2 * LZ4 - Fast LZ compression algorithm
4 * Copyright (C) 2011-2013, Yann Collet.
5 * BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions are
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
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.
30 * You can contact the author at :
31 * - LZ4 homepage : http://fastcompression.blogspot.com/p/lz4.html
32 * - LZ4 source repository : http://code.google.com/p/lz4/
35 #include <sys/types.h>
36 #include <sys/byteorder.h>
41 size_t lz4_compress(void *, void *, size_t, size_t, int);
42 int lz4_decompress(void *, void *, size_t, size_t, int);
43 static int real_LZ4_compress(const char *source
, char *dest
, int isize
,
45 static int LZ4_uncompress_unknownOutputSize(const char *source
, char *dest
,
46 int isize
, int maxOutputSize
);
47 static int LZ4_compressCtx(void *ctx
, const char *source
, char *dest
,
48 int isize
, int osize
);
49 static int LZ4_compress64kCtx(void *ctx
, const char *source
, char *dest
,
50 int isize
, int osize
);
54 lz4_compress(void *s_start
, void *d_start
, size_t s_len
, size_t d_len
, int n
)
59 assert(d_len
>= sizeof (bufsiz
));
61 bufsiz
= real_LZ4_compress(s_start
, &dest
[sizeof (bufsiz
)], s_len
,
62 d_len
- sizeof (bufsiz
));
64 /* Signal an error if the compression routine returned zero. */
69 * Encode the compresed buffer size at the start. We'll need this in
70 * decompression to counter the effects of padding which might be
71 * added to the compressed buffer and which, if unhandled, would
72 * confuse the hell out of our decompression function.
74 *(uint32_t *)dest
= BE_32(bufsiz
);
76 return (bufsiz
+ sizeof (bufsiz
));
81 lz4_decompress(void *s_start
, void *d_start
, size_t s_len
, size_t d_len
, int n
)
83 const char *src
= s_start
;
84 uint32_t bufsiz
= BE_IN32(src
);
86 /* invalid compressed buffer size encoded at start */
87 if (bufsiz
+ sizeof (bufsiz
) > s_len
)
91 * Returns 0 on success (decompression function returned non-negative)
92 * and non-zero on failure (decompression function returned negative.
94 return (LZ4_uncompress_unknownOutputSize(&src
[sizeof (bufsiz
)],
95 d_start
, bufsiz
, d_len
) < 0);
99 * LZ4 API Description:
102 * real_LZ4_compress() :
103 * isize : is the input size. Max supported value is ~1.9GB
104 * return : the number of bytes written in buffer dest
105 * or 0 if the compression fails (if LZ4_COMPRESSMIN is set).
106 * note : destination buffer must be already allocated.
107 * destination buffer must be sized to handle worst cases
108 * situations (input data not compressible)
112 * LZ4_uncompress_unknownOutputSize() :
113 * isize : is the input size, therefore the compressed size
114 * maxOutputSize : is the size of the destination buffer (which must be
116 * return : the number of bytes decoded in the destination buffer
117 * (necessarily <= maxOutputSize). If the source stream is
118 * malformed, the function will stop decoding and return a
119 * negative result, indicating the byte position of the faulty
120 * instruction. This function never writes beyond dest +
121 * maxOutputSize, and is therefore protected against malicious
123 * note : Destination buffer must be already allocated.
125 * LZ4_compressCtx() :
126 * This function explicitly handles the CTX memory structure.
128 * ILLUMOS CHANGES: the CTX memory structure must be explicitly allocated
129 * by the caller (either on the stack or using kmem_zalloc). Passing NULL
132 * LZ4_compress64kCtx() :
133 * Same as LZ4_compressCtx(), but specific to small inputs (<64KB).
134 * isize *Must* be <64KB, otherwise the output will be corrupted.
136 * ILLUMOS CHANGES: the CTX memory structure must be explicitly allocated
137 * by the caller (either on the stack or using kmem_zalloc). Passing NULL
146 * COMPRESSIONLEVEL: Increasing this value improves compression ratio
147 * Lowering this value reduces memory usage. Reduced memory usage
148 * typically improves speed, due to cache effect (ex: L1 32KB for Intel,
149 * L1 64KB for AMD). Memory usage formula : N->2^(N+2) Bytes
150 * (examples : 12 -> 16KB ; 17 -> 512KB)
152 #define COMPRESSIONLEVEL 12
155 * NOTCOMPRESSIBLE_CONFIRMATION: Decreasing this value will make the
156 * algorithm skip faster data segments considered "incompressible".
157 * This may decrease compression ratio dramatically, but will be
158 * faster on incompressible data. Increasing this value will make
159 * the algorithm search more before declaring a segment "incompressible".
160 * This could improve compression a bit, but will be slower on
161 * incompressible data. The default value (6) is recommended.
163 #define NOTCOMPRESSIBLE_CONFIRMATION 6
166 * BIG_ENDIAN_NATIVE_BUT_INCOMPATIBLE: This will provide a boost to
167 * performance for big endian cpu, but the resulting compressed stream
168 * will be incompatible with little-endian CPU. You can set this option
169 * to 1 in situations where data will stay within closed environment.
170 * This option is useless on Little_Endian CPU (such as x86).
172 /* #define BIG_ENDIAN_NATIVE_BUT_INCOMPATIBLE 1 */
175 * CPU Feature Detection
178 /* 32 or 64 bits ? */
179 #if (defined(__x86_64__) || defined(__x86_64) || defined(__amd64__) || \
180 defined(__amd64) || defined(__ppc64__) || defined(_WIN64) || \
181 defined(__LP64__) || defined(_LP64))
188 * Limits the amount of stack space that the algorithm may consume to hold
189 * the compression lookup table. The value `9' here means we'll never use
190 * more than 2k of stack (see above for a description of COMPRESSIONLEVEL).
191 * If more memory is needed, it is allocated from the heap.
196 * Little Endian or Big Endian?
197 * Note: overwrite the below #define if you know your architecture endianess.
199 #if (defined(__BIG_ENDIAN__) || defined(__BIG_ENDIAN) || \
200 defined(_BIG_ENDIAN) || defined(_ARCH_PPC) || defined(__PPC__) || \
201 defined(__PPC) || defined(PPC) || defined(__powerpc__) || \
202 defined(__powerpc) || defined(powerpc) || \
203 ((defined(__BYTE_ORDER__)&&(__BYTE_ORDER__ == __ORDER_BIG_ENDIAN__))))
204 #define LZ4_BIG_ENDIAN 1
207 * Little Endian assumed. PDP Endian and other very rare endian format
213 * Unaligned memory access is automatically enabled for "common" CPU,
214 * such as x86. For others CPU, the compiler will be more cautious, and
215 * insert extra code to ensure aligned access is respected. If you know
216 * your target CPU supports unaligned memory access, you may want to
217 * force this option manually to improve performance
219 #if defined(__ARM_FEATURE_UNALIGNED)
220 #define LZ4_FORCE_UNALIGNED_ACCESS 1
227 #if __STDC_VERSION__ >= 199901L /* C99 */
228 /* "restrict" is a known keyword */
230 /* Disable restrict */
234 #define GCC_VERSION (__GNUC__ * 100 + __GNUC_MINOR__)
238 /* Visual is not C99, but supports some kind of inline */
239 #define inline __forceinline
241 /* For Visual 2005 */
242 #pragma intrinsic(_BitScanForward64)
243 #pragma intrinsic(_BitScanReverse64)
244 #else /* !LZ4_ARCH64 */
245 /* For Visual 2005 */
246 #pragma intrinsic(_BitScanForward)
247 #pragma intrinsic(_BitScanReverse)
248 #endif /* !LZ4_ARCH64 */
249 #endif /* _MSC_VER */
252 #define lz4_bswap16(x) _byteswap_ushort(x)
253 #else /* !_MSC_VER */
254 #define lz4_bswap16(x) ((unsigned short int) ((((x) >> 8) & 0xffu) | \
255 (((x) & 0xffu) << 8)))
256 #endif /* !_MSC_VER */
258 #if (GCC_VERSION >= 302) || (__INTEL_COMPILER >= 800) || defined(__clang__)
259 #define expect(expr, value) (__builtin_expect((expr), (value)))
261 #define expect(expr, value) (expr)
264 #define likely(expr) expect((expr) != 0, 1)
265 #define unlikely(expr) expect((expr) != 0, 0)
268 #if defined(_MSC_VER)
269 /* Visual Studio does not support 'stdint' natively */
270 #define BYTE unsigned __int8
271 #define U16 unsigned __int16
272 #define U32 unsigned __int32
274 #define U64 unsigned __int64
275 #else /* !defined(_MSC_VER) */
281 #endif /* !defined(_MSC_VER) */
283 #ifndef LZ4_FORCE_UNALIGNED_ACCESS
287 typedef struct _U16_S
{
290 typedef struct _U32_S
{
293 typedef struct _U64_S
{
297 #ifndef LZ4_FORCE_UNALIGNED_ACCESS
301 #define A64(x) (((U64_S *)(x))->v)
302 #define A32(x) (((U32_S *)(x))->v)
303 #define A16(x) (((U16_S *)(x))->v)
310 #define HASH_LOG COMPRESSIONLEVEL
311 #define HASHTABLESIZE (1 << HASH_LOG)
312 #define HASH_MASK (HASHTABLESIZE - 1)
314 #define SKIPSTRENGTH (NOTCOMPRESSIBLE_CONFIRMATION > 2 ? \
315 NOTCOMPRESSIBLE_CONFIRMATION : 2)
318 * Defines if memory is allocated into the stack (local variable),
319 * or into the heap (kmem_alloc()).
321 #define HEAPMODE (HASH_LOG > STACKLIMIT)
323 #define LASTLITERALS 5
324 #define MFLIMIT (COPYLENGTH + MINMATCH)
325 #define MINLENGTH (MFLIMIT + 1)
328 #define MAX_DISTANCE ((1 << MAXD_LOG) - 1)
331 #define ML_MASK ((1U<<ML_BITS)-1)
332 #define RUN_BITS (8-ML_BITS)
333 #define RUN_MASK ((1U<<RUN_BITS)-1)
337 * Architecture-specific macros
343 #define LZ4_COPYSTEP(s, d) A64(d) = A64(s); d += 8; s += 8;
344 #define LZ4_COPYPACKET(s, d) LZ4_COPYSTEP(s, d)
345 #define LZ4_SECURECOPY(s, d, e) if (d < e) LZ4_WILDCOPY(s, d, e)
347 #define INITBASE(base) const BYTE* const base = ip
348 #else /* !LZ4_ARCH64 */
352 #define LZ4_COPYSTEP(s, d) A32(d) = A32(s); d += 4; s += 4;
353 #define LZ4_COPYPACKET(s, d) LZ4_COPYSTEP(s, d); LZ4_COPYSTEP(s, d);
354 #define LZ4_SECURECOPY LZ4_WILDCOPY
355 #define HTYPE const BYTE *
356 #define INITBASE(base) const int base = 0
357 #endif /* !LZ4_ARCH64 */
359 #if (defined(LZ4_BIG_ENDIAN) && !defined(BIG_ENDIAN_NATIVE_BUT_INCOMPATIBLE))
360 #define LZ4_READ_LITTLEENDIAN_16(d, s, p) \
361 { U16 v = A16(p); v = lz4_bswap16(v); d = (s) - v; }
362 #define LZ4_WRITE_LITTLEENDIAN_16(p, i) \
363 { U16 v = (U16)(i); v = lz4_bswap16(v); A16(p) = v; p += 2; }
365 #define LZ4_READ_LITTLEENDIAN_16(d, s, p) { d = (s) - A16(p); }
366 #define LZ4_WRITE_LITTLEENDIAN_16(p, v) { A16(p) = v; p += 2; }
370 /* Local structures */
372 HTYPE hashTable
[HASHTABLESIZE
];
377 #define LZ4_HASH_FUNCTION(i) (((i) * 2654435761U) >> ((MINMATCH * 8) - \
379 #define LZ4_HASH_VALUE(p) LZ4_HASH_FUNCTION(A32(p))
380 #define LZ4_WILDCOPY(s, d, e) do { LZ4_COPYPACKET(s, d) } while (d < e);
381 #define LZ4_BLINDCOPY(s, d, l) { BYTE* e = (d) + l; LZ4_WILDCOPY(s, d, e); \
385 /* Private functions */
389 LZ4_NbCommonBytes(register U64 val
)
391 #if defined(LZ4_BIG_ENDIAN)
392 #if defined(_MSC_VER) && !defined(LZ4_FORCE_SW_BITCOUNT)
394 _BitScanReverse64(&r
, val
);
395 return (int)(r
>> 3);
396 #elif defined(__GNUC__) && (GCC_VERSION >= 304) && \
397 !defined(LZ4_FORCE_SW_BITCOUNT)
398 return (__builtin_clzll(val
) >> 3);
417 #if defined(_MSC_VER) && !defined(LZ4_FORCE_SW_BITCOUNT)
419 _BitScanForward64(&r
, val
);
420 return (int)(r
>> 3);
421 #elif defined(__GNUC__) && (GCC_VERSION >= 304) && \
422 !defined(LZ4_FORCE_SW_BITCOUNT)
423 return (__builtin_ctzll(val
) >> 3);
425 static const int DeBruijnBytePos
[64] =
426 { 0, 0, 0, 0, 0, 1, 1, 2, 0, 3, 1, 3, 1, 4, 2, 7, 0, 2, 3, 6, 1, 5,
427 3, 5, 1, 3, 4, 4, 2, 5, 6, 7, 7, 0, 1, 2, 3, 3, 4, 6, 2, 6, 5,
428 5, 3, 4, 5, 6, 7, 1, 2, 4, 6, 4,
429 4, 5, 7, 2, 6, 5, 7, 6, 7, 7
431 return DeBruijnBytePos
[((U64
) ((val
& -val
) * 0x0218A392CDABBD3F)) >>
440 LZ4_NbCommonBytes(register U32 val
)
442 #if defined(LZ4_BIG_ENDIAN)
443 #if defined(_MSC_VER) && !defined(LZ4_FORCE_SW_BITCOUNT)
445 _BitScanReverse(&r
, val
);
446 return (int)(r
>> 3);
447 #elif defined(__GNUC__) && (GCC_VERSION >= 304) && \
448 !defined(LZ4_FORCE_SW_BITCOUNT)
449 return (__builtin_clz(val
) >> 3);
463 #if defined(_MSC_VER) && !defined(LZ4_FORCE_SW_BITCOUNT)
465 _BitScanForward(&r
, val
);
466 return (int)(r
>> 3);
467 #elif defined(__GNUC__) && (GCC_VERSION >= 304) && \
468 !defined(LZ4_FORCE_SW_BITCOUNT)
469 return (__builtin_ctz(val
) >> 3);
471 static const int DeBruijnBytePos
[32] = {
472 0, 0, 3, 0, 3, 1, 3, 0,
473 3, 2, 2, 1, 3, 2, 0, 1,
474 3, 3, 1, 2, 2, 2, 2, 0,
475 3, 1, 2, 0, 1, 0, 1, 1
477 return DeBruijnBytePos
[((U32
) ((val
& -(S32
) val
) * 0x077CB531U
)) >>
485 /* Public functions */
487 /* Compression functions */
491 LZ4_compressCtx(void *ctx
, const char *source
, char *dest
, int isize
,
495 struct refTables
*srt
= (struct refTables
*)ctx
;
496 HTYPE
*HashTable
= (HTYPE
*) (srt
->hashTable
);
498 HTYPE HashTable
[HASHTABLESIZE
] = { 0 };
501 const BYTE
*ip
= (BYTE
*) source
;
503 const BYTE
*anchor
= ip
;
504 const BYTE
*const iend
= ip
+ isize
;
505 const BYTE
*const oend
= (BYTE
*) dest
+ osize
;
506 const BYTE
*const mflimit
= iend
- MFLIMIT
;
507 #define matchlimit (iend - LASTLITERALS)
509 BYTE
*op
= (BYTE
*) dest
;
512 const int skipStrength
= SKIPSTRENGTH
;
517 if (isize
< MINLENGTH
)
521 HashTable
[LZ4_HASH_VALUE(ip
)] = ip
- base
;
523 forwardH
= LZ4_HASH_VALUE(ip
);
527 int findMatchAttempts
= (1U << skipStrength
) + 3;
528 const BYTE
*forwardIp
= ip
;
535 int step
= findMatchAttempts
++ >> skipStrength
;
537 forwardIp
= ip
+ step
;
539 if unlikely(forwardIp
> mflimit
) {
543 forwardH
= LZ4_HASH_VALUE(forwardIp
);
544 ref
= base
+ HashTable
[h
];
545 HashTable
[h
] = ip
- base
;
547 } while ((ref
< ip
- MAX_DISTANCE
) || (A32(ref
) != A32(ip
)));
550 while ((ip
> anchor
) && (ref
> (BYTE
*) source
) &&
551 unlikely(ip
[-1] == ref
[-1])) {
556 /* Encode Literal length */
557 length
= ip
- anchor
;
560 /* Check output limit */
561 if unlikely(op
+ length
+ (2 + 1 + LASTLITERALS
) +
562 (length
>> 8) > oend
)
565 if (length
>= (int)RUN_MASK
) {
566 *token
= (RUN_MASK
<< ML_BITS
);
567 len
= length
- RUN_MASK
;
568 for (; len
> 254; len
-= 255)
572 *token
= (length
<< ML_BITS
);
575 LZ4_BLINDCOPY(anchor
, op
, length
);
579 LZ4_WRITE_LITTLEENDIAN_16(op
, ip
- ref
);
583 ref
+= MINMATCH
; /* MinMatch verified */
585 while likely(ip
< matchlimit
- (STEPSIZE
- 1)) {
586 UARCH diff
= AARCH(ref
) ^ AARCH(ip
);
592 ip
+= LZ4_NbCommonBytes(diff
);
596 if ((ip
< (matchlimit
- 3)) && (A32(ref
) == A32(ip
))) {
601 if ((ip
< (matchlimit
- 1)) && (A16(ref
) == A16(ip
))) {
605 if ((ip
< matchlimit
) && (*ref
== *ip
))
609 /* Encode MatchLength */
611 /* Check output limit */
612 if unlikely(op
+ (1 + LASTLITERALS
) + (len
>> 8) > oend
)
614 if (len
>= (int)ML_MASK
) {
617 for (; len
> 509; len
-= 510) {
629 /* Test end of chunk */
635 HashTable
[LZ4_HASH_VALUE(ip
- 2)] = ip
- 2 - base
;
637 /* Test next position */
638 ref
= base
+ HashTable
[LZ4_HASH_VALUE(ip
)];
639 HashTable
[LZ4_HASH_VALUE(ip
)] = ip
- base
;
640 if ((ref
> ip
- (MAX_DISTANCE
+ 1)) && (A32(ref
) == A32(ip
))) {
645 /* Prepare next loop */
647 forwardH
= LZ4_HASH_VALUE(ip
);
651 /* Encode Last Literals */
653 int lastRun
= iend
- anchor
;
654 if (op
+ lastRun
+ 1 + ((lastRun
+ 255 - RUN_MASK
) / 255) >
657 if (lastRun
>= (int)RUN_MASK
) {
658 *op
++ = (RUN_MASK
<< ML_BITS
);
660 for (; lastRun
> 254; lastRun
-= 255) {
663 *op
++ = (BYTE
)lastRun
;
665 *op
++ = (lastRun
<< ML_BITS
);
666 (void) memcpy(op
, anchor
, iend
- anchor
);
671 return (int)(((char *)op
) - dest
);
676 /* Note : this function is valid only if isize < LZ4_64KLIMIT */
677 #define LZ4_64KLIMIT ((1 << 16) + (MFLIMIT - 1))
678 #define HASHLOG64K (HASH_LOG + 1)
679 #define HASH64KTABLESIZE (1U << HASHLOG64K)
680 #define LZ4_HASH64K_FUNCTION(i) (((i) * 2654435761U) >> ((MINMATCH*8) - \
682 #define LZ4_HASH64K_VALUE(p) LZ4_HASH64K_FUNCTION(A32(p))
686 LZ4_compress64kCtx(void *ctx
, const char *source
, char *dest
, int isize
,
690 struct refTables
*srt
= (struct refTables
*)ctx
;
691 U16
*HashTable
= (U16
*) (srt
->hashTable
);
693 U16 HashTable
[HASH64KTABLESIZE
] = { 0 };
696 const BYTE
*ip
= (BYTE
*) source
;
697 const BYTE
*anchor
= ip
;
698 const BYTE
*const base
= ip
;
699 const BYTE
*const iend
= ip
+ isize
;
700 const BYTE
*const oend
= (BYTE
*) dest
+ osize
;
701 const BYTE
*const mflimit
= iend
- MFLIMIT
;
702 #define matchlimit (iend - LASTLITERALS)
704 BYTE
*op
= (BYTE
*) dest
;
707 const int skipStrength
= SKIPSTRENGTH
;
711 if (isize
< MINLENGTH
)
716 forwardH
= LZ4_HASH64K_VALUE(ip
);
720 int findMatchAttempts
= (1U << skipStrength
) + 3;
721 const BYTE
*forwardIp
= ip
;
728 int step
= findMatchAttempts
++ >> skipStrength
;
730 forwardIp
= ip
+ step
;
732 if (forwardIp
> mflimit
) {
736 forwardH
= LZ4_HASH64K_VALUE(forwardIp
);
737 ref
= base
+ HashTable
[h
];
738 HashTable
[h
] = ip
- base
;
740 } while (A32(ref
) != A32(ip
));
743 while ((ip
> anchor
) && (ref
> (BYTE
*) source
) &&
744 (ip
[-1] == ref
[-1])) {
749 /* Encode Literal length */
750 length
= ip
- anchor
;
753 /* Check output limit */
754 if unlikely(op
+ length
+ (2 + 1 + LASTLITERALS
) +
755 (length
>> 8) > oend
)
758 if (length
>= (int)RUN_MASK
) {
759 *token
= (RUN_MASK
<< ML_BITS
);
760 len
= length
- RUN_MASK
;
761 for (; len
> 254; len
-= 255)
765 *token
= (length
<< ML_BITS
);
768 LZ4_BLINDCOPY(anchor
, op
, length
);
772 LZ4_WRITE_LITTLEENDIAN_16(op
, ip
- ref
);
776 ref
+= MINMATCH
; /* MinMatch verified */
778 while (ip
< matchlimit
- (STEPSIZE
- 1)) {
779 UARCH diff
= AARCH(ref
) ^ AARCH(ip
);
785 ip
+= LZ4_NbCommonBytes(diff
);
789 if ((ip
< (matchlimit
- 3)) && (A32(ref
) == A32(ip
))) {
794 if ((ip
< (matchlimit
- 1)) && (A16(ref
) == A16(ip
))) {
798 if ((ip
< matchlimit
) && (*ref
== *ip
))
802 /* Encode MatchLength */
804 /* Check output limit */
805 if unlikely(op
+ (1 + LASTLITERALS
) + (len
>> 8) > oend
)
807 if (len
>= (int)ML_MASK
) {
810 for (; len
> 509; len
-= 510) {
822 /* Test end of chunk */
828 HashTable
[LZ4_HASH64K_VALUE(ip
- 2)] = ip
- 2 - base
;
830 /* Test next position */
831 ref
= base
+ HashTable
[LZ4_HASH64K_VALUE(ip
)];
832 HashTable
[LZ4_HASH64K_VALUE(ip
)] = ip
- base
;
833 if (A32(ref
) == A32(ip
)) {
838 /* Prepare next loop */
840 forwardH
= LZ4_HASH64K_VALUE(ip
);
844 /* Encode Last Literals */
846 int lastRun
= iend
- anchor
;
847 if (op
+ lastRun
+ 1 + ((lastRun
+ 255 - RUN_MASK
) / 255) >
850 if (lastRun
>= (int)RUN_MASK
) {
851 *op
++ = (RUN_MASK
<< ML_BITS
);
853 for (; lastRun
> 254; lastRun
-= 255)
855 *op
++ = (BYTE
)lastRun
;
857 *op
++ = (lastRun
<< ML_BITS
);
858 (void) memcpy(op
, anchor
, iend
- anchor
);
863 return (int)(((char *)op
) - dest
);
867 real_LZ4_compress(const char *source
, char *dest
, int isize
, int osize
)
870 void *ctx
= umem_zalloc(sizeof (struct refTables
), UMEM_DEFAULT
);
874 * out of kernel memory, gently fall through - this will disable
875 * compression in zio_compress_data
880 if (isize
< LZ4_64KLIMIT
)
881 result
= LZ4_compress64kCtx(ctx
, source
, dest
, isize
, osize
);
883 result
= LZ4_compressCtx(ctx
, source
, dest
, isize
, osize
);
885 umem_free(ctx
, sizeof (struct refTables
));
888 if (isize
< (int)LZ4_64KLIMIT
)
889 return (LZ4_compress64kCtx(NULL
, source
, dest
, isize
, osize
));
890 return (LZ4_compressCtx(NULL
, source
, dest
, isize
, osize
));
894 /* Decompression functions */
897 * Note: The decoding function LZ4_uncompress_unknownOutputSize() is safe
898 * against "buffer overflow" attack type.
899 * LZ4_uncompress_unknownOutputSize() insures that it will never read
900 * outside of the input buffer. A corrupted input will produce an error
901 * result, a negative int, indicating the position of the error within
906 LZ4_uncompress_unknownOutputSize(const char *source
, char *dest
, int isize
,
909 /* Local Variables */
910 const BYTE
*restrict ip
= (const BYTE
*) source
;
911 const BYTE
*const iend
= ip
+ isize
;
914 BYTE
*op
= (BYTE
*) dest
;
915 BYTE
*const oend
= op
+ maxOutputSize
;
918 size_t dec32table
[] = {0, 3, 2, 3, 0, 0, 0, 0};
920 size_t dec64table
[] = {0, 0, 0, (size_t)-1, 0, 1, 2, 3};
930 if ((length
= (token
>> ML_BITS
)) == RUN_MASK
) {
932 while ((ip
< iend
) && (s
== 255)) {
939 /* CORNER-CASE: cpy might overflow. */
941 goto _output_error
; /* cpy was overflowed, bail! */
942 if ((cpy
> oend
- COPYLENGTH
) ||
943 (ip
+ length
> iend
- COPYLENGTH
)) {
945 /* Error: writes beyond output buffer */
947 if (ip
+ length
!= iend
)
949 * Error: LZ4 format requires to consume all
950 * input at this stage
953 (void) memcpy(op
, ip
, length
);
955 /* Necessarily EOF, due to parsing restrictions */
958 LZ4_WILDCOPY(ip
, op
, cpy
);
963 LZ4_READ_LITTLEENDIAN_16(ref
, cpy
, ip
);
965 if (ref
< (BYTE
* const) dest
)
967 * Error: offset creates reference outside of
972 /* get matchlength */
973 if ((length
= (token
& ML_MASK
)) == ML_MASK
) {
982 /* copy repeated sequence */
983 if unlikely(op
- ref
< STEPSIZE
) {
985 size_t dec64
= dec64table
[op
-ref
];
995 ref
-= dec32table
[op
-ref
];
1000 LZ4_COPYSTEP(ref
, op
);
1002 cpy
= op
+ length
- (STEPSIZE
- 4);
1003 if (cpy
> oend
- COPYLENGTH
) {
1006 * Error: request to write outside of
1007 * destination buffer
1010 LZ4_SECURECOPY(ref
, op
, (oend
- COPYLENGTH
));
1016 * Check EOF (should never happen, since
1017 * last 5 bytes are supposed to be literals)
1022 LZ4_SECURECOPY(ref
, op
, cpy
);
1023 op
= cpy
; /* correction */
1026 /* end of decoding */
1027 return (int)(((char *)op
) - dest
);
1029 /* write overflow error detected */
1031 return (int)(-(((char *)ip
) - source
));