EvalynOS | Operating System Development | Page 7

hazy saddle Dec 11, 2025, 8:51 AM

#

For USB I might have USB mass storage and USB keyboard and mouse in kernel

#

But other usb drivers would be userspace

#

And userspace will have a way to send and retrieve whatever it wants to usb ports

devout path Dec 11, 2025, 8:52 AM

#

hazy saddle For USB I might have USB mass storage and USB keyboard and mouse in kernel

ooh

hazy saddle Dec 11, 2025, 8:53 AM

#

So you can boot from USB and have usb input always without depending on userspace

devout path Dec 11, 2025, 8:54 AM

#

noicee yeah that woudl be nice

#

im close to giving up

#

i swear

#

aaaaaaaaaaaaaaaaaaaaaaaaaa

hazy saddle Dec 11, 2025, 8:57 AM

#

aaaaaaaa

#

Do
You have your collage friends that can look at this crap?

devout path Dec 11, 2025, 8:58 AM

#

they're all in vacation

hazy saddle Dec 11, 2025, 9:00 AM

#

Damn

#

Honestly @devout path

#

Do you wana just rewrite this from scratch?

#

And put this commit to shreds

#

https://tenor.com/view/futurama-farnsworth-shreds-to-shreds-you-say-gif-19143723

Tenor

#

It might be better to just recode this from scratch 💀

devout path Dec 11, 2025, 9:01 AM

#

hazy saddle Do you wana just rewrite this from scratch?

heh

#

the syscall?

hazy saddle Dec 11, 2025, 9:01 AM

#

The syscall stuff

#

I think the GDT part is fine?

devout path Dec 11, 2025, 9:01 AM

#

y e s

hazy saddle Dec 11, 2025, 9:01 AM

#

That can be cherry picked for a new attempt

#

And the MSRs

devout path Dec 11, 2025, 9:01 AM

#

can u send me a zip file of a clean without the syscal version

#

or is it in the git

hazy saddle Dec 11, 2025, 9:01 AM

#

Uhhhhhh

#

Ermmmmm

devout path Dec 11, 2025, 9:01 AM

#

hub

hazy saddle Dec 11, 2025, 9:01 AM

#

Uhhhhhhhhhh

devout path Dec 11, 2025, 9:02 AM

#

oh nvm

#

well iz fine

hazy saddle Dec 11, 2025, 9:02 AM

#

I never made a commit

devout path Dec 11, 2025, 9:02 AM

#

XD

#

weeel

hazy saddle Dec 11, 2025, 9:02 AM

#

I was gonna but

devout path Dec 11, 2025, 9:02 AM

#

this is why u make commits XD

hazy saddle Dec 11, 2025, 9:02 AM

#

I thought

#

Oh yeah easy 20 minute adventure

#

Just set some MSRs

devout path Dec 11, 2025, 9:02 AM

#

xd yes

hazy saddle Dec 11, 2025, 9:05 AM

#

@devout path this would mean you could probably fully PR this commit tbh

#

Because it’s a lot more of a minimal change your your doing most of it heh

devout path Dec 11, 2025, 9:06 AM

#

yeeeeeeeeeeeeeeeeeeaaaaaaaaaaaaaaaaaaaaaaaaahhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhh

#

i m eaaan ;-;

hazy saddle Dec 11, 2025, 9:06 AM

#

What

devout path Dec 11, 2025, 9:06 AM

#

yeah why not XD

devout path Dec 11, 2025, 9:06 AM

#

devout path i m eaaan ;-;

it'll take me a while if i don't get toolazy

hazy saddle Dec 11, 2025, 9:06 AM

#

Get your name in the open source

devout path Dec 11, 2025, 9:06 AM

#

hmm?

hazy saddle Dec 11, 2025, 9:06 AM

#

Or somthing I dunno heh

#

Just like your GitHub account would be linked to the commits

#

But yeah wtf is up with these bugs

devout path Dec 11, 2025, 9:07 AM

#

i have 0 clue

hazy saddle Dec 11, 2025, 9:07 AM

#

Like where the fuck in the code is causing it

devout path Dec 11, 2025, 9:08 AM

#

hooooooonestly

hazy saddle Dec 11, 2025, 9:08 AM

#

Maybe use a per thread global instead of gs @devout path and refer to it from assembly and for many cores make it an array

devout path Dec 11, 2025, 9:08 AM

#

hmmm yes yes

hazy saddle Dec 11, 2025, 9:08 AM

#

And then index by current core ID from reading the MSR that has that

#

Wait wait

#

@devout path what if you use the kernel base gs

devout path Dec 11, 2025, 9:09 AM

#

hmmmmmmmm

hazy saddle Dec 11, 2025, 9:09 AM

#

And change the write to the KBASEGS MSR

devout path Dec 11, 2025, 9:09 AM

#

could work

hazy saddle Dec 11, 2025, 9:09 AM

#

And then in the start of the kernel run a swapgs

#

@viral bison mentioned UBASEGS caused issues

devout path Dec 11, 2025, 9:11 AM

#

hmm ye syes

hazy saddle Dec 11, 2025, 9:12 AM

#

Plus the kernel using the KERNEL base gs would make more sense

devout path Dec 11, 2025, 9:12 AM

#

yeah true

hazy saddle Dec 11, 2025, 9:15 AM

#

@devout path is it working?

#

Or should I test that

devout path Dec 11, 2025, 9:17 AM

#

hazy saddle <@346576675183788034> is it working?

hmm?

#

wdym

#

can u test it for me pls

#

am not doin it rn i have life stuff xd

hazy saddle Dec 11, 2025, 9:17 AM

#

its not super paniing

#

#

shit

devout path Dec 11, 2025, 9:18 AM

#

all roads lead to albuquerqe

devout path Dec 11, 2025, 9:18 AM

#

hazy saddle shit

wat

hazy saddle Dec 11, 2025, 9:18 AM

#

what

#

if i rever it

#

its still 0s

#

??

#

#

#

#

wait

devout path Dec 11, 2025, 9:19 AM

#

hhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhuh

hazy saddle Dec 11, 2025, 9:19 AM

#

no yeah

devout path Dec 11, 2025, 9:20 AM

#

errrr

hazy saddle Dec 11, 2025, 9:21 AM

#

btw whys 0x06 on the IP
CR2=0x0000000000000006

#

of all things

#

f
mov [gs:32], rsp
mov rsp, [gs:24]
...
mov [gs:24], rsp
mov rsp, [gs:32]

#

this is correct rigjt

devout path Dec 11, 2025, 9:22 AM

#

hazy saddle f mov [gs:32], rsp mov rsp, [gs:24] ... mov [gs:24], rsp mov...

yes

#

but

#

h u h

hazy saddle Dec 11, 2025, 9:22 AM

#

w h a

devout path Dec 11, 2025, 9:22 AM

#

hazy saddle btw whys 0x06 on the IP CR2=0x0000000000000006

honestly atp

#

i have no clue whatsover

hazy saddle Dec 11, 2025, 9:23 AM

#

is there anyone we know that can help

devout path Dec 11, 2025, 9:23 AM

#

heh ;-;

#

idk

rancid viper Dec 11, 2025, 9:24 AM

#

whats up

hazy saddle Dec 11, 2025, 9:27 AM

#

SYSRET is a companion instruction to the SYSCALL instruction. It returns from an OS system-call handler to user code at privilege level 3. It does so by loading RIP from RCX and loading RFLAGS from R11.1 With a 64-bit operand size, SYSRET remains in 64-bit mode; otherwise, it enters compatibility mode and only the low 32 bits of the registers are loaded.
hm @devout path

devout path Dec 11, 2025, 9:27 AM

#

hazy saddle `SYSRET is a companion instruction to the SYSCALL instruction. It returns from a...

hmmmmmmmm

hazy saddle Dec 11, 2025, 9:27 AM

#

SYSCALL invokes an OS system-call handler at privilege level 0. It does so by loading RIP from the IA32_LSTAR MSR (after saving the address of the instruction following SYSCALL into RCX). (The WRMSR instruction ensures that the IA32_LSTAR MSR always contain a canonical address.)

SYSCALL also saves RFLAGS into R11 and then masks RFLAGS using the IA32_FMASK MSR (MSR address C0000084H); specifically, the processor clears in RFLAGS every bit corresponding to a bit that is set in the IA32_FMASK MSR.

#

do we clobber RCX anywhere

#

or R11

devout path Dec 11, 2025, 9:28 AM

#

wat

#

in the kernel?

hazy saddle Dec 11, 2025, 9:28 AM

#

yes

devout path Dec 11, 2025, 9:28 AM

#

i have 0 clue

#

XD

#

i mean

#

isn't that what syscall should do

hazy saddle Dec 11, 2025, 9:29 AM

#

if i change from sysretq to sysret i fer a diffrent error

devout path Dec 11, 2025, 9:30 AM

#

wat error

hazy saddle Dec 11, 2025, 9:30 AM

#

this is sysretq

#

normal sysret

devout path Dec 11, 2025, 9:31 AM

#

huh

#

isn't sysretq explicit for x64

#

sysretq

hazy saddle Dec 11, 2025, 9:31 AM

#

idk

devout path Dec 11, 2025, 9:32 AM

#

hazy saddle normal sysret

is this returningfrom a 32bit syscall?

#

h u h ?

hazy saddle Dec 11, 2025, 9:32 AM

#

no

#

i dont think s o

devout path Dec 11, 2025, 9:32 AM

#

then y is it different... <

#

https://tenor.com/view/azumanga-daiho-azumanga-osaka-ayumu-kasuga-gif-14774670791157440244

Tenor

#

i refactored the syscall a bit but

rancid viper Dec 11, 2025, 9:33 AM

#

sysretq is not valid

#

nasm thinks its a label u fucked up

#

so it outputs nothing

devout path Dec 11, 2025, 9:33 AM

#

true..

rancid viper Dec 11, 2025, 9:34 AM

#

u really should run with -Werror

#

o64 sysret

hazy saddle Dec 11, 2025, 9:35 AM

#

src/kernel/interupts/syscall.asm:55: error: label sysretq' alone on a line without a colon might be in error [-w+error=label-orphan]`

#

h u h

#

n e a t

rancid viper Dec 11, 2025, 9:35 AM

#

yes

devout path Dec 11, 2025, 9:35 AM

#

hazy saddle `src/kernel/interupts/syscall.asm:55: error: label `sysretq' alone on a line wit...

neat

hazy saddle Dec 11, 2025, 9:35 AM

#

whats these

src/kernel/interupts/interupts.asm:38: error: 32-bit relative section-crossing relocation [-w+error=reloc-rel-dword]
make[1]: *** [GNUmakefile:235: obj-x86_64/src/kernel/interupts/interupts.asm.o] Error 1
make[1]: *** Waiting for unfinished jobs....
src/kernel/interupts/syscall.asm:32: error: 32-bit relative section-crossing relocation [-w+error=reloc-rel-dword]
make[1]: *** [GNUmakefile:235: obj-x86_64/src/kernel/interupts/syscall.asm.o] Error 1

#

call dispatch_interupt

devout path Dec 11, 2025, 9:36 AM

#

devout path i refactored the syscall a bit but

doesn't work

#

prolly has dsometing to do with that

viral bison Dec 11, 2025, 9:38 AM

#

hazy saddle <@573903151472836609> mentioned UBASEGS caused issues

idk if that is actually the prob

viral bison Dec 11, 2025, 9:39 AM

#

hazy saddle n e a t

works now?

hazy saddle Dec 11, 2025, 9:39 AM

#

fixing the assembly

viral bison Dec 11, 2025, 9:39 AM

#

what was wrong with it?

hazy saddle Dec 11, 2025, 9:39 AM

#

aparently i need this

#

because

#

viral bison Dec 11, 2025, 9:40 AM

#

why not just call dispatch_int?

hazy saddle Dec 11, 2025, 9:40 AM

#

then it gives the error

#

nope still errors pain

rancid viper Dec 11, 2025, 9:40 AM

#

how do you compile your assembly

hazy saddle Dec 11, 2025, 9:41 AM

#

nasm

rancid viper Dec 11, 2025, 9:41 AM

#

yes.....

hazy saddle Dec 11, 2025, 9:41 AM

#

viral bison works now?

also no if igore the warnings it just errors diffrently

#

#

devout path Dec 11, 2025, 9:41 AM

#

hazy saddle aparently i need this

cld
mov rdi, rsp
call dispatch_interrupt

#

no?

hazy saddle Dec 11, 2025, 9:42 AM

#

it errors

#

src/kernel/interupts/syscall.asm:32: error: 32-bit relative section-crossing relocation [-w+error=reloc-rel-dword]

#

and the other way will too
64-bit absolute section-crossing relocation [-w+error=reloc-abs-qword]

#

pain

devout path Dec 11, 2025, 9:42 AM

#

hazy saddle `src/kernel/interupts/syscall.asm:32: error: 32-bit relative section-crossing re...

what is ur line 32

rancid viper Dec 11, 2025, 9:43 AM

#

are you like putting this in some other section

hazy saddle Dec 11, 2025, 9:43 AM

#

call execute_syscall

#

what section

#

what is a section and where would it be going

rancid viper Dec 11, 2025, 9:43 AM

#

elf section

hazy saddle Dec 11, 2025, 9:43 AM

#

default?

rancid viper Dec 11, 2025, 9:43 AM

#

default should be text

#

which is correct

hazy saddle Dec 11, 2025, 9:45 AM

#

it works if i ignore errors for interupts but not syscalls

#

but

devout path Dec 11, 2025, 9:51 AM

#

imma be back in an hour

#

imma just burn stress

hazy saddle Dec 11, 2025, 9:53 AM

#

Same

viral bison Dec 11, 2025, 9:55 AM

#

@hazy saddle try to compile your OS with clang

#

assembly too

#

maybe it'll work better by using just 1 compiler for everything

rancid viper Dec 11, 2025, 10:13 AM

#

in general you need to narrow the issue

#

it really seems you guys are just searching in the dark

viral bison Dec 11, 2025, 10:14 AM

#

well, we did find that it is somehow jumping to rdi

#

when it's pushed to the stack

#

or a bit later

hazy saddle Dec 11, 2025, 10:16 AM

#

viral bison assembly too

Can clang compile NASM code?

viral bison Dec 11, 2025, 10:16 AM

#

yes, ofc

hazy saddle Dec 11, 2025, 10:16 AM

#

What flags should I change in the makefile

viral bison Dec 11, 2025, 10:16 AM

#

sec, i'm finishing to edit the makefile

#

then i'll send it here

rancid viper Dec 11, 2025, 10:20 AM

#

viral bison yes, ofc

no

#

nasm syntax does happen to be very close to intel syntax so it might work

viral bison Dec 11, 2025, 10:21 AM

#

yeah i found something that needs fixing for it

rancid viper Dec 11, 2025, 10:21 AM

#

and why change assemblers

viral bison Dec 11, 2025, 10:21 AM

#

#

i tought the syntax was the same

rancid viper Dec 11, 2025, 10:21 AM

#

its not

hazy saddle Dec 11, 2025, 10:24 AM

#

rancid viper and why change assemblers

I dunno

#

It’s diamonds idea

viral bison Dec 11, 2025, 10:25 AM

#

lemme finish to convert them

rancid viper Dec 11, 2025, 10:25 AM

#

well ill just put it this way

#

if changing assemblers fixes something

#

there is some deeper issue

#

doesnt seem very productive

hazy saddle Dec 11, 2025, 10:35 AM

#

viral bison well, we did find that it is somehow jumping to rdi

Does this happen with using the proper sysret instruction?

viral bison Dec 11, 2025, 10:36 AM

#

idk

hazy saddle Dec 11, 2025, 10:36 AM

#

Instead of whatever the fuck sysretq is

#

Maybe we just ignore the 32bit whatever NASM error garbage is

floral kettle Dec 11, 2025, 10:36 AM

#

sysretq is what you should use (or in nasm o64 sysret, though sysretq probably works in nasm too idr)

#

because by default sysret is going to return to compatibility mode if you don't do that

hazy saddle Dec 11, 2025, 10:36 AM

#

h u h

#

sysretq did not work in nasm

#

it thought it was a lebel

floral kettle Dec 11, 2025, 10:38 AM

#

ah then ig it only supports the o64 sysret variant

hazy saddle Dec 11, 2025, 10:38 AM

#

how do i fix this aswell?
src/kernel/interupts/interupts.asm:40: error: 32-bit relative section-crossing relocation [-w+error=reloc-rel-dword]

#

so i need another o64 on iretq?

floral kettle Dec 11, 2025, 10:38 AM

#

probably yeah o64 iret

hazy saddle Dec 11, 2025, 10:39 AM

#

nope

src/kernel/interupts/syscall.asm:32: error: 32-bit relative section-crossing relocation [-w+error=reloc-rel-dword]
src/kernel/interupts/interupts.asm:40: error: 32-bit relative section-crossing relocation [-w+error=reloc-rel-dword]

#

both on the calls to c functions

#

leme just mute that

#

this is new

#

#

looks like its executing thr wrong systemcall

#

eg this heh

rancid viper Dec 11, 2025, 10:41 AM

#

floral kettle sysretq is what you should use (or in nasm `o64 sysret`, though `sysretq` probab...

I once spent a full day on debugging why my sysret didnt work. turns out nasm interpreted it as an incorrect label and threw a warning I didnt see

#

moral of the story is werror

hazy saddle Dec 11, 2025, 10:43 AM

#

GPF here now @devout path @viral bison

0xffffffff8000c1ea
dispatch_interupt_asm
src/kernel/interupts/interupts.asm:?

viral bison Dec 11, 2025, 10:43 AM

#

hmmm

rancid viper Dec 11, 2025, 10:44 AM

#

whats the error code of the gpf

hazy saddle Dec 11, 2025, 10:44 AM

#

ERROR : 00000000 00000000 00000000 00010000

viral bison Dec 11, 2025, 10:44 AM

#

your IDT is so fucked btw

hazy saddle Dec 11, 2025, 10:44 AM

#

hm?

viral bison Dec 11, 2025, 10:44 AM

#

i'm fixing all of it

hazy saddle Dec 11, 2025, 10:44 AM

#

heh

viral bison Dec 11, 2025, 10:44 AM

#

the IDT entry is completely wrong

hazy saddle Dec 11, 2025, 10:45 AM

#

bwuh

rancid viper Dec 11, 2025, 10:45 AM

#

hazy saddle ``` ERROR : 00000000 00000000 00000000 00010000 ```

I dont believe you

#

give me qemu log

hazy saddle Dec 11, 2025, 10:45 AM

#

#

leme generate

rancid viper Dec 11, 2025, 10:45 AM

#

I dont trust ur print

#

that is not a valid error code for a gpf

hazy saddle Dec 11, 2025, 10:45 AM

#

heh

#

ooo my stack tarce worked for once here kinda

#

#

in debug mode it dosnt do tht garbage

#

@rancid viper

📎 log.txt

rancid viper Dec 11, 2025, 10:48 AM

#

yeah look at that

#

0x10

hazy saddle Dec 11, 2025, 10:48 AM

#

whats 0x10?

rancid viper Dec 11, 2025, 10:49 AM

#

2nd index of the gdt

#

is what that error code means

hazy saddle Dec 11, 2025, 10:49 AM

#

ah

rancid viper Dec 11, 2025, 10:49 AM

#

so segment error on index 2 of the gdt

hazy saddle Dec 11, 2025, 10:49 AM

#

gdt_fill_entry (2, 0x92, 0xC); // 64 Bit Kernel Data | 0x10
hm

rancid viper Dec 11, 2025, 10:50 AM

#

can you dump the gdt

hazy saddle Dec 11, 2025, 10:50 AM

#

how

rancid viper Dec 11, 2025, 10:51 AM

#

whatever is convenient

#

qemu monitor?

hazy saddle Dec 11, 2025, 10:51 AM

#

i dont know any methods

#

what do i enter there

viral bison Dec 11, 2025, 10:51 AM

#

also, why were you using 32 bit flags for your IDT?

hazy saddle Dec 11, 2025, 10:51 AM

#

etc etc

hazy saddle Dec 11, 2025, 10:51 AM

#

viral bison also, why were you using 32 bit flags for your IDT?

wut

rancid viper Dec 11, 2025, 10:51 AM

#

tbh this would be much easier if u just used constants for the gdt

#

xd

viral bison Dec 11, 2025, 10:51 AM

#

0x8E is for 32 bit

#

64 bit is 0xE

#

same for the F

hazy saddle Dec 11, 2025, 10:51 AM

#

is that not for user callable interupts?

viral bison Dec 11, 2025, 10:51 AM

#

no

hazy saddle Dec 11, 2025, 10:51 AM

#

yaaar

#

how did it even fucking work then

viral bison Dec 11, 2025, 10:52 AM

#

DPL is for that

#

or well, selector in your case

rancid viper Dec 11, 2025, 10:52 AM

#

viral bison 0x8E is for 32 bit

?????

viral bison Dec 11, 2025, 10:52 AM

#

#

rancid viper Dec 11, 2025, 10:53 AM

#

send me a link to their idt coide

viral bison Dec 11, 2025, 10:54 AM

#

it's in the zip

#

there is no link

#

lol

rancid viper Dec 11, 2025, 10:54 AM

#

jesus im not doing allat

#

anyway at least I have a struct that looks like this

#

viral bison Dec 11, 2025, 10:55 AM

#

yes it is

rancid viper Dec 11, 2025, 10:55 AM

#

correct value for an interrupt gate here is to set flags to 0x8E

viral bison Dec 11, 2025, 10:55 AM

#

but how the flags are set was wrong

#

huh?

rancid viper Dec 11, 2025, 10:55 AM

#

viral bison

this describes the gate type

#

which is one flag

viral bison Dec 11, 2025, 10:56 AM

#

ah wait

rancid viper Dec 11, 2025, 10:56 AM

#

flags covers present, dpl, gate type

viral bison Dec 11, 2025, 10:56 AM

#

i split the DPL that's why

rancid viper Dec 11, 2025, 10:56 AM

#

bit 7 is present

#

bit 7 of a byte is 0x80

#

0x80 | 0xE = 0x8E

rancid viper Dec 11, 2025, 10:56 AM

#

viral bison i split the DPL that's why

yeah

#

I dont use bit fields anywhere

hazy saddle Dec 11, 2025, 11:02 AM

#

viral bison there is no link

in mean its on github too

#

my patchset isnt touching idt

viral bison Dec 11, 2025, 11:02 AM

#

ehm

#

fuck

hazy saddle Dec 11, 2025, 11:06 AM

#

hazy saddle this is new

Well because the random error code it gave I had to make the quick edit heh

#

And because some how the error code is the only thing that did not break

viral bison Dec 11, 2025, 11:07 AM

#

IT WORKS

#

well, not really works

#

but at least it doesn't pagefault anymore

hazy saddle Dec 11, 2025, 11:08 AM

#

What’s the fault now

#

GPF?

viral bison Dec 11, 2025, 11:08 AM

#

triple fault after it runs a bit

hazy saddle Dec 11, 2025, 11:08 AM

#

But bad apple runs?

viral bison Dec 11, 2025, 11:08 AM

#

prob something wrong when executing syscalls

viral bison Dec 11, 2025, 11:08 AM

#

hazy saddle But bad apple runs?

it spams the console

hazy saddle Dec 11, 2025, 11:08 AM

#

viral bison triple fault after it runs a bit

https://tenor.com/view/wii-wii-sports-wii-bowling-bowling-bowling-strike-gif-24891879

Tenor

viral bison Dec 11, 2025, 11:08 AM

#

i think it's your printf

devout path Dec 11, 2025, 11:08 AM

#

hazy saddle https://tenor.com/view/wii-wii-sports-wii-bowling-bowling-bowling-strike-gif-248...

lamao

#

i go back

#

then i see this XD

#

wild

hazy saddle Dec 11, 2025, 11:08 AM

#

Probaly the printf systemcall

hazy saddle Dec 11, 2025, 11:09 AM

#

viral bison i think it's your printf

It’s probably getting a non null terminated string

viral bison Dec 11, 2025, 11:10 AM

#

enjoy

📎 evalynOS.zip

devout path Dec 11, 2025, 11:11 AM

#

wat

#

is it fixed

hazy saddle Dec 11, 2025, 11:11 AM

#

TF still maybe

viral bison Dec 11, 2025, 11:11 AM

#

Somewhat

#

It does run a bit tho

#

So it might be some syscall impl

#

Ans it uses clang now

hazy saddle Dec 11, 2025, 11:12 AM

#

i hate intel syntax

#

the macros are disgusting

#

your zip seems to be bad

#

viral bison Dec 11, 2025, 11:16 AM

#

Hmm

#

Lemme try tar

hazy saddle Dec 11, 2025, 11:17 AM

#

you also forgot all the dot files

#

like .git and .gitignore

viral bison Dec 11, 2025, 11:17 AM

#

I told it to include everything

#

F

rancid viper Dec 11, 2025, 11:18 AM

#

zip -r dir dir

#

: )

hazy saddle Dec 11, 2025, 11:18 AM

#

what else did you change besides the IDT btw

viral bison Dec 11, 2025, 11:18 AM

#

Most of the asm files

rancid viper Dec 11, 2025, 11:19 AM

#

git diff HEAD~1 HEAD

hazy saddle Dec 11, 2025, 11:19 AM

#

@viral bison gimme a patch file to apply to main

devout path Dec 11, 2025, 11:19 AM

#

ima be back tmr i forgor i have a shit ton of things todo

#

if this not fixed till tmr im gonna try

viral bison Dec 11, 2025, 11:20 AM

#

hazy saddle <@573903151472836609> gimme a patch file to apply to main

try this

📎 evalynos.tar.gz

hazy saddle Dec 11, 2025, 11:20 AM

#

#

yaaar

viral bison Dec 11, 2025, 11:21 AM

#

yeah well

#

that might be bc of the archive you gave me

hazy saddle Dec 11, 2025, 11:21 AM

#

huh

#

can you should me the diffs of the file in your ide

viral bison Dec 11, 2025, 11:21 AM

#

after i unzipped it, it showed all the files edited

#

📎 patchesDiam.patch

#

patch too

#

the edits seems to be in the file modes

#

oh well

rancid viper Dec 11, 2025, 11:22 AM

#

WSL?

viral bison Dec 11, 2025, 11:22 AM

#

no

#

i'm on arch

rancid viper Dec 11, 2025, 11:22 AM

#

evalyn? wsl?

devout path Dec 11, 2025, 11:22 AM

#

who uses wsl nowadays

rancid viper Dec 11, 2025, 11:22 AM

#

: )

hazy saddle Dec 11, 2025, 11:22 AM

#

rancid viper evalyn? wsl?

CachyOS

rancid viper Dec 11, 2025, 11:22 AM

#

well

#

someones got fucked up shit

#

lol

hazy saddle Dec 11, 2025, 11:22 AM

#

hmmCat

#

yaaar

#

I’ve been awake a few hours

#

And il already sleepy as fuck

devout path Dec 11, 2025, 11:23 AM

#

rancid viper someones got fucked up shit

wat wdym

viral bison Dec 11, 2025, 11:23 AM

#

It should work anyway

devout path Dec 11, 2025, 11:23 AM

#

hazy saddle And il already sleepy as fuck

that's the stress

#

take nap it 'll go away

hazy saddle Dec 11, 2025, 11:23 AM

#

Okay okay

#

New plan

#

https://tenor.com/view/panda-flip-desk-freak-out-mad-gif-15503485

Tenor

#

Trash this code

#

And get a minimum working code

viral bison Dec 11, 2025, 11:24 AM

#

And use the one i sent?

hazy saddle Dec 11, 2025, 11:24 AM

#

Il be using bits from all of these

#

What was wrong with my idt

#

I want to keep NASM

viral bison Dec 11, 2025, 11:25 AM

#

Nasm is pain

hazy saddle Dec 11, 2025, 11:25 AM

#

The syntax is better for my mush brain

devout path Dec 11, 2025, 11:25 AM

#

wait why idt change

viral bison Dec 11, 2025, 11:25 AM

#

I used macros to auto gen all the 256 interrupts

#

And then get the address of the handlers from an array

#

Set via asm

#

some exception flags were wrong

hazy saddle Dec 11, 2025, 11:26 AM

#

I think the other way is fine and more explicit imo?

hazy saddle Dec 11, 2025, 11:26 AM

#

viral bison + some exception flags were wrong

These need to be changed

#

I mean take those fixes

viral bison Dec 11, 2025, 11:26 AM

#

hazy saddle I think the other way is fine and more explicit imo?

It would make a very long file

#

It's worse the way it was before

rancid viper Dec 11, 2025, 11:27 AM

#

I'd help but I am currently working so 🙂

hazy saddle Dec 11, 2025, 11:28 AM

#

I’m gonna take a break

#

And play a game maybe

#

Maybe eat food

#

I haven’t eaten

#

In today

devout path Dec 11, 2025, 11:28 AM

#

tbh

#

i suggest everyone to take a break from this os for a day XD

#

this made my brain mush more than it was

rancid viper Dec 11, 2025, 11:29 AM

#

you should keep an up-to-date dev branch

#

because these patches and zip files makes it just unecessarily difficult to help

devout path Dec 11, 2025, 11:29 AM

#

rancid viper you should keep an up-to-date dev branch

@hazy saddle pls :D

hazy saddle Dec 11, 2025, 11:29 AM

#

true

devout path Dec 11, 2025, 11:30 AM

#

u can commit everytime u do a minor change

hazy saddle Dec 11, 2025, 11:30 AM

#

What do I name it

devout path Dec 11, 2025, 11:30 AM

#

so i can develop on top of that with the block driver that im making

rancid viper Dec 11, 2025, 11:30 AM

#

dev

devout path Dec 11, 2025, 11:30 AM

#

make it weedbranch

hazy saddle Dec 11, 2025, 11:30 AM

#

true

rancid viper Dec 11, 2025, 11:30 AM

#

or something

#

lmao

devout path Dec 11, 2025, 11:30 AM

#

weeddev

#

cuz it'll be high

#

yes

#

most of the time

rancid viper Dec 11, 2025, 11:30 AM

#

then you can just cherry-pick and rebase everything u want to main

#

like I can understand wanting to keep a clean main

#

but this current system aint it

hazy saddle Dec 11, 2025, 11:31 AM

#

Yeahhhh

#

What if I call it unstable?

rancid viper Dec 11, 2025, 11:31 AM

#

it is your thing

#

call it what u want 😄

devout path Dec 11, 2025, 11:31 AM

#

yes

hazy saddle Dec 11, 2025, 11:31 AM

#

true

devout path Dec 11, 2025, 11:31 AM

#

https://tenor.com/view/osaka-spin-azumanga-gif-1737943182099768947

Tenor

#

https://tenor.com/view/azumanga-osaka-happy-osaka-azumanga-gif-27497987

Tenor

hazy saddle Dec 11, 2025, 11:32 AM

#

Il call it dev

devout path Dec 11, 2025, 11:32 AM

#

hazy saddle Il call it dev

perfect

hazy saddle Dec 11, 2025, 11:32 AM

#

All commits Goto dev

#

And when it’s time for main

#

Il squash and then merge into main

devout path Dec 11, 2025, 11:32 AM

#

iz it fine if u make me a collaborator

hazy saddle Dec 11, 2025, 11:33 AM

#

true

#

I should

devout path Dec 11, 2025, 11:33 AM

#

yez

rancid viper Dec 11, 2025, 11:33 AM

#

prs exist too 🙂

hazy saddle Dec 11, 2025, 11:33 AM

#

Can I give you perms on the dev branch but not main?

devout path Dec 11, 2025, 11:33 AM

#

yes

#

ofc

rancid viper Dec 11, 2025, 11:33 AM

#

on a public thingh

#

yes

#

for private u need to pay

hazy saddle Dec 11, 2025, 11:33 AM

#

It’s public

rancid viper Dec 11, 2025, 11:33 AM

#

yep

#

basically u just setup branch protection for mai

#

n

hazy saddle Dec 11, 2025, 11:34 AM

#

Okay I’m gonna get food and then get a commit with the gdt changes that were good

#

And then get the somewhat not working syscalls in dev

hazy saddle Dec 11, 2025, 11:34 AM

#

hazy saddle Okay I’m gonna get food and then get a commit with the gdt changes that were goo...

(In dev)

#

And then get @devout path and @viral bison as collabs on the dev branch

devout path Dec 11, 2025, 11:35 AM

#

yes

#

pls don't foret to make it always up to date though ;-;

#

so i can alwys check it

hazy saddle Dec 11, 2025, 11:36 AM

#

I am making the executive decision to stick with NASM

#

As it’s what I’m used to and I prefer the syntax

hazy saddle Dec 11, 2025, 11:36 AM

#

devout path pls don't foret to make it always up to date though ;-;

Any changes I make il be sending to dev instead of a patch file

#

And then once we get a feature working it gets squashed to main

devout path Dec 11, 2025, 11:37 AM

#

yes

#

good devops

#

xd

#

i ain't a devops engineer but that sounds great

hazy saddle Dec 11, 2025, 11:38 AM

#

Also you want to get added to my personal dev server? so I can setup a bot with git notifications and for other generic chatting that may not be good for here lmao

#

Possibly helpful hmmCat

devout path Dec 11, 2025, 11:39 AM

#

yes yes

hazy saddle Dec 11, 2025, 11:39 AM

#

devout path i ain't a devops engineer but that sounds great

And later il have an unstable branch once main is stable

#

So new features go there instead of main

#

Just incase

#

Then things go dev > unstable > main

devout path Dec 11, 2025, 11:39 AM

#

hazy saddle Then things go dev > unstable > main

yes yes perfefct

hazy saddle Dec 11, 2025, 11:40 AM

#

I think that milestone should be core utils + shell + nano + doom and ext2 and sata

devout path Dec 11, 2025, 11:41 AM

#

yes yes

#

@viral bison

#

would u like to be a part of the

#

dev server

viral bison Dec 11, 2025, 11:45 AM

#

devout path dev server

oh, there is a dev server?

#

ok then

#

@hazy saddle

hazy saddle Dec 11, 2025, 11:47 AM

#

viral bison <@618537432002068491>

yes yes

#

dm me if you want the inv

#

its for dev notis and non evalynos chat because yall friends :3

devout path Dec 11, 2025, 11:52 AM

#

hazy saddle its for dev notis and non evalynos chat because yall friends :3

yippie

hazy saddle Dec 11, 2025, 1:16 PM

#

https://github.com/EvalynGoemer/evalynOS/tree/dev

GitHub

GitHub - EvalynGoemer/evalynOS at dev

A WIP 64 Bit Operating System. Contribute to EvalynGoemer/evalynOS development by creating an account on GitHub.

#

new proper dev branch for changes that may ior may not be buggy

rancid viper Dec 11, 2025, 1:19 PM

#

the other option (which you cant do atm) is to fork and upstream

#

but that might be excessive for a personal project

hazy saddle Dec 11, 2025, 2:12 PM

#

@rancid viper progress and im about to push to dev

#

a few syscalls can be ran before death

#

and they are ones i expect

#

it is that invalid GPF error code

#

leme make a quick qemu log file

#

📎 qemu_log.txt

rancid viper Dec 11, 2025, 2:19 PM

#

something is wrong w ur error print

hazy saddle Dec 11, 2025, 2:20 PM

#

    if (frame != NULL) {
        printf("ERROR : " BYTE_TO_BINARY_PATTERN " " BYTE_TO_BINARY_PATTERN
        " " BYTE_TO_BINARY_PATTERN " " BYTE_TO_BINARY_PATTERN " \n",
        BYTE_TO_BINARY(frame->error >> 24), BYTE_TO_BINARY(frame->error > 16),
               BYTE_TO_BINARY(frame->error >> 8), BYTE_TO_BINARY(frame->error));
    }

rancid viper Dec 11, 2025, 2:20 PM

#

yeah error is still 0x10

hazy saddle Dec 11, 2025, 2:20 PM

#

but some systemcalls do get to be ran

rancid viper Dec 11, 2025, 2:20 PM

#

oh wtf it's binary

hazy saddle Dec 11, 2025, 2:20 PM

#

yes

rancid viper Dec 11, 2025, 2:20 PM

#

sadist

#

either way

#

error code still points at a segment error on gdt index 2

hazy saddle Dec 11, 2025, 2:21 PM

#

hmmCat

#

What happens if limit is 0 or max Fs

rancid viper Dec 11, 2025, 2:22 PM

#

limit is ignored on x86-64

hazy saddle Dec 11, 2025, 2:22 PM

#

gdt_fill_entry (2, 0x92, 0xC); // 64 Bit Kernel Data | 0x10

rancid viper Dec 11, 2025, 2:22 PM

#

why don't u make them constants

hazy saddle Dec 11, 2025, 2:23 PM

#

i dunno how

rancid viper Dec 11, 2025, 2:23 PM

#

I'm on my phone atm and can't really investigate what that function does

#

xd

hazy saddle Dec 11, 2025, 2:23 PM

#

rancid viper Dec 11, 2025, 2:23 PM

#

hazy saddle i dunno how

like u are calling a function that writes a value calculated from constants

hazy saddle Dec 11, 2025, 2:23 PM

#

#

it does this

hazy saddle Dec 11, 2025, 2:29 PM

#

rancid viper like u are calling a function that writes a value calculated from constants

Basically right?

rancid viper Dec 11, 2025, 2:30 PM

#

wait ur flags are weird

hazy saddle Dec 11, 2025, 2:30 PM

#

huh

rancid viper Dec 11, 2025, 2:30 PM

#

0xC

#

bits 2, 3

hazy saddle Dec 11, 2025, 2:30 PM

#

yes

rancid viper Dec 11, 2025, 2:30 PM

#

so granularity

#

ok

hazy saddle Dec 11, 2025, 2:31 PM

#

i took the flags and accsess values from here

rancid viper Dec 11, 2025, 2:31 PM

#

uh bit 2 I don't think should be set

#

hm

#

maybe I'm missing something

floral kettle Dec 11, 2025, 2:33 PM

#

bit 2 of the flags for data segment should be set, its the 16-bit vs 32-bit data bit

#

and bit 2 of the flags for 64-bit code segment should be cleared (and instead bit 1 set)

hazy saddle Dec 11, 2025, 2:34 PM

#

this is a 64bit data segment

floral kettle Dec 11, 2025, 2:35 PM

#

there is no such thing as a 64-bit data segment, it has the same format as a 32-bit data segment

hazy saddle Dec 11, 2025, 2:35 PM

#

ah

#

so this is correct then?

rancid viper Dec 11, 2025, 2:37 PM

#

floral kettle and bit 2 of the flags for 64-bit code segment should be cleared (and instead bi...

bit 1 is only for code segments no?

#

afaik bit 0, 1, 2 should be clear for a 64bit data segment?

#

oh I misread ur thing

#

lol

floral kettle Dec 11, 2025, 2:38 PM

#

rancid viper bit 1 is only for code segments no?

yes

#

lol

rancid viper Dec 11, 2025, 2:39 PM

#

Damn I have flags set to zero for my data segments

#

idk

floral kettle Dec 11, 2025, 2:39 PM

#

idk if it matters in 64-bit tbh, it might not

hazy saddle Dec 11, 2025, 2:39 PM

#

what bits should 0x10 have on flags

rancid viper Dec 11, 2025, 2:39 PM

#

floral kettle idk if it matters in 64-bit tbh, it might not

granularity at least prob doesnt? since limit and base are ignored

rancid viper Dec 11, 2025, 2:40 PM

#

hazy saddle what bits should 0x10 have on flags

no u have it correct

#

qwinci corrected me

hazy saddle Dec 11, 2025, 2:40 PM

#

ah

#

then whats up with the GPF

rancid viper Dec 11, 2025, 2:40 PM

#

well start by dumping the instructions where it happened so we know at least what it was trying to do

hazy saddle Dec 11, 2025, 2:43 PM

#

i dont think this is the right IP?

#

the IP lands somwhere here

#

wait no

#

its just a random match

#

 evalyn@EvalynPC  ~/Documents/Programming/C/evalynOS   master ±✚  addr2line -fai -e  kernel.elf 0xffffffff8000c1ea
0xffffffff8000c1ea
dispatch_interupt_asm
src/kernel/interupts/interupts.asm:?

#

but yeah the IP seems widly off

rancid viper Dec 11, 2025, 2:48 PM

#

https://github.com/elysium-os/elysium/blob/main/tools%2Fdump.py

#

modify this or something

#

and dump like a couple bytes before the rip

hazy saddle Dec 11, 2025, 2:53 PM

#

https://github.com/EvalynGoemer/evalynOS/tree/dev

GitHub

GitHub - EvalynGoemer/evalynOS at dev

A WIP 64 Bit Operating System. Contribute to EvalynGoemer/evalynOS development by creating an account on GitHub.

#

pushed to dev

#

here we go

rancid viper Dec 11, 2025, 3:12 PM

#

ah

#

well can u dump the stack at the time of exception

#

most likely something is wrong with the iret frame ig

hazy saddle Dec 11, 2025, 3:29 PM

#

Is there a good way to set a breakpoint there but only have it trigger when it’s about to do the GPF?

rancid viper Dec 11, 2025, 3:34 PM

#

u can just dump using qemu monitor after the fact

#

I don't think it consumes the frame if u fault

#

thought not 100% sure on that one

hazy saddle Dec 11, 2025, 3:42 PM

#

#

#

@rancid viper is this what you needed

rancid viper Dec 11, 2025, 4:03 PM

#

I'll look later

#

still at work need to finish some shit

hazy saddle Dec 11, 2025, 4:04 PM

#

rancid viper still at work need to finish some shit

Completely fair

rancid viper Dec 11, 2025, 6:57 PM

#

alright finally

#

9pm

#

got off work

hazy saddle Dec 11, 2025, 7:58 PM

#

nice!

#

i got CPUID working in the meantime

#

And if on physical hardware the frame buffer will get WC enabled to speed it up

hazy saddle Dec 11, 2025, 9:58 PM

#

hmmmmmmmmmmmmmmmm

#

huh wtf

#

between the printf and the switch case

#

the stack is fucked

#

#

its either nanoprintf or flantern maybe?

hazy saddle Dec 11, 2025, 10:55 PM

#

uhhhhhh

#

i get a page fault if i single step

#

#

?????

hazy saddle Dec 11, 2025, 11:12 PM

#

HUH

hazy saddle Dec 15, 2025, 12:14 PM

#

okay okay

#

ive made some more progress

#

syscalls can kinda work

#

but they then crash horridly

#

#

and the original error before it fully died from seriel is as here

FREE WILL IS A FLAW
General Protection Fault
General Registers:
RAX=0x0000000000000000 RBX=0x000000000000000b
RCX=0x00000000a0000000 RDX=0x000000000000002b
RSI=0x0000000000000001 RDI=0x0000000000000000
RBP=0x000000000000006c RSP=0xffff810000020998
R8 =0x00000000004ac200 R9 =0x0000000000000008
R10=0x0000000000000029 R11=0x000000000000002e
R12=0x0000000000000024 R13=0x00000000004ac210
R14=0x00000000a0035c00 R15=0x000000000000d200
Interrupt Frame:
IP=0xffffffff8000c4aa SP=0xffff810000020998
SS=0x0000000000000010 CS=0x0000000000000008
FLAGS : 00000000 00000001 00000000 10000010 
Control Registers:
CR0: 10000000 00000001 00000000 00010001 
CR4: 00000000 00000000 00000010 00100000 
CR8: 00000000 00000000 00000000 00000000 
CR2=0x0000000000000000 CR3=0x0000000001d94000
Misc:
ERROR : 00000000 00000001 00000000 00100000 
RFLAGS: 00000000 00000001 00000000 10000010 
GDTR Base=0xffffffff80016340 GDTR Limit=0x0000003f
IDTR Base=0xffffffff800193a0 IDTR Limit=0x00000fff
LDT=00000000 TR=00000030

#

the STAR MSR was being set wrong

#

which did some funky things

#

because i forgot one < yaaar

#

im unsure on SFMASK

#

i tried to look at other implementations for a proper mask and i tried some and they did not work. only 0 worked whoch others tell me isnt correct

#

i also properly told the thing it wold have RCX and R11 clobberd heh

#

aswell this made things a lil more stable

#

i may push to dev in a bit and keep working on this

#

but i am a lil confused

hazy saddle Dec 15, 2025, 12:18 PM

#

hazy saddle and the original error before it fully died from seriel is as here ``` FREE WILL...

the flags are printed as binary btw, i may add proper parsing for this later?

#

https://tenor.com/view/slam-smash-computer-slow-rage-gif-15806211

Tenor

#

but the user space app manages to survive a few seconds before death

hazy saddle Dec 15, 2025, 12:38 PM

#

Oops. Undoing that change

hazy saddle Dec 15, 2025, 2:13 PM

#

fault seems to be coming from here

/home/evalyn/Documents/Programming/C/evalynOS/src/kernel/interupts//src/kernel/interupts/interupts.asm:57

which is the iretq here

    pop rcx
    pop rbx
    pop rax

    add rsp, 16
    iretq

viral bison Dec 15, 2025, 2:14 PM

#

hazy saddle fault seems to be coming from here ``` evalyn@EvalynPC  ~/Documents/Programming...

is that the syscall code?

hazy saddle Dec 15, 2025, 2:14 PM

#

viral bison is that the syscall code?

interupts.asm:57

viral bison Dec 15, 2025, 2:15 PM

#

yeah i forgot you are using normal syscalls rn

hazy saddle Dec 15, 2025, 2:15 PM

#

Something is fucking up shit when it system calls and then comes back to interupt

#

Also that’s the same IP as when I had the mask be 0

#

So that seems to delay the inevitable

viral bison Dec 15, 2025, 2:15 PM

#

maybe a race condition?

hazy saddle Dec 15, 2025, 2:15 PM

#

maybe?

#

ive tried to add sti/cli in both handlers but no luck

viral bison Dec 15, 2025, 2:16 PM

#

i think i'm fighting with something similar in my scheduler

#

lol

hazy saddle Dec 15, 2025, 2:17 PM

#

I wasn’t able to get your scheduler trick to work

#

But this way shouldddd be fine too

#

It’s just usureaboutthat

#

Acting up

#

yaaar

viral bison Dec 15, 2025, 2:21 PM

#

hazy saddle I wasn’t able to get your scheduler trick to work

calling the schedule function from the int handler can go bad very fast

hazy saddle Dec 15, 2025, 2:23 PM

#

viral bison calling the schedule function from the int handler can go bad very fast

And making it the int handler also went bad?

#

The int handler calls the PIT code which calls the secedular which then calls the assembly… which then never pops the things off the stack right?

#

Meaning stack overflow

#

yaaar

#

But why did this never overflow before like this?

#

Even with running the kernel for hours?

viral bison Dec 15, 2025, 2:33 PM

#

hazy saddle And making it the int handler also went bad?

I've found that it never changed rsp correctly when using interrupts

#

So i'm making a dedicated handler

hazy saddle Dec 15, 2025, 2:34 PM

#

is there a way i can track the RSP to check if its doing that to me

#

maybe it could explain musterious stack coruption?

#

wana go over again with me how it should be domne?
its an assembly ISR that on trigger passes the current RSP to a C function and then switch to the RSP returned by said function @viral bison?

viral bison Dec 15, 2025, 2:43 PM

#

hazy saddle is there a way i can track the RSP to check if its doing that to me

single step

hazy saddle Dec 15, 2025, 2:48 PM

#

true

#

What’s the GDB command to make or print that every step?

viral bison Dec 15, 2025, 2:48 PM

#

get gef

hazy saddle Dec 15, 2025, 2:48 PM

#

true

viral bison Dec 15, 2025, 2:48 PM

#

and the ni to single step instructions (in asm)

#

or n for c code

#

s to step in

#

and fin to finish the current function

hazy saddle Dec 15, 2025, 2:49 PM

#

And I assume I have a breakpoint on the syscall handler

viral bison Dec 15, 2025, 2:49 PM

#

if you use kvm you have to use hb

#

and you can have max 10 hbs

hazy saddle Dec 15, 2025, 2:49 PM

#

I use TCG when debugging

#

What’s an HB?

viral bison Dec 15, 2025, 2:50 PM

#

hardware break

hazy saddle Dec 15, 2025, 2:50 PM

#

Hardware breakpoint

#

Ah

hazy saddle Dec 15, 2025, 2:50 PM

#

hazy saddle wana go over again with me how it should be domne? its an assembly ISR that on t...

And is this correct

viral bison Dec 15, 2025, 2:50 PM

#

it should

#

but remember that if you get the rip inside the int handler, you'll get the code inside the int handler

#

but if you have a dedicated handler, you get the code from where that int was called

hazy saddle Dec 15, 2025, 2:52 PM

#

Where would I be using the RIP here?

viral bison Dec 15, 2025, 2:52 PM

#

on the ret / iret

hazy saddle Dec 15, 2025, 2:52 PM

#

Would this new all assembly one still need to push and pop every register?

viral bison Dec 15, 2025, 2:53 PM

#

not every

#

you have 11 regs in total

hazy saddle Dec 15, 2025, 2:53 PM

#

Just the normal ones I already do

viral bison Dec 15, 2025, 2:53 PM

#

5 from iret

#

and 6 that you do currently

hazy saddle Dec 15, 2025, 2:55 PM

#

somthing like this @viral bison ?

pit_isr_schedule:
    pushfq
    push rbx
    push rbp
    push r12
    push r13
    push r14
    push r15

    mov rdi, rsp
    call the_c_function
    mov rsp, rax

    pop r15
    pop r14
    pop r13
    pop r12
    pop rbp
    pop rbx
    popfq

    ret

viral bison Dec 15, 2025, 2:57 PM

#

yes

#

tho prob pushfq is not needed

hazy saddle Dec 15, 2025, 2:58 PM

#

yeah but removing it before borked shit

viral bison Dec 15, 2025, 2:58 PM

#

also, you prob want a load cr6 before mov rsp, rsi

#

and you also don't have rsi btw

hazy saddle Dec 15, 2025, 2:58 PM

#

cr3? and would the C function do that

viral bison Dec 15, 2025, 2:58 PM

#

just do mov cr3, regs

hazy saddle Dec 15, 2025, 2:58 PM

#

true

#

but i have a C function that handles that with some extra stuff. and cant i only return one thing?

#

and the return is in rax right

#

so i changed that

viral bison Dec 15, 2025, 3:01 PM

#

it should be done in asm if you don't want a pagefault

viral bison Dec 15, 2025, 3:01 PM

#

hazy saddle and the return is in rax right

rax and rdx

hazy saddle Dec 15, 2025, 3:02 PM

#

so how do i return 2 values properly?

viral bison Dec 15, 2025, 3:02 PM

#

return an uint128

hazy saddle Dec 15, 2025, 3:02 PM

#

true

viral bison Dec 15, 2025, 3:02 PM

#

lower 64 bits are on rax

#

top are on rdx

hazy saddle Dec 15, 2025, 3:16 PM

#

@viral bison i dont have a uint128

viral bison Dec 15, 2025, 3:16 PM

#

make it

hazy saddle Dec 15, 2025, 3:16 PM

#

how

viral bison Dec 15, 2025, 3:17 PM

#

#

hazy saddle Dec 15, 2025, 3:17 PM

#

wut

viral bison Dec 15, 2025, 3:18 PM

#

look at how i did it

hazy saddle Dec 15, 2025, 3:18 PM

#

i am?

#

im confused

#

how does that even work

viral bison Dec 15, 2025, 3:19 PM

#

it's a compiler macro

hazy saddle Dec 15, 2025, 3:19 PM

#

yes

#

but where is the 128 bit integer primative coming from

viral bison Dec 15, 2025, 3:19 PM

#

macro

#

it's builtin in the compiler

#

dw about it

hazy saddle Dec 15, 2025, 3:19 PM

#

i dont have that

#

where

#

of course

#

autocomplete wont show it

#

but it works

#

fml

#

@viral bison so like this
return ((__uint128_t)current_thread->pagemap->top_level << 64) | current_thread->krsp;

viral bison Dec 15, 2025, 3:24 PM

#

yes

hazy saddle Dec 15, 2025, 3:29 PM

#

@viral bison

#

goddamnit

#

__uint128_t schedule(uint64_t old_krsp) {
    pitInteruptsTriggered++;
    if (shouldSchedule) {
        struct thread *previous_thread = threads->thread;
        threads = threads->next_thread;
        struct thread *current_thread = threads->thread;

        vmm_switch_to(current_thread->pagemap);

        tss.rsp0 = (uint64_t)current_thread->stack_top;
        wrmsr(UGSBAS, (uint64_t)current_thread);

        previous_thread->krsp = old_krsp;
        return ((__uint128_t)current_thread->pagemap->top_level << 64) | current_thread->krsp;
    } else {
        return ((__uint128_t)kernel_pagemap->top_level << 64) | old_krsp;
    }
}

#

global isr0x20
extern schedule
isr0x20:
    pushfq
    push rbx
    push rbp
    push r12
    push r13
    push r14
    push r15

    mov rdi, rsp
    call schedule
    mov rsp, rax
    mov cr3, rdx

    pop r15
    pop r14
    pop r13
    pop r12
    pop rbp
    pop rbx

    ; send EOI
    push rax
    mov al, 0x20
    out 0x20, al
    pop rax

    iretq

#

where is it going to 0x0 yaaar

viral bison Dec 15, 2025, 3:30 PM

#

how are you steting the initial stack?

hazy saddle Dec 15, 2025, 3:30 PM

#

true leme check

#

    *--stack = (uint64_t)task_quit;
    *--stack = (uint64_t)entry_point;
    *--stack = 0x202;

    *--stack = 0;
    *--stack = 0;
    *--stack = 0;
    *--stack = 0;
    *--stack = 0;
    *--stack = 0;

viral bison Dec 15, 2025, 3:31 PM

#

yeah well

#

you are missing quite a few registers

hazy saddle Dec 15, 2025, 3:31 PM

#

wait wha

viral bison Dec 15, 2025, 3:31 PM

#

first of all

#

you forgot a popfq

hazy saddle Dec 15, 2025, 3:31 PM

#

fuk

viral bison Dec 15, 2025, 3:31 PM

#

2nd, iretq uses 5 registers

#

rsi = rip

#

[rdi] = stack pointer

hazy saddle Dec 15, 2025, 3:32 PM

#

so do i want ret?

viral bison Dec 15, 2025, 3:32 PM

#

no

#

just add 4 regs to your stack setting

#

after the rip

hazy saddle Dec 15, 2025, 3:33 PM

#

so this and this

#

#

yaaar

#

wait after rip

#

so

#

#

nope

#

goddamnit

#

viral bison Dec 15, 2025, 3:35 PM

#

BRUH

hazy saddle Dec 15, 2025, 3:35 PM

#

what did i do 😭

viral bison Dec 15, 2025, 3:35 PM

#

look at how i set them

hazy saddle Dec 15, 2025, 3:36 PM

#

i thought i did yaaar

viral bison Dec 15, 2025, 3:36 PM

#

and i told you to add them before entry point

hazy saddle Dec 15, 2025, 3:36 PM

#

hazy saddle

FUCK

viral bison Dec 15, 2025, 3:36 PM

#

also, flags are not needed (pushfq and popfq)

#

bc the interrupt does it all for you

#

and iret restores the flags

hazy saddle Dec 15, 2025, 3:36 PM

#

so this pwease

viral bison Dec 15, 2025, 3:36 PM

#

in my case the flags are 0x292

viral bison Dec 15, 2025, 3:37 PM

#

hazy saddle so this <:pwease:1338294429282336879>

wrong values

hazy saddle Dec 15, 2025, 3:37 PM

#

dont i do this in assembly?

viral bison Dec 15, 2025, 3:37 PM

#

the stack you are setting is all 0

#

lol

hazy saddle Dec 15, 2025, 3:38 PM

#

yes?

viral bison Dec 15, 2025, 3:38 PM

#

yeah, and that values are meant to be in the stack

#

lol

hazy saddle Dec 15, 2025, 3:38 PM

#

goddamnit what do i need to put there yaaar

viral bison Dec 15, 2025, 3:38 PM

#

*--stack=0x10;
*--stack=rsp;
*--stack=0x292 (or your rflags);
*--stack=0x08;
*--stack=rip;

hazy saddle Dec 15, 2025, 3:38 PM

#

my brain isnt working

#

thank you

#

wait what rsp do i put there?

viral bison Dec 15, 2025, 3:39 PM

#

the thread one

#

kernel

hazy saddle Dec 15, 2025, 3:39 PM

#

what

#

thats in C right?????

viral bison Dec 15, 2025, 3:39 PM

#

the kernel thread stack

hazy saddle Dec 15, 2025, 3:39 PM

#

;-;

viral bison Dec 15, 2025, 3:39 PM

#

i think you save the thread rsp, no?

hazy saddle Dec 15, 2025, 3:39 PM

#

fuk

#

im dum

#

#

whyyyy

#

wait

#

i need to offset that

#

right

#

dosnt help yaaar

#

10 11 or 12 breaks

viral bison Dec 15, 2025, 3:44 PM

#

ehm

#

don't shift maybe?

#

just put stack top

#

the values will be popped anyway when iret reaches that rsp

hazy saddle Dec 15, 2025, 3:46 PM

#

Even without shifting

#

It breaks

hazy saddle Dec 15, 2025, 3:46 PM

#

hazy saddle

Here

viral bison Dec 15, 2025, 3:49 PM

#

single step then

#

and check values

hazy saddle Dec 15, 2025, 3:49 PM

#

yaaar

#

To be clear

#

This crashes before the schedules even runs

viral bison Dec 15, 2025, 3:49 PM

#

lol what

#

then it's not an iret problem

hazy saddle Dec 15, 2025, 3:49 PM

#

Yes

viral bison Dec 15, 2025, 3:49 PM

#

find where it crashes then

hazy saddle Dec 15, 2025, 3:50 PM

#

return ((__uint128_t)kernel_pagemap->top_level << 64) | old_krsp;

#

shit

#

thats not valid

viral bison Dec 15, 2025, 3:51 PM

#

huh?

hazy saddle Dec 15, 2025, 3:51 PM

#

the PIT is setup before the vmm

viral bison Dec 15, 2025, 3:51 PM

#

loooool

#

just do sti when you are done initializing

hazy saddle Dec 15, 2025, 3:51 PM

#

now it only GPFs

#

wtf??????

#

im done

#

wtf is this

viral bison Dec 15, 2025, 3:53 PM

#

?

hazy saddle Dec 15, 2025, 3:53 PM

#

#

i changed the code

#

no change

hazy saddle Dec 15, 2025, 4:50 PM

#

@viral bison update this will not cause a crash

#

but it locks up all interupts??

#

global isr0x20
extern schedule
extern shouldSchedule
extern pitInteruptsTriggered
isr0x20:
    add dword [rel pitInteruptsTriggered], 1
    push rax
    mov rax, [rel shouldSchedule]
    cmp rax, 1
    pop rax
    je do_schedule
    jne dont_schedule

    do_schedule:
    push rbx
    push rbp
    push r12
    push r13
    push r14
    push r15

    mov rdi, rsp
    call schedule
    mov rsp, rax
    mov cr3, rdx

    pop r15
    pop r14
    pop r13
    pop r12
    pop rbp
    pop rbx

    ; send EOI
    push rax
    mov al, 0x20
    out 0x20, al
    pop rax

    push 0x10
    push [rdi]
    push 0x292
    push 0x08
    push rsi

    dont_schedule:

    sti

    iretq

#

even just this will?????

#

isr0x20:
    add dword [rel pitInteruptsTriggered], 1

    push rax
    mov al, 0x20
    out 0x20, al
    pop rax
    sti
    iretq

hazy saddle Dec 15, 2025, 5:48 PM

#

yeah @viral bison im never gonna get your way to work.
even making my normal PIT isr in assemblu without the new functon is still fucked

#

hazy saddle Dec 15, 2025, 7:37 PM

#

viral bison [rdi] = stack pointer

okay i think im getting somwhere. where do i get the RIP in the ISR?

viral bison Dec 15, 2025, 7:38 PM

#

hazy saddle okay i think im getting somwhere. where do i get the RIP in the ISR?

it's the first item in the stack

hazy saddle Dec 15, 2025, 7:39 PM

#

so from here what should i be doing?
pop that into a register?

viral bison Dec 15, 2025, 7:40 PM

#

i told you to not push flags

hazy saddle Dec 15, 2025, 7:40 PM

#

FUCK

viral bison Dec 15, 2025, 7:40 PM

#

it won't do anything really

hazy saddle Dec 15, 2025, 7:40 PM

#

i completly refid the section

#

how did that even stay there 😭

viral bison Dec 15, 2025, 7:40 PM

#

it's just an useless push / pop

hazy saddle Dec 15, 2025, 7:41 PM

#

ok ok

#

but what do I need to do

#

yaaar

viral bison Dec 15, 2025, 7:41 PM

#

these here are completely wrong

hazy saddle Dec 15, 2025, 7:41 PM

#

why tf is it tripple faulting

#

is this also fucked

viral bison Dec 15, 2025, 7:42 PM

#

viral bison these here are completely wrong

they shouldn't be here

viral bison Dec 15, 2025, 7:42 PM

#

hazy saddle is this also fucked

this is correct

#

but you are pushing wrong values

hazy saddle Dec 15, 2025, 7:42 PM

#

then why did ytou tell me to put thoes there before

viral bison Dec 15, 2025, 7:42 PM

#

i push them in my stack init...

#

i told you

viral bison Dec 15, 2025, 7:42 PM

#

hazy saddle is this also fucked

this is your stack init

hazy saddle Dec 15, 2025, 7:43 PM

#

viral bison

i thought this was part of the isr ;-;

viral bison Dec 15, 2025, 7:43 PM

#

f

hazy saddle Dec 15, 2025, 7:43 PM

#

whats the proper values?

viral bison Dec 15, 2025, 7:43 PM

#

you have to remove those 5 pushes in the isr

hazy saddle Dec 15, 2025, 7:44 PM

#

i did

#

viral bison Dec 15, 2025, 7:44 PM

#

k, good

hazy saddle Dec 15, 2025, 7:44 PM

#

it dosnt crash but it locks up

#

do i need a sti?

#

no

#

it dosnt fix it

#

hmmCat

#

it seems to not be scheduling @viral bison??
but if i add a second thread it kinda goes but it page faults

viral bison Dec 15, 2025, 7:49 PM

#

bruh

hazy saddle Dec 15, 2025, 7:49 PM

#

and its only because its two of the same

#

otherwise still halts?

#

im sorry

#

what did i fuck up this time im so sorry

#

@viral bison somehow somewhere interupts get disabled goddamnit

viral bison Dec 15, 2025, 7:57 PM

#

hazy saddle <@573903151472836609> somehow somewhere interupts get disabled <:goddamnit:14034...

forever?

hazy saddle Dec 15, 2025, 7:57 PM

#

yes

viral bison Dec 15, 2025, 7:57 PM

#

oof

hazy saddle Dec 15, 2025, 7:57 PM

#

even this wont fix it

#

whyyyyy

#

the EOI is right here too

#

I genuinely don’t know what the fuck to do anymore

#

I’m fucking cursed

#

@viral bison is this correct too or am i being dumb??

viral bison Dec 15, 2025, 8:14 PM

#

should be

hazy saddle Dec 15, 2025, 8:16 PM

#

so any ideas?

floral kettle Dec 15, 2025, 8:16 PM

#

fwiw I wouldn't tie scheduling to interrupts, it just ends up being messy

#

if you are doing the same thing as what diamondmaster did that is

viral bison Dec 15, 2025, 8:16 PM

#

floral kettle fwiw I wouldn't tie scheduling to interrupts, it just ends up being messy

how do you schedule via lapic then?

floral kettle Dec 15, 2025, 8:17 PM

#

you just call sched() in your timer handler

hazy saddle Dec 15, 2025, 8:17 PM

#

i mean thats what @viral bison said for me to do
i used to have the PIT ISR call schedule()

#

but aparently that caused issues

viral bison Dec 15, 2025, 8:17 PM

#

floral kettle you just call sched() in your timer handler

timer handler is an isr tho

hazy saddle Dec 15, 2025, 8:17 PM

#

acording to diamond

floral kettle Dec 15, 2025, 8:17 PM

#

yes?

viral bison Dec 15, 2025, 8:17 PM

#

bc the lapic sends interrupts

dense cargo Dec 15, 2025, 8:17 PM

#

hazy saddle <@573903151472836609> is this correct too or am i being dumb??

uint128?

viral bison Dec 15, 2025, 8:18 PM

#

ye

floral kettle Dec 15, 2025, 8:18 PM

#

what I mean is that imo its cleaner if its a normal function you can just call anywhere to switch between threads

#

and then resume at that exact same point later on

viral bison Dec 15, 2025, 8:18 PM

#

oh, i have a function that calls the interrupt, lol

dense cargo Dec 15, 2025, 8:18 PM

#

dense cargo uint128?

why not a struct

viral bison Dec 15, 2025, 8:18 PM

#

i had problems with structs

hazy saddle Dec 15, 2025, 8:18 PM

#

dense cargo why not a struct

sysv abi to assembly

#

utint128 splits the two halfs into registers

dense cargo Dec 15, 2025, 8:19 PM

#

dense cargo why not a struct

the compiler would still split the return value into rax and rdx

floral kettle Dec 15, 2025, 8:21 PM

#

// void sched_switch_thread(uintptr_t* old_sp_loc, uintptr_t new_sp)
sched_switch_thread:
push rbx
push rbp
push r12
push r13
push r14
push r15

mov [rdi], rsp
mov rsp, rsi

pop r15
pop r14
pop r13
pop r12
pop rbp
pop rbx
ret

viral bison Dec 15, 2025, 8:22 PM

#

tho if this is called from an isr, it will then go back and finish that isr, potentially messing things up

floral kettle Dec 15, 2025, 8:22 PM

#

void sched() {
    // choose next_thread, update msrs and load page map etc
    // switch to there
    sched_switch_thread(&current_thread->sp, new_thread->sp);
}

hazy saddle Dec 15, 2025, 8:22 PM

#

yeah i had this before
removing pushfq broke things though

thread_switch:
    pushfq
    push rbx
    push rbp
    push r12
    push r13
    push r14
    push r15

    mov [rdi], rsp
    mov rsp, rsi

    pop r15
    pop r14
    pop r13
    pop r12
    pop rbp
    pop rbx
    popfq

    ret

hazy saddle Dec 15, 2025, 8:22 PM

#

floral kettle ```c void sched() { // choose next_thread, update msrs and load page map etc...

had this too

floral kettle Dec 15, 2025, 8:23 PM

#

viral bison tho if this is called from an isr, it will then go back and finish that isr, pot...

it doesn't mess anything up, it returns to the end of the isr where the full regs saved to the irq stack are restored + iret is done to return to wherever the preemption irq was triggered

dense cargo Dec 15, 2025, 8:23 PM

#

dense cargo the compiler would still split the return value into rax and rdx

viral bison Dec 15, 2025, 8:24 PM

#

dense cargo

that is c

#

in c++ it complains

#

the struct i mean

dense cargo Dec 15, 2025, 8:24 PM

#

viral bison in c++ it complains

yikes

floral kettle Dec 15, 2025, 8:24 PM

#

it was only because you had the initializers

viral bison Dec 15, 2025, 8:24 PM

#

oh

hazy saddle Dec 15, 2025, 8:25 PM

#

floral kettle it doesn't mess anything up, it returns to the end of the isr where the full reg...

so i should undo all this mess and go back to where i was when it worked but still broken syscalls?

floral kettle Dec 15, 2025, 8:25 PM

#

idk, or you could try fixing it somehow

viral bison Dec 15, 2025, 8:26 PM

#

i'm trying to fix mine too

dense cargo Dec 15, 2025, 8:26 PM

#

viral bison oh

compiler extensions like uint128 aren't really a good workaround

viral bison Dec 15, 2025, 8:26 PM

#

somehow the stack gets fucked up if i don't put a print on the function XD

dense cargo Dec 15, 2025, 8:26 PM

#

dense cargo compiler extensions like uint128 aren't really a good workaround

if a simple 16 bytes struct (which complies the same way as a 128bit integer) causes problems there's a deeper problem

#

but if the problem was caused just by initializers then it would be a good idea to go back to that implementation

hazy saddle Dec 15, 2025, 8:27 PM

#

floral kettle idk, or you could try fixing it somehow

i mean i liked the other way with the PIT timer calling the schedule

#

and that worked

#

but syscall is still fucked

#

but ive had some minor fixes and progress

floral kettle Dec 15, 2025, 8:28 PM

#

is it related to the sched stuff though or is it unrelated

dense cargo Dec 15, 2025, 8:28 PM

#

hazy saddle but syscall is still fucked

do you switch to rsp0 and/or a dedicated thread kernel stack

#

it's a common mistake

hazy saddle Dec 15, 2025, 8:29 PM

#

dense cargo do you switch to rsp0 and/or a dedicated thread kernel stack

it can run a few times. and then borks. and if i change SFMASK to zero from ~0x2 it ran for a few seconds somtimes

#

and i dont think so

#

dense cargo Dec 15, 2025, 8:31 PM

#

hazy saddle

should be the mov rsp, [gs:24] part

#

so you're using a thread kernel stack

hazy saddle Dec 15, 2025, 8:31 PM

#

yes

#

is that bad?

dense cargo Dec 15, 2025, 8:31 PM

#

hazy saddle is that bad?

nope, it's good, saves you from lots of troubles

#

like stacks being overwritten when a syscall handler yields and another thread requests a syscall

hazy saddle Dec 15, 2025, 8:33 PM

#

yes

#

so what is going on with it yaaar

#

pretty sure it is setup right

#

and my gdt is as such

dense cargo Dec 15, 2025, 8:35 PM

#

32bit user code?

#

why is that?

hazy saddle Dec 15, 2025, 8:36 PM

#

its what @viral bison told me i needed yaaar

viral bison Dec 15, 2025, 8:36 PM

#

it could be skipped

#

but i do recommend it

dense cargo Dec 15, 2025, 8:36 PM

#

compability mode?

viral bison Dec 15, 2025, 8:36 PM

#

ye

hazy saddle Dec 15, 2025, 8:37 PM

#

i dont do compat mode but aparently its needed to be that format for STAR?

viral bison Dec 15, 2025, 8:37 PM

#

you could skip it

#

and change that value to the one before the user data 64

#

tho you have to then be careful to not call syscall from 32 bit mode

hazy saddle Dec 15, 2025, 8:39 PM

#

viral bison and change that value to the one before the user data 64

what????? thats the same as it is now

dense cargo Dec 15, 2025, 8:39 PM

#

hazy saddle pretty sure it is setup right

won't star (0x18 << 48) be wrong?

viral bison Dec 15, 2025, 8:39 PM

#

i copied it from someone else

#

that had syscalls working

hazy saddle Dec 15, 2025, 8:39 PM

#

i copied it from diamond

dense cargo Dec 15, 2025, 8:39 PM

#

it needs to be coherent with gdt segments

hazy saddle Dec 15, 2025, 8:40 PM

#

documentation on this is so ass for syscall/sysret

hazy saddle Dec 15, 2025, 8:40 PM

#

dense cargo it needs to be coherent with gdt segments

what gdt segments is it meant to be

dense cargo Dec 15, 2025, 8:40 PM

#

hazy saddle pretty sure it is setup right

the 48 bits shifted value must be the segment offset for 64bit user code

#

which should come 8 bytes after the 64 bit user data segment

#

(iirc)

hazy saddle Dec 15, 2025, 8:41 PM

#

then why do some syscalls manage to work before failing i womder

floral kettle Dec 15, 2025, 8:41 PM

#

the value at star 63:48 should point to the user code32 gdt entry if you have that (and your gdt is of the form user code32, user data, user code64)

hazy saddle Dec 15, 2025, 8:42 PM

#

floral kettle the value at star 63:48 should point to the user code32 gdt entry if you have th...

yes thats how mine is set?

dense cargo Dec 15, 2025, 8:42 PM

#

floral kettle the value at star 63:48 should point to the user code32 gdt entry if you have th...

won't SS become 0x10 kernel data?

#

isn't SS = CS - 8

floral kettle Dec 15, 2025, 8:42 PM

#

no its cs + 8

#

actually wait no

dense cargo Dec 15, 2025, 8:43 PM

#

dense cargo isn't SS = CS - 8

I clearly remember having to reorder the segments because of it

floral kettle Dec 15, 2025, 8:43 PM

#

yeah its cs - 8 but star 63:48 doesn't point to cs64

#

if you write 0x18 to star then user cs is at 0x18 + 16 (0x28) and user ss will be 0x18 + 8 (0x20)

#

but anyway it should be correct yeah

hazy saddle Dec 15, 2025, 8:47 PM

#

So then what else is causing issues?

floral kettle Dec 15, 2025, 8:47 PM

#

I can try to take a look at it if I can get it triggered again lol

hazy saddle Dec 15, 2025, 8:48 PM

#

hazy saddle fault seems to be coming from here ``` evalyn@EvalynPC  ~/Documents/Programming...

This is where I was talking about previous crashes related to the syscall issue

#

I’m gonna put that all together into a proper commit and push to dev and remove this pit scheduler nonsense

hazy saddle Dec 15, 2025, 8:50 PM

#

floral kettle I can try to take a look at it if I can get it triggered again lol

Get what triggered again if I may ask?

floral kettle Dec 15, 2025, 8:54 PM

#

some random page fault when running the badapple cmd a bunch of times, is this different from that now?

hazy saddle Dec 15, 2025, 8:56 PM

#

If too many system calls run it faults

hazy saddle Dec 15, 2025, 8:56 PM

#

floral kettle some random page fault when running the badapple cmd a bunch of times, is this d...

This was fixed and if using int 0x69 they work

#

And I can spawn it as many as I want

#

That’s on main branch

#

But the dev branch is trying to get syscall to work but it’s broken

#

I tried adding more fixes and had some progress

#

But did not fully work and then I spent all day trying to get @viral bison s idea to work

gentle quest Dec 17, 2025, 3:57 PM

#

viral bison

Why did you wrap the typedefs into extern C?

viral bison Dec 17, 2025, 3:58 PM

#

gentle quest Why did you wrap the typedefs into extern C?

i had some functions in there

#

and later got rid of them

gentle quest Dec 17, 2025, 3:58 PM

#

Oh

viral bison Dec 17, 2025, 3:58 PM

#

and never removed extern "C"

hazy saddle Dec 17, 2025, 10:43 PM

#

project is now on codeberg.
may still commit to github but probally not
https://codeberg.org/EvalynGoemer/evalynOS

Codeberg.org

evalynOS

A WIP 64 Bit Operating System

hazy saddle Dec 19, 2025, 2:42 AM

#

okay well they 100% work once or twice

#

but somthing still gets messed up

#EvalynOS