X-Git-Url: https://wimlib.net/git/?a=blobdiff_plain;f=src%2Flzx-compress.c;h=bc50a858bc9b62cf26d9d1b03d7b822591a1eb33;hb=bcf3ccecc8071729f34fe5b393fab64fc02e3d47;hp=fbe255688f5004cda4e2d5d477464caeae0152af;hpb=dbe9b4a4b81485ba2a52a307a08adc048ea22bdd;p=wimlib diff --git a/src/lzx-compress.c b/src/lzx-compress.c index fbe25568..bc50a858 100644 --- a/src/lzx-compress.c +++ b/src/lzx-compress.c @@ -1,5 +1,7 @@ /* * lzx-compress.c + * + * A compressor that produces output compatible with the LZX compression format. */ /* @@ -107,46 +109,16 @@ * that position at previous positions in the window. With LZX, the minimum * match length is 2 and the maximum match length is 257. The only restriction * on offsets is that LZX does not allow the last 2 bytes of the window to match - * the the beginning of the window. - * - * Depending on how good a compression ratio we want (see the "Match-choosing" - * section), we may want to find: (a) all matches, or (b) just the longest - * match, or (c) just some "promising" matches that we are able to find quickly, - * or (d) just the longest match that we're able to find quickly. Below we - * introduce the match-finding methods that the code currently uses or has - * previously used: - * - * - Hash chains. Maintain a table that maps hash codes, computed from - * fixed-length byte sequences, to linked lists containing previous window - * positions. To search for matches, compute the hash for the current - * position in the window and search the appropriate hash chain. When - * advancing to the next position, prepend the current position to the - * appropriate hash list. This is a good approach for producing matches with - * stategy (d) and is useful for fast compression. Therefore, we provide an - * option to use this method for LZX compression. See lz_hash.c for the - * implementation. - * - * - Binary trees. Similar to hash chains, but each hash bucket contains a - * binary tree of previous window positions rather than a linked list. This - * is a good approach for producing matches with stategy (c) and is useful for - * achieving a good compression ratio. Therefore, we provide an option to use - * this method; see lz_bt.c for the implementation. - * - * - Suffix arrays. This code previously used this method to produce matches - * with stategy (c), but I've dropped it because it was slower than the binary - * trees approach, used more memory, and did not improve the compression ratio - * enough to compensate. Download wimlib v1.6.2 if you want the code. - * However, the suffix array method was basically as follows. Build the - * suffix array for the entire window. The suffix array contains each - * possible window position, sorted by the lexicographic order of the strings - * that begin at those positions. Find the matches at a given position by - * searching the suffix array outwards, in both directions, from the suffix - * array slot for that position. This produces the longest matches first, but - * "matches" that actually occur at later positions in the window must be - * skipped. To do this skipping, use an auxiliary array with dynamically - * constructed linked lists. Also, use the inverse suffix array to quickly - * find the suffix array slot for a given position without doing a binary - * search. + * the beginning of the window. + * + * There are a number of algorithms that can be used for this, including hash + * chains, binary trees, and suffix arrays. Binary trees generally work well + * for LZX compression since it uses medium-size windows (2^15 to 2^21 bytes). + * However, when compressing in a fast mode where many positions are skipped + * (not searched for matches), hash chains are faster. + * + * Since the match-finders are not specific to LZX, I will not explain them in + * detail here. Instead, see lz_hash_chains.c and lz_binary_trees.c. * * ---------------------------------------------------------------------------- * @@ -216,46 +188,31 @@ * for example. Therefore, for fast compression we combine lazy parsing with * the hash chain max-finder. For normal/high compression we combine * near-optimal parsing with the binary tree match-finder. - * - * Anyway, if you've read through this comment, you hopefully should have a - * better idea of why things are done in a certain way in this LZX compressor, - * as well as in other compressors for LZ77-based formats (including third-party - * ones). In my opinion, the phrase "compression algorithm" is often mis-used - * in place of "compression format", since there can be many different - * algorithms that all generate compressed data in the same format. The - * challenge is to design an algorithm that is efficient but still gives a good - * compression ratio. */ #ifdef HAVE_CONFIG_H # include "config.h" #endif -#include "wimlib.h" #include "wimlib/compressor_ops.h" #include "wimlib/compress_common.h" #include "wimlib/endianness.h" #include "wimlib/error.h" -#include "wimlib/lz.h" -#include "wimlib/lz_hash.h" -#include "wimlib/lz_bt.h" +#include "wimlib/lz_mf.h" +#include "wimlib/lz_repsearch.h" #include "wimlib/lzx.h" #include "wimlib/util.h" #include -#ifdef ENABLE_LZX_DEBUG -# include "wimlib/decompress_common.h" -#endif - -#define LZX_OPTIM_ARRAY_SIZE 4096 +#define LZX_OPTIM_ARRAY_LENGTH 4096 #define LZX_DIV_BLOCK_SIZE 32768 #define LZX_CACHE_PER_POS 8 +#define LZX_MAX_MATCHES_PER_POS (LZX_MAX_MATCH_LEN - LZX_MIN_MATCH_LEN + 1) + #define LZX_CACHE_LEN (LZX_DIV_BLOCK_SIZE * (LZX_CACHE_PER_POS + 1)) -#define LZX_CACHE_SIZE (LZX_CACHE_LEN * sizeof(struct lz_match)) -#define LZX_MAX_MATCHES_PER_POS (LZX_MAX_MATCH_LEN - LZX_MIN_MATCH_LEN + 1) /* Codewords for the LZX main, length, and aligned offset Huffman codes */ struct lzx_codewords { @@ -279,7 +236,7 @@ struct lzx_lens { * * If a codeword has zero frequency, it must still be assigned some nonzero cost * --- generally a high cost, since even if it gets used in the next iteration, - * it probably will not be used very times. */ + * it probably will not be used very many times. */ struct lzx_costs { u8 main[LZX_MAINCODE_MAX_NUM_SYMBOLS]; u8 len[LZX_LENCODE_NUM_SYMBOLS]; @@ -294,9 +251,9 @@ struct lzx_codes { /* Tables for tallying symbol frequencies in the three LZX alphabets */ struct lzx_freqs { - input_idx_t main[LZX_MAINCODE_MAX_NUM_SYMBOLS]; - input_idx_t len[LZX_LENCODE_NUM_SYMBOLS]; - input_idx_t aligned[LZX_ALIGNEDCODE_NUM_SYMBOLS]; + u32 main[LZX_MAINCODE_MAX_NUM_SYMBOLS]; + u32 len[LZX_LENCODE_NUM_SYMBOLS]; + u32 aligned[LZX_ALIGNEDCODE_NUM_SYMBOLS]; }; /* LZX intermediate match/literal format */ @@ -325,27 +282,35 @@ struct lzx_block_spec { int block_type; /* 0-based position in the window at which this block starts. */ - input_idx_t window_pos; + u32 window_pos; /* The number of bytes of uncompressed data this block represents. */ - input_idx_t block_size; + u32 block_size; /* The match/literal sequence for this block. */ struct lzx_item *chosen_items; /* The length of the @chosen_items sequence. */ - input_idx_t num_chosen_items; + u32 num_chosen_items; /* Huffman codes for this block. */ struct lzx_codes codes; }; +struct lzx_compressor; + +struct lzx_compressor_params { + struct lz_match (*choose_item_func)(struct lzx_compressor *); + enum lz_mf_algo mf_algo; + u32 num_optim_passes; + u32 min_match_length; + u32 nice_match_length; + u32 max_search_depth; +}; + /* State of the LZX compressor. */ struct lzx_compressor { - /* The parameters that were used to create the compressor. */ - struct wimlib_lzx_compressor_params params; - /* The buffer of data to be compressed. * * 0xe8 byte preprocessing is done directly on the data here before @@ -354,22 +319,34 @@ struct lzx_compressor { * Note that this compressor does *not* use a real sliding window!!!! * It's not needed in the WIM format, since every chunk is compressed * independently. This is by design, to allow random access to the - * chunks. - * - * We reserve a few extra bytes to potentially allow reading off the end - * of the array in the match-finding code for optimization purposes - * (currently only needed for the hash chain match-finder). */ - u8 *window; + * chunks. */ + u8 *cur_window; /* Number of bytes of data to be compressed, which is the number of - * bytes of data in @window that are actually valid. */ - input_idx_t window_size; + * bytes of data in @cur_window that are actually valid. */ + u32 cur_window_size; - /* Allocated size of the @window. */ - input_idx_t max_window_size; + /* Allocated size of @cur_window. */ + u32 max_window_size; - /* Number of symbols in the main alphabet (depends on the - * @max_window_size since it determines the maximum allowed offset). */ + /* log2 order of the LZX window size for LZ match offset encoding + * purposes. Will be >= LZX_MIN_WINDOW_ORDER and <= + * LZX_MAX_WINDOW_ORDER. + * + * Note: 1 << @window_order is normally equal to @max_window_size, but + * it will be greater than @max_window_size in the event that the + * compressor was created with a non-power-of-2 block size. (See + * lzx_get_window_order().) */ + unsigned window_order; + + /* Compression parameters. */ + struct lzx_compressor_params params; + + unsigned (*get_matches_func)(struct lzx_compressor *, const struct lz_match **); + void (*skip_bytes_func)(struct lzx_compressor *, unsigned n); + + /* Number of symbols in the main alphabet (depends on the @window_order + * since it determines the maximum allowed offset). */ unsigned num_main_syms; /* The current match offset LRU queue. */ @@ -396,30 +373,28 @@ struct lzx_compressor { /* The current cost model. */ struct lzx_costs costs; - /* Fast algorithm only: Array of hash table links. */ - input_idx_t *prev_tab; - - /* Slow algorithm only: Binary tree match-finder. */ - struct lz_bt mf; + /* Lempel-Ziv match-finder. */ + struct lz_mf *mf; /* Position in window of next match to return. */ - input_idx_t match_window_pos; + u32 match_window_pos; /* The end-of-block position. We can't allow any matches to span this * position. */ - input_idx_t match_window_end; + u32 match_window_end; - /* Matches found by the match-finder are cached in the following array - * to achieve a slight speedup when the same matches are needed on + /* When doing more than one match-choosing pass over the data, matches + * found by the match-finder are cached in the following array to + * achieve a slight speedup when the same matches are needed on * subsequent passes. This is suboptimal because different matches may * be preferred with different cost models, but seems to be a worthwhile * speedup. */ struct lz_match *cached_matches; struct lz_match *cache_ptr; - bool matches_cached; struct lz_match *cache_limit; - /* Match-chooser state. + /* Match-chooser state, used when doing near-optimal parsing. + * * When matches have been chosen, optimum_cur_idx is set to the position * in the window of the next match/literal to return and optimum_end_idx * is set to the position in the window at the end of the last @@ -427,6 +402,9 @@ struct lzx_compressor { struct lzx_mc_pos_data *optimum; unsigned optimum_cur_idx; unsigned optimum_end_idx; + + /* Previous match, used when doing lazy parsing. */ + struct lz_match prev_match; }; /* @@ -453,30 +431,165 @@ struct lzx_mc_pos_data { /* Position of the start of the match or literal that * was taken to get to this position in the approximate * minimum-cost parse. */ - input_idx_t link; + u32 link; /* Offset (as in an LZ (length, offset) pair) of the * match or literal that was taken to get to this * position in the approximate minimum-cost parse. */ - input_idx_t match_offset; + u32 match_offset; } prev; struct { /* Position at which the match or literal starting at * this position ends in the minimum-cost parse. */ - input_idx_t link; + u32 link; /* Offset (as in an LZ (length, offset) pair) of the * match or literal starting at this position in the * approximate minimum-cost parse. */ - input_idx_t match_offset; + u32 match_offset; } next; }; /* Adaptive state that exists after an approximate minimum-cost path to - * reach this position is taken. */ + * reach this position is taken. + * + * Note: we update this whenever we update the pending minimum-cost + * path. This is in contrast to LZMA, which also has an optimal parser + * that maintains a repeat offset queue per position, but will only + * compute the queue once that position is actually reached in the + * parse, meaning that matches are being considered *starting* at that + * position. However, the two methods seem to have approximately the + * same performance if appropriate optimizations are used. Intuitively + * the LZMA method seems faster, but it actually suffers from 1-2 extra + * hard-to-predict branches at each position. Probably it works better + * for LZMA than LZX because LZMA has a larger adaptive state than LZX, + * and the LZMA encoder considers more possibilities. */ struct lzx_lru_queue queue; }; + +/* + * Structure to keep track of the current state of sending bits to the + * compressed output buffer. + * + * The LZX bitstream is encoded as a sequence of 16-bit coding units. + */ +struct lzx_output_bitstream { + + /* Bits that haven't yet been written to the output buffer. */ + u32 bitbuf; + + /* Number of bits currently held in @bitbuf. */ + u32 bitcount; + + /* Pointer to the start of the output buffer. */ + le16 *start; + + /* Pointer to the position in the output buffer at which the next coding + * unit should be written. */ + le16 *next; + + /* Pointer past the end of the output buffer. */ + le16 *end; +}; + +/* + * Initialize the output bitstream. + * + * @os + * The output bitstream structure to initialize. + * @buffer + * The buffer being written to. + * @size + * Size of @buffer, in bytes. + */ +static void +lzx_init_output(struct lzx_output_bitstream *os, void *buffer, u32 size) +{ + os->bitbuf = 0; + os->bitcount = 0; + os->start = buffer; + os->next = os->start; + os->end = os->start + size / sizeof(le16); +} + +/* + * Write some bits to the output bitstream. + * + * The bits are given by the low-order @num_bits bits of @bits. Higher-order + * bits in @bits cannot be set. At most 17 bits can be written at once. + * + * @max_bits is a compile-time constant that specifies the maximum number of + * bits that can ever be written at the call site. Currently, it is used to + * optimize away the conditional code for writing a second 16-bit coding unit + * when writing fewer than 17 bits. + * + * If the output buffer space is exhausted, then the bits will be ignored, and + * lzx_flush_output() will return 0 when it gets called. + */ +static _always_inline_attribute void +lzx_write_varbits(struct lzx_output_bitstream *os, + const u32 bits, const unsigned int num_bits, + const unsigned int max_num_bits) +{ + /* This code is optimized for LZX, which never needs to write more than + * 17 bits at once. */ + LZX_ASSERT(num_bits <= 17); + LZX_ASSERT(num_bits <= max_num_bits); + LZX_ASSERT(os->bitcount <= 15); + + /* Add the bits to the bit buffer variable. @bitcount will be at most + * 15, so there will be just enough space for the maximum possible + * @num_bits of 17. */ + os->bitcount += num_bits; + os->bitbuf = (os->bitbuf << num_bits) | bits; + + /* Check whether any coding units need to be written. */ + if (os->bitcount >= 16) { + + os->bitcount -= 16; + + /* Write a coding unit, unless it would overflow the buffer. */ + if (os->next != os->end) + *os->next++ = cpu_to_le16(os->bitbuf >> os->bitcount); + + /* If writing 17 bits, a second coding unit might need to be + * written. But because 'max_num_bits' is a compile-time + * constant, the compiler will optimize away this code at most + * call sites. */ + if (max_num_bits == 17 && os->bitcount == 16) { + if (os->next != os->end) + *os->next++ = cpu_to_le16(os->bitbuf); + os->bitcount = 0; + } + } +} + +/* Use when @num_bits is a compile-time constant. Otherwise use + * lzx_write_varbits(). */ +static _always_inline_attribute void +lzx_write_bits(struct lzx_output_bitstream *os, + const u32 bits, const unsigned int num_bits) +{ + lzx_write_varbits(os, bits, num_bits, num_bits); +} + +/* + * Flush the last coding unit to the output buffer if needed. Return the total + * number of bytes written to the output buffer, or 0 if an overflow occurred. + */ +static u32 +lzx_flush_output(struct lzx_output_bitstream *os) +{ + if (os->next == os->end) + return 0; + + if (os->bitcount != 0) + *os->next++ = cpu_to_le16(os->bitbuf << (16 - os->bitcount)); + + return (const u8 *)os->next - (const u8 *)os->start; +} + /* Returns the LZX position slot that corresponds to a given match offset, * taking into account the recent offset queue and updating it if the offset is * found in it. */ @@ -544,42 +657,35 @@ lzx_make_huffman_codes(const struct lzx_freqs *freqs, /* * Output a precomputed LZX match. * - * @out: + * @os: * The bitstream to which to write the match. - * @block_type: - * The type of the LZX block (LZX_BLOCKTYPE_ALIGNED or - * LZX_BLOCKTYPE_VERBATIM) + * @ones_if_aligned + * A mask of all ones if the block is of type LZX_BLOCKTYPE_ALIGNED, + * otherwise 0. * @match: - * The match, as a (length, offset) pair. + * The match data. * @codes: * Pointer to a structure that contains the codewords for the main, length, * and aligned offset Huffman codes for the current LZX compressed block. */ static void -lzx_write_match(struct output_bitstream *out, int block_type, +lzx_write_match(struct lzx_output_bitstream *os, unsigned ones_if_aligned, struct lzx_item match, const struct lzx_codes *codes) { - /* low 8 bits are the match length minus 2 */ unsigned match_len_minus_2 = match.data & 0xff; - /* Next 17 bits are the position footer */ - unsigned position_footer = (match.data >> 8) & 0x1ffff; /* 17 bits */ - /* Next 6 bits are the position slot. */ - unsigned position_slot = (match.data >> 25) & 0x3f; /* 6 bits */ + u32 position_footer = (match.data >> 8) & 0x1ffff; + unsigned position_slot = (match.data >> 25) & 0x3f; unsigned len_header; unsigned len_footer; unsigned main_symbol; unsigned num_extra_bits; - unsigned verbatim_bits; - unsigned aligned_bits; /* If the match length is less than MIN_MATCH_LEN (= 2) + - * NUM_PRIMARY_LENS (= 7), the length header contains - * the match length minus MIN_MATCH_LEN, and there is no - * length footer. + * NUM_PRIMARY_LENS (= 7), the length header contains the match length + * minus MIN_MATCH_LEN, and there is no length footer. * - * Otherwise, the length header contains - * NUM_PRIMARY_LENS, and the length footer contains - * the match length minus NUM_PRIMARY_LENS minus + * Otherwise, the length header contains NUM_PRIMARY_LENS, and the + * length footer contains the match length minus NUM_PRIMARY_LENS minus * MIN_MATCH_LEN. */ if (match_len_minus_2 < LZX_NUM_PRIMARY_LENS) { len_header = match_len_minus_2; @@ -597,184 +703,138 @@ lzx_write_match(struct output_bitstream *out, int block_type, main_symbol = ((position_slot << 3) | len_header) + LZX_NUM_CHARS; /* Output main symbol. */ - bitstream_put_bits(out, codes->codewords.main[main_symbol], - codes->lens.main[main_symbol]); + lzx_write_varbits(os, codes->codewords.main[main_symbol], + codes->lens.main[main_symbol], + LZX_MAX_MAIN_CODEWORD_LEN); /* If there is a length footer, output it using the * length Huffman code. */ - if (len_header == LZX_NUM_PRIMARY_LENS) - bitstream_put_bits(out, codes->codewords.len[len_footer], - codes->lens.len[len_footer]); + if (len_header == LZX_NUM_PRIMARY_LENS) { + lzx_write_varbits(os, codes->codewords.len[len_footer], + codes->lens.len[len_footer], + LZX_MAX_LEN_CODEWORD_LEN); + } + + /* Output the position footer. */ num_extra_bits = lzx_get_num_extra_bits(position_slot); - /* For aligned offset blocks with at least 3 extra bits, output the - * verbatim bits literally, then the aligned bits encoded using the - * aligned offset code. Otherwise, only the verbatim bits need to be - * output. */ - if ((block_type == LZX_BLOCKTYPE_ALIGNED) && (num_extra_bits >= 3)) { + if ((num_extra_bits & ones_if_aligned) >= 3) { - verbatim_bits = position_footer >> 3; - bitstream_put_bits(out, verbatim_bits, - num_extra_bits - 3); + /* Aligned offset blocks: The low 3 bits of the position footer + * are Huffman-encoded using the aligned offset code. The + * remaining bits are output literally. */ - aligned_bits = (position_footer & 7); - bitstream_put_bits(out, - codes->codewords.aligned[aligned_bits], - codes->lens.aligned[aligned_bits]); + lzx_write_varbits(os, + position_footer >> 3, num_extra_bits - 3, 14); + + lzx_write_varbits(os, + codes->codewords.aligned[position_footer & 7], + codes->lens.aligned[position_footer & 7], + LZX_MAX_ALIGNED_CODEWORD_LEN); } else { - /* verbatim bits is the same as the position - * footer, in this case. */ - bitstream_put_bits(out, position_footer, num_extra_bits); + /* Verbatim blocks, or fewer than 3 extra bits: All position + * footer bits are output literally. */ + lzx_write_varbits(os, position_footer, num_extra_bits, 17); } } /* Output an LZX literal (encoded with the main Huffman code). */ static void -lzx_write_literal(struct output_bitstream *out, u8 literal, +lzx_write_literal(struct lzx_output_bitstream *os, unsigned literal, const struct lzx_codes *codes) { - bitstream_put_bits(out, - codes->codewords.main[literal], - codes->lens.main[literal]); + lzx_write_varbits(os, codes->codewords.main[literal], + codes->lens.main[literal], LZX_MAX_MAIN_CODEWORD_LEN); } static unsigned -lzx_build_precode(const u8 lens[restrict], - const u8 prev_lens[restrict], - const unsigned num_syms, - input_idx_t precode_freqs[restrict LZX_PRECODE_NUM_SYMBOLS], - u8 output_syms[restrict num_syms], - u8 precode_lens[restrict LZX_PRECODE_NUM_SYMBOLS], - u32 precode_codewords[restrict LZX_PRECODE_NUM_SYMBOLS], - unsigned *num_additional_bits_ret) +lzx_compute_precode_items(const u8 lens[restrict], + const u8 prev_lens[restrict], + const unsigned num_lens, + u32 precode_freqs[restrict], + unsigned precode_items[restrict]) { - memset(precode_freqs, 0, - LZX_PRECODE_NUM_SYMBOLS * sizeof(precode_freqs[0])); - - /* Since the code word lengths use a form of RLE encoding, the goal here - * is to find each run of identical lengths when going through them in - * symbol order (including runs of length 1). For each run, as many - * lengths are encoded using RLE as possible, and the rest are output - * literally. - * - * output_syms[] will be filled in with the length symbols that will be - * output, including RLE codes, not yet encoded using the precode. - * - * cur_run_len keeps track of how many code word lengths are in the - * current run of identical lengths. */ - unsigned output_syms_idx = 0; - unsigned cur_run_len = 1; - unsigned num_additional_bits = 0; - for (unsigned i = 1; i <= num_syms; i++) { - - if (i != num_syms && lens[i] == lens[i - 1]) { - /* Still in a run--- keep going. */ - cur_run_len++; - continue; - } + unsigned *itemptr; + unsigned run_start; + unsigned run_end; + unsigned extra_bits; + int delta; + u8 len; + + itemptr = precode_items; + run_start = 0; + do { + /* Find the next run of codeword lengths. */ - /* Run ended! Check if it is a run of zeroes or a run of - * nonzeroes. */ + /* len = the length being repeated */ + len = lens[run_start]; - /* The symbol that was repeated in the run--- not to be confused - * with the length *of* the run (cur_run_len) */ - unsigned len_in_run = lens[i - 1]; + run_end = run_start + 1; - if (len_in_run == 0) { - /* A run of 0's. Encode it in as few length - * codes as we can. */ + /* Fast case for a single length. */ + if (likely(run_end == num_lens || len != lens[run_end])) { + delta = prev_lens[run_start] - len; + if (delta < 0) + delta += 17; + precode_freqs[delta]++; + *itemptr++ = delta; + run_start++; + continue; + } - /* The magic length 18 indicates a run of 20 + n zeroes, - * where n is an uncompressed literal 5-bit integer that - * follows the magic length. */ - while (cur_run_len >= 20) { - unsigned additional_bits; + /* Extend the run. */ + do { + run_end++; + } while (run_end != num_lens && len == lens[run_end]); + + if (len == 0) { + /* Run of zeroes. */ - additional_bits = min(cur_run_len - 20, 0x1f); - num_additional_bits += 5; + /* Symbol 18: RLE 20 to 51 zeroes at a time. */ + while ((run_end - run_start) >= 20) { + extra_bits = min((run_end - run_start) - 20, 0x1f); precode_freqs[18]++; - output_syms[output_syms_idx++] = 18; - output_syms[output_syms_idx++] = additional_bits; - cur_run_len -= 20 + additional_bits; + *itemptr++ = 18 | (extra_bits << 5); + run_start += 20 + extra_bits; } - /* The magic length 17 indicates a run of 4 + n zeroes, - * where n is an uncompressed literal 4-bit integer that - * follows the magic length. */ - while (cur_run_len >= 4) { - unsigned additional_bits; - - additional_bits = min(cur_run_len - 4, 0xf); - num_additional_bits += 4; + /* Symbol 17: RLE 4 to 19 zeroes at a time. */ + if ((run_end - run_start) >= 4) { + extra_bits = min((run_end - run_start) - 4, 0xf); precode_freqs[17]++; - output_syms[output_syms_idx++] = 17; - output_syms[output_syms_idx++] = additional_bits; - cur_run_len -= 4 + additional_bits; + *itemptr++ = 17 | (extra_bits << 5); + run_start += 4 + extra_bits; } - } else { /* A run of nonzero lengths. */ - /* The magic length 19 indicates a run of 4 + n - * nonzeroes, where n is a literal bit that follows the - * magic length, and where the value of the lengths in - * the run is given by an extra length symbol, encoded - * with the precode, that follows the literal bit. - * - * The extra length symbol is encoded as a difference - * from the length of the codeword for the first symbol - * in the run in the previous code. - * */ - while (cur_run_len >= 4) { - unsigned additional_bits; - signed char delta; - - additional_bits = (cur_run_len > 4); - num_additional_bits += 1; - delta = (signed char)prev_lens[i - cur_run_len] - - (signed char)len_in_run; + /* Symbol 19: RLE 4 to 5 of any length at a time. */ + while ((run_end - run_start) >= 4) { + extra_bits = (run_end - run_start) > 4; + delta = prev_lens[run_start] - len; if (delta < 0) delta += 17; precode_freqs[19]++; - precode_freqs[(unsigned char)delta]++; - output_syms[output_syms_idx++] = 19; - output_syms[output_syms_idx++] = additional_bits; - output_syms[output_syms_idx++] = delta; - cur_run_len -= 4 + additional_bits; + precode_freqs[delta]++; + *itemptr++ = 19 | (extra_bits << 5) | (delta << 6); + run_start += 4 + extra_bits; } } - /* Any remaining lengths in the run are outputted without RLE, - * as a difference from the length of that codeword in the - * previous code. */ - while (cur_run_len > 0) { - signed char delta; - - delta = (signed char)prev_lens[i - cur_run_len] - - (signed char)len_in_run; + /* Output any remaining lengths without RLE. */ + while (run_start != run_end) { + delta = prev_lens[run_start] - len; if (delta < 0) delta += 17; - - precode_freqs[(unsigned char)delta]++; - output_syms[output_syms_idx++] = delta; - cur_run_len--; + precode_freqs[delta]++; + *itemptr++ = delta; + run_start++; } + } while (run_start != num_lens); - cur_run_len = 1; - } - - /* Build the precode from the frequencies of the length symbols. */ - - make_canonical_huffman_code(LZX_PRECODE_NUM_SYMBOLS, - LZX_MAX_PRE_CODEWORD_LEN, - precode_freqs, precode_lens, - precode_codewords); - - *num_additional_bits_ret = num_additional_bits; - - return output_syms_idx; + return itemptr - precode_items; } /* @@ -796,68 +856,71 @@ lzx_build_precode(const u8 lens[restrict], * as deltas from the codeword lengths of the corresponding code in the previous * block. * - * @out: + * @os: * Bitstream to which to write the compressed Huffman code. * @lens: * The codeword lengths, indexed by symbol, in the Huffman code. * @prev_lens: * The codeword lengths, indexed by symbol, in the corresponding Huffman * code in the previous block, or all zeroes if this is the first block. - * @num_syms: + * @num_lens: * The number of symbols in the Huffman code. */ static void -lzx_write_compressed_code(struct output_bitstream *out, +lzx_write_compressed_code(struct lzx_output_bitstream *os, const u8 lens[restrict], const u8 prev_lens[restrict], - unsigned num_syms) + unsigned num_lens) { - input_idx_t precode_freqs[LZX_PRECODE_NUM_SYMBOLS]; - u8 output_syms[num_syms]; + u32 precode_freqs[LZX_PRECODE_NUM_SYMBOLS]; u8 precode_lens[LZX_PRECODE_NUM_SYMBOLS]; u32 precode_codewords[LZX_PRECODE_NUM_SYMBOLS]; + unsigned precode_items[num_lens]; + unsigned num_precode_items; + unsigned precode_item; + unsigned precode_sym; unsigned i; - unsigned num_output_syms; - u8 precode_sym; - unsigned dummy; - - num_output_syms = lzx_build_precode(lens, - prev_lens, - num_syms, - precode_freqs, - output_syms, - precode_lens, - precode_codewords, - &dummy); - - /* Write the lengths of the precode codes to the output. */ + + for (i = 0; i < LZX_PRECODE_NUM_SYMBOLS; i++) + precode_freqs[i] = 0; + + /* Compute the "items" (RLE / literal tokens and extra bits) with which + * the codeword lengths in the larger code will be output. */ + num_precode_items = lzx_compute_precode_items(lens, + prev_lens, + num_lens, + precode_freqs, + precode_items); + + /* Build the precode. */ + make_canonical_huffman_code(LZX_PRECODE_NUM_SYMBOLS, + LZX_MAX_PRE_CODEWORD_LEN, + precode_freqs, precode_lens, + precode_codewords); + + /* Output the lengths of the codewords in the precode. */ for (i = 0; i < LZX_PRECODE_NUM_SYMBOLS; i++) - bitstream_put_bits(out, precode_lens[i], - LZX_PRECODE_ELEMENT_SIZE); - - /* Write the length symbols, encoded with the precode, to the output. */ - - for (i = 0; i < num_output_syms; ) { - precode_sym = output_syms[i++]; - - bitstream_put_bits(out, precode_codewords[precode_sym], - precode_lens[precode_sym]); - switch (precode_sym) { - case 17: - bitstream_put_bits(out, output_syms[i++], 4); - break; - case 18: - bitstream_put_bits(out, output_syms[i++], 5); - break; - case 19: - bitstream_put_bits(out, output_syms[i++], 1); - bitstream_put_bits(out, - precode_codewords[output_syms[i]], - precode_lens[output_syms[i]]); - i++; - break; - default: - break; + lzx_write_bits(os, precode_lens[i], LZX_PRECODE_ELEMENT_SIZE); + + /* Output the encoded lengths of the codewords in the larger code. */ + for (i = 0; i < num_precode_items; i++) { + precode_item = precode_items[i]; + precode_sym = precode_item & 0x1F; + lzx_write_varbits(os, precode_codewords[precode_sym], + precode_lens[precode_sym], + LZX_MAX_PRE_CODEWORD_LEN); + if (precode_sym >= 17) { + if (precode_sym == 17) { + lzx_write_bits(os, precode_item >> 5, 4); + } else if (precode_sym == 18) { + lzx_write_bits(os, precode_item >> 5, 5); + } else { + lzx_write_bits(os, (precode_item >> 5) & 1, 1); + precode_sym = precode_item >> 6; + lzx_write_varbits(os, precode_codewords[precode_sym], + precode_lens[precode_sym], + LZX_MAX_PRE_CODEWORD_LEN); + } } } } @@ -867,97 +930,55 @@ lzx_write_compressed_code(struct output_bitstream *out, * compressed block to the output bitstream in the final compressed * representation. * - * @ostream + * @os * The output bitstream. * @block_type * The chosen type of the LZX compressed block (LZX_BLOCKTYPE_ALIGNED or * LZX_BLOCKTYPE_VERBATIM). - * @match_tab + * @items * The array of matches/literals to output. - * @match_count - * Number of matches/literals to output (length of @match_tab). + * @num_items + * Number of matches/literals to output (length of @items). * @codes * The main, length, and aligned offset Huffman codes for the current * LZX compressed block. */ static void -lzx_write_matches_and_literals(struct output_bitstream *ostream, - int block_type, - const struct lzx_item match_tab[], - unsigned match_count, - const struct lzx_codes *codes) +lzx_write_items(struct lzx_output_bitstream *os, int block_type, + const struct lzx_item items[], u32 num_items, + const struct lzx_codes *codes) { - for (unsigned i = 0; i < match_count; i++) { - struct lzx_item match = match_tab[i]; + unsigned ones_if_aligned = 0U - (block_type == LZX_BLOCKTYPE_ALIGNED); + for (u32 i = 0; i < num_items; i++) { /* The high bit of the 32-bit intermediate representation * indicates whether the item is an actual LZ-style match (1) or * a literal byte (0). */ - if (match.data & 0x80000000) - lzx_write_match(ostream, block_type, match, codes); + if (items[i].data & 0x80000000) + lzx_write_match(os, ones_if_aligned, items[i], codes); else - lzx_write_literal(ostream, match.data, codes); + lzx_write_literal(os, items[i].data, codes); } } -static void -lzx_assert_codes_valid(const struct lzx_codes * codes, unsigned num_main_syms) -{ -#ifdef ENABLE_LZX_DEBUG - unsigned i; - - for (i = 0; i < num_main_syms; i++) - LZX_ASSERT(codes->lens.main[i] <= LZX_MAX_MAIN_CODEWORD_LEN); - - for (i = 0; i < LZX_LENCODE_NUM_SYMBOLS; i++) - LZX_ASSERT(codes->lens.len[i] <= LZX_MAX_LEN_CODEWORD_LEN); - - for (i = 0; i < LZX_ALIGNEDCODE_NUM_SYMBOLS; i++) - LZX_ASSERT(codes->lens.aligned[i] <= LZX_MAX_ALIGNED_CODEWORD_LEN); - - const unsigned tablebits = 10; - u16 decode_table[(1 << tablebits) + - (2 * max(num_main_syms, LZX_LENCODE_NUM_SYMBOLS))] - _aligned_attribute(DECODE_TABLE_ALIGNMENT); - LZX_ASSERT(0 == make_huffman_decode_table(decode_table, - num_main_syms, - min(tablebits, LZX_MAINCODE_TABLEBITS), - codes->lens.main, - LZX_MAX_MAIN_CODEWORD_LEN)); - LZX_ASSERT(0 == make_huffman_decode_table(decode_table, - LZX_LENCODE_NUM_SYMBOLS, - min(tablebits, LZX_LENCODE_TABLEBITS), - codes->lens.len, - LZX_MAX_LEN_CODEWORD_LEN)); - LZX_ASSERT(0 == make_huffman_decode_table(decode_table, - LZX_ALIGNEDCODE_NUM_SYMBOLS, - min(tablebits, LZX_ALIGNEDCODE_TABLEBITS), - codes->lens.aligned, - LZX_MAX_ALIGNED_CODEWORD_LEN)); -#endif /* ENABLE_LZX_DEBUG */ -} - /* Write an LZX aligned offset or verbatim block to the output. */ static void lzx_write_compressed_block(int block_type, - unsigned block_size, - unsigned max_window_size, + u32 block_size, + unsigned window_order, unsigned num_main_syms, struct lzx_item * chosen_items, - unsigned num_chosen_items, + u32 num_chosen_items, const struct lzx_codes * codes, const struct lzx_codes * prev_codes, - struct output_bitstream * ostream) + struct lzx_output_bitstream * os) { - unsigned i; - LZX_ASSERT(block_type == LZX_BLOCKTYPE_ALIGNED || block_type == LZX_BLOCKTYPE_VERBATIM); - lzx_assert_codes_valid(codes, num_main_syms); /* The first three bits indicate the type of block and are one of the * LZX_BLOCKTYPE_* constants. */ - bitstream_put_bits(ostream, block_type, 3); + lzx_write_bits(os, block_type, 3); /* Output the block size. * @@ -975,82 +996,59 @@ lzx_write_compressed_block(int block_type, * because WIMs created with chunk size greater than 32768 can seemingly * only be opened by wimlib anyway. */ if (block_size == LZX_DEFAULT_BLOCK_SIZE) { - bitstream_put_bits(ostream, 1, 1); + lzx_write_bits(os, 1, 1); } else { - bitstream_put_bits(ostream, 0, 1); + lzx_write_bits(os, 0, 1); - if (max_window_size >= 65536) - bitstream_put_bits(ostream, block_size >> 16, 8); + if (window_order >= 16) + lzx_write_bits(os, block_size >> 16, 8); - bitstream_put_bits(ostream, block_size, 16); + lzx_write_bits(os, block_size & 0xFFFF, 16); } - /* Write out lengths of the main code. Note that the LZX specification - * incorrectly states that the aligned offset code comes after the - * length code, but in fact it is the very first code to be written - * (before the main code). */ - if (block_type == LZX_BLOCKTYPE_ALIGNED) - for (i = 0; i < LZX_ALIGNEDCODE_NUM_SYMBOLS; i++) - bitstream_put_bits(ostream, codes->lens.aligned[i], - LZX_ALIGNEDCODE_ELEMENT_SIZE); - - LZX_DEBUG("Writing main code..."); - - /* Write the precode and lengths for the first LZX_NUM_CHARS symbols in - * the main code, which are the codewords for literal bytes. */ - lzx_write_compressed_code(ostream, - codes->lens.main, + /* Output the aligned offset code. */ + if (block_type == LZX_BLOCKTYPE_ALIGNED) { + for (int i = 0; i < LZX_ALIGNEDCODE_NUM_SYMBOLS; i++) { + lzx_write_bits(os, codes->lens.aligned[i], + LZX_ALIGNEDCODE_ELEMENT_SIZE); + } + } + + /* Output the main code (two parts). */ + lzx_write_compressed_code(os, codes->lens.main, prev_codes->lens.main, LZX_NUM_CHARS); - - /* Write the precode and lengths for the rest of the main code, which - * are the codewords for match headers. */ - lzx_write_compressed_code(ostream, - codes->lens.main + LZX_NUM_CHARS, + lzx_write_compressed_code(os, codes->lens.main + LZX_NUM_CHARS, prev_codes->lens.main + LZX_NUM_CHARS, num_main_syms - LZX_NUM_CHARS); - LZX_DEBUG("Writing length code..."); - - /* Write the precode and lengths for the length code. */ - lzx_write_compressed_code(ostream, - codes->lens.len, + /* Output the length code. */ + lzx_write_compressed_code(os, codes->lens.len, prev_codes->lens.len, LZX_LENCODE_NUM_SYMBOLS); - LZX_DEBUG("Writing matches and literals..."); - - /* Write the actual matches and literals. */ - lzx_write_matches_and_literals(ostream, block_type, - chosen_items, num_chosen_items, - codes); - - LZX_DEBUG("Done writing block."); + /* Output the compressed matches and literals. */ + lzx_write_items(os, block_type, chosen_items, num_chosen_items, codes); } /* Write out the LZX blocks that were computed. */ static void -lzx_write_all_blocks(struct lzx_compressor *ctx, struct output_bitstream *ostream) +lzx_write_all_blocks(struct lzx_compressor *c, struct lzx_output_bitstream *os) { - const struct lzx_codes *prev_codes = &ctx->zero_codes; - for (unsigned i = 0; i < ctx->num_blocks; i++) { - const struct lzx_block_spec *spec = &ctx->block_specs[i]; - - LZX_DEBUG("Writing block %u/%u (type=%d, size=%u, num_chosen_items=%u)...", - i + 1, ctx->num_blocks, - spec->block_type, spec->block_size, - spec->num_chosen_items); + const struct lzx_codes *prev_codes = &c->zero_codes; + for (unsigned i = 0; i < c->num_blocks; i++) { + const struct lzx_block_spec *spec = &c->block_specs[i]; lzx_write_compressed_block(spec->block_type, spec->block_size, - ctx->max_window_size, - ctx->num_main_syms, + c->window_order, + c->num_main_syms, spec->chosen_items, spec->num_chosen_items, &spec->codes, prev_codes, - ostream); + os); prev_codes = &spec->codes; } @@ -1074,7 +1072,7 @@ lzx_tally_match(unsigned match_len, u32 match_offset, struct lzx_freqs *freqs, struct lzx_lru_queue *queue) { unsigned position_slot; - unsigned position_footer; + u32 position_footer; u32 len_header; unsigned main_symbol; unsigned len_footer; @@ -1086,7 +1084,7 @@ lzx_tally_match(unsigned match_len, u32 match_offset, * as part of the main symbol) and a position footer. */ position_slot = lzx_get_position_slot(match_offset, queue); position_footer = (match_offset + LZX_OFFSET_OFFSET) & - ((1U << lzx_get_num_extra_bits(position_slot)) - 1); + (((u32)1 << lzx_get_num_extra_bits(position_slot)) - 1); /* The match length shall be encoded as a length header (itself encoded * as part of the main symbol) and an optional length footer. */ @@ -1136,28 +1134,6 @@ lzx_tally_match(unsigned match_len, u32 match_offset, (adjusted_match_len); } -struct lzx_record_ctx { - struct lzx_freqs freqs; - struct lzx_lru_queue queue; - struct lzx_item *matches; -}; - -static void -lzx_record_match(unsigned len, unsigned offset, void *_ctx) -{ - struct lzx_record_ctx *ctx = _ctx; - - (ctx->matches++)->data = lzx_tally_match(len, offset, &ctx->freqs, &ctx->queue); -} - -static void -lzx_record_literal(u8 lit, void *_ctx) -{ - struct lzx_record_ctx *ctx = _ctx; - - (ctx->matches++)->data = lzx_tally_literal(lit, &ctx->freqs); -} - /* Returns the cost, in bits, to output a literal byte using the specified cost * model. */ static u32 @@ -1166,46 +1142,28 @@ lzx_literal_cost(u8 c, const struct lzx_costs * costs) return costs->main[c]; } -/* Given a (length, offset) pair that could be turned into a valid LZX match as - * well as costs for the codewords in the main, length, and aligned Huffman - * codes, return the approximate number of bits it will take to represent this - * match in the compressed output. Take into account the match offset LRU - * queue and also updates it. */ +/* Returns the cost, in bits, to output a repeat offset match of the specified + * length and position slot (repeat index) using the specified cost model. */ static u32 -lzx_match_cost(unsigned length, u32 offset, const struct lzx_costs *costs, - struct lzx_lru_queue *queue) +lzx_repmatch_cost(u32 len, unsigned position_slot, const struct lzx_costs *costs) { - unsigned position_slot; unsigned len_header, main_symbol; - unsigned num_extra_bits; u32 cost = 0; - position_slot = lzx_get_position_slot(offset, queue); - - len_header = min(length - LZX_MIN_MATCH_LEN, LZX_NUM_PRIMARY_LENS); + len_header = min(len - LZX_MIN_MATCH_LEN, LZX_NUM_PRIMARY_LENS); main_symbol = ((position_slot << 3) | len_header) + LZX_NUM_CHARS; /* Account for main symbol. */ cost += costs->main[main_symbol]; - /* Account for extra position information. */ - num_extra_bits = lzx_get_num_extra_bits(position_slot); - if (num_extra_bits >= 3) { - cost += num_extra_bits - 3; - cost += costs->aligned[(offset + LZX_OFFSET_OFFSET) & 7]; - } else { - cost += num_extra_bits; - } - /* Account for extra length information. */ if (len_header == LZX_NUM_PRIMARY_LENS) - cost += costs->len[length - LZX_MIN_MATCH_LEN - LZX_NUM_PRIMARY_LENS]; + cost += costs->len[len - LZX_MIN_MATCH_LEN - LZX_NUM_PRIMARY_LENS]; return cost; - } -/* Set the cost model @ctx->costs from the Huffman codeword lengths specified in +/* Set the cost model @c->costs from the Huffman codeword lengths specified in * @lens. * * The cost model and codeword lengths are almost the same thing, but the @@ -1215,67 +1173,35 @@ lzx_match_cost(unsigned length, u32 offset, const struct lzx_costs *costs, * length) to take into account the fact that uses of these symbols are expected * to be rare. */ static void -lzx_set_costs(struct lzx_compressor * ctx, const struct lzx_lens * lens) +lzx_set_costs(struct lzx_compressor *c, const struct lzx_lens * lens, + unsigned nostat) { unsigned i; - unsigned num_main_syms = ctx->num_main_syms; /* Main code */ - for (i = 0; i < num_main_syms; i++) { - ctx->costs.main[i] = lens->main[i]; - if (ctx->costs.main[i] == 0) - ctx->costs.main[i] = ctx->params.alg_params.slow.main_nostat_cost; - } + for (i = 0; i < c->num_main_syms; i++) + c->costs.main[i] = lens->main[i] ? lens->main[i] : nostat; /* Length code */ - for (i = 0; i < LZX_LENCODE_NUM_SYMBOLS; i++) { - ctx->costs.len[i] = lens->len[i]; - if (ctx->costs.len[i] == 0) - ctx->costs.len[i] = ctx->params.alg_params.slow.len_nostat_cost; - } + for (i = 0; i < LZX_LENCODE_NUM_SYMBOLS; i++) + c->costs.len[i] = lens->len[i] ? lens->len[i] : nostat; /* Aligned offset code */ - for (i = 0; i < LZX_ALIGNEDCODE_NUM_SYMBOLS; i++) { - ctx->costs.aligned[i] = lens->aligned[i]; - if (ctx->costs.aligned[i] == 0) - ctx->costs.aligned[i] = ctx->params.alg_params.slow.aligned_nostat_cost; - } + for (i = 0; i < LZX_ALIGNEDCODE_NUM_SYMBOLS; i++) + c->costs.aligned[i] = lens->aligned[i] ? lens->aligned[i] : nostat / 2; } -/* Retrieve a list of matches available at the next position in the input. - * - * A pointer to the matches array is written into @matches_ret, and the return - * value is the number of matches found. */ -static unsigned -lzx_get_matches(struct lzx_compressor *ctx, - const struct lz_match **matches_ret) +/* Don't allow matches to span the end of an LZX block. */ +static inline u32 +maybe_truncate_matches(struct lz_match matches[], u32 num_matches, + struct lzx_compressor *c) { - struct lz_match *cache_ptr; - struct lz_match *matches; - unsigned num_matches; - - LZX_ASSERT(ctx->match_window_pos < ctx->match_window_end); - - cache_ptr = ctx->cache_ptr; - matches = cache_ptr + 1; - if (likely(cache_ptr <= ctx->cache_limit)) { - if (ctx->matches_cached) { - num_matches = cache_ptr->len; - } else { - num_matches = lz_bt_get_matches(&ctx->mf, matches); - cache_ptr->len = num_matches; - } - } else { - num_matches = 0; - } - - /* Don't allow matches to span the end of an LZX block. */ - if (ctx->match_window_end < ctx->window_size && num_matches != 0) { - unsigned limit = ctx->match_window_end - ctx->match_window_pos; + if (c->match_window_end < c->cur_window_size && num_matches != 0) { + u32 limit = c->match_window_end - c->match_window_pos; if (limit >= LZX_MIN_MATCH_LEN) { - unsigned i = num_matches - 1; + u32 i = num_matches - 1; do { if (matches[i].len >= limit) { matches[i].len = limit; @@ -1289,58 +1215,196 @@ lzx_get_matches(struct lzx_compressor *ctx, } else { num_matches = 0; } - cache_ptr->len = num_matches; } + return num_matches; +} -#if 0 - fprintf(stderr, "Pos %u/%u: %u matches\n", - ctx->match_window_pos, ctx->window_size, num_matches); - for (unsigned i = 0; i < num_matches; i++) - fprintf(stderr, "\tLen %u Offset %u\n", matches[i].len, matches[i].offset); -#endif +static unsigned +lzx_get_matches_fillcache_singleblock(struct lzx_compressor *c, + const struct lz_match **matches_ret) +{ + struct lz_match *cache_ptr; + struct lz_match *matches; + unsigned num_matches; -#ifdef ENABLE_LZX_DEBUG - for (unsigned i = 0; i < num_matches; i++) { - LZX_ASSERT(matches[i].len >= LZX_MIN_MATCH_LEN); - LZX_ASSERT(matches[i].len <= LZX_MAX_MATCH_LEN); - LZX_ASSERT(matches[i].len <= ctx->match_window_end - ctx->match_window_pos); - LZX_ASSERT(matches[i].offset > 0); - LZX_ASSERT(matches[i].offset <= ctx->match_window_pos); - LZX_ASSERT(!memcmp(&ctx->window[ctx->match_window_pos], - &ctx->window[ctx->match_window_pos - matches[i].offset], - matches[i].len)); - if (i) { - LZX_ASSERT(matches[i].len > matches[i - 1].len); - LZX_ASSERT(matches[i].offset > matches[i - 1].offset); - } + cache_ptr = c->cache_ptr; + matches = cache_ptr + 1; + if (likely(cache_ptr <= c->cache_limit)) { + num_matches = lz_mf_get_matches(c->mf, matches); + cache_ptr->len = num_matches; + c->cache_ptr = matches + num_matches; + } else { + num_matches = 0; } -#endif - ctx->match_window_pos++; - ctx->cache_ptr = matches + num_matches; + c->match_window_pos++; *matches_ret = matches; return num_matches; } -static void -lzx_skip_bytes(struct lzx_compressor *ctx, unsigned n) +static unsigned +lzx_get_matches_fillcache_multiblock(struct lzx_compressor *c, + const struct lz_match **matches_ret) { struct lz_match *cache_ptr; + struct lz_match *matches; + unsigned num_matches; - LZX_ASSERT(n <= ctx->match_window_end - ctx->match_window_pos); + cache_ptr = c->cache_ptr; + matches = cache_ptr + 1; + if (likely(cache_ptr <= c->cache_limit)) { + num_matches = lz_mf_get_matches(c->mf, matches); + num_matches = maybe_truncate_matches(matches, num_matches, c); + cache_ptr->len = num_matches; + c->cache_ptr = matches + num_matches; + } else { + num_matches = 0; + } + c->match_window_pos++; + *matches_ret = matches; + return num_matches; +} - cache_ptr = ctx->cache_ptr; - ctx->match_window_pos += n; - if (ctx->matches_cached) { - while (n-- && cache_ptr <= ctx->cache_limit) - cache_ptr += 1 + cache_ptr->len; +static unsigned +lzx_get_matches_usecache(struct lzx_compressor *c, + const struct lz_match **matches_ret) +{ + struct lz_match *cache_ptr; + struct lz_match *matches; + unsigned num_matches; + + cache_ptr = c->cache_ptr; + matches = cache_ptr + 1; + if (cache_ptr <= c->cache_limit) { + num_matches = cache_ptr->len; + c->cache_ptr = matches + num_matches; } else { - lz_bt_skip_positions(&ctx->mf, n); - while (n-- && cache_ptr <= ctx->cache_limit) { + num_matches = 0; + } + c->match_window_pos++; + *matches_ret = matches; + return num_matches; +} + +static unsigned +lzx_get_matches_usecache_nocheck(struct lzx_compressor *c, + const struct lz_match **matches_ret) +{ + struct lz_match *cache_ptr; + struct lz_match *matches; + unsigned num_matches; + + cache_ptr = c->cache_ptr; + matches = cache_ptr + 1; + num_matches = cache_ptr->len; + c->cache_ptr = matches + num_matches; + c->match_window_pos++; + *matches_ret = matches; + return num_matches; +} + +static unsigned +lzx_get_matches_nocache_singleblock(struct lzx_compressor *c, + const struct lz_match **matches_ret) +{ + struct lz_match *matches; + unsigned num_matches; + + matches = c->cache_ptr; + num_matches = lz_mf_get_matches(c->mf, matches); + c->match_window_pos++; + *matches_ret = matches; + return num_matches; +} + +static unsigned +lzx_get_matches_nocache_multiblock(struct lzx_compressor *c, + const struct lz_match **matches_ret) +{ + struct lz_match *matches; + unsigned num_matches; + + matches = c->cache_ptr; + num_matches = lz_mf_get_matches(c->mf, matches); + num_matches = maybe_truncate_matches(matches, num_matches, c); + c->match_window_pos++; + *matches_ret = matches; + return num_matches; +} + +/* + * Find matches at the next position in the window. + * + * Returns the number of matches found and sets *matches_ret to point to the + * matches array. The matches will be sorted by strictly increasing length and + * offset. + */ +static inline unsigned +lzx_get_matches(struct lzx_compressor *c, + const struct lz_match **matches_ret) +{ + return (*c->get_matches_func)(c, matches_ret); +} + +static void +lzx_skip_bytes_fillcache(struct lzx_compressor *c, unsigned n) +{ + struct lz_match *cache_ptr; + + cache_ptr = c->cache_ptr; + c->match_window_pos += n; + lz_mf_skip_positions(c->mf, n); + if (cache_ptr <= c->cache_limit) { + do { cache_ptr->len = 0; cache_ptr += 1; - } + } while (--n && cache_ptr <= c->cache_limit); + } + c->cache_ptr = cache_ptr; +} + +static void +lzx_skip_bytes_usecache(struct lzx_compressor *c, unsigned n) +{ + struct lz_match *cache_ptr; + + cache_ptr = c->cache_ptr; + c->match_window_pos += n; + if (cache_ptr <= c->cache_limit) { + do { + cache_ptr += 1 + cache_ptr->len; + } while (--n && cache_ptr <= c->cache_limit); } - ctx->cache_ptr = cache_ptr; + c->cache_ptr = cache_ptr; +} + +static void +lzx_skip_bytes_usecache_nocheck(struct lzx_compressor *c, unsigned n) +{ + struct lz_match *cache_ptr; + + cache_ptr = c->cache_ptr; + c->match_window_pos += n; + do { + cache_ptr += 1 + cache_ptr->len; + } while (--n); + c->cache_ptr = cache_ptr; +} + +static void +lzx_skip_bytes_nocache(struct lzx_compressor *c, unsigned n) +{ + c->match_window_pos += n; + lz_mf_skip_positions(c->mf, n); +} + +/* + * Skip the specified number of positions in the window (don't search for + * matches at them). + */ +static inline void +lzx_skip_bytes(struct lzx_compressor *c, unsigned n) +{ + return (*c->skip_bytes_func)(c, n); } /* @@ -1350,39 +1414,56 @@ lzx_skip_bytes(struct lzx_compressor *ctx, unsigned n) * Returns the first match in the list. */ static struct lz_match -lzx_match_chooser_reverse_list(struct lzx_compressor *ctx, unsigned cur_pos) +lzx_match_chooser_reverse_list(struct lzx_compressor *c, unsigned cur_pos) { unsigned prev_link, saved_prev_link; unsigned prev_match_offset, saved_prev_match_offset; - ctx->optimum_end_idx = cur_pos; + c->optimum_end_idx = cur_pos; - saved_prev_link = ctx->optimum[cur_pos].prev.link; - saved_prev_match_offset = ctx->optimum[cur_pos].prev.match_offset; + saved_prev_link = c->optimum[cur_pos].prev.link; + saved_prev_match_offset = c->optimum[cur_pos].prev.match_offset; do { prev_link = saved_prev_link; prev_match_offset = saved_prev_match_offset; - saved_prev_link = ctx->optimum[prev_link].prev.link; - saved_prev_match_offset = ctx->optimum[prev_link].prev.match_offset; + saved_prev_link = c->optimum[prev_link].prev.link; + saved_prev_match_offset = c->optimum[prev_link].prev.match_offset; - ctx->optimum[prev_link].next.link = cur_pos; - ctx->optimum[prev_link].next.match_offset = prev_match_offset; + c->optimum[prev_link].next.link = cur_pos; + c->optimum[prev_link].next.match_offset = prev_match_offset; cur_pos = prev_link; } while (cur_pos != 0); - ctx->optimum_cur_idx = ctx->optimum[0].next.link; + c->optimum_cur_idx = c->optimum[0].next.link; return (struct lz_match) - { .len = ctx->optimum_cur_idx, - .offset = ctx->optimum[0].next.match_offset, + { .len = c->optimum_cur_idx, + .offset = c->optimum[0].next.match_offset, }; } /* - * lzx_get_near_optimal_match() - + * Find the longest repeat offset match. + * + * If no match of at least LZX_MIN_MATCH_LEN bytes is found, then return 0. + * + * If a match of at least LZX_MIN_MATCH_LEN bytes is found, then return its + * length and set *slot_ret to the index of its offset in @queue. + */ +static inline u32 +lzx_repsearch(const u8 * const strptr, const u32 bytes_remaining, + const struct lzx_lru_queue *queue, unsigned *slot_ret) +{ + BUILD_BUG_ON(LZX_MIN_MATCH_LEN != 2); + return lz_repsearch(strptr, bytes_remaining, LZX_MAX_MATCH_LEN, + queue->R, LZX_NUM_RECENT_OFFSETS, slot_ret); +} + +/* + * lzx_choose_near_optimal_item() - * * Choose an approximately optimal match or literal to use at the next position * in the string, or "window", being LZ-encoded. @@ -1444,81 +1525,72 @@ lzx_match_chooser_reverse_list(struct lzx_compressor *ctx, unsigned cur_pos) * chosen. For literals, the length is 0 or 1 and the offset is meaningless. */ static struct lz_match -lzx_get_near_optimal_match(struct lzx_compressor *ctx) +lzx_choose_near_optimal_item(struct lzx_compressor *c) { unsigned num_matches; const struct lz_match *matches; struct lz_match match; - unsigned longest_len; - unsigned longest_rep_len; - u32 longest_rep_offset; + u32 longest_len; + u32 longest_rep_len; + unsigned longest_rep_slot; unsigned cur_pos; unsigned end_pos; + struct lzx_mc_pos_data *optimum = c->optimum; - if (ctx->optimum_cur_idx != ctx->optimum_end_idx) { + if (c->optimum_cur_idx != c->optimum_end_idx) { /* Case 2: Return the next match/literal already found. */ - match.len = ctx->optimum[ctx->optimum_cur_idx].next.link - - ctx->optimum_cur_idx; - match.offset = ctx->optimum[ctx->optimum_cur_idx].next.match_offset; + match.len = optimum[c->optimum_cur_idx].next.link - + c->optimum_cur_idx; + match.offset = optimum[c->optimum_cur_idx].next.match_offset; - ctx->optimum_cur_idx = ctx->optimum[ctx->optimum_cur_idx].next.link; + c->optimum_cur_idx = optimum[c->optimum_cur_idx].next.link; return match; } /* Case 1: Compute a new list of matches/literals to return. */ - ctx->optimum_cur_idx = 0; - ctx->optimum_end_idx = 0; - - /* Search for matches at recent offsets. Only keep the one with the - * longest match length. */ - longest_rep_len = LZX_MIN_MATCH_LEN - 1; - if (ctx->match_window_pos >= 1) { - unsigned limit = min(LZX_MAX_MATCH_LEN, - ctx->match_window_end - ctx->match_window_pos); - for (int i = 0; i < LZX_NUM_RECENT_OFFSETS; i++) { - u32 offset = ctx->queue.R[i]; - const u8 *strptr = &ctx->window[ctx->match_window_pos]; - const u8 *matchptr = strptr - offset; - unsigned len = 0; - while (len < limit && strptr[len] == matchptr[len]) - len++; - if (len > longest_rep_len) { - longest_rep_len = len; - longest_rep_offset = offset; - } - } + c->optimum_cur_idx = 0; + c->optimum_end_idx = 0; + + /* Search for matches at repeat offsets. As a heuristic, we only keep + * the one with the longest match length. */ + if (likely(c->match_window_pos >= 1)) { + longest_rep_len = lzx_repsearch(&c->cur_window[c->match_window_pos], + c->match_window_end - c->match_window_pos, + &c->queue, + &longest_rep_slot); + } else { + longest_rep_len = 0; } - /* If there's a long match with a recent offset, take it. */ - if (longest_rep_len >= ctx->params.alg_params.slow.nice_match_length) { - lzx_skip_bytes(ctx, longest_rep_len); + /* If there's a long match with a repeat offset, choose it immediately. */ + if (longest_rep_len >= c->params.nice_match_length) { + lzx_skip_bytes(c, longest_rep_len); return (struct lz_match) { .len = longest_rep_len, - .offset = longest_rep_offset, + .offset = c->queue.R[longest_rep_slot], }; } - /* Search other matches. */ - num_matches = lzx_get_matches(ctx, &matches); + /* Find other matches. */ + num_matches = lzx_get_matches(c, &matches); - /* If there's a long match, take it. */ + /* If there's a long match, choose it immediately. */ if (num_matches) { longest_len = matches[num_matches - 1].len; - if (longest_len >= ctx->params.alg_params.slow.nice_match_length) { - lzx_skip_bytes(ctx, longest_len - 1); + if (longest_len >= c->params.nice_match_length) { + lzx_skip_bytes(c, longest_len - 1); return matches[num_matches - 1]; } } else { longest_len = 1; } - /* Calculate the cost to reach the next position by coding a literal. - */ - ctx->optimum[1].queue = ctx->queue; - ctx->optimum[1].cost = lzx_literal_cost(ctx->window[ctx->match_window_pos - 1], - &ctx->costs); - ctx->optimum[1].prev.link = 0; + /* Calculate the cost to reach the next position by coding a literal. */ + optimum[1].queue = c->queue; + optimum[1].cost = lzx_literal_cost(c->cur_window[c->match_window_pos - 1], + &c->costs); + optimum[1].prev.link = 0; /* Calculate the cost to reach any position up to and including that * reached by the longest match. @@ -1539,54 +1611,62 @@ lzx_get_near_optimal_match(struct lzx_compressor *ctx) unsigned num_extra_bits; offset = matches[i].offset; - queue = ctx->queue; + queue = c->queue; position_cost = 0; position_slot = lzx_get_position_slot(offset, &queue); num_extra_bits = lzx_get_num_extra_bits(position_slot); if (num_extra_bits >= 3) { position_cost += num_extra_bits - 3; - position_cost += ctx->costs.aligned[(offset + LZX_OFFSET_OFFSET) & 7]; + position_cost += c->costs.aligned[(offset + LZX_OFFSET_OFFSET) & 7]; } else { position_cost += num_extra_bits; } do { + u32 cost; unsigned len_header; unsigned main_symbol; - u32 cost; cost = position_cost; - len_header = min(len - LZX_MIN_MATCH_LEN, LZX_NUM_PRIMARY_LENS); + if (len - LZX_MIN_MATCH_LEN < LZX_NUM_PRIMARY_LENS) { + len_header = len - LZX_MIN_MATCH_LEN; + } else { + len_header = LZX_NUM_PRIMARY_LENS; + cost += c->costs.len[len - LZX_MIN_MATCH_LEN - LZX_NUM_PRIMARY_LENS]; + } + main_symbol = ((position_slot << 3) | len_header) + LZX_NUM_CHARS; - cost += ctx->costs.main[main_symbol]; - if (len_header == LZX_NUM_PRIMARY_LENS) - cost += ctx->costs.len[len - LZX_MIN_MATCH_LEN - LZX_NUM_PRIMARY_LENS]; - - ctx->optimum[len].queue = queue; - ctx->optimum[len].prev.link = 0; - ctx->optimum[len].prev.match_offset = offset; - ctx->optimum[len].cost = cost; + cost += c->costs.main[main_symbol]; + + optimum[len].queue = queue; + optimum[len].prev.link = 0; + optimum[len].prev.match_offset = offset; + optimum[len].cost = cost; } while (++len <= matches[i].len); } end_pos = longest_len; - if (longest_rep_len >= LZX_MIN_MATCH_LEN) { - struct lzx_lru_queue queue; + if (longest_rep_len) { + + LZX_ASSERT(longest_rep_len >= LZX_MIN_MATCH_LEN); + u32 cost; while (end_pos < longest_rep_len) - ctx->optimum[++end_pos].cost = MC_INFINITE_COST; - - queue = ctx->queue; - cost = lzx_match_cost(longest_rep_len, longest_rep_offset, - &ctx->costs, &queue); - if (cost <= ctx->optimum[longest_rep_len].cost) { - ctx->optimum[longest_rep_len].queue = queue; - ctx->optimum[longest_rep_len].prev.link = 0; - ctx->optimum[longest_rep_len].prev.match_offset = longest_rep_offset; - ctx->optimum[longest_rep_len].cost = cost; + optimum[++end_pos].cost = MC_INFINITE_COST; + + cost = lzx_repmatch_cost(longest_rep_len, longest_rep_slot, + &c->costs); + if (cost <= optimum[longest_rep_len].cost) { + optimum[longest_rep_len].queue = c->queue; + swap(optimum[longest_rep_len].queue.R[0], + optimum[longest_rep_len].queue.R[longest_rep_slot]); + optimum[longest_rep_len].prev.link = 0; + optimum[longest_rep_len].prev.match_offset = + optimum[longest_rep_len].queue.R[0]; + optimum[longest_rep_len].cost = cost; } } @@ -1594,17 +1674,16 @@ lzx_get_near_optimal_match(struct lzx_compressor *ctx) * position. The algorithm may find multiple paths to reach each * position; only the lowest-cost path is saved. * - * The progress of the parse is tracked in the @ctx->optimum array, which - * for each position contains the minimum cost to reach that position, - * the index of the start of the match/literal taken to reach that - * position through the minimum-cost path, the offset of the match taken - * (not relevant for literals), and the adaptive state that will exist - * at that position after the minimum-cost path is taken. The @cur_pos + * The progress of the parse is tracked in the @optimum array, which for + * each position contains the minimum cost to reach that position, the + * index of the start of the match/literal taken to reach that position + * through the minimum-cost path, the offset of the match taken (not + * relevant for literals), and the adaptive state that will exist at + * that position after the minimum-cost path is taken. The @cur_pos * variable stores the position at which the algorithm is currently * considering coding choices, and the @end_pos variable stores the * greatest position at which the costs of coding choices have been - * saved. (Actually, the algorithm guarantees that all positions up to - * and including @end_pos are reachable by at least one path.) + * saved. * * The loop terminates when any one of the following conditions occurs: * @@ -1620,7 +1699,7 @@ lzx_get_near_optimal_match(struct lzx_compressor *ctx) * match/literal list. * * 3. Failing either of the above in a degenerate case, the loop - * terminates when space in the @ctx->optimum array is exhausted. + * terminates when space in the @optimum array is exhausted. * This terminates the algorithm and forces it to start returning * matches/literals even though they may not be globally optimal. * @@ -1639,64 +1718,55 @@ lzx_get_near_optimal_match(struct lzx_compressor *ctx) cur_pos++; /* Check termination conditions (2) and (3) noted above. */ - if (cur_pos == end_pos || cur_pos == LZX_OPTIM_ARRAY_SIZE) - return lzx_match_chooser_reverse_list(ctx, cur_pos); - - /* Search for matches at recent offsets. */ - longest_rep_len = LZX_MIN_MATCH_LEN - 1; - unsigned limit = min(LZX_MAX_MATCH_LEN, - ctx->match_window_end - ctx->match_window_pos); - for (int i = 0; i < LZX_NUM_RECENT_OFFSETS; i++) { - u32 offset = ctx->optimum[cur_pos].queue.R[i]; - const u8 *strptr = &ctx->window[ctx->match_window_pos]; - const u8 *matchptr = strptr - offset; - unsigned len = 0; - while (len < limit && strptr[len] == matchptr[len]) - len++; - if (len > longest_rep_len) { - longest_rep_len = len; - longest_rep_offset = offset; - } - } + if (cur_pos == end_pos || cur_pos == LZX_OPTIM_ARRAY_LENGTH) + return lzx_match_chooser_reverse_list(c, cur_pos); - /* If we found a long match at a recent offset, choose it + /* Search for matches at repeat offsets. Again, as a heuristic + * we only keep the longest one. */ + longest_rep_len = lzx_repsearch(&c->cur_window[c->match_window_pos], + c->match_window_end - c->match_window_pos, + &optimum[cur_pos].queue, + &longest_rep_slot); + + /* If we found a long match at a repeat offset, choose it * immediately. */ - if (longest_rep_len >= ctx->params.alg_params.slow.nice_match_length) { + if (longest_rep_len >= c->params.nice_match_length) { /* Build the list of matches to return and get * the first one. */ - match = lzx_match_chooser_reverse_list(ctx, cur_pos); + match = lzx_match_chooser_reverse_list(c, cur_pos); /* Append the long match to the end of the list. */ - ctx->optimum[cur_pos].next.match_offset = longest_rep_offset; - ctx->optimum[cur_pos].next.link = cur_pos + longest_rep_len; - ctx->optimum_end_idx = cur_pos + longest_rep_len; + optimum[cur_pos].next.match_offset = + optimum[cur_pos].queue.R[longest_rep_slot]; + optimum[cur_pos].next.link = cur_pos + longest_rep_len; + c->optimum_end_idx = cur_pos + longest_rep_len; /* Skip over the remaining bytes of the long match. */ - lzx_skip_bytes(ctx, longest_rep_len); + lzx_skip_bytes(c, longest_rep_len); /* Return first match in the list. */ return match; } - /* Search other matches. */ - num_matches = lzx_get_matches(ctx, &matches); + /* Find other matches. */ + num_matches = lzx_get_matches(c, &matches); - /* If there's a long match, take it. */ + /* If there's a long match, choose it immediately. */ if (num_matches) { longest_len = matches[num_matches - 1].len; - if (longest_len >= ctx->params.alg_params.slow.nice_match_length) { + if (longest_len >= c->params.nice_match_length) { /* Build the list of matches to return and get * the first one. */ - match = lzx_match_chooser_reverse_list(ctx, cur_pos); + match = lzx_match_chooser_reverse_list(c, cur_pos); /* Append the long match to the end of the list. */ - ctx->optimum[cur_pos].next.match_offset = + optimum[cur_pos].next.match_offset = matches[num_matches - 1].offset; - ctx->optimum[cur_pos].next.link = cur_pos + longest_len; - ctx->optimum_end_idx = cur_pos + longest_len; + optimum[cur_pos].next.link = cur_pos + longest_len; + c->optimum_end_idx = cur_pos + longest_len; /* Skip over the remaining bytes of the long match. */ - lzx_skip_bytes(ctx, longest_len - 1); + lzx_skip_bytes(c, longest_len - 1); /* Return first match in the list. */ return match; @@ -1705,17 +1775,19 @@ lzx_get_near_optimal_match(struct lzx_compressor *ctx) longest_len = 1; } + /* If we are reaching any positions for the first time, we need + * to initialize their costs to infinity. */ while (end_pos < cur_pos + longest_len) - ctx->optimum[++end_pos].cost = MC_INFINITE_COST; + optimum[++end_pos].cost = MC_INFINITE_COST; /* Consider coding a literal. */ - cost = ctx->optimum[cur_pos].cost + - lzx_literal_cost(ctx->window[ctx->match_window_pos - 1], - &ctx->costs); - if (cost < ctx->optimum[cur_pos + 1].cost) { - ctx->optimum[cur_pos + 1].queue = ctx->optimum[cur_pos].queue; - ctx->optimum[cur_pos + 1].cost = cost; - ctx->optimum[cur_pos + 1].prev.link = cur_pos; + cost = optimum[cur_pos].cost + + lzx_literal_cost(c->cur_window[c->match_window_pos - 1], + &c->costs); + if (cost < optimum[cur_pos + 1].cost) { + optimum[cur_pos + 1].queue = optimum[cur_pos].queue; + optimum[cur_pos + 1].cost = cost; + optimum[cur_pos + 1].prev.link = cur_pos; } /* Consider coding a match. @@ -1728,76 +1800,186 @@ lzx_get_near_optimal_match(struct lzx_compressor *ctx) * length. */ for (unsigned i = 0, len = 2; i < num_matches; i++) { u32 offset; - struct lzx_lru_queue queue; u32 position_cost; unsigned position_slot; unsigned num_extra_bits; offset = matches[i].offset; - queue = ctx->optimum[cur_pos].queue; - position_cost = ctx->optimum[cur_pos].cost; + position_cost = optimum[cur_pos].cost; + + /* Yet another optimization: instead of calling + * lzx_get_position_slot(), hand-inline the search of + * the repeat offset queue. Then we can omit the + * extra_bits calculation for repeat offset matches, and + * also only compute the updated queue if we actually do + * find a new lowest cost path. */ + for (position_slot = 0; position_slot < LZX_NUM_RECENT_OFFSETS; position_slot++) + if (offset == optimum[cur_pos].queue.R[position_slot]) + goto have_position_cost; + + position_slot = lzx_get_position_slot_raw(offset + LZX_OFFSET_OFFSET); - position_slot = lzx_get_position_slot(offset, &queue); num_extra_bits = lzx_get_num_extra_bits(position_slot); if (num_extra_bits >= 3) { position_cost += num_extra_bits - 3; - position_cost += ctx->costs.aligned[ + position_cost += c->costs.aligned[ (offset + LZX_OFFSET_OFFSET) & 7]; } else { position_cost += num_extra_bits; } + have_position_cost: + do { + u32 cost; unsigned len_header; unsigned main_symbol; - u32 cost; cost = position_cost; - len_header = min(len - LZX_MIN_MATCH_LEN, - LZX_NUM_PRIMARY_LENS); - main_symbol = ((position_slot << 3) | len_header) + - LZX_NUM_CHARS; - cost += ctx->costs.main[main_symbol]; - if (len_header == LZX_NUM_PRIMARY_LENS) { - cost += ctx->costs.len[len - + if (len - LZX_MIN_MATCH_LEN < LZX_NUM_PRIMARY_LENS) { + len_header = len - LZX_MIN_MATCH_LEN; + } else { + len_header = LZX_NUM_PRIMARY_LENS; + cost += c->costs.len[len - LZX_MIN_MATCH_LEN - LZX_NUM_PRIMARY_LENS]; } - if (cost < ctx->optimum[cur_pos + len].cost) { - ctx->optimum[cur_pos + len].queue = queue; - ctx->optimum[cur_pos + len].prev.link = cur_pos; - ctx->optimum[cur_pos + len].prev.match_offset = offset; - ctx->optimum[cur_pos + len].cost = cost; + + main_symbol = ((position_slot << 3) | len_header) + + LZX_NUM_CHARS; + cost += c->costs.main[main_symbol]; + + if (cost < optimum[cur_pos + len].cost) { + if (position_slot < LZX_NUM_RECENT_OFFSETS) { + optimum[cur_pos + len].queue = optimum[cur_pos].queue; + swap(optimum[cur_pos + len].queue.R[0], + optimum[cur_pos + len].queue.R[position_slot]); + } else { + optimum[cur_pos + len].queue.R[0] = offset; + optimum[cur_pos + len].queue.R[1] = optimum[cur_pos].queue.R[0]; + optimum[cur_pos + len].queue.R[2] = optimum[cur_pos].queue.R[1]; + } + optimum[cur_pos + len].prev.link = cur_pos; + optimum[cur_pos + len].prev.match_offset = offset; + optimum[cur_pos + len].cost = cost; } } while (++len <= matches[i].len); } - if (longest_rep_len >= LZX_MIN_MATCH_LEN) { - struct lzx_lru_queue queue; - - while (end_pos < cur_pos + longest_rep_len) - ctx->optimum[++end_pos].cost = MC_INFINITE_COST; + /* Consider coding a repeat offset match. + * + * As a heuristic, we only consider the longest length of the + * longest repeat offset match. This does not, however, + * necessarily mean that we will never consider any other repeat + * offsets, because above we detect repeat offset matches that + * were found by the regular match-finder. Therefore, this + * special handling of the longest repeat-offset match is only + * helpful for coding a repeat offset match that was *not* found + * by the match-finder, e.g. due to being obscured by a less + * distant match that is at least as long. + * + * Note: an alternative, used in LZMA, is to consider every + * length of every repeat offset match. This is a more thorough + * search, and it makes it unnecessary to detect repeat offset + * matches that were found by the regular match-finder. But by + * my tests, for LZX the LZMA method slows down the compressor + * by ~10% and doesn't actually help the compression ratio too + * much. + * + * Also tested a compromise approach: consider every 3rd length + * of the longest repeat offset match. Still didn't seem quite + * worth it, though. + */ + if (longest_rep_len) { - queue = ctx->optimum[cur_pos].queue; + LZX_ASSERT(longest_rep_len >= LZX_MIN_MATCH_LEN); - cost = ctx->optimum[cur_pos].cost + - lzx_match_cost(longest_rep_len, longest_rep_offset, - &ctx->costs, &queue); - if (cost <= ctx->optimum[cur_pos + longest_rep_len].cost) { - ctx->optimum[cur_pos + longest_rep_len].queue = - queue; - ctx->optimum[cur_pos + longest_rep_len].prev.link = + while (end_pos < cur_pos + longest_rep_len) + optimum[++end_pos].cost = MC_INFINITE_COST; + + cost = optimum[cur_pos].cost + + lzx_repmatch_cost(longest_rep_len, longest_rep_slot, + &c->costs); + if (cost <= optimum[cur_pos + longest_rep_len].cost) { + optimum[cur_pos + longest_rep_len].queue = + optimum[cur_pos].queue; + swap(optimum[cur_pos + longest_rep_len].queue.R[0], + optimum[cur_pos + longest_rep_len].queue.R[longest_rep_slot]); + optimum[cur_pos + longest_rep_len].prev.link = cur_pos; - ctx->optimum[cur_pos + longest_rep_len].prev.match_offset = - longest_rep_offset; - ctx->optimum[cur_pos + longest_rep_len].cost = + optimum[cur_pos + longest_rep_len].prev.match_offset = + optimum[cur_pos + longest_rep_len].queue.R[0]; + optimum[cur_pos + longest_rep_len].cost = cost; } } } } +static struct lz_match +lzx_choose_lazy_item(struct lzx_compressor *c) +{ + const struct lz_match *matches; + struct lz_match cur_match; + struct lz_match next_match; + u32 num_matches; + + if (c->prev_match.len) { + cur_match = c->prev_match; + c->prev_match.len = 0; + } else { + num_matches = lzx_get_matches(c, &matches); + if (num_matches == 0 || + (matches[num_matches - 1].len <= 3 && + (matches[num_matches - 1].len <= 2 || + matches[num_matches - 1].offset > 4096))) + { + return (struct lz_match) { }; + } + + cur_match = matches[num_matches - 1]; + } + + if (cur_match.len >= c->params.nice_match_length) { + lzx_skip_bytes(c, cur_match.len - 1); + return cur_match; + } + + num_matches = lzx_get_matches(c, &matches); + if (num_matches == 0 || + (matches[num_matches - 1].len <= 3 && + (matches[num_matches - 1].len <= 2 || + matches[num_matches - 1].offset > 4096))) + { + lzx_skip_bytes(c, cur_match.len - 2); + return cur_match; + } + + next_match = matches[num_matches - 1]; + + if (next_match.len <= cur_match.len) { + lzx_skip_bytes(c, cur_match.len - 2); + return cur_match; + } else { + c->prev_match = next_match; + return (struct lz_match) { }; + } +} + +/* + * Return the next match or literal to use, delegating to the currently selected + * match-choosing algorithm. + * + * If the length of the returned 'struct lz_match' is less than + * LZX_MIN_MATCH_LEN, then it is really a literal. + */ +static inline struct lz_match +lzx_choose_item(struct lzx_compressor *c) +{ + return (*c->params.choose_item_func)(c); +} + /* Set default symbol costs for the LZX Huffman codes. */ static void lzx_set_default_costs(struct lzx_costs * costs, unsigned num_main_syms) @@ -1847,532 +2029,365 @@ lzx_choose_verbatim_or_aligned(const struct lzx_freqs * freqs, return LZX_BLOCKTYPE_VERBATIM; } -/* Find a near-optimal sequence of matches/literals with which to output the - * specified LZX block, then set the block's type to that which has the minimum - * cost to output (either verbatim or aligned). */ +/* Find a sequence of matches/literals with which to output the specified LZX + * block, then set the block's type to that which has the minimum cost to output + * (either verbatim or aligned). */ static void -lzx_optimize_block(struct lzx_compressor *ctx, struct lzx_block_spec *spec, - unsigned num_passes) +lzx_choose_items_for_block(struct lzx_compressor *c, struct lzx_block_spec *spec) { - const struct lzx_lru_queue orig_queue = ctx->queue; - unsigned num_passes_remaining = num_passes; + const struct lzx_lru_queue orig_queue = c->queue; + u32 num_passes_remaining = c->params.num_optim_passes; struct lzx_freqs freqs; const u8 *window_ptr; const u8 *window_end; - struct lzx_item *next_chosen_match; + struct lzx_item *next_chosen_item; struct lz_match lz_match; struct lzx_item lzx_item; - LZX_ASSERT(num_passes >= 1); - LZX_ASSERT(lz_bt_get_position(&ctx->mf) == spec->window_pos); + LZX_ASSERT(num_passes_remaining >= 1); + LZX_ASSERT(lz_mf_get_position(c->mf) == spec->window_pos); - ctx->match_window_end = spec->window_pos + spec->block_size; - ctx->matches_cached = false; + c->match_window_end = spec->window_pos + spec->block_size; + + if (c->params.num_optim_passes > 1) { + if (spec->block_size == c->cur_window_size) + c->get_matches_func = lzx_get_matches_fillcache_singleblock; + else + c->get_matches_func = lzx_get_matches_fillcache_multiblock; + c->skip_bytes_func = lzx_skip_bytes_fillcache; + } else { + if (spec->block_size == c->cur_window_size) + c->get_matches_func = lzx_get_matches_nocache_singleblock; + else + c->get_matches_func = lzx_get_matches_nocache_multiblock; + c->skip_bytes_func = lzx_skip_bytes_nocache; + } /* The first optimal parsing pass is done using the cost model already - * set in ctx->costs. Each later pass is done using a cost model + * set in c->costs. Each later pass is done using a cost model * computed from the previous pass. * * To improve performance we only generate the array containing the * matches and literals in intermediate form on the final pass. */ while (--num_passes_remaining) { - ctx->match_window_pos = spec->window_pos; - ctx->cache_ptr = ctx->cached_matches; + c->match_window_pos = spec->window_pos; + c->cache_ptr = c->cached_matches; memset(&freqs, 0, sizeof(freqs)); - window_ptr = &ctx->window[spec->window_pos]; + window_ptr = &c->cur_window[spec->window_pos]; window_end = window_ptr + spec->block_size; while (window_ptr != window_end) { - lz_match = lzx_get_near_optimal_match(ctx); + lz_match = lzx_choose_item(c); LZX_ASSERT(!(lz_match.len == LZX_MIN_MATCH_LEN && - lz_match.offset == ctx->max_window_size - + lz_match.offset == c->max_window_size - LZX_MIN_MATCH_LEN)); if (lz_match.len >= LZX_MIN_MATCH_LEN) { lzx_tally_match(lz_match.len, lz_match.offset, - &freqs, &ctx->queue); + &freqs, &c->queue); window_ptr += lz_match.len; } else { lzx_tally_literal(*window_ptr, &freqs); window_ptr += 1; } } - lzx_make_huffman_codes(&freqs, &spec->codes, ctx->num_main_syms); - lzx_set_costs(ctx, &spec->codes.lens); - ctx->queue = orig_queue; - ctx->matches_cached = true; + lzx_make_huffman_codes(&freqs, &spec->codes, c->num_main_syms); + lzx_set_costs(c, &spec->codes.lens, 15); + c->queue = orig_queue; + if (c->cache_ptr <= c->cache_limit) { + c->get_matches_func = lzx_get_matches_usecache_nocheck; + c->skip_bytes_func = lzx_skip_bytes_usecache_nocheck; + } else { + c->get_matches_func = lzx_get_matches_usecache; + c->skip_bytes_func = lzx_skip_bytes_usecache; + } } - ctx->match_window_pos = spec->window_pos; - ctx->cache_ptr = ctx->cached_matches; + c->match_window_pos = spec->window_pos; + c->cache_ptr = c->cached_matches; memset(&freqs, 0, sizeof(freqs)); - window_ptr = &ctx->window[spec->window_pos]; + window_ptr = &c->cur_window[spec->window_pos]; window_end = window_ptr + spec->block_size; - spec->chosen_items = &ctx->chosen_items[spec->window_pos]; - next_chosen_match = spec->chosen_items; + spec->chosen_items = &c->chosen_items[spec->window_pos]; + next_chosen_item = spec->chosen_items; + unsigned unseen_cost = 9; while (window_ptr != window_end) { - lz_match = lzx_get_near_optimal_match(ctx); + + lz_match = lzx_choose_item(c); LZX_ASSERT(!(lz_match.len == LZX_MIN_MATCH_LEN && - lz_match.offset == ctx->max_window_size - + lz_match.offset == c->max_window_size - LZX_MIN_MATCH_LEN)); if (lz_match.len >= LZX_MIN_MATCH_LEN) { lzx_item.data = lzx_tally_match(lz_match.len, lz_match.offset, - &freqs, &ctx->queue); + &freqs, &c->queue); window_ptr += lz_match.len; } else { lzx_item.data = lzx_tally_literal(*window_ptr, &freqs); window_ptr += 1; } - *next_chosen_match++ = lzx_item; + *next_chosen_item++ = lzx_item; + + /* When doing one-pass "near-optimal" parsing, update the cost + * model occassionally. */ + if (unlikely((next_chosen_item - spec->chosen_items) % 2048 == 0) && + c->params.choose_item_func == lzx_choose_near_optimal_item && + c->params.num_optim_passes == 1) + { + lzx_make_huffman_codes(&freqs, &spec->codes, c->num_main_syms); + lzx_set_costs(c, &spec->codes.lens, unseen_cost); + if (unseen_cost < 15) + unseen_cost++; + } } - spec->num_chosen_items = next_chosen_match - spec->chosen_items; - lzx_make_huffman_codes(&freqs, &spec->codes, ctx->num_main_syms); + spec->num_chosen_items = next_chosen_item - spec->chosen_items; + lzx_make_huffman_codes(&freqs, &spec->codes, c->num_main_syms); spec->block_type = lzx_choose_verbatim_or_aligned(&freqs, &spec->codes); } /* Prepare the input window into one or more LZX blocks ready to be output. */ static void -lzx_prepare_blocks(struct lzx_compressor * ctx) +lzx_prepare_blocks(struct lzx_compressor *c) { /* Set up a default cost model. */ - lzx_set_default_costs(&ctx->costs, ctx->num_main_syms); + if (c->params.choose_item_func == lzx_choose_near_optimal_item) + lzx_set_default_costs(&c->costs, c->num_main_syms); /* Set up the block specifications. * TODO: The compression ratio could be slightly improved by performing * data-dependent block splitting instead of using fixed-size blocks. * Doing so well is a computationally hard problem, however. */ - ctx->num_blocks = DIV_ROUND_UP(ctx->window_size, LZX_DIV_BLOCK_SIZE); - for (unsigned i = 0; i < ctx->num_blocks; i++) { - unsigned pos = LZX_DIV_BLOCK_SIZE * i; - ctx->block_specs[i].window_pos = pos; - ctx->block_specs[i].block_size = min(ctx->window_size - pos, - LZX_DIV_BLOCK_SIZE); + c->num_blocks = DIV_ROUND_UP(c->cur_window_size, LZX_DIV_BLOCK_SIZE); + for (unsigned i = 0; i < c->num_blocks; i++) { + u32 pos = LZX_DIV_BLOCK_SIZE * i; + c->block_specs[i].window_pos = pos; + c->block_specs[i].block_size = min(c->cur_window_size - pos, + LZX_DIV_BLOCK_SIZE); } /* Load the window into the match-finder. */ - lz_bt_load_window(&ctx->mf, ctx->window, ctx->window_size); + lz_mf_load_window(c->mf, c->cur_window, c->cur_window_size); /* Determine sequence of matches/literals to output for each block. */ - lzx_lru_queue_init(&ctx->queue); - ctx->optimum_cur_idx = 0; - ctx->optimum_end_idx = 0; - for (unsigned i = 0; i < ctx->num_blocks; i++) { - lzx_optimize_block(ctx, &ctx->block_specs[i], - ctx->params.alg_params.slow.num_optim_passes); - } + lzx_lru_queue_init(&c->queue); + c->optimum_cur_idx = 0; + c->optimum_end_idx = 0; + c->prev_match.len = 0; + for (unsigned i = 0; i < c->num_blocks; i++) + lzx_choose_items_for_block(c, &c->block_specs[i]); } -/* - * This is the fast version of lzx_prepare_blocks(). This version "quickly" - * prepares a single compressed block containing the entire input. See the - * description of the "Fast algorithm" at the beginning of this file for more - * information. - * - * Input --- the preprocessed data: - * - * ctx->window[] - * ctx->window_size - * - * Output --- the block specification and the corresponding match/literal data: - * - * ctx->block_specs[] - * ctx->num_blocks - * ctx->chosen_items[] - */ static void -lzx_prepare_block_fast(struct lzx_compressor * ctx) +lzx_build_params(unsigned int compression_level, + u32 max_window_size, + struct lzx_compressor_params *lzx_params) { - struct lzx_record_ctx record_ctx; - struct lzx_block_spec *spec; - - /* Parameters to hash chain LZ match finder - * (lazy with 1 match lookahead) */ - static const struct lz_params lzx_lz_params = { - /* Although LZX_MIN_MATCH_LEN == 2, length 2 matches typically - * aren't worth choosing when using greedy or lazy parsing. */ - .min_match = 3, - .max_match = LZX_MAX_MATCH_LEN, - .max_offset = LZX_MAX_WINDOW_SIZE, - .good_match = LZX_MAX_MATCH_LEN, - .nice_match = LZX_MAX_MATCH_LEN, - .max_chain_len = LZX_MAX_MATCH_LEN, - .max_lazy_match = LZX_MAX_MATCH_LEN, - .too_far = 4096, - }; - - /* Initialize symbol frequencies and match offset LRU queue. */ - memset(&record_ctx.freqs, 0, sizeof(struct lzx_freqs)); - lzx_lru_queue_init(&record_ctx.queue); - record_ctx.matches = ctx->chosen_items; - - /* Determine series of matches/literals to output. */ - lz_analyze_block(ctx->window, - ctx->window_size, - lzx_record_match, - lzx_record_literal, - &record_ctx, - &lzx_lz_params, - ctx->prev_tab); - - /* Set up block specification. */ - spec = &ctx->block_specs[0]; - spec->block_type = LZX_BLOCKTYPE_ALIGNED; - spec->window_pos = 0; - spec->block_size = ctx->window_size; - spec->num_chosen_items = (record_ctx.matches - ctx->chosen_items); - spec->chosen_items = ctx->chosen_items; - lzx_make_huffman_codes(&record_ctx.freqs, &spec->codes, - ctx->num_main_syms); - ctx->num_blocks = 1; + if (compression_level < 25) { + lzx_params->choose_item_func = lzx_choose_lazy_item; + lzx_params->num_optim_passes = 1; + if (max_window_size <= 262144) + lzx_params->mf_algo = LZ_MF_HASH_CHAINS; + else + lzx_params->mf_algo = LZ_MF_BINARY_TREES; + lzx_params->min_match_length = 3; + lzx_params->nice_match_length = 25 + compression_level * 2; + lzx_params->max_search_depth = 25 + compression_level; + } else { + lzx_params->choose_item_func = lzx_choose_near_optimal_item; + lzx_params->num_optim_passes = compression_level / 20; + if (max_window_size <= 32768 && lzx_params->num_optim_passes == 1) + lzx_params->mf_algo = LZ_MF_HASH_CHAINS; + else + lzx_params->mf_algo = LZ_MF_BINARY_TREES; + lzx_params->min_match_length = (compression_level >= 45) ? 2 : 3; + lzx_params->nice_match_length = min(((u64)compression_level * 32) / 50, + LZX_MAX_MATCH_LEN); + lzx_params->max_search_depth = min(((u64)compression_level * 50) / 50, + LZX_MAX_MATCH_LEN); + } } -static size_t -lzx_compress(const void *uncompressed_data, size_t uncompressed_size, - void *compressed_data, size_t compressed_size_avail, void *_ctx) +static void +lzx_build_mf_params(const struct lzx_compressor_params *lzx_params, + u32 max_window_size, struct lz_mf_params *mf_params) { - struct lzx_compressor *ctx = _ctx; - struct output_bitstream ostream; - size_t compressed_size; - - if (uncompressed_size < 100) { - LZX_DEBUG("Too small to bother compressing."); - return 0; - } - - if (uncompressed_size > ctx->max_window_size) { - LZX_DEBUG("Can't compress %zu bytes using window of %u bytes!", - uncompressed_size, ctx->max_window_size); - return 0; - } - - LZX_DEBUG("Attempting to compress %zu bytes...", - uncompressed_size); - - /* The input data must be preprocessed. To avoid changing the original - * input, copy it to a temporary buffer. */ - memcpy(ctx->window, uncompressed_data, uncompressed_size); - ctx->window_size = uncompressed_size; - - /* This line is unnecessary; it just avoids inconsequential accesses of - * uninitialized memory that would show up in memory-checking tools such - * as valgrind. */ - memset(&ctx->window[ctx->window_size], 0, 12); - - LZX_DEBUG("Preprocessing data..."); - - /* Before doing any actual compression, do the call instruction (0xe8 - * byte) translation on the uncompressed data. */ - lzx_do_e8_preprocessing(ctx->window, ctx->window_size); - - LZX_DEBUG("Preparing blocks..."); - - /* Prepare the compressed data. */ - if (ctx->params.algorithm == WIMLIB_LZX_ALGORITHM_FAST) - lzx_prepare_block_fast(ctx); - else - lzx_prepare_blocks(ctx); + memset(mf_params, 0, sizeof(*mf_params)); + + mf_params->algorithm = lzx_params->mf_algo; + mf_params->max_window_size = max_window_size; + mf_params->min_match_len = lzx_params->min_match_length; + mf_params->max_match_len = LZX_MAX_MATCH_LEN; + mf_params->max_search_depth = lzx_params->max_search_depth; + mf_params->nice_match_len = lzx_params->nice_match_length; +} - LZX_DEBUG("Writing compressed blocks..."); +static void +lzx_free_compressor(void *_c); - /* Generate the compressed data. */ - init_output_bitstream(&ostream, compressed_data, compressed_size_avail); - lzx_write_all_blocks(ctx, &ostream); +static u64 +lzx_get_needed_memory(size_t max_block_size, unsigned int compression_level) +{ + struct lzx_compressor_params params; + u64 size = 0; + unsigned window_order; + u32 max_window_size; - LZX_DEBUG("Flushing bitstream..."); - compressed_size = flush_output_bitstream(&ostream); - if (compressed_size == ~(input_idx_t)0) { - LZX_DEBUG("Data did not compress to %zu bytes or less!", - compressed_size_avail); + window_order = lzx_get_window_order(max_block_size); + if (window_order == 0) return 0; - } + max_window_size = max_block_size; - LZX_DEBUG("Done: compressed %zu => %zu bytes.", - uncompressed_size, compressed_size); - - /* Verify that we really get the same thing back when decompressing. - * Although this could be disabled by default in all cases, it only - * takes around 2-3% of the running time of the slow algorithm to do the - * verification. */ - if (ctx->params.algorithm == WIMLIB_LZX_ALGORITHM_SLOW - #if defined(ENABLE_LZX_DEBUG) || defined(ENABLE_VERIFY_COMPRESSION) - || 1 - #endif - ) - { - struct wimlib_decompressor *decompressor; + lzx_build_params(compression_level, max_window_size, ¶ms); - if (0 == wimlib_create_decompressor(WIMLIB_COMPRESSION_TYPE_LZX, - ctx->max_window_size, - NULL, - &decompressor)) - { - int ret; - ret = wimlib_decompress(compressed_data, - compressed_size, - ctx->window, - uncompressed_size, - decompressor); - wimlib_free_decompressor(decompressor); - - if (ret) { - ERROR("Failed to decompress data we " - "compressed using LZX algorithm"); - wimlib_assert(0); - return 0; - } - if (memcmp(uncompressed_data, ctx->window, uncompressed_size)) { - ERROR("Data we compressed using LZX algorithm " - "didn't decompress to original"); - wimlib_assert(0); - return 0; - } - } else { - WARNING("Failed to create decompressor for " - "data verification!"); - } - } - return compressed_size; -} + size += sizeof(struct lzx_compressor); -static void -lzx_free_compressor(void *_ctx) -{ - struct lzx_compressor *ctx = _ctx; - - if (ctx) { - FREE(ctx->chosen_items); - FREE(ctx->cached_matches); - FREE(ctx->optimum); - lz_bt_destroy(&ctx->mf); - FREE(ctx->block_specs); - FREE(ctx->prev_tab); - FREE(ctx->window); - FREE(ctx); - } -} + size += max_window_size; -static const struct wimlib_lzx_compressor_params lzx_fast_default = { - .hdr = { - .size = sizeof(struct wimlib_lzx_compressor_params), - }, - .algorithm = WIMLIB_LZX_ALGORITHM_FAST, - .use_defaults = 0, - .alg_params = { - .fast = { - }, - }, -}; -static const struct wimlib_lzx_compressor_params lzx_slow_default = { - .hdr = { - .size = sizeof(struct wimlib_lzx_compressor_params), - }, - .algorithm = WIMLIB_LZX_ALGORITHM_SLOW, - .use_defaults = 0, - .alg_params = { - .slow = { - .use_len2_matches = 1, - .nice_match_length = 32, - .num_optim_passes = 2, - .max_search_depth = 50, - .main_nostat_cost = 15, - .len_nostat_cost = 15, - .aligned_nostat_cost = 7, - }, - }, -}; + size += DIV_ROUND_UP(max_window_size, LZX_DIV_BLOCK_SIZE) * + sizeof(struct lzx_block_spec); -static const struct wimlib_lzx_compressor_params * -lzx_get_params(const struct wimlib_compressor_params_header *_params) -{ - const struct wimlib_lzx_compressor_params *params = - (const struct wimlib_lzx_compressor_params*)_params; + size += max_window_size * sizeof(struct lzx_item); - if (params == NULL) { - LZX_DEBUG("Using default algorithm and parameters."); - params = &lzx_slow_default; - } else { - if (params->use_defaults) { - if (params->algorithm == WIMLIB_LZX_ALGORITHM_SLOW) - params = &lzx_slow_default; - else - params = &lzx_fast_default; - } + size += lz_mf_get_needed_memory(params.mf_algo, max_window_size); + if (params.choose_item_func == lzx_choose_near_optimal_item) { + size += (LZX_OPTIM_ARRAY_LENGTH + params.nice_match_length) * + sizeof(struct lzx_mc_pos_data); } - return params; + if (params.num_optim_passes > 1) + size += LZX_CACHE_LEN * sizeof(struct lz_match); + else + size += LZX_MAX_MATCHES_PER_POS * sizeof(struct lz_match); + return size; } static int -lzx_create_compressor(size_t window_size, - const struct wimlib_compressor_params_header *_params, - void **ctx_ret) +lzx_create_compressor(size_t max_block_size, unsigned int compression_level, + void **c_ret) { - const struct wimlib_lzx_compressor_params *params = lzx_get_params(_params); - struct lzx_compressor *ctx; - - LZX_DEBUG("Allocating LZX context..."); + struct lzx_compressor *c; + struct lzx_compressor_params params; + struct lz_mf_params mf_params; + unsigned window_order; + u32 max_window_size; + + window_order = lzx_get_window_order(max_block_size); + if (window_order == 0) + return WIMLIB_ERR_INVALID_PARAM; + max_window_size = max_block_size; - if (!lzx_window_size_valid(window_size)) + lzx_build_params(compression_level, max_window_size, ¶ms); + lzx_build_mf_params(¶ms, max_window_size, &mf_params); + if (!lz_mf_params_valid(&mf_params)) return WIMLIB_ERR_INVALID_PARAM; - LZX_DEBUG("Allocating memory."); + c = CALLOC(1, sizeof(struct lzx_compressor)); + if (!c) + goto oom; + + c->params = params; + c->num_main_syms = lzx_get_num_main_syms(window_order); + c->max_window_size = max_window_size; + c->window_order = window_order; - ctx = CALLOC(1, sizeof(struct lzx_compressor)); - if (ctx == NULL) + c->cur_window = ALIGNED_MALLOC(max_window_size, 16); + if (!c->cur_window) goto oom; - ctx->num_main_syms = lzx_get_num_main_syms(window_size); - ctx->max_window_size = window_size; - ctx->window = MALLOC(window_size + 12); - if (ctx->window == NULL) + c->block_specs = MALLOC(DIV_ROUND_UP(max_window_size, + LZX_DIV_BLOCK_SIZE) * + sizeof(struct lzx_block_spec)); + if (!c->block_specs) goto oom; - if (params->algorithm == WIMLIB_LZX_ALGORITHM_FAST) { - ctx->prev_tab = MALLOC(window_size * sizeof(ctx->prev_tab[0])); - if (ctx->prev_tab == NULL) - goto oom; - } + c->chosen_items = MALLOC(max_window_size * sizeof(struct lzx_item)); + if (!c->chosen_items) + goto oom; - size_t block_specs_length = DIV_ROUND_UP(window_size, LZX_DIV_BLOCK_SIZE); - ctx->block_specs = MALLOC(block_specs_length * sizeof(ctx->block_specs[0])); - if (ctx->block_specs == NULL) + c->mf = lz_mf_alloc(&mf_params); + if (!c->mf) goto oom; - if (params->algorithm == WIMLIB_LZX_ALGORITHM_SLOW) { - unsigned min_match_len = LZX_MIN_MATCH_LEN; - if (!params->alg_params.slow.use_len2_matches) - min_match_len = max(min_match_len, 3); - - if (!lz_bt_init(&ctx->mf, - window_size, - min_match_len, - LZX_MAX_MATCH_LEN, - params->alg_params.slow.nice_match_length, - params->alg_params.slow.max_search_depth)) + if (params.choose_item_func == lzx_choose_near_optimal_item) { + c->optimum = MALLOC((LZX_OPTIM_ARRAY_LENGTH + + params.nice_match_length) * + sizeof(struct lzx_mc_pos_data)); + if (!c->optimum) goto oom; } - if (params->algorithm == WIMLIB_LZX_ALGORITHM_SLOW) { - ctx->optimum = MALLOC((LZX_OPTIM_ARRAY_SIZE + - min(params->alg_params.slow.nice_match_length, - LZX_MAX_MATCH_LEN)) * - sizeof(ctx->optimum[0])); - if (!ctx->optimum) + if (params.num_optim_passes > 1) { + c->cached_matches = MALLOC(LZX_CACHE_LEN * + sizeof(struct lz_match)); + if (!c->cached_matches) goto oom; - } - - if (params->algorithm == WIMLIB_LZX_ALGORITHM_SLOW) { - ctx->cached_matches = MALLOC(LZX_CACHE_SIZE); - if (ctx->cached_matches == NULL) + c->cache_limit = c->cached_matches + LZX_CACHE_LEN - + (LZX_MAX_MATCHES_PER_POS + 1); + } else { + c->cached_matches = MALLOC(LZX_MAX_MATCHES_PER_POS * + sizeof(struct lz_match)); + if (!c->cached_matches) goto oom; - ctx->cache_limit = ctx->cached_matches + - LZX_CACHE_LEN - (LZX_MAX_MATCHES_PER_POS + 1); } - ctx->chosen_items = MALLOC(window_size * sizeof(ctx->chosen_items[0])); - if (ctx->chosen_items == NULL) - goto oom; - - memcpy(&ctx->params, params, sizeof(struct wimlib_lzx_compressor_params)); - memset(&ctx->zero_codes, 0, sizeof(ctx->zero_codes)); - - LZX_DEBUG("Successfully allocated new LZX context."); - - *ctx_ret = ctx; + *c_ret = c; return 0; oom: - lzx_free_compressor(ctx); + lzx_free_compressor(c); return WIMLIB_ERR_NOMEM; } -static u64 -lzx_get_needed_memory(size_t max_block_size, - const struct wimlib_compressor_params_header *_params) +static size_t +lzx_compress(const void *uncompressed_data, size_t uncompressed_size, + void *compressed_data, size_t compressed_size_avail, void *_c) { - const struct wimlib_lzx_compressor_params *params = lzx_get_params(_params); + struct lzx_compressor *c = _c; + struct lzx_output_bitstream os; - u64 size = 0; + /* Don't bother compressing very small inputs. */ + if (uncompressed_size < 100) + return 0; - size += sizeof(struct lzx_compressor); + /* The input data must be preprocessed. To avoid changing the original + * input, copy it to a temporary buffer. */ + memcpy(c->cur_window, uncompressed_data, uncompressed_size); + c->cur_window_size = uncompressed_size; - size += max_block_size + 12; + /* Preprocess the data. */ + lzx_do_e8_preprocessing(c->cur_window, c->cur_window_size); - size += DIV_ROUND_UP(max_block_size, LZX_DIV_BLOCK_SIZE) * - sizeof(((struct lzx_compressor*)0)->block_specs[0]); + /* Prepare the compressed data. */ + lzx_prepare_blocks(c); - if (params->algorithm == WIMLIB_LZX_ALGORITHM_SLOW) { - size += max_block_size * sizeof(((struct lzx_compressor*)0)->chosen_items[0]); - size += lz_bt_get_needed_memory(max_block_size); - size += (LZX_OPTIM_ARRAY_SIZE + - min(params->alg_params.slow.nice_match_length, - LZX_MAX_MATCH_LEN)) * - sizeof(((struct lzx_compressor *)0)->optimum[0]); - size += LZX_CACHE_SIZE; - } else { - size += max_block_size * sizeof(((struct lzx_compressor*)0)->prev_tab[0]); - } - return size; + /* Generate the compressed data and return its size, or 0 if an overflow + * occurred. */ + lzx_init_output(&os, compressed_data, compressed_size_avail); + lzx_write_all_blocks(c, &os); + return lzx_flush_output(&os); } -static bool -lzx_params_valid(const struct wimlib_compressor_params_header *_params) +static void +lzx_free_compressor(void *_c) { - const struct wimlib_lzx_compressor_params *params = - (const struct wimlib_lzx_compressor_params*)_params; - - if (params->hdr.size != sizeof(struct wimlib_lzx_compressor_params)) { - LZX_DEBUG("Invalid parameter structure size!"); - return false; - } - - if (params->algorithm != WIMLIB_LZX_ALGORITHM_SLOW && - params->algorithm != WIMLIB_LZX_ALGORITHM_FAST) - { - LZX_DEBUG("Invalid algorithm."); - return false; - } - - if (params->algorithm == WIMLIB_LZX_ALGORITHM_SLOW && - !params->use_defaults) - { - if (params->alg_params.slow.num_optim_passes < 1) - { - LZX_DEBUG("Invalid number of optimization passes!"); - return false; - } - - if (params->alg_params.slow.main_nostat_cost < 1 || - params->alg_params.slow.main_nostat_cost > 16) - { - LZX_DEBUG("Invalid main_nostat_cost!"); - return false; - } - - if (params->alg_params.slow.len_nostat_cost < 1 || - params->alg_params.slow.len_nostat_cost > 16) - { - LZX_DEBUG("Invalid len_nostat_cost!"); - return false; - } - - if (params->alg_params.slow.aligned_nostat_cost < 1 || - params->alg_params.slow.aligned_nostat_cost > 8) - { - LZX_DEBUG("Invalid aligned_nostat_cost!"); - return false; - } + struct lzx_compressor *c = _c; + + if (c) { + ALIGNED_FREE(c->cur_window); + FREE(c->block_specs); + FREE(c->chosen_items); + lz_mf_free(c->mf); + FREE(c->optimum); + FREE(c->cached_matches); + FREE(c); } - return true; } const struct compressor_ops lzx_compressor_ops = { - .params_valid = lzx_params_valid, .get_needed_memory = lzx_get_needed_memory, .create_compressor = lzx_create_compressor, .compress = lzx_compress,