davidgm94 / birth Goto Github PK

• Windows

• MacOS

This is due most probably to runner code not exiting properly, but it could be Zig.

• Implementation
• Integrate it with Limine
• Integrate it with Rise (BIOS)
• Integrate it with Rise (UEFI)

Related: #74

Advantages:

• Filesystem-independent
• Compression
• Less scattered disk reads

Disadvantages:

• CPU time for compression/decompression (which is not that bad)
• You have to dig into the bundle to read the files (yet less costly than disk operations)

Currently the bootloader has a lot of common functionality expressed differently in each bootloader. I think it would be nice to refactor it so that common code is written only once and used in every bootloader had (at the moment just Limine and our own, Rise).

• Run default test locally
• Run all tests locally
• Wait for the Zig PR to be merged into master
• Run default test in the CI
• Run all tests in the CI

Currently the kernel is panicking in release modes

Currently, the behavior is to spin waiting for a device to driver interrupt (in the GPU and block drivers).
Sometimes, specially on Windows, the interrupts never happen and the kernel is stuck there.

Hello, is there a specific zig compiler version you are using? I tried to build a couple weeks ago, as well as again today, and things didn't build.

Could you add the version to the README as well, just so others will know which compiler version to use ;)

Thanks

Currently there is a race condition in the NVMe driver, at least with ReleaseFast mode. This is making the kernel unable to load a minimal userspace program since it can't load a file from disk. Hardly ever happens.

• Support all sorts of name, not just the most basic ones.

Currently the way we are using Limine is to load the kernel again from it as we don't use the default version loaded by this bootloader. It would make sense then to have a little prekernel that truly loads the kernel instead of this special "entry" which is forced to load the kernel twice.