The boot sequence varies in details among systems but can be roughly divided to the following steps: (i) hardware boot, (ii) OS loader, (iii) kernel startup, (iv) init and inittab, (v) boot scripts. We will describe each of these in more detail below.
This program normally makes a basic self-test of the machine and accesses non-volatile memory to read further parameters. This memory in the PC is battery-backed CMOS memory, so most people refer to it as the CMOS, although outside of the PC world, it is usually called nvram (non-volatile ram).
The parameters stored in the nvram vary between systems, but as a minimum, the hardware boot program should know what is the boot device, or which devices to probe as possible boot devices.
Then the hardware boot stage accesses the boot device, loads the OS Loader, which is located on a fixed position on the boot device, and transfers control to it.
In most systems, this primary loader is very limited due to various constraints. Even on non-PC systems there are some limitations to the size and complexity of this loader, but the size limitation of the PC MBR (512 bytes including the partition table) makes it almost impossible to squeeze a full OS Loader into it.
Therefore, most operating systems make the primary loader call a secondary OS loader which may be located on a specified disk partition.
In Linux the OS loader is normally lilo(8) or grub(8). Both of them may install either as secondary loaders (where the DOS installed MBR points to them), or as a two part loader where they provide special MBR containing the bootstrap code to load the second part of the loader from the root partition.
The main job of the OS Loader is to locate the kernel on the disk, load it and run it. Most OS loaders allow interactive use, to enable specification of alternative kernel (maybe a backup in case the last compiled one isn't functioning) and to pass optional parameters to the kernel.
Some of the parameters that may be passed to the kernel relate to these activities (e.g: You can override the default root file system). For further information on Linux kernel parameters read bootparam(7).
This gives the system administrator an easy management scheme, where each run-level is associated with a set of services (e.g: S is single-user, on 2 most network services start, etc.). The administrator may change the current run-level via init(8) and query the current run-level via runlevel(8).
For each managed service (mail, nfs server, cron, etc.) there is a single startup script located in a specific directory (/etc/init.d in most versions of Linux). Each of these scripts accepts as a single argument the word "start" -- causing it to start the service, or the word "stop" -- causing it to stop the service. The script may optionally accept other "convenience" parameters (e.g: "restart", to stop and then start, "status" do display the service status). Running the script without parameters displays the possible arguments.
A primary script (usually /etc/rc) is called from inittab(5) and calls the services scripts via the links in the sequencing directories. All links with names that begin with aqSaq are being called with the argument "start" (thereby starting the service). All links with names that begin with aqKaq are being called with the argument "stop" (thereby stopping the service).
To define the starting or stopping order within the same run-level, the names of the links contain order-numbers. Also, to make the names clearer, they usually end with the name of the service they refer to. Example: the link /etc/rc2.d/S80sendmail starts the sendmail service on runlevel 2. This happens after /etc/rc2.d/S12syslog is run but before /etc/rc2.d/S90xfs is run.
To manage the boot order and run-levels, we have to manage these links. However, on many versions of Linux, there are tools to help with this task (e.g: chkconfig(8)).
In older Unix systems, these files contained the actual command line options for the daemons, but in modern Linux systems (and also in HP-UX), these files just contain shell variables. The boot scripts in /etc/init.d source the configuration files, and then use the variable values.