OSDev.org

Nothing impressive or of the other world, has no paging, multitasking, or executables files yet. (well, sincerely I have no idea how to do it, but well ... ) At least it's something more than a simple bootloader that just prints "Hello world".
Only haves an 80x25 screen resolution, text only. Who needs GUI? A simple shell is enough for everyone! You can have a fancy OS with beautiful GUI, but the kernel can be buggy, or not fully functional. (there are exceptions) Or you can have a ugly shell, but a fully functional kernel.

Now CHicago have it's own assembler + linker for the userspace, with a custom executable format.

All the non-efi stuff (except the loaders) are written for my OS. Now just to find the framebuffer, load the driver and kernel files, and set up virtual memory.

@K3achas Lol, nice messages

First time posting to this forum
This is an image of my kernel loading a kernel object

I've not posted a screenshot for a little while since I've largely been working on "invisible" back-end stuff, like hardware autodetection, package management, etc. and haven't had much to show... I've recently returned to working on UI-related stuff, so here goes:



This shows a few "prototype" GUI apps I built relatively quickly based on a port of SDL Widgets (whose stock test program is also visible). At the bottom of the screen, you can see my very early work-in-progress on my own GUI widget system.

EDIT: Here's a bonus screenshot of a much higher resolution display with (nearly; there is evidence of the missing one...) every GUI application that exists for BT/OS visible, along with a bit more work on my widgets:


(Click for full size)

mallard, I'm very impressed, nice progress. keep going!

Hi guys,

I decided to update my code base a little. The following is a quick screen shot, nothing fancy, but (if sound was included) you can see that I now (mostly) support the AC97 sound card.



As a side note, I decided to let it boot at 640x480x32 and of course the background image is a little larger and "bleeds" to the top. I will have to fix this.

Anyway, after the interest in the AC97 Post, and after reading over the specs, I decided to see what I could come up with. Surprisingly, the AC97 is actually quite simple to program, once I clarified a few things with different documents.

Anyway, I have an updated (USB) bootable hard drive image (20 Meg) at the home page or a direct link (7 meg zipped).

It is a Lean FS image with the boot and kernel files, fonts, various image formats and now, various sound formats as shown in the listing in the image above.

The GUI runs fine (though a bit slow) in an emulator, i.e.: all images are visible, even the animated .GIFs and animated .PNGs. It runs a lot faster on real hardware since it takes advantage of the hardware's fast memory move features, but some of the images now don't display at all. uuhhmmm. I will have to investigate this.

Anyway, just thought I would put up an update. Thanks to everyone here for keeping this hobby interesting.

Ben
- http://www.fysnet.net/osdesign_book_series.htm

Excelent work Ben!

After I've started rewriting the whole scheduler/syscall/task/vfs thing, there's not much to show, but now the kernel supports kernel-mode tasks.

And with vesa framebuffer console:


Here's also a screenshot from before I've started the rewrite - signal handling in userspace:

Looks like Windows 3.x and 9x have been mixed together.

Hello, World!
Here's the culmination of about 3 to 4 weeks of work in my free time:



Of course it's still missing tons of basic functionality, but since it can load and run an executable, I believe it has just reached the point of being an operating system.

A new, UEFI based, kernel
And yes, the APs are all started up (trying to boot another OS afterwards hangs, since the SIPIs are lost)




https://github.com/ChaiSoft/ChaiOS

In the last few months, I did not do much work on ports. Nevertheless, today I decided to implement the missing stuff for ncurses and nano.

Apart from that, a new major feature in managarm is that userspace drivers can now upload eBPF-like programs to the kernel. The programs are written using C++ expression templates (as in this example), compiled to x86_64 ELF shared libraries by my new SSA compiler library (called lewis), and inserted as kernel modules. This is used to process interrupt handlers synchronously (which previously had to be asynchronously in a mask-irq/wake-drivers/unmask-irq cycle). Hence, this feature improves performance by closing the gap in IRQ latency with respect to monolithic kernels.

I'm trying to learn how to write a simple OS.. here it is! MiaOS
UEFI app style OS written in C and gnu-efi

https://imgur.com/a/bQMU5Ff