X-Git-Url: https://wimlib.net/git/?p=wimlib;a=blobdiff_plain;f=src%2Flzx-decompress.c;h=4047f956c376abf0e2d35dbe02d5a0ef37adc6ec;hp=1cc5113c4515b0f1fa60065f35a1b714982f13e6;hb=157d002da341c9109c5c065893ae82c6dbf5d4e8;hpb=9e56d04309e3e6a896319225288f0c86bd36d34e diff --git a/src/lzx-decompress.c b/src/lzx-decompress.c index 1cc5113c..4047f956 100644 --- a/src/lzx-decompress.c +++ b/src/lzx-decompress.c @@ -38,13 +38,13 @@ * * A compressed WIM resource consists of a table of chunk offsets followed by * the compressed chunks themselves. All compressed chunks except possibly the - * last decompress to WIM_CHUNK_SIZE (= 32768) bytes. This is quite similar to - * the cabinet (.cab) file format, but they are not the same. According to the - * cabinet format documentation, the LZX block size is independent from the - * CFDATA blocks, and a LZX block may span several CFDATA blocks. However, in - * WIMs, LZX blocks do not appear to ever span multiple WIM chunks. Note that - * this means any WIM chunk may be decompressed or compressed independently from - * any other chunk, which is convenient. + * last decompress to 32768 bytes. This is quite similar to the cabinet (.cab) + * file format, but they are not the same. According to the cabinet format + * documentation, the LZX block size is independent from the CFDATA blocks, and + * a LZX block may span several CFDATA blocks. However, in WIMs, LZX blocks do + * not appear to ever span multiple WIM chunks. Note that this means any WIM + * chunk may be decompressed or compressed independently from any other chunk, + * which is convenient. * * A LZX compressed WIM chunk contains one or more LZX blocks of the aligned, * verbatim, or uncompressed block types. For aligned and verbatim blocks, the @@ -106,71 +106,86 @@ * succeed. */ -#include "util.h" -#include "lzx.h" -#include "decompress.h" +#ifdef HAVE_CONFIG_H +# include "config.h" +#endif + +#include "wimlib.h" +#include "wimlib/decompress.h" +#include "wimlib/lzx.h" +#include "wimlib/util.h" + #include /* Huffman decoding tables and maps from symbols to code lengths. */ struct lzx_tables { - u16 maintree_decode_table[(1 << LZX_MAINTREE_TABLEBITS) + - (LZX_MAINTREE_NUM_SYMBOLS * 2)]; - u8 maintree_lens[LZX_MAINTREE_NUM_SYMBOLS]; + u16 maintree_decode_table[(1 << LZX_MAINCODE_TABLEBITS) + + (LZX_MAINCODE_MAX_NUM_SYMBOLS * 2)] + _aligned_attribute(DECODE_TABLE_ALIGNMENT); + u8 maintree_lens[LZX_MAINCODE_MAX_NUM_SYMBOLS]; - u16 lentree_decode_table[(1 << LZX_LENTREE_TABLEBITS) + - (LZX_LENTREE_NUM_SYMBOLS * 2)]; - u8 lentree_lens[LZX_LENTREE_NUM_SYMBOLS]; + u16 lentree_decode_table[(1 << LZX_LENCODE_TABLEBITS) + + (LZX_LENCODE_NUM_SYMBOLS * 2)] + _aligned_attribute(DECODE_TABLE_ALIGNMENT); + u8 lentree_lens[LZX_LENCODE_NUM_SYMBOLS]; - u16 alignedtree_decode_table[(1 << LZX_ALIGNEDTREE_TABLEBITS) + - (LZX_ALIGNEDTREE_NUM_SYMBOLS * 2)]; - u8 alignedtree_lens[LZX_ALIGNEDTREE_NUM_SYMBOLS]; -}; + u16 alignedtree_decode_table[(1 << LZX_ALIGNEDCODE_TABLEBITS) + + (LZX_ALIGNEDCODE_NUM_SYMBOLS * 2)] + _aligned_attribute(DECODE_TABLE_ALIGNMENT); + u8 alignedtree_lens[LZX_ALIGNEDCODE_NUM_SYMBOLS]; +} _aligned_attribute(DECODE_TABLE_ALIGNMENT); /* * Reads a Huffman-encoded symbol using the pre-tree. */ -static inline int read_huffsym_using_pretree(struct input_bitstream *istream, - const u16 pretree_decode_table[], - const u8 pretree_lens[], unsigned *n) +static inline int +read_huffsym_using_pretree(struct input_bitstream *istream, + const u16 pretree_decode_table[], + const u8 pretree_lens[], unsigned *n) { return read_huffsym(istream, pretree_decode_table, pretree_lens, - LZX_PRETREE_NUM_SYMBOLS, LZX_PRETREE_TABLEBITS, n, - LZX_MAX_CODEWORD_LEN); + LZX_PRECODE_NUM_SYMBOLS, LZX_PRECODE_TABLEBITS, n, + LZX_MAX_PRE_CODEWORD_LEN); } /* Reads a Huffman-encoded symbol using the main tree. */ -static inline int read_huffsym_using_maintree(struct input_bitstream *istream, - const struct lzx_tables *tables, - unsigned *n) +static inline int +read_huffsym_using_maintree(struct input_bitstream *istream, + const struct lzx_tables *tables, + unsigned *n, + unsigned num_main_syms) { return read_huffsym(istream, tables->maintree_decode_table, - tables->maintree_lens, LZX_MAINTREE_NUM_SYMBOLS, - LZX_MAINTREE_TABLEBITS, n, LZX_MAX_CODEWORD_LEN); + tables->maintree_lens, num_main_syms, + LZX_MAINCODE_TABLEBITS, n, LZX_MAX_MAIN_CODEWORD_LEN); } /* Reads a Huffman-encoded symbol using the length tree. */ -static inline int read_huffsym_using_lentree(struct input_bitstream *istream, - const struct lzx_tables *tables, - unsigned *n) +static inline int +read_huffsym_using_lentree(struct input_bitstream *istream, + const struct lzx_tables *tables, + unsigned *n) { return read_huffsym(istream, tables->lentree_decode_table, - tables->lentree_lens, LZX_LENTREE_NUM_SYMBOLS, - LZX_LENTREE_TABLEBITS, n, LZX_MAX_CODEWORD_LEN); + tables->lentree_lens, LZX_LENCODE_NUM_SYMBOLS, + LZX_LENCODE_TABLEBITS, n, LZX_MAX_LEN_CODEWORD_LEN); } /* Reads a Huffman-encoded symbol using the aligned offset tree. */ -static inline int read_huffsym_using_alignedtree(struct input_bitstream *istream, - const struct lzx_tables *tables, - unsigned *n) +static inline int +read_huffsym_using_alignedtree(struct input_bitstream *istream, + const struct lzx_tables *tables, + unsigned *n) { return read_huffsym(istream, tables->alignedtree_decode_table, tables->alignedtree_lens, - LZX_ALIGNEDTREE_NUM_SYMBOLS, - LZX_ALIGNEDTREE_TABLEBITS, n, 8); + LZX_ALIGNEDCODE_NUM_SYMBOLS, + LZX_ALIGNEDCODE_TABLEBITS, n, + LZX_MAX_ALIGNED_CODEWORD_LEN); } /* @@ -185,34 +200,36 @@ static inline int read_huffsym_using_alignedtree(struct input_bitstream *istream * @num_lens: Number of length values to decode and return. * */ -static int lzx_read_code_lens(struct input_bitstream *istream, u8 lens[], - unsigned num_lens) +static int +lzx_read_code_lens(struct input_bitstream *istream, u8 lens[], + unsigned num_lens) { /* Declare the decoding table and length table for the pretree. */ - u16 pretree_decode_table[(1 << LZX_PRETREE_TABLEBITS) + - (LZX_PRETREE_NUM_SYMBOLS * 2)]; - u8 pretree_lens[LZX_PRETREE_NUM_SYMBOLS]; + u16 pretree_decode_table[(1 << LZX_PRECODE_TABLEBITS) + + (LZX_PRECODE_NUM_SYMBOLS * 2)] + _aligned_attribute(DECODE_TABLE_ALIGNMENT); + u8 pretree_lens[LZX_PRECODE_NUM_SYMBOLS]; unsigned i; unsigned len; int ret; /* Read the code lengths of the pretree codes. There are 20 lengths of * 4 bits each. */ - for (i = 0; i < LZX_PRETREE_NUM_SYMBOLS; i++) { - ret = bitstream_read_bits(istream, LZX_PRETREE_ELEMENT_SIZE, + for (i = 0; i < LZX_PRECODE_NUM_SYMBOLS; i++) { + ret = bitstream_read_bits(istream, LZX_PRECODE_ELEMENT_SIZE, &len); - if (ret != 0) + if (ret) return ret; pretree_lens[i] = len; } /* Make the decoding table for the pretree. */ ret = make_huffman_decode_table(pretree_decode_table, - LZX_PRETREE_NUM_SYMBOLS, - LZX_PRETREE_TABLEBITS, + LZX_PRECODE_NUM_SYMBOLS, + LZX_PRECODE_TABLEBITS, pretree_lens, - LZX_MAX_CODEWORD_LEN); - if (ret != 0) + LZX_MAX_PRE_CODEWORD_LEN); + if (ret) return ret; /* Pointer past the last length value that needs to be filled in. */ @@ -230,16 +247,16 @@ static int lzx_read_code_lens(struct input_bitstream *istream, u8 lens[], unsigned num_zeroes; unsigned code; unsigned num_same; - char value; + signed char value; ret = read_huffsym_using_pretree(istream, pretree_decode_table, pretree_lens, &tree_code); - if (ret != 0) + if (ret) return ret; switch (tree_code) { case 17: /* Run of 0's */ ret = bitstream_read_bits(istream, 4, &num_zeroes); - if (ret != 0) + if (ret) return ret; num_zeroes += 4; while (num_zeroes--) { @@ -250,7 +267,7 @@ static int lzx_read_code_lens(struct input_bitstream *istream, u8 lens[], break; case 18: /* Longer run of 0's */ ret = bitstream_read_bits(istream, 5, &num_zeroes); - if (ret != 0) + if (ret) return ret; num_zeroes += 20; while (num_zeroes--) { @@ -261,16 +278,16 @@ static int lzx_read_code_lens(struct input_bitstream *istream, u8 lens[], break; case 19: /* Run of identical lengths */ ret = bitstream_read_bits(istream, 1, &num_same); - if (ret != 0) + if (ret) return ret; num_same += 4; ret = read_huffsym_using_pretree(istream, pretree_decode_table, pretree_lens, &code); - if (ret != 0) + if (ret) return ret; - value = (char)*lens - (char)code; + value = (signed char)*lens - (signed char)code; if (value < 0) value += 17; while (num_same--) { @@ -280,7 +297,7 @@ static int lzx_read_code_lens(struct input_bitstream *istream, u8 lens[], } break; default: /* Difference from old length. */ - value = (char)*lens - (char)tree_code; + value = (signed char)*lens - (signed char)tree_code; if (value < 0) value += 17; *lens = value; @@ -306,11 +323,14 @@ static int lzx_read_code_lens(struct input_bitstream *istream, u8 lens[], * R0, R1, and R2 will be written (only for uncompressed * blocks, which contain this information in the header) */ -static int lzx_read_block_header(struct input_bitstream *istream, - unsigned *block_size_ret, - unsigned *block_type_ret, - struct lzx_tables *tables, - struct lru_queue *queue) +static int +lzx_read_block_header(struct input_bitstream *istream, + unsigned num_main_syms, + unsigned max_window_size, + unsigned *block_size_ret, + unsigned *block_type_ret, + struct lzx_tables *tables, + struct lzx_lru_queue *queue) { int ret; unsigned block_type; @@ -320,8 +340,8 @@ static int lzx_read_block_header(struct input_bitstream *istream, unsigned len; ret = bitstream_ensure_bits(istream, 4); - if (ret != 0) { - ERROR("LZX input stream overrun"); + if (ret) { + LZX_DEBUG("LZX input stream overrun"); return ret; } @@ -329,18 +349,34 @@ static int lzx_read_block_header(struct input_bitstream *istream, * of the LZX_BLOCKTYPE_* values. */ block_type = bitstream_read_bits_nocheck(istream, 3); - /* The next bit indicates whether the block size is the default (32768), - * indicated by a 1 bit, or whether the block size is given by the next - * 16 bits, indicated by a 0 bit. */ + /* Read the block size. This mirrors the behavior + * lzx_write_compressed_block() in lzx-compress.c; see that for more + * details. */ s = bitstream_read_bits_nocheck(istream, 1); - if (s) { - block_size = 32768; + block_size = LZX_DEFAULT_BLOCK_SIZE; } else { - ret = bitstream_read_bits(istream, 16, &block_size); - if (ret != 0) + unsigned tmp; + block_size = 0; + + ret = bitstream_read_bits(istream, 8, &tmp); + if (ret) + return ret; + block_size |= tmp; + + ret = bitstream_read_bits(istream, 8, &tmp); + if (ret) return ret; - block_size = le16_to_cpu(block_size); + block_size <<= 8; + block_size |= tmp; + + if (max_window_size >= 65536) { + ret = bitstream_read_bits(istream, 8, &tmp); + if (ret) + return ret; + block_size <<= 8; + block_size |= tmp; + } } switch (block_type) { @@ -348,24 +384,24 @@ static int lzx_read_block_header(struct input_bitstream *istream, /* Read the path lengths for the elements of the aligned tree, * then build it. */ - for (i = 0; i < LZX_ALIGNEDTREE_NUM_SYMBOLS; i++) { + for (i = 0; i < LZX_ALIGNEDCODE_NUM_SYMBOLS; i++) { ret = bitstream_read_bits(istream, - LZX_ALIGNEDTREE_ELEMENT_SIZE, + LZX_ALIGNEDCODE_ELEMENT_SIZE, &len); - if (ret != 0) + if (ret) return ret; tables->alignedtree_lens[i] = len; } LZX_DEBUG("Building the aligned tree."); ret = make_huffman_decode_table(tables->alignedtree_decode_table, - LZX_ALIGNEDTREE_NUM_SYMBOLS, - LZX_ALIGNEDTREE_TABLEBITS, + LZX_ALIGNEDCODE_NUM_SYMBOLS, + LZX_ALIGNEDCODE_TABLEBITS, tables->alignedtree_lens, - 8); - if (ret != 0) { - ERROR("lzx_decompress(): Failed to make the decode " - "table for the aligned offset tree"); + LZX_MAX_ALIGNED_CODEWORD_LEN); + if (ret) { + LZX_DEBUG("Failed to make the decode table for the " + "aligned offset tree"); return ret; } @@ -381,10 +417,9 @@ static int lzx_read_block_header(struct input_bitstream *istream, * tree. */ ret = lzx_read_code_lens(istream, tables->maintree_lens, LZX_NUM_CHARS); - if (ret != 0) { - ERROR("lzx_decompress(): Failed to read the code " - "lengths for the first 256 elements of the " - "main tree"); + if (ret) { + LZX_DEBUG("Failed to read the code lengths for the " + "first 256 elements of the main tree"); return ret; } @@ -392,14 +427,13 @@ static int lzx_read_block_header(struct input_bitstream *istream, * tree. */ LZX_DEBUG("Reading path lengths for remaining elements of " "main tree (%d elements).", - LZX_MAINTREE_NUM_SYMBOLS - LZX_NUM_CHARS); + num_main_syms - LZX_NUM_CHARS); ret = lzx_read_code_lens(istream, tables->maintree_lens + LZX_NUM_CHARS, - LZX_MAINTREE_NUM_SYMBOLS - LZX_NUM_CHARS); - if (ret != 0) { - ERROR("lzx_decompress(): Failed to read the path " - "lengths for the remaining elements of the main " - "tree"); + num_main_syms - LZX_NUM_CHARS); + if (ret) { + LZX_DEBUG("Failed to read the path lengths for the " + "remaining elements of the main tree"); return ret; } @@ -407,34 +441,33 @@ static int lzx_read_block_header(struct input_bitstream *istream, "table for the main tree."); ret = make_huffman_decode_table(tables->maintree_decode_table, - LZX_MAINTREE_NUM_SYMBOLS, - LZX_MAINTREE_TABLEBITS, + num_main_syms, + LZX_MAINCODE_TABLEBITS, tables->maintree_lens, - LZX_MAX_CODEWORD_LEN); - if (ret != 0) { - ERROR("lzx_decompress(): Failed to make the decode " - "table for the main tree"); + LZX_MAX_MAIN_CODEWORD_LEN); + if (ret) { + LZX_DEBUG("Failed to make the decode " + "table for the main tree"); return ret; } LZX_DEBUG("Reading path lengths for the length tree."); ret = lzx_read_code_lens(istream, tables->lentree_lens, - LZX_LENTREE_NUM_SYMBOLS); - if (ret != 0) { - ERROR("lzx_decompress(): Failed to read the path " - "lengths for the length tree"); + LZX_LENCODE_NUM_SYMBOLS); + if (ret) { + LZX_DEBUG("Failed to read the path " + "lengths for the length tree"); return ret; } LZX_DEBUG("Building the length tree."); ret = make_huffman_decode_table(tables->lentree_decode_table, - LZX_LENTREE_NUM_SYMBOLS, - LZX_LENTREE_TABLEBITS, + LZX_LENCODE_NUM_SYMBOLS, + LZX_LENCODE_TABLEBITS, tables->lentree_lens, - LZX_MAX_CODEWORD_LEN); - if (ret != 0) { - ERROR("lzx_decompress(): Failed to build the length " - "Huffman tree"); + LZX_MAX_LEN_CODEWORD_LEN); + if (ret) { + LZX_DEBUG("Failed to build the length Huffman tree"); return ret; } /* The bitstream of compressed literals and matches for this @@ -449,26 +482,32 @@ static int lzx_read_block_header(struct input_bitstream *istream, * *already* aligned, the correct thing to do is to throw away * the next 16 bits. */ if (istream->bitsleft == 0) { - if (istream->data_bytes_left < 14) + if (istream->data_bytes_left < 14) { + LZX_DEBUG("Insufficient length in " + "uncompressed block"); return -1; + } istream->data += 2; istream->data_bytes_left -= 2; } else { - if (istream->data_bytes_left < 12) + if (istream->data_bytes_left < 12) { + LZX_DEBUG("Insufficient length in " + "uncompressed block"); return -1; + } istream->bitsleft = 0; istream->bitbuf = 0; } - queue->R0 = le32_to_cpu(*(u32*)(istream->data + 0)); - queue->R1 = le32_to_cpu(*(u32*)(istream->data + 4)); - queue->R2 = le32_to_cpu(*(u32*)(istream->data + 8)); + queue->R[0] = le32_to_cpu(*(u32*)(istream->data + 0)); + queue->R[1] = le32_to_cpu(*(u32*)(istream->data + 4)); + queue->R[2] = le32_to_cpu(*(u32*)(istream->data + 8)); istream->data += 12; istream->data_bytes_left -= 12; /* The uncompressed data of this block directly follows and will * be read in lzx_decompress(). */ break; default: - ERROR("lzx_decompress(): Found invalid block"); + LZX_DEBUG("Found invalid block"); return -1; } *block_type_ret = block_type; @@ -511,12 +550,13 @@ static int lzx_read_block_header(struct input_bitstream *istream, * - Match refers to data before the window. * - The input bitstream ended unexpectedly. */ -static int lzx_decode_match(unsigned main_element, int block_type, - unsigned bytes_remaining, u8 *window, - unsigned window_pos, - const struct lzx_tables *tables, - struct lru_queue *queue, - struct input_bitstream *istream) +static int +lzx_decode_match(unsigned main_element, int block_type, + unsigned bytes_remaining, u8 *window, + unsigned window_pos, + const struct lzx_tables *tables, + struct lzx_lru_queue *queue, + struct input_bitstream *istream) { unsigned length_header; unsigned position_slot; @@ -541,14 +581,15 @@ static int lzx_decode_match(unsigned main_element, int block_type, position_slot = main_element >> 3; /* If the length_header is less than LZX_NUM_PRIMARY_LENS (= 7), it - * gives the match length as the offset from LZX_MIN_MATCH. Otherwise, - * the length is given by an additional symbol encoded using the length - * tree, offset by 9 (LZX_MIN_MATCH + LZX_NUM_PRIMARY_LENS) */ - match_len = LZX_MIN_MATCH + length_header; + * gives the match length as the offset from LZX_MIN_MATCH_LEN. + * Otherwise, the length is given by an additional symbol encoded using + * the length tree, offset by 9 (LZX_MIN_MATCH_LEN + + * LZX_NUM_PRIMARY_LENS) */ + match_len = LZX_MIN_MATCH_LEN + length_header; if (length_header == LZX_NUM_PRIMARY_LENS) { ret = read_huffsym_using_lentree(istream, tables, &additional_len); - if (ret != 0) + if (ret) return ret; match_len += additional_len; } @@ -559,18 +600,18 @@ static int lzx_decode_match(unsigned main_element, int block_type, * queue. */ switch (position_slot) { case 0: - match_offset = queue->R0; + match_offset = queue->R[0]; break; case 1: - match_offset = queue->R1; - swap(queue->R0, queue->R1); + match_offset = queue->R[1]; + swap(queue->R[0], queue->R[1]); break; case 2: /* The queue doesn't work quite the same as a real LRU queue, * since using the R2 offset doesn't bump the R1 offset down to * R2. */ - match_offset = queue->R2; - swap(queue->R0, queue->R2); + match_offset = queue->R[2]; + swap(queue->R[0], queue->R[2]); break; default: /* Otherwise, the offset was not encoded as one the offsets in @@ -597,14 +638,14 @@ static int lzx_decode_match(unsigned main_element, int block_type, * will just set it to 0. ) */ ret = bitstream_read_bits(istream, num_extra_bits - 3, &verbatim_bits); - if (ret != 0) + if (ret) return ret; verbatim_bits <<= 3; ret = read_huffsym_using_alignedtree(istream, tables, &aligned_bits); - if (ret != 0) + if (ret) return ret; } else { /* For non-aligned blocks, or for aligned blocks with @@ -613,7 +654,7 @@ static int lzx_decode_match(unsigned main_element, int block_type, * the alignment is taken to be 0. */ ret = bitstream_read_bits(istream, num_extra_bits, &verbatim_bits); - if (ret != 0) + if (ret) return ret; aligned_bits = 0; @@ -621,34 +662,36 @@ static int lzx_decode_match(unsigned main_element, int block_type, /* Calculate the match offset. */ match_offset = lzx_position_base[position_slot] + - verbatim_bits + aligned_bits - 2; + verbatim_bits + aligned_bits - LZX_OFFSET_OFFSET; /* Update the LRU queue. */ - queue->R2 = queue->R1; - queue->R1 = queue->R0; - queue->R0 = match_offset; + queue->R[2] = queue->R[1]; + queue->R[1] = queue->R[0]; + queue->R[0] = match_offset; break; } /* Verify that the match is in the bounds of the part of the window * currently in use, then copy the source of the match to the current * position. */ - match_dest = window + window_pos; - match_src = match_dest - match_offset; if (match_len > bytes_remaining) { - ERROR("lzx_decode_match(): Match of length %u bytes overflows " - "uncompressed block size", match_len); + LZX_DEBUG("Match of length %u bytes overflows " + "uncompressed block size", match_len); return -1; } - if (match_src < window) { - ERROR("lzx_decode_match(): Match of length %u bytes references " - "data before window (match_offset = %u, window_pos = %u)", - match_len, match_offset, window_pos); + if (match_offset > window_pos) { + LZX_DEBUG("Match of length %u bytes references " + "data before window (match_offset = %u, " + "window_pos = %u)", + match_len, match_offset, window_pos); return -1; } + match_dest = window + window_pos; + match_src = match_dest - match_offset; + #if 0 printf("Match: src %u, dst %u, len %u\n", match_src - window, match_dest - window, @@ -668,11 +711,12 @@ static int lzx_decode_match(unsigned main_element, int block_type, return match_len; } -static void undo_call_insn_translation(u32 *call_insn_target, int input_pos, - int32_t file_size) +static void +undo_call_insn_translation(u32 *call_insn_target, int input_pos, + s32 file_size) { - int32_t abs_offset; - int32_t rel_offset; + s32 abs_offset; + s32 rel_offset; abs_offset = le32_to_cpu(*call_insn_target); if (abs_offset >= -input_pos && abs_offset < file_size) { @@ -709,8 +753,8 @@ static void undo_call_insn_translation(u32 *call_insn_target, int input_pos, * Call instruction processing is supposed to take the file size as a parameter, * as it is used in calculating the translated jump targets. But in WIM files, * this file size is always the same (LZX_WIM_MAGIC_FILESIZE == 12000000).*/ -static void undo_call_insn_preprocessing(u8 uncompressed_data[], - int uncompressed_data_len) +static void +undo_call_insn_preprocessing(u8 uncompressed_data[], int uncompressed_data_len) { for (int i = 0; i < uncompressed_data_len - 10; i++) { if (uncompressed_data[i] == 0xe8) { @@ -729,6 +773,7 @@ static void undo_call_insn_preprocessing(u8 uncompressed_data[], * @block_type: The type of the block (LZX_BLOCKTYPE_VERBATIM or * LZX_BLOCKTYPE_ALIGNED) * @block_size: The size of the block, in bytes. + * @num_main_syms: Number of symbols in the main alphabet. * @window: Pointer to the decompression window. * @window_pos: The current position in the window. Will be 0 for the first * block. @@ -737,12 +782,14 @@ static void undo_call_insn_preprocessing(u8 uncompressed_data[], * @queue: The least-recently-used queue for match offsets. * @istream: The input bitstream for the compressed literals. */ -static int lzx_decompress_block(int block_type, unsigned block_size, - u8 *window, - unsigned window_pos, - const struct lzx_tables *tables, - struct lru_queue *queue, - struct input_bitstream *istream) +static int +lzx_decompress_block(int block_type, unsigned block_size, + unsigned num_main_syms, + u8 *window, + unsigned window_pos, + const struct lzx_tables *tables, + struct lzx_lru_queue *queue, + struct input_bitstream *istream) { unsigned main_element; unsigned end; @@ -752,15 +799,16 @@ static int lzx_decompress_block(int block_type, unsigned block_size, end = window_pos + block_size; while (window_pos < end) { ret = read_huffsym_using_maintree(istream, tables, - &main_element); - if (ret != 0) + &main_element, + num_main_syms); + if (ret) return ret; if (main_element < LZX_NUM_CHARS) { /* literal: 0 to LZX_NUM_CHARS - 1 */ window[window_pos++] = main_element; } else { - /* match: LZX_NUM_CHARS to LZX_MAINTREE_NUM_SYMBOLS - 1 */ + /* match: LZX_NUM_CHARS to num_main_syms - 1 */ match_len = lzx_decode_match(main_element, block_type, end - window_pos, @@ -777,48 +825,45 @@ static int lzx_decompress_block(int block_type, unsigned block_size, return 0; } -/* - * Decompresses a block of LZX-compressed data as used in the WIM file format. - * - * Note that this will NOT work unmodified for LZX as used in the cabinet - * format, which is not the same as in the WIM format! - * - * @compressed_data: A pointer to the compressed data. - * - * @compressed_len: The length of the compressed data, in bytes. - * - * @uncompressed_data: A pointer to the buffer into which to write the - * uncompressed data. - * - * @uncompressed_len: The length of the uncompressed data. It must be - * 32768 bytes or less. - * - * Return 0 on success; non-zero on failure. - */ -int lzx_decompress(const void *compressed_data, unsigned compressed_len, - void *uncompressed_data, unsigned uncompressed_len) +WIMLIBAPI int +wimlib_lzx_decompress2(const void *compressed_data, unsigned compressed_len, + void *uncompressed_data, unsigned uncompressed_len, + u32 max_window_size) { struct lzx_tables tables; struct input_bitstream istream; - struct lru_queue queue; + struct lzx_lru_queue queue; unsigned window_pos; unsigned block_size; unsigned block_type; + unsigned num_main_syms; int ret; bool e8_preprocessing_done; - LZX_DEBUG("lzx_decompress (compressed_data = %p, compressed_len = %d, " - "uncompressed_data = %p, uncompressed_len = %d).", + LZX_DEBUG("compressed_data = %p, compressed_len = %u, " + "uncompressed_data = %p, uncompressed_len = %u, " + "max_window_size=%u).", compressed_data, compressed_len, - uncompressed_data, uncompressed_len); + uncompressed_data, uncompressed_len, max_window_size); + + if (!lzx_window_size_valid(max_window_size)) { + LZX_DEBUG("Window size of %u is invalid!", + max_window_size); + return -1; + } + + num_main_syms = lzx_get_num_main_syms(max_window_size); - wimlib_assert(uncompressed_len <= 32768); + if (uncompressed_len > max_window_size) { + LZX_DEBUG("Uncompressed chunk size of %u exceeds " + "window size of %u!", + uncompressed_len, max_window_size); + return -1; + } memset(tables.maintree_lens, 0, sizeof(tables.maintree_lens)); memset(tables.lentree_lens, 0, sizeof(tables.lentree_lens)); - queue.R0 = 1; - queue.R1 = 1; - queue.R2 = 1; + lzx_lru_queue_init(&queue); init_input_bitstream(&istream, compressed_data, compressed_len); e8_preprocessing_done = false; /* Set to true if there may be 0xe8 bytes @@ -834,18 +879,19 @@ int lzx_decompress(const void *compressed_data, unsigned compressed_len, window_pos += block_size) { LZX_DEBUG("Reading block header."); - ret = lzx_read_block_header(&istream, &block_size, + ret = lzx_read_block_header(&istream, num_main_syms, + max_window_size, &block_size, &block_type, &tables, &queue); - if (ret != 0) + if (ret) return ret; LZX_DEBUG("block_size = %u, window_pos = %u", block_size, window_pos); if (block_size > uncompressed_len - window_pos) { - ERROR("lzx_decompress(): Expected a block size of at " - "most %u bytes (found %u bytes)", - uncompressed_len - window_pos, block_size); + LZX_DEBUG("Expected a block size of at " + "most %u bytes (found %u bytes)", + uncompressed_len - window_pos, block_size); return -1; } @@ -858,23 +904,25 @@ int lzx_decompress(const void *compressed_data, unsigned compressed_len, LZX_DEBUG("LZX_BLOCKTYPE_ALIGNED"); ret = lzx_decompress_block(block_type, block_size, + num_main_syms, uncompressed_data, window_pos, &tables, &queue, &istream); - if (ret != 0) + if (ret) return ret; + if (tables.maintree_lens[0xe8] != 0) e8_preprocessing_done = true; break; case LZX_BLOCKTYPE_UNCOMPRESSED: LZX_DEBUG("LZX_BLOCKTYPE_UNCOMPRESSED"); if (istream.data_bytes_left < block_size) { - ERROR("Unexpected end of input when " - "reading %u bytes from LZX bitstream " - "(only have %u bytes left)", - block_size, istream.data_bytes_left); + LZX_DEBUG("Unexpected end of input when " + "reading %u bytes from LZX bitstream " + "(only have %u bytes left)", + block_size, istream.data_bytes_left); return -1; } memcpy(&((u8*)uncompressed_data)[window_pos], istream.data, @@ -895,3 +943,13 @@ int lzx_decompress(const void *compressed_data, unsigned compressed_len, undo_call_insn_preprocessing(uncompressed_data, uncompressed_len); return 0; } + +/* API function documented in wimlib.h */ +WIMLIBAPI int +wimlib_lzx_decompress(const void *compressed_data, unsigned compressed_len, + void *uncompressed_data, unsigned uncompressed_len) +{ + return wimlib_lzx_decompress2(compressed_data, compressed_len, + uncompressed_data, uncompressed_len, + 32768); +}