ISO 9660

The system area, the first 32,768 data bytes of the disc (16 sectors of 2,048 bytes each), is unused by ISO 9660 and therefore available for other uses.[16] While it is suggested that they are reserved for use by bootable media,[17] a CD-ROM may contain an alternative file system descriptor in this area, and it is often used by hybrid CDs to offer classic Mac OS-specific and macOS-specific content.

The data area begins with the volume descriptor set, a set of one or more volume descriptors terminated with a volume descriptor set terminator. These collectively act as a header for the data area, describing its content (similar to the BIOS parameter block used by FAT, HPFS and NTFS formatted disks).

Each volume descriptor is 2048 bytes in size, fitting perfectly into a single Mode 1 or Mode 2 Form 1 sector. They have the following structure:

The data field of a volume descriptor may be subdivided into several fields, with the exact content depending on the type. Redundant copies of each volume descriptor can also be included in case the first copy of the descriptor becomes corrupt.

Standard volume descriptor types are the following:

An ISO 9660 compliant disc must contain at least one primary volume descriptor describing the file system and a volume descriptor set terminator for indicating the end of the descriptor sequence. The volume descriptor set terminator is simply a particular type of volume descriptor with the purpose of marking the end of this set of structures. The primary volume descriptor provides information about the volume, characteristics and metadata, including a root directory record that indicates in which sector the root directory is located. Other fields contain the description or name of the volume, and information about who created it and with which application. The size of the logical blocks which the file system uses to segment the volume is also stored in a field inside the primary volume descriptor, as well as the amount of space occupied by the volume (measured in number of logical blocks).

In addition to the primary volume descriptor(s), supplementary volume descriptors or enhanced volume descriptors may be present. Supplementary volume descriptors describe the same volume as the primary volume descriptor does, and are normally used for providing additional code page support when the standard code tables are insufficient. The standard specifies that ISO 2022 is used for managing code sets that are wider than 8 bytes, and that ISO 2375 escape sequences are used to identify each particular code page used. Consequently, ISO 9660 supports international single-byte and multi-byte character sets, provided they fit into the framework of the referenced standards. However, ISO 9660 does not specify any code pages that are guaranteed to be supported: all use of code tables other than those defined in the standard itself are subject to agreement between the originator and the recipient of the volume. Enhanced volume descriptors were introduced in ISO 9660, Amendment 1. They relax some of the requirements of the other volume descriptors and the directory records referenced by them: for example, the directory depth can exceed eight, file identifiers need not contain '.' or file version number, the length of a file and directory identifier is maximized to 207.

Path tables summarize the directory structure of the relevant directory hierarchy. For each directory in the image, the path table provides the directory identifier, the location of the extent in which the directory is recorded, the length of any extended attributes associated with the directory, and the index of its parent directory path table entry. The parent directory number is a 16-bit number, limiting its range from 1 to 65,535.[18]

Overview of the ISO 9660 directory structure

Directory entries are stored following the location of the root directory entry, where evaluation of filenames is begun. Both directories and files are stored as extents, which are sequential series of sectors. Files and directories are differentiated only by a file attribute that indicates its nature (similar to Unix). The attributes of a file are stored in the directory entry that describes the file, and optionally in the extended attribute record. To locate a file, the directory names in the file's path can be checked sequentially, going to the location of each directory to obtain the location of the subsequent subdirectory. However, a file can also be located through the path table provided by the file system. This path table stores information about each directory, its parent, and its location on disc. Since the path table is stored in a contiguous region, it can be searched much faster than jumping to the particular locations of each directory in the file's path, thus reducing seek time.

The standard specifies three nested levels of interchange (paraphrased from section 10):

• Level 1: File names are limited to eight characters with a three-character extension. Directory names are limited to eight characters. Files may contain one single file section.
• Level 2: Files may contain one single file section.
• Level 3: No additional restrictions than those stipulated in the main body of the standard. That is, directory identifiers may not exceed 31 characters in length, and file name + '.' + file name extension may not exceed 30 characters in length (sections 7.5 and 7.6). Files are also allowed to consist of multiple non-contiguous sections (with some restrictions as to order).

Additional restrictions in the body of the standard: The depth of the directory hierarchy must not exceed 8 (root directory being at level 1), and the path length of any file must not exceed 255. (section 6.8.2.1).

The standard also specifies the following name restrictions (sections 7.5 and 7.6):[4]

• All levels restrict file names in the mandatory file hierarchy to upper case letters, digits, underscores ("_"), and a dot. (see also section 7.4.4 and Annex A),
• If no characters are specified for the File Name then the File Name Extension shall consist of at least one character.
• If no characters are specified for the File Name Extension then the File Name shall consist of at least one character.
• File names shall not have more than one dot.
• Directory names shall not use dots at all.

A CD-ROM producer may choose one of the lower Levels of Interchange specified in chapter 10 of the standard, and further restrict file name length from 30 characters to only 8+3 in file identifiers, and 8 in directory identifiers in order to promote interchangeability with implementations that do not implement the full standard.

All numbers in ISO 9660 file systems except the single byte value used for the GMT offset are unsigned numbers. As the length of a file's extent on disc is stored in a 32 bit value,[19] it allows for a maximum length of just over 4.2 GB (more precisely, one byte less than 4 GiB). It is possible to circumvent this limitation by using the multi-extent (fragmentation) feature of ISO 9660 Level 3 to create ISO 9660 file systems and single files up to 8 TiB. With this, files larger than 4 GiB can be split up into multiple extents (sequential series of sectors), each not exceeding the 4 GiB limit. For example, the free software such as InfraRecorder, ImgBurn and mkisofs as well as Roxio Toast are able to create ISO 9660 file systems that use multi-extent files to store files larger than 4 GiB on appropriate media such as recordable DVDs. Linux supports multiple extents.[20]

System Use Sharing Protocol (SUSP, IEEE P1281) provides a generic way of including additional properties for any directory entry reachable from the primary volume descriptor (PVD). In an ISO 9660 volume, every directory entry has an optional system use area whose contents are undefined and left to be interpreted by the system. SUSP defines a method to subdivide that area into multiple system use fields, each identified by a two-character signature tag. The idea behind SUSP was that it would enable any number of independent extensions to ISO 9660 to be created and included on a volume without conflicting. It also allows for the inclusion of property data that would otherwise be too large to fit within the limits of the system use area.

SUSP defines several common tags and system use fields:

• CE: Continuation area
• PD: Padding field
• SP: System use sharing protocol indicator
• ST: System use sharing protocol terminator
• ER: Extensions reference
• ES: Extension selector

Other known SUSP fields include:

• AA: Apple extension, preferred
• BA: Apple extension, old (length attribute is missing)
• AS: Amiga file properties
• ZF: zisofs compressed file, usually produced by program mkzftree or by libisofs. Transparently decompressed by Linux kernel if built with CONFIG_ZISOFS.[21]
• AL: records Extended File Attributes, including ACLs. Proposed by liburnia, supported by libisofs.[22]

The Apple extensions do not technically follow the SUSP standard; however the basic structure of the AA and AB fields defined by Apple are forward compatible with SUSP; so that, with care, a volume can use both Apple extensions as well as RRIP extensions.

The Rock Ridge Interchange Protocol (RRIP, IEEE P1282) is an extension which adds POSIX file system semantics. The availability of these extension properties allows for better integration with Unix and Unix-like operating systems.[23] The standard takes its name from the fictional town Rock Ridge in Mel Brooks' film Blazing Saddles.[24] The RRIP extensions are, briefly:

• Longer file names (up to 255 bytes) and fewer restrictions on allowed characters (support for lowercase, etc.)
• UNIX-style file modes, user ids and group ids, and file timestamps
• Support for Symbolic links and device files
• Deeper directory hierarchy (more than 8 levels)
• Efficient storage of sparse files

The RRIP extensions are built upon SUSP, defining additional tags for support of POSIX semantics, along with the format and meaning of the corresponding system use fields:

• RR: Rock Ridge extensions in-use indicator (note: dropped from standard after version 1.09)
• PX: POSIX file attributes
• PN: POSIX device numbers
• SL: symbolic link
• NM: alternate name
• CL: child link
• PL: parent link
• RE: relocated directory
• TF: time stamp
• SF: sparse file data

Amiga Rock Ridge is similar to RRIP, except it provides additional properties used by AmigaOS. It too is built on the SUSP standard by defining an "AS"-tagged system use field. Thus both Amiga Rock Ridge and the POSIX RRIP may be used simultaneously on the same volume. Some of the specific properties supported by this extension are the additional Amiga-bits for files. There is support for attribute "P" that stands for "pure" bit (indicating re-entrant command) and attribute "S" for script bit (indicating batch file). This includes the protection flags plus an optional comment field. These extensions were introduced by Angela Schmidt with the help of Andrew Young, the primary author of the Rock Ridge Interchange Protocol and System Use Sharing Protocol. The first publicly available software to master a CD-ROM with Amiga extensions was MakeCD, an Amiga software which Angela Schmidt developed together with Patrick Ohly.[25]

El Torito is an extension designed to allow a computer to boot from a CD-ROM. It was announced in November 1994[26] and first issued in January 1995 as a joint proposal by IBM and BIOS manufacturer Phoenix Technologies. According to legend, the El Torito CD/DVD extension to ISO 9660 gained its name because its design originated in an El Torito restaurant in Irvine, California (33.684722°N 117.852547°W / 33.684722; -117.852547).[27] This is supported by a claim by Jack Allweiss, founder of Future Domain Inc. who states that Future Domain helped establish the standard.[28] The initial two authors were Curtis Stevens, of Phoenix Technologies, and Stan Merkin, of IBM.[27]

A 32-bit PC BIOS will search for boot code on an ISO 9660 CD-ROM. The standard allows for booting in two different modes. Either in hard disk emulation when the boot information can be accessed directly from the CD media, or in floppy emulation mode where the boot information is stored in an image file of a floppy disk, which is loaded from the CD and then behaves as a virtual floppy disk. This is useful for computers built before about 1999, which were designed to boot only from floppy drive. For modern computers the "no emulation" mode is generally the more reliable method. The BIOS will assign a BIOS drive number to the CD drive. The drive number (for INT 13H) assigned is any of 80_hex (hard disk emulation), 00_hex (floppy disk emulation) or an arbitrary number if the BIOS should not provide emulation. Emulation allows older operating systems to be booted from a CD, by making it appear to them as if they were booted from a hard or floppy disk.

El Torito can also be used to produce CDs which can boot up Linux operating systems, by including the GRUB bootloader on the CD and following the Multiboot Specification.[29] While the El Torito spec alludes to a "Mac" platform ID, PowerPC-based Apple Macintosh computers don't use it.[30]

Joliet is an extension specified and endorsed by Microsoft and has been supported by all versions of its Windows operating system since Windows 95[31] and Windows NT 4.0.[32] Its primary focus is the relaxation of the filename restrictions inherent with full ISO 9660 compliance. Joliet accomplishes this by supplying an additional set of filenames that are encoded in UCS-2BE (UTF-16BE in practice since Windows 2000). These filenames are stored in a special supplementary volume descriptor, that is safely ignored by ISO 9660-compliant software, thus preserving backward compatibility.[33] The specification only allows filenames to be up to 64 Unicode characters in length. However, the documentation for mkisofs states filenames up to 103 characters in length do not appear to cause problems.[34] Microsoft has documented it "can use up to 110 characters."[35]

Joliet allows Unicode characters to be used for all text fields, which includes file names and the volume name. A "Secondary" volume descriptor with type 2 contains the same information as the Primary one (sector 16 offset 40 bytes), but in UCS-2BE in sector 17, offset 40 bytes. As a result of this, the volume name is limited to 16 characters.

Many current PC operating systems are able to read Joliet-formatted media, thus allowing exchange of files between those operating systems even if non-Roman characters are involved (such as Arabic, Japanese or Cyrillic), which was formerly not possible with plain ISO 9660-formatted media. Operating systems which can read Joliet media include:

• Microsoft Windows;[31] Microsoft recommends the use of the Joliet extension for developers targeting Windows.[36]
• Linux[37]
• macOS[38]
• FreeBSD[39]
• OpenSolaris[40]
• Haiku[41]
• AmigaOS
• RISC OS

Romeo was developed by Adaptec and allows the use of long filenames up to 128 characters. However, Romeo is not backwards compatible with ISO 9660 and discs authored using this file system can only be read under the Windows 9x and Windows NT platforms, thus not allowing exchange of files between those operating systems if non-Roman characters are involved (such as Arabic, Japanese or Cyrillic), for example ü becomes ³.[42]

Apple Computer authored a set of extensions that add ProDOS or HFS/HFS+ (the primary contemporary file system for Mac OS) properties to the filesystem. Some of the additional metadata properties include:[43]

• Date of last backup
• File type
• Creator code
• Flags and data for display
• Reference to a resource fork

In order to allow non-Macintosh systems to access Macintosh files on CD-ROMs, Apple chose to use an extension of the standard ISO 9660 format. Most of the data, other than the Apple specific metadata, remains visible to operating systems that are able to read ISO 9660.

For operating systems which do not support any extensions, a name translation file TRANS.TBL must be used. The TRANS.TBL file is a plain ASCII text file. Each line contains three fields, separated by an arbitrary amount of whitespace:

• The file type ("F" for file or "D" for directory);
• The ISO 9660 filename (including the usually hidden ";1" for files); and
• The extended filename, which may contain spaces.

Most implementations that create TRANS.TBL files put a single space between the file type and ISO 9660 name and some arbitrary number of tabs between the ISO 9660 filename and the extended filename.

Native support for using TRANS.TBL still exists in many ISO 9660 implementations, particularly those related to Unix. However, it has long since been superseded by other extensions, and modern utilities that create ISO 9660 images either cannot create TRANS.TBL files at all, or no longer create them unless explicitly requested by the user. Since a TRANS.TBL file has no special identification other than its name, it can also be created separately and included in the directory before filesystem creation.

The ISO 13490 standard is an extension to the ISO 9660 format that adds support for multiple sessions on a disc. Since ISO 9660 is by design a read-only, pre-mastered file system, all the data has to be written in one go or "session" to the medium. Once written, there is no provision for altering the stored content. ISO 13490 was created to allow adding more files to a writeable disc such as CD-R in multiple sessions.

JIS X 0606:1998, also known as ISO 9660:1999, is a Japanese Industrial Standard draft created by the Japanese National Body (JTC1 N4222) in order to make some improvements and remove some limitations from the original ISO 9660 standard.[44] This draft was submitted in 1998, but it has not been ratified as an ISO standard yet. Some of its changes includes the removal of some restrictions imposed by the original standard by extending the maximum file name length to 207 characters, removing the eight-level maximum directory nesting limit, and removing the special meaning of the dot character in filenames. Some operating systems allow these relaxations as well when reading optical discs. Several disc authoring tools (such as Nero Burning ROM, mkisofs and ImgBurn) support a so-called "ISO 9660:1999" mode (sometimes called "ISO 9660 v2" or "ISO 9660 Level 4" mode) that removes restrictions following the guidelines in the ISO 9660:1999 draft.

The ISO 13346/ECMA-167 standard was designed in conjunction to the ISO 13490 standard. This new format addresses most of the shortcomings of ISO 9660, and a subset of it evolved into the Universal Disk Format (UDF), which was adopted for DVDs. The volume descriptor table retains the ISO9660 layout, but the identifier has been updated.[45][46]