Nyaux | Operating System Development | Page 27

hollow goblet Oct 26, 2024, 12:47 PM

#

for example rust for linux uses a fork of alloc that has stuff like fallible allocations and generally avoids panicking

finite summit Oct 26, 2024, 12:51 PM

#

hollow goblet for example rust for linux uses a fork of `alloc` that has stuff like fallible a...

Lmao

#

He is using C now

hollow goblet Oct 26, 2024, 12:51 PM

#

WHAAA

#

i thought nyaux was still rust

spice yarrow Oct 26, 2024, 3:31 PM

#

I need to know the first exception that was part of the triple fault, not the last one

#

Also you cropped off all the useful info like the error code

surreal path Oct 26, 2024, 4:21 PM

#

im using C

surreal path Oct 26, 2024, 4:21 PM

#

hollow goblet i thought nyaux was still rust

bro i got sick of the UB

hollow goblet Oct 26, 2024, 4:22 PM

#

it's easier to cause UB in C lol

#

rust basically says "uphold normal safe rust guarantees in ur unsafe code (keep it "sound")"

surreal path Oct 26, 2024, 4:26 PM

#

hollow goblet it's easier to cause UB in C lol

no bro

#

the llvm compiler

#

fucking hated me

hollow goblet Oct 26, 2024, 4:26 PM

#

what

hollow goblet Oct 26, 2024, 4:27 PM

#

surreal path no bro

i prefer sis <3

surreal path Oct 26, 2024, 4:27 PM

#

bro i did everything correct

surreal path Oct 26, 2024, 4:27 PM

#

hollow goblet i prefer sis <3

sorry :3

#

sis

hollow goblet Oct 26, 2024, 4:27 PM

#

np

surreal path Oct 26, 2024, 4:27 PM

#

surreal path bro i did everything correct

yet the compiler still threw some ub in uacpi or smthin idfk

#

im not debugging that

#

im sick of that

#

rust skill issue

#

idc

hollow goblet Oct 26, 2024, 4:27 PM

#

ffi safety is a thing but rustc doesn't compile uacpi :3

#

ffi safety: don't pass invalid values through ffi

molten grotto Oct 26, 2024, 4:28 PM

#

more likely the issues were just caused by you using rust like it was c and some other random unrelated issues

#

anyway yeah you are probably better doing it in c

surreal path Oct 26, 2024, 4:29 PM

#

molten grotto anyway yeah you are probably better doing it in c

yea

#

tbh

hollow goblet Oct 26, 2024, 4:30 PM

#

molten grotto more likely the issues were just caused by you using rust like it was c and some...

surely if i cast everything to references it's safe rust :clueless:

surreal path Oct 26, 2024, 4:41 PM

#

wtf

#

lol

#

sir

#

what

#

no

#

tf u on about 😭

#

okay so

#

whatever is going on

#

its happening here

#

no that takes too much ram

surreal path Oct 26, 2024, 4:43 PM

#

surreal path okay so

oh yeah im dumb dumb

#

wait no

#

im not

#

tf is going on

#

😭 ?

#

how did bro 0xd with this

#

i saw that

#

😡

#

oki i will slow down qemu

#

and see

#

now it suddenly works

#

huh?

#

TF???

#

it only works when i print it

#

bro tf is the compiler doing

#

bro

#

now C UB

#

fucking hell

#

am i a bad programmer

#

result pmm_init() {
  spinlock_lock(&pmmlock);
  volatile pnode *cur = &head;
  result ok = {.type = ERR, .err_msg = "pmm_init() failed"};
  kprintf("pmm_init(): entry count %d\n", memmap_request.response->entry_count);
  kprintf("%s(): The HHDM is 0x%lx\n", __FUNCTION__,
          hhdm_request.response->offset);
  for (uint64_t i = 0; i < memmap_request.response->entry_count; i++) {
    struct limine_memmap_entry *entry = memmap_request.response->entries[i];
    switch (entry->type) {
    case LIMINE_MEMMAP_USABLE:
      for (uint64_t i = 0; i < entry->length; i += 4096) {
        volatile pnode *hi =
            (pnode *)(entry->base + i + hhdm_request.response->offset);

        cur->next = (struct pnode *)hi;

        cur = hi;
      }
      break;
    }
  }
  kprintf("FreeList Created\n");
  kprintf("Free Pages %ju\n", get_free_pages());
  spinlock_unlock(&pmmlock);
  ok.type = OKAY;
  ok.okay = true;
  return ok;
}
uint64_t get_free_pages() {
  pnode *cur = &head;
  uint64_t page = 0;
  while (cur != NULL) {
    page += 1;
    cur = (pnode *)cur->next;
  };
  return page;
}

#

i dont understand

#

whats wrong with my code

#

nooo

#

this SHOULD work

#

all the addresses looked fine when i PRINTED them

#

but when i dont print them

#

fucking 0xd with some weird ah cr2

#

what is the compiler doing

#

bro

#

nooo

kind root Oct 26, 2024, 5:01 PM

#

the compiler is doing what u told it

surreal path Oct 26, 2024, 5:01 PM

#

uint64_t get_free_pages() {
  pnode *cur = &head;
  uint64_t page = 0;
  while (cur != NULL) {
    page += 1;
    cur = (pnode *)cur->next;
  };
  return page;
}

#

it 0xd'd here

#

CR2=8948fffffb58858b

#

that shouldnt happen how tf am i making a weird ah address

#

like wtf?

#

nooo nooo

kind root Oct 26, 2024, 5:03 PM

#

what is head

surreal path Oct 26, 2024, 5:03 PM

#

0x0 initally

#

which is what i set it to

#

NULL

#

oh wait

#

wait

#

wait

#

wait

#

wait

safe spireBOT Oct 26, 2024, 5:04 PM

#

Anti-spam Message

Possible spam detected for user: linuxmaster2.0. Please contact a moderator to be unmuted.

finite summit Oct 26, 2024, 5:04 PM

#

.!unmute @surreal path

safe spireBOT Oct 26, 2024, 5:04 PM

#

Command executed by oberrow

Command Error

Type

Permission Error

Details

You must have MODERATE_MEMBERS to execute this command.

kind root Oct 26, 2024, 5:04 PM

#

.!unmute @surreal path

safe spireBOT Oct 26, 2024, 5:04 PM

#

@surreal path is not muted.

finite summit Oct 26, 2024, 5:04 PM

#

kind root .!unmute <@787341689333350450>

qwinci beat you to it

#

@surreal path you are unmuted now

surreal path Oct 26, 2024, 5:05 PM

#

okay

#

same general protection fault

#

nooo nooo nooo

#

i changed the code aroudn a bit

kind root Oct 26, 2024, 5:08 PM

#

same skill issue

surreal path Oct 26, 2024, 5:08 PM

#

bro how

#

nooo

kind root Oct 26, 2024, 5:08 PM

#

do u have warnings enabled

surreal path Oct 26, 2024, 5:08 PM

#

yes

#

-Werror

kind root Oct 26, 2024, 5:08 PM

#

thats not warnigns enabled

surreal path Oct 26, 2024, 5:08 PM

#

oh

#

how to enable them

kind root Oct 26, 2024, 5:08 PM

#

-Wall -Wextra -Wpedantic

surreal path Oct 26, 2024, 5:08 PM

#

okay

#

#

i only got this

#

and this

#

so um

#

yea

#

tf going on

#

nooo

#

these how i compile

kind root Oct 26, 2024, 5:11 PM

#

still

#

probably treating some uninitialized memory as a pointer

surreal path Oct 26, 2024, 5:11 PM

#

kind root probably treating some uninitialized memory as a pointer

noooo ?

#

look

#

sec

surreal path Oct 26, 2024, 5:11 PM

#

kind root probably treating some uninitialized memory as a pointer

wait

#

u mean

#

memzero right @kind root

kind root Oct 26, 2024, 5:13 PM

#

i dont mean anything particular

surreal path Oct 26, 2024, 5:17 PM

#

kind root i dont mean anything particular

u were right

#

look

#

!!!!

#

#

i forgot to memset by stuff

#

so happy :)

#

i love C

finite summit Oct 26, 2024, 5:18 PM

#

https://tenor.com/view/c-c-my-beloved-azedev-programming-c-my-beloved-gif-15333289406142266880

Tenor

#

https://tenor.com/view/fan-discord-gif-26557005

Tenor

surreal path Oct 26, 2024, 5:19 PM

#

C :)

upbeat turret Oct 26, 2024, 5:23 PM

#

spice yarrow Also you cropped off all the useful info like the error code

mb

surreal path Oct 26, 2024, 5:23 PM

#

pmm done

upbeat turret Oct 26, 2024, 5:33 PM

#

upbeat turret mb

this good?

surreal path Oct 26, 2024, 5:38 PM

#

bro got 0xb

#

😭

thorn bramble Oct 26, 2024, 5:39 PM

#

0xb?

surreal path Oct 26, 2024, 5:39 PM

#

yea 😭

thorn bramble Oct 26, 2024, 5:40 PM

#

where???

surreal path Oct 26, 2024, 5:40 PM

#

thorn bramble Oct 26, 2024, 5:41 PM

#

oh

#

i was looking at

#

v=08

surreal path Oct 26, 2024, 5:48 PM

#

thorn bramble v=08

imagine

flat nymph Oct 26, 2024, 6:42 PM

#

surreal path yet the compiler still threw some ub in uacpi or smthin idfk

Just have uACPI in userspace meme

surreal path Oct 26, 2024, 6:43 PM

#

flat nymph Just have uACPI in userspace <:meme:575445094589661215>

lol

flat nymph Oct 26, 2024, 6:46 PM

#

Why did you need ACPI at that stage in first place?

surreal path Oct 26, 2024, 6:46 PM

#

flat nymph Why did you need ACPI at that stage in first place?

dw

#

heap stress test

#

:)

kind root Oct 26, 2024, 6:53 PM

#

flat nymph Just have uACPI in userspace <:meme:575445094589661215>

very few people know u in uACPI means userspace

#

meme

flat nymph Oct 26, 2024, 6:54 PM

#

I mean the interface is basically POSIX

kind root Oct 26, 2024, 6:54 PM

#

close

flat nymph Oct 26, 2024, 6:54 PM

#

It was easy to plug in

crystal scarab Oct 26, 2024, 6:54 PM

#

kind root very few people know u in uACPI means userspace

i always thought the u meant universal

kind root Oct 26, 2024, 6:55 PM

#

its either ultra or micro

#

but everyone has their own meaning for the u

cinder plinth Oct 26, 2024, 6:55 PM

#

in my head canon it's μ

kind root Oct 26, 2024, 7:11 PM

#

to do what

unkempt relic Oct 26, 2024, 7:13 PM

#

kind root its either ultra or micro

i thought it was UltraOS

#

like how LAI was the LUX ACPI implementation

kind root Oct 26, 2024, 7:15 PM

#

unkempt relic like how LAI was the LUX ACPI implementation

didnt know that

unkempt relic Oct 26, 2024, 7:16 PM

#

kind root didnt know that

i thought you were an osdev old server veteran

kind root Oct 26, 2024, 7:16 PM

#

i was there probably before lux was a thing

#

hmm, good question, i dont remember

#

i think i was curious about acpi and how complex it seemed

unkempt relic Oct 26, 2024, 7:19 PM

#

kind root i was there probably before lux was a thing

2019 was when omarx exited the community it seems

kind root Oct 26, 2024, 7:19 PM

#

hmm

elder shoal Oct 26, 2024, 7:27 PM

#

kind root but everyone has their own meaning for the u

it just means u

#

there is no special meaning

#

troll

kind root Oct 26, 2024, 7:30 PM

#

who knows maybe thats true

finite summit Oct 26, 2024, 7:31 PM

#

kind root its either ultra or micro

It's ultra

cinder plinth Oct 26, 2024, 7:31 PM

#

u = you

#

uACPI is human

glossy walrus Oct 26, 2024, 10:48 PM

#

that's just called depression

surreal path Oct 27, 2024, 2:13 AM

#

working on slab allocator

#

should be easy

finite summit Oct 27, 2024, 2:48 AM

#

C ub is only a problem when the compiler leverages it unexpectedly for optimizations

#

imo

#

ahem ahem

#

strict aliasing

#

ahem ahem

surreal path Oct 27, 2024, 2:50 AM

#

slab allocator done

#

:)

#

no UB

#

https://images-ext-1.discordapp.net/external/a3392X4tpxUv5qN2VLQ935mpLmSjVTnekyzmoAWXme8/https/media.tenor.com/1MrJecNevgAAAAPo/c-c-my-beloved.mp4

▶ Play video

finite summit Oct 27, 2024, 2:51 AM

#

waow

#

👏 👏 👏

surreal path Oct 27, 2024, 2:51 AM

#

thank

#

:)

finite summit Oct 27, 2024, 2:51 AM

#

joys of using C

surreal path Oct 27, 2024, 2:51 AM

#

indeed

surreal path Oct 27, 2024, 5:15 AM

#

@thorn bramble i just realized i found a bug within old nyauxkt when reading the code about the page tables

#

i solved them in my C version now

#

im doing a similar implementation cause i think it was good

#

i might use a linked list for the vmm regions this time

#

but um

#

yea

#

bug found

#

bug squashed

#

yippie

#

also btw

#

you shouldnt do this check when mapping a 2mb page

#

right?

#

cause like

#

that would NEVER happen

#

am i correct

#

cause its only on like page level 2 (0, 1, 2, 3)

#

that happens

hollow herald Oct 27, 2024, 5:23 AM

#

surreal path <@796462670995390524> i just realized i found a bug within old nyauxkt when read...

! is not bitwise not in C

surreal path Oct 27, 2024, 5:24 AM

#

hollow herald `!` is not bitwise not in C

sorry ~

#

thanks for catching that

hollow goblet Oct 27, 2024, 1:44 PM

#

it may have happened to me

surreal path Oct 27, 2024, 5:14 PM

#

hollow goblet it may have happened to me

congrats!!!

hollow goblet Oct 27, 2024, 5:14 PM

#

:3

#

thx

surreal path Oct 27, 2024, 8:28 PM

#

i nap some more then we work on nyaux, i have page table shit mostly done i just need to write a function to setup a kernel pagemap and switch to such pagemap

#

then its vmm regions

#

then uacpi to test the heap

#

then scheduling straight away

surreal path Oct 28, 2024, 1:44 AM

#

bro what nooo

#

tf are u on about

#

does anyone know how to fix this???

#

this shouldnt be happening?

finite summit Oct 28, 2024, 1:51 AM

#

surreal path bro what <:nooo:750466870993223750>

just make an asm stub

#

or rather

surreal path Oct 28, 2024, 1:51 AM

#

finite summit just make an asm stub

for ONE instruction

#

😭

finite summit Oct 28, 2024, 1:51 AM

#

a function to invlpg in a separate asm file

finite summit Oct 28, 2024, 1:52 AM

#

surreal path for ONE instruction

what I have is just a bunch of instructions in one file

#

I have out* and in* for port manip

#

getCR*

#

getEFER

#

getDR6 (idk what happened with the other debug regs

#

cpuid

#

*msr

#

pause

#

invlpg

#

wbinvd

#

xsave

#

cli

#

sti

#

hlt

surreal path Oct 28, 2024, 1:53 AM

#

ugh fine

surreal path Oct 28, 2024, 2:13 AM

#

check_exception old: 0xffffffff new 0xe
   133: v=0e e=0010 i=0 cpl=0 IP=0028:ffffffff8000e1c7 pc=ffffffff8000e1c7 SP=0030:ffff800006683f98 CR2=ffffffff8000e1c7
RIP=ffffffff8000e1c7 RFL=00000082 [--S----] CPL=0 II=0 A20=1 SMM=0 HLT=0
CR0=80010011 CR2=ffffffff8000e1c7 CR3=0000000000008000 CR4=00000020
check_exception old: 0xe new 0xe
   134: v=08 e=0000 i=0 cpl=0 IP=0028:ffffffff8000e1c7 pc=ffffffff8000e1c7 SP=0030:ffff800006683f98 env->regs[R_EAX]=000000000000000d
RIP=ffffffff8000e1c7 RFL=00000082 [--S----] CPL=0 II=0 A20=1 SMM=0 HLT=0
CR0=80010011 CR2=ffffffff800e9ba0 CR3=0000000000008000 CR4=00000020
check_exception old: 0x8 new 0xe

fun

#

when switching to the new pagemap

#

this happens

#

!!!!!

#

void map(uint64_t *pt, uint64_t phys, uint64_t virt, uint64_t flags) {
  uint64_t *f = find_pte_and_allocate(pt, virt);
  *f = phys | flags;
}
uint64_t *find_pte_and_allocate(uint64_t *pt, uint64_t virt) {
  uint64_t shift = 48;
  for (int i = 0; i < 4; i++) {
    shift -= 9;
    uint64_t idx = (virt >> shift) & 0x1ff;
    uint64_t *page_table =
        (uint64_t *)((uint64_t)pt + hhdm_request.response->offset);
    if (i == 3) {
      return (uint64_t *)((uint64_t)page_table + idx);
    }
    if (page_table[idx] & ~PRESENT) {
      uint64_t *guy =
          (uint64_t *)((uint64_t)pmm_alloc() - hhdm_request.response->offset);
      page_table[idx] = (uint64_t)guy | PRESENT | RWALLOWED;
      pt = guy;
    } else if (page_table[idx] & PAGE2MB) {
      uint64_t *guy =
          (uint64_t *)((uint64_t)pmm_alloc() - hhdm_request.response->offset);
      uint64_t old_phys = page_table[idx] & 0x000ffffffffff000;
      uint64_t old_flags = page_table[idx] & ~0x000ffffffffff000;
      for (int j = 0; j < 512; j++) {
        guy[j] = (old_phys + j * 4096) | (old_flags & ~PAGE2MB);
      }
    } else {
      pt = (uint64_t *)(page_table[idx] & 0x000ffffffffff000);
    }
  }
  return 0;
}

#

i dont see an issue

#

or the issue*

#

nooo

finite summit Oct 28, 2024, 2:15 AM

#

fun

surreal path Oct 28, 2024, 2:15 AM

#

uint64_t *find_pte_and_allocate2mb(uint64_t *pt, uint64_t virt) {
  uint64_t shift = 48;
  for (int i = 0; i < 4; i++) {
    shift -= 9;
    uint64_t idx = (virt >> shift) & 0x1ff;
    uint64_t *page_table =
        (uint64_t *)((uint64_t)pt + hhdm_request.response->offset);
    if (i == 2) {
      return (uint64_t *)((uint64_t)page_table + idx);
    }
    if (page_table[idx] & ~PRESENT) {
      uint64_t *guy =
          (uint64_t *)((uint64_t)pmm_alloc() - hhdm_request.response->offset);
      page_table[idx] = (uint64_t)guy | PRESENT | RWALLOWED;
      pt = guy;
    } else {
      pt = (uint64_t *)(page_table[idx] & 0x000ffffffffff000);
    }
  }
  return 0;
}

finite summit Oct 28, 2024, 2:15 AM

#

gl

surreal path Oct 28, 2024, 2:15 AM

#

nooo nooo

cinder plinth Oct 28, 2024, 2:16 AM

#

surreal path bro what <:nooo:750466870993223750>

the problem is ()

surreal path Oct 28, 2024, 2:16 AM

#

()?

cinder plinth Oct 28, 2024, 2:16 AM

#

you're dereferencing an address

#

yes

surreal path Oct 28, 2024, 2:16 AM

#

( )?

cinder plinth Oct 28, 2024, 2:16 AM

#

yes

#

around %0

spice yarrow Oct 28, 2024, 2:16 AM

#

att syntax is weird

cinder plinth Oct 28, 2024, 2:16 AM

#

surreal path ```C void map(uint64_t *pt, uint64_t phys, uint64_t virt, uint64_t flags) { ui...

this code is fucking ugly

surreal path Oct 28, 2024, 2:17 AM

#

ik

surreal path Oct 28, 2024, 2:17 AM

#

cinder plinth the problem is ()

arent u supposed to

#

oberrow does it

surreal path Oct 28, 2024, 2:17 AM

#

cinder plinth this code is fucking ugly

ALSO

#

i dont know whats wrong with this 😔

cinder plinth Oct 28, 2024, 2:17 AM

#

surreal path

that seems wrong

surreal path Oct 28, 2024, 2:17 AM

#

@finite summit

#

bug for u

finite summit Oct 28, 2024, 2:18 AM

#

hai

spice yarrow Oct 28, 2024, 2:18 AM

#

surreal path

What's the definition of it in the header file?

finite summit Oct 28, 2024, 2:18 AM

#

no skill issue for you

surreal path Oct 28, 2024, 2:18 AM

#

spice yarrow What's the definition of it in the header file?

no its an asm file

cinder plinth Oct 28, 2024, 2:18 AM

#

its fine if oberrow does something like

surreal path Oct 28, 2024, 2:18 AM

#

compiled in nasm

cinder plinth Oct 28, 2024, 2:18 AM

#

&addr

finite summit Oct 28, 2024, 2:18 AM

#

huh

surreal path Oct 28, 2024, 2:18 AM

#

huh

cinder plinth Oct 28, 2024, 2:18 AM

#

invplg [rdi] dereferences what is at rdi

spice yarrow Oct 28, 2024, 2:18 AM

#

surreal path no its an asm file

You need a header file so that c knows how to call the asm

surreal path Oct 28, 2024, 2:18 AM

#

spice yarrow You need a header file so that c knows how to call the asm

yea i extern it

finite summit Oct 28, 2024, 2:18 AM

#

cinder plinth invplg [rdi] dereferences what is at rdi

wrong

cinder plinth Oct 28, 2024, 2:18 AM

#

no

surreal path Oct 28, 2024, 2:19 AM

#

surreal path yea i extern it

ofc i do

finite summit Oct 28, 2024, 2:19 AM

#

invlpg rdi```

cinder plinth Oct 28, 2024, 2:19 AM

#

invplg [rdi] is invlpg *rdi in C terms

finite summit Oct 28, 2024, 2:19 AM

#

is invalid

#

error: invalid combination of opcode and operands

#

and I directly pass the address of the page to invlpg

cinder plinth Oct 28, 2024, 2:19 AM

#

https://www.felixcloutier.com/x86/invlpg

finite summit Oct 28, 2024, 2:20 AM

#

containing m

finite summit Oct 28, 2024, 2:20 AM

#

cinder plinth https://www.felixcloutier.com/x86/invlpg

also nasm literally rejects invlpg rdi

surreal path Oct 28, 2024, 2:21 AM

#

this can be a work around for u oberrow

cinder plinth Oct 28, 2024, 2:21 AM

#

finite summit also nasm literally rejects `invlpg rdi`

idk it doesnt really make sense

finite summit Oct 28, 2024, 2:21 AM

#

bro wtf do you mean workaround

#

it's not broken

finite summit Oct 28, 2024, 2:21 AM

#

surreal path this can be a work around for u oberrow

&virt is wrong

#

or rather

#

not what you intended

surreal path Oct 28, 2024, 2:22 AM

#

ok i see

cinder plinth Oct 28, 2024, 2:22 AM

#

[rdi] means "invlpg *rdi"

surreal path Oct 28, 2024, 2:22 AM

#

i know

finite summit Oct 28, 2024, 2:22 AM

#

yes invlpg the address pointed to by rdi

cinder plinth Oct 28, 2024, 2:22 AM

#

so passing an address as rdi would read whatever is at that address

finite summit Oct 28, 2024, 2:22 AM

#

no

#

think of it like a mov [rdi], 0

#

it accesses the memory at rdi

#

kinda

#

but instead of moving anything in or out of it

#

it invalidates the relavant tlb entry

cinder plinth Oct 28, 2024, 2:23 AM

#

Invalidates any translation lookaside buffer (TLB) entries specified with the source operand. The source operand is a memory address. The processor determines the page that contains that address and flushes all TLB entries for that page.

#

the source operand is a memory address

spice yarrow Oct 28, 2024, 2:24 AM

#

It's a memory address, not an integer

cinder plinth Oct 28, 2024, 2:24 AM

#

same thing

finite summit Oct 28, 2024, 2:24 AM

#

let me open up the sdm quickly

cinder plinth Oct 28, 2024, 2:24 AM

#

this is the sdm

finite summit Oct 28, 2024, 2:24 AM

#

yes I know

#

I am looking for a different section

spice yarrow Oct 28, 2024, 2:25 AM

#

It's the same as lea

surreal path Oct 28, 2024, 2:25 AM

#

im too depressed to debug this rn i am feeling sucidal thoughts rn so im just ognna try to distract myself or smthin for a bit

cinder plinth Oct 28, 2024, 2:26 AM

#

but the at&t code does seem fine tho 🤔 I really wonder if that's just a syntax quirk of assemblers

cinder plinth Oct 28, 2024, 2:26 AM

#

surreal path im too depressed to debug this rn i am feeling sucidal thoughts rn so im just og...

hope you get better soon

finite summit Oct 28, 2024, 2:26 AM

#

they would assemble to the same thing

cinder plinth Oct 28, 2024, 2:27 AM

#

yeah I'm just saying maybe assemblers have some kind of check for invlpg

finite summit Oct 28, 2024, 2:27 AM

#

no at&t syntax

#

nvm

#

that was for immediates

#

I was gonna say it auto derefs

cinder plinth Oct 28, 2024, 2:27 AM

#

if (%0) is correct then presumably [rdi] is too

#

so that's weird

spice yarrow Oct 28, 2024, 2:28 AM

#

Wait so what does your invlpg do?

finite summit Oct 28, 2024, 2:28 AM

#

invlpg:
    invlpg [rdi]
    ret```

#

@cinder plinth it seems like it takes an effective address

#

which is why you wouldn't just give it rdi

cinder plinth Oct 28, 2024, 2:29 AM

#

yeah so I chalk it up as weird syntax

#

[rdi] typically means take whatever is there

finite summit Oct 28, 2024, 2:29 AM

#

well I guess it's kinda like lea

#

whatever

#

I think we can all agree that we have better stuff to be doing than arguing about invlpg

#

at least I do

spice yarrow Oct 28, 2024, 2:30 AM

#

Just reset cr3

finite summit Oct 28, 2024, 2:30 AM

#

genius

spice yarrow Oct 28, 2024, 2:30 AM

#

Best invlpg

finite summit Oct 28, 2024, 2:30 AM

#

nah actually

#

just go into _S3

#

each time

surreal path Oct 28, 2024, 2:33 AM

#

i asked for help on the osdev fourms cause i do not wanna debug this rn, im gonna go play some val then nap again ig

spice yarrow Oct 28, 2024, 2:37 AM

#

surreal path i asked for help on the osdev fourms cause i do not wanna debug this rn, im gonn...

Have you pushed this to GitHub?

surreal path Oct 28, 2024, 2:37 AM

#

spice yarrow Have you pushed this to GitHub?

yea

surreal path Oct 28, 2024, 2:37 AM

#

spice yarrow Have you pushed this to GitHub?

https://github.com/rayanmargham/NyauxKC/blob/master/kernel/src/mem/vmm.c

GitHub

NyauxKC/kernel/src/mem/vmm.c at master · rayanmargham/NyauxKC

Contribute to rayanmargham/NyauxKC development by creating an account on GitHub.

spice yarrow Oct 28, 2024, 2:37 AM

#

Okay I'll take a look

surreal path Oct 28, 2024, 2:38 AM

#

thank you

cinder plinth Oct 28, 2024, 2:38 AM

#

didnt you already have working code for this

surreal path Oct 28, 2024, 2:39 AM

#

cinder plinth didnt you already have working code for this

yea

cinder plinth Oct 28, 2024, 2:40 AM

#

cant you just copy it

#

if it doesnt work then it's likely the pmm's fault

surreal path Oct 28, 2024, 2:40 AM

#

its rust

#

and i copied most of the implementation

cinder plinth Oct 28, 2024, 2:41 AM

#

you had old nyaux before tho

spice yarrow Oct 28, 2024, 2:41 AM

#

You can call rust from c

cinder plinth Oct 28, 2024, 2:41 AM

#

don't

surreal path Oct 28, 2024, 2:41 AM

#

cinder plinth you had old nyaux before tho

thats disgusting code i dont wanna touch

#

please

#

for the love of god

spice yarrow Oct 28, 2024, 2:42 AM

#

spice yarrow You can call rust from c

You can... it's not a good idea but you can

#

But actually... what do you see when you use gdb pt dump?

surreal path Oct 28, 2024, 2:42 AM

#

i told u i really dont wanna debug this rn

#

im not feeling good at all

#

im just trying to distract myself with val

#

just take a look at it if u want to, if u dont ill try tmrw if im in the mood

spice yarrow Oct 28, 2024, 2:43 AM

#

Alright

#

I really hope you feel better

surreal path Oct 28, 2024, 8:48 AM

#

i woke up

#

thats the address

#

thats valid memory

#

so thats not the issue

edgy pilot Oct 28, 2024, 8:49 AM

#

what is this

page_table[idx] & ~PRESENT

#

!(page_table[idx] & PRESENT)

#

@surreal path

surreal path Oct 28, 2024, 9:01 AM

#

wdym

edgy pilot Oct 28, 2024, 9:01 AM

#

wdym wdim

surreal path Oct 28, 2024, 9:02 AM

#

i invert present

edgy pilot Oct 28, 2024, 9:02 AM

#

edgy pilot ``` !(page_table[idx] & PRESENT) ```

this is how it's done

edgy pilot Oct 28, 2024, 9:02 AM

#

surreal path i invert present

yes and why?

surreal path Oct 28, 2024, 9:02 AM

#

to check if it doesnt have present

#

whats wrong with that

edgy pilot Oct 28, 2024, 9:02 AM

#

that's not how it works

surreal path Oct 28, 2024, 9:02 AM

#

wdym

edgy pilot Oct 28, 2024, 9:02 AM

#

surreal path to check if it doesnt have present

you check if it has everything other than present

#

~0b00110010 = 0b11001101

surreal path Oct 28, 2024, 9:03 AM

#

oh

edgy pilot Oct 28, 2024, 9:04 AM

#

inb4 nyaux is fixed

surreal path Oct 28, 2024, 9:04 AM

#

nope

#

lol

#

error code is 0x18

#

0xe exception

#

@edgy pilot

edgy pilot Oct 28, 2024, 9:05 AM

#

push the code

surreal path Oct 28, 2024, 9:05 AM

#

kk

#

pushed

#

https://github.com/rayanmargham/NyauxKC/blob/master/kernel/src/mem/vmm.c

GitHub

NyauxKC/kernel/src/mem/vmm.c at master · rayanmargham/NyauxKC

Contribute to rayanmargham/NyauxKC development by creating an account on GitHub.

#

@edgy pilot

#

take a look

edgy pilot Oct 28, 2024, 9:07 AM

#

2097152

#

why

#

use hex

surreal path Oct 28, 2024, 9:07 AM

#

ehhh it dont matter

edgy pilot Oct 28, 2024, 9:08 AM

#

why do you have so many find_pte functions

surreal path Oct 28, 2024, 9:09 AM

#

idk man

#

ill fix that later just

#

idk

edgy pilot Oct 28, 2024, 9:09 AM

#

make it into a single one

surreal path Oct 28, 2024, 9:09 AM

#

later

edgy pilot Oct 28, 2024, 9:09 AM

#

duplicated code introduces multiple failure points

#

later often means "never"

surreal path Oct 28, 2024, 9:09 AM

#

no i will

#

pin that

surreal path Oct 28, 2024, 9:10 AM

#

edgy pilot make it into a single one

pin that

edgy pilot Oct 28, 2024, 9:10 AM

#

I can't

surreal path Oct 28, 2024, 9:10 AM

#

any mods here

#

lurking

#

or smthin

#

guessss not

edgy pilot Oct 28, 2024, 9:11 AM

#

surreal path ill fix that later just

auto pagemap::getlvl(entry &entry, bool allocate) -> table *
{
    table *ret = nullptr;
    if (!entry.getflags(valid_table_flags))
    {
        if (allocate == false)
            return nullptr;

        entry.clear();
        entry.setaddr(reinterpret_cast<std::uintptr_t>(ret = new_table()));
        entry.setflags(new_table_flags, true);
    }
    else ret = reinterpret_cast<table *>(entry.getaddr());

    return lib::tohh(ret);
}
// ...
auto pagemap::getpte(std::uintptr_t vaddr, page_size psize, bool allocate) -> std::expected<std::reference_wrapper<entry>, error>
{
    static constexpr std::uintptr_t bits = 0b111111111;
    static constexpr std::size_t levels = 4; // We don't need more than 256 tib of addresses
    static constexpr std::size_t shift_start = 12 + (levels - 1) * 9;

    auto pml = lib::tohh(_table);

    auto retidx = levels - static_cast<std::size_t>(psize) - 1;
    auto shift = shift_start;

    for (std::size_t i = 0; i < levels; i++)
    {
        auto &entry = pml->entries[(vaddr >> shift) & bits];

        if (i == retidx)
            return std::ref(entry);

        pml = getlvl(entry, allocate);
        if (pml == nullptr)
            return std::unexpected { error::invalid_entry };

        shift -= 9;
    }
    std::unreachable();
}

psize is 0=4kib 1=2mib or 2=1gib

surreal path Oct 28, 2024, 9:11 AM

#

this is c++ code

edgy pilot Oct 28, 2024, 9:12 AM

#

yes
use it as a reference

surreal path Oct 28, 2024, 9:12 AM

#

also arent u supposed to be finding this weird ah bug 😭

surreal path Oct 28, 2024, 9:13 AM

#

surreal path also arent u supposed to be finding this weird ah bug 😭

^^

edgy pilot Oct 28, 2024, 9:13 AM

#

I'm supposed to be writing an essay

#

not finding a weird bug

surreal path Oct 28, 2024, 9:14 AM

#

oh

#

sorry

#

im sorry man

#

im sorry

#

im rlly sorry

#

im sorry

#

im sorry

edgy pilot Oct 28, 2024, 9:14 AM

#

one sorry was enough :p

surreal path Oct 28, 2024, 9:14 AM

#

still im sorry

median goblet Oct 28, 2024, 9:48 AM

#

edgy pilot ``` !(page_table[idx] & PRESENT) ```

or even better

#

~page_table[idx] & PRESENT

edgy pilot Oct 28, 2024, 9:48 AM

#

lol

#

yeah that works too

#

I have never seen that used though

molten grotto Oct 28, 2024, 9:50 AM

#

its also pretty unreadable when compared to the alternative

surreal path Oct 28, 2024, 6:00 PM

#

idk why but i still feel actually rlly fucking awful

#

nooo

#

maybe its the bad sleep idfk

#

maybe ill feel better if i sleep some more

tender gorge Oct 28, 2024, 6:59 PM

#

surreal path idk why but i still feel actually rlly fucking awful

itll be alright hug

tender gorge Oct 28, 2024, 7:01 PM

#

surreal path maybe ill feel better if i sleep some more

sleeping too much isnt good either

glossy walrus Oct 28, 2024, 7:01 PM

#

surreal path maybe ill feel better if i sleep some more

that's not how that works

tender gorge Oct 28, 2024, 7:01 PM

#

sleeping is not really a good coping mechanism

median goblet Oct 28, 2024, 7:57 PM

#

edgy pilot I have never seen that used though

i saw one guy do it and learned it from him

median goblet Oct 28, 2024, 7:57 PM

#

molten grotto its also pretty unreadable when compared to the alternative

not really?

#

i feel like !(a & b) is more cumbersome than ~a & b

edgy pilot Oct 28, 2024, 7:57 PM

#

it looks a bit like -

median goblet Oct 28, 2024, 7:57 PM

#

maybe

molten grotto Oct 28, 2024, 7:58 PM

#

at least to me ~a & b takes more time because I have to actually think about what it does

edgy pilot Oct 28, 2024, 7:58 PM

#

^

median goblet Oct 28, 2024, 7:58 PM

#

btw i see a lot less mods, and those former mods having the secret private club access

#

just an observation

#

has there been a reform of mods lately?

edgy pilot Oct 28, 2024, 7:58 PM

#

no minors

median goblet Oct 28, 2024, 7:58 PM

#

like, reform of the mod system

median goblet Oct 28, 2024, 7:59 PM

#

edgy pilot no minors

so there were a lot of mods who were minors before

molten grotto Oct 28, 2024, 7:59 PM

#

well not that much

#

couple afaik

median goblet Oct 28, 2024, 7:59 PM

#

i didn't know czapek/iretq was a minor

frigid cliff Oct 28, 2024, 8:05 PM

#

tender gorge sleeping is not really a good coping mechanism

I mean at least for me sleeping has to be the least problematic coping mechanism

tender gorge Oct 28, 2024, 8:05 PM

#

sleeping as a coping mechanism can spiral you into a even deeper depression

#

source: me

frigid cliff Oct 28, 2024, 8:08 PM

#

I mean that’s kinda what makes it bad

tender gorge Oct 28, 2024, 8:08 PM

#

i mean yeah

#

must admit i dont really have any other suggestions for coping mechanisms that are good on account of not having had any when i was depressed

frigid cliff Oct 28, 2024, 8:09 PM

#

Yeah I don’t have any either 😓

tender gorge Oct 28, 2024, 8:10 PM

#

as much as im past it now, hindsight says i probably should have seen a psychiatrist lol

#

but yeah, the only real advice i can give is that it will be alright, and that recovery is not linear

frigid cliff Oct 28, 2024, 8:10 PM

#

That is generally the advice but speaking to someone can be really hard

tender gorge Oct 28, 2024, 8:11 PM

#

yeah, but its also what one should do, as hard as it is

#

i mean im saying that while having never actually done that to anyone irl, but yea

frigid cliff Oct 28, 2024, 8:12 PM

#

Because we know what the correct thing is, but the correct thing is sometimes very hard to first come to the realization that’s it’s the correct thing and then to actually do that

#

Idk I’m just saying shit I have been up for over 24 hours

tender gorge Oct 28, 2024, 8:13 PM

#

frigid cliff Idk I’m just saying shit I have been up for over 24 hours

ok you need to go to sleep

#

lol

tender gorge Oct 28, 2024, 8:14 PM

#

frigid cliff Because we know what the *correct* thing is, but the correct thing is sometimes ...

if i can decipher that correctly then yeah, agreed

#

(i think youre trying to say that we know what the correct thing to do is, but that its sometimes very hard to come to the realization that that is the correct thing to do)

frigid cliff Oct 28, 2024, 8:14 PM

#

Yesss

thorn bramble Oct 28, 2024, 8:37 PM

#

median goblet i didn't know czapek/iretq was a minor

no im not lol, i just didnt want to be mod and that was made true with the secret club reform™️

surreal path Oct 29, 2024, 1:20 AM

#

glossy walrus that's not how that works

yea ur right i still feel like shit

#

nooo

#

oh yeah it came to me in a dream what i was doing wrong with the page tables

#

i was doing return (uint64_t *)((uint64_t)page_table + idx)

#

i should be doing

#

return page_table + idx

finite summit Oct 29, 2024, 1:23 AM

#

it's a sign!

surreal path Oct 29, 2024, 1:23 AM

#

indeed !!!

#

it fixed the issue

finite summit Oct 29, 2024, 1:23 AM

#

and it's damn right

surreal path Oct 29, 2024, 1:23 AM

#

so :)

surreal path Oct 29, 2024, 1:24 AM

#

surreal path i was doing return (uint64_t *)((uint64_t)page_table + idx)

the reason this is wrong is im adding the idx to the address and this is without taking into account the fact that idx needs to be multiplied by 8 bytes because every entry is 8 bytes. doing page_table + idx; C does ((uint64_t)pagetable + (idx * 8)) basically

#

anyways page tables done now

#

#

i need to do vmm regions but i feel shit rn so idfk what to do

tender gorge Oct 29, 2024, 2:35 AM

#

surreal path i need to do vmm regions but i feel shit rn so idfk what to do

Take a break, have a drink, watch some YouTube, play some games maybe?

surreal path Oct 29, 2024, 2:35 AM

#

tender gorge Take a break, have a drink, watch some YouTube, play some games maybe?

okii

tender gorge Oct 29, 2024, 2:35 AM

#

surreal path yea ur right i still feel like shit

Not surprised

#

I'd recommend talking with your psychiatrist about your sleeping problems

#

If you haven't already

rigid fable Oct 29, 2024, 2:36 AM

#

tender gorge Take a break, have a drink, watch some YouTube, play some games maybe?

opposite

surreal path Oct 29, 2024, 2:36 AM

#

i did

rigid fable Oct 29, 2024, 2:36 AM

#

do no use screens

surreal path Oct 29, 2024, 2:36 AM

#

he gave me sleep tablets

#

they dont work

#

anymore

#

used to

#

not anymore

rigid fable Oct 29, 2024, 2:36 AM

#

how much time do you spend on screens?

tender gorge Oct 29, 2024, 2:36 AM

#

surreal path used to

Maybe talk to them again then

surreal path Oct 29, 2024, 2:36 AM

#

tender gorge Maybe talk to them again then

i did they said they cannot give anything else

#

caus eim under 18

tender gorge Oct 29, 2024, 2:37 AM

#

It's normal that meds sometimes stop working and need changing/dose adjustments

surreal path Oct 29, 2024, 2:37 AM

#

ill try to ask him again

tender gorge Oct 29, 2024, 2:37 AM

#

surreal path i did they said they cannot give anything else

Ah

tender gorge Oct 29, 2024, 2:40 AM

#

rigid fable do no use screens

I mean yeah, but I'm gonna assume nyaux is currently in a psych ward based on what happened a few days ago, and if my knowledge of them is correct that's not really much of a option

rigid fable Oct 29, 2024, 2:40 AM

#

oh he like broke a leg or?

#

iirc

tender gorge Oct 29, 2024, 2:41 AM

#

rigid fable oh he like broke a leg or?

If you want to know, scroll up a few days, other than that, it's for nyaux to share, not me

surreal path Oct 29, 2024, 2:45 AM

#

tender gorge I mean yeah, but I'm gonna assume nyaux is currently in a psych ward based on wh...

no im not

#

but if i do anything

#

i will go to one

#

and they'll take away my stuff

#

which means

#

no more nyaux

#

ever

#

nooo

#

so i cannot try to hurt myself

#

again

#

or

#

yk

tender gorge Oct 29, 2024, 2:46 AM

#

surreal path no more nyaux

If that's your worry, I'd talk to the psychiatrist again

surreal path Oct 29, 2024, 2:46 AM

#

tender gorge If that's your worry, I'd talk to the psychiatrist again

oki

tender gorge Oct 29, 2024, 2:46 AM

#

Since they might be able to organise you working on your job/hobbies from within hospital

#

Source: a person I knew who was in a UK psych ward

surreal path Oct 29, 2024, 2:52 AM

#

i will ask

tender gorge Oct 29, 2024, 2:52 AM

#

Alright, sounds good

#

It'll be alright 🫂

surreal path Oct 29, 2024, 4:43 AM

#

got vmm regions working

#

#

this test works

#

page faults at memset

#

because it is

#

unmapped!!!

surreal path Oct 29, 2024, 5:09 AM

#

this doesnt cause a page fault

#

is this bad

#

im sure its fine

#

time to integrate uacpi

#

ummmm

#

idk what this means ???

#

wha

#

nvm

#

i got that working

#

time to impl kernel_api.h

#

nooo

surreal path Oct 29, 2024, 6:00 AM

#

welp

#

thats fun

#

and it fails using my memset so

#

probs memory corruption

#

yippie !

#

nyaux hell 2015i0925890125901285th time

rigid fable Oct 29, 2024, 6:54 AM

#

more colors on logs is what we need

#

obos and astral will be no more

surreal path Oct 29, 2024, 7:11 AM

#

surreal path nyaux hell 2015i0925890125901285th time

cant figure this out im going sleep

finite summit Oct 29, 2024, 12:17 PM

#

rigid fable obos and astral will be no more

All my logs have colors

#

Jokes on you

edgy pilot Oct 29, 2024, 2:58 PM

#

every line should have a different random rgb value

#

and panic when you're out of colours

rigid fable Oct 29, 2024, 3:16 PM

#

edgy pilot and panic when you're out of colours

the panic should also use the colors

edgy pilot Oct 29, 2024, 3:17 PM

#

but every letter is of a random one

edgy pilot Oct 29, 2024, 3:18 PM

#

edgy pilot every line should have a different random rgb value

how would you do this efficiently

rigid fable Oct 29, 2024, 3:18 PM

#

the random function?

edgy pilot Oct 29, 2024, 3:18 PM

#

yes

#

but every time it's different

#

for 255*255*255 calls

rigid fable Oct 29, 2024, 3:19 PM

#

I recall you can use the timer with a specific algorithm. idk if that actually works but there are probably better methods

edgy pilot Oct 29, 2024, 3:19 PM

#

any algorithm that guarantees that?

elder shoal Oct 29, 2024, 3:19 PM

#

Define a bitmap 255255255 big troll

edgy pilot Oct 29, 2024, 3:20 PM

#

based

rigid fable Oct 29, 2024, 3:20 PM

#

edgy pilot any algorithm that guarantees that?

well a true rng would return different values each time, and it is unlikely to get the same value (like using a call for each color channel each time)

#

but what is likely is to get similar colors

#

iirc ToaruOS had a screen with some random colored chars, maybe I am wrong but whatever

rigid fable Oct 29, 2024, 3:22 PM

#

rigid fable well a true rng would return different values each time, and it is unlikely to g...

And I specify "true rng" because some unknown company did

int rand() {
        return 4;
}``` for an key encryption algorithm

edgy pilot Oct 29, 2024, 3:26 PM

#

I believe that was on playstation 2

rigid fable Oct 29, 2024, 3:27 PM

#

edgy pilot I believe that was on playstation 2

ps3

#

ECDSA

edgy pilot Oct 29, 2024, 3:27 PM

#

ah

rigid fable Oct 29, 2024, 3:27 PM

#

reason why you can't cfw a super slim

elder shoal Oct 29, 2024, 3:27 PM

#

rigid fable And I specify "true rng" because some unknown company did ```c int rand() { ...

Chosen by fair dice roll

rigid fable Oct 29, 2024, 3:27 PM

#

elder shoal Chosen by fair dice roll

Again, I have this on my gallery smh

edgy pilot Oct 29, 2024, 3:27 PM

#

there's xkcd about that

rigid fable Oct 29, 2024, 3:28 PM

#

#

Literally from the like 10 photos I have on my phone

edgy pilot Oct 29, 2024, 3:28 PM

#

so it wasn't 4

edgy pilot Oct 29, 2024, 3:29 PM

#

rigid fable Literally from the like 10 photos I have on my phone

I have more than 7 gib of memes

#

but most of that storage is taken up by videos

rigid fable Oct 29, 2024, 3:31 PM

#

edgy pilot so it wasn't 4

I actually have 4 memes on my phone 💀

#

depending on what you'd consider meme

edgy pilot Oct 29, 2024, 3:34 PM

#

everything from memes channels or Reddit

#

and is funny

cinder plinth Oct 29, 2024, 5:01 PM

#

edgy pilot every line should have a different random rgb value

I do that but not really

#

I assign a random color based on pid for every task

edgy pilot Oct 29, 2024, 5:03 PM

#

randomise pids on every context switch

cinder plinth Oct 29, 2024, 5:05 PM

#

Openbsd is scared of this man

surreal path Oct 29, 2024, 5:41 PM

#

hi chat need more sleep

#

eta 1am my time i wake up

#

(irish time)

edgy pilot Oct 29, 2024, 8:28 PM

#

to be fair I don't think it's a bad idea if you use a good hash map implementation

#

has better lookup times than a rb tree

cinder plinth Oct 29, 2024, 8:28 PM

#

a good hashmap implementation is implemented using a rbtree or avl

#

I'd just use an array

#

which is instant

#

the problem is sparse memory areas

edgy pilot Oct 29, 2024, 8:29 PM

#

then you need to resize that array

cinder plinth Oct 29, 2024, 8:29 PM

#

yeah so

edgy pilot Oct 29, 2024, 8:30 PM

#

how would you map addresses to index

#

that's still a mini hash map lol

cinder plinth Oct 29, 2024, 8:31 PM

#

no

#

you just do

#

page / page size

#

or rather

#

offset / page size

edgy pilot Oct 29, 2024, 8:32 PM

#

that becomes very inefficient if address isn't small

#

please correct me if i'm wrong

cinder plinth Oct 29, 2024, 8:34 PM

#

I'm not sure what you mean

#

you keep an array of pages per anonymous mapping

#

the indices are based on the offset into that mapping

#

and for sparse areas, you can mitigate that using chunking

edgy pilot Oct 29, 2024, 8:36 PM

#

wasn't the previous conversation about mapping when page faulting

#

and you would need to look up the data structure with the faulting address

cinder plinth Oct 29, 2024, 8:37 PM

#

so

#

you keep an rbtree or something of mappings

#

then in that mapping you can find a single page using the array

#

that's like

#

O(log n) + O(1) amortized

edgy pilot Oct 29, 2024, 8:38 PM

#

cinder plinth you keep an rbtree or something of mappings

I'm talking about that first part

#

lookup would be faster with a proper hashmap

#

wouldn't it?

cinder plinth Oct 29, 2024, 8:39 PM

#

well

#

trees have better worst-case performance

#

and resizing a hashtable is very inefficient

#

and self-balanced trees are sorted too

edgy pilot Oct 29, 2024, 8:40 PM

#

you said good hashmaps are implemented using a rbtree or an avl

#

so resizing it wouldn't be so bad

cinder plinth Oct 29, 2024, 8:41 PM

#

pretty sure you still need to rehash everything

edgy pilot Oct 29, 2024, 8:41 PM

#

not always

#

only when a new hash conflicts with some other

#

iirc

cinder plinth Oct 29, 2024, 8:43 PM

#

you can also lookup ranges better

#

in a tree

#

since it's ordered

#

and less overhead probs

edgy pilot Oct 29, 2024, 8:43 PM

#

when would you need to lookup ranges though

#

in this case

cinder plinth Oct 29, 2024, 8:44 PM

#

unmap

#

let's say you have a mapping from [0, 0x1000) and another one [0x1000, 0x2000)

#

now you unmap 0 with size 0x2000

#

also I think trees are preferred because worst case hashtables kinda suck while performance can be more easily predicted in a tree

#

yeah they're O(n) worst case

#

rbtree is always O(log n)

#

anyway I'm not a data structure expert but that's probably the biggest reason

surreal path Oct 29, 2024, 8:47 PM

#

hi i woke up (early this time !!!)

#

anyways

#

im gonna play one val comp game

#

then uhhh figure out whats going on lol

edgy pilot Oct 29, 2024, 8:48 PM

#

cinder plinth anyway I'm not a data structure expert but that's probably the biggest reason

me neither lol
I'm just trying to find the best way to do this

#

(in the future)

cinder plinth Oct 29, 2024, 8:49 PM

#

but yeah I think tldr hashmaps are less predictable than trees

edgy pilot Oct 29, 2024, 8:49 PM

#

abseil::btree_map looks interesting

cinder plinth Oct 29, 2024, 8:51 PM

#

std::map is tree-based

edgy pilot Oct 29, 2024, 8:51 PM

#

stl has pretty inefficient implementations though

#

as it needs to work with everything

cinder plinth Oct 29, 2024, 8:52 PM

#

yeah

#

but you dont have access to abseil do you

#

is it even freestanding

edgy pilot Oct 29, 2024, 8:53 PM

#

I have this https://github.com/greg7mdp/parallel-hashmap

#

it has abseil btee structures ported

#

basically the same thing

surreal path Oct 29, 2024, 8:53 PM

#

chat what are we talking abt :3

edgy pilot Oct 29, 2024, 8:53 PM

#

I was already using flat hash map in my "old" kernel

edgy pilot Oct 29, 2024, 8:54 PM

#

surreal path chat what are we talking abt :3

data structures

surreal path Oct 29, 2024, 8:54 PM

#

also im losing so bad in this val game

#

nooo

surreal path Oct 29, 2024, 8:54 PM

#

edgy pilot data structures

fair

#

tbh

#

hashmap >= anything else

#

fight me

#

😤

#

guys multidimensional hashmap

#

NEED

cinder plinth Oct 29, 2024, 8:54 PM

#

edgy pilot I was already using flat hash map in my "old" kernel

I plan on using the rbtree+array mechanism I described earlier

#

might scale better

edgy pilot Oct 29, 2024, 8:55 PM

#

edgy pilot I was already using flat hash map in my "old" kernel

not for this use case

cinder plinth Oct 29, 2024, 8:55 PM

#

then again, how many mappings are there in a process

surreal path Oct 29, 2024, 8:55 PM

#

guys

#

what do yall think of multidimensional hashmaps

edgy pilot Oct 29, 2024, 8:55 PM

#

cinder plinth then again, how many mappings are there in a process

not much

cinder plinth Oct 29, 2024, 8:55 PM

#

surreal path what do yall think of multidimensional hashmaps

what does that mean

#

Hashmap<Hashmap<>, Hashmap<>>

edgy pilot Oct 29, 2024, 8:55 PM

#

surreal path guys

multi maps?

cinder plinth Oct 29, 2024, 8:55 PM

#

meme

surreal path Oct 29, 2024, 8:55 PM

#

cinder plinth what does that mean

Hashmap<Hashmap<usize>, Hashmap<usize>>

#

meme

edgy pilot Oct 29, 2024, 8:56 PM

#

why?

#

just std::pair<val1, val2> as a key

cinder plinth Oct 29, 2024, 8:56 PM

#

edgy pilot not much

would you just hash a pointer to a struct describing the mapping?

#

then that's overhead too

#

because this not intrusive like rbtree

edgy pilot Oct 29, 2024, 8:56 PM

#

cinder plinth would you just hash a pointer to a struct describing the mapping?

if the structure is large then maybe

#

but I don't think it'll be that large

cinder plinth Oct 29, 2024, 8:56 PM

#

I don't think you can reasonably uniquely hash a structure like that

edgy pilot Oct 29, 2024, 8:57 PM

#

just have a good hashing function meme

surreal path Oct 29, 2024, 8:57 PM

#

edgy pilot why?

CAUSE ITS FUNNY

#

AND ALSO

#

YEA

cinder plinth Oct 29, 2024, 8:57 PM

#

wait no

#

you can

#

nvm

#

I'm stupid

#

you hash using the address

#

not the data

edgy pilot Oct 29, 2024, 8:57 PM

#

aren't we all

#

yeah

cinder plinth Oct 29, 2024, 8:57 PM

#

then yeah maybe it'd work but

#

the main two issues are overhead and predictability/scalability

surreal path Oct 29, 2024, 8:58 PM

#

btw i just said in a val game

#

WE ARE VALORANT

#

i made the entire team cringe

edgy pilot Oct 29, 2024, 8:58 PM

#

with btree_map or a hashmap

#

which one are you talking about

cinder plinth Oct 29, 2024, 8:58 PM

#

hashmap

edgy pilot Oct 29, 2024, 8:58 PM

#

yes

cinder plinth Oct 29, 2024, 8:58 PM

#

best-case I think a hashmap would be pretty good tho

surreal path Oct 29, 2024, 8:58 PM

#

surreal path i made the entire team cringe

what is uropion on this information

edgy pilot Oct 29, 2024, 8:59 PM

#

cinder plinth best-case I think a hashmap would be pretty good tho

yes but the need for a sorted structure you brought up makes me thing the btree one is better

#

might use that

cinder plinth Oct 29, 2024, 8:59 PM

#

yes

#

then again, how common is that

#

idk

edgy pilot Oct 29, 2024, 9:00 PM

#

one issue is that btree moves items a lot so the structure needs to not be very large

#

what else would you store in it:

paddr
vaddr
offset
length
prot/flags

#

not that much

#

there are probably way more knowledgeable people here

cinder plinth Oct 29, 2024, 9:03 PM

#

why offset

edgy pilot Oct 29, 2024, 9:04 PM

#

wait what was I thinking about

#

I forgor why

#

nvm that

cinder plinth Oct 29, 2024, 9:04 PM

#

"and hash tables (which are not kept sorted to be easily traversed in order, and must be tuned for a specific size and hash function where rbtrees scale gracefully storing arbitrary keys)."

#

according to linux

#

https://www.kernel.org/doc/html/v5.9/core-api/rbtree.html

edgy pilot Oct 29, 2024, 9:06 PM

#

cinder plinth why offset

ah offset if a file is mmaped

#

and I froggot length

#

40 bytes max

edgy pilot Oct 29, 2024, 9:12 PM

#

edgy pilot ah offset if a file is mmaped

offset could be stored in the paddr field
and you would use a single byte to differentiate between them

#

saving 7 bytes

surreal path Oct 29, 2024, 11:47 PM

#

maybe its my slab allocator?

#

wait nvm

#

idfk

#

tried watching it

#

got nothing rlly

#

cause

#

the address changes everytime

#

what it faults with in cr2

#

nooo

#

after memsetting with my slabs the cr2 is now consistent

#

okay interesting

#

theres no base that exists with this address

#

so um

#

wtf

edgy pilot Oct 30, 2024, 12:07 AM

#

what's at rip

surreal path Oct 30, 2024, 12:09 AM

#

memset

#

my memset function which came with the limine c template

edgy pilot Oct 30, 2024, 12:09 AM

#

what is it memsetting

surreal path Oct 30, 2024, 12:09 AM

#

i dont know

#

im trying to figure that out rn

edgy pilot Oct 30, 2024, 12:09 AM

#

printf

surreal path Oct 30, 2024, 12:09 AM

#

dude

#

its uacpi doing itr

#

not me

edgy pilot Oct 30, 2024, 12:09 AM

#

you are doing it

surreal path Oct 30, 2024, 12:10 AM

#

no im not?

edgy pilot Oct 30, 2024, 12:10 AM

#

and uacpi is failing because of that

surreal path Oct 30, 2024, 12:10 AM

#

uacpi is using MY memset

#

i printf'd all the addresses

#

being allocated and deallocated

#

nothing found

edgy pilot Oct 30, 2024, 12:10 AM

#

then you are allocating or mapping incorrectly

#

what's the error code

#

in binary

thorn bramble Oct 30, 2024, 12:11 AM

#

maybe it is uacpi failing, but it's your fault it's doing that

surreal path Oct 30, 2024, 12:11 AM

#

0x2

surreal path Oct 30, 2024, 12:11 AM

#

thorn bramble maybe it is uacpi failing, but it's your fault it's doing that

i know

thorn bramble Oct 30, 2024, 12:11 AM

#

you are clearly not upholding something

edgy pilot Oct 30, 2024, 12:11 AM

#

surreal path 0x2

0b10

thorn bramble Oct 30, 2024, 12:11 AM

#

so yeah, print!

edgy pilot Oct 30, 2024, 12:11 AM

#

edgy pilot 0b10

lookup error code on osdev

thorn bramble Oct 30, 2024, 12:11 AM

#

make sure your malloc is up to spec

edgy pilot Oct 30, 2024, 12:11 AM

#

see what it means

surreal path Oct 30, 2024, 12:11 AM

#

i know what that means

#

it means non present write

edgy pilot Oct 30, 2024, 12:12 AM

#

check your mapper

surreal path Oct 30, 2024, 12:12 AM

#

thorn bramble so yeah, print!

print what else? i printed all the addresses coming in and out

edgy pilot Oct 30, 2024, 12:12 AM

#

and combine those functions

surreal path Oct 30, 2024, 12:12 AM

#

im not combining them

#

that isnt gonna solve the issue

edgy pilot Oct 30, 2024, 12:12 AM

#

combine them

surreal path Oct 30, 2024, 12:12 AM

#

no

edgy pilot Oct 30, 2024, 12:12 AM

#

please

surreal path Oct 30, 2024, 12:12 AM

#

bru

edgy pilot Oct 30, 2024, 12:12 AM

#

or I'll be sad

surreal path Oct 30, 2024, 12:12 AM

#

maybe later

#

promise

#

not rn tho

#

i wanna get rid of this bug

edgy pilot Oct 30, 2024, 12:13 AM

#

nooo

surreal path Oct 30, 2024, 12:13 AM

#

#

this should be correct

#

bru

edgy pilot Oct 30, 2024, 12:13 AM

#

what is that

surreal path Oct 30, 2024, 12:13 AM

#

part of my vmm region allocation

#

wait one sec i have an idea

edgy pilot Oct 30, 2024, 12:14 AM

#

good night

surreal path Oct 30, 2024, 12:14 AM

#

taking away both free functions

#

causes a different

#

result

#

fucking hell

#

nooo

#

uint64_t *find_pte_and_allocate(uint64_t *pt, uint64_t virt) {
  uint64_t shift = 48;
  for (int i = 0; i < 4; i++) {
    shift -= 9;
    uint64_t idx = (virt >> shift) & 0x1ff;
    uint64_t *page_table =
        (uint64_t *)((uint64_t)pt + hhdm_request.response->offset);
    if (i == 3) {
      return page_table + idx;
    }
    if (!(page_table[idx] & PRESENT)) {
      uint64_t *guy =
          (uint64_t *)((uint64_t)pmm_alloc() - hhdm_request.response->offset);
      page_table[idx] = (uint64_t)guy | PRESENT | RWALLOWED;
      pt = guy;
    } else if (page_table[idx] & PAGE2MB) {
      uint64_t *guy = (uint64_t *)((uint64_t)pmm_alloc());
      uint64_t old_phys = page_table[idx] & 0x000ffffffffff000;
      uint64_t old_flags = page_table[idx] & ~0x000ffffffffff000;
      for (int j = 0; j < 512; j++) {
        guy[j] = (old_phys + j * 4096) | (old_flags & ~PAGE2MB);
      }
      pt = (uint64_t *)((uint64_t)guy - hhdm_request.response->offset);
    } else {
      pt = (uint64_t *)(page_table[idx] & 0x000ffffffffff000);
    }
  }
  return 0;
}

my mapping function

#

void map(uint64_t *pt, uint64_t phys, uint64_t virt, uint64_t flags) {
  uint64_t *f = find_pte_and_allocate(pt, virt);
  *f = phys | flags;
}

#

void kvmm_region_dealloc(void *addr) {
  if (addr == NULL) {
    return;
  }
  VMMRegion *cur = (VMMRegion *)ker_map.head;
  while (cur != NULL) {
    if (cur->base == (uint64_t)addr) {
      memset((void *)cur->base, 0, cur->length);
      uint64_t amount_to_allocateinpages = cur->length / 4096;
      for (uint64_t i = 0; i != amount_to_allocateinpages; i++) {
        uint64_t phys = unmap(ker_map.pml4, cur->base + (i * 4096));
        pmm_dealloc((void *)(phys + hhdm_request.response->offset));
      }
      slabfree(cur);
      return;
    } else {
      cur = (VMMRegion *)cur->next;
      continue;
    }
  }
}

#

region deallocator

#

void *kvmm_region_alloc(uint64_t amount, uint64_t flags) {
  assert(ker_map.head != NULL);
  assert(ker_map.pml4 != NULL);
  VMMRegion *cur = (VMMRegion *)ker_map.head;
  VMMRegion *prev = NULL;
  while (cur != NULL) {
    if (prev == NULL) {
      prev = cur;
      cur = (VMMRegion *)cur->next;
      continue;
    }
    if ((cur->base - (prev->base + prev->length)) >=
        align_up(amount, 4096) + 0x1000) {
      VMMRegion *new =
          create_region((prev->base + prev->length), align_up(amount, 4096));
      prev->next = (struct VMMRegion *)new;
      new->next = (struct VMMRegion *)cur;
      uint64_t amount_to_allocateinpages = new->length / 4096;
      for (uint64_t i = 0; i != amount_to_allocateinpages; i++) {
        void *page = pmm_alloc();
        map(ker_map.pml4, (uint64_t)page, new->base + (i * 4096), flags);
      }
      memset((void *)new->base, 0, new->length);
      return (void *)new->base;
    } else {
      prev = cur;
      cur = (VMMRegion *)cur->next;
      continue;
    }
  };
  kprintf("No free Regions, Too Much Memory being used!!!\n");
  panic("Sir madamm this should never occur");
  return NULL;
}

region allcocator

#

void init_cache(cache *mod, uint64_t size) {
  mod->size = size;
  mod->slabs = (struct slab *)init_slab(size);
}
void *slab_alloc(cache *mod) {
  spinlock_lock(&pmmlock);
  if (mod->slabs == NULL || mod->size == 0) {
    return NULL;
  }
  slab *cur = (slab *)mod->slabs;
  while (true) {
    if (cur->freelist == NULL) {
      if (cur->next != NULL) {
        cur = (slab *)cur->next;
        continue;
      }
      break;
    }
    void *guy = cur->freelist;
    cur->freelist = ((pnode *)cur->freelist)->next;
    memset(guy, 0, mod->size);
    spinlock_unlock(&pmmlock);
    return guy;
  };
  cur->next = (struct slab *)init_slab(mod->size);
  cur = (slab *)cur->next;
  void *guy = cur->freelist;
  cur->freelist = ((pnode *)cur->freelist)->next;
  memset(guy, 0, mod->size);
  spinlock_unlock(&pmmlock);
  return guy;
}
void *slaballocate(uint64_t amount) {
  amount = next_pow2(amount);
  for (int i = 0; i != 7; i++) {
    cache *c = &caches[i];
    assert(c->size != 0);
    return slab_alloc(c);
  }
  return NULL;
}
void slabfree(void *addr) {
  spinlock_lock(&pmmlock);
  uint64_t real_addr = (uint64_t)addr;
  if (real_addr == 0) {
    spinlock_unlock(&pmmlock);
    return;
  }
  slab *guy = (slab *)(real_addr & ~0xFFF);
  memset(addr, 0, sizeof(pnode));
  pnode *node = (pnode *)addr;
  pnode *old = (pnode *)guy->freelist;
  node->next = (struct pnode *)old;
  guy->freelist = (struct pnode *)node;
  spinlock_unlock(&pmmlock);
}
slab *init_slab(uint64_t size) {
  void *page = pmm_alloc();
  uint64_t obj_amount = (4096 - sizeof(slab)) / size;
  slab *hdr = (slab *)page;
  pnode *first = (pnode *)((uint64_t)page + sizeof(slab));
  pnode *cur = first;
  hdr->freelist = (struct pnode *)first;
  for (uint64_t i = 1; i < obj_amount; i++) {
    pnode *new = (pnode *)((uint64_t)first + (i * size));
    new->next = (struct pnode *)cur;
    cur = new;
  };
  return hdr;
}

#

my slab allocator

#

https://github.com/rayanmargham/NyauxKC

GitHub

GitHub - rayanmargham/NyauxKC

Contribute to rayanmargham/NyauxKC development by creating an account on GitHub.

#

all on github

#

and btw

#

i cannot even figure out whats calling what because memset is being used thousands of times from uacpi

#

#

i tried back tracing i got nothinggggg

#

wtf

#

wtf is that

#

why is that

#

all garbage data

#

wtf

#

but C sees it fine

#

wtf

#

MORE BUGS??

#

of course because a shitty programmer like me cannot even fucking do something as simple as a memory allocator

#

i fuck everything, the stack, the heap

#

what next

#

nooo

#

i cannot even go up because the stack is corrupt???

#

wtf

#

fucking kill me

#

im gonna legit cry

#

doesnt matter what language i write in

#

i always make UB

spice yarrow Oct 30, 2024, 1:25 AM

#

surreal path Hashmap<Hashmap<usize>, Hashmap<usize>>

I love using hashmaps as the key to my hashmap

spice yarrow Oct 30, 2024, 1:29 AM

#

surreal path of course because a shitty programmer like me cannot even fucking do something a...

osdev is hard, struggling with it doesn't make you a bad programmer

finite summit Oct 30, 2024, 1:33 AM

#

surreal path theres no base that exists with this address

That looks like it's offsetting by the hhdm base

steady flume Oct 30, 2024, 1:42 AM

#

surreal path what is uropion on this information

i think that's pretty funny

median goblet Oct 30, 2024, 7:39 AM

#

thorn bramble no im not lol, i just didnt want to be mod and that was made true with the secre...

🗣️

surreal path Oct 31, 2024, 1:18 AM

#

hi i think i need a break for today

#

sorry

finite summit Oct 31, 2024, 1:19 AM

#

it's np

steady flume Oct 31, 2024, 1:49 AM

#

all good :)

tender gorge Oct 31, 2024, 2:28 AM

#

surreal path hi i think i need a break for today

All good, focus on yourself pls

surreal path Oct 31, 2024, 4:37 AM

#

btw i found something cause i was bored and decided to debug for like 15 minutes

...
ffff8000ffc00000: 00000000ffc00000 --P-----W
ffff8000ffe00000: 00000000ffe00000 --P-----W
ffff800100400000: 0003800000033000 X-------W    <------- Messed up
ffff800100401000: 0003800000036000 X-------W    <------- Messed up
ffffffff80000000: 0000000002125000 ----A---W
ffffffff80001000: 0000000002126000 ----A---W
ffffffff80002000: 0000000002127000 ----A---W
...

something is wrong with my mapping function

#

eh

#

anyways im gonna go back on my break for today

#

tmrw we figure this out

surreal path Oct 31, 2024, 10:36 PM

#

hi chat todays halloween and i still kinda want to spend this time with my family and go out trick or treating

#

i didnt realize yesterday that tmrw was halloween

#

promise tmrw ill get something done

#

im srry

tender gorge Oct 31, 2024, 10:40 PM

#

surreal path im srry

you do not have to be sorry for not working on your hobby

tender gorge Oct 31, 2024, 10:40 PM

#

surreal path hi chat todays halloween and i still kinda want to spend this time with my famil...

have fun!

steady flume Nov 1, 2024, 1:56 AM

#

are you feeling pressured to work on this project? if so, i don't think that's particularly healthly

spice yarrow Nov 1, 2024, 3:20 AM

#

It's not like GitHub is gonna fire you cause you didn't do enough work on Nyaux
Just do whatever you want to, if you decide to abandon this project that's totally fine

surreal path Nov 1, 2024, 9:32 PM

#

im not gonna abandon

#

im okay now

#

ill work on it!!!

#

rn im opening vscode

#

time to debug this shit

surreal path Nov 1, 2024, 9:33 PM

#

steady flume are you feeling pressured to work on this project? if so, i don't think that's p...

no not rlly

#

im just letting you guys know whats going on

#

okay so time for epic nyaux debugging

surreal path Nov 1, 2024, 11:26 PM

#

i found more information

#

it seems my page entries get messed up at levels 1 and 2

#

im unsure what is causing this still but

#

that is useful information to know

#

it seems like the upper 17 bits get all set?

#

tf?

#

lower bits seems fine

#

but

#

wtf

#

is going on

#

nooo

thorn bramble Nov 2, 2024, 1:47 AM

#

check_exception old: 0xffffffff new 0xe
     0: v=0e e=0008 i=0 cpl=0 IP=0028:ffffffff8000114e pc=ffffffff8000114e SP=0030:ffff800007f38fe0 CR2=ffff800007f38fe0```

#

the nyaux curse 😭😭😭

#

lol i love operator precedence! :)

#

apparently pte_flags |= PTE_P * (flags & MMU_PRESENT) != 0; isn't equivalent to pte_flags |= PTE_P * ((flags & MMU_PRESENT) != 0); =w=

#

it actually makes sense tho, now that i think about it

tender gorge Nov 2, 2024, 6:33 AM

#

surreal path it seems like the upper 17 bits get all set?

Sounds like some kind of signed integer bug somehow

surreal path Nov 2, 2024, 1:48 PM

#

tender gorge Sounds like some kind of signed integer bug somehow

i solved that issue, was with the hddm offset

#

but now a new issue

#

nooo

#

as usual

#

nyaux curse continues

#

Page Fault! CR2 0xffff800100c001d8
RIP is 0xffffffff80010666
NYAUX Panic! Reason: Page Fault:c

#

happens in uacpi

#

probs cause my allocator shit or smthin idfk

surreal path Nov 2, 2024, 1:49 PM

#

thorn bramble ``` check_exception old: 0xffffffff new 0xe 0: v=0e e=0008 i=0 cpl=0 IP=002...

uhm what da sigma

surreal path Nov 2, 2024, 1:49 PM

#

surreal path ```C Page Fault! CR2 0xffff800100c001d8 RIP is 0xffffffff80010666 NYAUX Panic! R...

error code is non present

#

happens here so def allocator bug

#

nooo

#

more bug as usual

#

the nyaux curse continues

surreal path Nov 2, 2024, 7:41 PM

#

welp at least we made progress today

surreal path Nov 3, 2024, 2:48 AM

#

man this is tiring

#

but maybe its find_pte

#

to be honest

#

AND ALSO

#

sorry for the very slow progress, just this bug is persisent

#

and honestly i dont wanna burn myself out at the same time so i kinda work on nyaux in bursts now

#

whenever i feel not horrible

#

i think its more sustainable

#

ima sleep

#

then tmrw ill look into find_pte and see if it is encountering a 2mb entry or smthin

#

maybe that be the issue who knows

surreal path Nov 3, 2024, 4:18 AM

#

it is not find_pte

#

im afraid

surreal path Nov 3, 2024, 4:55 AM

#

nooo nooo

steady flume Nov 3, 2024, 6:51 AM

#

poor nyaux

surreal path Nov 3, 2024, 8:55 AM

#

steady flume poor nyaux

fr fr

#

its because

#

uhhh

#

idk man

#

🤷‍♂️

#

i just get tortured by bug after bugi

#

nooo

#

yea i know

#

it is how i talk 😎

#

how u know 😱

#

also @fa#

#

@unkempt relic i see u lurking!!

#

say hiii

#

hiii

#

i am NOT used to this 60% keyboard 😭

#

i cannot type for shit with this

#

anyways more nyaux debugging after i take a nap!!

#

also i was debugging all night

#

i didnt find much

#

😔

#

all i know is that its just another page table bug of some kind i THINK

#

ill keep inspecting every entry and

#

figure it out

surreal path Nov 3, 2024, 9:00 AM

#

surreal path ```C Page Fault! CR2 0xffff800100c001d8 RIP is 0xffffffff80010666 NYAUX Panic! R...

whats interesting tho

#

is that the cr2 isnt some null deref

#

is it something with memory corruption cause memset?

#

no?

#

i memset all my pages and every piece of memory that goes in and out

#

maybe its with unmap

#

idk ima nap then try commenting out unmap

#

maybe thats ze issue

surreal path Nov 3, 2024, 7:00 PM

#

shit

#

i have school tmrw

#

ill debug this rn

#

i dont have much timne

#

cause schol

#

nooo

tender gorge Nov 3, 2024, 8:11 PM

#

but also not too much

edgy pilot Nov 3, 2024, 8:14 PM

#

I either sleep 4 hours or 12

surreal path Nov 3, 2024, 10:14 PM

#

me and oberrow joined forces to debug on it, a lot of bugs going on. we continue debugging after school !

#

obos x nyaux

#Nyaux