diff options
Diffstat (limited to 'deps/zlib/deflate.c')
| -rw-r--r-- | deps/zlib/deflate.c | 1548 |
1 files changed, 955 insertions, 593 deletions
diff --git a/deps/zlib/deflate.c b/deps/zlib/deflate.c index fb2d5f1835..799fb93cc0 100644 --- a/deps/zlib/deflate.c +++ b/deps/zlib/deflate.c @@ -1,5 +1,5 @@ /* deflate.c -- compress data using the deflation algorithm - * Copyright (C) 1995-2023 Jean-loup Gailly and Mark Adler + * Copyright (C) 1995-2022 Jean-loup Gailly and Mark Adler * For conditions of distribution and use, see copyright notice in zlib.h */ @@ -52,7 +52,7 @@ #include "deflate.h" const char deflate_copyright[] = - " deflate 1.3 Copyright 1995-2023 Jean-loup Gailly and Mark Adler "; + " deflate 1.2.12 Copyright 1995-2022 Jean-loup Gailly and Mark Adler "; /* If you use the zlib library in a product, an acknowledgment is welcome in the documentation of your product. If for some reason you cannot @@ -60,6 +60,9 @@ const char deflate_copyright[] = copyright string in the executable of your product. */ +/* =========================================================================== + * Function prototypes. + */ typedef enum { need_more, /* block not completed, need more input or more output */ block_done, /* block flush performed */ @@ -67,24 +70,34 @@ typedef enum { finish_done /* finish done, accept no more input or output */ } block_state; -typedef block_state (*compress_func)(deflate_state *s, int flush); +typedef block_state (*compress_func) OF((deflate_state *s, int flush)); /* Compression function. Returns the block state after the call. */ -static int deflateStateCheck(z_streamp strm); -static void fill_window(deflate_state *s); -static block_state deflate_stored(deflate_state *s, int flush); -static block_state deflate_fast(deflate_state *s, int flush); -static block_state deflate_slow(deflate_state *s, int flush); -static block_state deflate_rle(deflate_state *s, int flush); -static block_state deflate_huff(deflate_state *s, int flush); -static void lm_init(deflate_state *s); -static void putShortMSB(deflate_state *s, uint32_t b); -static void flush_pending(z_streamp strm); -static int read_buf(z_streamp strm, uint8_t *buf, uint32_t size); +local int deflateStateCheck OF((z_streamp strm)); +local void slide_hash OF((deflate_state *s)); +local void fill_window OF((deflate_state *s)); +local block_state deflate_stored OF((deflate_state *s, int flush)); +local block_state deflate_fast OF((deflate_state *s, int flush)); +#ifndef FASTEST +local block_state deflate_slow OF((deflate_state *s, int flush)); +#endif +local block_state deflate_rle OF((deflate_state *s, int flush)); +local block_state deflate_huff OF((deflate_state *s, int flush)); +local void lm_init OF((deflate_state *s)); +local void putShortMSB OF((deflate_state *s, uInt b)); +local void flush_pending OF((z_streamp strm)); +local unsigned read_buf OF((z_streamp strm, Bytef *buf, unsigned size)); +#ifdef ASMV +# pragma message("Assembler code may have bugs -- use at your own risk") + void match_init OF((void)); /* asm code initialization */ + uInt longest_match OF((deflate_state *s, IPos cur_match)); +#else +local uInt longest_match OF((deflate_state *s, IPos cur_match)); +#endif #ifdef ZLIB_DEBUG -static void check_match(deflate_state *s, IPos start, IPos match, - int length); +local void check_match OF((deflate_state *s, IPos start, IPos match, + int length)); #endif /* =========================================================================== @@ -93,21 +106,32 @@ static void check_match(deflate_state *s, IPos start, IPos match, #define NIL 0 /* Tail of hash chains */ -#define ACTUAL_MIN_MATCH 4 + +#ifndef TOO_FAR +# define TOO_FAR 4096 +#endif +/* Matches of length 3 are discarded if their distance exceeds TOO_FAR */ + /* Values for max_lazy_match, good_match and max_chain_length, depending on * the desired pack level (0..9). The values given below have been tuned to * exclude worst case performance for pathological files. Better values may be * found for specific files. */ typedef struct config_s { - uint16_t good_length; /* reduce lazy search above this match length */ - uint16_t max_lazy; /* do not perform lazy search above this match length */ - uint16_t nice_length; /* quit search above this match length */ - uint16_t max_chain; + ush good_length; /* reduce lazy search above this match length */ + ush max_lazy; /* do not perform lazy search above this match length */ + ush nice_length; /* quit search above this match length */ + ush max_chain; compress_func func; } config; -static const config configuration_table[10] = { +#ifdef FASTEST +local const config configuration_table[2] = { +/* good lazy nice chain */ +/* 0 */ {0, 0, 0, 0, deflate_stored}, /* store only */ +/* 1 */ {4, 4, 8, 4, deflate_fast}}; /* max speed, no lazy matches */ +#else +local const config configuration_table[10] = { /* good lazy nice chain */ /* 0 */ {0, 0, 0, 0, deflate_stored}, /* store only */ /* 1 */ {4, 4, 8, 4, deflate_fast}, /* max speed, no lazy matches */ @@ -120,6 +144,7 @@ static const config configuration_table[10] = { /* 7 */ {8, 32, 128, 256, deflate_slow}, /* 8 */ {32, 128, 258, 1024, deflate_slow}, /* 9 */ {32, 258, 258, 4096, deflate_slow}}; /* max compression */ +#endif /* Note: the deflate() code requires max_lazy >= MIN_MATCH and max_chain >= 4 * For deflate_fast() (levels <= 3) good is ignored and lazy has a different @@ -129,75 +154,103 @@ static const config configuration_table[10] = { /* rank Z_BLOCK between Z_NO_FLUSH and Z_PARTIAL_FLUSH */ #define RANK(f) (((f) * 2) - ((f) > 4 ? 9 : 0)) -#ifdef __aarch64__ - -#include <arm_neon.h> -#include <arm_acle.h> -static uint32_t hash_func(deflate_state *s, void* str) { - return __crc32cw(0, *(uint32_t*)str) & s->hash_mask; -} - -#elif defined HAS_SSE42 +/* =========================================================================== + * Update a hash value with the given input byte + * IN assertion: all calls to UPDATE_HASH are made with consecutive input + * characters, so that a running hash key can be computed from the previous + * key instead of complete recalculation each time. + */ +#define UPDATE_HASH(s,h,c) (h = (((h)<<s->hash_shift) ^ (c)) & s->hash_mask) -#include <immintrin.h> -static uint32_t hash_func(deflate_state *s, void* str) { - return _mm_crc32_u32(0, *(uint32_t*)str) & s->hash_mask; -} +/* =========================================================================== + * Insert string str in the dictionary and set match_head to the previous head + * of the hash chain (the most recent string with same hash key). Return + * the previous length of the hash chain. + * If this file is compiled with -DFASTEST, the compression level is forced + * to 1, and no hash chains are maintained. + * IN assertion: all calls to INSERT_STRING are made with consecutive input + * characters and the first MIN_MATCH bytes of str are valid (except for + * the last MIN_MATCH-1 bytes of the input file). + */ +#ifdef FASTEST +#define INSERT_STRING(s, str, match_head) \ + (UPDATE_HASH(s, s->ins_h, s->window[(str) + (MIN_MATCH-1)]), \ + match_head = s->head[s->ins_h], \ + s->head[s->ins_h] = (Pos)(str)) #else - -static uint32_t hash_func(deflate_state *s, void* str) { - uint32_t w; - zmemcpy(&w, str, sizeof(w)); - // generic multiply-xor hash, using some magic numbers from xxhash. - w *= 0x85ebca77u; - w ^= w >> 19; - return w & s->hash_mask; -} - +#define INSERT_STRING(s, str, match_head) \ + (UPDATE_HASH(s, s->ins_h, s->window[(str) + (MIN_MATCH-1)]), \ + match_head = s->prev[(str) & s->w_mask] = s->head[s->ins_h], \ + s->head[s->ins_h] = (Pos)(str)) #endif /* =========================================================================== - * Insert string str in the dictionary and return the previous head - * of the hash chain (the most recent string with same hash key). - * IN assertion: ACTUAL_MIN_MATCH bytes of str are valid - * (except for the last ACTUAL_MIN_MATCH-1 bytes of the input file). + * Initialize the hash table (avoiding 64K overflow for 16 bit systems). + * prev[] will be initialized on the fly. */ -static Pos insert_string(deflate_state *s, Pos str) { - Pos match_head; - s->ins_h = hash_func(s, &s->window[str]); - match_head = s->prev[(str) & s->w_mask] = s->head[s->ins_h]; - s->head[s->ins_h] = (Pos)str; - return match_head; -} - -static void bulk_insert_str(deflate_state *s, Pos startpos, uint32_t count) { - uint32_t idx; - for (idx = 0; idx < count; idx++) { - s->ins_h = hash_func(s, &s->window[startpos + idx]); - s->prev[(startpos + idx) & s->w_mask] = s->head[s->ins_h]; - s->head[s->ins_h] = (Pos)(startpos + idx); - } -} +#define CLEAR_HASH(s) \ + do { \ + s->head[s->hash_size-1] = NIL; \ + zmemzero((Bytef *)s->head, \ + (unsigned)(s->hash_size-1)*sizeof(*s->head)); \ + } while (0) /* =========================================================================== - * Initialize the hash table prev[] will be initialized on the fly. + * Slide the hash table when sliding the window down (could be avoided with 32 + * bit values at the expense of memory usage). We slide even when level == 0 to + * keep the hash table consistent if we switch back to level > 0 later. */ -#define CLEAR_HASH(s) \ - zmemzero((uint8_t *)s->head, (unsigned)(s->hash_size)*sizeof(*s->head)); +local void slide_hash(s) + deflate_state *s; +{ + unsigned n, m; + Posf *p; + uInt wsize = s->w_size; + + n = s->hash_size; + p = &s->head[n]; + do { + m = *--p; + *p = (Pos)(m >= wsize ? m - wsize : NIL); + } while (--n); + n = wsize; +#ifndef FASTEST + p = &s->prev[n]; + do { + m = *--p; + *p = (Pos)(m >= wsize ? m - wsize : NIL); + /* If n is not on any hash chain, prev[n] is garbage but + * its value will never be used. + */ + } while (--n); +#endif +} /* ========================================================================= */ -int ZEXPORT deflateInit_(z_streamp strm, int level, const char *version, - int stream_size) { +int ZEXPORT deflateInit_(strm, level, version, stream_size) + z_streamp strm; + int level; + const char *version; + int stream_size; +{ return deflateInit2_(strm, level, Z_DEFLATED, MAX_WBITS, DEF_MEM_LEVEL, Z_DEFAULT_STRATEGY, version, stream_size); /* To do: ignore strm->next_in if we use it as window */ } /* ========================================================================= */ -int ZEXPORT deflateInit2_(z_streamp strm, int level, int method, - int windowBits, int memLevel, int strategy, - const char *version, int stream_size) { +int ZEXPORT deflateInit2_(strm, level, method, windowBits, memLevel, strategy, + version, stream_size) + z_streamp strm; + int level; + int method; + int windowBits; + int memLevel; + int strategy; + const char *version; + int stream_size; +{ deflate_state *s; int wrap = 1; static const char my_version[] = ZLIB_VERSION; @@ -224,18 +277,22 @@ int ZEXPORT deflateInit2_(z_streamp strm, int level, int method, strm->zfree = zcfree; #endif +#ifdef FASTEST + if (level != 0) level = 1; +#else if (level == Z_DEFAULT_COMPRESSION) level = 6; +#endif if (windowBits < 0) { /* suppress zlib wrapper */ wrap = 0; - if (windowBits < -15) - return Z_STREAM_ERROR; windowBits = -windowBits; } +#ifdef GZIP else if (windowBits > 15) { wrap = 2; /* write gzip wrapper instead */ windowBits -= 16; } +#endif if (memLevel < 1 || memLevel > MAX_MEM_LEVEL || method != Z_DEFLATED || windowBits < 8 || windowBits > 15 || level < 0 || level > 9 || strategy < 0 || strategy > Z_FIXED || (windowBits == 8 && wrap != 1)) { @@ -244,23 +301,24 @@ int ZEXPORT deflateInit2_(z_streamp strm, int level, int method, if (windowBits == 8) windowBits = 9; /* until 256-byte window bug fixed */ s = (deflate_state *) ZALLOC(strm, 1, sizeof(deflate_state)); if (s == Z_NULL) return Z_MEM_ERROR; - strm->state = (struct internal_state *)s; + strm->state = (struct internal_state FAR *)s; s->strm = strm; s->status = INIT_STATE; /* to pass state test in deflateReset() */ s->wrap = wrap; s->gzhead = Z_NULL; - s->w_bits = windowBits; + s->w_bits = (uInt)windowBits; s->w_size = 1 << s->w_bits; s->w_mask = s->w_size - 1; - s->hash_bits = memLevel + 7; + s->hash_bits = (uInt)memLevel + 7; s->hash_size = 1 << s->hash_bits; s->hash_mask = s->hash_size - 1; + s->hash_shift = ((s->hash_bits+MIN_MATCH-1)/MIN_MATCH); - s->window = (uint8_t *) ZALLOC(strm, s->w_size, 2*sizeof(uint8_t)); - s->prev = (Pos *) ZALLOC(strm, s->w_size, sizeof(Pos)); - s->head = (Pos *) ZALLOC(strm, s->hash_size, sizeof(Pos)); + s->window = (Bytef *) ZALLOC(strm, s->w_size, 2*sizeof(Byte)); + s->prev = (Posf *) ZALLOC(strm, s->w_size, sizeof(Pos)); + s->head = (Posf *) ZALLOC(strm, s->hash_size, sizeof(Pos)); s->high_water = 0; /* nothing written to s->window yet */ @@ -282,11 +340,11 @@ int ZEXPORT deflateInit2_(z_streamp strm, int level, int method, * sym_buf value to read moves forward three bytes. From that symbol, up to * 31 bits are written to pending_buf. The closest the written pending_buf * bits gets to the next sym_buf symbol to read is just before the last - * code is written. At that time, 31*(n - 2) bits have been written, just - * after 24*(n - 2) bits have been consumed from sym_buf. sym_buf starts at - * 8*n bits into pending_buf. (Note that the symbol buffer fills when n - 1 + * code is written. At that time, 31*(n-2) bits have been written, just + * after 24*(n-2) bits have been consumed from sym_buf. sym_buf starts at + * 8*n bits into pending_buf. (Note that the symbol buffer fills when n-1 * symbols are written.) The closest the writing gets to what is unread is - * then n + 14 bits. Here n is lit_bufsize, which is 16384 by default, and + * then n+14 bits. Here n is lit_bufsize, which is 16384 by default, and * can range from 128 to 32768. * * Therefore, at a minimum, there are 142 bits of space between what is @@ -305,8 +363,8 @@ int ZEXPORT deflateInit2_(z_streamp strm, int level, int method, * symbols from which it is being constructed. */ - s->pending_buf = (uint8_t *) ZALLOC(strm, s->lit_bufsize, 4); - s->pending_buf_size = (uint64_t)s->lit_bufsize * 4; + s->pending_buf = (uchf *) ZALLOC(strm, s->lit_bufsize, 4); + s->pending_buf_size = (ulg)s->lit_bufsize * 4; if (s->window == Z_NULL || s->prev == Z_NULL || s->head == Z_NULL || s->pending_buf == Z_NULL) { @@ -324,7 +382,7 @@ int ZEXPORT deflateInit2_(z_streamp strm, int level, int method, s->level = level; s->strategy = strategy; - s->method = (uint8_t)method; + s->method = (Byte)method; return deflateReset(strm); } @@ -332,13 +390,18 @@ int ZEXPORT deflateInit2_(z_streamp strm, int level, int method, /* ========================================================================= * Check for a valid deflate stream state. Return 0 if ok, 1 if not. */ -static int deflateStateCheck(z_streamp strm) { +local int deflateStateCheck (strm) + z_streamp strm; +{ deflate_state *s; if (strm == Z_NULL || strm->zalloc == (alloc_func)0 || strm->zfree == (free_func)0) return 1; s = strm->state; if (s == Z_NULL || s->strm != strm || (s->status != INIT_STATE && +#ifdef GZIP + s->status != GZIP_STATE && +#endif s->status != EXTRA_STATE && s->status != NAME_STATE && s->status != COMMENT_STATE && @@ -350,13 +413,16 @@ static int deflateStateCheck(z_streamp strm) { } /* ========================================================================= */ -int ZEXPORT deflateSetDictionary (z_streamp strm, const uint8_t *dictionary, uint32_t dictLength) +int ZEXPORT deflateSetDictionary (strm, dictionary, dictLength) + z_streamp strm; + const Bytef *dictionary; + uInt dictLength; { deflate_state *s; - uint32_t str, n; + uInt str, n; int wrap; - uint32_t avail; - z_const uint8_t *next; + unsigned avail; + z_const unsigned char *next; if (deflateStateCheck(strm) || dictionary == Z_NULL) return Z_STREAM_ERROR; @@ -386,21 +452,28 @@ int ZEXPORT deflateSetDictionary (z_streamp strm, const uint8_t *dictionary, ui avail = strm->avail_in; next = strm->next_in; strm->avail_in = dictLength; - strm->next_in = (z_const uint8_t*)dictionary; + strm->next_in = (z_const Bytef *)dictionary; fill_window(s); - while (s->lookahead >= ACTUAL_MIN_MATCH) { + while (s->lookahead >= MIN_MATCH) { str = s->strstart; - n = s->lookahead - (ACTUAL_MIN_MATCH-1); - bulk_insert_str(s, str, n); - s->strstart = str + n; - s->lookahead = ACTUAL_MIN_MATCH-1; + n = s->lookahead - (MIN_MATCH-1); + do { + UPDATE_HASH(s, s->ins_h, s->window[str + MIN_MATCH-1]); +#ifndef FASTEST + s->prev[str & s->w_mask] = s->head[s->ins_h]; +#endif + s->head[s->ins_h] = (Pos)str; + str++; + } while (--n); + s->strstart = str; + s->lookahead = MIN_MATCH-1; fill_window(s); } s->strstart += s->lookahead; s->block_start = (long)s->strstart; s->insert = s->lookahead; s->lookahead = 0; - s->match_length = s->prev_length = ACTUAL_MIN_MATCH-1; + s->match_length = s->prev_length = MIN_MATCH-1; s->match_available = 0; strm->next_in = next; strm->avail_in = avail; @@ -409,7 +482,30 @@ int ZEXPORT deflateSetDictionary (z_streamp strm, const uint8_t *dictionary, ui } /* ========================================================================= */ -int ZEXPORT deflateResetKeep(z_streamp strm) +int ZEXPORT deflateGetDictionary (strm, dictionary, dictLength) + z_streamp strm; + Bytef *dictionary; + uInt *dictLength; +{ + deflate_state *s; + uInt len; + + if (deflateStateCheck(strm)) + return Z_STREAM_ERROR; + s = strm->state; + len = s->strstart + s->lookahead; + if (len > s->w_size) + len = s->w_size; + if (dictionary != Z_NULL && len) + zmemcpy(dictionary, s->window + s->strstart + s->lookahead - len, len); + if (dictLength != Z_NULL) + *dictLength = len; + return Z_OK; +} + +/* ========================================================================= */ +int ZEXPORT deflateResetKeep (strm) + z_streamp strm; { deflate_state *s; @@ -428,9 +524,15 @@ int ZEXPORT deflateResetKeep(z_streamp strm) if (s->wrap < 0) { s->wrap = -s->wrap; /* was made negative by deflate(..., Z_FINISH); */ } - s->status = s->wrap ? INIT_STATE : BUSY_STATE; + s->status = +#ifdef GZIP + s->wrap == 2 ? GZIP_STATE : +#endif + INIT_STATE; strm->adler = +#ifdef GZIP s->wrap == 2 ? crc32(0L, Z_NULL, 0) : +#endif adler32(0L, Z_NULL, 0); s->last_flush = -2; @@ -440,7 +542,9 @@ int ZEXPORT deflateResetKeep(z_streamp strm) } /* ========================================================================= */ -int ZEXPORT deflateReset(z_streamp strm) { +int ZEXPORT deflateReset (strm) + z_streamp strm; +{ int ret; ret = deflateResetKeep(strm); @@ -450,7 +554,10 @@ int ZEXPORT deflateReset(z_streamp strm) { } /* ========================================================================= */ -int ZEXPORT deflateSetHeader(z_streamp strm, gz_headerp head) { +int ZEXPORT deflateSetHeader (strm, head) + z_streamp strm; + gz_headerp head; +{ if (deflateStateCheck(strm) || strm->state->wrap != 2) return Z_STREAM_ERROR; strm->state->gzhead = head; @@ -458,7 +565,11 @@ int ZEXPORT deflateSetHeader(z_streamp strm, gz_headerp head) { } /* ========================================================================= */ -int ZEXPORT deflatePending(z_streamp strm, unsigned *pending, int *bits) { +int ZEXPORT deflatePending (strm, pending, bits) + unsigned *pending; + int *bits; + z_streamp strm; +{ if (deflateStateCheck(strm)) return Z_STREAM_ERROR; if (pending != Z_NULL) *pending = strm->state->pending; @@ -468,19 +579,24 @@ int ZEXPORT deflatePending(z_streamp strm, unsigned *pending, int *bits) { } /* ========================================================================= */ -int ZEXPORT deflatePrime(z_streamp strm, int bits, int value) { +int ZEXPORT deflatePrime (strm, bits, value) + z_streamp strm; + int bits; + int value; +{ deflate_state *s; int put; if (deflateStateCheck(strm)) return Z_STREAM_ERROR; s = strm->state; - if ((uint8_t *)(s->sym_buf) < s->pending_out + ((Buf_size + 7) >> 3)) + if (bits < 0 || bits > 16 || + s->sym_buf < s->pending_out + ((Buf_size + 7) >> 3)) return Z_BUF_ERROR; do { put = Buf_size - s->bi_valid; if (put > bits) put = bits; - s->bi_buf |= (uint16_t)((value & ((1 << put) - 1)) << s->bi_valid); + s->bi_buf |= (ush)((value & ((1 << put) - 1)) << s->bi_valid); s->bi_valid += put; _tr_flush_bits(s); value >>= put; @@ -490,14 +606,22 @@ int ZEXPORT deflatePrime(z_streamp strm, int bits, int value) { } /* ========================================================================= */ -int ZEXPORT deflateParams(z_streamp strm, int level, int strategy) { +int ZEXPORT deflateParams(strm, level, strategy) + z_streamp strm; + int level; + int strategy; +{ deflate_state *s; compress_func func; if (deflateStateCheck(strm)) return Z_STREAM_ERROR; s = strm->state; +#ifdef FASTEST + if (level != 0) level = 1; +#else if (level == Z_DEFAULT_COMPRESSION) level = 6; +#endif if (level < 0 || level > 9 || strategy < 0 || strategy > Z_FIXED) { return Z_STREAM_ERROR; } @@ -513,6 +637,13 @@ int ZEXPORT deflateParams(z_streamp strm, int level, int strategy) { return Z_BUF_ERROR; } if (s->level != level) { + if (s->level == 0 && s->matches != 0) { + if (s->matches == 1) + slide_hash(s); + else + CLEAR_HASH(s); + s->matches = 0; + } s->level = level; s->max_lazy_match = configuration_table[level].max_lazy; s->good_match = configuration_table[level].good_length; @@ -524,16 +655,21 @@ int ZEXPORT deflateParams(z_streamp strm, int level, int strategy) { } /* ========================================================================= */ -int ZEXPORT deflateTune(z_streamp strm, int good_length, int max_lazy, - int nice_length, int max_chain) { +int ZEXPORT deflateTune(strm, good_length, max_lazy, nice_length, max_chain) + z_streamp strm; + int good_length; + int max_lazy; + int nice_length; + int max_chain; +{ deflate_state *s; if (deflateStateCheck(strm)) return Z_STREAM_ERROR; s = strm->state; - s->good_match = good_length; - s->max_lazy_match = max_lazy; + s->good_match = (uInt)good_length; + s->max_lazy_match = (uInt)max_lazy; s->nice_match = nice_length; - s->max_chain_length = max_chain; + s->max_chain_length = (uInt)max_chain; return Z_OK; } @@ -554,11 +690,12 @@ int ZEXPORT deflateTune(z_streamp strm, int good_length, int max_lazy, * upper bound of about 14% expansion does not seem onerous for output buffer * allocation. */ -uint64_t ZEXPORT deflateBound(z_streamp strm, uint64_t sourceLen) +uLong ZEXPORT deflateBound(strm, sourceLen) + z_streamp strm; + uLong sourceLen; { deflate_state *s; - uint64_t complen, wraplen; - uint8_t *str; + uLong complen, wraplen; /* conservative upper bound for compressed data */ complen = sourceLen + @@ -577,9 +714,11 @@ uint64_t ZEXPORT deflateBound(z_streamp strm, uint64_t sourceLen) case 1: /* zlib wrapper */ wraplen = 6 + (s->strstart ? 4 : 0); break; +#ifdef GZIP case 2: /* gzip wrapper */ wraplen = 18; if (s->gzhead != Z_NULL) { /* user-supplied gzip header */ + Bytef *str; if (s->gzhead->extra != Z_NULL) wraplen += 2 + s->gzhead->extra_len; str = s->gzhead->name; @@ -596,6 +735,7 @@ uint64_t ZEXPORT deflateBound(z_streamp strm, uint64_t sourceLen) wraplen += 2; } break; +#endif default: /* for compiler happiness */ wraplen = 6; } @@ -614,20 +754,24 @@ uint64_t ZEXPORT deflateBound(z_streamp strm, uint64_t sourceLen) * IN assertion: the stream state is correct and there is enough room in * pending_buf. */ -static void putShortMSB (deflate_state *s, uint32_t b) +local void putShortMSB (s, b) + deflate_state *s; + uInt b; { - put_byte(s, (uint8_t)(b >> 8)); - put_byte(s, (uint8_t)(b & 0xff)); + put_byte(s, (Byte)(b >> 8)); + put_byte(s, (Byte)(b & 0xff)); } /* ========================================================================= - * Flush as much pending output as possible. All deflate() output goes - * through this function so some applications may wish to modify it - * to avoid allocating a large strm->next_out buffer and copying into it. - * (See also read_buf()). + * Flush as much pending output as possible. All deflate() output, except for + * some deflate_stored() output, goes through this function so some + * applications may wish to modify it to avoid allocating a large + * strm->next_out buffer and copying into it. (See also read_buf()). */ -static void flush_pending(z_streamp strm) { - uint32_t len; +local void flush_pending(strm) + z_streamp strm; +{ + unsigned len; deflate_state *s = strm->state; _tr_flush_bits(s); @@ -646,8 +790,21 @@ static void flush_pending(z_streamp strm) { } } +/* =========================================================================== + * Update the header CRC with the bytes s->pending_buf[beg..s->pending - 1]. + */ +#define HCRC_UPDATE(beg) \ + do { \ + if (s->gzhead->hcrc && s->pending > (beg)) \ + strm->adler = crc32(strm->adler, s->pending_buf + (beg), \ + s->pending - (beg)); \ + } while (0) + /* ========================================================================= */ -int ZEXPORT deflate(z_streamp strm, int flush) { +int ZEXPORT deflate (strm, flush) + z_streamp strm; + int flush; +{ int old_flush; /* value of flush param for previous deflate call */ deflate_state *s; @@ -663,213 +820,219 @@ int ZEXPORT deflate(z_streamp strm, int flush) { } if (strm->avail_out == 0) ERR_RETURN(strm, Z_BUF_ERROR); - s->strm = strm; /* just in case */ old_flush = s->last_flush; s->last_flush = flush; + /* Flush as much pending output as possible */ + if (s->pending != 0) { + flush_pending(strm); + if (strm->avail_out == 0) { + /* Since avail_out is 0, deflate will be called again with + * more output space, but possibly with both pending and + * avail_in equal to zero. There won't be anything to do, + * but this is not an error situation so make sure we + * return OK instead of BUF_ERROR at next call of deflate: + */ + s->last_flush = -1; + return Z_OK; + } + + /* Make sure there is something to do and avoid duplicate consecutive + * flushes. For repeated and useless calls with Z_FINISH, we keep + * returning Z_STREAM_END instead of Z_BUF_ERROR. + */ + } else if (strm->avail_in == 0 && RANK(flush) <= RANK(old_flush) && + flush != Z_FINISH) { + ERR_RETURN(strm, Z_BUF_ERROR); + } + + /* User must not provide more input after the first FINISH: */ + if (s->status == FINISH_STATE && strm->avail_in != 0) { + ERR_RETURN(strm, Z_BUF_ERROR); + } + /* Write the header */ + if (s->status == INIT_STATE && s->wrap == 0) + s->status = BUSY_STATE; if (s->status == INIT_STATE) { - if (s->wrap == 2) { - strm->adler = crc32(0L, Z_NULL, 0); - put_byte(s, 31); - put_byte(s, 139); - put_byte(s, 8); - if (s->gzhead == Z_NULL) { - put_byte(s, 0); - put_byte(s, 0); - put_byte(s, 0); - put_byte(s, 0); - put_byte(s, 0); - put_byte(s, s->level == 9 ? 2 : - (s->strategy >= Z_HUFFMAN_ONLY || s->level < 2 ? - 4 : 0)); - put_byte(s, OS_CODE); - s->status = BUSY_STATE; - } - else { - put_byte(s, (s->gzhead->text ? 1 : 0) + - (s->gzhead->hcrc ? 2 : 0) + - (s->gzhead->extra == Z_NULL ? 0 : 4) + - (s->gzhead->name == Z_NULL ? 0 : 8) + - (s->gzhead->comment == Z_NULL ? 0 : 16) - ); - put_byte(s, (uint8_t)(s->gzhead->time & 0xff)); - put_byte(s, (uint8_t)((s->gzhead->time >> 8) & 0xff)); - put_byte(s, (uint8_t)((s->gzhead->time >> 16) & 0xff)); - put_byte(s, (uint8_t)((s->gzhead->time >> 24) & 0xff)); - put_byte(s, s->level == 9 ? 2 : - (s->strategy >= Z_HUFFMAN_ONLY || s->level < 2 ? - 4 : 0)); - put_byte(s, s->gzhead->os & 0xff); - if (s->gzhead->extra != Z_NULL) { - put_byte(s, s->gzhead->extra_len & 0xff); - put_byte(s, (s->gzhead->extra_len >> 8) & 0xff); - } - if (s->gzhead->hcrc) - strm->adler = crc32(strm->adler, s->pending_buf, - s->pending); - s->gzindex = 0; - s->status = EXTRA_STATE; - } - } + /* zlib header */ + uInt header = (Z_DEFLATED + ((s->w_bits-8)<<4)) << 8; + uInt level_flags; + + if (s->strategy >= Z_HUFFMAN_ONLY || s->level < 2) + level_flags = 0; + else if (s->level < 6) + level_flags = 1; + else if (s->level == 6) + level_flags = 2; else - { - uint32_t header = (Z_DEFLATED + ((s->w_bits-8)<<4)) << 8; - uint32_t level_flags; - - if (s->strategy >= Z_HUFFMAN_ONLY || s->level < 2) - level_flags = 0; - else if (s->level < 6) - level_flags = 1; - else if (s->level == 6) - level_flags = 2; - else - level_flags = 3; - header |= (level_flags << 6); - if (s->strstart != 0) header |= PRESET_DICT; - header += 31 - (header % 31); + level_flags = 3; + header |= (level_flags << 6); + if (s->strstart != 0) header |= PRESET_DICT; + header += 31 - (header % 31); + + putShortMSB(s, header); + /* Save the adler32 of the preset dictionary: */ + if (s->strstart != 0) { + putShortMSB(s, (uInt)(strm->adler >> 16)); + putShortMSB(s, (uInt)(strm->adler & 0xffff)); + } + strm->adler = adler32(0L, Z_NULL, 0); + s->status = BUSY_STATE; + + /* Compression must start with an empty pending buffer */ + flush_pending(strm); + if (s->pending != 0) { + s->last_flush = -1; + return Z_OK; + } + } +#ifdef GZIP + if (s->status == GZIP_STATE) { + /* gzip header */ + strm->adler = crc32(0L, Z_NULL, 0); + put_byte(s, 31); + put_byte(s, 139); + put_byte(s, 8); + if (s->gzhead == Z_NULL) { + put_byte(s, 0); + put_byte(s, 0); + put_byte(s, 0); + put_byte(s, 0); + put_byte(s, 0); + put_byte(s, s->level == 9 ? 2 : + (s->strategy >= Z_HUFFMAN_ONLY || s->level < 2 ? + 4 : 0)); + put_byte(s, OS_CODE); s->status = BUSY_STATE; - putShortMSB(s, header); - /* Save the adler32 of the preset dictionary: */ - if (s->strstart != 0) { - putShortMSB(s, (uint32_t)(strm->adler >> 16)); - putShortMSB(s, (uint32_t)(strm->adler & 0xffff)); + /* Compression must start with an empty pending buffer */ + flush_pending(strm); + if (s->pending != 0) { + s->last_flush = -1; + return Z_OK; + } + } + else { + put_byte(s, (s->gzhead->text ? 1 : 0) + + (s->gzhead->hcrc ? 2 : 0) + + (s->gzhead->extra == Z_NULL ? 0 : 4) + + (s->gzhead->name == Z_NULL ? 0 : 8) + + (s->gzhead->comment == Z_NULL ? 0 : 16) + ); + put_byte(s, (Byte)(s->gzhead->time & 0xff)); + put_byte(s, (Byte)((s->gzhead->time >> 8) & 0xff)); + put_byte(s, (Byte)((s->gzhead->time >> 16) & 0xff)); + put_byte(s, (Byte)((s->gzhead->time >> 24) & 0xff)); + put_byte(s, s->level == 9 ? 2 : + (s->strategy >= Z_HUFFMAN_ONLY || s->level < 2 ? + 4 : 0)); + put_byte(s, s->gzhead->os & 0xff); + if (s->gzhead->extra != Z_NULL) { + put_byte(s, s->gzhead->extra_len & 0xff); + put_byte(s, (s->gzhead->extra_len >> 8) & 0xff); } - strm->adler = adler32(0L, Z_NULL, 0); + if (s->gzhead->hcrc) + strm->adler = crc32(strm->adler, s->pending_buf, + s->pending); + s->gzindex = 0; + s->status = EXTRA_STATE; } } if (s->status == EXTRA_STATE) { if (s->gzhead->extra != Z_NULL) { - uint32_t beg = s->pending; /* start of bytes to update crc */ - - while (s->gzindex < (s->gzhead->extra_len & 0xffff)) { - if (s->pending == s->pending_buf_size) { - if (s->gzhead->hcrc && s->pending > beg) - strm->adler = crc32(strm->adler, s->pending_buf + beg, - s->pending - beg); - flush_pending(strm); - beg = s->pending; - if (s->pending == s->pending_buf_size) - break; + ulg beg = s->pending; /* start of bytes to update crc */ + uInt left = (s->gzhead->extra_len & 0xffff) - s->gzindex; + while (s->pending + left > s->pending_buf_size) { + uInt copy = s->pending_buf_size - s->pending; + zmemcpy(s->pending_buf + s->pending, + s->gzhead->extra + s->gzindex, copy); + s->pending = s->pending_buf_size; + HCRC_UPDATE(beg); + s->gzindex += copy; + flush_pending(strm); + if (s->pending != 0) { + s->last_flush = -1; + return Z_OK; } - put_byte(s, s->gzhead->extra[s->gzindex]); - s->gzindex++; - } - if (s->gzhead->hcrc && s->pending > beg) - strm->adler = crc32(strm->adler, s->pending_buf + beg, - s->pending - beg); - if (s->gzindex == s->gzhead->extra_len) { - s->gzindex = 0; - s->status = NAME_STATE; + beg = 0; + left -= copy; } + zmemcpy(s->pending_buf + s->pending, + s->gzhead->extra + s->gzindex, left); + s->pending += left; + HCRC_UPDATE(beg); + s->gzindex = 0; } - else - s->status = NAME_STATE; + s->status = NAME_STATE; } if (s->status == NAME_STATE) { if (s->gzhead->name != Z_NULL) { - uint32_t beg = s->pending; /* start of bytes to update crc */ + ulg beg = s->pending; /* start of bytes to update crc */ int val; - do { if (s->pending == s->pending_buf_size) { - if (s->gzhead->hcrc && s->pending > beg) - strm->adler = crc32(strm->adler, s->pending_buf + beg, - s->pending - beg); + HCRC_UPDATE(beg); flush_pending(strm); - beg = s->pending; - if (s->pending == s->pending_buf_size) { - val = 1; - break; + if (s->pending != 0) { + s->last_flush = -1; + return Z_OK; } + beg = 0; } val = s->gzhead->name[s->gzindex++]; put_byte(s, val); } while (val != 0); - if (s->gzhead->hcrc && s->pending > beg) - strm->adler = crc32(strm->adler, s->pending_buf + beg, - s->pending - beg); - if (val == 0) { - s->gzindex = 0; - s->status = COMMENT_STATE; - } + HCRC_UPDATE(beg); + s->gzindex = 0; } - else - s->status = COMMENT_STATE; + s->status = COMMENT_STATE; } if (s->status == COMMENT_STATE) { if (s->gzhead->comment != Z_NULL) { - uint32_t beg = s->pending; /* start of bytes to update crc */ + ulg beg = s->pending; /* start of bytes to update crc */ int val; - do { if (s->pending == s->pending_buf_size) { - if (s->gzhead->hcrc && s->pending > beg) - strm->adler = crc32(strm->adler, s->pending_buf + beg, - s->pending - beg); + HCRC_UPDATE(beg); flush_pending(strm); - beg = s->pending; - if (s->pending == s->pending_buf_size) { - val = 1; - break; + if (s->pending != 0) { + s->last_flush = -1; + return Z_OK; } + beg = 0; } val = s->gzhead->comment[s->gzindex++]; put_byte(s, val); } while (val != 0); - if (s->gzhead->hcrc && s->pending > beg) - strm->adler = crc32(strm->adler, s->pending_buf + beg, - s->pending - beg); - if (val == 0) - s->status = HCRC_STATE; + HCRC_UPDATE(beg); } - else - s->status = HCRC_STATE; + s->status = HCRC_STATE; } if (s->status == HCRC_STATE) { if (s->gzhead->hcrc) { - if (s->pending + 2 > s->pending_buf_size) + if (s->pending + 2 > s->pending_buf_size) { flush_pending(strm); - if (s->pending + 2 <= s->pending_buf_size) { - put_byte(s, (uint8_t)(strm->adler & 0xff)); - put_byte(s, (uint8_t)((strm->adler >> 8) & 0xff)); - strm->adler = crc32(0L, Z_NULL, 0); - s->status = BUSY_STATE; + if (s->pending != 0) { + s->last_flush = -1; + return Z_OK; + } } + put_byte(s, (Byte)(strm->adler & 0xff)); + put_byte(s, (Byte)((strm->adler >> 8) & 0xff)); + strm->adler = crc32(0L, Z_NULL, 0); } - else - s->status = BUSY_STATE; - } + s->status = BUSY_STATE; - /* Flush as much pending output as possible */ - if (s->pending != 0) { + /* Compression must start with an empty pending buffer */ flush_pending(strm); - if (strm->avail_out == 0) { - /* Since avail_out is 0, deflate will be called again with - * more output space, but possibly with both pending and - * avail_in equal to zero. There won't be anything to do, - * but this is not an error situation so make sure we - * return OK instead of BUF_ERROR at next call of deflate: - */ + if (s->pending != 0) { s->last_flush = -1; return Z_OK; } - - /* Make sure there is something to do and avoid duplicate consecutive - * flushes. For repeated and useless calls with Z_FINISH, we keep - * returning Z_STREAM_END instead of Z_BUF_ERROR. - */ - } else if (strm->avail_in == 0 && RANK(flush) <= RANK(old_flush) && - flush != Z_FINISH) { - ERR_RETURN(strm, Z_BUF_ERROR); - } - - /* User must not provide more input after the first FINISH: */ - if (s->status == FINISH_STATE && strm->avail_in != 0) { - ERR_RETURN(strm, Z_BUF_ERROR); } +#endif /* Start a new block or continue the current one. */ @@ -902,7 +1065,7 @@ int ZEXPORT deflate(z_streamp strm, int flush) { if (flush == Z_PARTIAL_FLUSH) { _tr_align(s); } else if (flush != Z_BLOCK) { /* FULL_FLUSH or SYNC_FLUSH */ - _tr_stored_block(s, (uint8_t*)0, 0L, 0); + _tr_stored_block(s, (char*)0, 0L, 0); /* For a full flush, this empty block will be recognized * as a special marker by inflate_sync(). */ @@ -922,26 +1085,27 @@ int ZEXPORT deflate(z_streamp strm, int flush) { } } } - Assert(strm->avail_out > 0, "bug2"); if (flush != Z_FINISH) return Z_OK; if (s->wrap <= 0) return Z_STREAM_END; /* Write the trailer */ +#ifdef GZIP if (s->wrap == 2) { - put_byte(s, (uint8_t)(strm->adler & 0xff)); - put_byte(s, (uint8_t)((strm->adler >> 8) & 0xff)); - put_byte(s, (uint8_t)((strm->adler >> 16) & 0xff)); - put_byte(s, (uint8_t)((strm->adler >> 24) & 0xff)); - put_byte(s, (uint8_t)(strm->total_in & 0xff)); - put_byte(s, (uint8_t)((strm->total_in >> 8) & 0xff)); - put_byte(s, (uint8_t)((strm->total_in >> 16) & 0xff)); - put_byte(s, (uint8_t)((strm->total_in >> 24) & 0xff)); + put_byte(s, (Byte)(strm->adler & 0xff)); + put_byte(s, (Byte)((strm->adler >> 8) & 0xff)); + put_byte(s, (Byte)((strm->adler >> 16) & 0xff)); + put_byte(s, (Byte)((strm->adler >> 24) & 0xff)); + put_byte(s, (Byte)(strm->total_in & 0xff)); + put_byte(s, (Byte)((strm->total_in >> 8) & 0xff)); + put_byte(s, (Byte)((strm->total_in >> 16) & 0xff)); + put_byte(s, (Byte)((strm->total_in >> 24) & 0xff)); } else +#endif { - putShortMSB(s, (uint32_t)(strm->adler >> 16)); - putShortMSB(s, (uint32_t)(strm->adler & 0xffff)); + putShortMSB(s, (uInt)(strm->adler >> 16)); + putShortMSB(s, (uInt)(strm->adler & 0xffff)); } flush_pending(strm); /* If avail_out is zero, the application will call deflate again @@ -952,7 +1116,9 @@ int ZEXPORT deflate(z_streamp strm, int flush) { } /* ========================================================================= */ -int ZEXPORT deflateEnd(z_streamp strm) { +int ZEXPORT deflateEnd (strm) + z_streamp strm; +{ int status; if (deflateStateCheck(strm)) return Z_STREAM_ERROR; @@ -976,7 +1142,13 @@ int ZEXPORT deflateEnd(z_streamp strm) { * To simplify the source, this is not supported for 16-bit MSDOS (which * doesn't have enough memory anyway to duplicate compression states). */ -int ZEXPORT deflateCopy(z_streamp dest, z_streamp source) { +int ZEXPORT deflateCopy (dest, source) + z_streamp dest; + z_streamp source; +{ +#ifdef MAXSEG_64K + return Z_STREAM_ERROR; +#else deflate_state *ds; deflate_state *ss; @@ -991,14 +1163,14 @@ int ZEXPORT deflateCopy(z_streamp dest, z_streamp source) { ds = (deflate_state *) ZALLOC(dest, 1, sizeof(deflate_state)); if (ds == Z_NULL) return Z_MEM_ERROR; - dest->state = (struct internal_state *) ds; + dest->state = (struct internal_state FAR *) ds; zmemcpy((voidpf)ds, (voidpf)ss, sizeof(deflate_state)); ds->strm = dest; - ds->window = (uint8_t *) ZALLOC(dest, ds->w_size, 2*sizeof(uint8_t)); - ds->prev = (Pos *) ZALLOC(dest, ds->w_size, sizeof(Pos)); - ds->head = (Pos *) ZALLOC(dest, ds->hash_size, sizeof(Pos)); - ds->pending_buf = (uint8_t *) ZALLOC(dest, ds->lit_bufsize, sizeof(uint16_t)+2); + ds->window = (Bytef *) ZALLOC(dest, ds->w_size, 2*sizeof(Byte)); + ds->prev = (Posf *) ZALLOC(dest, ds->w_size, sizeof(Pos)); + ds->head = (Posf *) ZALLOC(dest, ds->hash_size, sizeof(Pos)); + ds->pending_buf = (uchf *) ZALLOC(dest, ds->lit_bufsize, 4); if (ds->window == Z_NULL || ds->prev == Z_NULL || ds->head == Z_NULL || ds->pending_buf == Z_NULL) { @@ -1006,10 +1178,10 @@ int ZEXPORT deflateCopy(z_streamp dest, z_streamp source) { return Z_MEM_ERROR; } /* following zmemcpy do not work for 16-bit MSDOS */ - zmemcpy(ds->window, ss->window, ds->w_size * 2 * sizeof(uint8_t)); + zmemcpy(ds->window, ss->window, ds->w_size * 2 * sizeof(Byte)); zmemcpy((voidpf)ds->prev, (voidpf)ss->prev, ds->w_size * sizeof(Pos)); zmemcpy((voidpf)ds->head, (voidpf)ss->head, ds->hash_size * sizeof(Pos)); - zmemcpy(ds->pending_buf, ss->pending_buf, (uint32_t)ds->pending_buf_size); + zmemcpy(ds->pending_buf, ss->pending_buf, (uInt)ds->pending_buf_size); ds->pending_out = ds->pending_buf + (ss->pending_out - ss->pending_buf); ds->sym_buf = ds->pending_buf + ds->lit_bufsize; @@ -1019,6 +1191,7 @@ int ZEXPORT deflateCopy(z_streamp dest, z_streamp source) { ds->bl_desc.dyn_tree = ds->bl_tree; return Z_OK; +#endif /* MAXSEG_64K */ } /* =========================================================================== @@ -1028,9 +1201,12 @@ int ZEXPORT deflateCopy(z_streamp dest, z_streamp source) { * allocating a large strm->next_in buffer and copying from it. * (See also flush_pending()). */ -static int read_buf(z_streamp strm, uint8_t *buf, uint32_t size) +local unsigned read_buf(strm, buf, size) + z_streamp strm; + Bytef *buf; + unsigned size; { - uint32_t len = strm->avail_in; + unsigned len = strm->avail_in; if (len > size) len = size; if (len == 0) return 0; @@ -1041,21 +1217,24 @@ static int read_buf(z_streamp strm, uint8_t *buf, uint32_t size) if (strm->state->wrap == 1) { strm->adler = adler32(strm->adler, buf, len); } +#ifdef GZIP else if (strm->state->wrap == 2) { strm->adler = crc32(strm->adler, buf, len); } +#endif strm->next_in += len; strm->total_in += len; - return (int)len; + return len; } /* =========================================================================== * Initialize the "longest match" routines for a new zlib stream */ -static void lm_init (deflate_state *s) +local void lm_init (s) + deflate_state *s; { - s->window_size = (uint64_t)2L*s->w_size; + s->window_size = (ulg)2L*s->w_size; CLEAR_HASH(s); @@ -1070,94 +1249,60 @@ static void lm_init (deflate_state *s) s->block_start = 0L; s->lookahead = 0; s->insert = 0; - s->match_length = s->prev_length = ACTUAL_MIN_MATCH-1; + s->match_length = s->prev_length = MIN_MATCH-1; s->match_available = 0; s->ins_h = 0; +#ifndef FASTEST +#ifdef ASMV + match_init(); /* initialize the asm code */ +#endif +#endif } -/* longest_match() with minor change to improve performance (in terms of - * execution time). - * - * The pristine longest_match() function is sketched bellow (strip the - * then-clause of the "#ifdef UNALIGNED_OK"-directive) - * - * ------------------------------------------------------------ - * uInt longest_match(...) { - * ... - * do { - * match = s->window + cur_match; //s0 - * if (*(ushf*)(match+best_len-1) != scan_end || //s1 - * *(ushf*)match != scan_start) continue; //s2 - * ... - * - * do { - * } while (*(ushf*)(scan+=2) == *(ushf*)(match+=2) && - * *(ushf*)(scan+=2) == *(ushf*)(match+=2) && - * *(ushf*)(scan+=2) == *(ushf*)(match+=2) && - * *(ushf*)(scan+=2) == *(ushf*)(match+=2) && - * scan < strend); //s3 - * - * ... - * } while(cond); //s4 - * - * ------------------------------------------------------------- - * - * The change include: - * - * 1) The hottest statements of the function is: s0, s1 and s4. Pull them - * together to form a new loop. The benefit is two-fold: - * - * o. Ease the compiler to yield good code layout: the conditional-branch - * corresponding to s1 and its biased target s4 become very close (likely, - * fit in the same cache-line), hence improving instruction-fetching - * efficiency. - * - * o. Ease the compiler to promote "s->window" into register. "s->window" - * is loop-invariant; it is supposed to be promoted into register and keep - * the value throughout the entire loop. However, there are many such - * loop-invariant, and x86-family has small register file; "s->window" is - * likely to be chosen as register-allocation victim such that its value - * is reloaded from memory in every single iteration. By forming a new loop, - * "s->window" is loop-invariant of that newly created tight loop. It is - * lot easier for compiler to promote this quantity to register and keep - * its value throughout the entire small loop. - * - * 2) Transfrom s3 such that it examines sizeof(long)-byte-match at a time. - * This is done by: - * ------------------------------------------------ - * v1 = load from "scan" by sizeof(long) bytes - * v2 = load from "match" by sizeof(lnog) bytes - * v3 = v1 xor v2 - * match-bit = little-endian-machine(yes-for-x86) ? - * count-trailing-zero(v3) : - * count-leading-zero(v3); - * - * match-byte = match-bit/8 - * - * "scan" and "match" advance if necessary - * ------------------------------------------------- +#ifndef FASTEST +/* =========================================================================== + * Set match_start to the longest match starting at the given string and + * return its length. Matches shorter or equal to prev_length are discarded, + * in which case the result is equal to prev_length and match_start is + * garbage. + * IN assertions: cur_match is the head of the hash chain for the current + * string (strstart) and its distance is <= MAX_DIST, and prev_length >= 1 + * OUT assertion: the match length is not greater than s->lookahead. */ - -static uint32_t longest_match(deflate_state *s, IPos cur_match /* current match */) +#ifndef ASMV +/* For 80x86 and 680x0, an optimized version will be provided in match.asm or + * match.S. The code will be functionally equivalent. + */ +local uInt longest_match(s, cur_match) + deflate_state *s; + IPos cur_match; /* current match */ { - uint32_t chain_length = s->max_chain_length; /* max hash chain length */ - register uint8_t *scan = s->window + s->strstart; /* current string */ - register uint8_t *match; /* matched string */ - register int len; /* length of current match */ - int best_len = (s->prev_length == 0) ? ACTUAL_MIN_MATCH-1 : s->prev_length; /* best match length so far */ - int nice_match = s->nice_match; /* stop if match long enough */ + unsigned chain_length = s->max_chain_length;/* max hash chain length */ + register Bytef *scan = s->window + s->strstart; /* current string */ + register Bytef *match; /* matched string */ + register int len; /* length of current match */ + int best_len = (int)s->prev_length; /* best match length so far */ + int nice_match = s->nice_match; /* stop if match long enough */ IPos limit = s->strstart > (IPos)MAX_DIST(s) ? s->strstart - (IPos)MAX_DIST(s) : NIL; /* Stop when cur_match becomes <= limit. To simplify the code, * we prevent matches with the string of window index 0. */ - Pos *prev = s->prev; - uint32_t wmask = s->w_mask; + Posf *prev = s->prev; + uInt wmask = s->w_mask; - register uint8_t *strend = s->window + s->strstart + MAX_MATCH; - /* We optimize for a minimal match of four bytes */ - register uint32_t scan_start = *(uint32_t*)scan; - register uint32_t scan_end = *(uint32_t*)(scan+best_len-3); +#ifdef UNALIGNED_OK + /* Compare two bytes at a time. Note: this is not always beneficial. + * Try with and without -DUNALIGNED_OK to check. + */ + register Bytef *strend = s->window + s->strstart + MAX_MATCH - 1; + register ush scan_start = *(ushf*)scan; + register ush scan_end = *(ushf*)(scan+best_len-1); +#else + register Bytef *strend = s->window + s->strstart + MAX_MATCH; + register Byte scan_end1 = scan[best_len-1]; + register Byte scan_end = scan[best_len]; +#endif /* The code is optimized for HASH_BITS >= 8 and MAX_MATCH-2 multiple of 16. * It is easy to get rid of this optimization if necessary. @@ -1171,13 +1316,13 @@ static uint32_t longest_match(deflate_state *s, IPos cur_match /* current match /* Do not look for matches beyond the end of the input. This is necessary * to make deflate deterministic. */ - if ((uint32_t)nice_match > s->lookahead) nice_match = s->lookahead; + if ((uInt)nice_match > s->lookahead) nice_match = (int)s->lookahead; - Assert((uint64_t)s->strstart <= s->window_size-MIN_LOOKAHEAD, "need lookahead"); + Assert((ulg)s->strstart <= s->window_size-MIN_LOOKAHEAD, "need lookahead"); do { - int cont ; Assert(cur_match < s->strstart, "no future"); + match = s->window + cur_match; /* Skip to next match if the match length cannot increase * or if the match length is less than 2. Note that the checks below @@ -1187,58 +1332,150 @@ static uint32_t longest_match(deflate_state *s, IPos cur_match /* current match * However the length of the match is limited to the lookahead, so * the output of deflate is not affected by the uninitialized values. */ - cont = 1; +#if (defined(UNALIGNED_OK) && MAX_MATCH == 258) + /* This code assumes sizeof(unsigned short) == 2. Do not use + * UNALIGNED_OK if your compiler uses a different size. + */ + if (*(ushf*)(match+best_len-1) != scan_end || + *(ushf*)match != scan_start) continue; + + /* It is not necessary to compare scan[2] and match[2] since they are + * always equal when the other bytes match, given that the hash keys + * are equal and that HASH_BITS >= 8. Compare 2 bytes at a time at + * strstart+3, +5, ... up to strstart+257. We check for insufficient + * lookahead only every 4th comparison; the 128th check will be made + * at strstart+257. If MAX_MATCH-2 is not a multiple of 8, it is + * necessary to put more guard bytes at the end of the window, or + * to check more often for insufficient lookahead. + */ + Assert(scan[2] == match[2], "scan[2]?"); + scan++, match++; do { - match = s->window + cur_match; - if (likely(*(uint32_t*)(match+best_len-3) != scan_end) || (*(uint32_t*)match != scan_start)) { - if ((cur_match = prev[cur_match & wmask]) > limit - && --chain_length != 0) { - continue; - } else - cont = 0; - } - break; - } while (1); - - if (!cont) - break; + } while (*(ushf*)(scan+=2) == *(ushf*)(match+=2) && + *(ushf*)(scan+=2) == *(ushf*)(match+=2) && + *(ushf*)(scan+=2) == *(ushf*)(match+=2) && + *(ushf*)(scan+=2) == *(ushf*)(match+=2) && + scan < strend); + /* The funny "do {}" generates better code on most compilers */ + + /* Here, scan <= window+strstart+257 */ + Assert(scan <= s->window+(unsigned)(s->window_size-1), "wild scan"); + if (*scan == *match) scan++; + + len = (MAX_MATCH - 1) - (int)(strend-scan); + scan = strend - (MAX_MATCH-1); + +#else /* UNALIGNED_OK */ + + if (match[best_len] != scan_end || + match[best_len-1] != scan_end1 || + *match != *scan || + *++match != scan[1]) continue; + + /* The check at best_len-1 can be removed because it will be made + * again later. (This heuristic is not always a win.) + * It is not necessary to compare scan[2] and match[2] since they + * are always equal when the other bytes match, given that + * the hash keys are equal and that HASH_BITS >= 8. + */ + scan += 2, match++; + Assert(*scan == *match, "match[2]?"); - scan += 4, match+=4; + /* We check for insufficient lookahead only every 8th comparison; + * the 256th check will be made at strstart+258. + */ do { - uint64_t sv = *(uint64_t*)(void*)scan; - uint64_t mv = *(uint64_t*)(void*)match; - uint64_t xor = sv ^ mv; - if (xor) { - int match_byte = __builtin_ctzl(xor) / 8; - scan += match_byte; - match += match_byte; - break; - } else { - scan += 8; - match += 8; - } - } while (scan < strend); - - if (scan > strend) - scan = strend; + } while (*++scan == *++match && *++scan == *++match && + *++scan == *++match && *++scan == *++match && + *++scan == *++match && *++scan == *++match && + *++scan == *++match && *++scan == *++match && + scan < strend); - Assert(scan <= s->window+(uint32_t)(s->window_size-1), "wild scan"); + Assert(scan <= s->window+(unsigned)(s->window_size-1), "wild scan"); len = MAX_MATCH - (int)(strend - scan); scan = strend - MAX_MATCH; +#endif /* UNALIGNED_OK */ + if (len > best_len) { s->match_start = cur_match; best_len = len; if (len >= nice_match) break; - scan_end = *(uint32_t*)(scan+best_len-3); +#ifdef UNALIGNED_OK + scan_end = *(ushf*)(scan+best_len-1); +#else + scan_end1 = scan[best_len-1]; + scan_end = scan[best_len]; +#endif } } while ((cur_match = prev[cur_match & wmask]) > limit && --chain_length != 0); - if ((uint32_t)best_len <= s->lookahead) return (uint32_t)best_len; + if ((uInt)best_len <= s->lookahead) return (uInt)best_len; return s->lookahead; } +#endif /* ASMV */ + +#else /* FASTEST */ + +/* --------------------------------------------------------------------------- + * Optimized version for FASTEST only + */ +local uInt longest_match(s, cur_match) + deflate_state *s; + IPos cur_match; /* current match */ +{ + register Bytef *scan = s->window + s->strstart; /* current string */ + register Bytef *match; /* matched string */ + register int len; /* length of current match */ + register Bytef *strend = s->window + s->strstart + MAX_MATCH; + + /* The code is optimized for HASH_BITS >= 8 and MAX_MATCH-2 multiple of 16. + * It is easy to get rid of this optimization if necessary. + */ + Assert(s->hash_bits >= 8 && MAX_MATCH == 258, "Code too clever"); + + Assert((ulg)s->strstart <= s->window_size-MIN_LOOKAHEAD, "need lookahead"); + + Assert(cur_match < s->strstart, "no future"); + + match = s->window + cur_match; + + /* Return failure if the match length is less than 2: + */ + if (match[0] != scan[0] || match[1] != scan[1]) return MIN_MATCH-1; + + /* The check at best_len-1 can be removed because it will be made + * again later. (This heuristic is not always a win.) + * It is not necessary to compare scan[2] and match[2] since they + * are always equal when the other bytes match, given that + * the hash keys are equal and that HASH_BITS >= 8. + */ + scan += 2, match += 2; + Assert(*scan == *match, "match[2]?"); + + /* We check for insufficient lookahead only every 8th comparison; + * the 256th check will be made at strstart+258. + */ + do { + } while (*++scan == *++match && *++scan == *++match && + *++scan == *++match && *++scan == *++match && + *++scan == *++match && *++scan == *++match && + *++scan == *++match && *++scan == *++match && + scan < strend); + + Assert(scan <= s->window+(unsigned)(s->window_size-1), "wild scan"); + + len = MAX_MATCH - (int)(strend - scan); + + if (len < MIN_MATCH) return MIN_MATCH - 1; + + s->match_start = cur_match; + return (uInt)len <= s->lookahead ? (uInt)len : s->lookahead; +} + +#endif /* FASTEST */ #ifdef ZLIB_DEBUG @@ -1248,7 +1485,11 @@ static uint32_t longest_match(deflate_state *s, IPos cur_match /* current match /* =========================================================================== * Check that the match at match_start is indeed a match. */ -static void check_match(deflate_state *s, IPos start, IPos match, int length) { +local void check_match(s, start, match, length) + deflate_state *s; + IPos start, match; + int length; +{ /* check that the match is indeed a match */ if (zmemcmp(s->window + match, s->window + start, length) != EQUAL) { @@ -1260,7 +1501,7 @@ static void check_match(deflate_state *s, IPos start, IPos match, int length) { z_error("invalid match"); } if (z_verbose > 1) { - fprintf(stderr,"\\[%d,%d]", start - match, length); + fprintf(stderr,"\\[%d,%d]", start-match, length); do { putc(s->window[start++], stderr); } while (--length != 0); } } @@ -1278,17 +1519,17 @@ static void check_match(deflate_state *s, IPos start, IPos match, int length) { * performed for at least two bytes (required for the zip translate_eol * option -- not supported here). */ -static void fill_window(deflate_state *s) +local void fill_window(s) + deflate_state *s; { - register uint32_t n, m; - register Pos *p; - uint32_t more; /* Amount of free space at the end of the window. */ - uint32_t wsize = s->w_size; + unsigned n; + unsigned more; /* Amount of free space at the end of the window. */ + uInt wsize = s->w_size; Assert(s->lookahead < MIN_LOOKAHEAD, "already enough lookahead"); do { - more = (unsigned)(s->window_size -(uint64_t)s->lookahead -(ulg)s->strstart); + more = (unsigned)(s->window_size -(ulg)s->lookahead -(ulg)s->strstart); /* Deal with !@#$% 64K limit: */ if (sizeof(int) <= 2) { @@ -1306,57 +1547,15 @@ static void fill_window(deflate_state *s) /* If the window is almost full and there is insufficient lookahead, * move the upper half to the lower one to make room in the upper half. */ - if (s->strstart >= wsize+MAX_DIST(s)) { - int i; - zmemcpy(s->window, s->window+wsize, (unsigned)wsize); + zmemcpy(s->window, s->window+wsize, (unsigned)wsize - more); s->match_start -= wsize; - s->strstart -= wsize; - s->block_start -= (int64_t) wsize; - n = s->hash_size; - -#ifdef __aarch64__ - - uint16x8_t W; - uint16_t *q ; - W = vmovq_n_u16(wsize); - q = (uint16_t*)s->head; - - for(i=0; i<n/8; i++) { - vst1q_u16(q, vqsubq_u16(vld1q_u16(q), W)); - q+=8; - } - - n = wsize; - q = (uint16_t*)s->prev; - - for(i=0; i<n/8; i++) { - vst1q_u16(q, vqsubq_u16(vld1q_u16(q), W)); - q+=8; - } - -#elif defined HAS_SSE2 - - __m128i W; - __m128i *q; - W = _mm_set1_epi16(wsize); - q = (__m128i*)s->head; - - for(i=0; i<n/8; i++) { - _mm_storeu_si128(q, _mm_subs_epu16(_mm_loadu_si128(q), W)); - q++; - } - - n = wsize; - q = (__m128i*)s->prev; - - for(i=0; i<n/8; i++) { - _mm_storeu_si128(q, _mm_subs_epu16(_mm_loadu_si128(q), W)); - q++; - } - -#endif + s->strstart -= wsize; /* we now have strstart >= MAX_DIST */ + s->block_start -= (long) wsize; + if (s->insert > s->strstart) + s->insert = s->strstart; + slide_hash(s); more += wsize; } if (s->strm->avail_in == 0) break; @@ -1378,21 +1577,26 @@ static void fill_window(deflate_state *s) s->lookahead += n; /* Initialize the hash value now that we have some input: */ - if (s->lookahead + s->insert >= ACTUAL_MIN_MATCH) { - uint32_t str = s->strstart - s->insert; - uint32_t ins_h = s->window[str]; + if (s->lookahead + s->insert >= MIN_MATCH) { + uInt str = s->strstart - s->insert; + s->ins_h = s->window[str]; + UPDATE_HASH(s, s->ins_h, s->window[str + 1]); +#if MIN_MATCH != 3 + Call UPDATE_HASH() MIN_MATCH-3 more times +#endif while (s->insert) { - ins_h = hash_func(s, &s->window[str]); - s->prev[str & s->w_mask] = s->head[ins_h]; - s->head[ins_h] = (Pos)str; + UPDATE_HASH(s, s->ins_h, s->window[str + MIN_MATCH-1]); +#ifndef FASTEST + s->prev[str & s->w_mask] = s->head[s->ins_h]; +#endif + s->head[s->ins_h] = (Pos)str; str++; s->insert--; - if (s->lookahead + s->insert < ACTUAL_MIN_MATCH) + if (s->lookahead + s->insert < MIN_MATCH) break; } - s->ins_h = ins_h; } - /* If the whole input has less than ACTUAL_MIN_MATCH bytes, ins_h is garbage, + /* If the whole input has less than MIN_MATCH bytes, ins_h is garbage, * but this is not important since only literal bytes will be emitted. */ @@ -1406,8 +1610,8 @@ static void fill_window(deflate_state *s) * routines allow scanning to strstart + MAX_MATCH, ignoring lookahead. */ if (s->high_water < s->window_size) { - uint64_t curr = s->strstart + (ulg)(s->lookahead); - uint64_t init; + ulg curr = s->strstart + (ulg)(s->lookahead); + ulg init; if (s->high_water < curr) { /* Previous high water mark below current data -- zero WIN_INIT @@ -1419,12 +1623,12 @@ static void fill_window(deflate_state *s) zmemzero(s->window + curr, (unsigned)init); s->high_water = curr + init; } - else if (s->high_water < (uint64_t)curr + WIN_INIT) { + else if (s->high_water < (ulg)curr + WIN_INIT) { /* High water mark at or above current data, but below current data * plus WIN_INIT -- zero out to current data plus WIN_INIT, or up * to end of window, whichever is less. */ - init = (uint64_t)curr + WIN_INIT - s->high_water; + init = (ulg)curr + WIN_INIT - s->high_water; if (init > s->window_size - s->high_water) init = s->window_size - s->high_water; zmemzero(s->window + s->high_water, (unsigned)init); @@ -1432,7 +1636,7 @@ static void fill_window(deflate_state *s) } } - Assert((uint64_t)s->strstart <= s->window_size - MIN_LOOKAHEAD, + Assert((ulg)s->strstart <= s->window_size - MIN_LOOKAHEAD, "not enough room for search"); } @@ -1442,9 +1646,9 @@ static void fill_window(deflate_state *s) */ #define FLUSH_BLOCK_ONLY(s, last) { \ _tr_flush_block(s, (s->block_start >= 0L ? \ - (uint8_t *)&s->window[(uint64_t)s->block_start] : \ - (uint8_t *)Z_NULL), \ - (uint64_t)((int64_t)s->strstart - s->block_start), \ + (charf *)&s->window[(unsigned)s->block_start] : \ + (charf *)Z_NULL), \ + (ulg)((long)s->strstart - s->block_start), \ (last)); \ s->block_start = s->strstart; \ flush_pending(s->strm); \ @@ -1457,68 +1661,205 @@ static void fill_window(deflate_state *s) if (s->strm->avail_out == 0) return (last) ? finish_started : need_more; \ } +/* Maximum stored block length in deflate format (not including header). */ +#define MAX_STORED 65535 + +/* Minimum of a and b. */ +#define MIN(a, b) ((a) > (b) ? (b) : (a)) + /* =========================================================================== * Copy without compression as much as possible from the input stream, return * the current block state. - * This function does not insert new strings in the dictionary since - * uncompressible data is probably not useful. This function is used - * only for the level=0 compression option. - * NOTE: this function should be optimized to avoid extra copying from - * window to pending_buf. + * + * In case deflateParams() is used to later switch to a non-zero compression + * level, s->matches (otherwise unused when storing) keeps track of the number + * of hash table slides to perform. If s->matches is 1, then one hash table + * slide will be done when switching. If s->matches is 2, the maximum value + * allowed here, then the hash table will be cleared, since two or more slides + * is the same as a clear. + * + * deflate_stored() is written to minimize the number of times an input byte is + * copied. It is most efficient with large input and output buffers, which + * maximizes the opportunites to have a single copy from next_in to next_out. */ -static block_state deflate_stored(deflate_state *s, int flush) +local block_state deflate_stored(s, flush) + deflate_state *s; + int flush; { - /* Stored blocks are limited to 0xffff bytes, pending_buf is limited - * to pending_buf_size, and each stored block has a 5 byte header: + /* Smallest worthy block size when not flushing or finishing. By default + * this is 32K. This can be as small as 507 bytes for memLevel == 1. For + * large input and output buffers, the stored block size will be larger. */ - uint64_t max_block_size = 0xffff; - uint64_t max_start; + unsigned min_block = MIN(s->pending_buf_size - 5, s->w_size); - if (max_block_size > s->pending_buf_size - 6) { - max_block_size = s->pending_buf_size - 6; - } + /* Copy as many min_block or larger stored blocks directly to next_out as + * possible. If flushing, copy the remaining available input to next_out as + * stored blocks, if there is enough space. + */ + unsigned len, left, have, last = 0; + unsigned used = s->strm->avail_in; + do { + /* Set len to the maximum size block that we can copy directly with the + * available input data and output space. Set left to how much of that + * would be copied from what's left in the window. + */ + len = MAX_STORED; /* maximum deflate stored block length */ + have = (s->bi_valid + 42) >> 3; /* number of header bytes */ + if (s->strm->avail_out < have) /* need room for header */ + break; + /* maximum stored block length that will fit in avail_out: */ + have = s->strm->avail_out - have; + left = s->strstart - s->block_start; /* bytes left in window */ + if (len > (ulg)left + s->strm->avail_in) + len = left + s->strm->avail_in; /* limit len to the input */ + if (len > have) + len = have; /* limit len to the output */ + + /* If the stored block would be less than min_block in length, or if + * unable to copy all of the available input when flushing, then try + * copying to the window and the pending buffer instead. Also don't + * write an empty block when flushing -- deflate() does that. + */ + if (len < min_block && ((len == 0 && flush != Z_FINISH) || + flush == Z_NO_FLUSH || + len != left + s->strm->avail_in)) + break; - /* Copy as much as possible from input to output: */ - for (;;) { - /* Fill the window as much as possible: */ - if (s->lookahead <= 1) { + /* Make a dummy stored block in pending to get the header bytes, + * including any pending bits. This also updates the debugging counts. + */ + last = flush == Z_FINISH && len == left + s->strm->avail_in ? 1 : 0; + _tr_stored_block(s, (char *)0, 0L, last); - Assert(s->strstart < s->w_size+MAX_DIST(s) || - s->block_start >= (int64_t)s->w_size, "slide too late"); + /* Replace the lengths in the dummy stored block with len. */ + s->pending_buf[s->pending - 4] = len; + s->pending_buf[s->pending - 3] = len >> 8; + s->pending_buf[s->pending - 2] = ~len; + s->pending_buf[s->pending - 1] = ~len >> 8; - fill_window(s); - if (s->lookahead == 0 && flush == Z_NO_FLUSH) return need_more; + /* Write the stored block header bytes. */ + flush_pending(s->strm); - if (s->lookahead == 0) break; /* flush the current block */ +#ifdef ZLIB_DEBUG + /* Update debugging counts for the data about to be copied. */ + s->compressed_len += len << 3; + s->bits_sent += len << 3; +#endif + + /* Copy uncompressed bytes from the window to next_out. */ + if (left) { + if (left > len) + left = len; + zmemcpy(s->strm->next_out, s->window + s->block_start, left); + s->strm->next_out += left; + s->strm->avail_out -= left; + s->strm->total_out += left; + s->block_start += left; + len -= left; } - Assert(s->block_start >= 0L, "block gone"); - - s->strstart += s->lookahead; - s->lookahead = 0; - - /* Emit a stored block if pending_buf will be full: */ - max_start = s->block_start + max_block_size; - if (s->strstart == 0 || (uint64_t)s->strstart >= max_start) { - /* strstart == 0 is possible when wraparound on 16-bit machine */ - s->lookahead = (uint32_t)(s->strstart - max_start); - s->strstart = (uint32_t)max_start; - FLUSH_BLOCK(s, 0); + + /* Copy uncompressed bytes directly from next_in to next_out, updating + * the check value. + */ + if (len) { + read_buf(s->strm, s->strm->next_out, len); + s->strm->next_out += len; + s->strm->avail_out -= len; + s->strm->total_out += len; } - /* Flush if we may have to slide, otherwise block_start may become - * negative and the data will be gone: + } while (last == 0); + + /* Update the sliding window with the last s->w_size bytes of the copied + * data, or append all of the copied data to the existing window if less + * than s->w_size bytes were copied. Also update the number of bytes to + * insert in the hash tables, in the event that deflateParams() switches to + * a non-zero compression level. + */ + used -= s->strm->avail_in; /* number of input bytes directly copied */ + if (used) { + /* If any input was used, then no unused input remains in the window, + * therefore s->block_start == s->strstart. */ - if (s->strstart - (uint32_t)s->block_start >= MAX_DIST(s)) { - FLUSH_BLOCK(s, 0); + if (used >= s->w_size) { /* supplant the previous history */ + s->matches = 2; /* clear hash */ + zmemcpy(s->window, s->strm->next_in - s->w_size, s->w_size); + s->strstart = s->w_size; + s->insert = s->strstart; + } + else { + if (s->window_size - s->strstart <= used) { + /* Slide the window down. */ + s->strstart -= s->w_size; + zmemcpy(s->window, s->window + s->w_size, s->strstart); + if (s->matches < 2) + s->matches++; /* add a pending slide_hash() */ + if (s->insert > s->strstart) + s->insert = s->strstart; + } + zmemcpy(s->window + s->strstart, s->strm->next_in - used, used); + s->strstart += used; + s->insert += MIN(used, s->w_size - s->insert); } + s->block_start = s->strstart; } - s->insert = 0; - if (flush == Z_FINISH) { - FLUSH_BLOCK(s, 1); + if (s->high_water < s->strstart) + s->high_water = s->strstart; + + /* If the last block was written to next_out, then done. */ + if (last) return finish_done; + + /* If flushing and all input has been consumed, then done. */ + if (flush != Z_NO_FLUSH && flush != Z_FINISH && + s->strm->avail_in == 0 && (long)s->strstart == s->block_start) + return block_done; + + /* Fill the window with any remaining input. */ + have = s->window_size - s->strstart; + if (s->strm->avail_in > have && s->block_start >= (long)s->w_size) { + /* Slide the window down. */ + s->block_start -= s->w_size; + s->strstart -= s->w_size; + zmemcpy(s->window, s->window + s->w_size, s->strstart); + if (s->matches < 2) + s->matches++; /* add a pending slide_hash() */ + have += s->w_size; /* more space now */ + if (s->insert > s->strstart) + s->insert = s->strstart; } - if ((int64_t)s->strstart > s->block_start) - FLUSH_BLOCK(s, 0); - return block_done; + if (have > s->strm->avail_in) + have = s->strm->avail_in; + if (have) { + read_buf(s->strm, s->window + s->strstart, have); + s->strstart += have; + s->insert += MIN(have, s->w_size - s->insert); + } + if (s->high_water < s->strstart) + s->high_water = s->strstart; + + /* There was not enough avail_out to write a complete worthy or flushed + * stored block to next_out. Write a stored block to pending instead, if we + * have enough input for a worthy block, or if flushing and there is enough + * room for the remaining input as a stored block in the pending buffer. + */ + have = (s->bi_valid + 42) >> 3; /* number of header bytes */ + /* maximum stored block length that will fit in pending: */ + have = MIN(s->pending_buf_size - have, MAX_STORED); + min_block = MIN(have, s->w_size); + left = s->strstart - s->block_start; + if (left >= min_block || + ((left || flush == Z_FINISH) && flush != Z_NO_FLUSH && + s->strm->avail_in == 0 && left <= have)) { + len = MIN(left, have); + last = flush == Z_FINISH && s->strm->avail_in == 0 && + len == left ? 1 : 0; + _tr_stored_block(s, (charf *)s->window + s->block_start, len, last); + s->block_start += len; + flush_pending(s->strm); + } + + /* We've done all we can with the available input and output. */ + return last ? finish_started : need_more; } /* =========================================================================== @@ -1528,14 +1869,17 @@ static block_state deflate_stored(deflate_state *s, int flush) * new strings in the dictionary only for unmatched strings or for short * matches. It is used only for the fast compression options. */ -static block_state deflate_fast(deflate_state *s, int flush) { +local block_state deflate_fast(s, flush) + deflate_state *s; + int flush; +{ IPos hash_head; /* head of the hash chain */ int bflush; /* set if current block must be flushed */ for (;;) { /* Make sure that we always have enough lookahead, except * at the end of the input file. We need MAX_MATCH bytes - * for the next match, plus ACTUAL_MIN_MATCH bytes to insert the + * for the next match, plus MIN_MATCH bytes to insert the * string following the next match. */ if (s->lookahead < MIN_LOOKAHEAD) { @@ -1546,16 +1890,16 @@ static block_state deflate_fast(deflate_state *s, int flush) { if (s->lookahead == 0) break; /* flush the current block */ } - /* Insert the string window[strstart .. strstart + 2] in the + /* Insert the string window[strstart .. strstart+2] in the * dictionary, and set hash_head to the head of the hash chain: */ hash_head = NIL; - if (s->lookahead >= ACTUAL_MIN_MATCH) { - hash_head = insert_string(s, s->strstart); + if (s->lookahead >= MIN_MATCH) { + INSERT_STRING(s, s->strstart, hash_head); } /* Find the longest match, discarding those <= prev_length. - * At this point we have always match_length < ACTUAL_MIN_MATCH + * At this point we have always match_length < MIN_MATCH */ if (hash_head != NIL && s->strstart - hash_head <= MAX_DIST(s)) { /* To simplify the code, we prevent matches with the string @@ -1565,7 +1909,7 @@ static block_state deflate_fast(deflate_state *s, int flush) { s->match_length = longest_match (s, hash_head); /* longest_match() sets match_start */ } - if (s->match_length >= ACTUAL_MIN_MATCH) { + if (s->match_length >= MIN_MATCH) { check_match(s, s->strstart, s->match_start, s->match_length); _tr_tally_dist(s, s->strstart - s->match_start, @@ -1576,34 +1920,42 @@ static block_state deflate_fast(deflate_state *s, int flush) { /* Insert new strings in the hash table only if the match length * is not too large. This saves time but degrades compression. */ +#ifndef FASTEST if (s->match_length <= s->max_insert_length && - s->lookahead >= ACTUAL_MIN_MATCH) { + s->lookahead >= MIN_MATCH) { s->match_length--; /* string at strstart already in table */ do { s->strstart++; - hash_head = insert_string(s, s->strstart); + INSERT_STRING(s, s->strstart, hash_head); /* strstart never exceeds WSIZE-MAX_MATCH, so there are - * always ACTUAL_MIN_MATCH bytes ahead. + * always MIN_MATCH bytes ahead. */ } while (--s->match_length != 0); s->strstart++; - } else { + } else +#endif + { s->strstart += s->match_length; s->match_length = 0; - /* If lookahead < ACTUAL_MIN_MATCH, ins_h is garbage, but it does not + s->ins_h = s->window[s->strstart]; + UPDATE_HASH(s, s->ins_h, s->window[s->strstart+1]); +#if MIN_MATCH != 3 + Call UPDATE_HASH() MIN_MATCH-3 more times +#endif + /* If lookahead < MIN_MATCH, ins_h is garbage, but it does not * matter since it will be recomputed at next deflate call. */ } } else { /* No match, output a literal byte */ Tracevv((stderr,"%c", s->window[s->strstart])); - _tr_tally_lit(s, s->window[s->strstart], bflush); + _tr_tally_lit (s, s->window[s->strstart], bflush); s->lookahead--; s->strstart++; } if (bflush) FLUSH_BLOCK(s, 0); } - s->insert = s->strstart < ACTUAL_MIN_MATCH-1 ? s->strstart : ACTUAL_MIN_MATCH-1; + s->insert = s->strstart < MIN_MATCH-1 ? s->strstart : MIN_MATCH-1; if (flush == Z_FINISH) { FLUSH_BLOCK(s, 1); return finish_done; @@ -1613,12 +1965,16 @@ static block_state deflate_fast(deflate_state *s, int flush) { return block_done; } +#ifndef FASTEST /* =========================================================================== * Same as above, but achieves better compression. We use a lazy * evaluation for matches: a match is finally adopted only if there is * no better match at the next window position. */ -static block_state deflate_slow(deflate_state *s, int flush) { +local block_state deflate_slow(s, flush) + deflate_state *s; + int flush; +{ IPos hash_head; /* head of hash chain */ int bflush; /* set if current block must be flushed */ @@ -1626,7 +1982,7 @@ static block_state deflate_slow(deflate_state *s, int flush) { for (;;) { /* Make sure that we always have enough lookahead, except * at the end of the input file. We need MAX_MATCH bytes - * for the next match, plus ACTUAL_MIN_MATCH bytes to insert the + * for the next match, plus MIN_MATCH bytes to insert the * string following the next match. */ if (s->lookahead < MIN_LOOKAHEAD) { @@ -1637,18 +1993,18 @@ static block_state deflate_slow(deflate_state *s, int flush) { if (s->lookahead == 0) break; /* flush the current block */ } - /* Insert the string window[strstart .. strstart+3] in the + /* Insert the string window[strstart .. strstart+2] in the * dictionary, and set hash_head to the head of the hash chain: */ hash_head = NIL; - if (s->lookahead >= ACTUAL_MIN_MATCH) { - hash_head = insert_string(s, s->strstart); + if (s->lookahead >= MIN_MATCH) { + INSERT_STRING(s, s->strstart, hash_head); } /* Find the longest match, discarding those <= prev_length. */ s->prev_length = s->match_length, s->prev_match = s->match_start; - s->match_length = ACTUAL_MIN_MATCH-1; + s->match_length = MIN_MATCH-1; if (hash_head != NIL && s->prev_length < s->max_lazy_match && s->strstart - hash_head <= MAX_DIST(s)) { @@ -1659,46 +2015,46 @@ static block_state deflate_slow(deflate_state *s, int flush) { s->match_length = longest_match (s, hash_head); /* longest_match() sets match_start */ - if (s->match_length <= 5 && (s->strategy == Z_FILTERED )) { + if (s->match_length <= 5 && (s->strategy == Z_FILTERED +#if TOO_FAR <= 32767 + || (s->match_length == MIN_MATCH && + s->strstart - s->match_start > TOO_FAR) +#endif + )) { - /* If prev_match is also ACTUAL_MIN_MATCH, match_start is garbage + /* If prev_match is also MIN_MATCH, match_start is garbage * but we will ignore the current match anyway. */ - s->match_length = ACTUAL_MIN_MATCH-1; + s->match_length = MIN_MATCH-1; } } /* If there was a match at the previous step and the current * match is not better, output the previous match: */ - if (s->prev_length >= ACTUAL_MIN_MATCH && s->match_length <= s->prev_length) { - uint32_t mov_fwd ; - uint32_t insert_cnt ; - - uint32_t max_insert = s->strstart + s->lookahead - ACTUAL_MIN_MATCH; + if (s->prev_length >= MIN_MATCH && s->match_length <= s->prev_length) { + uInt max_insert = s->strstart + s->lookahead - MIN_MATCH; /* Do not insert strings in hash table beyond this. */ - check_match(s, s->strstart - 1, s->prev_match, s->prev_length); + check_match(s, s->strstart-1, s->prev_match, s->prev_length); - _tr_tally_dist(s, s->strstart - 1 - s->prev_match, + _tr_tally_dist(s, s->strstart -1 - s->prev_match, s->prev_length - MIN_MATCH, bflush); /* Insert in hash table all strings up to the end of the match. - * strstart - 1 and strstart are already inserted. If there is not + * strstart-1 and strstart are already inserted. If there is not * enough lookahead, the last two strings are not inserted in * the hash table. */ s->lookahead -= s->prev_length-1; - - mov_fwd = s->prev_length - 2; - insert_cnt = mov_fwd; - if (unlikely(insert_cnt > max_insert - s->strstart)) - insert_cnt = max_insert - s->strstart; - - bulk_insert_str(s, s->strstart + 1, insert_cnt); - s->prev_length = ACTUAL_MIN_MATCH-1; + s->prev_length -= 2; + do { + if (++s->strstart <= max_insert) { + INSERT_STRING(s, s->strstart, hash_head); + } + } while (--s->prev_length != 0); s->match_available = 0; - s->match_length = ACTUAL_MIN_MATCH-1; - s->strstart += mov_fwd + 1; + s->match_length = MIN_MATCH-1; + s->strstart++; if (bflush) FLUSH_BLOCK(s, 0); @@ -1707,8 +2063,8 @@ static block_state deflate_slow(deflate_state *s, int flush) { * single literal. If there was a match but the current match * is longer, truncate the previous match to a single literal. */ - Tracevv((stderr,"%c", s->window[s->strstart - 1])); - _tr_tally_lit(s, s->window[s->strstart - 1], bflush); + Tracevv((stderr,"%c", s->window[s->strstart-1])); + _tr_tally_lit(s, s->window[s->strstart-1], bflush); if (bflush) { FLUSH_BLOCK_ONLY(s, 0); } @@ -1726,11 +2082,11 @@ static block_state deflate_slow(deflate_state *s, int flush) { } Assert (flush != Z_NO_FLUSH, "no flush?"); if (s->match_available) { - Tracevv((stderr,"%c", s->window[s->strstart - 1])); - _tr_tally_lit(s, s->window[s->strstart - 1], bflush); + Tracevv((stderr,"%c", s->window[s->strstart-1])); + _tr_tally_lit(s, s->window[s->strstart-1], bflush); s->match_available = 0; } - s->insert = s->strstart < ACTUAL_MIN_MATCH-1 ? s->strstart : ACTUAL_MIN_MATCH-1; + s->insert = s->strstart < MIN_MATCH-1 ? s->strstart : MIN_MATCH-1; if (flush == Z_FINISH) { FLUSH_BLOCK(s, 1); return finish_done; @@ -1739,17 +2095,20 @@ static block_state deflate_slow(deflate_state *s, int flush) { FLUSH_BLOCK(s, 0); return block_done; } +#endif /* FASTEST */ /* =========================================================================== * For Z_RLE, simply look for runs of bytes, generate matches only of distance * one. Do not maintain a hash table. (It will be regenerated if this run of * deflate switches away from Z_RLE.) */ -static block_state deflate_rle(deflate_state *s, int flush) +local block_state deflate_rle(s, flush) + deflate_state *s; + int flush; { - int bflush; /* set if current block must be flushed */ - uint32_t prev; /* byte at distance one to match */ - uint8_t *scan, *strend; /* scan goes up to strend for length of run */ + int bflush; /* set if current block must be flushed */ + uInt prev; /* byte at distance one to match */ + Bytef *scan, *strend; /* scan goes up to strend for length of run */ for (;;) { /* Make sure that we always have enough lookahead, except @@ -1766,7 +2125,7 @@ static block_state deflate_rle(deflate_state *s, int flush) /* See how many times the previous byte repeats */ s->match_length = 0; - if (s->lookahead >= ACTUAL_MIN_MATCH && s->strstart > 0) { + if (s->lookahead >= MIN_MATCH && s->strstart > 0) { scan = s->window + s->strstart - 1; prev = *scan; if (prev == *++scan && prev == *++scan && prev == *++scan) { @@ -1777,15 +2136,15 @@ static block_state deflate_rle(deflate_state *s, int flush) prev == *++scan && prev == *++scan && prev == *++scan && prev == *++scan && scan < strend); - s->match_length = MAX_MATCH - (int)(strend - scan); + s->match_length = MAX_MATCH - (uInt)(strend - scan); if (s->match_length > s->lookahead) s->match_length = s->lookahead; } - Assert(scan <= s->window+(uint32_t)(s->window_size-1), "wild scan"); + Assert(scan <= s->window+(uInt)(s->window_size-1), "wild scan"); } - /* Emit match if have run of ACTUAL_MIN_MATCH or longer, else emit literal */ - if (s->match_length >= ACTUAL_MIN_MATCH) { + /* Emit match if have run of MIN_MATCH or longer, else emit literal */ + if (s->match_length >= MIN_MATCH) { check_match(s, s->strstart, s->strstart - 1, s->match_length); _tr_tally_dist(s, 1, s->match_length - MIN_MATCH, bflush); @@ -1796,7 +2155,7 @@ static block_state deflate_rle(deflate_state *s, int flush) } else { /* No match, output a literal byte */ Tracevv((stderr,"%c", s->window[s->strstart])); - _tr_tally_lit(s, s->window[s->strstart], bflush); + _tr_tally_lit (s, s->window[s->strstart], bflush); s->lookahead--; s->strstart++; } @@ -1816,7 +2175,10 @@ static block_state deflate_rle(deflate_state *s, int flush) * For Z_HUFFMAN_ONLY, do not look for matches. Do not maintain a hash table. * (It will be regenerated if this run of deflate switches away from Huffman.) */ -static block_state deflate_huff(deflate_state *s, int flush) { +local block_state deflate_huff(s, flush) + deflate_state *s; + int flush; +{ int bflush; /* set if current block must be flushed */ for (;;) { @@ -1833,7 +2195,7 @@ static block_state deflate_huff(deflate_state *s, int flush) { /* Output a literal byte */ s->match_length = 0; Tracevv((stderr,"%c", s->window[s->strstart])); - _tr_tally_lit(s, s->window[s->strstart], bflush); + _tr_tally_lit (s, s->window[s->strstart], bflush); s->lookahead--; s->strstart++; if (bflush) FLUSH_BLOCK(s, 0); |