Let us say that I claim to know how to make an OS that is impossible to hack

But Tom is right here: you must make decisions and choose design options that will in fact affect your whole future evolution ... but you can not make the right choices unless you have your own vast coding and design experience.... this is the issue

It is more likely that you will choose blindly the "standard", "good design" and "well known" options and you will end (if ever) with either Linux or Windows.

Pre-emption is a no-brainer, it is very simple to implement. The scheduler simply selects a different saved state to restore when the kernel switches back to user code.

I think some people are confusing pre-emption with task isolation.

Yes, I believe you, but I would also not expect the OS to do much.

Insecurity comes with complexity. Security comes with simplicity.

I don't want another linux or windows. The differences will be in the actual implementations of things like VFS, Driver Interface, API, Ints, Syscalls.

Let me reformulate: A simple OS that is functional and can never be hacked. I was not thinking towards complex solutions. I prefer simplicity over complexity anytime but I was not thinking of an OS that does not do anything either.




Wrong. A difference in implementation is almost no difference at all.

Unfortunately, other than:
- fantasy and dreaming of unexperienced
- adding never ending complexity
- "good looks" but no contents

Everything else that could have been done in OS development was already done.

Of course it might still make a huge difference to your own skills and understanding but nothing else

Difference in sys call interface, see "Modern Operating Systems"

So you must be a beginner...

round robin is the fairest scheduling algorithm

See here for some ideas about scheduling algorithms and how to measure fairness.

when i see this thread, i instantlly remember this one

the beginner think he is the best because he begin to understand some things.
then, as i learn more and more about the asm coding, my own opinion about my knowledge is falling in the bottom, and i consider now that i am very null. the total opposite of my first self-opinion.

opinion is realative, the more you learn, the more you see your ignorance.

about the impossible to hack OS, i think that the OS that constantlly hack the others is a bit more protected than the one who always try to prevent attacks, exactlly like human being, the one who always try to control the others have less chances to be controled than the one who always try to evict controllers.

for example, the US airforce OS, this one (i don't know it's name and i don't know where to see screen shots) is the one who is pretty sure to be the more impossible to hack, because it always and constantlly hack the others, in the objective of spying and antiterrorism.

but possible i am wrong.

Not at all, I have read it though when I was just getting in to OS programming. I have read much on the topic. Many tuts also. I am confident in my skills.

Not proud, humbly confident

os.hacker64: Can you point us to an example of your work? How far have you progressed with the OS so far?

I keep my work to myself, I have a basic kernel with some text printing functions newline etc.

Then you can't expect others to help you.

An OS needs more than one person, If you want a starting point I'll give you one.

There are many OSes around written by only one person. Just to be clear, I won't be participating, I have already written too many OSes to start yet another one. But keeping all your code private is not going to encourage others to give you their code. You need to give and take.

exact, just look at my work.
the code i don't share is rare.

one day, i decided i'll share all my code.
to show the beginners errors, halp the beginners to learn, and enlarge the disponible code for fasm coding.

one day, i was thinking that i'll not share my code, in the fear of "if a m$ like coder stela my code, then make money on my back, and then fuck me".
this idea is in the cave now, becaus ei think that if somebody steal my code, make money with, and then fuck me, it means that he depend on me, and then, i can build a trap, in return to fuck him more and more with delation and pollution of his website.

and fancklly the chance that this can happen is very thin. cause now, companies don't know anything in asm, and nobody code in asm when it is destined to control the world.

bye bye.

ps:
there are many coders here that have a kernel with text printing funtions. including me.
and mine is better, it can control the keyboard as a joypad and display moving images, with cool effects and new entries creation via a command line.

http://www.oby.ro/os/index.html
I think Bogdan has done an excellent job of explaining the rationale for the various features of his operating system: SOL.
That is the sort of prologue that I have in mind, for you to elaborate, as opposed to simply furnishing a book title. Bogdan's work is mature, and very complete, you of course would not be expected to generate such a detailed design doc at the outset, but surely, a paragraph of three sentences ought not be too much trouble.
With regard to revolution's point, I share his opinion, and urge you to produce a simple overview paragraph, which need not divulge details, but which makes very clear your rationale for pre-emption, rather than cooperative multitasking. Your earlier post referenced a C compiler, and that notion suggests, to me at least, that you propose developing a non real-time operating system, if so, a phrase or two explaining the intended useage of this new OS could be helpful....

Some OS prevent the user from accessing ring zero, others permit applications to operate with the same level of privilege as the OS itself, so, it could be instructive for any prospective collaborator to know your intentions in that regard.....

I'll be back.