Description of the zisofs2 Format Revision 2.0-dev as of zisofs2-tools by Valentín KIVACHUK BURDÁ and Thomas SCHMITT 1 Oct 2020 The zisofs2 format was invented by Valentín KIVACHUK BURDÁ and Thomas SCHMITT (as extension of zisofs by H. Peter Anvin). It compresses data file content, marks it by a header and provides a pointer array for coarse random access. Within a RRIP enhanced ISO 9660 image the format is additionally marked by a System Use entry with signature "ZF" or "Z2". The uncompressed size of a single zisofs2 compressed file is restricted to 2^64 - 1 bytes. Larger files shall not be compressed. The format of version 1 of zisofs is supported by this specification. Using it for files with uncompressed size smaller than 4 GiB is friendly towards software which does not know about zisofs2. See section **LEGACY** for a summary of version 1 of zisofs. Data Types ISO 9660:7.3.1 - little endian 4-byte words ISO 9660:7.1.1 - unsigned single bytes ISO 9660:7.3.3 - 8-bytes value, first in little endian, then big endian. #uint64 - 8-bytes unsigned value in little endian Supported compressors The file header has this layout: @alg_id @alg_char Description 1 'PZ' (50)(5A) Zlib 2 'XZ' (78)(7A) XZ 3 'L4' (6C)(34) LZ4 4 'ZD' (7A)(64) Zstandard 5 'B2' (62)(32) Bzip2 @alg_id is a 7.1.1 value. @alg_char is 2 ASCII characters stored as 2 bytes Values of @alg_id = 0 and @alg_char = 'pz'(70)(7A) are reserved and must not be used. Other compressors are allowed and may be added to this list in the future Compressor strategy The default strategy for a compressor is to compress each input data block independently. The zisofs2 spec may define in the future other strategies, which will have a new @alg_id, @alg_char and a description in this section. File Header The file header has this layout: Offset Type Identifier Contents -------------------------------------------------------------------------- 0 (8 bytes) @hdr_magic Magic num (EF 22 55 A1 BC 1B 95 A0) 8 7.1.1 @hdr_version File header version (0) 9 7.1.1 @hdr_size header_size >> 2 (6) 10 7.1.1 @alg_id Algorithm Type (>=1) 11 7.1.1 @hdr_bsize log2(block_size) (15, 16, or 17) 12 #uint64 @size Uncompressed file size 20 (4 bytes) - Padding. Ignored So its size is 24. Readers shall be able to handle log2(block_size) values 15, 16 and 17 i.e. block sizes 32 kB, 64 kB, and 128 kB. Writers must not use other sizes. Block Pointers There are ceil(input_size / block_size) input resp. output blocks. Each input block is of fixed size whereas the output blocks have varying size (down to 0). For each output block there is an offset pointer giving its byte address in the overall file content. The next block pointer in the array tells the start of the next block which begins immediately after the end of its predecessor. A final pointer (*eob*) gives the first invalid byte address and thus marks the end of the last block. So there are ceil(input_size / block_size) + 1 block pointers. They are stored directly after the file header, i.e. beginning at byte 24, as an array of values in #uint64 format (8 bytes). Legacy format (zisofs) may be used, which is described in section *LEGACY* Data Part The data part begins immediately after the pointer array (*eob*). In principle it consists of the variable length output blocks as delivered by different compression algorithms when fed with the fixed size input blocks. A special case of input and output block is defined: Zero-length blocks represent a block full of 0-bytes. Such input blocks do not get processed by compress2() but shall be mapped to 0-sized output directly. Vice versa 0-sized blocks have to bypass uncompress() when being read. ZF System Use Entry Format The ZF entry follows the general layout of SUSP resp. RRIP. Its fields are: [1] "BP 1 to BP 2 - Signature Word" shall be (5A)(46) ("ZF"). [2] "BP 3 - Length" shall specify as an 8-bit number the length in bytes of the ZF entry recorded according to ISO 9660:7.1.1. This length is 16 decimal. Refer to **LEGACY** [3] "BP 4 - System Use Entry Version" shall be 2 as in ISO 9660:7.1.1. Refer to **LEGACY** [4] "BP 5 to BP 6 - Algorithm" shall be two chars to indicate the compression algorithm. For example, (50)(5A) ("PZ") (This is a copy of @alg_char). Refer to **LEGACY** [5] "BP 7 - Header Size Div 4" shall specify as an 8-bit number the number of 4-byte words in the header part of the file data recorded according to ISO 9660:7.1.1. (This is a copy of @hdr_size). [6] "BP 8 - Log2 of Block Size" shall specify as an 8-bit number the binary logarithm of the compression block size recorded according to ISO 9660:7.1.1. (This is a copy of header byte 13 (@hdr_bsize), resp. header BP 14. The value has to be 15, 16 or 17 i.e. 32 kiB, 64 kiB, or 128 kiB.) [7] "BP 9 to BP 16 - Virtual Uncompressed File Size" shall contain as a 64-bit unsigned little endian number the uncompressed file size represented by the given extent. Refer to **LEGACY** | 'Z' | 'F' | LENGTH | 2 | 'P' | 'Z' | HEADER SIZE DIV 4 | | LOG2 BLOCK SIZE | UNCOMPRESSED SIZE | Example (block size 128 kiB, uncompressed file size = 40 TB): { 'Z', 'F', 16, 2, 'P', 'Z', 8, 17, 0x00, 0x80, 0xCA, 0x39, 0x61, 0x24, 0x00, 0x00 } Z2 System Use Entry Format Linux kernels which are configured by CONFIG_ZISOFS to recognize zisofs but are not aware of zisofs2 will complain about ZF entries which announce algorithms other than "pz". The system log will show for each zisofs2 compressed file at first stat(2) or open(2) a line like this: isofs: Unknown ZF compression algorithm: PZ To avoid these complaints, it is possible to use [1] "BP 1 to BP 2 - Signature Word" shall be (5A)(32) ("Z2"). instead of "ZF" with the System Use Entry Format specified above. Everything else shall be like in ZF format version 2, including the version number itself: [3] "BP 4 - System Use Entry Version" shall be 2 as in ISO 9660:7.1.1. **LEGACY** zisofs2 supports old readers by respecting the zisofs format. This section describes which definitions from zisofs2 must change to be compatible with zisofs. - General behaviour The uncompressed size of a single zisofs compressed file is restricted to 4 GiB - 1. Larger files shall not be compressed. - Supported algorithms Only algorithm Zlib with default strategy is supported. - The file header must follow this structure: Offset Type Identifier Contents 0 (8 bytes) @hdr_magic Magic number (37 E4 53 96 C9 DB D6 07) 8 7.3.1 @size Uncompressed file size 12 7.1.1 @hdr_size header_size >> 2 (4) 13 7.1.1 @hdr_bsize log2(block_size) (15, 16, or 17) 14 (2 bytes) - Reserved, must be zero So its size is 16. - Block pointers The array must use ISO 9660:7.3.1 (4 bytes) values. - ZF entry Its fields are: [1] "BP 1 to BP 2 - Signature Word" shall be (5A)(46) ("ZF"). [2] "BP 3 - Length" must be 16 decimal. [3] "BP 4 - System Use Entry Version" must be 1. [4] "BP 5 to BP 6 - Algorithm" must be (70)(7A) ("pz"). [5] "BP 7 - Header Size Div 4" - same as zisofs2. [6] "BP 8 - Log2 of Block Size" - same as zisofs2. [7] "BP 9 to BP 16 - Uncompressed Size" This field shall be recorded according to ISO 9660:7.3.3. (This number is the same as @size ) | 'Z' | 'F' | LENGTH | 1 | 'p' | 'z' | HEADER SIZE DIV 4 | | LOG2 BLOCK SIZE | UNCOMPRESSED SIZE | Example (block size 32 kiB, uncompressed file size = 1,234,567 bytes): { 'Z', 'F', 16, 1, 'p', 'z', 4, 15, 0x87, 0xD6, 0x12, 0x00, 0x00, 0x12, 0xD6, 0x87 } References: zisofs2-tools https://github.com/vk496/zisofs2-tools zisofs-tools http://freshmeat.net/projects/zisofs-tools/ zlib: /usr/include/zlib.h cdrtools with mkisofs ftp://ftp.berlios.de/pub/cdrecord/alpha ECMA-119 aka ISO 9660 http://www.ecma-international.org/publications/files/ECMA-ST/Ecma-119.pdf SUSP 1.12 ftp://ftp.ymi.com/pub/rockridge/susp112.ps RRIP 1.12 ftp://ftp.ymi.com/pub/rockridge/rrip112.ps zisofs version 1 libisofs-*/doc/zisofs_format.txt https://dev.lovelyhq.com/libburnia/libisofs/raw/branch/master/doc/zisofs_format.txt --------------------------------------------------------------------------- This text is under Copyright (c) 2009 - 2010, 2020 Thomas SCHMITT Copyright (c) 2020 - Valentín KIVACHUK BURDÁ It shall reflect the effective technical specifications as implemented in zisofs2-tools and the Linux kernel. So please contact mailing list or to the copyright holders in private, if you want to make changes. Only if you cannot reach the copyright holder for at least one month it is permissible to modify and distribute this text under the license "GPLv3".