OSDev.org

Finally, finally I have a stable but still-needs-to-be-improved multitasking implementation.
(After asking ridiculous amount of questions)



The seven tasks are loaded from seven ELF binaries that prints A, r, d, u, i, n, o 255 times respectively that compiled with my i686-agola os specific toolchain.

Agola wrote:Finally, finally I have a stable but still-needs-to-be-improved multitasking implementation.
(After asking ridiculous amount of questions)



The seven tasks are loaded from seven ELF binaries that prints A, r, d, u, i, n, o 255 times respectively that compiled with my i686-agola os specific toolchain.

Nice to see that beast finally working. After all those questions.
Does it support any kind of synchronization or message passing or resource sharing?

Octacone wrote:

Agola wrote:Finally, finally I have a stable but still-needs-to-be-improved multitasking implementation.
(After asking ridiculous amount of questions)



The seven tasks are loaded from seven ELF binaries that prints A, r, d, u, i, n, o 255 times respectively that compiled with my i686-agola os specific toolchain.

Nice to see that beast finally working. After all those questions.
Does it support any kind of synchronization or message passing or resource sharing?

Yes, I've a very basic mutex implementation. I've implemented async message passing also but it is not *very* stable and it is very very ugly. I'm working on signals now, but it is kind of hard though.

I've totally rewritten my kernel with a far more portable / stable codebase. I've not yet escaped kernel-mode, but I do at least have fake "processes" and "threads" that, aside from being in kernel-mode, operate like the real thing.

Here I am cat-ing a file sitting in my initrd and requesting the system time.
Much to do!

here is the same chat proof of concept, this time via TCP. Sadly I was lazy in excitement to rename the executable. Only modified SOCK_DGRAM to SOCK_STREAM, so yeah.

http://imgur.com/b5sPyhP

Nothing fancy, just a kernel shell. Need to work on memory before moving on, though.
This is my 4th or 5th try at OS, and furthest I've been, I am pretty big noob at this, so I am very much excited.





Previous versions:

Chilling in IRC with fun people smiling...

http://i.imgur.com/xo3W0Cv.png

Why have i suddenly started working on a network stack after a few years of inactivity?

To do this: https://www.instagram.com/p/BYdy8Xmlq9r/

Once upon a time i found this old motherboard on a "old used crap" sale.
It was some sort of a self-contained thin client - some weird Via x86 CPU, Via Rhine network card, 32 Mb of RAM.
And a Disk-on-Chip by M-systems.

That DoC is only 32Mb, and is one of the earliest SSDs.
I long wondered what was on it, and recently decided to figure out what would it take to find out.

The chip was fairly well documented, so writing a driver for it was straightforward.
Getting the data out was a good excuse to finally make a network stack.

The network card is supported by Linux, so i cheated by peeking at it while i wrote a driver for Aprom.
The network stack itself took some time.
After i got a ping working, some time was wasted hunting old architectural bugs and rewriting a bunch of architectural stuff elsewhere.
And finally, after a few false starts i managed to dump the content of the DoC and send it over UDP via a real network card with real driver to another PC.

Sure, i could have just dug up an old Linux distro, slap a DoC patch on it, dump from there, and be done in an hour rather than a few weeks, but what's the fun in that?

And this is how i used my Aprom OS for a real purpose for the first time.

I'm porting Agola to 64-bit, after two hours of work, I've finally entered long mode. I will be able to load my kernel fully after building a 64-bit IDT and rewriting the paging code for 64-bit. Seperating arch specific things and common things in Makefile design before helped me a lot.

Agola wrote:I'm porting Agola to 64-bit, after two hours of work, I've finally entered long mode. I will be able to load my kernel fully after building a 64-bit IDT and rewriting the paging code for 64-bit. Seperating arch specific things and common things in Makefile design before helped me a lot.

Make sure you get get higher-half mapping working asap. It'll save you a lot of bother further down the road

Here is something that took ages to synchronize and in the end had to be rewritten.
It was definitely worth an entire morning of work.

Gramado - Shell App

Gramado: Message box.

Gramado: Help message on Boot Manager.

Gramado: cpuid