 |
Description  |
|
|
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to a file management system and more in detail to a
file management system, in which a plurality of workstations, each of
which includes a respective file, are linked through a network, and which
is suitable for a plurality of users to share distributed files.
2. Description of the Prior Art
A file management system, which enables a plurality of users to share files
distributedly stored in a plurality of memory devices, is disclosed e.g.
in USENIX 1986, Conference Proceedings, pp. 248-259.
FIG. 4A and 4B illustrate the concept of file sharing according to the
prior art. In FIG. 4A reference numerals 1a and 1b represent the structure
of files stored in disk devices on different workstations, node 1 and node
2. Each workstation independently manages its local files in the tree
structure form, and each structure of files 1a and 1b is equivalent to
what is viewed by users at each workstation when each workstation is used
as a stand-alone system. For example, a user at the node 1 views "A0",
called a root file, at the root position of the file structure, and three
files "A1","A2" and "A3" under "A0". These files are divided into two
category files. The content of the directory file is the information to
point other files and to construct tree structure. "A0", "B0", "B1" and
"B2" in FIG. 4A are examples of directory files. All files at leaf
positions of tree structure are not files in the other category except
said directory files. Since there exists a directory file which points no
file at present, the directory file may appear at leaf positions of tree
structure. "A1" and "A3" in FIG. 4A are the examples.
Files which are not directory files are in the other category. Generally
these files are called "ordinary files". They always appear at leaf
positions of tree structure. In FIG. 4A "A2", "B11", "B12", "B13", "B21"
and "B22" are the examples. Each of the files contains data base data, or
some object program. Arbitrary types of data may be stored in the files.
An user can read or write the files and if they contain object programs he
can also execute the files. There also exits a special file which
corresponds to a real device. For example, read or write request to a
special file corresponding to a terminal is treated as getting an input
from a keyboard or displaying image on a screen respectively.
A method for managing local files in each workstation is disclosed, e.g.,
in Communications of the ACM, Vol. 17, pp. 365-375, July 1974. FIG. 5
shows the method by using the example of the node 1 in FIG. 4A. To manage
local files, file control information 20a, directory information 21a and
file body 22a are stored in the local memory device 6a. File control
information 20a is a set of file control tables (30a, 30b, . . . )
corresponding to each file. Each file control table 30a consists of
following fields:
file type, owner, protection, file size and address etc.
The file type field 31a corresponding to the file "A0" indicates
"directory", and the file type field 31b corresponding to the file "A2"
indicates "normal (ordinary)". Each data in address fields 32a and 32b
indicates each address on the local memory device in which the content of
each the file resides. For example, data on address field 32 b indicates
that the content of the file "A2" resides in the area 40 in the file body
area 22a.
Directory information 21a is the content of directory files. Reference
numeral 40a shows the content of the directory file "A0". As stated above,
a directory file contains the information for each file which is pointed
by the directory file. Each information indicates what a name of the
pointed file is and where a file control table for the pointed file is. In
this example directory file "A0" contains the information for "A1", "A2"
and "A3". Directory information 21a is distinguished from the file body
22a in this figure, but it does not mean that directory information and
file body are physically separated.
Using the above method users can access any local file in each workstation
by specifying a path from the root file to it on the tree structure. For
example, the file "A2" (in FIG. 4A) can be specified with "A0/A2" from a
user at the node 1. Omitting "A0" and specifying only "A2" is allowable in
this example. A file control table for the root file "A0" is located at
the predetermined position and an operating system can trace the specified
path by reference file control tables and the contents of directory files
corresponding to the path.
The above method enables users only to access local files stored in a
workstation that the users are using, and the users cannot access remote
files in another workstation. The reference numeral 3a in FIG. 4A
represents a file link to extend the local file structure in node 1 to the
structure including remote files in the node 2. The file link 3a links a
subtree 2a, headed by a directory file "B2" in the structure of files 1b
in node 2, to a branch "A1" in the structure of files 1a in the node 1.
With this file link 3a, the file structure viewed from users at the node 1
becomes a new shape in which the subtree 2a is virtually attached under
the directory file "A1" as shown in FIG. 4B. In other words, the file link
3a has an effect that a user can directly access to the file "B2" in the
node 2. Therefore, users at the node 1 can specify and access the files
"B2" and "B21" in the subtree 2a with names "A1" and "A1/B21"
respectively.
FIG. 6 shows an embodiment of the file link. Before a file link is done,
definitions of linkable subtrees are made at the workstation in which the
subtrees reside. When a user program 12b makes a request of linkable
subtree definition by specifying the head file of the subtree, for
example, "B2" in FIG. 4A, the file system management program 90b searches
the specified head file by referring file control information 20b and
directory information 21h, and set the linkable subtree information to a
management table. The management table 70 has columns to indicate a
linkable subtree, accessibility to the subtree, etc. For example, the
information of pointing to a file control table for the specified head
file is registered in the column 71 to specify the linkable subtree.
Subtrees defined the above procedure are linkable from other workstations.
For example, when a user program 12a at the node 1 makes a file link
request, as shown in FIG. 4A, the file system management program 90a
searches a destination file "A1" by referring file control information 20a
and directory information 21a, and if found, sends a message to the file
system management program 90a that a subtree headed by "B2" is requested
to be linked, and waits for the answer from the file management program
90b. When the management program 90b receives the message, it checks the
management table 70 and sends back the answer to the management program
90a, whether the request is granted or not. Receiving the answer, the
management program 90a sets the file link information to a management
table 60 if the request is granted, and answers the result to the user
program 12a. The management table 60 has a column 61 for storing
information indicating a destination file in the workstation, a column 62
for storing information indicating an original file to be linked in
another workstation and a column 63 for storing information indicating the
accessibility, etc. As for the column 61, for example, the information
pointing to a file control table for a for a destination is used. As for
the column 62, the information indicating a workstation in which an
original file resides should be included as well as the information
pointing a file control table for the file.
After the file link has been done, a file access request from a user
program is treated as follows. When a user program 12a specified a file to
be accessed, the file system management program 90a searches along the
file tree structure by tracing a path name of the file one by one from a
root file "A0". At each point of the tree, the management program 90a
refers the management table 60 first. If there is no entry indicating said
accessed file in link destination column 61 at the position that the
management program 90a is currently processing, the management program 90a
advances next position in the tree structure by referring file control
information 20a and directory information 21a. If there is an entry
indicating said accessed file, the management program 90a sends a file
access message to a file system management program in another workstation
in which the subtree indicated by an original file column of the matched
entry resides, and waits for the answer. The file access message includes
the following information:
1. the original file indicated by the matches entry; and
2. the remaining path name of the specified file which is not processed by
the management program 90a.
When a file system management program in the another workstation receives
the file access message, it searches the requested file from the position
of the specified original file with the remaining path name passed by the
message, and sends the answer. The algorithm of management program in the
another workstation is the same as that in the management program 90a.
When the answer is received, the management program 90a returns the result
to the user program 12a. For example, when the user program 12a makes a
file access request of "A2", the request is locally processed by the
management program 90a. When the user program 12a makes a file access
request of "A1/B21", however, the management program 90a finds at the
position of "A1" that a file access message should be sent to a management
program 90b. The message that the original file is "B2" and that the
remaining path name is "B21". Receiving the message, the management
program 90b starts accessing "B21" through "B2", and sends back the answer
when the file management program 90b finds file "B21". In the above
explanation, a user program which accesses a remote file by using the file
link is the same program, but they may be different programs.
According to the conventional system described above, there is a problem
that the extension of the file structure by the link of file structures
has influence on all the users utilizing the same workstation. For
example, in the case where a plurality of users utilize a workstation, if
one of them effects the extension of the file structure, the file
structure viewed from the other users varies without their knowing it.
Further, in the case where a plurality of users utilize successively a
workstation, if a preceding user has extended the file structure and
leaves it as it is, this gives rise to a problem that the succeeding user
utilizes the extended file.
SUMMARY OF THE INVENTION
The principal object of this invention is to provide a file management
system permitting the construction of a file system belonging to only one
user without having any influence on other users by linking a plurality of
information processing devices provided with files of tree structure
through a communication line and by using an information processing
device.
In order to achieve the above object, the file management system according
to this invention comprises a plurality of data processing devices, each
of said data processing devices having file memory for storing some
ordinary files (e.g., data base data file program file, etc.) and
directory files for managing said ordinary files in tree structure;
linkage memory for storing linkage relation between a directory file
stored in each data processing device and a directory file stored in
another data processing device according to an user or a user program
(e.g., data base access program,) in response to a linkage request from
said user or said user program. In addition, each of data processing
devices has a central processing unit for judging, in response to a
request to access a file directed by the directory file stored in another
data processing device through the directory file in the data processing
device from a user or a user program, whether the data processing device
should request another data processing device to access the file directed
by the directory file in the other data processing device by checking
whether said directory file stored in the data processing device is linked
with the directory file stored in the other data processing device
according to the requesting user or user program.
In addition, the other data processing device comprises means for storing
an indicator which indicates a count of the number of users or user
programs accessing the file directed by the directory file stored in the
other data processing device through the directory file stored in the data
processing device by using the linkage relation, so that the linkage
relation is automatically deleted when the indicator indicates a
predetermined number (for example zero).
According to this invention, since the extension of the file structure
based on the group specification (according to a user or user program)
described above is perfectly invalid for the members outside of the group,
it is possible to define freely the file structure without influencing no
one other than the relating members of the group. Further, even if a user
has left the release operation for the link undone, since the file
structure becomes automatically invalid at the moment of termination of
the processing for all the group members, it has no bad influence on
succeeding processings.
The above-mentioned and other features and objects of this invention will
become more apparent by reference to the following description taken in
conjunction with the accompanying drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a scheme illustrating the construction of a signal processing
system, in which a plurality of workstations, where this invention is
realized, are linked through a communication line;
FIGS. 2A and 2B are schemes for showing conceptionally the link between
files for explaining the file management system according to this
invention;
FIG. 3 is a scheme illustrating the construction of an embodiment of the
file management system according to this invention;
FIGS. 4A and 4B are schemes for explaining conceptionally the link of files
in a prior art file management system.
FIG. 5 is a scheme for explaining a method to manage local files in each
workstation in said prior art file management system; and
FIG. 6 is a scheme for explaining a linking operation between two
workstations in said prior art file management system.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
FIG. 1 is a scheme illustrating the construction of an information
processing system, in which the file management system according to this
invention is realized.
In FIG. 1, reference numerals 4a-4c' represents information processing
devices, which are connected with a communication line 8 in common through
communication devices 7a-7c. Each of the information processing devices
4a-4c includes a memory device 6a-6c and thus files are stored in a
distributed form under control of a file system management program
provided in each of the central processing units 5a-5c. The stored files
contain several type files, for example, the files contain directory files
and normal files like files of data base or program files such as the
prior art described above. A file system management program communicates
with the file system management program in another information processing
device 4a-4c through the communication devices 7a-7c and the communication
line 8, and utilizes the files distributedly stored in memory devices
6a-6c of the other information processing devices 4a-4c. Users of each of
the information processing devices 4a-4c can extend virtually the
structure of files stored in the local memory device 6a-6c provided in
each of the information processing devices by using the communication
between file system management programs described above and access the
file stored in another information processing device 6a-6c.
FIGS. 2A and 2B are schemes for explaining conceptually the functions of
linking and extending files in the file management system according to
this invention.
In FIG. 2A, reference numerals 1a-1c represent the structures of files
which are stored in the memory devices 6a-6c, respectively, indicated in
FIG. 1, which are examples of the file structure visible as the initial
state from users of each of the information processing devices 4a-4c. Now,
taking the file structure viewed by a user of the information processing
device 4a as an example, it is explained for the case, where the file
structure is virtually extended by linking subtrees 2a-2c in the structure
of files 1b and 1c stored in the other information processing devices 4b
and 4c to a part of the structure of files la stored in the information
processing device 4a.
In FIG. 2 characters including an underline, for example "A2",
"B11"...etc., signify files of data base data or files of programs,
characters including no underline, for example "A1", "A3", ...etc.,
signify directory files like the above-mentioned prior art.
The line indicated by reference numeral 3a in FIG. 2A means that the
subtree 2a under B.sub.2 in the structure of files 1b is linked to the
branch A.sub.1 in the structure of files 1a. As a result the subtree 2a is
attached virtually under the branch A.sub.1. The line 3a described above
shows a link without group specification and any user of the information
processing device 1a can utilize the subtree 2a as if it is attached under
branch A.sub.1.
The broken lines indicated by reference numerals 3b and 3c in the figure
mean that the subtree 2b and 2c in the structure of files 1b and 1c,
respectively, are linked to the branch in the structure of files 1a. These
differ from the link 3a in that at the link specification for the former
the group is specified for restricting the valid range of the link. For
example, a group X is specified for the link 3b and another group Y is
specified for the link 3c. In the case of the link 3a stated above, the
result of the link has influence on all the users of the information
processing device 1a. On the contrary, when the group is specified as for
the link 3b and 3c, the link of the branch A.sub.3 with the subtree 2b or
that of the file A.sub.3 with the subtree 2c is valid only for the users
belonging to the group X or Y, as indicated in FIG. 2B. Consequently, when
the group is specified, it is possible to link subtrees to a branch
A.sub.3 which are different for every group. Here a group means an
assembly of users or an assembly of a plurality of programs for a series
of works. It is also possible to define each of users as an independent
group. Further, concerning the subtree, which will be an object of the
link, it may be prevented that the files in the subtree are utilized from
another without permission e.g. by declaring on the information processing
device including the relevant subtree that any link thereof with another
information processing device is not permissible
Next, an embodiment of the file management indicated in FIGS. 2A and 2B
will be explained below, referring to FIG. 3. In FIG. 3, reference
numerals 4a-4c represent information processing devices, each of which is
provided with a file system management program 9a-9c, user programs
12x-12z, 12b, 12c, and a management table 10, 11b, 11c. Now it is supposed
that users of the information processing devices 4b and 4c send previously
generated demands to the file system management programs 9b and 9c by
using the programs 12b and 12c, respectively in order to define the
subtrees 2a, 2b and 2c in FIG. 2A as linkable ones. The file system
management program 9b and 9c of the devices 4b and 4c register information
indicating their respective linkable subtrees in the management tables 11b
and 11c, respectively. Each of these tables consists of columns for
storing information indicating linkable subtrees, groups which can access
the defined subtrees (the entry "--" in this column means that all the
users can access) and a column for storing information indicating whether
it is writable. In these columns the information is set by the
specification from the programs 12b and 12c described above is set.
Further, in the table stated above there is a column for storing the
number of references indicating how many information processing devices
refer to the defined subtree. This column is cleared to zero in the
initial state. In the case where the subtree defined as a linkable file is
not referred to by another information processing device, i.e. in the case
where the number of references is zero, the link permission may be deleted
and the table may be amended so.
When the registration in the control tables 11b and 11c is terminated, the
specification processing of the link relation is effected in the following
way. For example, when the user program within the information processing
device 4a demands the file system management program 9a to link the
subtree 2b in FIG. 2A with A.sub.3, the file system management program 9a
effects an access right check to the branch A.sub.3 and thereafter send a
demand to link the subtree 2b to the file system management program 9b of
the information processing device 4b including the subtree 2b. The file
system management program 9b effects the access right check of the groups,
etc. according to the demand stated above. If there are no problems, it
increases the number of references of the entry corresponding to the
subtree 2b of the management table 11b by 1 and thereafter replies to the
file system management program 9a that the link has been allowed The file
system management program 9a, which has received the reply of permission,
prepares an entry (information) indicating the file link relation in the
management Table 10 on the basis of the information coming from the user
program 12x. This Table 10 consists of columns for storing information
indicating the branch position of the link destination, information
indicating the subtree with which it is to be linked, the group for which
this link is valid (the entry "--" in this column means the link is valid
for all the users), information indicating whether it is writable or not,
and the number of members which are using the relevant link at present.
The last mentioned column, say, the number of members can be used in such
a way as it is prevented that the link is deleted during the reference of
the file and at the same time, when the file is no more referred to, it is
possible to invalidate automatically the relevant entry. Management
similar to that described above for the subtree 2b is effected also for
the other subtrees 2a and 2c.
Now the management system will be explained for the case where a user
program accesses a file in the subtree linked. Here it is supposed that
the user program 12y demands an access to a file "C.sub.2 " specified
with" A.sub.3 /C.sub.12 (" / " represents a delimiter to separate the
branch or file names specifying the path from root in the tree structure
for each file) For this, file system management program searches the file
in the managed file structure from the root and verifies whether the
branch A.sub.3 exists within the management Table 10 or not. In this case,
since the group, to which user program 12y demanding the access belongs,
is Y, it is judged that the second entry within the management Table 10 is
invalid and the third entry is valid and thus the link with C.sub.1 is
permitted. Then the file management system program system program 9a
demands an access to C.sub.12, which is a subtree under C.sub.1, to the
file system management program 9c of the information processing device 4c.
The file system management program 9c, which has received this demand,
refers to the control Table 11c and carries out the demand after having
checked the access right and returns the result to the file system
management program 9a of the demander. The file system management program
9a, which has received the result, replies it to the user program 12y and
terminates the access demand processing Accordingly, the user program 12y
can treat the file " C12" within the processing device 4c as a file
directed by the directory file "A3". In the case where another user
program 12z demands an access to the same file "C.sub.12 " specifying
with" A.sub.3 /C.sub.12, since there is no entry of A.sub.3 corresponding
to group Z within the management Table 10, the search of file A.sub.3
/C.sub.12 within the memory device 6a (in FIG. 1) of the information
processing device 4a finally results in that it is an undefined file and
the access is treated as an access error. In the case where the user
program 12x tries to access the file "C.sub.12 " specifying with A.sub.3
/C.sub.12, too, since it is judged that C.sub.12 is in the subtree under
B.sub.1 on the management Table 10 and an access is demanded to the file
system management program 9b, no relevant file is found and the access is
treated similarly as an access error. On the other hand, concerning the
subtree 2a in FIG. 2A, which is linked for all the users in common,
considering e.g. an access to the file A.sub.1 /B.sub.21, since "--"
stands in the column for the group specification of the first entry within
the management Table 10, an access demand from either one of the programs,
12x, 12y and 12z is transmitted to the file system management program 9b
and treated normally.
The setting of the file groups may be effected by the management with pass
word e.g. by using a group identifier management file, etc. in the same
way as user identifiers or user group identifiers. Further, in the case
where the file access is allowed for a program executing a series of
works, for example, a system call for setting a unique group identifier to
these series of works is prepared.
According to the embodiment described above, since the file system 1a, 1b,
1c on each of the information processing devices is managed in an
independent and equalized form under the system management program, even
if an information processing device breaks down, an advantage can be
obtained that accesses to the file system of other devices are performed
normally.
According to this invention, when a file structure is virtually extended by
linking a part of another file structure distributedly managed in other
information processing devices with a local file structure, it is possible
to specify whether the extension is valid for all the users or it is valid
only for specified groups. Consequently not only the file structure can be
extended so that all the users can utilize it in common, but also a file
structure, which is utilized uniquely by a group, can be virtually
constructed without having any influence on the others than the users
concerned.
* * * * *
|
|
 |
|
 |
Description |
|
