The logical organization of the file system
File Attributes The concept of file includes not only the stored data and the
name and attributes. Attributes - that is, information that describes the
properties of the file. Examples of possible file attributes: file type (regular
file, directory, special file, etc.); the owner of the file; creator of the file;
a password for access to the file; Information on the operations allowed access
to the file; time of creation, last access and the latest developments; the
current size of the file; Maximum file size; o o sign read only; sign hidden
file; sign system file; sign archive file; sign binary / character; sign interim
(removed after the completion of the process); token lock; the length of an
entry in the file; a pointer to the key field in the record; key length. Set the
file attributes defined by the specifics of the file system: the file systems of
different types to describe the files can be used different sets of attributes.
For example, the file systems that support unstructured files, no need to use
the last three in the list of attributes associated with structuring the file.
In a single operating system in a set of attributes will not have
characteristics that are relevant to users and protection, such as file owner,
the creator of the file, the password to access the file, information about how
to access the file. Users can gain access to the attributes, using funds
provided for these purposes, the file system. Usually allowed to read the value
of any attribute, and change - just a few. For example, the user can change the
file access rights (provided that it has authorized to do so), but change the
date of creation or the current size of the file it is not permitted. The values
of the attributes of files can be directly found in catalogs, as in the file
system, MS-DOS The figure represented an entry in the directory structure
containing a mere symbolic name and file attributes. Here, the letters are
marked signs of the file: R - read-only, A - Archive H - hidden, S - system.
Another option is to place the attributes of special tables when directories
contain only links to the tables. This approach is realized, for example, a file
system ufs OS UNIX. In this file system directory structure is very simple.
Record of each file contains a short character file name and a pointer to inodes
file, so named ufs table, which focused the value of the file attributes In
either version of the catalog provides a link between file names and the actual
files. However, an approach where the file name is separated from its attributes,
makes the system more flexible. For example, the file can be easily incorporated
in several directories. The record of this file in different directories can
contain different simple names, but in the links will contain the same number of
inodes. The logical organization of the file In general, the data contained in
the file, have a certain logical structure. This structure is the basis for
policy, designed to process the data. For example, the text can be properly
displayed, the program should be able to isolate specific words, lines,
paragraphs, etc. The signs separating one from another building block, can be a
certain code sequence or a well-known program offset values of these structural
elements relative to the beginning of the file. Maintenance of the data
structure can be either entirely the responsibility of the application, or in
varying degrees, this work can take on the file system. In the first case, where
all actions related to the structuring and interpretation of the contents of the
file entirely the responsibility of the application file to the FS sequence of
unstructured data. Annex makes requests to the file system on the input-output
using common to all applications, system tools, for example, by specifying the
offset from the beginning of the file and the number of bytes that need to be
considered or write. Received by the application stream of bytes is interpreted
in accordance with the framework of the program logic. For example, the compiler
generates, and the editor of the links takes quite some object module format of
the program. At the same file format, which stores the object module is known
only to those programs. Emphasize that the interpretation of the data is not
linked with a valid way to store them in the file system. Model file, in which
the contents of the file to an unstructured sequence (stream) of bytes became
popular along with the operating system UNIX, and now it is widely used in most
modern operating systems, including MS-DOS, Windows NT/2000, NetWare.
Unstructured model file allows easy sharing of files among multiple applications:
different applications may in its own structure and interpret the data contained
in the file. Another model is a file that was used in the operating system OS/360,
DEC RSX and VMS, and currently is used very often - is a structured file. In
this case, the maintenance of the structure of a file is requested the file
system. A file system sees a file as an ordered sequence of logical records. An
application can apply to the Federal Assembly with requests for input-output-level
records, such as take record 25 in the file FILE.DOC. FS should have information
about the structure of the file, sufficient to identify any record. FS annex
provides access to the records, and all further processing of the data contained
in the record, the application. The development of this approach become a
database management system (DBMS) that supports not only a complex structure,
but also the relationship between them. The logical record is the smallest
element of data that can operate a programmer at the exchange with an external
device. Even if the physical exchange of the device being more than one
operating system to provide the programmer access to a single logical record. A
file system can use two ways to access the logical records: read or write
logical records sequentially (serial access) or to position the file on record
with a specified number (direct access). Obviously, the OS may not support all
possible ways to structure the data in the file, so in the operating system, in
which there is general support for the logical structuring of files, it is for a
small number of widespread schemes logical organization of the file. These
include ways to structure the true representation of data in the form of records,
the length of which is fixed in the files in this case, access to the n-th
recording is carried out either by reading the serial (n-1) prior to the records,
either directly at the address, calculated by its serial number . For example,
if L - length of the recording, the start address of n-th record is equal to L *
n. Note that this logical organization of the size of recording fixed within a
file and write in different files that belong to the same file system, may have
different sizes. Another way is to structure the data in the form of a sequence
of records, which varies in a single file. If the distance between the values of
lengths of records as shown in Fig. 7.7, b, to find specific records system
should consistently take all previous records. Calculate the address of its
record number in such a logical organization of the file is not possible, and
therefore can not be applied more efficient method of direct access. The files,
access to records which are consistently on the numbers of positions, called
neindeksirovannymi or consistent.
Another type of files are indexed files, they allow more rapid access to a
single logical record. In the indexed file records have one or more key (index)
of fields and can be addressed by specifying the values of these fields. For
quick retrieval of data in the Index provides a special index table, in which
the values of key fields is in line address external memory. This address can
point directly to the desired entry, or some areas of external memory, takes
only a few recordings, which include the required entry. In the latter case,
saying that the file is indexed posledovatelnuyu organization, as well as search
includes two parts: direct access to an index of the region drive, and then view
the records in this region. Keeping index table assumes the file system. It is
clear that the entries in the index files can be of any length. All of this is
more applicable to ordinary files, which can be as structured or unstructured.
As for other types of files, they have some structure, the well-known file
system. For example, the file system should understand the structure of data
stored in a file directory or file type symbolic link
|