Hornet(rewrite, do you remember?)

[tardy mist]

just load programs

Im testing in a test env

[tardy mist]

• learn to not confuse big endian and little endian

[tardy mist]

Im modifying flanterm to support beeg fonts

it now kinda works, it at least knows what glyph to display

holy shit it actually worked

clearing is still a bit buggy

oh, cuz I didnt modify plot_char_fast

nice

now it works

with any font I can imagine

the lack of AA is kinda annoying tho

anyways

now the terminal looks good

some of the letters in this font kinda look wierd

also its a bit too large

so can you suggest some better ones?

as close to 16x16 resolution as possible

[tight summit]

why did you need to modify it though?

couldn't you have told it to use a font of a different size?

[tardy mist]

the code creating ctx->font_bool does not work for font sizes over 8, so I made it use ctx->font_bits instead

I was too lazy to delete font_bool so its floating there unused :/

why does font_bool even exist?

[hardy merlin]

for quicker decision making I guess

[wind mortar]

you still need to load it and compare it so i don’t see how it’s quicker

if anything it just uses more memory because you’ve expanded it to use a byte per bit, but idk how flanterm uses that so i might be speaking out of my ass here

[hardy merlin]

honestly I don't know why that is either

I'd point out my font drawing routines that use bit ops are slow... but it's NOT because they use bitops; it's because they're written in a way that doesn't make them performant

[wind mortar]

hand write them in assembly

[hardy merlin]

C is good enough, I'm not saying that

but it has to recalculate a bunch at once per pixel of font data

[wind mortar]

i’m jk

[hardy merlin]

like the address OF the pixel to write

I really should revamp that, but I'm working on boron right now

[tardy mist]

so the font_bool is garbage?

[hardy merlin]

I don't know

I guess wait for mintsuki to explain the reasoning

[tight summit]

because it's faster

practically tested to be faster

not saying that there couldn't be bitwise pieces of code that are as fast or faster

but this is faster than a naïve bitwise routine

don't ask me why, i don't know for sure

[proven horizon]

flanterm_fb_init has a parameter for font scaling

unless i'm retarded

[tardy mist]

since when is the letter ï in the English alphabet?

[proven horizon]

probably a foreign word adopted by english?

[barren sundial]

yeah lol

[proven horizon]

yeah flanterm does support font scaling in the init function

[barren sundial]

a lot of english is just things we yoinked

[tardy mist]

but, if the font_width varible is greater than 8, the routine generating font_bool will fuck things up, not allowing you to load the high res font

so I made it use font_bits instead of font_bool

[tight summit]

you could also fix font_bool to not fuck things up i guess

that sounds like a change welcome upstream

either that or figuring out a way to make bits as fast as bool

[tardy mist]

that is my plan cuz the other one is harder

I see, it expects a VGA font

then, I just have to make it take whatever bitmap

[tardy mist]

lessgooo

works

I have to change some things before contributing tho

[tardy mist]

so, the builtin font bugs out my code

cuz its exactly 8 pixels wide xD

I need to fix it so VGA fonts actually load

almost working

[tardy mist]

I lost all my progress :/

[proven horizon]

wut how

[tardy mist]

git and stupidity

[tardy mist]

here it is:https://github.com/mintsuki/flanterm/pull/5

[proven horizon]

mfw when git reset --hard

[tardy mist]

Its refactoring day

[tardy mist]

Im finished refactoring

[tight summit]

i am not sure i fully understand your PR to flanterm lol

[tardy mist]

what is your question? is the code quality questionable?

[tight summit]

i am not sure whether you take into account the fact that glyphs 0xc0-0xdf need to be extended right

for one

[tardy mist]

what characters are there? and isnt that only for VGA fonts?

[tight summit]

this is VGA fonts

[tardy mist]

Cuz I alredy broke the VGA specification by letting any resolution be loaded

[tight summit]

nah

ok so

forget about the word VGA

codepage 437, which is what we use

those glyphs need to be right extended

they are mostly box drawing characters and similar

and no, that's not conditional

it has to be done

[tardy mist]

oh, so they have no spacing inbetween them and their neighbours?

[tight summit]

it's not that they have no spacing as much as it is that the last right most column of the glyph is repeated inside what would be the spacing

this allows something like a straight line to be drawn using box characters without any gaps

[tardy mist]

oh

Ill fix that rq

[tardy mist]

so I have to duplicate on theese lines?(basically does this look right?)

WHAT THE HELL FREETYPE, YOU CANT DO THIS THING WITHOUT GAPS?

works like a charm now

@tight summit done with fixing it

[tight summit]

alright ty

[tardy mist]

any other questions?

[tight summit]

no just give me time pls

[tardy mist]

ok

[tardy mist]

I made the page fault handler kill the current process instead of kernel panicing

also just fucking realized that my scheduler gives the same stack to all kernel processes

it wasnt a problem since I only had one kernel procass and 0 user processes

but it now is since I want to load programs

[tardy mist]

it now works

kinda

idk what that PID 267866096 is

added special case for the init system malfunctioning

[tardy mist]

Im alredy fucking stuck

its like @fallow pilot said, -1 motivation

[proven horizon]

what's wrong?

[tardy mist]

pid 0 crashes if theres another process running parallel to it

*user process

https://github.com/lolguy91/h0r.net/blob/rewrite/src/core/sched/sched.c

[proven horizon]

how can it run parallel with the kernel when you don't have SMP wat

[tardy mist]

just scheduler stuff

[proven horizon]

in create_process you're creating a new process structure without defining stuff and immediately setting cr3 to an uninitialized value

[tardy mist]

not uninitialized

just the current cr3

[proven horizon]

newproc.cr3 can be anything at the point of executing the movq

[tardy mist]

no cuz this is limine

it sets up a page table

*uising limine

[proven horizon]

wait wait wait what's the syntax for gas

mov dest, src or mov src, dest ?

[tardy mist]

src dest

I learned it the hard way

[proven horizon]

fuuuck okay i'm retarded sorry

[tardy mist]

petition: add nasm syntax option into GCC

[proven horizon]

nah it's not that big of a deal really

[fallow pilot]

intel syntax should be supported, might be wrong though

[tardy mist]

anything else sus in the code?

[proven horizon]

can't tell just from looking at my phone

i'll take a look when I get home

[tardy mist]

ok ty

[tardy mist]

are you a home?

[fallow pilot]

I can help you

[tardy mist]

really?

ty

[fallow pilot]

yeah

[tardy mist]

vc?

[fallow pilot]

text sorry

[tardy mist]

https://tenor.com/view/god-damn-it-rouxls-kaard-rouxls-kaard-damnit-gif-23440867

[fallow pilot]

why

[tardy mist]

Im tired of typing and screenshotting

[fallow pilot]

its not that bad

[tardy mist]

also you feel busy cuz you dont reply without a ping

[fallow pilot]

i am in a match in a game rn

but what do you need help with?

[tardy mist]

fucking inexplickable scheduler behaviour

Imma install simplescreenrecorder and make a vid for you

[fallow pilot]

ok

[tardy mist]

done

frick

wrong monitro

wait

now good @fallow pilot

[tardy mist]

sched code is here

well fuck I have to go

so just leave your findings here

[proven horizon]

jesus i can't even compile it

you're apparently including a missing stdio.h

oh wait you're supplying binaries in the repo thank god

it's flanterm

exactly this line in plot_char():
uint32_t bg = c->bg == 0xffffffff ? canvas_line[gx] : c->bg;
(line 467 in backends/fb.c)
@tight summit

the only times where plot_char() is called are here: (line numbers included and all in backends/fb.c)

747:    plot_char(_ctx, &c, ctx->cursor_x, ctx->cursor_y);
769:            plot_char_fast(_ctx, old, &q->c, q->x, q->y);
771:            plot_char(_ctx, &q->c, q->x, q->y);
779:            plot_char(_ctx, &ctx->grid[ctx->old_cursor_x + ctx->old_cursor_y * _ctx->cols], ctx->old_cursor_x, ctx->old_cursor_y);
832:        plot_char(_ctx, &ctx->grid[i], x, y);

yeah it's crashing absolutely randomly so it's not flanterm (sorry for the ping mint :p)

so you're getting a bunch of logs about a GPF in the process with a same PID, before the kernel gives up and throws a page fault, which does make me believe you're not actually killing the process

my blind guess: when you kill a process you don't actually reset the current_PID, but regardless of the process' RIP, you're doing this

if (not_first_tick)
        memcpy(&processes.data[current_PID].status, regs, sizeof(Registers));

which might be why you're getting the page fault, and not resetting the current_PID might be why you're getting the GPF repeatedly

and of course I might be wrong, I'm hella sleepy and my brain isn't braining as much

Braining xD

[tardy mist]

fucking hell

yk what caused ALL this

?

I switched up cs and ss

that is FUCKING IT

[tardy mist]

FUCK

just realized that my PMM is full of bugs

so I have to fix them

[tardy mist]

also there is a memory leak somewhere in my kernel, probably my shit testing shell

and yes, -5 points for baking in a shell

context: #1170316206935388231 message

[fallow pilot]

Can you show me the code for the shell?

[proven horizon]

src/core/interface/desh.c

just looked, lolguy gets -10 points for using if-else and strcmp for builtin commands

and -5 points for not building the shell as a separate executable

[tardy mist]

memory leak fixed

what took me so long was that I retrieved the size wrong in the free function so it just froze

[tardy mist]

to be fair this will be replaced the moment I have an ELF loader

[proven horizon]

what, the if-else or building it into the kernel?

[tardy mist]

both

this is a dirty test enviroment

[proven horizon]

good

i might be able to port over the hashmap used in psh later today if you want

[tardy mist]

sure

hashmaps are cool

and I accept anyy help

[tardy mist]

SOMEONE HELP ME

I HAVE NO CLUE WTF IS HAPPENING

seriously

my OS hangs on a single instruction

this one:

[fallow pilot]

well does rax point to valid page tables?

[tardy mist]

yes

and there is no interrupt

and qemu flickers between paused and unpaused

the code that generates the page table: https://github.com/lolguy91/h0r.net/blob/8fd700b00fcc76177310a2def1f42b5e40c52563/src/core/sched/sched.c#L88

[tardy mist]

do you have any idea why this is?

[fallow pilot]

idk

[tardy mist]

btw why did you block me?

[fallow pilot]

i didnt

[fallow pilot]

I set my dm's to be friends only

[tardy mist]

because I asked for help?

[fallow pilot]

no it was not because of you

[proven horizon]

Ded project

[proven horizon]

certainly not

[tardy mist]

nope

[tardy mist]

I hacked together an ELF loader

its horrible

it assumes that .code is phdr 1

it also doesnt do anything with the other phdrs

but works for my test binary

and probably nothing else lmao

anyways, this is not usermode yet

but I am in the perfect situation to start working on the syscall interface

so, syscalls

btw dont look at the code

youll cringe hard

[tight summit]

2035

[tardy mist]

first syscall I implemented is a printing one!

and thats it for today

[wind mortar]

16:81

🤔

[proven horizon]

17/17/2035 16:81

[tardy mist]

seems like Im decoding time incorrectly

or I need special flags for qemu to give the correct date

[hollow leaf]

falut handler

[proven horizon]

looks more like this being the issue

[fallow pilot]

maybe you forgot to convert from BCD time

[fallow pilot]

https://en.wikipedia.org/wiki/Binary-coded_decimal The BIOS time is in BCD encoding

[regal gale]

1301th message

[woven tide]

1302nd message

[fallow pilot]

1304th message

[steady merlin]

1305th message

[old sapphire]

Chain prevention! ⛓️💥

Mission failed! We'll get em next time..

[fallow pilot]

1307th message

[steady merlin]

1308th message

[woven tide]

1309th message

[proven horizon]

"Reminder" "remainder" and "reindeer" all look the goddamn same

Also why rain deer?

[tardy mist]

Raain deer

[tardy mist]

I do not thingók this is usable anymore

now it has both a flex cable and an LCD panel issue

I litterally dropped a salt lamp on it lmao

[fallow pilot]

Oh No! You ran your OS on real hardware and this happened????

[tardy mist]

worst accident ever

[tardy mist]

no, I dropped a salt lamp on my laptop

[fallow pilot]

how the fuck does that happen

[tardy mist]

cuz I forgot I have weak arms

[fallow pilot]

oh

[tardy mist]

very weak arms

well, now I can't multitask as well

I need to get a HDMI monitor now

well

[tardy mist]

(HDMI cuz thats the only other video port, I alredy use the VGA)

[fallow pilot]

why do you already use VGA?

[tardy mist]

thats my second monitor

[tardy mist]

(pwease dont ban me for custom css usage)

[old sapphire]

ban for the windows 7 theme instead /j

[tardy mist]

ban for the usage of tone indicators

[old sapphire]

ban for the blurry photo

[tardy mist]

ban for being giygas

[proven horizon]

ban for not having an emoji font installed

[tardy mist]

ban for changing your name every month

[proven horizon]

true that

[proven horizon]

[tardy mist]

I have no motivation to work on anything

[proven horizon]

That is what depression looks like