Nyaux

[surreal path]

its litearrly directly from limine

@wicked loom

^

[kind root]

flanterm_fb_queue_item is not related to the framebuffer

[surreal path]

yea i have no idea

thats not my problem

i just init

and ask flanterm to put a char

thats it

[silver yarrow]

I have the alignment sanitization turned off idk

[surreal path]

no UB?

WTF

THERE WAS

NO UB PRINT

NONE

@silver yarrow wtf

[silver yarrow]

unfortunate

ubsan doesnt catch everything

just some basic UB

prolly something more complex going on

what exactly is the issue anyway

[surreal path]

so on some hardware the page tables fails to load

it just hangs

on some other hardware

it works completely fine 😭

all vms load nyaux fine

[silver yarrow]

well things dont really "just hang"

[surreal path]

vbox

qemu

kvm

[surreal path]

i even stop loading the idt

it didnt triple fault

OH WAIT

IT DIDNT TRTIPLE FAULT

IT DIDNT TRIPLE FAULT???

framebuffer issue

its a framebuffer issue

fuck me

[silver yarrow]

ubsan catches ub "when it happens" so prolly worth running this on the hardware that dont work

[surreal path]

ah yes i have a serial port

to connect to

[silver yarrow]

well

[orchid dawn]

after intentionally trying to reset the system, it does reset?

[surreal path]

holding the power button

wait

[orchid dawn]

then it's possible its a framebuffer issue
or it's possible your code is crashing

[surreal path]

asking uacpi to reset

i could try

@kind root how to ask uacpi to shutdown

oh hold on

nvm

sorry for the ping

found the wiki

[kind root]

yeah

it wont work btw because you return error on install_interrupt_handler

[surreal path]

hold on ill change that

oh because i didnt load an idt

[molten grotto]

no

tbh I didn't look at it after I found those two bugs, I got busy hating google

[surreal path]

okay it resetted on qemu

[surreal path]

understandable

[kind root]

resetted?

[surreal path]

like

it shutdonw

as ud expect

[kind root]

ah

[surreal path]

@molten grotto since ur here

can u see if it shuts down on ur machine?

cause i have something important opened i cant restart

[molten grotto]

maybe I should plug the laptop back to the ethernet cable, it would make testing easier with pxe just that then my website is going to die lol

[surreal path]

😭

[molten grotto]

I should buy an another cable, not that it matters that much as the website is literally just empty git server + a text file with hello world + a main page with links to those

anyway now I can try it

[surreal path]

alr coolss

[molten grotto]

no it just gets stuck after giving address

[surreal path]

so it didnt reset

so its NOT a framebuffer issue

even tho

i removed the idt

tested it on my laptop

didnt triple fault

even tho

i put back the idt and at the end of init

just request to shutdown

WTF

yea something very complex is going on

we may be stuck on this for weeks

inb4

[flat nymph]

gdbserver from monitor

[molten grotto]

its in real hw

[surreal path]

^

[molten grotto]

with the git build it does reboot

[surreal path]

wait

it does

okay so

framebuffer issue

welp

FUN!!!

why isnt the framebuffer reappearing

wtf

[molten grotto]

where did you add the triple fault code

[surreal path]

huh

i removed the idt to test if its the page tables

it didnt triple fault

theres no triple fault code

[molten grotto]

what are you using for the reboot then?

[surreal path]

uacpi

i readded the idt

and then

had this

@molten grotto

[molten grotto]

does the git code even have some code that intentionally causes a reboot

[surreal path]

wait no

it has no idt

ohhh

so it did triple fault

thats the git code

okay

and ubsan

didnt say shit

@molten grotto weve got a complicated issue

fun

[elder shoal]

in case you're curious, last I checked astral can boot to bash with as low as like 29 mbs of ram if you consider memory used

I checked it once when removing the limine backgrounhd

it boots to fvwm with like 128 mb

[surreal path]

nice

[flat nymph]

Limine uses a lot of memory...

[surreal path]

how much

i should really reclaim it

WHEN THIS BUG

GETS FIXED

I CANNOT FIGURE IT OUT

KILL ME CHAT

SEND ME A GLOCK M14 OVER MAIL

[edgy pilot]

[surreal path]

it does the same

yep

@flat nymph

ill remove the idt

so it can triple fault

okay?

holy shit i just crashed wsl2

@flat nymph take this iso

does it triple fault yes or no

let me knoqw

[edgy pilot]

you are on wsl?

smh

[surreal path]

yes

sadly

😔

[flat nymph]

Me too

[edgy pilot]

smh x2

[flat nymph]

Though I installed Linux yesterday

[edgy pilot]

which distro

[flat nymph]

To put printfs inside the kernel

Fedora

[edgy pilot]

based

[flat nymph]

But it led to us discovering why the power button wasn't working with uACPI

[surreal path]

[edgy pilot]

why does it take so long to boot

[surreal path]

pmm

anyways

did u test on real hardware

[edgy pilot]

no

[surreal path]

pain

im going to sleep

im too frustrated

[edgy pilot]

.insert(0, 1, 's');

[flat nymph]

Why does your PMM take so long

[surreal path]

i insert a node into each page

lol

anyways

im too frustrated to figure anything out

im going to sleep

[edgy pilot]

.erase(anyways.length() - 1);

[surreal path]

i hope a miracle will happen and maybe one of you spot something but thats obviously not gonna happen we are gonna be stuck for fucking weeks

goodnight

[edgy pilot]

goodnight ||piss|| role haver

[flat nymph]

Triple faults after freelist created

[surreal path]

nvm we arent sleeping

HUH???

WAIT SERIOUSLY

triple faults after freelist created...

mhm.

[edgy pilot]

https://tenor.com/view/sleep-go-to-sleep-monster-inc-sally-gif-17177651

[surreal path]

okay so this is like advanced UB

[edgy pilot]

dream up some solutions

[flat nymph]

I think

[surreal path]

that ubsan didnt spot

[flat nymph]

It printed something and triple faulted

[flat nymph]

Why do you think it's UB

[surreal path]

idk

could be anything

[flat nymph]

It can be heap corruption or something

[surreal path]

memory corruption

yea

[edgy pilot]

it's shrimply an issue of krill

[surreal path]

could be litearrly anything

thats fucking impossible to debug

wish i could attach a gdb debugger to real hardware chat

[flat nymph]

Like something shitting in all the pointers

[surreal path]

i took gdb for granted

[edgy pilot]

you can

[surreal path]

i dont have a serial port

so no prints

ubsan is enabled

[edgy pilot]

have you checked your mobo?

[surreal path]

btw

yep

[edgy pilot]

internal pins

[surreal path]

its a laptop

[flat nymph]

Put a bunch of while (1) ;'s

See when it freezes and when it crashes

[surreal path]

i cant do that

[edgy pilot]

yes you can

[flat nymph]

(but with asm inside)

[surreal path]

because i need to sleep

[edgy pilot]

eep first

debug later

[flat nymph]

That's how I was debugging my OS at some point

[surreal path]

wait i have an idea

hold on

[edgy pilot]

sleep?

I'm telling people to sleep at 3am smh

good night

[surreal path]

@flat nymph try

i put it here

before switching page tables

just to verify

[edgy pilot]

tripple

[flat nymph]

Why don't you install a double fault handler at least?

[surreal path]

i already have one

it just hangs

doesnt execute anything

when an idt is installed

and it switches page tables

nothing else js hangs

so js try please

[edgy pilot]

private nyauxmaster
you are given a direct order to sleep

[elder shoal]

go to sleep, programming tired wont do any good

[surreal path]

yep its the page tables

for which some hardware doesnt care

some does

like ur one and my laptop

[surreal path]

fine

goodnight for real this time

[tawdry mirage]

are you sure the physical mem you're using is actually free memory

[surreal path]

yep

i ignore everything but usable

if u need further context or have any questions u can look at the code here im sleeping

https://github.com/rayanmargham/NyauxKC

[wicked loom]

what

[surreal path]

ubsan is saying a bunch of stuff with flanterm for some reason

[wicked loom]

ur issue

[surreal path]

how

im not even giving it my allocator or anything

[wicked loom]

i ran ubsan and asan and fuzzed flanterm many times

in userland

without an issue

[surreal path]

maybe its an old version then

to be fair i havent updated flanterm in a while

[wicked loom]

update it

[surreal path]

will do when i finish sleeping

lol

[wicked loom]

ok lol

[kind root]

this man can chat about osdev while sleeping

[surreal path]

its because i have my phone notifying me cause i have my phone under my pillow

still havent slept but eyes are really tired

ima power off my phone

[surreal path]

morning

soooo

i guess we will

update flanterm

ask i said i would

as*

hopefully that will clear all ubsan messages

then idk

figure out SOME way to get closer to the actual issue

does anyone here have a computer with a serial port?

and something to read that serial port?

also tbh

its very out of spec for some cpus to outright refuse to load my page tables

and hang

but some do

ill try to make an issue on the forums as well

lets hope mint didnt change the api for the billionth time

💀

okay good

damn no commit in 2 months

ehhhhh

[flat nymph]

I do

[surreal path]

do u have something to read that serial port?

[flat nymph]

Two computers with serial ports

And a modem cable

One with AMD A10 and one with Ryzen 9 5900X

[surreal path]

okay

[flat nymph]

So relatively new and 10 years old hardware

[surreal path]

should i add a .note.GNU.stack or smthin

ill js disable the warning

[molten grotto]

yes or alternatively just pass -z noexecstack to the linker

[surreal path]

yea

that solved it

hold on while i enable ubsan

and remove the flanterm checkj

its still mad?

mint said its my problem

okay

[kind root]

perhaps it is a flanterm bug

[surreal path]

no because mints code is never wrong!

[kind root]

maybe its relying on some alignment that is not guaranteed

[surreal path]

she did say she did it multiple times with santiziers or smthin

[surreal path]

that ^

so

🤷‍♂️

[kind root]

doesnt guarantee anything

[surreal path]

yea i guess

[kind root]

i also run uacpi with both but still have ub uncovered every once in a while

[surreal path]

should tell mint this

obos reference?

anyways

how are we gonna figure out our stupid hanging issue

any ideas

guessing not

come on there has to be SOMETHING we can do

gdb pt dump?

maybe?

well i will try

i have no other options lmao

BRO ITS A PYTHON FILE

[haughty kite]

fucking what

you tried running a file with a .py file extension as a shell script

also its a gdb plugin iirc

[surreal path]

they said source it

so im sourcing it

[haughty kite]

IN GDB

[surreal path]

ohhh

thanks for the heads up

[molten grotto]

btw the flanterm issues are very likely because the flanterm bump alloc doesn't guarantee any kind of alignment from the looks of it

[surreal path]

understandable

mint should fix that ngl

[kind root]

yes

[surreal path]

but thats not our issue rn

[kind root]

btw i recently uncovered -Wstrict-prototypes, very good flag u should enable it

[surreal path]

heres what it looks like

anything weird?

[kind root]

you tell us

[surreal path]

looks fine

ill walk through each one

huh

oh thats 2mb

also with this?

this looks fine

that as well

and this too

okay

the only weird one

[surreal path]

is this

why is it 0x0

tf

and why doesnt the vm care

[molten grotto]

because its the start of the hhdm that maps to physical addr 0?

[surreal path]

oh yea

thanks

sorry

[surreal path]

but why is this also mapped 2mb?

oh its the memory map

yea so

gdb pt dump says the page tables are fine

okay

so wtf

[surreal path]

enabling now

yea doing that changed nothing

[kind root]

do u have -Wpedantic?

[surreal path]

no ill enable that

[cinder plinth]

That's just for standards, no?

[surreal path]

yea it is

[kind root]

maybe

[surreal path]

yea i dont see anything much

fun

okay if pt dump said my page tables were fine

i dont have any more ideas

someone give me ideas please

[flat nymph]

Are you clearing pages when building page tables?

[surreal path]

yep

[flat nymph]

(also, limine had memory randomization)

[haughty kite]

what does it do?

[kind root]

it errors out if u have non-static definition without a previous declaration, or if u have void do_stuff(<no void here>)

[haughty kite]

oh cool

[flat nymph]

-Wall?

[kind root]

not included there

not even in -Wextra

[flat nymph]

Wtf, how was it even a feature

The old style declarations

[thorn bramble]

that's not a thing

you realize right

moving into cr3 does not cause the page tables to be read

[surreal path]

ik

[thorn bramble]

if it works in a virtual machine it means you're not handling memory correctly somewhere

which is what i gathered from two days worth of messages so far

works in qeme under tcg and kvm but doesnt on real hw

make sure you're zeroing out the pages you allocate for page tables before using them

[surreal path]

i am tho

[thorn bramble]

and make sure you are constructing the page table entries correctly

[surreal path]

ii even randomized the ram and used it as memory on qemu

i am

gdb pt dump says nothing suspicous

looks all good

i dont know what to do

[flat nymph]

Put while (1) ;'s

[surreal path]

I DID

[flat nymph]

Put printfs everythere

[surreal path]

i know where it happens

switching to cr3

that where it dies

i put printfs everywhere too as well

[flat nymph]

Then you're not mapping kernel regions correctly

[surreal path]

but gdb pt dump says everything is fine?

so like

how

[finite summit]

You're leaving stuff out

[surreal path]

wdym

say rn

@finite summit

what am i leaving out

[finite summit]

Forgetting to map stuff

[surreal path]

like what

[finite summit]

Up to you to find out

[surreal path]

i mapped the memory map, kernel, hhdm, and i map vmm regions

[rigid fable]

why nyaux is archived

[surreal path]

how would that even work, wouldnt the vm also die

[surreal path]

its not?

[finite summit]

Nyauxkc is the new repo

[surreal path]

yea

[surreal path]

please use nyauxkc

https://github.com/rayanmargham/NyauxKC

[rigid fable]

I didn't know

[surreal path]

i have gone through many rewrites

[surreal path]

?

[rigid fable]

you got a star from duck 🦆

[surreal path]

thanks duck

[rigid fable]

create

another pml4

and put into it all usable entries and etc

[surreal path]

thats what i do

[rigid fable]

ye

i have kernel pml4 and pml4 for pmm and vmm (all mapped)

[surreal path]

wha

[rigid fable]

like

[surreal path]

@finite summit

how would it work

[rigid fable]

i have two pml4

[surreal path]

if im missing something to map

wouldnt the vm also die

like

why does it happen only on real hardware

[rigid fable]

page fault ?

[surreal path]

and well

im mapping everything

look at that clearly default case

like

i dont understand

[rigid fable]

why not

[surreal path]

i dont understand why my kernel dies on real hw

[rigid fable]

void vmmMapEntry(uint64_t* pml4,uint16_t type,uint64_t flags) {
    struct limine_memmap_entry* current_entry;
    for(uint64_t i = 0; i < getMemMap()->response->entry_count;i++) {
        current_entry = getMemMap()->response->entries[i];
        if(current_entry->type == type) {
            for(uint64_t i = current_entry->base;i < current_entry->base + current_entry->length;i+=PAGE_SIZE) {
                vmmMapPage(pml4,i,(uint64_t)phys2Virt(i),flags);

            }
        }
    }
}

[surreal path]

i know what im doing

i already map the memory map

[rigid fable]

i always testing my os on real hw

[surreal path]

thats not the issue

[rigid fable]

check cr2

if it page faults

[surreal path]

it doesnt

it hangs when switching to cr3

and triple faults with no idt

so

[rigid fable]

then your kernel is not mapped

[surreal path]

it is

[rigid fable]

map kernel

[surreal path]

i clearly map it

[rigid fable]

send code idk

[surreal path]

fine

hold on

https://github.com/rayanmargham/NyauxKC/blob/master/kernel/src/arch/x86_64/page_tables/pt.c

[thorn bramble]

two isn't enough

[rigid fable]

try

to map kernel after mapping a entrys

[surreal path]

💀

[surreal path]

what difference does this make

[thorn bramble]

6 here and still counting

get on my level

[surreal path]

[rigid fable]

just try

after mapping entrys map kernel

[surreal path]

im not gonna flash a usb and reboot my laptop for this

i know that wont change anything

[rigid fable]

why not ventoy

[surreal path]

it doesnt matter what order

[rigid fable]

?

it comfortable

[surreal path]

its not when u have a billion tabs open

then i have to save my session

then i have to fucking idk

[thorn bramble]

excuses excuses

[surreal path]

so ur agreeing with duck that the order of how u map things matters 💀

[rigid fable]

what address pmm alloc returns ?

physical or virtual

[surreal path]

a hhdm address

virtual

i have to unadd the hhdm offset before utilizing it

to which i do in every instance

if i didnt

page fault

which the vm would catch

[thorn bramble]

it could depending on how shit ur code is

[surreal path]

bro

[thorn bramble]

it shouldn't tho

[surreal path]

my code doesnt wrok like that

[rigid fable]

my eyes burn on pt code

[surreal path]

even my code isnt that shit

[rigid fable]

i think this is anti copy paste defence

[surreal path]

good

its messy on purpose i swear

[molten grotto]

the style that you use is annoying to read

like who tf unironically uses 2 space tabs in their own project or literally no newlines whatsoever lol

[surreal path]

😭

idk man

[rigid fable]

you don't understand this is copy paste protection!!!

[surreal path]

vscode has formatting enabled so

[rigid fable]

[surreal path]

okay

step 1

figure out why some real hardware works no problem

but some real hardware doesnt

maybe by finding this out we could get closer

okay so

@orchid dawn @flat nymph did u try nyaux on a laptop or desktop, cause i know yall reported nyaux hanged as well

@molten grotto as well as u qwinci

[flat nymph]

laptop

[surreal path]

okay

2 laptops

interesting

[finite summit]

If it hangs on smp

[surreal path]

not on smp

hangs on switching page tables

[finite summit]

Well it might hang on smp sometime later

[molten grotto]

yes I tested it on my laptop, the build that you sent just hanged but the build that I made from git just triple faulted

[finite summit]

Because hyper threading

[surreal path]

no cause we know its triple faulting from the tables

[finite summit]

Lul

[surreal path]

yea the build that i sent had an idt installed

the git didnt

okay thats 3 laptops

[rigid fable]

rewrite it

[surreal path]

no

[rigid fable]

and test on real hw

!

[surreal path]

im not rewriting shit

im done with that hell

[surreal path]

so we far as i know no one has said a desktop in which nyaux hangs...

[rigid fable]

nooo

[surreal path]

that is interesting

[rigid fable]

😭

[molten grotto]

well idk if anyone has even tested it on a desktop lol

[surreal path]

i have

works fine

on my hp desktop that my sister has

[molten grotto]

I also could but its the pc that I normally use so its annoying

and you don't really gain anything from it even if it works

[surreal path]

well

we would know that there is a connection between hanging and laptops

that would be helpful maybe

idfk

[molten grotto]

it very likely doesn't really help with finding the problem

[surreal path]

man im so out of ideas for how to fix this ive gone schizo

💀

[surreal path]

well

do u have better ideas?

no?

[flat nymph]

Intel/AMD?

[surreal path]

both intel cpus the laptop and desktop i tried

13th gen on my laptop

some 9th gen i5 on the desktop\

[flat nymph]

Most of my PCs are AMD

[molten grotto]

no

[surreal path]

exactly

whos the debugging god now

nobody!

[molten grotto]

or well except looking more at the code lol

[surreal path]

i barely slept looking at the code

yea im lost

[flat nymph]

2MB?

It's 1MB on AMD64

[surreal path]

2MB?

1MB?

huh?

[flat nymph]

You're passing 2MB everywhere

[molten grotto]

no its 2mb pages on x86_64

1mb is for 32-bit iirc

[surreal path]

yuh

okay

heres what im gonna do

nothing

but

just beg a bunch of forums or something for help

💀

yea idfk

[flat nymph]

Use 4KB pages and that's it

[surreal path]

2MB pages are fine

plus kernel dies without them for the hhdm

cause we run out of memory

which is expected

for how much i map

[flat nymph]

Oh I'm stupid

[surreal path]

lol

okay time to ask #osdev-misc-0 , osdev forums and the osdev irc channel

there has to be ONE answer that might get us closer

i hope

i will ask all three when i arrive at my moms, currently waiting for my step mom to collect me

[flat nymph]

No you're mapping 2mb pages

void map2mb(uint64_t *pt, uint64_t phys, uint64_t virt, uint64_t flags) {
  uint64_t *f = find_pte_and_allocate2mb(pt, virt);
  *f = (phys & ~0x1ffffful) | flags | PAGE2MB;
}

[surreal path]

yea

what about that

lol

[flat nymph]

This (1fffff) masks 21 bits

[surreal path]

uhhhhh its the same on x86-64

lol

i think for both

[flat nymph]

No

Think

Hmm

[molten grotto]

there are 21 offset bits for 2mb

[flat nymph]

Never mind

It's 4MB on ia32

I need some sleep

Sorry

[surreal path]

its fine dw

at least ur trying to help

[flat nymph]

It's 1GB 2MB 4KB

[surreal path]

which i appreciate a lot

[molten grotto]

btw the memset is still wrong, rn you don't tell the compiler that the input is modified (but this is very likely unrelated as I don't think the output from it is used by the compiler)

[flat nymph]

Are you only mapping 4GB?

[molten grotto]

it should be something like ```c
asm volatile("rep stosb"
: "+D"(s), "+c"(n)
: "a"(c)
: "memory");

[flat nymph]

you're mapping everything

Btw, your hhdm_pages accounting is probably wrong

But it shouldn't affect anything

Can you make a function which prints the page mappings and print that in VM and on physical hw?

Your paging code is very convoluted

Like walking the page tables

[edgy pilot]

didn't you try to do that before too?

I think it might have been a different script

[flat nymph]

Like the code looks correct enough to me

But like if it's crashing loading cr3, then you have garbage in page tables

[rigid fable]

I think there's problems in his pmm

[flat nymph]

or that

But still, it's something that leads to garbage in page tables

[rigid fable]

tvoi glaza gorat na ego kode ?

[flat nymph]

нет

[rigid fable]

ia dumayu on nepravilno zamapil yadro

[flat nymph]

@surreal path how do I compile your kernel

[surreal path]

make

sorry im back

[rigid fable]

doesn't he have a makefile???

[surreal path]

thanks

[flat nymph]

Your makefiles are broken

[surreal path]

how

they work fine

show me error

[flat nymph]

misha@Yoga:~/NyauxKC$ make
./kernel/get-deps
+ dirname ./kernel/get-deps
+ srcdir=./kernel
+ test -z ./kernel
+ cd ./kernel
+ clone_repo_commit https://github.com/osdev0/freestnd-c-hdrs-0bsd.git freestnd-c-hdrs-0bsd 0353851fdebe0eb6a4d2c608c5393040d310bf35
+ test -d freestnd-c-hdrs-0bsd/.git
+ git -C freestnd-c-hdrs-0bsd reset --hard
HEAD is now at 0353851 Update macro names and include guards
+ git -C freestnd-c-hdrs-0bsd clean -fd
+ git -C freestnd-c-hdrs-0bsd checkout 0353851fdebe0eb6a4d2c608c5393040d310bf35
HEAD is now at 0353851 Update macro names and include guards
+ test -d freestnd-c-hdrs-0bsd
+ clone_repo_commit https://github.com/osdev0/cc-runtime.git cc-runtime 9c281448fe40498a6341d74f55c9af4f642c0b1d
+ test -d cc-runtime/.git
+ test -d cc-runtime
+ set +x
error: 'cc-runtime' is not a Git repository
make: *** [GNUmakefile:161: kernel-deps] Error 1

[rigid fable]

[surreal path]

thats the ./get-deps

not mine

from mints

try cleaning everything

and try again

if u have any more problems hop into vc ill try to help

[rigid fable]

have .gitignore

[surreal path]

i already have one

in the root of the directory

i will add it for cc-runtime

[rigid fable]

also are you mapping a reserved memory ?

[surreal path]

yep

[rigid fable]

if yes then it's bad

[molten grotto]

you put it in the wrong place

[surreal path]

wdym

[molten grotto]

+D and +c should be in the output section

[surreal path]

no u should map reversed memory

[rigid fable]

because it reduces the speed of paging initialization, etc.

[molten grotto]

doesn't really matter

[flat nymph]

Wtf is THE_REAL and AMONG_US?

Why are you doing it this way?

[surreal path]

okay thanks

[rigid fable]

depending on the map

[surreal path]

check the linker script

its stupid but

it works

[flat nymph]

Are you sure the long part is the pmm initialization and not mappings?

[rigid fable]

among us

[surreal path]

yep

@molten grotto

[rigid fable]

if you need it then map later

[surreal path]

wait hold on

[molten grotto]

...

[surreal path]

HOLD ON

QWINCI

RELAX

😭

[rigid fable]

but for now you only need a framebuffer, usable memory and bootloader reclaimable

[molten grotto]

• stack

actually nvm that's within the bl reclaim

[rigid fable]

I usually put it into the kernel

[surreal path]

okay i did it properly

but compiler still mad

[rigid fable]

why limine putting all responses on bl reclaimable

[molten grotto]

yeah but nyaux is using the limine provided stack

[surreal path]

okay i fixed

[molten grotto]

well they are reclaimable after all

[surreal path]

is the memcpy also wrong qwinici?

[thorn bramble]

rule 7 or something

[surreal path]

or is it fine

[thorn bramble]

also i love how i can understand that weird language

[rigid fable]

real

[surreal path]

nah memcpy is fine it seems

[thorn bramble]

why u write it like that

[rigid fable]

sorry 🙏

[thorn bramble]

instead of using cyrillic

[molten grotto]

yes you already fixed it

[surreal path]

yea

i forgor lol

[rigid fable]

why not

[thorn bramble]

fair

[surreal path]

are u still having trouble compiling?

if so hop in a vc and screenshare ill try to help

[rigid fable]

are you a stole makefile from template

[surreal path]

yes but modified slightly

ish

not rlly

but wtv

[rigid fable]

write yourself

[surreal path]

i just added a few flags

[thorn bramble]

iirc his kernel is based on the limine c template

[rigid fable]

its easy

[thorn bramble]

don't write it yourself

don't listen to him pls

[surreal path]

i hate writing makefiles

[surreal path]

ik

i would never

[rigid fable]

why

[flat nymph]

I'm in a lecture lol

[surreal path]

okay but if u have any issues js ping me

[thorn bramble]

because mint already wrote a better one than anyone will ever write in 5 minutes

[surreal path]

mhm

and mint will insta flam me

if i tried

[rigid fable]

and

🦆

[flat nymph]

It's not compiling

[molten grotto]

is it still the same thing

[surreal path]

hold on

[surreal path]

whats wrong

im upstreaming uacpi changes

hold on

can u pull my commit rq

remove this folder

go into

kernel

./git-deps

get*

and then try to make it

[rigid fable]

just write makefile yourself

[flat nymph]

fatal: No url found for submodule path 'kernel/cc-runtime' in .gitmodules

[surreal path]

hold on

[flat nymph]

Idk

[surreal path]

just wait

if u mind lol

[flat nymph]

Have you tried this?

[surreal path]

it isnt

ive had that issue before

its been solved

[flat nymph]

Entries almost certainly overlap

[surreal path]

how are u sure

[flat nymph]

my kernel prints mappings at the beginnings

[surreal path]

ive looked at them with gdb pt dump

if there was something wrong

it wouldve shown

on the vm

[surreal path]

can u show me the full context

like show me the full output

[rigid fable]

then problem in your pmm

[surreal path]

no not really

its really simple

i doubt anything with it

[rigid fable]

try to test it on real hw

[surreal path]

we've already done so

[rigid fable]

stress test pmm 😈

[haughty kite]

does it still hang if you turn off uacpi shit

[surreal path]

wdym

like not load uacpi

[haughty kite]

yeah

[surreal path]

i dont load uacpi until after i switch page tables

[haughty kite]

does it still break

[flat nymph]

That's the key

[haughty kite]

ah okay

[flat nymph]

Look at it on physical hw

[surreal path]

there will be too much mappings to print

[rigid fable]

virtual machines sometimes show incorrect results

[surreal path]

ngl

[rigid fable]

same timing

[flat nymph]

Wdym

[surreal path]

also @flat nymph please show the full output of ur command please

[rigid fable]

or caching

[surreal path]

let me help

[flat nymph]

There will be a lot of contiguous ones

[surreal path]

like it should compile

[flat nymph]

Finding the largest memory region...
 Memory map region 0: 0x1000 - 0x2000, type 5
 Memory map region 1: 0x2000 - 0xA0000, type 0
 Memory map region 2: 0x100000 - 0x800000, type 0
 Memory map region 3: 0x800000 - 0x808000, type 3
 Memory map region 4: 0x808000 - 0x80B000, type 0
 Memory map region 5: 0x80B000 - 0x80C000, type 3
 Memory map region 6: 0x80C000 - 0x811000, type 0
 Memory map region 7: 0x811000 - 0x900000, type 3
 Memory map region 8: 0x900000 - 0x796FE000, type 0
 Memory map region 9: 0x796FE000 - 0x79704000, type 5
 Memory map region 10: 0x79704000 - 0x79826000, type 6
 Memory map region 11: 0x79826000 - 0x79827000, type 5
 Memory map region 12: 0x79827000 - 0x79829000, type 0
 Memory map region 13: 0x79829000 - 0x7982A000, type 5
 Memory map region 14: 0x7982A000 - 0x79900000, type 6
 Memory map region 15: 0x79900000 - 0x79905000, type 5
 Memory map region 16: 0x79905000 - 0x79A6F000, type 6
 Memory map region 17: 0x79A6F000 - 0x79A74000, type 5
 Memory map region 18: 0x79A74000 - 0x79B71000, type 6
 Memory map region 19: 0x79B71000 - 0x79B76000, type 5
 Memory map region 20: 0x79B76000 - 0x79C68000, type 6
 Memory map region 21: 0x79C68000 - 0x79C6D000, type 5
 Memory map region 22: 0x79C6D000 - 0x79DA7000, type 6
 Memory map region 23: 0x79DA7000 - 0x79DAC000, type 5
 Memory map region 24: 0x79DAC000 - 0x79E98000, type 6
 Memory map region 25: 0x79E98000 - 0x79E9D000, type 5
 Memory map region 26: 0x79E9D000 - 0x79EFF000, type 6
 Memory map region 27: 0x79EFF000 - 0x79F04000, type 5
 Memory map region 28: 0x79F04000 - 0x79F6A000, type 6
 Memory map region 29: 0x79F6A000 - 0x79F6D000, type 5
 Memory map region 30: 0x79F6D000 - 0x79FC8000, type 6
 Memory map region 31: 0x79FC8000 - 0x7A019000, type 6
 Memory map region 32: 0x7A019000 - 0x7A204000, type 5
 Memory map region 33: 0x7A204000 - 0x7A207000, type 0
 Memory map region 34: 0x7A207000 - 0x7A23C000, type 5
 Memory map region 35: 0x7A23C000 - 0x7BB5C000, type 0
 Memory map region 36: 0x7BB5C000 - 0x7BBA4000, type 5
 Memory map region 37: 0x7BBA4000 - 0x7E28E000, type 0
 Memory map region 38: 0x7E28E000 - 0x7E293000, type 1
 Memory map region 39: 0x7E293000 - 0x7E2A4000, type 0
 Memory map region 40: 0x7E2A4000 - 0x7E311000, type 5
 Memory map region 41: 0x7E311000 - 0x7E31C000, type 1
 Memory map region 42: 0x7E31C000 - 0x7EA7E000, type 0
 Memory map region 43: 0x7EA7E000 - 0x7EB08000, type 1
 Memory map region 44: 0x7EB08000 - 0x7F4ED000, type 0
 Memory map region 45: 0x7F4ED000 - 0x7F5ED000, type 1
 Memory map region 46: 0x7F5ED000 - 0x7F6ED000, type 1
 Memory map region 47: 0x7F6ED000 - 0x7F76D000, type 1
 Memory map region 48: 0x7F76D000 - 0x7F77F000, type 2
 Memory map region 49: 0x7F77F000 - 0x7F7FF000, type 3
 Memory map region 50: 0x7F7FF000 - 0x7FE00000, type 0
 Memory map region 51: 0x7FE00000 - 0x7FE4F000, type 5
 Memory map region 52: 0x7FE4F000 - 0x7FE92000, type 0
 Memory map region 53: 0x7FE92000 - 0x7FE96000, type 1
 Memory map region 54: 0x7FE96000 - 0x7FE98000, type 3
 Memory map region 55: 0x7FE98000 - 0x7FEDC000, type 0
 Memory map region 56: 0x7FEDC000 - 0x7FF60000, type 1
 Memory map region 57: 0x7FF60000 - 0x80000000, type 3
 Memory map region 58: 0x80000000 - 0x803E8000, type 7
 Memory map region 59: 0xE0000000 - 0xF0000000, type 1
 Memory map region 60: 0xFD00000000 - 0x10000000000, type 1
Found the largest memory region at 0x900000, size 0x78DFE000

Limine allocates a lot of size 0x1000 stuff

[rigid fable]

why so many

[surreal path]

no i mean ur compiling command

[flat nymph]

On my PC at home it had 130 entries lol

[surreal path]

what did u do

[flat nymph]

misha@Yoga:~/NyauxKC$ make
./kernel/get-deps
+ dirname ./kernel/get-deps
+ srcdir=./kernel
+ test -z ./kernel
+ cd ./kernel
+ clone_repo_commit https://github.com/osdev0/freestnd-c-hdrs-0bsd.git freestnd-c-hdrs-0bsd 0353851fdebe0eb6a4d2c608c5393040d310bf35
+ test -d freestnd-c-hdrs-0bsd/.git
+ git -C freestnd-c-hdrs-0bsd reset --hard
HEAD is now at 0353851 Update macro names and include guards
+ git -C freestnd-c-hdrs-0bsd clean -fd
+ git -C freestnd-c-hdrs-0bsd checkout 0353851fdebe0eb6a4d2c608c5393040d310bf35
HEAD is now at 0353851 Update macro names and include guards
+ test -d freestnd-c-hdrs-0bsd
+ clone_repo_commit https://github.com/osdev0/cc-runtime.git cc-runtime 9c281448fe40498a6341d74f55c9af4f642c0b1d
+ test -d cc-runtime/.git
+ test -d cc-runtime
+ set +x
error: 'cc-runtime' is not a Git repository
make: *** [GNUmakefile:161: kernel-deps] Error 1

[rigid fable]

your motherboard is prankster

[flat nymph]

Nothing, it's what limine is passing

[surreal path]

i will remove the test .git

[molten grotto]

you can just mkdir cc-runtime/.git

[surreal path]

or that

js do that

[molten grotto]

the script is kinda stupid tbh

[surreal path]

yea

[flat nymph]

It woekd

[surreal path]

cool

[flat nymph]

src/term/../flanterm/backends/../flanterm.h:61:5: error: unknown type name ‘bool’
   61 |     bool autoflush;
      |     ^~~~
src/term/../flanterm/backends/../flanterm.h:62:5: error: unknown type name ‘bool’
   62 |     bool cursor_enabled;

[rigid fable]

[surreal path]

lol what?

[molten grotto]

what gcc version do you have

[surreal path]

run gcc --version

show the output

[rigid fable]

why not clang

[surreal path]

nyaux is compiled by default on -std=gnu2x

if this is causing issues

go into the makefile and change that

its js a flag

[surreal path]

i dont like clang

or well

i dont use it for C

[flat nymph]

gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0
Copyright (C) 2021 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

[rigid fable]

write makefile yourself: ❌
fix it yourself: ✅

[surreal path]

yea change that gnu flag

[molten grotto]

yeah that's def too old, has to be at least 13 or whatever for c23

[surreal path]

kernel/makefile

change

[molten grotto]

changing the flag isn't going to fix it if you use c23 features

[surreal path]

i dont rlly use c23 features atm

but i plan to in the future

ANYWAYs

change that flag for now

gnu11 or whatever u wanna use

[rigid fable]

what c23 have fearures

I never thought about it

[flat nymph]

Anyway

Why am I doing debugging for you

[surreal path]

because ur a kind person

[flat nymph]

Printf page mappings

Coalesce them so there aren't too many of them

The error will become immediately appartent

[surreal path]

fine u want me to print them so bad

fine

ill print every mapping

but can u still try to compile nyaux

prewtty plewease

guess not

[rigid fable]

great nyaux master don't be sad soon nyaux will be better than windows linux and macos in 9999 times

[surreal path]

okay i printed every mapping

complete dump

[kind root]

nyaux spoonfeeding challenge

[surreal path]

bru

i am not getting spoonfed

hes just

a nice person whos gonna debug the entire kernel for me instead of me having to do any work

/j

[kind root]

tru

[flat nymph]

Well here's your answer ```
vmm(x86_64)(): mapped 0xffff800000800000 to 0xffff800002000000 : from phys 0x800000 to 0x2000000
vmm(x86_64)(): mapped 0xffff800001e00000 to 0xffff800002000000 : from phys 0x1e00000 to 0x2000000
vmm(x86_64)(): mapped 0xffff800001e00000 to 0xffff800002200000 : from phys 0x1e00000 to 0x2200000

[kind root]

like if u dont learn to debug ez stuff like this its only downhill from here

[surreal path]

multiple overlapping mappings?

[surreal path]

i know how to debug

im js lazy sometimes

[surreal path]

yea shouldnt they just overwrite each other @flat nymph

[rigid fable]

try to debug qemu with all the ram of your pc

[surreal path]

hold on i need to piss

hold that thought

[rigid fable]

😩

[hexed mango]

I have some serial ported x86 servers

[kind root]

btw how is your progress with itanium os?

[surreal path]

ok back

[surreal path]

??

[flat nymph]

Yes but if you've mapping 2 different physical addresses to the same virtual one

[hexed mango]

It's going kinda meh
I can't test any of the itanium parts due to no working machines

[flat nymph]

So one of the mappings is gotta be incorrect

[surreal path]

understandable

so ummm

[kind root]

did they break or something?

[surreal path]

this is a good question

but why do i have overlapping mappings

i js map the 4GiB like a good boy

[rigid fable]

MAYBE his mapping entry's overwriting kernel map

[hexed mango]

The system board still doesn't work for some reason and I'm waiting on someone to hire me so I can buy more parts

[rigid fable]

I am always mapping the kernel after entry map

[kind root]

aw damn that sucks

[flat nymph]

No, kernel should be at -2GB

[surreal path]

@flat nymph is it because i do this unwillingly

is this bad

[rigid fable]

what if

Anything can happen

[surreal path]

becauase i both map 4Gib and the memory map

am i doing something wrong

obviously yes but what is it exactly

[flat nymph]

Idk ask @surreal path

[rigid fable]

wait

[surreal path]

no the kernel is higherhalf

[rigid fable]

you mapping 4 gb

and after mapping all entries ?

[surreal path]

yea

is that bad

😭

[rigid fable]

it's useless

[surreal path]

no because if i didnt do it nyaux dies

look ill remove the mapping 4gib code

[rigid fable]

That's how you map these miserable 4 GB

[flat nymph]

Could it be that 4GB mappings are correct and it works in a VM, but on my PC with 32GB of ram there's a bunch of stuff that's beyond it and so it breaks?

[surreal path]

maybe?

[rigid fable]

he after mapping a all entrys

[surreal path]

i have 16gb of ram

that desktop had 8

hold on im gonna give the vm like an absurb amount of ra

ram

[flat nymph]

No

Find a PC where it breaks

Print the mappings

And check that they are correct

[surreal path]

i.e. my laptop

oh yeah removing the 4gib mappings obviously breaks nyaux

for obvious reasons

[flat nymph]

Optionally, look at the code and fix the overlapping mappings

[surreal path]

i am looking at the code

i js dont understand why its overrlapping

now hold on ill reboot my pc into nyaux

[rigid fable]

lmao

[flat nymph]

One of the mappings here is not where it's supposed to be

[surreal path]

lemme give some of the debug messages difference

so i know if its coming from the memory map or

the 4gib mappings

[flat nymph]

Because your mappings are garbage

You don't need 4gb

[surreal path]

so what do i need to do then

i need to map the hhdm

this seems to be from the memorymap

[rigid fable]

think with brain

[surreal path]

not the 4gib mappings

[rigid fable]

you mapping ALL entries to hhdm

absolutely all physical memory

you don't need 4gb mapping

[surreal path]

yea arent u supposed to do that thats the point of ze hhdm lol

[surreal path]

oh

right

wait tho

wait

nonono

hold on

[flat nymph]

Idk I don't even use hhdm

[rigid fable]

direct mapping user

[surreal path]

nvm

[flat nymph]

No

Hhdm is direct mapping

[rigid fable]

I mean

0x1000 phys to 0x1000 virtual and etc

[flat nymph]

No

I don't use that

[rigid fable]

hooray

[surreal path]

as per suggestions i added this as well