OSDev.org

cloudapio wrote:[...]
Hello everyone! This is my first time posting on this forum, but I have been working on my OS for more than a year now. Today, I finally managed to make GCC work on my OS (after four months of it segfaulting every time I tried to run it).

Turns out the culprit was a bug in my memmove implementation, of all things:
[...]

The first i++ should have been i--. I have no idea how I managed to have a functioning OS for so long with that bug there, I fixed it after a GUI text editor I was working on started behaving a bit weirdly.

It's always great when you can start compiling on your own OS. It means it's close to being self-hosting.

The memmove bug reminds me of a problem I had with Linux on PPC around 2005-2006. Where Linux on x86 was rock-solid, many of the same programs segfaulted randomly on PPC. I think it had to have been something as deep as memmove if not within the kernel itself. It got better, then 8-10 years later got bad again. I wasn't too surprised by it getting bad again because Linux is heavily optimized and changes frequently enough that it needs "many eyes" constantly on the lookout for bugs. Less popular platforms are going to have problems. There's a lot to be said for not optimizing or changing that which works well enough.

There's also a lot to be said for testing, but it's not always easy to test well and it's awfully tempting to just write code and only fix the most obvious problems.

What was really only a dream when I started programming my OS has finally come true. I can view ODT and DOCX files. Of course, only on very simple level, but it is still great thing.

Klakap wrote:What was really only a dream when I started programming my OS has finally come true. I can view ODT and DOCX files. Of course, only on very simple level, but it is still great thing.

Congrats! But your screenshot isn't visible. Github isn't serving the image; it's not even giving an error.

Huh. I hope this will work better:

Klakap wrote:Huh. I hope this will work better

It does. That text rendering isn't bad.

Here's what my OS, NanoShell, looks like currently.




I'm also working on another OS, named Boron, but that one currently doesn't have much to show yet.

This is absolutely amazing, IProgramInCpp. I like this GUI very much. Just a mind blowing achievement. Jealous in a good way.

Looks great, iProgramInCpp!
And it's running SM64

pvc wrote:Looks great, iProgramInCpp!
And it's running SM64

Yep, I ended up "borrowing" the software renderer fgsfdsfgs wrote for the DOS port.

iProgramInCpp wrote:Here's what my OS, NanoShell, looks like currently.

It looks complete! You've come a long way.

eekee wrote:

iProgramInCpp wrote:Here's what my OS, NanoShell, looks like currently.

It looks complete! You've come a long way.

Only looks. Under the hood there are many changes I still need to make. For one, my libc is far from complete, and what already exists is probably pretty buggy as well. I also want to add dynamic linking at some point, among other things.

iProgramInCpp wrote:

eekee wrote:

iProgramInCpp wrote:Here's what my OS, NanoShell, looks like currently.

It looks complete! You've come a long way.

Only looks. Under the hood there are many changes I still need to make. For one, my libc is far from complete, and what already exists is probably pretty buggy as well. I also want to add dynamic linking at some point, among other things.

Oh, yeah... I'm actually kind-of dreading reaching that point, having so much behind-the-scenes stuff to fix. It makes me kind-of glad there is no standard for the basic library of the language I've chosen. I'm also not implementing dynamic linking so I have to design for efficient communication up front.

XenevaOS, this is what it looks like currently, March 2024 update.




Manas Kamal,

https://github.com/manaskamal/XenevaOS

I've been rewriting my kernel over the past year. So far I have:
- A good bit of POSIX implemented
- Ports such as Bash, GCC, X.org, Quake and Fvwm3
- Networking (missing TCP though)
- Filesystems such as ext2, tmpfs and devfs
- Disks drivers for virtio-blk and nvme (with caching done by a page cache instead of a buffer cache)
- SMP support (kinda buggy at the moment but it is able to boot to fvwm3)

Gramado OS in 2024