Nyaux

[shut laurel]

i mean what do you actually fault on?

[surreal path]

i just told you

[shut laurel]

nvm saw the cr2

[surreal path]

maybe my scheduler code is bugged?

idk

no idea

everything looks correct??

tf

[flat nymph]

That's not how that works

[surreal path]

sorry

wait wdym

[flat nymph]

When you enter the kernel you set rbp to 0 after saving it

I think

Idr

It's too late

But stack frames make a linked list

[shut laurel]

you‘ll just set it to 0, after that entry function returns it would be 0 as functions just push rbp

so i guess it should work like that

[surreal path]

yea so

why isnt it working

because i have a check in my stack tracing code

if its 0

stop

[shut laurel]

wait no, I zero rbp, then push rbp

[surreal path]

oh so after saving rbp i must set it to 0 yes?

in the asm code?

like this

or no

[flat nymph]

Yes

No

Not here

[surreal path]

where

?

this one

or the interrupt code

[flat nymph]

Yes

But in the other half of it

When you enter the kernel

[surreal path]

oh so the interrupt

i must set rbp to 0

after saivng it yes

[flat nymph]

Yes

[shut laurel]

oh yeah

[surreal path]

okay there

thats it?

[flat nymph]

Also make sure the stack is aligned

[surreal path]

it should be as its coming from my allocator

so it will always be aligned

[flat nymph]

To 16 bytes

[surreal path]

it still dies

btw

[flat nymph]

That's what I do

[surreal path]

oh before pushing it???????????????????????????????????????????????

what

why

[flat nymph]

No

So the stack frame is constructed

Ideally you push all the reggisters, then you set %rbp to 0 and push it

(I'm pushing it twice because of alignment)

Then you do ```x86asm
pushq %rbp
movq %rsp, %rbp

[surreal path]

thats my stackframe

[flat nymph]

As function prelude

[flat nymph]

Doen't matter

[surreal path]

is it aligned to 16 bytes at least

[flat nymph]

Idk

What does your allocator do

Although it probably doesn't matter

[surreal path]

always has good alignment so its fine

[surreal path]

okay

[flat nymph]

Unless you're saving SSE stuff

[surreal path]

so make rbp 0 then push it then at function preload make the rbp rsp

[flat nymph]

Idk read system v spec

[surreal path]

bru

i will try this

[flat nymph]

That's what I do, but I don't remember why and maybe it's brokern

[flat nymph]

No

This will override rbp and break userspace in weird ways

[surreal path]

oh

[flat nymph]

And kernelspace as well

[surreal path]

but it solved it?

weird

but okay

yea no idea what to do

[flat nymph]

Yeah you'll not see stack trace at all

Read the spec

[surreal path]

bru

[flat nymph]

You also probably need movq %rsp, %rbp

After pushq

[surreal path]

isnt that generated when u call a c func

[flat nymph]

It is

Then there is no need to push I guess

[surreal path]

yea im finding nothing in spec useful

[flat nymph]

Idk I haven't read it and don't know if it's any good but this is the first link from google https://textbook.cs161.org/memory-safety/x86.html

[surreal path]

i literarly know this

i dont get how any of this is gonna help my issue

[flat nymph]

ChatGPT should know this well

It will stop your stack unrolling

[surreal path]

but i already know all of this

its saying basic shit

[flat nymph]

Yeah it's just a linked list

[surreal path]

yea and somehow my stack is corrupted or something

no idea fucking why

[flat nymph]

I wouln't be surprised if it is, but it's probably not

[surreal path]

it prob is

because why else is stacktrace dying

[flat nymph]

What makes you think so

What's your stacktrace code

[surreal path]

here

[flat nymph]

Are you doing this?

[surreal path]

nyauxsymbol h = find_from_rip(frame->rip);
    uint64_t* base_ptr = 0;
    uint64_t temp = 0;
    temp = frame->rbp;
    kprintf_symbol(h);
    base_ptr = (uint64_t*)temp;
    while (base_ptr != 0)
    {
        uint64_t ret_addr = *(uint64_t*)((uint64_t)base_ptr + 8);
        if (ret_addr != 0)
        {
            h = find_from_rip(ret_addr);
            kprintf_symbol(h);
        }
        else
        {
            kprintf("-> Function: none -- 0x0\n");
        }

        base_ptr = (uint64_t*)*(uint64_t*)(base_ptr);
    }

[flat nymph]

No

[surreal path]

wdym no

[flat nymph]

Wdym find_from_rip

Oh nvm

[surreal path]

searches my symbol array

[flat nymph]

Still

rbp points to saved rbp

When it's 0, you stop

[surreal path]

yes

[flat nymph]

Why are you doing if (ret_addr != 0)

Does some magic function at the last virtual address make a function call?

[surreal path]

what

[flat nymph]

Hmm

I need to sleep

[surreal path]

yea i dont think its the stacktrace code

[flat nymph]

The code is ok

[surreal path]

yea

told you

[flat nymph]

Still

mov 0 to rbp before calling C I guess

[surreal path]

i did

nothing changed!

[flat nymph]

I think Limine should do that for you

Although idr

Maybe it doesn't

[surreal path]

either way i still have an issue

[flat nymph]

If you're still using its stack

[flat nymph]

Sorry

[surreal path]

why are you apologising

[flat nymph]

I'm being very arrogant

[surreal path]

its fine even then, at least ur trying to help

which im grateful for

😄

i think u should go to sleep

[flat nymph]

I'm trying to study

[surreal path]

okay, well focus on studying then. ill try to figure this issue out

somehow

[surreal path]

still cant figure it out chat

looking at the code back and fourth

dont get why its having this

[surreal path]

[tender gorge]

Have you tried politely asking the code to work?

[surreal path]

ah yes

but seriously tho

i have absolutely NO idea whats going on

[cinder plinth]

less complaining more fixing

🔫

[surreal path]

bro thats what ive been trying to do for the past few hours

💀

[surreal path]

yea ive tried so much to figure this out

its 4:30am

i need to sleepn

goodnight

[finite summit]

Gm

Gn

[surreal path]

morning

back to debugging nyaux

[surreal path]

we are running limine stack

we get first timer interrupt, push stackframe

disregard this stackframe cause old thread is NULL initially

run do_savekstackandloadkstack

push kernel registers rbx - r15

load kstack from thread

call change rsp0

pop those k registers

ret into return_from_kernel_in_new_thread
pop userspace regs

iretq into thread

get timer interrupt

push stackframe

run do_saveksatckandloadkstack
push k registers
save old userspace regs into thread

load new userspace regs from thread
call change_rsp0
pop k registers

ret into return_from_kernel_in_new_thread

pop userspace regs

iretq into thread

yea no idea

i made this keeping track of the stack

i see no area where stack would get corrupted

[flat nymph]

call change_rsp0

[surreal path]

how

[flat nymph]

What does it do

[surreal path]

that

[flat nymph]

You create a new thread inside the timer interrupt, right?

[surreal path]

no

[flat nymph]

Then why do you have ret into return_from_kernel_in_new_thread

[surreal path]

because that is setup by create_kentry

before we go into the timer interrupt

[flat nymph]

And

After you call do_savekstackandloadkstack, you return to there on the next call

If you're not calling create_kentry

Which you're hopefully not doing

[surreal path]

im not

[flat nymph]

So you create two threads

[surreal path]

no

one

[flat nymph]

Then what are you switching to inside the timer interrupt

[surreal path]

hold on ill explain this

[surreal path]

schedd

the new thread

[flat nymph]

You are inside the only thread there is

You call schedd

[surreal path]

yes

then

[flat nymph]

What does it switch to

[surreal path]

i save the old thread

switch to the same thread

go back into it

[flat nymph]

Where does it pagefault?

[surreal path]

stacktrace

[flat nymph]

Still, after you pop k registers, you don't return to ret into return_from_kernel_in_new_thread

You only do that when you enter userspace for the first time in a new thread

[surreal path]

am i supposed to do that everytiem

[flat nymph]

No

You're only supposed to do that once

[surreal path]

okay

so whats the problem

[flat nymph]

It's just that this isn't what's really happening

What calls your timer interrupt?

[surreal path]

lapic timer

every 10ms

[flat nymph]

Yes, which function do you put in idt and where is it in your code

[surreal path]

sched

that

[flat nymph]

which file is it defined in?

[surreal path]

idt.c

[flat nymph]

That's not how you're supposed to be doing that

You need to be saving all registers

[surreal path]

it does in isr stub

😭

[flat nymph]

Why are you doing cli and sti?

[surreal path]

maybe a timer interrupt happens while im executing this code

[flat nymph]

Yeah

[surreal path]

still tho

no idea whats going on

to cause the stacktrace to die

[flat nymph]

You're not setting rbp to 0 in isr stub

[surreal path]

even if i do that

it still dies

[molten grotto]

well yeah because you take the rbp from the frame (also zeroing rbp there would make the whole backtrace useless)

[flat nymph]

Your struct StackFrame is in wrong order I think

[surreal path]

how

its in the right order

[finite summit]

it's rbp field first then rip

if this is

like stack trace

[molten grotto]

you just have to somehow make rbp be zero at kentry

[surreal path]

look

[flat nymph]

Your intnum is on the top

[surreal path]

thats a dummy error code

[flat nymph]

Or not

Wait

[surreal path]

error_code

[flat nymph]

Yeah

[surreal path]

wrong reply

[surreal path]

so what the hell am i supposed to do

[molten grotto]

where do you switch to kentry?

[surreal path]

wdym

in the schedd task function

?

[molten grotto]

where do you call that from

[surreal path]

timer interrupt

@molten grotto

[flat nymph]

Also, you can just try halting the kernel, and looking at the trace with GDB

[surreal path]

yes i did before

look

^

ignore the ubsan

but still

and btw @flat nymph i had forgot to add a send eoi before switching into the thread

and an sti

but even that doesnt solve anything ofc

i dont know how to fix this

[flat nymph]

Are you sure it's stack corruption?

Like you can place 2 breakpoints inside the isr stub

And check that the registers are the same on entry and on exit

[surreal path]

no

[molten grotto]

your stack trace is broken and tries to get the rip of the first frame with rbp 0

which ends up accessing out of bounds of the thread's stack

[surreal path]

so break if rbp is 0 instantly?

guess what

solves nothing lol

[molten grotto]

you did it in the wrong place

you need to have smth like ```c
uint64_t* next = (uint64_t*) (uint64_t) (base_ptr);
if (!next) {
break;
}

[surreal path]

nope

[molten grotto]

you need to load it from the base_ptr not check base_ptr itself...

the while already checks base_ptr

[surreal path]

that?

solved it??

so there was no bug?

it was just my stacktrace

being stupid?????????????????????????????

[molten grotto]

yes

[surreal path]

wow

so all that fucking debugging

was for a bug that didnt exist

[molten grotto]

a good advise for debugging: don't use macros for large pieces of code like that

[flat nymph]

I said there was no stack corruption

[molten grotto]

it makes debugging harder for no reason, just make it a normal function you can step through line by line

[surreal path]

hi im going to shoot myself

i debugged for 7-8 hours

fucking gotta be kidding me

[surreal path]

thanks for that advice

@molten grotto how do i get 8 hours of my life back

[molten grotto]

by speedrunning whatever the next thing you wanted to do on nyaux was

[surreal path]

[flat nymph]

bash when

[surreal path]

soon ™️

ioapic shanngains to finish

then gotta remmeber the vfs

[molten grotto]

you are likely going to get it before I at least

[surreal path]

[molten grotto]

unless I am going to speedrun a new unix-like kernel

[surreal path]

then tmpfs, devfs, ustar parsing

syscalls

xsave and friends

elf loading

and um

tty

[molten grotto]

I kinda actually want to try making an unix-like kernel, I just don't think I want to do it with crescent tho

[surreal path]

if bro speedruns a unix like kernel faster then i write the vfs

im cooked

anyways im kinda thinking of givings threads like

tids

so like a thread could have tid 0 or 1 or wtv tf

pid exists too

maybe giving threads names too

i think linux does that too

as well as tids

does linux give threads names

[molten grotto]

yes

[surreal path]

whats the max thread name limit

if there is one

[molten grotto]

looks like it has been 15 at some point at least

[surreal path]

wdym some point

did they change it?

[molten grotto]

idk

but you don't have to have it be the same

just choose some sensible limit

[surreal path]

fair

tids are probs stored in a u64 or smt

[molten grotto]

you don't want thread names that are 128 chars long

[surreal path]

what if i do

/j

[molten grotto]

well then you do ig

but that would be kinda weird

[surreal path]

is 256kib for a kernel stack TOO much

i choose that number cause limine defaults to giving 256kib stacks

[flat nymph]

My thread names have no limit in length

[kind root]

kernel stacks are usually 4-16k

[surreal path]

kb or kib

[kind root]

whats the difference

[surreal path]

ima give nyaux 16kib kstacks

[kind root]

are u going to make them exactly 4000 bytes if i tell u its kb

[surreal path]

no

16384

thats 16kib

[kind root]

yes

[kind root]

nah

on linux pid == tid for main thread

[surreal path]

nyaux running on 16kib kstack (page fault is intentionally forgot to remove it)

[kind root]

then tid is just in the same namespace as pid

just incremented

or wahtever

[surreal path]

so if u had pid 16

it would be tid 16 for the main thread

some other thread being 17

[kind root]

yes

[surreal path]

then 18 etc etc

[kind root]

yes

[surreal path]

fair

[surreal path]

wanna change the flanterm font ngl

[kind root]

u have better things to do

[surreal path]

but looks

/j

yes i have better things to do ur right tho

gotta work on ioapic

then

vfs and etcetc

[kind root]

EC 😡

[surreal path]

AND EC

😭

(please dont kill my family infy)

[thorn bramble]

flanterm font: not important
ec driver: very important

[kind root]

im in your walls

[surreal path]

[kind root]

important to get priorities straight

[surreal path]

actually ill do EC straight after ioapic

[finite summit]

he'll end up having some bug with his driver bricking his PC

[kind root]

lmao

[finite summit]

I thought you were in my walls 🥺

[kind root]

sure

[cinder plinth]

i have a patch for psf fonts if u want

oops wrong message

anyway

for fonts

[surreal path]

hold on

take a quick nap then ioapic

[surreal path]

this makes sense

need an route irq function then im done

[flat nymph]

Also interrupt allocation

[surreal path]

i have that but its not per cpu

no per cpu idt

cry abt it i have 256 vectors

[flat nymph]

"division by 0"

[surreal path]

minus the exceptions ofc

[flat nymph]

me neither

[surreal path]

this is genuinely so cringe

i want to work on nyaux but

the loneliness of not having a relationship anymore

fucking hell why i always get these feelings

😡

[finite summit]

t'is natural

[surreal path]

its annoying

genuinely

[kind root]

What happened

[surreal path]

happened like a week ago didnt wanna say anything

teenage girls are really mean

like

REALLY

[finite summit]

osdev has no end
girlfriend has end

choose one

[surreal path]

osdev lol

but im human so i cannot escape these kind of feelings

which sucks ass

[kind root]

Be an ai

[surreal path]

lol

[finite summit]

program until you transcend feeling

until it's all you think about

get a tumor in your amygdala

[hexed mango]

teenagers in general aren't that great tbf

Hope you feel better soon

[plush hearth]

Yep

[finite summit]

fr

[surreal path]

i napped and now its 2am

[finite summit]

uh oh

[surreal path]

yep

ruined my sleep schedule further

😭

[surreal path]

i had the single worst sleep

known to existence

[flat nymph]

You haven't had university entrance exams yet

(and university in general if you're gonna do that)

[elder shoal]

The joy of spending 5h30m in one day and 5h the other day doing an exam

[flat nymph]

Although maybe I had the best sleep after that

I remember being knocked out for 3 days after them

With no memory of them

[cinder plinth]

meanwhile during exams I started at like 8

so I couldnt sleep

[surreal path]

anyone have a throwaway ec laptop

my only laptop has EC

but

i dont want to brick it

im not taking the risk

@kind root do u know anyone with a throwaway ec laptop

also do u have resources on EC

[flat nymph]

Dw

Unless you do something stupid

(which now if you come to think of it could totally happen in nyaux)

[green elbow]

bro rip

[flat nymph]

Nothing bad should happen

[surreal path]

i hope.

[surreal path]

yea

😔

[surreal path]

yes.

it probs would

but

ill have to take the risk with my laptop

if i brick it tho

no more osdev for a long time, ill have to save up for a new computer

and ill only have a phone

[surreal path]

ill ask more abt this after i finish my route irq function

[elder shoal]

Dw girls are temporary NYAUX IS FOREVER 💪

[surreal path]

YESSS

[kind root]

dont worry u wont corrupt it to an extent where its not bootable

worst case you'll have to pull out the battery

[surreal path]

i dont have a removable battery laptop

[kind root]

lmao

it must have some hard reset thing

[surreal path]

no

it does not

[molten grotto]

no hole at the bottom or anything?

my laptop has a tiny hole and a button under it

[flat nymph]

I think that laptops should have the same thing as phones, where if you hold a power button for a long time it should disconnect the board from power

[surreal path]

no

[silver yarrow]

mine does

it takes like a minute

[surreal path]

lucky bro

anyways

route irq function time

[flat nymph]

You can just use MSI

[surreal path]

hold on lemme js test nyaux on my laptop rq

[surreal path]

no i mean

for the ioapic

route irq function for the ioapic that is all

then its EC

i have lost my usb.

fuck

sooo

anyone willing to flash and test nyaux on real hw rq.

i have lost my usb.

i want to check something with systems with multiple ioapics

[flat nymph]

Maybe tomorrow

[surreal path]

hold on

we maybe be saved

i found a usb sd card reader

and i have an sd card in my drone

we are saved

nyaux development continues

nvm

we are not saved

my only sd card reader just broke apart

like the entire shell.

broke

pcb came off the usb connector too

JUST MY FUCKING LUCK

😭

soooo chatttt

i not only need a person with a usb drive

i need a person whos willing to test nyaux on real hw a billion times

not just for the ioapic

but for even EC

@rigid fable mr duck i need ur help

how

the actual fuck

am i going to work on nyaux

now

[silver yarrow]

cant u pickup another usb

[surreal path]

i dont have another usb

[silver yarrow]

I meant buy one

[surreal path]

im not an adult

no 1

no 2

my dad wont let me leave the house til he comes home

[silver yarrow]

how did u lose the usb had in the first placea

[surreal path]

no idea

just disappeared

lol

anyways i found a victim friend to help me test it

some random guy who has a usb and a recent computer

[rigid fable]

who ping

[thorn bramble]

there's this cool thing to find that out

in top right of your discord client

[surreal path]

!!

vm

nvm

its goign nowhere

[surreal path]

mr duck are u still alive

i have a favor

to ask you

flash this on a usb or smth and test it on real hw please

[silver yarrow]

this is a shot in the dark but do u have a rooted android phone by any chance?

[surreal path]

no

i have an iphone

[silver yarrow]

rip, couldve used that as a bootable drive

[surreal path]

yea

:(

[rigid fable]

I am scared

[silver yarrow]

lol

[surreal path]

please just trust me its a normal iso

view it under 7zip or smt

straight outta the makefile

i am beggin

😭

[rigid fable]

okY

I'll put it on ventoy

[silver yarrow]

Nyaux automated testing pipeline: Nyauxmaster -> discord -> 🦆 -> discord

[surreal path]

yes

[surreal path]

😎

[rigid fable]

i'll test it on 2 pc

[surreal path]

okay

[surreal path]

damn

no multiple ioapics

😔

im fucked

[rigid fable]

crap limine doesn't support bios ventoy

[surreal path]

damn @wicked loom that is normal right?

i think that is?

[rigid fable]

it supports only uefi

[surreal path]

no?

[finite summit]

Huh?

[surreal path]

it supports both?

[finite summit]

The crap here is ventoy

Idk why people even use that crap

[rigid fable]

putting iso in filesystems is safer

[kind root]

i recommend implementing the entire kernel api, especially pci io before running on real hw

[surreal path]

i dont have pci yet

lol

i am just testing something

[kind root]

you will end up with half initialized firmware

[surreal path]

i know but im still gonna write the EC driver

[rigid fable]

what is EC

and why this is extists

[surreal path]

embeeded controlerl

[rigid fable]

and why it exists

[wicked loom]

i mean yeah

Limine doesn't need to support Ventoy

Ventoy needs to support Limine

[rigid fable]

it support uefi

[rigid fable]

👆

[surreal path]

so

does anyone

have a computer with mutiple apics

[wicked loom]

probably only bigger servers/workstations do

[surreal path]

my laptop does

[wicked loom]

maybe you can tell QEMU to emulate that

ah

[surreal path]

to which i cant test it

[wicked loom]

well that's cool

[surreal path]

BECAUSE I DONT HAVE A USB

😭

[rigid fable]

do you have

[surreal path]

i lost it

[rigid fable]

external drives

[surreal path]

no

[rigid fable]

what is your bootloader

[surreal path]

i dont have a usb

either way

[rigid fable]

on your

main hard drive

what is bootloader

[surreal path]

windows boot manager

and grub

ig

[rigid fable]

you can

[cinder plinth]

You could just make a partition on your disk

And flash the drive there

[rigid fable]

load iso file to ram

with grub

[cinder plinth]

And boot from there

[rigid fable]

and chainload efi file

[surreal path]

how so

[cinder plinth]

wdym

[rigid fable]

load iso to ram and chainload it

[cinder plinth]

You just reserve a few gigs

[rigid fable]

grub can do it

[cinder plinth]

And dd your iso there

[surreal path]

im on windows rn

hold on

[rigid fable]

windows 😡

[kind root]

ikr

ventoy never works

[rigid fable]

why are you hating ventoy

[wicked loom]

Ventoy is a hack, he is right in that

but there is no way to make something like what Ventoy wants to achieve without being a hack that barely works

[surreal path]

fair

[rigid fable]

😭

[flat nymph]

AMD PCs seem to have it

ventoy is convenient though

[wicked loom]

for what it is, it's a miracle it works as well as it does, yes

[flat nymph]

what's missing though

ventoy is open source (?)

[tawdry mirage]

amd machines usually have 2 ioapics

[flat nymph]

fine i'll test nyaux on my desktop

[wicked loom]

that's interesting, now that it was mentioned, i kind of remember that being the case

[tawdry mirage]

on my laptop: ```
[ 0.035718] IOAPIC[0]: apic_id 7, version 33, address 0xfec00000, GSI 0-23
[ 0.035722] IOAPIC[1]: apic_id 8, version 33, address 0xfec01000, GSI 24-55

[surreal path]

thank you

so much

[surreal path]

i needed this thanks qookie

ur a life saver

[surreal path]

thank you

[flat nymph]

i think it's instantly tripple faulting

That PC has a serial port so I can share the logs if you write them there

[kind root]

at least it doesnt brick it

[flat nymph]

It doesn't have EC so nothing to worry about

@surreal path

[surreal path]

FUCK

share the logs

please

read from com1

[surreal path]

theres no ec yet

in nyaux

just share the logs @flat nymph please

we may have another nyaux bug

[kind root]

no way

maybe his pc is just broken

[surreal path]

no im assuming its my code

it prob is

some weird obsecure bug that only happens on hardware like his because of one device maybe idfk

[flat nymph]

I'm booting the second PC

[surreal path]

okay

[flat nymph]

pmOS works on so it can't be

Even the framebuffer

[kind root]

/s

[flat nymph]

(another secretly fucked os)

[surreal path]

yep

cant get enough of these bugs

i swear

[flat nymph]

It takes like half an hour to boot though

[kind root]

wtf why

[flat nymph]

It has two broken HDDs

With bad sectors

[surreal path]

im gonna

[flat nymph]

And stuff

One was pulled from a laptop

[surreal path]

why is it theres always a new bug in nyaux

ugh

[flat nymph]

Because it was dying

So it's reading at 10MB/s

With seek times of seconds

But Linux boots on it so that's fine

[surreal path]

whats the cpu on the pc that triple faulted

how much ram

[flat nymph]

Ryzen 9 5900X

[surreal path]

wait ram shoulnt matter

[flat nymph]

64GB ram

[surreal path]

well fuck

so

nyaux hates mne

[flat nymph]

Resizeable BAR

A.k.a. MMIO mapped above 4GB

[surreal path]

maybe thats the problem?

no

it would have just said it faulted

i install an idt immeditely after boot

idfk

[flat nymph]

fun stuff

(I think it booted)

(Linux that is)

[surreal path]

can u try booting nyaux

[flat nymph]

What's your com port parameters

[surreal path]

theres no parameters

i just write characters to com1

[flat nymph]

Wdym

[surreal path]

literarly just write chars

[flat nymph]

That's not how that works

[surreal path]

i dont even init it

[flat nymph]

Lol

[surreal path]

it works on qemu

[flat nymph]

Yeah no

[thorn bramble]

ofc it works on qemu

everything works on qemu

-# even nyaux

[surreal path]

well time to debug another bug

[thorn bramble]

what bug are u having

[surreal path]

:)\

[surreal path]

nyaux triple faults on his pc

[flat nymph]

It's not a bug

Oh, that is a bug

[surreal path]

i tried nyaux just minutes ago on my laptop

i got lucky

found a usb

that kinda works

its shitty tho

and sometimes just refuses to work

but it ran nyaux fine

no issues

no triple fault or any nonsense

and i learned my laptop does not have 2 ioapics

[thorn bramble]

lmao

[surreal path]

ofc it is

nothing in osdev is simple

[finite summit]

nyaux triple faulting is a feature confirmed

[surreal path]

@flat nymph are u sure u flashed the iso

and are not using an older version of nyaux

[flat nymph]

Wdym am I sure

[surreal path]

i just want to make sure

[flat nymph]

I've been testing pmOS previously

[surreal path]

understandable

im gonna actually crt

cry*

@flat nymph did nayaux boot yet on that pc

nyaux*

[flat nymph]

What do you want me to do

What's the rate of the serial port

[surreal path]

theres no rate

i dont init

the serial port

i just write bytes

[flat nymph]

then there's no serial port

[surreal path]

fun

hold on while i properly init the serial port

ill write some quick code

im actually just going to yoink this init code from the wiki

i need to see this now

[flat nymph]

You also need to be waiting for the port to be empty when writing

[surreal path]

yea

stole code for that too

from the wiki

try this please

@flat nymph

baud rate is 38400

[flat nymph]

balenaetcher is refusing to work

I'm gonna flash ventoy

[surreal path]

just use rufus

if ur on windows

if ur on linux js dd the iso

[flat nymph]

it's headless

[surreal path]

dd it

[flat nymph]

�z��2c���Y�d�)2�8QY�

[surreal path]

im going to kill myself

cant even fucking debug it

im going to actually go into tears

now we have a bug thats not even debugable

@flat nymph i dont know what to do anymore

[flat nymph]

Do you do this?

[surreal path]

yes

[flat nymph]

I'm gonna try pmOS just to make sure the PC is sane

Since I've gotten a new motherboard recently

[tawdry mirage]

did you set the baud rate to 38400?

[surreal path]

yes its lit copied osdev wiki code

[tawdry mirage]

that's what the code nyaux copied sets it to

[surreal path]

i didnt even bother to write it myself

[flat nymph]

sudo screen /dev/ttyS0 38400

When do you start printing logs?

[surreal path]

literary after init temr

[flat nymph]

?

[flat nymph]

have you considered doing it before that?

Hmm pmOS is loosing half of the logs

[surreal path]

i dont print anything before init term

and im actually starting to cry because i thought i was so done with these bugs and now we have a huge bug thats probs gonna be undebuggable because even the fucking serial port doesnt work

[flat nymph]

My serial cable is skill issued for some reason

I'm blaming the PSU

[surreal path]

it doesnt matter

nyaux is bugged

i have no way of testing it

i want to bang my head againist the wall

i dont know what to do

@flat nymph any ideas at all?

[thorn bramble]

what is the bug

sorry

i think i asked this earlier

but i forgot to check in

[surreal path]

it doesnt work on his machine

[thorn bramble]

ah the triple fault

[surreal path]

works on my machine

fine

works on ducks machine

fine

but his machine

NOOO

[thorn bramble]

find another machine it doesnt work on that u have access to

[surreal path]

i dont have

at all

[thorn bramble]

didnt it also crash in vbox?

[surreal path]

no

it didnt

[flat nymph]

Another PC which is mine

[thorn bramble]

i remember you had some issues with vbox

hmm

whatever

[surreal path]

very old bug

that was solved a while ago

[thorn bramble]

yeah if mishakov is fine with it then you can just start by commenting out everything

[kind root]

sorry, but why cant u just get serial logs?

[thorn bramble]

then comment back in one thing

[flat nymph]

 map regFF2000 - 0xC13CCmory map region 14: 0xC1 0xC16CC region  0xFEDD658: 0xFF 59: 0x1mory map - 0x102C000 - 0000, type 6
 Memory map region 2F1DD000 - 0x102F23F000,mory map2F2A5000F300000,0000 - 0x1030000000, typ region 00000000
                                                                                                           Found tst memor at 0x100000000,F2EA60000
Initializing region 00xA20E0000000
Sp PMM re020000 -PMM region 0xC1600
80000007 6799000 - 0xC327E00ng up PMM regionzing interrupts
[Info: TSant bit lizing pidle tasinit
Moline: inclaimablC1C000 -ng bootle
CPU 1e
CPU 1PU 1B enryspsaca4C5P)U  1Fte r(i6s5e5r0.de)r
                                                 spa/idle
CPU 1 enteDerbsupsyscall failed)

[thorn bramble]

and put a for(;;) asm volatile ("hlt")

[flat nymph]

logs from pmOS

[kind root]

most stable nyaux driver

[flat nymph]

which were working before

[thorn bramble]

that's pmos

not nyaux

[flat nymph]

I'm blaming the power supply

[surreal path]

his serial port is ass

[thorn bramble]

.

[thorn bramble]

.

[surreal path]

k

[thorn bramble]

do this until you find the thing that crashes

[surreal path]

will do

[kind root]

ah

[thorn bramble]

if you narrow it down like that it will be easier to find the bug yourself

because eventually you find the function that leads to the crash

and you can audit it yourself

even without logs it should be easy to find what's going on

[surreal path]

try this @flat nymph

i commented out everything but one print

[thorn bramble]

unless it's some sort of scheduler bug then that won't work

[surreal path]

also mishakov

prepare for a lot of flashing

[kind root]

dont flash him twf

[surreal path]

💀

[kind root]

first a comment about fucking nyaux now this ☠️

[surreal path]

nahhhh 💀

ur dirty minded

nahhh

[finite summit]

ban him

[surreal path]

[finite summit]

.!naenae @surreal path being inappropriate

[safe spire]

Command executed by oberrow

Command Error

Type

Permission Error

Details

You must have BAN_MEMBERS to execute this command.

Dec 27, 2024, 8:16 PM

[flat nymph]

btw

[surreal path]

yea?

[flat nymph]

while the funny PC is booted

I can the ACPI dump

[surreal path]

yea but its def fnot the aml

[flat nymph]

Tripple fault

[surreal path]

I COMMENTED OUT EVERYTHING

TF

WHAT

WHAT

WHAT

[flat nymph]

���Y�d�)�z��2c���Y�d�)8Q

[kind root]

most stable entrypoint

[flat nymph]

At least it's consistent

[surreal path]

BRO

FINE LETS COMMENT OUT THE KPRINTF TOO

[kind root]

are u sure u recompiled it

[surreal path]

yes

im sure

u can see by the file size

[thorn bramble]

what limine base revision are you using

[surreal path]

2

[thorn bramble]

hm ok

well

the framebuffer might be >4G

and that might crash lol

[surreal path]

why

[flat nymph]

It probably is

[thorn bramble]

well take a guess?

lol

unless limine uncoditionally makes the framebuffer

even if >4G

[flat nymph]

Fuck why does it boot into windows all the time

[thorn bramble]

try commenting out the flanterm init

[surreal path]

k

[thorn bramble]

and ofc every usage of the flanterm context

because if u use it while its uninitialized you will crash

duh

[surreal path]

what

[thorn bramble]

but thats the only thing i can think of that might cause issues

what what

[surreal path]

i dont get

why would it crash

[kind root]

bruh

[thorn bramble]

limine only maps first 4G of memory for you

[kind root]

think harder

[thorn bramble]

if the framebuffer is above 4G you will page fault

and without an idt you crash

lol

[surreal path]

oh

yea

[thorn bramble]

UNLESS

[flat nymph]

Does it not map all memory?

[thorn bramble]

as i said

[surreal path]

forgot

[thorn bramble]

limine maps the framebuffer

which I DONT KNOW if it does

i think it maps all usable entries + 0x1000-4G unconditionally

[flat nymph]

tripple fault

[surreal path]

if the other iso crashed, try this. no flanterm init here

[surreal path]

okay so not even my code

[thorn bramble]

no, it is your code

[surreal path]

bro

[flat nymph]

Can you comment everything?

[surreal path]

i did

[thorn bramble]

you are calling flanterm_fb_init

[flat nymph]

kmain() { hlt(); }

[surreal path]

i did

[kind root]

by that logic uacpi crashing because of a broken allocator would also not be your code

[surreal path]

V

[thorn bramble]

isnt it a headless machine

it wont have a framebuffer

[surreal path]

flanterm fb init not my code

is mints

not mine

i dont own flanterm

[thorn bramble]

but ur giving it bad arguments

so its ur fault

[surreal path]

what

[flat nymph]

It's not, though it does probably have some weirdness with framebuffers

Since it has 2 4K screens

[thorn bramble]

that's not weird

[flat nymph]

One of which is 144Hz

So the firmware sets some obnoxious resolution

[thorn bramble]

check for fb response before dereferencing anything

[surreal path]

i do

[flat nymph]

With one of them, it sets it to 4K

[thorn bramble]

at least its not a 2004 laptop that barely knows what a vbe is

[surreal path]

framebuffer_request.response

[thorn bramble]

you do?

you sure?

[surreal path]

before

[thorn bramble]

Ah

[flat nymph]

Framebuffer does work in pmOS though

[thorn bramble]

<--- Blind ass mf

[flat nymph]

But I throw away Limine's hhdm very early on