X-Git-Url: https://git.gag.com/?a=blobdiff_plain;f=deflate.c;h=951d7af4e895b2285d1617c05a4940c6b7b336b6;hb=92249085071a973e2c0621b0415b93d2e48bb00d;hp=0950391ee88ba44a67a36ce4f1b14c87f79ebaa3;hpb=a2016c1de6e4884f6c8ed5cc498f3bf821c25ca4;p=debian%2Fgzip diff --git a/deflate.c b/deflate.c index 0950391..951d7af 100644 --- a/deflate.c +++ b/deflate.c @@ -1,6 +1,6 @@ /* deflate.c -- compress data using the deflation algorithm - Copyright (C) 1999, 2006, 2009-2010 Free Software Foundation, Inc. + Copyright (C) 1999, 2006, 2009-2018 Free Software Foundation, Inc. Copyright (C) 1992-1993 Jean-loup Gailly This program is free software; you can redistribute it and/or modify @@ -80,6 +80,7 @@ #include "tailor.h" #include "gzip.h" #include "lzw.h" /* just for consistency checking */ +#include "verify.h" /* =========================================================================== * Configuration parameters @@ -131,6 +132,14 @@ #endif /* Matches of length 3 are discarded if their distance exceeds TOO_FAR */ +#ifndef RSYNC_WIN +# define RSYNC_WIN 4096 +#endif +verify(RSYNC_WIN < MAX_DIST); + +#define RSYNC_SUM_MATCH(sum) ((sum) % RSYNC_WIN == 0) +/* Whether window sum matches magic value */ + /* =========================================================================== * Local data used by the "longest match" routines. */ @@ -161,7 +170,7 @@ typedef unsigned IPos; /* DECLARE(Pos, head, 1<= HASH_BITS */ -unsigned int near prev_length; + unsigned int near prev_length; /* Length of the best match at previous step. Matches not greater than this * are discarded. This is used in the lazy match evaluation. */ @@ -190,7 +199,7 @@ unsigned int near prev_length; local int eofile; /* flag set at end of input file */ local unsigned lookahead; /* number of valid bytes ahead in window */ -unsigned near max_chain_length; + unsigned max_chain_length; /* To speed up deflation, hash chains are never searched beyond this length. * A higher limit improves compression ratio but degrades the speed. */ @@ -209,9 +218,11 @@ local unsigned int max_lazy_match; local int compr_level; /* compression level (1..9) */ -unsigned near good_match; +unsigned good_match; /* Use a faster search when the previous match is longer than this */ +local ulg rsync_sum; /* rolling sum of rsync window */ +local ulg rsync_chunk_end; /* next rsync sequence point */ /* 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 @@ -226,10 +237,17 @@ typedef struct config { ush max_chain; } config; +#ifdef ASMV +# define static_unless_ASMV +#else +# define static_unless_ASMV static +#endif + #ifdef FULL_SEARCH # define nice_match MAX_MATCH #else - int near nice_match; /* Stop searching when current match exceeds this */ + /* Stop searching when current match exceeds this */ + static_unless_ASMV int nice_match; #endif local config configuration_table[10] = { @@ -254,21 +272,21 @@ local config configuration_table[10] = { /* =========================================================================== * Prototypes for local functions. */ -local void fill_window OF((void)); -local off_t deflate_fast OF((void)); +local void fill_window (void); +local off_t deflate_fast (void); - int longest_match OF((IPos cur_match)); #ifdef ASMV - void match_init OF((void)); /* asm code initialization */ + int longest_match (IPos cur_match); + void match_init (void); /* asm code initialization */ #endif #ifdef DEBUG -local void check_match OF((IPos start, IPos match, int length)); +local void check_match (IPos start, IPos match, int length); #endif /* =========================================================================== * Update a hash value with the given input byte - * IN assertion: all calls to to UPDATE_HASH are made with consecutive + * 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. */ @@ -278,7 +296,7 @@ local void check_match OF((IPos start, IPos match, int length)); * Insert string s 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. - * IN assertion: all calls to to INSERT_STRING are made with consecutive + * IN assertion: all calls to INSERT_STRING are made with consecutive * input characters and the first MIN_MATCH bytes of s are valid * (except for the last MIN_MATCH-1 bytes of the input file). */ @@ -300,13 +318,17 @@ void lm_init (pack_level, flags) compr_level = pack_level; /* Initialize the hash table. */ -#if defined(MAXSEG_64K) && HASH_BITS == 15 +#if defined MAXSEG_64K && HASH_BITS == 15 for (j = 0; j < HASH_SIZE; j++) head[j] = NIL; #else memzero((char*)head, HASH_SIZE*sizeof(*head)); #endif /* prev will be initialized on the fly */ + /* rsync params */ + rsync_chunk_end = 0xFFFFFFFFUL; + rsync_sum = 0; + /* Set the default configuration parameters: */ max_lazy_match = configuration_table[pack_level].max_lazy; @@ -329,7 +351,7 @@ void lm_init (pack_level, flags) #endif lookahead = read_buf((char*)window, - sizeof(int) <= 2 ? (unsigned)WSIZE : 2*WSIZE); + sizeof(int) <= 2 ? (unsigned)WSIZE : 2*WSIZE); if (lookahead == 0 || lookahead == (unsigned)EOF) { eofile = 1, lookahead = 0; @@ -361,8 +383,8 @@ void lm_init (pack_level, flags) * match.s. The code is functionally equivalent, so you can use the C version * if desired. */ -int longest_match(cur_match) - IPos cur_match; /* current match */ +static int +longest_match(IPos cur_match) { unsigned chain_length = max_chain_length; /* max hash chain length */ register uch *scan = window + strstart; /* current string */ @@ -407,7 +429,7 @@ int longest_match(cur_match) /* Skip to next match if the match length cannot increase * or if the match length is less than 2: */ -#if (defined(UNALIGNED_OK) && MAX_MATCH == 258) +#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. */ @@ -481,7 +503,7 @@ int longest_match(cur_match) #endif } } while ((cur_match = prev[cur_match & WMASK]) > limit - && --chain_length != 0); + && --chain_length != 0); return best_len; } @@ -543,6 +565,8 @@ local void fill_window() memcpy((char*)window, (char*)window+WSIZE, (unsigned)WSIZE); match_start -= WSIZE; strstart -= WSIZE; /* we now have strstart >= MAX_DIST: */ + if (rsync_chunk_end != 0xFFFFFFFFUL) + rsync_chunk_end -= WSIZE; block_start -= (long) WSIZE; @@ -564,19 +588,55 @@ local void fill_window() n = read_buf((char*)window+strstart+lookahead, more); if (n == 0 || n == (unsigned)EOF) { eofile = 1; + /* Don't let garbage pollute the dictionary. */ + memzero (window + strstart + lookahead, MIN_MATCH - 1); } else { lookahead += n; } } } +/* With an initial offset of START, advance rsync's rolling checksum + by NUM bytes. */ +local void rsync_roll(unsigned int start, unsigned int num) +{ + unsigned i; + + if (start < RSYNC_WIN) { + /* before window fills. */ + for (i = start; i < RSYNC_WIN; i++) { + if (i == start + num) + return; + rsync_sum += (ulg)window[i]; + } + num -= (RSYNC_WIN - start); + start = RSYNC_WIN; + } + + /* buffer after window full */ + for (i = start; i < start+num; i++) { + /* New character in */ + rsync_sum += (ulg)window[i]; + /* Old character out */ + rsync_sum -= (ulg)window[i - RSYNC_WIN]; + if (rsync_chunk_end == 0xFFFFFFFFUL && RSYNC_SUM_MATCH(rsync_sum)) + rsync_chunk_end = i; + } +} + +/* =========================================================================== + * Set rsync_chunk_end if window sum matches magic value. + */ +#define RSYNC_ROLL(s, n) \ + do { if (rsync) rsync_roll((s), (n)); } while(0) + /* =========================================================================== * Flush the current block, with given end-of-file flag. * IN assertion: strstart is set to the end of the current match. */ #define FLUSH_BLOCK(eof) \ flush_block(block_start >= 0L ? (char*)&window[(unsigned)block_start] : \ - (char*)NULL, (long)strstart - block_start, (eof)) + (char*)NULL, (long)strstart - block_start, flush-1, (eof)) /* =========================================================================== * Processes a new input file and return its compressed length. This @@ -587,7 +647,7 @@ local void fill_window() local off_t deflate_fast() { IPos hash_head; /* head of the hash chain */ - int flush; /* set if current block must be flushed */ + int flush = 0; /* set if current block must be flushed, 2=>and padded */ unsigned match_length = 0; /* length of best match */ prev_length = MIN_MATCH-1; @@ -601,7 +661,7 @@ local off_t deflate_fast() * At this point we have always match_length < MIN_MATCH */ if (hash_head != NIL && strstart - hash_head <= MAX_DIST - && strstart <= window_size - MIN_LOOKAHEAD) { + && strstart <= window_size - MIN_LOOKAHEAD) { /* To simplify the code, we prevent matches with the string * of window index 0 (in particular we have to avoid a match * of the string with itself at the start of the input file). @@ -617,7 +677,8 @@ local off_t deflate_fast() lookahead -= match_length; - /* Insert new strings in the hash table only if the match length + RSYNC_ROLL(strstart, match_length); + /* Insert new strings in the hash table only if the match length * is not too large. This saves time but degrades compression. */ if (match_length <= max_insert_length) { @@ -631,12 +692,12 @@ local off_t deflate_fast() * the next lookahead bytes will be emitted as literals. */ } while (--match_length != 0); - strstart++; + strstart++; } else { - strstart += match_length; - match_length = 0; - ins_h = window[strstart]; - UPDATE_HASH(ins_h, window[strstart+1]); + strstart += match_length; + match_length = 0; + ins_h = window[strstart]; + UPDATE_HASH(ins_h, window[strstart+1]); #if MIN_MATCH != 3 Call UPDATE_HASH() MIN_MATCH-3 more times #endif @@ -645,8 +706,13 @@ local off_t deflate_fast() /* No match, output a literal byte */ Tracevv((stderr,"%c",window[strstart])); flush = ct_tally (0, window[strstart]); + RSYNC_ROLL(strstart, 1); lookahead--; - strstart++; + strstart++; + } + if (rsync && strstart > rsync_chunk_end) { + rsync_chunk_end = 0xFFFFFFFFUL; + flush = 2; } if (flush) FLUSH_BLOCK(0), block_start = strstart; @@ -670,7 +736,7 @@ off_t deflate() { IPos hash_head; /* head of hash chain */ IPos prev_match; /* previous match */ - int flush; /* set if current block must be flushed */ + int flush = 0; /* set if current block must be flushed */ int match_available = 0; /* set if previous match exists */ register unsigned match_length = MIN_MATCH-1; /* length of best match */ @@ -721,6 +787,7 @@ off_t deflate() */ lookahead -= prev_length-1; prev_length -= 2; + RSYNC_ROLL(strstart, prev_length+1); do { strstart++; INSERT_STRING(strstart, hash_head); @@ -733,24 +800,40 @@ off_t deflate() match_available = 0; match_length = MIN_MATCH-1; strstart++; - if (flush) FLUSH_BLOCK(0), block_start = strstart; + if (rsync && strstart > rsync_chunk_end) { + rsync_chunk_end = 0xFFFFFFFFUL; + flush = 2; + } + if (flush) FLUSH_BLOCK(0), block_start = strstart; } else if (match_available) { /* If there was no match at the previous position, output a * single literal. If there was a match but the current match * is longer, truncate the previous match to a single literal. */ Tracevv((stderr,"%c",window[strstart-1])); - if (ct_tally (0, window[strstart-1])) { - FLUSH_BLOCK(0), block_start = strstart; + flush = ct_tally (0, window[strstart-1]); + if (rsync && strstart > rsync_chunk_end) { + rsync_chunk_end = 0xFFFFFFFFUL; + flush = 2; } + if (flush) FLUSH_BLOCK(0), block_start = strstart; + RSYNC_ROLL(strstart, 1); strstart++; lookahead--; } else { /* There is no previous match to compare with, wait for * the next step to decide. */ + if (rsync && strstart > rsync_chunk_end) { + /* Reset huffman tree */ + rsync_chunk_end = 0xFFFFFFFFUL; + flush = 2; + FLUSH_BLOCK(0), block_start = strstart; + } + match_available = 1; + RSYNC_ROLL(strstart, 1); strstart++; lookahead--; }