Chapter 6, Processes

• Please read chapter 6.
• I intend to cover this chapter lightly.
• The loader
  • Historically a loader was a process that
    • Reads the contents of an executable file.
    • Places it into memory in the appropriate location.
    • Copy command line arguments to the correct location in the stack.
    • Initialize registers
    • Jump to the entry point
  • At one time this was a process that the operator had to run
  • This is integrated into the os through the exec family of system/library calls.
    • fork creates a new process
    • exec loads an executable into memory and begins execution.
• Binary file formats
  • There have been a number of binary file formats
    • a.out format : original unix, assembler output
    • coeff: Common object file format
    • elf: executable and linking format, the current standard
    • file /bin/ls
  • This file contains
    • Machine language instructions for the program
    • The address where the program starts executing (the address of main)
    • The data used by the program (global variables and constants)
    • Symbol and relocation tables.
    • Shared Library and dynamic linking information
• Once the loader reads the program from the file and has the OS jump to the entry point, the program becomes a process.
• As we have discussed, each process has a unique ID.
  • The kernel keeps track of the last pid issued
  • When a new process is started, it searches for the next available pid
  • The max value is cat /proc/sys/kernel/pid_max
  • Changing the value in this file will increase the number of pid (hence processes) available.
    • 2¹⁵ or 32,768 is the max on 32 bit systems.
    • 2²² or 4,194,304 is the max on 64 bit systems.
  • This is linux specific
• Recycling of pids starts at 300 to avoid searching seldom unused lower numbers.
• remember init == systemd for the purpose of most of this discussion.
• pstree is a system command that will show (some) of the system processes and their relationship to each other.