Managing Processes In UNIX and Linux

The process identifier ( PID ) is unique throughout the system and is used to
access this process. The init process identifier is always one. The process
ID of the ancestor ( ppid ) is specified when it is created. If the ancestor of
process P completes execution, the ancestor of that process automatically
becomes init, so ppid for P will be unity.
With the process as fl ' related set of security attributes.
❖ 
The real user and process group identifiers ( uid , gid ) correspond
to the user who started the program, resulting in the corresponding
process appearing on the system
❖ 
Effective user and process group identifiers ( euid , egid ) are used
in a special process execution mode - are owned by the owner
Process Control Unit
Consider the structure of the process control unit in Linux.
The process control block on Linux is displayed by the task _ struct data
structure. The most important fields in this structure are fields that contain
the following information:
❖ 
Identification data (including pid - process identifier)
❖ 
Process status (execution, expectations, etc.)
❖ 
Pointers to ancestral and descendant structures
❖ 
Process creation time and total execution time (so-called process
timers)
❖ 
Processor state (contents of registers and counter of instructions)
❖ 
Process security attributes ( id , gid, euid , egid )
Note that the Linux kernel does not have a separate data structure for the
stream, so the processor status information is contained in the control block
of the process. In addition to the above, task _ struct has several custom
fields required for different Linux subsystems :

❖ 
Information for signal processing
❖ 
Process planning information
❖ 
About files and directories floor ' related to the process
❖ 
Data structure for managing memory subsystems
The task _ struct field data can be shared by several special purpose
processes, in which case these processes are actually threads. The control
units of the process are stored in the kernel in a special data structure. Prior
to the release of kernel 2.4, this structure was called a system process
table; it was an array whose maximum length could not exceed 4K. In the
core processes, it has been replaced by two dynamic data structures that do
not have such a restriction:
❖ 
A hash table (where the pid of the process acts as a hash ), this
structure allows you to quickly find the process by its ID
❖ 
Ring list, this structure ensures the implementation of actions in the
loop for all system processe s
Now the limit on the maximum number of processes is checked only within
the implementation of the fork () function and depends on the amount of
available memory (for example, there is information that on a system with
512 MB of memory it is possible to create about 32000 processes).
The implementation of a control unit in Linux is different from its
traditional implementation in UNIX systems. In most versions
of UNIX (System V, BSD), the process control block consists of two data
layers - a process structure ( proc ) and a user structure ( u ).

Download 6,34 Mb.