• Choose a category
• All
• Classic-Fiction

Processes

At the heart of Unix lies the concept of a process. Understanding this concept will help you keep control of your login session as a user. If you are also a system administrator, the concept is even more important.

A process is an independently running program that has its own set of resources. For instance, we showed in an earlier section how you could direct the output of a program to a file while your shell continued to direct output to your screen. The reason that the shell and the other program can send output to different places is that they are separate processes .

On Unix, the finite resources of the system, such as the memory and the disks, are managed by one all-powerful program called the kernel. Everything else on the system is a process.

Thus, before you log in, your terminal is monitored by a getty process. After you log in, the getty process dies (a new one is started by the kernel when you log out) and your terminal is managed by your shell, which is a different process. The shell then creates a new process each time you enter a command. The creation of a new process is called forking because one process splits into two.

If you are using the X Window System , each process starts up one or more windows. Thus, the window in which you are typing commands is owned by an xterm process or a reloaded terminal program. That process forks a shell to run within the window. And that shell forks yet more processes as you enter commands.

To see the processes you are running, enter the command ps. Figure 10-2 shows some typical output and what each field means. You may be surprised how many processes you are running, especially if you are using X. One of the processes is the ps command itself, which of course dies as soon as the output is displayed.

Figure 10-2. Output of ps command

The first field in the ps output is a unique identifier for the process. If you have a runaway process that you can't get rid of through Ctrl-C or other means, you can kill it by going to a different virtual console or X window and entering:

$ killprocess-id

The TTY field shows which terminal the process is running on, if any. (Everything run from a shell uses a terminal, of course, but background daemons don't have a terminal.)

The STAT field shows what state the process is in. The shell is currently suspended, so this field shows an S. An Emacs editing session is running, but it's suspended using Ctrl-Z. This is shown by the T in its STAT field. The last process shown is the ps that is generating all this input; its state, of course, is R because it is running.

The TIME field shows how much CPU time the processes have used. Because both bash and Emacs are interactive, they actually don't use much of the CPU.

You aren't restricted to seeing your own processes. Look for a minute at all the processes on the system. The a option stands for all processes, while the x option includes processes that have no controlling terminal (such as daemons started at runtime):

$ ps ax | more

Now you can see the daemons that we mentioned in the previous section.

Recent versions of the ps command have a nice additional option. If you are looking for a certain process whose name (or at least parts of it) you know, you can use the option -C, followed by the name to see only the processes whose names match the name you specify:

$ ps -C httpd

And here, with a breathtaking view of the entire Linux system at work, we end our discussion of processes (the lines are cut off at column 76; if you want to see the command lines in their full glory, add the option -w to the ps command):

kalle@owl:~ > ps aux

USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND

root 1 0.0 0.0 588 240 ? S 14:49 0:05 init [3]

root 2 0.0 0.0 0 0 ? S 14:49 0:00 [migration/0]

root 3 0.0 0.0 0 0 ? SN 14:49 0:00 [ksoftirqd/0]

root 4 0.0 0.0 0 0 ? S 14:49 0:00 [migration/1]

root 5 0.0 0.0 0 0 ? SN 14:49 0:00 [ksoftirqd/1]

root