diff options
Diffstat (limited to '')
| -rw-r--r-- | doc/file-format.txt | 105 |
1 files changed, 85 insertions, 20 deletions
diff --git a/doc/file-format.txt b/doc/file-format.txt index 49c9a75f..951e3943 100644 --- a/doc/file-format.txt +++ b/doc/file-format.txt @@ -43,10 +43,11 @@ The .lzma File Format 5.1. Alignment 5.2. Security 5.3. Filters - 5.3.1. LZMA2 - 5.3.2. Branch/Call/Jump Filters for Executables - 5.3.3. Delta - 5.3.3.1. Format of the Encoded Output + 5.3.1. LZMA + 5.3.2. LZMA2 + 5.3.3. Branch/Call/Jump Filters for Executables + 5.3.4. Delta + 5.3.4.1. Format of the Encoded Output 5.4. Custom Filter IDs 5.4.1. Reserved Custom Filter ID Ranges 6. Cyclic Redundancy Checks @@ -85,7 +86,7 @@ The .lzma File Format 0.2. Changes - Last modified: 2008-06-17 14:10+0300 + Last modified: 2008-09-03 14:10+0300 (A changelog will be kept once the first official version is made.) @@ -530,6 +531,10 @@ The .lzma File Format officially defined Filter IDs and the formats of their Filter Properties are described in Section 5.3. + Filter IDs greater than or equal to 0x4000_0000_0000_0000 + (2^62) are reserved for implementation-specific internal use. + These Filter IDs must never be used in List of Filter Flags. + 3.1.6. Header Padding @@ -765,20 +770,15 @@ The .lzma File Format 5.3. Filters -5.3.1. LZMA2 +5.3.1. LZMA LZMA (Lempel-Ziv-Markov chain-Algorithm) is a general-purporse compression algorithm with high compression ratio and fast decompression. LZMA is based on LZ77 and range coding algorithms. - LZMA2 uses LZMA internally, but adds support for uncompressed - chunks, eases stateful decoder implementations, and improves - support for multithreading. Thus, the plain LZMA will not be - supported in this file format. - - Filter ID: 0x21 - Size of Filter Properties: 1 byte + Filter ID: 0x40 + Size of Filter Properties: 5 bytes Changes size of data: Yes Allow as a non-last filter: No Allow as the last filter: Yes @@ -793,6 +793,66 @@ The .lzma File Format a separate document, because including the documentation here would lengthen this document considerably. + The format of the Filter Properties field is as follows: + + +-----------------+----+----+----+----+ + | LZMA Properties | Dictionary Size | + +-----------------+----+----+----+----+ + + The LZMA Properties field contains three properties. An + abbreviation is given in parentheses, followed by the value + range of the property. The field consists of + + 1) the number of literal context bits (lc, [0, 4]); + 2) the number of literal position bits (lp, [0, 4]); and + 3) the number of position bits (pb, [0, 4]). + + In addition to above ranges, the sum of lc and lp must not + exceed four. Note that this limit didn't exist in the old + LZMA_Alone format, which allowed lc to be in the range [0, 8]. + + The properties are encoded using the following formula: + + LZMA Properties = (pb * 5 + lp) * 9 + lc + + The following C code illustrates a straightforward way to + decode the properties: + + uint8_t lc, lp, pb; + uint8_t prop = get_lzma_properties(); + if (prop > (4 * 5 + 4) * 9 + 8) + return LZMA_PROPERTIES_ERROR; + + pb = prop / (9 * 5); + prop -= pb * 9 * 5; + lp = prop / 9; + lc = prop - lp * 9; + + if (lc + lp > 4) + return LZMA_PROPERTIES_ERROR; + + Dictionary Size is encoded as unsigned 32-bit little endian + integer. + + +5.3.2. LZMA2 + + LZMA2 is an extensions on top of the original LZMA. LZMA2 uses + LZMA internally, but adds support for flushing the encoder, + uncompressed chunks, eases stateful decoder implementations, + and improves support for multithreading. For most uses, it is + recommended to use LZMA2 instead of LZMA. + + Filter ID: 0x21 + Size of Filter Properties: 1 byte + Changes size of data: Yes + Allow as a non-last filter: No + Allow as the last filter: Yes + + Preferred alignment: + Input data: Adjustable to 1/2/4/8/16 byte(s) + Output data: 1 byte + The format of the one-byte Filter Properties field is as follows: @@ -818,7 +878,7 @@ The .lzma File Format 37 3 29 1536 MiB 38 2 30 2048 MiB 39 3 30 3072 MiB - 40 2 31 4096 MiB + 40 2 31 4096 MiB - 1 B Instead of having a table in the decoder, the dictionary size can be decoded using the following C code: @@ -827,11 +887,16 @@ The .lzma File Format if (bits > 40) return DICTIONARY_TOO_BIG; // Bigger than 4 GiB - uint32_t dictionary_size = 2 | (bits & 1); - dictionary_size <<= bits / 2 + 11; + uint32_t dictionary_size; + if (bits == 40) { + dictionary_size = UINT32_MAX; + } else { + dictionary_size = 2 | (bits & 1); + dictionary_size <<= bits / 2 + 11; + } -5.3.2. Branch/Call/Jump Filters for Executables +5.3.3. Branch/Call/Jump Filters for Executables These filters convert relative branch, call, and jump instructions to their absolute counterparts in executable @@ -871,7 +936,7 @@ The .lzma File Format the Subblock filter. -5.3.3. Delta +5.3.4. Delta The Delta filter may increase compression ratio when the value of the next byte correlates with the value of an earlier byte @@ -892,7 +957,7 @@ The .lzma File Format distance of 1 byte and 0xFF distance of 256 bytes. -5.3.3.1. Format of the Encoded Output +5.3.4.1. Format of the Encoded Output The code below illustrates both encoding and decoding with the Delta filter. @@ -944,7 +1009,7 @@ The .lzma File Format Bits Mask Description 0-15 0x0000_0000_0000_FFFF Filter ID 16-55 0x00FF_FFFF_FFFF_0000 Developer ID - 56-62 0x7F00_0000_0000_0000 Static prefix: 0x7F + 56-62 0x3F00_0000_0000_0000 Static prefix: 0x3F The resulting 63-bit integer will use 9 bytes of space when stored using the encoding described in Section 1.2. To get |