15 /* in token */ - 1 /* to avoid needing a zero byte */ + ((U32)(olimit - op) - 1 - LASTLITERALS) * 255;
1163 ip -= matchCode - newMatchCode;
1164 assert(newMatchCode < matchCode);
1165 matchCode = newMatchCode;
1166 if (unlikely(ip <= filledIp)) {
1167 /* We have already filled up to filledIp so if ip ends up less than filledIp
1168 * we have positions in the hash table beyond the current position. This is
1169 * a problem if we reuse the hash table. So we have to remove these positions
1170 * from the hash table.
1171 */
1172 const BYTE* ptr;
1173 DEBUGLOG(5, "Clearing %u positions", (U32)(filledIp - ip));
1174 for (ptr = ip; ptr <= filledIp; ++ptr) {
1175 U32 const h = LZ4_hashPosition(ptr, tableType);
1176 LZ4_clearHash(h, cctx->hashTable, tableType);
1177 }
1178 }
1179 } else {
1180 assert(outputDirective == limitedOutput);
1181 return 0; /* cannot compress within `dst` budget. Stored indexes in hash table are nonetheless fine */
1182 }
1183 }
1184 if (matchCode >= ML_MASK) {
1185 *token += ML_MASK;
1186 matchCode -= ML_MASK;
1187 LZ4_write32(op, 0xFFFFFFFF);
1188 while (matchCode >= 4*255) {
1189 op+=4;
1190 LZ4_write32(op, 0xFFFFFFFF);
1191 matchCode -= 4*255;
1192 }
1193 op += matchCode / 255;
1194 *op++ = (BYTE)(matchCode % 255);
1195 } else
1196 *token += (BYTE)(matchCode);
1197 }
1198 /* Ensure we have enough space for the last literals. */
1199 assert(!(outputDirective == fillOutput && op + 1 + LASTLITERALS > olimit));
1200
1201 anchor = ip;
1202
1203 /* Test end of chunk */
1204 if (ip >= mflimitPlusOne) break;
1205
1206 /* Fill table */
1207 LZ4_putPosition(ip-2, cctx->hashTable, tableType, base);
1208
1209 /* Test next position */
1210 if (tableType == byPtr) {
1211
1212 match = LZ4_getPosition(ip, cctx->hashTable, tableType, base);
1213 LZ4_putPosition(ip, cctx->hashTable, tableType, base);
1214 if ( (match+LZ4_DISTANCE_MAX >= ip)
1215 && (LZ4_read32(match) == LZ4_read32(ip)) )
1216 { token=op++; *token=0; goto _next_match; }
1217
1218 } else { /* byU32, byU16 */
1219
1220 U32 const h = LZ4_hashPosition(ip, tableType);
1221 U32 const current = (U32)(ip-base);
1222 U32 matchIndex = LZ4_getIndexOnHash(h, cctx->hashTable, tableType);
1223 assert(matchIndex < current);
1224 if (dictDirective == usingDictCtx) {
1225 if (matchIndex < startIndex) {
1226 /* there was no match, try the dictionary */
1227 matchIndex = LZ4_getIndexOnHash(h, dictCtx->hashTable, byU32);
1228 match = dictBase + matchIndex;
1229 lowLimit = dictionary; /* required for match length counter */
1230 matchIndex += dictDelta;
1231 } else {
1232 match = base + matchIndex;
1233 lowLimit = (const BYTE*)source; /* required for match length counter */
1234 }
1235 } else if (dictDirective==usingExtDict) {
1236 if (matchIndex < startIndex) {
1237 assert(dictBase);
1238 match = dictBase + matchIndex;
1239 lowLimit = dictionary; /* required for match length counter */
1240 } else {
1241 match = base + matchIndex;
1242 lowLimit = (const BYTE*)source; /* required for match length counter */
1243 }
1244 } else { /* single memory segment */
1245 match = base + matchIndex;
1246 }
1247 LZ4_putIndexOnHash(current, h, cctx->hashTable, tableType);
1248 assert(matchIndex < current);
1249 if ( ((dictIssue==dictSmall) ? (matchIndex >= prefixIdxLimit) : 1)
1250 && (((tableType==byU16) && (LZ4_DISTANCE_MAX == LZ4_DISTANCE_ABSOLUTE_MAX)) ? 1 : (matchIndex+LZ4_DISTANCE_MAX >= current))
1251 && (LZ4_read32(match) == LZ4_read32(ip)) ) {
1252 token=op++;
1253 *token=0;
1254 if (maybe_extMem) offset = current - matchIndex;
1255 DEBUGLOG(6, "seq.start:%i, literals=%u, match.start:%i",
1256 (int)(anchor-(const BYTE*)source), 0, (int)(ip-(const BYTE*)source));
1257 goto _next_match;
1258 }
1259 }
1260
1261 /* Prepare next loop */
1262 forwardH = LZ4_hashPosition(++ip, tableType);
1263
1264 }
1265
1266 _last_literals:
1267 /* Encode Last Literals */
1268 { size_t lastRun = (size_t)(iend - anchor);
1269 if ( (outputDirective) && /* Check output buffer overflow */
1270 (op + lastRun + 1 + ((lastRun+255-RUN_MASK)/255) > olimit)) {
1271 if (outputDirective == fillOutput) {
1272 /* adapt lastRun to fill 'dst' */
1273 assert(olimit >= op);
1274 lastRun = (size_t)(olimit-op) - 1/*token*/;
1275 lastRun -= (lastRun + 256 - RUN_MASK) / 256; /*additional length tokens*/
1276 } else {
1277 assert(outputDirective == limitedOutput);
1278 return 0; /* cannot compress within `dst` budget. Stored indexes in hash table are nonetheless fine */
1279 }
1280 }
1281 DEBUGLOG(6, "Final literal run : %i literals", (int)lastRun);
1282 if (lastRun >= RUN_MASK) {
1283 size_t accumulator = lastRun - RUN_MASK;
1284 *op++ = RUN_MASK << ML_BITS;
1285 for(; accumulator >= 255 ; accumulator-=255) *op++ = 255;
1286 *op++ = (BYTE) accumulator;
1287 } else {
1288 *op++ = (BYTE)(lastRun<<ML_BITS);
1289 }
1290 LZ4_memcpy(op, anchor, lastRun);
1291 ip = anchor + lastRun;
1292 op += lastRun;
1293 }
1294
1295 if (outputDirective == fillOutput) {
1296 *inputConsumed = (int) (((const char*)ip)-source);
1297 }
1298 result = (int)(((char*)op) - dest);
1299 assert(result > 0);
1300 DEBUGLOG(5, "LZ4_compress_generic: compressed %i bytes into %i bytes", inputSize, result);
1301 return result;
1302 }
1303
1304 /** LZ4_compress_generic() :
1305 * inlined, to ensure branches are decided at compilation time;
1306 * takes care of src == (NULL, 0)
1307 * and forward the rest to LZ4_compress_generic_validated */
LZ4_compress_generic(LZ4_stream_t_internal * const cctx,const char * const src,char * const dst,const int srcSize,int * inputConsumed,const int dstCapacity,const limitedOutput_directive outputDirective,const tableType_t tableType,const dict_directive dictDirective,const dictIssue_directive dictIssue,const int acceleration)1308 LZ4_FORCE_INLINE int LZ4_compress_generic(
1309 LZ4_stream_t_internal* const cctx,
1310 const char* const src,
1311 char* const dst,
1312 const int srcSize,
1313 int *inputConsumed, /* only written when outputDirective == fillOutput */
1314 const int dstCapacity,
1315 const limitedOutput_directive outputDirective,
1316 const tableType_t tableType,
1317 const dict_directive dictDirective,
1318 const dictIssue_directive dictIssue,
1319 const int acceleration)
1320 {
1321 DEBUGLOG(5, "LZ4_compress_generic: srcSize=%i, dstCapacity=%i",
1322 srcSize, dstCapacity);
1323
1324 if ((U32)srcSize > (U32)LZ4_MAX_INPUT_SIZE) { return 0; } /* Unsupported srcSize, too large (or negative) */
1325 if (srcSize == 0) { /* src == NULL supported if srcSize == 0 */
1326 if (outputDirective != notLimited && dstCapacity <= 0) return 0; /* no output, can't write anything */
1327 DEBUGLOG(5, "Generating an empty block");
1328 assert(outputDirective == notLimited || dstCapacity >= 1);
1329 assert(dst != NULL);
1330 dst[0] = 0;
1331 if (outputDirective == fillOutput) {
1332 assert (inputConsumed != NULL);
1333 *inputConsumed = 0;
1334 }
1335 return 1;
1336 }
1337 assert(src != NULL);
1338
1339 return LZ4_compress_generic_validated(cctx, src, dst, srcSize,
1340 inputConsumed, /* only written into if outputDirective == fillOutput */
1341 dstCapacity, outputDirective,
1342 tableType, dictDirective, dictIssue, acceleration);
1343 }
1344
1345
LZ4_compress_fast_extState(void * state,const char * source,char * dest,int inputSize,int maxOutputSize,int acceleration)1346 int LZ4_compress_fast_extState(void* state, const char* source, char* dest, int inputSize, int maxOutputSize, int acceleration)
1347 {
1348 LZ4_stream_t_internal* const ctx = & LZ4_initStream(state, sizeof(LZ4_stream_t)) -> internal_donotuse;
1349 assert(ctx != NULL);
1350 if (acceleration < 1) acceleration = LZ4_ACCELERATION_DEFAULT;
1351 if (acceleration > LZ4_ACCELERATION_MAX) acceleration = LZ4_ACCELERATION_MAX;
1352 if (maxOutputSize >= LZ4_compressBound(inputSize)) {
1353 if (inputSize < LZ4_64Klimit) {
1354 return LZ4_compress_generic(ctx, source, dest, inputSize, NULL, 0, notLimited, byU16, noDict, noDictIssue, acceleration);
1355 } else {
1356 const tableType_t tableType = ((sizeof(void*)==4) && ((uptrval)source > LZ4_DISTANCE_MAX)) ? byPtr : byU32;
1357 return LZ4_compress_generic(ctx, source, dest, inputSize, NULL, 0, notLimited, tableType, noDict, noDictIssue, acceleration);
1358 }
1359 } else {
1360 if (inputSize < LZ4_64Klimit) {
1361 return LZ4_compress_generic(ctx, source, dest, inputSize, NULL, maxOutputSize, limitedOutput, byU16, noDict, noDictIssue, acceleration);
1362 } else {
1363 const tableType_t tableType = ((sizeof(void*)==4) && ((uptrval)source > LZ4_DISTANCE_MAX)) ? byPtr : byU32;
1364 return LZ4_compress_generic(ctx, source, dest, inputSize, NULL, maxOutputSize, limitedOutput, tableType, noDict, noDictIssue, acceleration);
1365 }
1366 }
1367 }
1368
1369 /**
1370 * LZ4_compress_fast_extState_fastReset() :
1371 * A variant of LZ4_compress_fast_extState().
1372 *
1373 * Using this variant avoids an expensive initialization step. It is only safe
1374 * to call if the state buffer is known to be correctly initialized already
1375 * (see comment in lz4.h on LZ4_resetStream_fast() for a definition of
1376 * "correctly initialized").
1377 */
LZ4_compress_fast_extState_fastReset(void * state,const char * src,char * dst,int srcSize,int dstCapacity,int acceleration)1378 int LZ4_compress_fast_extState_fastReset(void* state, const char* src, char* dst, int srcSize, int dstCapacity, int acceleration)
1379 {
1380 LZ4_stream_t_internal* ctx = &((LZ4_stream_t*)state)->internal_donotuse;
1381 if (acceleration < 1) acceleration = LZ4_ACCELERATION_DEFAULT;
1382 if (acceleration > LZ4_ACCELERATION_MAX) acceleration = LZ4_ACCELERATION_MAX;
1383
1384 if (dstCapacity >= LZ4_compressBound(srcSize)) {
1385 if (srcSize < LZ4_64Klimit) {
1386 const tableType_t tableType = byU16;
1387 LZ4_prepareTable(ctx, srcSize, tableType);
1388 if (ctx->currentOffset) {
1389 return LZ4_compress_generic(ctx, src, dst, srcSize, NULL, 0, notLimited, tableType, noDict, dictSmall, acceleration);
1390 } else {
1391 return LZ4_compress_generic(ctx, src, dst, srcSize, NULL, 0, notLimited, tableType, noDict, noDictIssue, acceleration);
1392 }
1393 } else {
1394 const tableType_t tableType = ((sizeof(void*)==4) && ((uptrval)src > LZ4_DISTANCE_MAX)) ? byPtr : byU32;
1395 LZ4_prepareTable(ctx, srcSize, tableType);
1396 return LZ4_compress_generic(ctx, src, dst, srcSize, NULL, 0, notLimited, tableType, noDict, noDictIssue, acceleration);
1397 }
1398 } else {
1399 if (srcSize < LZ4_64Klimit) {
1400 const tableType_t tableType = byU16;
1401 LZ4_prepareTable(ctx, srcSize, tableType);
1402 if (ctx->currentOffset) {
1403 return LZ4_compress_generic(ctx, src, dst, srcSize, NULL, dstCapacity, limitedOutput, tableType, noDict, dictSmall, acceleration);
1404 } else {
1405 return LZ4_compress_generic(ctx, src, dst, srcSize, NULL, dstCapacity, limitedOutput, tableType, noDict, noDictIssue, acceleration);
1406 }
1407 } else {
1408 const tableType_t tableType = ((sizeof(void*)==4) && ((uptrval)src > LZ4_DISTANCE_MAX)) ? byPtr : byU32;
1409 LZ4_prepareTable(ctx, srcSize, tableType);
1410 return LZ4_compress_generic(ctx, src, dst, srcSize, NULL, dstCapacity, limitedOutput, tableType, noDict, noDictIssue, acceleration);
1411 }
1412 }
1413 }
1414
1415
LZ4_compress_fast(const char * source,char * dest,int inputSize,int maxOutputSize,int acceleration)1416 int LZ4_compress_fast(const char* source, char* dest, int inputSize, int maxOutputSize, int acceleration)
1417 {
1418 int result;
1419 #if (LZ4_HEAPMODE)
1420 LZ4_stream_t* ctxPtr = (LZ4_stream_t*)ALLOC(sizeof(LZ4_stream_t)); /* malloc-calloc always properly aligned */
1421 if (ctxPtr == NULL) return 0;
1422 #else
1423 LZ4_stream_t ctx;
1424 LZ4_stream_t* const ctxPtr = &ctx;
1425 #endif
1426 result = LZ4_compress_fast_extState(ctxPtr, source, dest, inputSize, maxOutputSize, acceleration);
1427
1428 #if (LZ4_HEAPMODE)
1429 FREEMEM(ctxPtr);
1430 #endif
1431 return result;
1432 }
1433
1434
LZ4_compress_default(const char * src,char * dst,int srcSize,int maxOutputSize)1435 int LZ4_compress_default(const char* src, char* dst, int srcSize, int maxOutputSize)
1436 {
1437 return LZ4_compress_fast(src, dst, srcSize, maxOutputSize, 1);
1438 }
1439
1440
1441 /* Note!: This function leaves the stream in an unclean/broken state!
1442 * It is not safe to subsequently use the same state with a _fastReset() or
1443 * _continue() call without resetting it. */
LZ4_compress_destSize_extState(LZ4_stream_t * state,const char * src,char * dst,int * srcSizePtr,int targetDstSize)1444 static int LZ4_compress_destSize_extState (LZ4_stream_t* state, const char* src, char* dst, int* srcSizePtr, int targetDstSize)
1445 {
1446 void* const s = LZ4_initStream(state, sizeof (*state));
1447 assert(s != NULL); (void)s;
1448
1449 if (targetDstSize >= LZ4_compressBound(*srcSizePtr)) { /* compression success is guaranteed */
1450 return LZ4_compress_fast_extState(state, src, dst, *srcSizePtr, targetDstSize, 1);
1451 } else {
1452 if (*srcSizePtr < LZ4_64Klimit) {
1453 return LZ4_compress_generic(&state->internal_donotuse, src, dst, *srcSizePtr, srcSizePtr, targetDstSize, fillOutput, byU16, noDict, noDictIssue, 1);
1454 } else {
1455 tableType_t const addrMode = ((sizeof(void*)==4) && ((uptrval)src > LZ4_DISTANCE_MAX)) ? byPtr : byU32;
1456 return LZ4_compress_generic(&state->internal_donotuse, src, dst, *srcSizePtr, srcSizePtr, targetDstSize, fillOutput, addrMode, noDict, noDictIssue, 1);
1457 } }
1458 }
1459
1460
LZ4_compress_destSize(const char * src,char * dst,int * srcSizePtr,int targetDstSize)1461 int LZ4_compress_destSize(const char* src, char* dst, int* srcSizePtr, int targetDstSize)
1462 {
1463 #if (LZ4_HEAPMODE)
1464 LZ4_stream_t* ctx = (LZ4_stream_t*)ALLOC(sizeof(LZ4_stream_t)); /* malloc-calloc always properly aligned */
1465 if (ctx == NULL) return 0;
1466 #else
1467 LZ4_stream_t ctxBody;
1468 LZ4_stream_t* ctx = &ctxBody;
1469 #endif
1470
1471 int result = LZ4_compress_destSize_extState(ctx, src, dst, srcSizePtr, targetDstSize);
1472
1473 #if (LZ4_HEAPMODE)
1474 FREEMEM(ctx);
1475 #endif
1476 return result;
1477 }
1478
1479
1480
1481 /*-******************************
1482 * Streaming functions
1483 ********************************/
1484
1485 #if !defined(LZ4_STATIC_LINKING_ONLY_DISABLE_MEMORY_ALLOCATION)
LZ4_createStream(void)1486 LZ4_stream_t* LZ4_createStream(void)
1487 {
1488 LZ4_stream_t* const lz4s = (LZ4_stream_t*)ALLOC(sizeof(LZ4_stream_t));
1489 LZ4_STATIC_ASSERT(sizeof(LZ4_stream_t) >= sizeof(LZ4_stream_t_internal));
1490 DEBUGLOG(4, "LZ4_createStream %p", lz4s);
1491 if (lz4s == NULL) return NULL;
1492 LZ4_initStream(lz4s, sizeof(*lz4s));
1493 return lz4s;
1494 }
1495 #endif
1496
LZ4_stream_t_alignment(void)1497 static size_t LZ4_stream_t_alignment(void)
1498 {
1499 #if LZ4_ALIGN_TEST
1500 typedef struct { char c; LZ4_stream_t t; } t_a;
1501 return sizeof(t_a) - sizeof(LZ4_stream_t);
1502 #else
1503 return 1; /* effectively disabled */
1504 #endif
1505 }
1506
LZ4_initStream(void * buffer,size_t size)1507 LZ4_stream_t* LZ4_initStream (void* buffer, size_t size)
1508 {
1509 DEBUGLOG(5, "LZ4_initStream");
1510 if (buffer == NULL) { return NULL; }
1511 if (size < sizeof(LZ4_stream_t)) { return NULL; }
1512 if (!LZ4_isAligned(buffer, LZ4_stream_t_alignment())) return NULL;
1513 MEM_INIT(buffer, 0, sizeof(LZ4_stream_t_internal));
1514 return (LZ4_stream_t*)buffer;
1515 }
1516
1517 /* resetStream is now deprecated,
1518 * prefer initStream() which is more general */
LZ4_resetStream(LZ4_stream_t * LZ4_stream)1519 void LZ4_resetStream (LZ4_stream_t* LZ4_stream)
1520 {
1521 DEBUGLOG(5, "LZ4_resetStream (ctx:%p)", LZ4_stream);
1522 MEM_INIT(LZ4_stream, 0, sizeof(LZ4_stream_t_internal));
1523 }
1524
LZ4_resetStream_fast(LZ4_stream_t * ctx)1525 void LZ4_resetStream_fast(LZ4_stream_t* ctx) {
1526 LZ4_prepareTable(&(ctx->internal_donotuse), 0, byU32);
1527 }
1528
1529 #if !defined(LZ4_STATIC_LINKING_ONLY_DISABLE_MEMORY_ALLOCATION)
LZ4_freeStream(LZ4_stream_t * LZ4_stream)1530 int LZ4_freeStream (LZ4_stream_t* LZ4_stream)
1531 {
1532 if (!LZ4_stream) return 0; /* support free on NULL */
1533 DEBUGLOG(5, "LZ4_freeStream %p", LZ4_stream);
1534 FREEMEM(LZ4_stream);
1535 return (0);
1536 }
1537 #endif
1538
1539
1540 #define HASH_UNIT sizeof(reg_t)
LZ4_loadDict(LZ4_stream_t * LZ4_dict,const char * dictionary,int dictSize)1541 int LZ4_loadDict (LZ4_stream_t* LZ4_dict, const char* dictionary, int dictSize)
1542 {
1543 LZ4_stream_t_internal* dict = &LZ4_dict->internal_donotuse;
1544 const tableType_t tableType = byU32;
1545 const BYTE* p = (const BYTE*)dictionary;
1546 const BYTE* const dictEnd = p + dictSize;
1547 const BYTE* base;
1548
1549 DEBUGLOG(4, "LZ4_loadDict (%i bytes from %p into %p)", dictSize, dictionary, LZ4_dict);
1550
1551 /* It's necessary to reset the context,
1552 * and not just continue it with prepareTable()
1553 * to avoid any risk of generating overflowing matchIndex
1554 * when compressing using this dictionary */
1555 LZ4_resetStream(LZ4_dict);
1556
1557 /* We always increment the offset by 64 KB, since, if the dict is longer,
1558 * we truncate it to the last 64k, and if it's shorter, we still want to
1559 * advance by a whole window length so we can provide the guarantee that
1560 * there are only valid offsets in the window, which allows an optimization
1561 * in LZ4_compress_fast_continue() where it uses noDictIssue even when the
1562 * dictionary isn't a full 64k. */
1563 dict->currentOffset += 64 KB;
1564
1565 if (dictSize < (int)HASH_UNIT) {
1566 return 0;
1567 }
1568
1569 if ((dictEnd - p) > 64 KB) p = dictEnd - 64 KB;
1570 base = dictEnd - dict->currentOffset;
1571 dict->dictionary = p;
1572 dict->dictSize = (U32)(dictEnd - p);
1573 dict->tableType = (U32)tableType;
1574
1575 while (p <= dictEnd-HASH_UNIT) {
1576 LZ4_putPosition(p, dict->hashTable, tableType, base);
1577 p+=3;
1578 }
1579
1580 return (int)dict->dictSize;
1581 }
1582
LZ4_attach_dictionary(LZ4_stream_t * workingStream,const LZ4_stream_t * dictionaryStream)1583 void LZ4_attach_dictionary(LZ4_stream_t* workingStream, const LZ4_stream_t* dictionaryStream)
1584 {
1585 const LZ4_stream_t_internal* dictCtx = (dictionaryStream == NULL) ? NULL :
1586 &(dictionaryStream->internal_donotuse);
1587
1588 DEBUGLOG(4, "LZ4_attach_dictionary (%p, %p, size %u)",
1589 workingStream, dictionaryStream,
1590 dictCtx != NULL ? dictCtx->dictSize : 0);
1591
1592 if (dictCtx != NULL) {
1593 /* If the current offset is zero, we will never look in the
1594 * external dictionary context, since there is no value a table
1595 * entry can take that indicate a miss. In that case, we need
1596 * to bump the offset to something non-zero.
1597 */
1598 if (workingStream->internal_donotuse.currentOffset == 0) {
1599 workingStream->internal_donotuse.currentOffset = 64 KB;
1600 }
1601
1602 /* Don't actually attach an empty dictionary.
1603 */
1604 if (dictCtx->dictSize == 0) {
1605 dictCtx = NULL;
1606 }
1607 }
1608 workingStream->internal_donotuse.dictCtx = dictCtx;
1609 }
1610
1611
LZ4_renormDictT(LZ4_stream_t_internal * LZ4_dict,int nextSize)1612 static void LZ4_renormDictT(LZ4_stream_t_internal* LZ4_dict, int nextSize)
1613 {
1614 assert(nextSize >= 0);
1615 if (LZ4_dict->currentOffset + (unsigned)nextSize > 0x80000000) { /* potential ptrdiff_t overflow (32-bits mode) */
1616 /* rescale hash table */
1617 U32 const delta = LZ4_dict->currentOffset - 64 KB;
1618 const BYTE* dictEnd = LZ4_dict->dictionary + LZ4_dict->dictSize;
1619 int i;
1620 DEBUGLOG(4, "LZ4_renormDictT");
1621 for (i=0; i<LZ4_HASH_SIZE_U32; i++) {
1622 if (LZ4_dict->hashTable[i] < delta) LZ4_dict->hashTable[i]=0;
1623 else LZ4_dict->hashTable[i] -= delta;
1624 }
1625 LZ4_dict->currentOffset = 64 KB;
1626 if (LZ4_dict->dictSize > 64 KB) LZ4_dict->dictSize = 64 KB;
1627 LZ4_dict->dictionary = dictEnd - LZ4_dict->dictSize;
1628 }
1629 }
1630
1631
LZ4_compress_fast_continue(LZ4_stream_t * LZ4_stream,const char * source,char * dest,int inputSize,int maxOutputSize,int acceleration)1632 int LZ4_compress_fast_continue (LZ4_stream_t* LZ4_stream,
1633 const char* source, char* dest,
1634 int inputSize, int maxOutputSize,
1635 int acceleration)
1636 {
1637 const tableType_t tableType = byU32;
1638 LZ4_stream_t_internal* const streamPtr = &LZ4_stream->internal_donotuse;
1639 const char* dictEnd = streamPtr->dictSize ? (const char*)streamPtr->dictionary + streamPtr->dictSize : NULL;
1640
1641 DEBUGLOG(5, "LZ4_compress_fast_continue (inputSize=%i, dictSize=%u)", inputSize, streamPtr->dictSize);
1642
1643 LZ4_renormDictT(streamPtr, inputSize); /* fix index overflow */
1644 if (acceleration < 1) acceleration = LZ4_ACCELERATION_DEFAULT;
1645 if (acceleration > LZ4_ACCELERATION_MAX) acceleration = LZ4_ACCELERATION_MAX;
1646
1647 /* invalidate tiny dictionaries */
1648 if ( (streamPtr->dictSize < 4) /* tiny dictionary : not enough for a hash */
1649 && (dictEnd != source) /* prefix mode */
1650 && (inputSize > 0) /* tolerance : don't lose history, in case next invocation would use prefix mode */
1651 && (streamPtr->dictCtx == NULL) /* usingDictCtx */
1652 ) {
1653 DEBUGLOG(5, "LZ4_compress_fast_continue: dictSize(%u) at addr:%p is too small", streamPtr->dictSize, streamPtr->dictionary);
1654 /* remove dictionary existence from history, to employ faster prefix mode */
1655 streamPtr->dictSize = 0;
1656 streamPtr->dictionary = (const BYTE*)source;
1657 dictEnd = source;
1658 }
1659
1660 /* Check overlapping input/dictionary space */
1661 { const char* const sourceEnd = source + inputSize;
1662 if ((sourceEnd > (const char*)streamPtr->dictionary) && (sourceEnd < dictEnd)) {
1663 streamPtr->dictSize = (U32)(dictEnd - sourceEnd);
1664 if (streamPtr->dictSize > 64 KB) streamPtr->dictSize = 64 KB;
1665 if (streamPtr->dictSize < 4) streamPtr->dictSize = 0;
1666 streamPtr->dictionary = (const BYTE*)dictEnd - streamPtr->dictSize;
1667 }
1668 }
1669
1670 /* prefix mode : source data follows dictionary */
1671 if (dictEnd == source) {
1672 if ((streamPtr->dictSize < 64 KB) && (streamPtr->dictSize < streamPtr->currentOffset))
1673 return LZ4_compress_generic(streamPtr, source, dest, inputSize, NULL, maxOutputSize, limitedOutput, tableType, withPrefix64k, dictSmall, acceleration);
1674 else
1675 return LZ4_compress_generic(streamPtr, source, dest, inputSize, NULL, maxOutputSize, limitedOutput, tableType, withPrefix64k, noDictIssue, acceleration);
1676 }
1677
1678 /* external dictionary mode */
1679 { int result;
1680 if (streamPtr->dictCtx) {
1681 /* We depend here on the fact that dictCtx'es (produced by
1682 * LZ4_loadDict) guarantee that their tables contain no references
1683 * to offsets between dictCtx->currentOffset - 64 KB and
1684 * dictCtx->currentOffset - dictCtx->dictSize. This makes it safe
1685 * to use noDictIssue even when the dict isn't a full 64 KB.
1686 */
1687 if (inputSize > 4 KB) {
1688 /* For compressing large blobs, it is faster to pay the setup
1689 * cost to copy the dictionary's tables into the active context,
1690 * so that the compression loop is only looking into one table.
1691 */
1692 LZ4_memcpy(streamPtr, streamPtr->dictCtx, sizeof(*streamPtr));
1693 result = LZ4_compress_generic(streamPtr, source, dest, inputSize, NULL, maxOutputSize, limitedOutput, tableType, usingExtDict, noDictIssue, acceleration);
1694 } else {
1695 result = LZ4_compress_generic(streamPtr, source, dest, inputSize, NULL, maxOutputSize, limitedOutput, tableType, usingDictCtx, noDictIssue, acceleration);
1696 }
1697 } else { /* small data <= 4 KB */
1698 if ((streamPtr->dictSize < 64 KB) && (streamPtr->dictSize < streamPtr->currentOffset)) {
1699 result = LZ4_compress_generic(streamPtr, source, dest, inputSize, NULL, maxOutputSize, limitedOutput, tableType, usingExtDict, dictSmall, acceleration);
1700 } else {
1701 result = LZ4_compress_generic(streamPtr, source, dest, inputSize, NULL, maxOutputSize, limitedOutput, tableType, usingExtDict, noDictIssue, acceleration);
1702 }
1703 }
1704 streamPtr->dictionary = (const BYTE*)source;
1705 streamPtr->dictSize = (U32)inputSize;
1706 return result;
1707 }
1708 }
1709
1710
1711 /* Hidden debug function, to force-test external dictionary mode */
LZ4_compress_forceExtDict(LZ4_stream_t * LZ4_dict,const char * source,char * dest,int srcSize)1712 int LZ4_compress_forceExtDict (LZ4_stream_t* LZ4_dict, const char* source, char* dest, int srcSize)
1713 {
1714 LZ4_stream_t_internal* streamPtr = &LZ4_dict->internal_donotuse;
1715 int result;
1716
1717 LZ4_renormDictT(streamPtr, srcSize);
1718
1719 if ((streamPtr->dictSize < 64 KB) && (streamPtr->dictSize < streamPtr->currentOffset)) {
1720 result = LZ4_compress_generic(streamPtr, source, dest, srcSize, NULL, 0, notLimited, byU32, usingExtDict, dictSmall, 1);
1721 } else {
1722 result = LZ4_compress_generic(streamPtr, source, dest, srcSize, NULL, 0, notLimited, byU32, usingExtDict, noDictIssue, 1);
1723 }
1724
1725 streamPtr->dictionary = (const BYTE*)source;
1726 streamPtr->dictSize = (U32)srcSize;
1727
1728 return result;
1729 }
1730
1731
1732 /*! LZ4_saveDict() :
1733 * If previously compressed data block is not guaranteed to remain available at its memory location,
1734 * save it into a safer place (char* safeBuffer).
1735 * Note : no need to call LZ4_loadDict() afterwards, dictionary is immediately usable,
1736 * one can therefore call LZ4_compress_fast_continue() right after.
1737 * @return : saved dictionary size in bytes (necessarily <= dictSize), or 0 if error.
1738 */
LZ4_saveDict(LZ4_stream_t * LZ4_dict,char * safeBuffer,int dictSize)1739 int LZ4_saveDict (LZ4_stream_t* LZ4_dict, char* safeBuffer, int dictSize)
1740 {
1741 LZ4_stream_t_internal* const dict = &LZ4_dict->internal_donotuse;
1742
1743 DEBUGLOG(5, "LZ4_saveDict : dictSize=%i, safeBuffer=%p", dictSize, safeBuffer);
1744
1745 if ((U32)dictSize > 64 KB) { dictSize = 64 KB; } /* useless to define a dictionary > 64 KB */
1746 if ((U32)dictSize > dict->dictSize) { dictSize = (int)dict->dictSize; }
1747
1748 if (safeBuffer == NULL) assert(dictSize == 0);
1749 if (dictSize > 0) {
1750 const BYTE* const previousDictEnd = dict->dictionary + dict->dictSize;
1751 assert(dict->dictionary);
1752 LZ4_memmove(safeBuffer, previousDictEnd - dictSize, (size_t)dictSize);
1753 }
1754
1755 dict->dictionary = (const BYTE*)safeBuffer;
1756 dict->dictSize = (U32)dictSize;
1757
1758 return dictSize;
1759 }
1760
1761
1762
1763 /*-*******************************
1764 * Decompression functions
1765 ********************************/
1766
1767 typedef enum { decode_full_block = 0, partial_decode = 1 } earlyEnd_directive;
1768
1769 #undef MIN
1770 #define MIN(a,b) ( (a) < (b) ? (a) : (b) )
1771
1772
1773 /* variant for decompress_unsafe()
1774 * does not know end of input
1775 * presumes input is well formed
1776 * note : will consume at least one byte */
read_long_length_no_check(const BYTE ** pp)1777 size_t read_long_length_no_check(const BYTE** pp)
1778 {
1779 size_t b, l = 0;
1780 do { b = **pp; (*pp)++; l += b; } while (b==255);
1781 DEBUGLOG(6, "read_long_length_no_check: +length=%zu using %zu input bytes", l, l/255 + 1)
1782 return l;
1783 }
1784
1785 /* core decoder variant for LZ4_decompress_fast*()
1786 * for legacy support only : these entry points are deprecated.
1787 * - Presumes input is correctly formed (no defense vs malformed inputs)
1788 * - Does not know input size (presume input buffer is "large enough")
1789 * - Decompress a full block (only)
1790 * @return : nb of bytes read from input.
1791 * Note : this variant is not optimized for speed, just for maintenance.
1792 * the goal is to remove support of decompress_fast*() variants by v2.0
1793 **/
1794 LZ4_FORCE_INLINE int
LZ4_decompress_unsafe_generic(const BYTE * const istart,BYTE * const ostart,int decompressedSize,size_t prefixSize,const BYTE * const dictStart,const size_t dictSize)1795 LZ4_decompress_unsafe_generic(
1796 const BYTE* const istart,
1797 BYTE* const ostart,
1798 int decompressedSize,
1799
1800 size_t prefixSize,
1801 const BYTE* const dictStart, /* only if dict==usingExtDict */
1802 const size_t dictSize /* note: =0 if dictStart==NULL */
1803 )
1804 {
1805 const BYTE* ip = istart;
1806 BYTE* op = (BYTE*)ostart;
1807 BYTE* const oend = ostart + decompressedSize;
1808 const BYTE* const prefixStart = ostart - prefixSize;
1809
1810 DEBUGLOG(5, "LZ4_decompress_unsafe_generic");
1811 if (dictStart == NULL) assert(dictSize == 0);
1812
1813 while (1) {
1814 /* start new sequence */
1815 unsigned token = *ip++;
1816
1817 /* literals */
1818 { size_t ll = token >> ML_BITS;
1819 if (ll==15) {
1820 /* long literal length */
1821 ll += read_long_length_no_check(&ip);
1822 }
1823 if ((size_t)(oend-op) < ll) return -1; /* output buffer overflow */
1824 LZ4_memmove(op, ip, ll); /* support in-place decompression */
1825 op += ll;
1826 ip += ll;
1827 if ((size_t)(oend-op) < MFLIMIT) {
1828 if (op==oend) break; /* end of block */
1829 DEBUGLOG(5, "invalid: literals end at distance %zi from end of block", oend-op);
1830 /* incorrect end of block :
1831 * last match must start at least MFLIMIT==12 bytes before end of output block */
1832 return -1;
1833 } }
1834
1835 /* match */
1836 { size_t ml = token & 15;
1837 size_t const offset = LZ4_readLE16(ip);
1838 ip+=2;
1839
1840 if (ml==15) {
1841 /* long literal length */
1842 ml += read_long_length_no_check(&ip);
1843 }
1844 ml += MINMATCH;
1845
1846 if ((size_t)(oend-op) < ml) return -1; /* output buffer overflow */
1847
1848 { const BYTE* match = op - offset;
1849
1850 /* out of range */
1851 if (offset > (size_t)(op - prefixStart) + dictSize) {
1852 DEBUGLOG(6, "offset out of range");
1853 return -1;
1854 }
1855
1856 /* check special case : extDict */
1857 if (offset > (size_t)(op - prefixStart)) {
1858 /* extDict scenario */
1859 const BYTE* const dictEnd = dictStart + dictSize;
1860 const BYTE* extMatch = dictEnd - (offset - (size_t)(op-prefixStart));
1861 size_t const extml = (size_t)(dictEnd - extMatch);
1862 if (extml > ml) {
1863 /* match entirely within extDict */
1864 LZ4_memmove(op, extMatch, ml);
1865 op += ml;
1866 ml = 0;
1867 } else {
1868 /* match split between extDict & prefix */
1869 LZ4_memmove(op, extMatch, extml);
1870 op += extml;
1871 ml -= extml;
1872 }
1873 match = prefixStart;
1874 }
1875
1876 /* match copy - slow variant, supporting overlap copy */
1877 { size_t u;
1878 for (u=0; u<ml; u++) {
1879 op[u] = match[u];
1880 } } }
1881 op += ml;
1882 if ((size_t)(oend-op) < LASTLITERALS) {
1883 DEBUGLOG(5, "invalid: match ends at distance %zi from end of block", oend-op);
1884 /* incorrect end of block :
1885 * last match must stop at least LASTLITERALS==5 bytes before end of output block */
1886 return -1;
1887 }
1888 } /* match */
1889 } /* main loop */
1890 return (int)(ip - istart);
1891 }
1892
1893
1894 /* Read the variable-length literal or match length.
1895 *
1896 * @ip : input pointer
1897 * @ilimit : position after which if length is not decoded, the input is necessarily corrupted.
1898 * @initial_check - check ip >= ipmax before start of loop. Returns initial_error if so.
1899 * @error (output) - error code. Must be set to 0 before call.
1900 **/
1901 typedef size_t Rvl_t;
1902 static const Rvl_t rvl_error = (Rvl_t)(-1);
1903 LZ4_FORCE_INLINE Rvl_t
read_variable_length(const BYTE ** ip,const BYTE * ilimit,int initial_check)1904 read_variable_length(const BYTE** ip, const BYTE* ilimit,
1905 int initial_check)
1906 {
1907 Rvl_t s, length = 0;
1908 assert(ip != NULL);
1909 assert(*ip != NULL);
1910 assert(ilimit != NULL);
1911 if (initial_check && unlikely((*ip) >= ilimit)) { /* read limit reached */
1912 return rvl_error;
1913 }
1914 do {
1915 s = **ip;
1916 (*ip)++;
1917 length += s;
1918 if (unlikely((*ip) > ilimit)) { /* read limit reached */
1919 return rvl_error;
1920 }
1921 /* accumulator overflow detection (32-bit mode only) */
1922 if ((sizeof(length)<8) && unlikely(length > ((Rvl_t)(-1)/2)) ) {
1923 return rvl_error;
1924 }
1925 } while (s==255);
1926
1927 return length;
1928 }
1929
1930 /*! LZ4_decompress_generic() :
1931 * This generic decompression function covers all use cases.
1932 * It shall be instantiated several times, using different sets of directives.
1933 * Note that it is important for performance that this function really get inlined,
1934 * in order to remove useless branches during compilation optimization.
1935 */
1936 LZ4_FORCE_INLINE int
LZ4_decompress_generic(const char * const src,char * const dst,int srcSize,int outputSize,earlyEnd_directive partialDecoding,dict_directive dict,const BYTE * const lowPrefix,const BYTE * const dictStart,const size_t dictSize)1937 LZ4_decompress_generic(
1938 const char* const src,
1939 char* const dst,
1940 int srcSize,
1941 int outputSize, /* If endOnInput==endOnInputSize, this value is `dstCapacity` */
1942
1943 earlyEnd_directive partialDecoding, /* full, partial */
1944 dict_directive dict, /* noDict, withPrefix64k, usingExtDict */
1945 const BYTE* const lowPrefix, /* always <= dst, == dst when no prefix */
1946 const BYTE* const dictStart, /* only if dict==usingExtDict */
1947 const size_t dictSize /* note : = 0 if noDict */
1948 )
1949 {
1950 if ((src == NULL) || (outputSize < 0)) { return -1; }
1951
1952 { const BYTE* ip = (const BYTE*) src;
1953 const BYTE* const iend = ip + srcSize;
1954
1955 BYTE* op = (BYTE*) dst;
1956 BYTE* const oend = op + outputSize;
1957 BYTE* cpy;
1958
1959 const BYTE* const dictEnd = (dictStart == NULL) ? NULL : dictStart + dictSize;
1960
1961 const int checkOffset = (dictSize < (int)(64 KB));
1962
1963
1964 /* Set up the "end" pointers for the shortcut. */
1965 const BYTE* const shortiend = iend - 14 /*maxLL*/ - 2 /*offset*/;
1966 const BYTE* const shortoend = oend - 14 /*maxLL*/ - 18 /*maxML*/;
1967
1968 const BYTE* match;
1969 size_t offset;
1970 unsigned token;
1971 size_t length;
1972
1973
1974 DEBUGLOG(5, "LZ4_decompress_generic (srcSize:%i, dstSize:%i)", srcSize, outputSize);
1975
1976 /* Special cases */
1977 assert(lowPrefix <= op);
1978 if (unlikely(outputSize==0)) {
1979 /* Empty output buffer */
1980 if (partialDecoding) return 0;
1981 return ((srcSize==1) && (*ip==0)) ? 0 : -1;
1982 }
1983 if (unlikely(srcSize==0)) { return -1; }
1984
1985 /* LZ4_FAST_DEC_LOOP:
1986 * designed for modern OoO performance cpus,
1987 * where copying reliably 32-bytes is preferable to an unpredictable branch.
1988 * note : fast loop may show a regression for some client arm chips. */
1989 #if LZ4_FAST_DEC_LOOP
1990 if ((oend - op) < FASTLOOP_SAFE_DISTANCE) {
1991 DEBUGLOG(6, "skip fast decode loop");
1992 goto safe_decode;
1993 }
1994
1995 /* Fast loop : decode sequences as long as output < oend-FASTLOOP_SAFE_DISTANCE */
1996 while (1) {
1997 /* Main fastloop assertion: We can always wildcopy FASTLOOP_SAFE_DISTANCE */
1998 assert(oend - op >= FASTLOOP_SAFE_DISTANCE);
1999 assert(ip < iend);
2000 token = *ip++;
2001 length = token >> ML_BITS; /* literal length */
2002
2003 /* decode literal length */
2004 if (length == RUN_MASK) {
2005 size_t const addl = read_variable_length(&ip, iend-RUN_MASK, 1);
2006 if (addl == rvl_error) { goto _output_error; }
2007 length += addl;
2008 if (unlikely((uptrval)(op)+length<(uptrval)(op))) { goto _output_error; } /* overflow detection */
2009 if (unlikely((uptrval)(ip)+length<(uptrval)(ip))) { goto _output_error; } /* overflow detection */
2010
2011 /* copy literals */
2012 cpy = op+length;
2013 LZ4_STATIC_ASSERT(MFLIMIT >= WILDCOPYLENGTH);
2014 if ((cpy>oend-32) || (ip+length>iend-32)) { goto safe_literal_copy; }
2015 LZ4_wildCopy32(op, ip, cpy);
2016 ip += length; op = cpy;
2017 } else {
2018 cpy = op+length;
2019 DEBUGLOG(7, "copy %u bytes in a 16-bytes stripe", (unsigned)length);
2020 /* We don't need to check oend, since we check it once for each loop below */
2021 if (ip > iend-(16 + 1/*max lit + offset + nextToken*/)) { goto safe_literal_copy; }
2022 /* Literals can only be <= 14, but hope compilers optimize better when copy by a register size */
2023 LZ4_memcpy(op, ip, 16);
2024 ip += length; op = cpy;
2025 }
2026
2027 /* get offset */
2028 offset = LZ4_readLE16(ip); ip+=2;
2029 match = op - offset;
2030 assert(match <= op); /* overflow check */
2031
2032 /* get matchlength */
2033 length = token & ML_MASK;
2034
2035 if (length == ML_MASK) {
2036 size_t const addl = read_variable_length(&ip, iend - LASTLITERALS + 1, 0);
2037 if (addl == rvl_error) { goto _output_error; }
2038 length += addl;
2039 length += MINMATCH;
2040 if (unlikely((uptrval)(op)+length<(uptrval)op)) { goto _output_error; } /* overflow detection */
2041 if ((checkOffset) && (unlikely(match + dictSize < lowPrefix))) { goto _output_error; } /* Error : offset outside buffers */
2042 if (op + length >= oend - FASTLOOP_SAFE_DISTANCE) {
2043 goto safe_match_copy;
2044 }
2045 } else {
2046 length += MINMATCH;
2047 if (op + length >= oend - FASTLOOP_SAFE_DISTANCE) {
2048 goto safe_match_copy;
2049 }
2050
2051 /* Fastpath check: skip LZ4_wildCopy32 when true */
2052 if ((dict == withPrefix64k) || (match >= lowPrefix)) {
2053 if (offset >= 8) {
2054 assert(match >= lowPrefix);
2055 assert(match <= op);
2056 assert(op + 18 <= oend);
2057
2058 LZ4_memcpy(op, match, 8);
2059 LZ4_memcpy(op+8, match+8, 8);
2060 LZ4_memcpy(op+16, match+16, 2);
2061 op += length;
2062 continue;
2063 } } }
2064
2065 if (checkOffset && (unlikely(match + dictSize < lowPrefix))) { goto _output_error; } /* Error : offset outside buffers */
2066 /* match starting within external dictionary */
2067 if ((dict==usingExtDict) && (match < lowPrefix)) {
2068 assert(dictEnd != NULL);
2069 if (unlikely(op+length > oend-LASTLITERALS)) {
2070 if (partialDecoding) {
2071 DEBUGLOG(7, "partialDecoding: dictionary match, close to dstEnd");
2072 length = MIN(length, (size_t)(oend-op));
2073 } else {
2074 goto _output_error; /* end-of-block condition violated */
2075 } }
2076
2077 if (length <= (size_t)(lowPrefix-match)) {
2078 /* match fits entirely within external dictionary : just copy */
2079 LZ4_memmove(op, dictEnd - (lowPrefix-match), length);
2080 op += length;
2081 } else {
2082 /* match stretches into both external dictionary and current block */
2083 size_t const copySize = (size_t)(lowPrefix - match);
2084 size_t const restSize = length - copySize;
2085 LZ4_memcpy(op, dictEnd - copySize, copySize);
2086 op += copySize;
2087 if (restSize > (size_t)(op - lowPrefix)) { /* overlap copy */
2088 BYTE* const endOfMatch = op + restSize;
2089 const BYTE* copyFrom = lowPrefix;
2090 while (op < endOfMatch) { *op++ = *copyFrom++; }
2091 } else {
2092 LZ4_memcpy(op, lowPrefix, restSize);
2093 op += restSize;
2094 } }
2095 continue;
2096 }
2097
2098 /* copy match within block */
2099 cpy = op + length;
2100
2101 assert((op <= oend) && (oend-op >= 32));
2102 if (unlikely(offset<16)) {
2103 LZ4_memcpy_using_offset(op, match, cpy, offset);
2104 } else {
2105 LZ4_wildCopy32(op, match, cpy);
2106 }
2107
2108 op = cpy; /* wildcopy correction */
2109 }
2110 safe_decode:
2111 #endif
2112
2113 /* Main Loop : decode remaining sequences where output < FASTLOOP_SAFE_DISTANCE */
2114 while (1) {
2115 assert(ip < iend);
2116 token = *ip++;
2117 length = token >> ML_BITS; /* literal length */
2118
2119 /* A two-stage shortcut for the most common case:
2120 * 1) If the literal length is 0..14, and there is enough space,
2121 * enter the shortcut and copy 16 bytes on behalf of the literals
2122 * (in the fast mode, only 8 bytes can be safely copied this way).
2123 * 2) Further if the match length is 4..18, copy 18 bytes in a similar
2124 * manner; but we ensure that there's enough space in the output for
2125 * those 18 bytes earlier, upon entering the shortcut (in other words,
2126 * there is a combined check for both stages).
2127 */
2128 if ( (length != RUN_MASK)
2129 /* strictly "less than" on input, to re-enter the loop with at least one byte */
2130 && likely((ip < shortiend) & (op <= shortoend)) ) {
2131 /* Copy the literals */
2132 LZ4_memcpy(op, ip, 16);
2133 op += length; ip += length;
2134
2135 /* The second stage: prepare for match copying, decode full info.
2136 * If it doesn't work out, the info won't be wasted. */
2137 length = token & ML_MASK; /* match length */
2138 offset = LZ4_readLE16(ip); ip += 2;
2139 match = op - offset;
2140 assert(match <= op); /* check overflow */
2141
2142 /* Do not deal with overlapping matches. */
2143 if ( (length != ML_MASK)
2144 && (offset >= 8)
2145 && (dict==withPrefix64k || match >= lowPrefix) ) {
2146 /* Copy the match. */
2147 LZ4_memcpy(op + 0, match + 0, 8);
2148 LZ4_memcpy(op + 8, match + 8, 8);
2149 LZ4_memcpy(op +16, match +16, 2);
2150 op += length + MINMATCH;
2151 /* Both stages worked, load the next token. */
2152 continue;
2153 }
2154
2155 /* The second stage didn't work out, but the info is ready.
2156 * Propel it right to the point of match copying. */
2157 goto _copy_match;
2158 }
2159
2160 /* decode literal length */
2161 if (length == RUN_MASK) {
2162 size_t const addl = read_variable_length(&ip, iend-RUN_MASK, 1);
2163 if (addl == rvl_error) { goto _output_error; }
2164 length += addl;
2165 if (unlikely((uptrval)(op)+length<(uptrval)(op))) { goto _output_error; } /* overflow detection */
2166 if (unlikely((uptrval)(ip)+length<(uptrval)(ip))) { goto _output_error; } /* overflow detection */
2167 }
2168
2169 /* copy literals */
2170 cpy = op+length;
2171 #if LZ4_FAST_DEC_LOOP
2172 safe_literal_copy:
2173 #endif
2174 LZ4_STATIC_ASSERT(MFLIMIT >= WILDCOPYLENGTH);
2175 if ((cpy>oend-MFLIMIT) || (ip+length>iend-(2+1+LASTLITERALS))) {
2176 /* We've either hit the input parsing restriction or the output parsing restriction.
2177 * In the normal scenario, decoding a full block, it must be the last sequence,
2178 * otherwise it's an error (invalid input or dimensions).
2179 * In partialDecoding scenario, it's necessary to ensure there is no buffer overflow.
2180 */
2181 if (partialDecoding) {
2182 /* Since we are partial decoding we may be in this block because of the output parsing
2183 * restriction, which is not valid since the output buffer is allowed to be undersized.
2184 */
2185 DEBUGLOG(7, "partialDecoding: copying literals, close to input or output end")
2186 DEBUGLOG(7, "partialDecoding: literal length = %u", (unsigned)length);
2187 DEBUGLOG(7, "partialDecoding: remaining space in dstBuffer : %i", (int)(oend - op));
2188 DEBUGLOG(7, "partialDecoding: remaining space in srcBuffer : %i", (int)(iend - ip));
2189 /* Finishing in the middle of a literals segment,
2190 * due to lack of input.
2191 */
2192 if (ip+length > iend) {
2193 length = (size_t)(iend-ip);
2194 cpy = op + length;
2195 }
2196 /* Finishing in the middle of a literals segment,
2197 * due to lack of output space.
2198 */
2199 if (cpy > oend) {
2200 cpy = oend;
2201 assert(op<=oend);
2202 length = (size_t)(oend-op);
2203 }
2204 } else {
2205 /* We must be on the last sequence (or invalid) because of the parsing limitations
2206 * so check that we exactly consume the input and don't overrun the output buffer.
2207 */
2208 if ((ip+length != iend) || (cpy > oend)) {
2209 DEBUGLOG(6, "should have been last run of literals")
2210 DEBUGLOG(6, "ip(%p) + length(%i) = %p != iend (%p)", ip, (int)length, ip+length, iend);
2211 DEBUGLOG(6, "or cpy(%p) > oend(%p)", cpy, oend);
2212 goto _output_error;
2213 }
2214 }
2215 LZ4_memmove(op, ip, length); /* supports overlapping memory regions, for in-place decompression scenarios */
2216 ip += length;
2217 op += length;
2218 /* Necessarily EOF when !partialDecoding.
2219 * When partialDecoding, it is EOF if we've either
2220 * filled the output buffer or
2221 * can't proceed with reading an offset for following match.
2222 */
2223 if (!partialDecoding || (cpy == oend) || (ip >= (iend-2))) {
2224 break;
2225 }
2226 } else {
2227 LZ4_wildCopy8(op, ip, cpy); /* can overwrite up to 8 bytes beyond cpy */
2228 ip += length; op = cpy;
2229 }
2230
2231 /* get offset */
2232 offset = LZ4_readLE16(ip); ip+=2;
2233 match = op - offset;
2234
2235 /* get matchlength */
2236 length = token & ML_MASK;
2237
2238 _copy_match:
2239 if (length == ML_MASK) {
2240 size_t const addl = read_variable_length(&ip, iend - LASTLITERALS + 1, 0);
2241 if (addl == rvl_error) { goto _output_error; }
2242 length += addl;
2243 if (unlikely((uptrval)(op)+length<(uptrval)op)) goto _output_error; /* overflow detection */
2244 }
2245 length += MINMATCH;
2246
2247 #if LZ4_FAST_DEC_LOOP
2248 safe_match_copy:
2249 #endif
2250 if ((checkOffset) && (unlikely(match + dictSize < lowPrefix))) goto _output_error; /* Error : offset outside buffers */
2251 /* match starting within external dictionary */
2252 if ((dict==usingExtDict) && (match < lowPrefix)) {
2253 assert(dictEnd != NULL);
2254 if (unlikely(op+length > oend-LASTLITERALS)) {
2255 if (partialDecoding) length = MIN(length, (size_t)(oend-op));
2256 else goto _output_error; /* doesn't respect parsing restriction */
2257 }
2258
2259 if (length <= (size_t)(lowPrefix-match)) {
2260 /* match fits entirely within external dictionary : just copy */
2261 LZ4_memmove(op, dictEnd - (lowPrefix-match), length);
2262 op += length;
2263 } else {
2264 /* match stretches into both external dictionary and current block */
2265 size_t const copySize = (size_t)(lowPrefix - match);
2266 size_t const restSize = length - copySize;
2267 LZ4_memcpy(op, dictEnd - copySize, copySize);
2268 op += copySize;
2269 if (restSize > (size_t)(op - lowPrefix)) { /* overlap copy */
2270 BYTE* const endOfMatch = op + restSize;
2271 const BYTE* copyFrom = lowPrefix;
2272 while (op < endOfMatch) *op++ = *copyFrom++;
2273 } else {
2274 LZ4_memcpy(op, lowPrefix, restSize);
2275 op += restSize;
2276 } }
2277 continue;
2278 }
2279 assert(match >= lowPrefix);
2280
2281 /* copy match within block */
2282 cpy = op + length;
2283
2284 /* partialDecoding : may end anywhere within the block */
2285 assert(op<=oend);
2286 if (partialDecoding && (cpy > oend-MATCH_SAFEGUARD_DISTANCE)) {
2287 size_t const mlen = MIN(length, (size_t)(oend-op));
2288 const BYTE* const matchEnd = match + mlen;
2289 BYTE* const copyEnd = op + mlen;
2290 if (matchEnd > op) { /* overlap copy */
2291 while (op < copyEnd) { *op++ = *match++; }
2292 } else {
2293 LZ4_memcpy(op, match, mlen);
2294 }
2295 op = copyEnd;
2296 if (op == oend) { break; }
2297 continue;
2298 }
2299
2300 if (unlikely(offset<8)) {
2301 LZ4_write32(op, 0); /* silence msan warning when offset==0 */
2302 op[0] = match[0];
2303 op[1] = match[1];
2304 op[2] = match[2];
2305 op[3] = match[3];
2306 match += inc32table[offset];
2307 LZ4_memcpy(op+4, match, 4);
2308 match -= dec64table[offset];
2309 } else {
2310 LZ4_memcpy(op, match, 8);
2311 match += 8;
2312 }
2313 op += 8;
2314
2315 if (unlikely(cpy > oend-MATCH_SAFEGUARD_DISTANCE)) {
2316 BYTE* const oCopyLimit = oend - (WILDCOPYLENGTH-1);
2317 if (cpy > oend-LASTLITERALS) { goto _output_error; } /* Error : last LASTLITERALS bytes must be literals (uncompressed) */
2318 if (op < oCopyLimit) {
2319 LZ4_wildCopy8(op, match, oCopyLimit);
2320 match += oCopyLimit - op;
2321 op = oCopyLimit;
2322 }
2323 while (op < cpy) { *op++ = *match++; }
2324 } else {
2325 LZ4_memcpy(op, match, 8);
2326 if (length > 16) { LZ4_wildCopy8(op+8, match+8, cpy); }
2327 }
2328 op = cpy; /* wildcopy correction */
2329 }
2330
2331 /* end of decoding */
2332 DEBUGLOG(5, "decoded %i bytes", (int) (((char*)op)-dst));
2333 return (int) (((char*)op)-dst); /* Nb of output bytes decoded */
2334
2335 /* Overflow error detected */
2336 _output_error:
2337 return (int) (-(((const char*)ip)-src))-1;
2338 }
2339 }
2340
2341
2342 /*===== Instantiate the API decoding functions. =====*/
2343
2344 LZ4_FORCE_O2
LZ4_decompress_safe(const char * source,char * dest,int compressedSize,int maxDecompressedSize)2345 int LZ4_decompress_safe(const char* source, char* dest, int compressedSize, int maxDecompressedSize)
2346 {
2347 return LZ4_decompress_generic(source, dest, compressedSize, maxDecompressedSize,
2348 decode_full_block, noDict,
2349 (BYTE*)dest, NULL, 0);
2350 }
2351
2352 LZ4_FORCE_O2
LZ4_decompress_safe_partial(const char * src,char * dst,int compressedSize,int targetOutputSize,int dstCapacity)2353 int LZ4_decompress_safe_partial(const char* src, char* dst, int compressedSize, int targetOutputSize, int dstCapacity)
2354 {
2355 dstCapacity = MIN(targetOutputSize, dstCapacity);
2356 return LZ4_decompress_generic(src, dst, compressedSize, dstCapacity,
2357 partial_decode,
2358 noDict, (BYTE*)dst, NULL, 0);
2359 }
2360
2361 LZ4_FORCE_O2
LZ4_decompress_fast(const char * source,char * dest,int originalSize)2362 int LZ4_decompress_fast(const char* source, char* dest, int originalSize)
2363 {
2364 DEBUGLOG(5, "LZ4_decompress_fast");
2365 return LZ4_decompress_unsafe_generic(
2366 (const BYTE*)source, (BYTE*)dest, originalSize,
2367 0, NULL, 0);
2368 }
2369
2370 /*===== Instantiate a few more decoding cases, used more than once. =====*/
2371
2372 LZ4_FORCE_O2 /* Exported, an obsolete API function. */
LZ4_decompress_safe_withPrefix64k(const char * source,char * dest,int compressedSize,int maxOutputSize)2373 int LZ4_decompress_safe_withPrefix64k(const char* source, char* dest, int compressedSize, int maxOutputSize)
2374 {
2375 return LZ4_decompress_generic(source, dest, compressedSize, maxOutputSize,
2376 decode_full_block, withPrefix64k,
2377 (BYTE*)dest - 64 KB, NULL, 0);
2378 }
2379
2380 LZ4_FORCE_O2
LZ4_decompress_safe_partial_withPrefix64k(const char * source,char * dest,int compressedSize,int targetOutputSize,int dstCapacity)2381 static int LZ4_decompress_safe_partial_withPrefix64k(const char* source, char* dest, int compressedSize, int targetOutputSize, int dstCapacity)
2382 {
2383 dstCapacity = MIN(targetOutputSize, dstCapacity);
2384 return LZ4_decompress_generic(source, dest, compressedSize, dstCapacity,
2385 partial_decode, withPrefix64k,
2386 (BYTE*)dest - 64 KB, NULL, 0);
2387 }
2388
2389 /* Another obsolete API function, paired with the previous one. */
LZ4_decompress_fast_withPrefix64k(const char * source,char * dest,int originalSize)2390 int LZ4_decompress_fast_withPrefix64k(const char* source, char* dest, int originalSize)
2391 {
2392 return LZ4_decompress_unsafe_generic(
2393 (const BYTE*)source, (BYTE*)dest, originalSize,
2394 64 KB, NULL, 0);
2395 }
2396
2397 LZ4_FORCE_O2
LZ4_decompress_safe_withSmallPrefix(const char * source,char * dest,int compressedSize,int maxOutputSize,size_t prefixSize)2398 static int LZ4_decompress_safe_withSmallPrefix(const char* source, char* dest, int compressedSize, int maxOutputSize,
2399 size_t prefixSize)
2400 {
2401 return LZ4_decompress_generic(source, dest, compressedSize, maxOutputSize,
2402 decode_full_block, noDict,
2403 (BYTE*)dest-prefixSize, NULL, 0);
2404 }
2405
2406 LZ4_FORCE_O2
LZ4_decompress_safe_partial_withSmallPrefix(const char * source,char * dest,int compressedSize,int targetOutputSize,int dstCapacity,size_t prefixSize)2407 static int LZ4_decompress_safe_partial_withSmallPrefix(const char* source, char* dest, int compressedSize, int targetOutputSize, int dstCapacity,
2408 size_t prefixSize)
2409 {
2410 dstCapacity = MIN(targetOutputSize, dstCapacity);
2411 return LZ4_decompress_generic(source, dest, compressedSize, dstCapacity,
2412 partial_decode, noDict,
2413 (BYTE*)dest-prefixSize, NULL, 0);
2414 }
2415
2416 LZ4_FORCE_O2
LZ4_decompress_safe_forceExtDict(const char * source,char * dest,int compressedSize,int maxOutputSize,const void * dictStart,size_t dictSize)2417 int LZ4_decompress_safe_forceExtDict(const char* source, char* dest,
2418 int compressedSize, int maxOutputSize,
2419 const void* dictStart, size_t dictSize)
2420 {
2421 return LZ4_decompress_generic(source, dest, compressedSize, maxOutputSize,
2422 decode_full_block, usingExtDict,
2423 (BYTE*)dest, (const BYTE*)dictStart, dictSize);
2424 }
2425
2426 LZ4_FORCE_O2
LZ4_decompress_safe_partial_forceExtDict(const char * source,char * dest,int compressedSize,int targetOutputSize,int dstCapacity,const void * dictStart,size_t dictSize)2427 int LZ4_decompress_safe_partial_forceExtDict(const char* source, char* dest,
2428 int compressedSize, int targetOutputSize, int dstCapacity,
2429 const void* dictStart, size_t dictSize)
2430 {
2431 dstCapacity = MIN(targetOutputSize, dstCapacity);
2432 return LZ4_decompress_generic(source, dest, compressedSize, dstCapacity,
2433 partial_decode, usingExtDict,
2434 (BYTE*)dest, (const BYTE*)dictStart, dictSize);
2435 }
2436
2437 LZ4_FORCE_O2
LZ4_decompress_fast_extDict(const char * source,char * dest,int originalSize,const void * dictStart,size_t dictSize)2438 static int LZ4_decompress_fast_extDict(const char* source, char* dest, int originalSize,
2439 const void* dictStart, size_t dictSize)
2440 {
2441 return LZ4_decompress_unsafe_generic(
2442 (const BYTE*)source, (BYTE*)dest, originalSize,
2443 0, (const BYTE*)dictStart, dictSize);
2444 }
2445
2446 /* The "double dictionary" mode, for use with e.g. ring buffers: the first part
2447 * of the dictionary is passed as prefix, and the second via dictStart + dictSize.
2448 * These routines are used only once, in LZ4_decompress_*_continue().
2449 */
2450 LZ4_FORCE_INLINE
LZ4_decompress_safe_doubleDict(const char * source,char * dest,int compressedSize,int maxOutputSize,size_t prefixSize,const void * dictStart,size_t dictSize)2451 int LZ4_decompress_safe_doubleDict(const char* source, char* dest, int compressedSize, int maxOutputSize,
2452 size_t prefixSize, const void* dictStart, size_t dictSize)
2453 {
2454 return LZ4_decompress_generic(source, dest, compressedSize, maxOutputSize,
2455 decode_full_block, usingExtDict,
2456 (BYTE*)dest-prefixSize, (const BYTE*)dictStart, dictSize);
2457 }
2458
2459 /*===== streaming decompression functions =====*/
2460
2461 #if !defined(LZ4_STATIC_LINKING_ONLY_DISABLE_MEMORY_ALLOCATION)
LZ4_createStreamDecode(void)2462 LZ4_streamDecode_t* LZ4_createStreamDecode(void)
2463 {
2464 LZ4_STATIC_ASSERT(sizeof(LZ4_streamDecode_t) >= sizeof(LZ4_streamDecode_t_internal));
2465 return (LZ4_streamDecode_t*) ALLOC_AND_ZERO(sizeof(LZ4_streamDecode_t));
2466 }
2467
LZ4_freeStreamDecode(LZ4_streamDecode_t * LZ4_stream)2468 int LZ4_freeStreamDecode (LZ4_streamDecode_t* LZ4_stream)
2469 {
2470 if (LZ4_stream == NULL) { return 0; } /* support free on NULL */
2471 FREEMEM(LZ4_stream);
2472 return 0;
2473 }
2474 #endif
2475
2476 /*! LZ4_setStreamDecode() :
2477 * Use this function to instruct where to find the dictionary.
2478 * This function is not necessary if previous data is still available where it was decoded.
2479 * Loading a size of 0 is allowed (same effect as no dictionary).
2480 * @return : 1 if OK, 0 if error
2481 */
LZ4_setStreamDecode(LZ4_streamDecode_t * LZ4_streamDecode,const char * dictionary,int dictSize)2482 int LZ4_setStreamDecode (LZ4_streamDecode_t* LZ4_streamDecode, const char* dictionary, int dictSize)
2483 {
2484 LZ4_streamDecode_t_internal* lz4sd = &LZ4_streamDecode->internal_donotuse;
2485 lz4sd->prefixSize = (size_t)dictSize;
2486 if (dictSize) {
2487 assert(dictionary != NULL);
2488 lz4sd->prefixEnd = (const BYTE*) dictionary + dictSize;
2489 } else {
2490 lz4sd->prefixEnd = (const BYTE*) dictionary;
2491 }
2492 lz4sd->externalDict = NULL;
2493 lz4sd->extDictSize = 0;
2494 return 1;
2495 }
2496
2497 /*! LZ4_decoderRingBufferSize() :
2498 * when setting a ring buffer for streaming decompression (optional scenario),
2499 * provides the minimum size of this ring buffer
2500 * to be compatible with any source respecting maxBlockSize condition.
2501 * Note : in a ring buffer scenario,
2502 * blocks are presumed decompressed next to each other.
2503 * When not enough space remains for next block (remainingSize < maxBlockSize),
2504 * decoding resumes from beginning of ring buffer.
2505 * @return : minimum ring buffer size,
2506 * or 0 if there is an error (invalid maxBlockSize).
2507 */
LZ4_decoderRingBufferSize(int maxBlockSize)2508 int LZ4_decoderRingBufferSize(int maxBlockSize)
2509 {
2510 if (maxBlockSize < 0) return 0;
2511 if (maxBlockSize > LZ4_MAX_INPUT_SIZE) return 0;
2512 if (maxBlockSize < 16) maxBlockSize = 16;
2513 return LZ4_DECODER_RING_BUFFER_SIZE(maxBlockSize);
2514 }
2515
2516 /*
2517 *_continue() :
2518 These decoding functions allow decompression of multiple blocks in "streaming" mode.
2519 Previously decoded blocks must still be available at the memory position where they were decoded.
2520 If it's not possible, save the relevant part of decoded data into a safe buffer,
2521 and indicate where it stands using LZ4_setStreamDecode()
2522 */
2523 LZ4_FORCE_O2
LZ4_decompress_safe_continue(LZ4_streamDecode_t * LZ4_streamDecode,const char * source,char * dest,int compressedSize,int maxOutputSize)2524 int LZ4_decompress_safe_continue (LZ4_streamDecode_t* LZ4_streamDecode, const char* source, char* dest, int compressedSize, int maxOutputSize)
2525 {
2526 LZ4_streamDecode_t_internal* lz4sd = &LZ4_streamDecode->internal_donotuse;
2527 int result;
2528
2529 if (lz4sd->prefixSize == 0) {
2530 /* The first call, no dictionary yet. */
2531 assert(lz4sd->extDictSize == 0);
2532 result = LZ4_decompress_safe(source, dest, compressedSize, maxOutputSize);
2533 if (result <= 0) return result;
2534 lz4sd->prefixSize = (size_t)result;
2535 lz4sd->prefixEnd = (BYTE*)dest + result;
2536 } else if (lz4sd->prefixEnd == (BYTE*)dest) {
2537 /* They're rolling the current segment. */
2538 if (lz4sd->prefixSize >= 64 KB - 1)
2539 result = LZ4_decompress_safe_withPrefix64k(source, dest, compressedSize, maxOutputSize);
2540 else if (lz4sd->extDictSize == 0)
2541 result = LZ4_decompress_safe_withSmallPrefix(source, dest, compressedSize, maxOutputSize,
2542 lz4sd->prefixSize);
2543 else
2544 result = LZ4_decompress_safe_doubleDict(source, dest, compressedSize, maxOutputSize,
2545 lz4sd->prefixSize, lz4sd->externalDict, lz4sd->extDictSize);
2546 if (result <= 0) return result;
2547 lz4sd->prefixSize += (size_t)result;
2548 lz4sd->prefixEnd += result;
2549 } else {
2550 /* The buffer wraps around, or they're switching to another buffer. */
2551 lz4sd->extDictSize = lz4sd->prefixSize;
2552 lz4sd->externalDict = lz4sd->prefixEnd - lz4sd->extDictSize;
2553 result = LZ4_decompress_safe_forceExtDict(source, dest, compressedSize, maxOutputSize,
2554 lz4sd->externalDict, lz4sd->extDictSize);
2555 if (result <= 0) return result;
2556 lz4sd->prefixSize = (size_t)result;
2557 lz4sd->prefixEnd = (BYTE*)dest + result;
2558 }
2559
2560 return result;
2561 }
2562
2563 LZ4_FORCE_O2 int
LZ4_decompress_fast_continue(LZ4_streamDecode_t * LZ4_streamDecode,const char * source,char * dest,int originalSize)2564 LZ4_decompress_fast_continue (LZ4_streamDecode_t* LZ4_streamDecode,
2565 const char* source, char* dest, int originalSize)
2566 {
2567 LZ4_streamDecode_t_internal* const lz4sd =
2568 (assert(LZ4_streamDecode!=NULL), &LZ4_streamDecode->internal_donotuse);
2569 int result;
2570
2571 DEBUGLOG(5, "LZ4_decompress_fast_continue (toDecodeSize=%i)", originalSize);
2572 assert(originalSize >= 0);
2573
2574 if (lz4sd->prefixSize == 0) {
2575 DEBUGLOG(5, "first invocation : no prefix nor extDict");
2576 assert(lz4sd->extDictSize == 0);
2577 result = LZ4_decompress_fast(source, dest, originalSize);
2578 if (result <= 0) return result;
2579 lz4sd->prefixSize = (size_t)originalSize;
2580 lz4sd->prefixEnd = (BYTE*)dest + originalSize;
2581 } else if (lz4sd->prefixEnd == (BYTE*)dest) {
2582 DEBUGLOG(5, "continue using existing prefix");
2583 result = LZ4_decompress_unsafe_generic(
2584 (const BYTE*)source, (BYTE*)dest, originalSize,
2585 lz4sd->prefixSize,
2586 lz4sd->externalDict, lz4sd->extDictSize);
2587 if (result <= 0) return result;
2588 lz4sd->prefixSize += (size_t)originalSize;
2589 lz4sd->prefixEnd += originalSize;
2590 } else {
2591 DEBUGLOG(5, "prefix becomes extDict");
2592 lz4sd->extDictSize = lz4sd->prefixSize;
2593 lz4sd->externalDict = lz4sd->prefixEnd - lz4sd->extDictSize;
2594 result = LZ4_decompress_fast_extDict(source, dest, originalSize,
2595 lz4sd->externalDict, lz4sd->extDictSize);
2596 if (result <= 0) return result;
2597 lz4sd->prefixSize = (size_t)originalSize;
2598 lz4sd->prefixEnd = (BYTE*)dest + originalSize;
2599 }
2600
2601 return result;
2602 }
2603
2604
2605 /*
2606 Advanced decoding functions :
2607 *_usingDict() :
2608 These decoding functions work the same as "_continue" ones,
2609 the dictionary must be explicitly provided within parameters
2610 */
2611
LZ4_decompress_safe_usingDict(const char * source,char * dest,int compressedSize,int maxOutputSize,const char * dictStart,int dictSize)2612 int LZ4_decompress_safe_usingDict(const char* source, char* dest, int compressedSize, int maxOutputSize, const char* dictStart, int dictSize)
2613 {
2614 if (dictSize==0)
2615 return LZ4_decompress_safe(source, dest, compressedSize, maxOutputSize);
2616 if (dictStart+dictSize == dest) {
2617 if (dictSize >= 64 KB - 1) {
2618 return LZ4_decompress_safe_withPrefix64k(source, dest, compressedSize, maxOutputSize);
2619 }
2620 assert(dictSize >= 0);
2621 return LZ4_decompress_safe_withSmallPrefix(source, dest, compressedSize, maxOutputSize, (size_t)dictSize);
2622 }
2623 assert(dictSize >= 0);
2624 return LZ4_decompress_safe_forceExtDict(source, dest, compressedSize, maxOutputSize, dictStart, (size_t)dictSize);
2625 }
2626
LZ4_decompress_safe_partial_usingDict(const char * source,char * dest,int compressedSize,int targetOutputSize,int dstCapacity,const char * dictStart,int dictSize)2627 int LZ4_decompress_safe_partial_usingDict(const char* source, char* dest, int compressedSize, int targetOutputSize, int dstCapacity, const char* dictStart, int dictSize)
2628 {
2629 if (dictSize==0)
2630 return LZ4_decompress_safe_partial(source, dest, compressedSize, targetOutputSize, dstCapacity);
2631 if (dictStart+dictSize == dest) {
2632 if (dictSize >= 64 KB - 1) {
2633 return LZ4_decompress_safe_partial_withPrefix64k(source, dest, compressedSize, targetOutputSize, dstCapacity);
2634 }
2635 assert(dictSize >= 0);
2636 return LZ4_decompress_safe_partial_withSmallPrefix(source, dest, compressedSize, targetOutputSize, dstCapacity, (size_t)dictSize);
2637 }
2638 assert(dictSize >= 0);
2639 return LZ4_decompress_safe_partial_forceExtDict(source, dest, compressedSize, targetOutputSize, dstCapacity, dictStart, (size_t)dictSize);
2640 }
2641
LZ4_decompress_fast_usingDict(const char * source,char * dest,int originalSize,const char * dictStart,int dictSize)2642 int LZ4_decompress_fast_usingDict(const char* source, char* dest, int originalSize, const char* dictStart, int dictSize)
2643 {
2644 if (dictSize==0 || dictStart+dictSize == dest)
2645 return LZ4_decompress_unsafe_generic(
2646 (const BYTE*)source, (BYTE*)dest, originalSize,
2647 (size_t)dictSize, NULL, 0);
2648 assert(dictSize >= 0);
2649 return LZ4_decompress_fast_extDict(source, dest, originalSize, dictStart, (size_t)dictSize);
2650 }
2651
2652
2653 /*=*************************************************
2654 * Obsolete Functions
2655 ***************************************************/
2656 /* obsolete compression functions */
LZ4_compress_limitedOutput(const char * source,char * dest,int inputSize,int maxOutputSize)2657 int LZ4_compress_limitedOutput(const char* source, char* dest, int inputSize, int maxOutputSize)
2658 {
2659 return LZ4_compress_default(source, dest, inputSize, maxOutputSize);
2660 }
LZ4_compress(const char * src,char * dest,int srcSize)2661 int LZ4_compress(const char* src, char* dest, int srcSize)
2662 {
2663 return LZ4_compress_default(src, dest, srcSize, LZ4_compressBound(srcSize));
2664 }
LZ4_compress_limitedOutput_withState(void * state,const char * src,char * dst,int srcSize,int dstSize)2665 int LZ4_compress_limitedOutput_withState (void* state, const char* src, char* dst, int srcSize, int dstSize)
2666 {
2667 return LZ4_compress_fast_extState(state, src, dst, srcSize, dstSize, 1);
2668 }
LZ4_compress_withState(void * state,const char * src,char * dst,int srcSize)2669 int LZ4_compress_withState (void* state, const char* src, char* dst, int srcSize)
2670 {
2671 return LZ4_compress_fast_extState(state, src, dst, srcSize, LZ4_compressBound(srcSize), 1);
2672 }
LZ4_compress_limitedOutput_continue(LZ4_stream_t * LZ4_stream,const char * src,char * dst,int srcSize,int dstCapacity)2673 int LZ4_compress_limitedOutput_continue (LZ4_stream_t* LZ4_stream, const char* src, char* dst, int srcSize, int dstCapacity)
2674 {
2675 return LZ4_compress_fast_continue(LZ4_stream, src, dst, srcSize, dstCapacity, 1);
2676 }
LZ4_compress_continue(LZ4_stream_t * LZ4_stream,const char * source,char * dest,int inputSize)2677 int LZ4_compress_continue (LZ4_stream_t* LZ4_stream, const char* source, char* dest, int inputSize)
2678 {
2679 return LZ4_compress_fast_continue(LZ4_stream, source, dest, inputSize, LZ4_compressBound(inputSize), 1);
2680 }
2681
2682 /*
2683 These decompression functions are deprecated and should no longer be used.
2684 They are only provided here for compatibility with older user programs.
2685 - LZ4_uncompress is totally equivalent to LZ4_decompress_fast
2686 - LZ4_uncompress_unknownOutputSize is totally equivalent to LZ4_decompress_safe
2687 */
LZ4_uncompress(const char * source,char * dest,int outputSize)2688 int LZ4_uncompress (const char* source, char* dest, int outputSize)
2689 {
2690 return LZ4_decompress_fast(source, dest, outputSize);
2691 }
LZ4_uncompress_unknownOutputSize(const char * source,char * dest,int isize,int maxOutputSize)2692 int LZ4_uncompress_unknownOutputSize (const char* source, char* dest, int isize, int maxOutputSize)
2693 {
2694 return LZ4_decompress_safe(source, dest, isize, maxOutputSize);
2695 }
2696
2697 /* Obsolete Streaming functions */
2698
LZ4_sizeofStreamState(void)2699 int LZ4_sizeofStreamState(void) { return sizeof(LZ4_stream_t); }
2700
LZ4_resetStreamState(void * state,char * inputBuffer)2701 int LZ4_resetStreamState(void* state, char* inputBuffer)
2702 {
2703 (void)inputBuffer;
2704 LZ4_resetStream((LZ4_stream_t*)state);
2705 return 0;
2706 }
2707
2708 #if !defined(LZ4_STATIC_LINKING_ONLY_DISABLE_MEMORY_ALLOCATION)
LZ4_create(char * inputBuffer)2709 void* LZ4_create (char* inputBuffer)
2710 {
2711 (void)inputBuffer;
2712 return LZ4_createStream();
2713 }
2714 #endif
2715
LZ4_slideInputBuffer(void * state)2716 char* LZ4_slideInputBuffer (void* state)
2717 {
2718 /* avoid const char * -> char * conversion warning */
2719 return (char *)(uptrval)((LZ4_stream_t*)state)->internal_donotuse.dictionary;
2720 }
2721
2722 #endif /* LZ4_COMMONDEFS_ONLY */
2723