Show HN: My from-scratch OS kernel that runs DOOM

3 months ago (github.com)

Hi there! I've been on-and-off working on TacOS for a few months, which follows some UNIX-derived concepts (exec/fork, unix-style VFS, etc) and is now able to run a port of Doom, with a fairly small amount of modifications, using my from-scratch libc. The performance is actually decent compared to what I expected. Very interested to hear your thoughts. Thank you!

Hi, congratulations! You must be feeling proud. Nice choice of proof of concept (DOOM).

Sorry to disappoint you but all I have are some noob questions.

What would be the steps to run this on a laptop? I take it that after building it there would be a process similar to, say, setting up dual-boot in a Windows PC? (Whoa I'm asking a stranger on the Internet how to run dangerous software on my computer...)

If one wanted to undertake such a project, do you have any recommendations of textbooks or other reading material? I had some OS & related courses in university (I'm EE, so computer-adjacent), but they were all very abstract / high level / conceptually-focused. I'd love something more concrete. It doesn't have to be x64.

  • Not disappointed at all to answer! Running it on my laptop was literally just formatting a USB with the ISO and booting from the USB.

    I would recommend checking out https://osdev.wiki to start out if you want to write a kernel, as well as reading relevant specifications (such as Intel Developer Manual and the specs for any drivers you write). I don't really know much about non-x86 kernel dev but most of the concepts are the same as far as I know, just different technical implementations. There's a link to a discord server on the project's readme, there are some very smart people in there who I'm sure would be more than happy to help you out.

Ok, but can your tacos run DOOM??

I kid I kid ;) That's a commendable effort and nice job! Question though: was it an effort to make TacOS using DOOM as a "standard" or was it an effort to make an OS dedicated to running DOOM run from scratch?

And I don't ask from any place but actual curiosity. I made an absolute bare-bones-cant-do-anything-but-boot type OS way back "in the day" (like almost 30 years ago, ack!!!) just for my own education/fun, but the idea of having a dedicated OS that can basically only run DOOM, yet be ported to anything would just make the "can it run DOOM" meme so much more ironic and fun!

Nice stuff! Keep it up!!

  • Well, it can do a little more than run Doom, that's just the latest milestone as to what was ported last. It wasn't a huge effort, maybe took a week or so to get Doom itself running including adding libc requirements for Doom, but what came before that was a fair bit more groundwork. I used DoomGeneric, which is basically a fork of Doom which is made to be very portable. I hope I answered your question, I might have misunderstood.

From-scratch kernel straight to DOOM is some top-tier hacker cred stuff. You must be very happy with it running on real hardware. Very cool.

  • Yes, quite happy with real hardware. The kernel isn't exactly straight to Doom, it boots into a shell which you can just run Doom from, but yeah lol.

Slight tangent but I've wondered about something similar to this, has there been much initiative to make games that directly boot on modern PC hardware? So not load a full OS but just go directly to the game. Similar to older generation gaming consoles. It should be possible, granted if you want to stay simple, things like wifi, bt, GPU would be hard to utilize without modern drivers, but a keyboard and mouse should be fairly doable as they seem to have some sort of default BIOS access? (probably wrong terms there but hopefully my point is understandable)

  • I don't know if it's been much used but it is known and works. I was doing this early on in my x86-16 assembler experiments but ended up using DOS as a program launcher for an easier emulator to use than qemu (dosbox-staging).

    The big limits if you don't want to get into disk IO, is 512 bytes or less since you're basically running your program as a master boot record. To get more you'll need to load some LBAs from disk which yes there is an interrupt for and osdev has even better stuff.

    Other than that, the difference between a .com file (usual limit 64kb single segment) and an MBR style bootable program is pretty minimal

Great work, I would love to have the skills to do something like this, but I can see you had to read lots of specifications to achieve this and thats my weakest point.

One silly question you may know: Imagine you wanted to use GPU acceleration, even in the smallest form. How hard would it be to build a driver for the GPU? Do you think there is good documentation about it?

  • Umm that's probably the extreme end of OSDev which I likely wouldn't be able to do, at least not for a driver you can buy. Qemu's emulated GPU is documented decently and could be possible, but things like nvidia GPUs are badly documented (and until recently, the docs were fully closed source) - even Linux has issues with this (and I actually see a few other hobby OS devs who just use Linux's GPU drivers in the end). There aren't a lot of things I've marked as near impossible but writing a genuinely decent GPU driver for a common GPU isn't really something I imagine I'll ever be able to do sadly.

    • The docs for intel’s integrated offerings seem pretty good, I’m planning to do a driver for those on my toy OS. Anything else is hit and miss in terms of availability of information unfortunately.

    • thanks for your response. Thats what I heard but I was wishing things to change...

      I was also thinking about integrated GPUs like the one in a RPi or other SoCs

Hey unmapped (I am ThatOSDeveloper on GitHub and discord it's my display name) I never knew you got doom running on it, pretty cool, but I have a few questions, is it the original doom, is it on disk or an initramfs and do you use freedom or the shareware doom wad with the engine you use?

  • It's doomgeneric, as can be seen in the article, with a fairly small amount of modifications, as can be seen at the top of this page.

  • Hi, I use DoomGeneric which is a portable fork of Doom. It's on a TempFS loaded from an initrd. I use doom1.wad.

Very cool, but why choose an unsafe language when today there exists low-level memory-safe languages? We all know already that most of security bugs are memory related.

I get that this is an hobby project, but still why not deprecate unsafe languages where there are much better alternatives?

  • Mostly just because C is a lot simpler, and in kernel dev, simplicity is everything. I've used rust for other projects but I feel like in kernel dev I would much rather use a simple and readable language than a safe language.

Very cool project! How are you handling process isolation and scheduling in TacOS?

  • I use paging for virtual memory, which gives each process it's own address space. I have a round-robin scheduler connected to the PIT driver, so every 10ms the PIT fires an interrupt which triggers the scheduler, which selects the next task, saves the current state of the previous task, switches to the new address space, switches the stack, restores registers of the task, then uses the iretq instruction to switch to ring 3 user mode and jump to the instruction pointer.

    • Thanks for that explanation. I've been doing some low-level programming lately, and I'm getting interested on running stuff bare-metal. Every previous description of multitasking I've seen has been very hand-wavy.

      2 replies →

Dude, I'm impressed cause I don't the energy to build my own OS kernel!

But here's a wacky idea. Just set it up so that the OS only runs DOOM as default. Boots directly into Qemu and runs DOOM, nothing to select or change. Maybe something you could fold other games into so that they can run off a USB boot loader. Might be appealing to people who don't want to install or compromise their base setup.

  • I could probably do that. The init program, which is a userspace program called by the kernel which spawns everything else, currently enters the shell but it'd be quite simple to make it just boot into Doom immediately. It's still not quite stable enough for actual usage, but it is able to boot off a USB at the moment, yes.

Great project. Mmap implementation is a cheat :).

  • Yeah mmap is quite a stub lol, I def cheated on that. I just wanted framebuffer access, but the user heap uses sbrk internally instead of mmap anyway so I properly

What did you do make the project interactive enough so you don’t end up copying existing implementations?

  • I'm not too sure what you mean. I kinda just avoided looking at existed implementations because it's a bit more interesting to do it myself.

Why is running doom one of your milestones, given that doom will run fine on bare metal?

  • That __is__ true, Doom can run on bare metal, but that's a fairly hacky solution in many ways. Doing it properly in userspace with a LibC and conceptually POSIX syscalls require a bit more effort. It requires a list of LibC functions and POSIX utilities as well as a few OS specific implementations.

  • Doom is a fine milestone for an OS that intends to have graphical capabilities. Maybe Doom in text mode for other OSes :P

    It's a known quantity, and doom has been ported to everything, so it shouldn't be too hard to make work either (he says, not having done it).

    An OS that can run Doom is clearly capable of graphical interfaces with user interactivity, and maybe sound (although sound would be easy to leave out).

    • Yeah, that's perfectly fair. I raised my question because it occurred to me this kind of sells the operating system short. In other words: even DOS could run Doom (with some help with a memory extender I guess, which is a technology we can all very happily not worry about anymore).

      1 reply →

Dude, getting Doom to run on your own kernel is epic - I gotta try building some wild stuff like this someday

  • Yeah definitely an achievement I'm happy with. I've got a bit of refactoring to do, ANSI parsing etc then I'd like to port more - perhaps even Vim (or another portable Vim-like editor called Dim)

    • Could you explain more about why you'd need to port things? If you have a libc, shouldn't it "just work"? Do you have to scatter #define or #ifdef all over? What about if you wanted to support Golang? Would you have to implement a bunch of syscall?

      1 reply →

A few months work by one guy and already more capable than the Hurd.

Imagine what you could accomplish given 35 years.

  • > A few months work by one guy and already more capable than the Hurd

    It is no way capable than Hurd. It is a cool project though. Have you used Hurd recently? It can run a modern desktop.

    • I searched YouTube for actual evidence of Hurd booting to a desktop and only found two videos of Hurd freezing during boot, and a third video of RMS explaining to a very confused convention attendee that he's "never installed GNU slash lynn-ox" because he could just ask someone else to do it.

      No videos of Hurd running Doom either, but anyone is welcome to create one and share.

      6 replies →

    • FWIW, GNU/Hurd has been able to run a desktop environment the last 20+ years.

      And as someone who portedgor Doom working on GNU/Hurd decades ago … I’m going to wager that the GP has never used it. :-)

  • Haha not quite at the point Hurd has gotten. I'd be very happy if I could last that long on one project though, I have a tendency to get bored after a few months sadly.

    • > Haha not quite at the point Hurd has gotten

      that's true, you've only shipped to one computer, while they've shipped to dozens!