Unix & Linux Stack Exchange is a question and answer site for users of Linux, FreeBSD and other Un*x-like operating systems. It only takes a minute to sign up.
Sign up to join this community
I often hear people refer to the Linux kernel as the Linux kernel image and I can't seem to find an answers on any search engines as to why its called an image.
When I think of an image I can only think of two things either a copy of a disk in or a photo. It sure as hell isn't a photo image so why is it referred to as an image?
The Unix boot process has (had) only limited capabilities of intelligently loading a program (relocating it, loading libraries etc). Therefore the initial program was an exact image, stored on disc, of what needed to be loaded into memory and "called" to get the kernel going.
Only much later things like (de-)compression were added and although more powerful bootloaders are now in place, the image name has stuck.
The word image also has a definition "A file that contains all information needed to produce a live working copy."
It doesn't mean an "image" is just a 1:1 copy of a disk. Just as a photograph represents the reality exactly as it was when shooting, an image of an executable program (or kernel) represents the program in a state, where it can be loaded (or unpacked) in the systems memory exactly as it is and then given control to it. That program can then start running from that state in a consistent manner. So the Linux kernel image is an image (a picture of the state) of the Linux kernel that is able to run by itself after giving the control to it.
Nowadays, the bootloader loads such an image from the hard disk’s filesystem (driver is needed), replaces itself with it and so gives the control to it. The booting process of a computer does that several times until the operating system finally runs. This is called chain loading. Or if a smaller program (chain-)loads a more complex one, it is called bootstrapping.
The BIOS loads the bootloader that is also an image, for example called boot.img in case of grub. That boot.img is not a file (if grub is installed); it is the name for the part that is in the Master Boot Record (MBR). If you dump that to a file it would then be an image in form of a file not rawly written to disk, but rawly written in a file. This is also a representation (image) of the earliest state where grub is able to load the rest of itself. grub then has its own mechanism how to fully load itself by loading other images. This is represented by the different stages in grub. After that, the bootloader loads the kernel image by replacing itself with the extracted contents of that file.
Ancient History. the term image comes from an old Digital Equipment Corporation term for the output from the compiler-> linker. the file is an image created by interpreting the code and so on through the linker to make an executable "Image" of your design.
In math the kernel is the inverse image of a subset of the image of a some map, were the subset is equal to the identity element in the codomain. I'm certain these names derive from mathematical concepts as they are related significantly in various fields in mathematics. Considering Unix was derived in an academic environment it may be possible that it's use of these word's kernel and image are the same.
If you have a set which represents some level of information about the "complete" O.S., if that information also forms a group then you can define group homomorphism's on that set or basically maps to other sets having different sizes then the original set so long as they "respect" the orginal set's structure that made it a group. You can see it may be in one's favor to map the set to a smaller set or a subset of some set where the subset is smaller.
Image - The image of a group homomorphism and in general functions and maps, are just a subset of some set, who's elements actually get mapped to. The function may not map to every single element and those elements would not be included in the image.
Kernel - Basically just the elements from the original set that map to the image, but only map to the identity element in the image. Basically the elements that map to 0 like thing in the image.
If the image is smaller in size then the original set then we can see multiple items must map to one single element. So for example there may be multiple elements from the kernel that map to the image and we already know they all have to map to 0.
We can see that if we choose the original set to be finite sequences of binary or 1's and 0's and the codomain (set mapped to) to be also sequences of binary, then we can construct such things if and only if, a suitable group structure can be defined (this little bit in depth and unrelated to question asked).
So we see with complete certainty that "kernel" and "image" of an O.S. are completely defined and have mathematical meaning. Independent from perhaps other uses of the terms.
