Word | Operating System Development | Page 1

weak karma Aug 17, 2025, 12:39 AM

#

32-bit x86 kernel that I work on because its awesome and I have nothing better to do. I kinda just do whatever I feel like with it because that is fun to do. I do not have a todo list anywhere because I just figure it all out along the way. Perhaps one day I'll write my own bootloader for it but probably not because Multiboot2 is fine enough I feel.

#

Got my scheduler working fine enough with the 2 tests I've done. Hooray!

#

And by 2 tests I mean I had a process run while another one was running and then deleted it and then went back to process 1

#

and it didn't crash 100% of the time

haughty bough Aug 17, 2025, 12:52 AM

#

this is fire, keep it up 🙏

wild ferry Aug 17, 2025, 9:34 AM

#

cool, is the source available anywhere?

weak karma Aug 17, 2025, 1:54 PM

#

It is available here: https://github.com/ajmoritz2/wordOS

GitHub

GitHub - ajmoritz2/wordOS: Word

Word. Contribute to ajmoritz2/wordOS development by creating an account on GitHub.

#

Whether or not it compiles for anyone but myself is something I have not tested yet and do slightly doubt that it will

weak karma Aug 19, 2025, 10:14 PM

#

Been trying to get this to run on hardware

#

VMs really do just make it all so easy huh

weak karma Aug 22, 2025, 2:40 AM

#

Okay so progress on the real hardware part. My initial boot page tables decided to not map on real hardware due to some logic i put in

weak karma Sep 8, 2025, 11:34 AM

#

Yep so I gotta use PAE

#

my computers map the fb above 4 Gi

#

Might cry or explode or both

uneven talon Sep 8, 2025, 11:34 AM

#

weak karma Yep so I gotta use PAE

Do long mode instead :^)

weak karma Sep 8, 2025, 11:35 AM

#

Probably a better idea, but for some silly/stupid reason 32bit calls my name

#

Maybe it is a siren song....

weak karma Sep 10, 2025, 3:54 AM

#

Okay ngl PAE is kinda fun

#

I'm enjoying putting it in

#

It just kinda makes sense and juggling around pages is so whimsical

#

Especially because I haven't gotten to the annoying part where I make up for hardcoded mistakes past

coral robin Sep 10, 2025, 8:38 AM

#

weak karma Probably a better idea, but for some silly/stupid reason 32bit calls my name

32 bit is better than 64 bit

#

don't listen to the glowies

young crystal Sep 10, 2025, 10:07 AM

#

coral robin 32 bit is better than 64 bit

https://tenor.com/view/this-post-was-fact-checked-gif-13039434031513975568

Tenor

weak karma Sep 10, 2025, 1:56 PM

#

I don't know if I would say it's better

#

I just think more random ass shit

coral robin Sep 10, 2025, 2:33 PM

#

young crystal https://tenor.com/view/this-post-was-fact-checked-gif-13039434031513975568

who cares

#

fsb are also glowies

weak karma Sep 11, 2025, 1:26 AM

#

So I got my stuff transferred over now but now I gotta figure out how recursive paging is going to work

#

Or even if I want to do that

#

Since I still only get 4 gigs of address space, do I want all of that memory just sitting there

#

Well I guess it's only 8 mini bytes I'm missing out on

weak karma Sep 13, 2025, 3:10 AM

#

Okay I'm gonna explode if I keep procrastinating this

#

And I don't want to explode

weak karma Sep 13, 2025, 9:11 PM

#

Okay I got the fb working when I jerryrig

#

Now lets try not to be Jerry and start being tom

weak karma Sep 13, 2025, 9:52 PM

#

Okayyyyyy no more Jerry

#

weak karma Sep 14, 2025, 5:10 AM

#

Okay time to rework my scheduling

#

Although tbf I think it will be easier this time because I don't have to fuck with cr3

#

I'll just keep the 3rd entry in the pdpt the same for all processes

#

And from there change the bottom 3 entries where necessary

#

Perhaps PAE is a blessing in disguise

#

Kinda like Shrek 2

weak karma Sep 22, 2025, 4:04 AM

#

#

We are getting places a bit

#

Scheduling works on qemu now so just gotta get it to work on hardware

#

Ngl not sure why this happened but that's a tomorrow problem, not a now problem

weak karma Sep 25, 2025, 3:13 AM

#

okayyyyy stuff kinda works-ish now

#

uhhhh only issue is spaghetti code didnt set up ps/2 correctly

#

but ofc it works in qemmu

#

soooo now comes the fun part of debugging why my ps/2 keyboard doesnt send interrupts on hardware

#

I know my ioapic works fine, and there isnt an override, so thats good at least

#

and i also know the ps/2 exists because I get a keyboard scancode and whatnot

#

which leaves me with I just didn't set it up because a year ago me was more focused on quantity rather than quality work

#

and since I didn't start doing stuff on hardware until recently, it worked unfortunately

#

oh yea and my ACPI code needs to be completely reworked at some point because of how ass it is

#

like cmon why did I do less than the bare minimum when I wrote it the first time

weak karma Sep 25, 2025, 10:24 PM

#

sooooo

#

now with my properly tested and I guess initialized ps/2 controllor i now have another issue

#

I think I have to disable legacy usb because I am being told my ps/2 is non-functional

#

well not really non-functional, but it fails the self test

#

or whatever the A9 send is called

#

nvm my code is just ass i guess

#

0x45 is NOT a valid response

#

okay yea no its definitly the legacy usb stuff because I'm getting undefined behavior that is not the same every single time

#

aaaand my bios doesnt have an option dis disable it...

#

so I guess i have to write a pci driver and whatnot now

#

wait no im just stupid once again

#

well possibly

#

I think what I actually need to do is get rid of function wrappers because they might be fucking with my serial communication too much

#

or i could also just be stupid +10

young crystal Sep 26, 2025, 4:35 AM

#

weak karma I think I have to disable legacy usb because I am being told my ps/2 is non-func...

is legacy usb is ps/2 keyboard emulation

weak karma Sep 26, 2025, 4:36 AM

#

young crystal is legacy usb is ps/2 keyboard emulation

I think it's supposed to be, but I've read it's hella glitchy and rarely works

weak karma Sep 26, 2025, 10:34 PM

#

ok so i forgot polling the ps/2 existed

#

but now i remember and the irq not firing is the issue...

#

not really sure why, so now i guess i check if its enabled or something like that

#

okay it just straight up wasnt enabled l0000l

#

yea im a bit slow...

#

This might be awesome

weak karma Sep 29, 2025, 3:10 AM

#

Okay lmao I should probably zero out memory before I use it

#

That would be why it make the yellowing

#

Now all I gotta do is fix up my scheduling to have it work proper and then time for pci!

weak karma Sep 29, 2025, 3:53 AM

#

Ooo interesting bug I've just ran into

#

I forgot to invalidate the tlb cache but my code works on real hw

#

It doesn't work on qemu however

#

This tells me they do their tlbs differently

#

Kinda funny

weak karma Sep 30, 2025, 4:09 AM

#

Okay so funny issue now

#

Because my processes are quite separate, I cannot use the terminal unless directly in the terminal process

#

I feel like I now need to add some sort of IPC, at least at a base level so this issue does not persist

#

I forgot I would have issues with this

#

...

weak karma Oct 2, 2025, 3:18 AM

#

okay i've been wrong about all my guesses, and now i m no closer to figuring this out

#

its something with the scheduling messing with memory i think

weak karma Oct 3, 2025, 3:42 AM

#

Ok I am closer

#

The pdpt does not change when the process changes

#

This might be causing issues

#

(It most definitely is)

#

I will fix this tomorrow or probably not tomorrow but at least I know a possible cause!

weak karma Oct 8, 2025, 4:00 AM

#

I did not fix this tomorrow

#

BUTTT

#

I found out why for reals

#

My function returns are not saved on process switch or something

#

So my function caller gets back garbage

weak karma Oct 9, 2025, 4:03 AM

#

Working on stuff is hard

#

I just haven't

#

HELP

#

Well I don't actually need help

#

I just need time....................,.

weak karma Oct 9, 2025, 3:47 PM

#

I'm gonna freak out

#

It was exactly what I thought this entire time

#

Stupid tlb

#

I was not clearing it at all

#

So the stack I was using was just not the right one

weak karma Oct 10, 2025, 4:03 AM

#

Now I could either PCIe time or remake my ACPI parser

#

I think I know which one I'll do

weak karma Oct 11, 2025, 4:29 PM

#

okay actually I'm still debugging this problem l00l

#

it keeps appearing in new ways

#

i think the stack is being silly now and switching before I want it to but I can't prove it just yet

#

and I also just need to spend more time

#

so hopefully this weekend it should be good because I have time

#

please

#

I just wanna rework my acpi parser

#

and then pcie

#

and then fs

#

and then fat or something maybe tar first idk

weak karma Oct 15, 2025, 3:35 AM

#

Tlb is my enemy

#

:'(

weak karma Oct 15, 2025, 2:12 PM

#

okay its not my enemy anymore

#

i forgot my stacks switch, but my esp doesnt for a bit and then I was trying to write to the new one with the old stack pointer, referencing variables from the old one, and then calling a function return with the new, with the frame still on the old. Therefore I did not do anything correctly

#

will probably have to rewrite this for the 5th time

weak karma Oct 23, 2025, 3:58 AM

#

Man I have made almost no progress ngl

#

Qemu crashes now

#

So yay I guess

weak karma Nov 2, 2025, 2:15 PM

#

I was accidentally enabling interrupts again in the middle of my process creation code

#

Figures the stack would be fucked

#

Welp now I'm just gonna figure out why sometimes it just forward an instruction with a fast timer speed

weak karma Nov 6, 2025, 2:29 PM

#

Yea so it's because of interrupt gates vs trap gates

#

Anyhow my memory allocator is bugged I found out, so I gotta go fix that

#

It doesn't not in fact use memory that was already used but freed under certain circumstances

#

But now I gotta do hw instead of working on this stuff

#

Story of the past 2 months

#

I actually don't know how to find time to do this

#

It's like once per week I get a chance and it's so sad :(

weak karma Nov 7, 2025, 6:34 PM

#

Forgot to set the new size of memory nodes when splitting nodes...

#

Time to rework my acpi table parsing now!

#

Then I can ACTUALLY for real start file systems

weak karma Dec 5, 2025, 3:58 AM

#

oop

#

did all that, now working on implementing ahci

#

got pci in a "working" state

#

pcie because im fancy B)

#

then yea gonna do ahci but its finals so not yet

#

idt the code is pushed either...

#

oh well

keen junco Dec 5, 2025, 6:19 AM

#

WOW

#

You are back

weak karma Dec 24, 2025, 10:57 PM

#

til you can segfault printf if you redefine malloc in the code

#

figured this out testing my realloc fn

#

took me a second to figure out this was the problem

#

god I love c

#

i should probably start posting little itty bitty code snippets in here sometimes huh

#

Someone might steal my horrible already on the internet code though!!!!!

weak karma Dec 27, 2025, 7:34 PM

#

SATA does not make any sense to me at all

#

I just cannot wrap my head around what any of the specification means

#

and then how I combine it all with AHCI

#

makes no sense whatsoever to me

weak karma Dec 28, 2025, 9:31 PM

#

Nvm it does I've just been putting virtual addresses in spots where they should have been physical

#

Like it says in the specs

#

I should probably read them more carefully or something but idc

weak karma Dec 30, 2025, 8:19 PM

#

AHCI is "working"

#

My MSIs are not

#

Idk why

#

Probably I don't have a bit set somewhere

#

Time to comb through the 2000000 manuals

#

Checking every single bit

weak karma Dec 31, 2025, 1:50 AM

#

okay the are working now

#

me when i dont set the pxie bits...

#

now i gotta have it go to the apic to generate an interrupt on write

weak karma Jan 1, 2026, 6:33 PM

#

okay hooray it works

#

read write time

weak karma Jan 3, 2026, 10:32 PM

#

Read command is sending but I'm not getting nothing in my data sad face emoji

weak karma Jan 4, 2026, 2:54 AM

#

i cant figure out why im not getting any data

#

all my bits are set correctly according to my understanding

#

which is obviosly wrong but dang

#

do I need to set something else up for DMA access

#

because i know there exists a DMA setup FIS

#

but from what I've read that shouldn't matter because it sets up the DMA on the first read request i think???

weak karma Jan 4, 2026, 7:31 PM

#

The read is going through but the data isn't being read

#

I know it's all happening because qemu tells me in the block stats

#

But like ts does not transfer over

#

I also know there is data because I had qemu copy the block to another file and there was more than 0s

weak karma Jan 4, 2026, 8:25 PM

#

Okay I can read it with command 0x20 which is regular sector read

#

But I really want DMA to be implemented

#

And it is supported im pretty sure

#

I will check again, but I know 48 bit addressing is

weak karma Jan 4, 2026, 9:05 PM

#

tspmo im gonna play runescape for a bit now

weak karma Jan 6, 2026, 4:33 AM

#

okay last thing imma try before i explode is reseting the port

#

but after that i honestly dont know what else to do

weak karma Jan 6, 2026, 8:39 PM

#

okay well I figured it out???

#

something about my prd table's last entry needs to have the MSB set whatever that means. something something I shifted it over 9 and now it can work???

#

This isn't specified in ANY of the specs I've read, maybe its in ATA-x but I dont have $70 to spend on a stupid pdf file and I cant find it anywhere online

weak karma Jan 15, 2026, 2:46 PM

#

Busy semester. This will all be on the farthest back simmer burner until it ends

#

They hate hobbies here

#

And non-required fun

weak karma Jan 28, 2026, 2:07 AM

#

This does not pertain to my os but I wanna yap about it

#

I saw a video on lots and lots of recursion

#

Trillions even

#

And the end got me thinking

#

Would it be possible to reduce a return address even farther by jumping into 16 bit mode of the cpu

#

But I don't think it would be because iirc there isn't paging for 16 bit and it's a whole debacle to jump into it again

#

You'd have to set up proper segmentation, although that's not tooo hard to do for just this purpose

#

Maybe this summer I make a kernel specifically tailored to generating the largest amount of recursion possible

#

Probably not, but maybe...

#

I mean you wouldn't need an allocator or anything and maybe you'd want apic parsing just to get a basic disk filesystem going

#

And maybe a compression algorithm although that wouldn't be too hard considering what data would be stored

#

Idk

#

Food for thought

iron hill Jan 28, 2026, 2:55 PM

#

who bro talking too 🙏, same i also do that but nice os

weak karma Jan 28, 2026, 8:20 PM

#

I talk to the void in here...

#

But that's okay because I love talking to the void

weak karma Feb 19, 2026, 2:12 PM

#

I tried pushing a 1gb file on accident

#

Then I couldn't figure out how to go backwards in commits

#

I spent two hours last night trying to recover everything and I think it's all bueno now

#

Now I know for next time!

weak karma Feb 19, 2026, 11:08 PM

#

Hmm my hardware has an ahci controller but idt it has anything connected to it

#

None of the ports seem to have anything going on

#

Or at least there are no active ports that don't have a signature 0xffffffff

#

There is an nvme controller though!

#

I suspect that is what the drive is connected to, but I really cba writing an nvme driver rn

mild socket Feb 19, 2026, 11:12 PM

#

weak karma I tried pushing a 1gb file on accident

relatable

weak karma Feb 24, 2026, 4:50 PM

#

I have been chipping away at implementing fat32 as my first filesystem

#

Oops

#

I'll just have incurable brain damage in my abstractions and the number of bad decisions I've made will increase by 1

#

Or actually more than 1

weak karma Mar 1, 2026, 7:42 AM

#

Working with the structs and offsets being weird

#

I forgot to give them packed attribute

#

Took me too long to figure out why I was getting nonsensical values from the data

#

Compiler just added padding

iron hill Mar 1, 2026, 11:36 AM

#

For some reason the cpu needs it to be "even"

weak karma Mar 1, 2026, 4:15 PM

#

Doesn't really need it to be aligned, it's just more efficient

#

But the compiler will make sure it is the case for efficiency sake and then it makes me sad

weak karma Mar 12, 2026, 4:19 AM

#

I've been reading the Abraham operating systems book and it's so cool to read about all the theory

#

I've been looking into file systems and idk there are just so many things I've never thought about that have huge impacts

#

And like they make sense after I read them but I haven't put much thought into it

#

Just as an easy off the dome example, ssd and hdd are best used with different file systems

#

Obviously this makes sense, but I've just never thought about it

#

#Word