BACKGROUND MATERIAL: UNDERSTANDING A PROCESS IN LINUX

  Before attempting system coding you should become familiar with the important 
  attributes and parameters attached to a user process from the perspective of
  the kernel. In particular, you should

  o  understand the output of ps and top
  o  understand the /proc pseudo-file system

  A Unix process is an image in memory with a thread of execution and a context
  that restricts resources that the process can access. Since everything in 
  Unix is a file or a process, the information that the kernel maintains to
  manage a process is stored in files.

  Much of the kernel information can be displayed with the 'ps' command. The ps
  command grabs information stored in a pseudo-file system mounted at /proc. 
  Before starting the coding portion of this lab, read through the man pages 
  for proc, ps and top.

  You should know how to gather information on a process and understand what 
  that information means. For example, if you execute a process that runs
  continuously you can run 'ps' with various switches to get information.

  The output of a process can be written to stderr (descriptor 2) or written
  to stdout (descriptor 1). You can redirect both to a file:

       ./foo >log 2> err

  Know how to read the man page for Unix commands. In the ps examples below 
  '|' means OR and anything enclosed in square brackets '[]' is optional:

  Usage: ps -[f | ly | j]u username  

  Useful Commands (sspade is a sample username):

   $ top -u sspade
   $ ps -fjHlyu sspade
   $ ps -fu sspade    # full listing of all processes owned by sspade 
   $ ps -jHu sspade   # display process tree
   $ ps -lyu sspade   # displays long format including rss 

  You can also get useful information with the top command:

       $ top -u sspade

  You can also get the memory map with the command 

       $ pmap -x [process_id]

  The pmap command can find things you can't as a user because it is allowed 
  to read kernel memory (/dev/kmem). 

  You can also get a map of processes with the 'pstree' command. (Pstree will 
  not work on a vt100 terminal unless the emulator supports Unicode.) Pstree 
  will work under xterm. To start an xterm session:

      $ ssh -Y username@sleipnir "/usr/bin/xterm -ls -geometry 80x50"
 
  In the xterm window:

       $ pstree

  Finally, explore /proc/{pid}. Everything the OS needs to execute the process
  is stored here. Read the manpage for proc(5).
   
      $ man 5 proc

  Using the tools discussed above you can find the following information about 
  a process. There may be differences between systems in the abbreviations and 
  units of measurement -- refer to ps(2) manpage for help.

  UID (user ID)
  PID (processID)
  PPID (parent processID)
  TTY or TT (controlling terminal, if any)
  RSS (resident set size)
  VSZ or SZ (virtual size)
  WCHAN (address of event for which process is waiting)  
  PRI (priority in system-dependent units)
  CMD (command line invocation via execve())
  STAT or S (process status: R-runnable, T-stopped, S-sleeping, Z-zombie, ...) 
  STIME (starting time)
  TIME (profiling time - user plus system)