and getpage
#Astral
robust geyser
i want to implement procfs but a) idk how procfs works in general b) idk what those functions are supposed to do
grand shadow
its for writing pages to/from the page cache
robust geyser
while b) is a problem a) is probably a bigger one 
robust geyser
okay so i can just ENOSYS them or something?
or just return 0?
i dont think these are needed for procfs
grand shadow
okay so i can just ENOSYS them or something?
all regular file I/O goes through it so uh
robust geyser
with content being generated on the fly
oh
well those aren't regular files
__assert(!"tmpfs_read is handled by the page cache");
hmm
grand shadow
I guess I could make pagecache support FS specific and have regular files for non supporting filesystems go through the normal route
hmm
grand shadow
lemme see
I could add a helper function in the vfs or vmmcache or whatever that says if the page cache would be available for a specific vnode and add those checks everywhere
like in vfs_read/write or in vmm_pagefault etc
robust geyser
not sure what the right choice would be 😅
i tried running startwm as user 🤔
works as root, hmm
grand shadow
weird, it should work as both
grand shadow
hell yeah
grand shadow
i tried running startwm as user 🤔
works on my machine
grand shadow
maybe it was just a bit slow
craggy tide
yep lol
it's really bad and i wasn't in the mood to fix that, but i was in the mood to have the build run flawlessly on aarch64, so i nuked it
grand shadow
I wonder if how I did it would make any difference
@robust geyser should be ok now
it'll go through the page cache if its either a block device or a regular file AND it exposes the VOP_GETPAGE and VOP_PUTPAGE vnode ops
so just have VOP_GETPAGE and VOP_PUTPAGE = NULL and you should be fine
also @prime juniper mr portmaster what do you think of the shit I pulled here
prime juniper
it's nasty sure
but have you seen this
(it's yaml but over the discord character limit)
grand shadow
what the fuck
prime juniper
or this
yeah I have some nasty shit
but chromium is a real piece of work
firefox is a bit better on the eyes recipe wise
grand shadow
5head build system
prime juniper
NSS is fucking nasty
prime juniper
prime juniper
seems like any packages that end with *SS have stupid build systems
contain *SS
openssl is also stupid as fuck
prime juniper
webkitgtk by a metric shitload of advantage
chromium compiled and ran a webpage once upon a time
might actually work now
but it bitrotted and doesn't build
the cross compilation process took me a full month of hard work (make a patch and watch it compile for 343943 hours, repeat until done and better not touch any gn files)
firefox didn't compile fully and I never debugged that
links beats webkit but that's text mode only
and I'm assuming proper browsers here
grand shadow
yeah links isnt really good
I need to get around to seeing how one ports webkitgtk to a new platform
prime juniper
oh it's fine
links in gui mode will even load pictures
I can and have read news on it
inside Managarm
grand shadow
true
robust geyser
webkitgtk probably depends on gtk, so get that first
prime juniper
well
yes
but
it's webkit
and the gtk port is webkitgtk
soooo
maybe there's a better port for you out there
or you write your own completely
prime juniper
GTK3 will do X for you
robust geyser
(half of linux gui apps use it)
prime juniper
managarm only keeps it around for gtklife (broken anyway) and hexchat
where hexchat is due to be replaced by konversation (Qt based)
grand shadow
idk if GTK4 still supports that
it'll only support wayland?
prime juniper
unsure
robust geyser
webkit/wpe is apparently a thing
prime juniper
but wouldn't be surprised
robust geyser
wpe webkit
robust geyser
its an official webkit thing too, from what i can read
prime juniper
and will use wpe if available and told to do so (which is recommended by BLFS)
yes
grand shadow
I need my pretty xscreensaver demos grrrrr
prime juniper
but I agree we need it
prime juniper
I need my pretty xscreensaver demos grrrrr
xclock needs to go brrrrrrrrrr
robust geyser
webkit wpe seems to want libwpe, which on arch depends on not much (namely libegl and libxkbcommon)
grand shadow
xedit also needs it I think
prime juniper
webkit wpe seems to want libwpe, which on arch depends on not much (namely libeg...
libwpe isn't bad no
robust geyser
might be a good choice if you want to avoid gtk/qt then
which i dont think you shouldl
prime juniper
- name: libwpe
metadata:
summary: Platform-agnostic interfaces for WPE WebKit
description: The libwpe package contains a general purpose library for WPE WebKit and the WPE Renderer.
spdx: 'BSD-2-Clause'
website: 'https://wpewebkit.org/'
maintainer: "Dennis Bonke <[email protected]>"
categories: ['gui-libs']
source:
subdir: 'ports'
git: 'https://github.com/WebPlatformForEmbedded/libwpe.git'
tag: '1.14.1'
version: '1.14.1'
tools_required:
- system-gcc
- host-pkg-config
- virtual: pkgconfig-for-target
triple: "@OPTION:arch-triple@"
pkgs_required:
- mlibc
- mesa
- libxkbcommon
revision: 1
configure:
- args:
- 'meson'
- 'setup'
- '--cross-file'
- '@SOURCE_ROOT@/scripts/meson-@OPTION:[email protected]'
- '--prefix=/usr'
- '--libdir=lib'
- '--buildtype=debugoptimized'
- '-Denable-xkb=true'
- '-Dbuild-docs=false'
- '@THIS_SOURCE_DIR@'
build:
- args: ['ninja']
- args: ['ninja', 'install']
environ:
DESTDIR: '@THIS_COLLECT_DIR@'
enjoy
you also want this
prime juniper
- name: wpebackend-fdo
architecture: '@OPTION:arch@'
metadata:
summary: WPE backend designed for Linux desktop systems
description: The wpebackend-fdo package contains the Freedesktop.org backend for WPE WebKit and the WPE renderer.
spdx: 'BSD-2-Clause'
website: 'https://wpewebkit.org/'
maintainer: "Dennis Bonke <[email protected]>"
categories: ['gui-libs']
source:
subdir: 'ports'
git: 'https://github.com/Igalia/WPEBackend-fdo.git'
tag: '1.14.2'
version: '1.14.2'
tools_required:
- system-gcc
- host-pkg-config
- virtual: pkgconfig-for-target
triple: "@OPTION:arch-triple@"
- wayland-scanner
pkgs_required:
- mlibc
- libwpe
- wayland-protocols
- wayland
- libepoxy
- glib
revision: 1
configure:
- args:
- 'meson'
- 'setup'
- '--cross-file'
- '@SOURCE_ROOT@/scripts/meson-@OPTION:[email protected]'
- '--prefix=/usr'
- '--libdir=lib'
- '--buildtype=debugoptimized'
- '-Dbuild_docs=false'
- '@THIS_SOURCE_DIR@'
build:
- args: ['ninja']
- args: ['ninja', 'install']
environ:
DESTDIR: '@THIS_COLLECT_DIR@'
idr if I upstreamed these two
grand shadow
```yml
- name: wpebackend-fdo
architecture: '@OPTION:arch@'
metadata:
...
wayland
prime juniper
lol
grand shadow
:(
prime juniper
yes wayland
prime juniper
oh BLFS webkitgtk no longer mentions wpe
does have an optional wayland dep tho
what
hard dep on libgudev?
gentoo disagrees
grand shadow
day 12031283219321 of telling myself to implement auth checking for kill()
robust geyser
i should do the astral vm thing i wanted to do
robust geyser
yeah yeah
i would just
have it print some byte sequence over e9
and the host would listen for that
and restart if needed
i would print that on panic
grand shadow
true
robust geyser
i could log everything to a separate file
each "session"
and if it panics i would save that file for analysis or something
since i would use qmp to interact with the vm i could even backtrace on panic
im just not too sure how to make spinning up multiple vms without anyone being able to fill up the disk space
i guess the root image would be small, each vm gets like 128M and thats the only storage they get
prime juniper
hmm
managarm can do that too if I fix ssh and get someone to give the netstack some server love
prime juniper
I'll keep that in mind for later
robust geyser
:D
grand shadow
managarm can do that too if I fix ssh and get someone to give the netstack some ...
does managarm have fd passing in unix sockets
prime juniper
I'm 99.9% sure we do
grand shadow
openssh needs that
prime juniper
you mean SCM_RIGHTS right?
SCM_RIGHTS
Send or receive a set of open file descriptors from another process. The data portion contains an integer array of the file descriptors.
Commonly, this operation is referred to as "passing a file descriptor" to another process. However, more accurately, what is being passed is a reference to an open file description (see open(2)),
and in the receiving process it is likely that a different file descriptor number will be used. Semantically, this operation is equivalent to duplicating (dup(2)) a file descriptor into the file
descriptor table of another process.
this shit?
grand shadow
yeah
prime juniper
if(!packet->files.empty()) {
if(ctrl.message(SOL_SOCKET, SCM_RIGHTS, sizeof(int) * packet->files.size())) {
for(auto &file : packet->files)
ctrl.write<int>(process->fileContext()->attachFile(std::move(file),
flags & MSG_CMSG_CLOEXEC));
}else{
throw std::runtime_error("posix: CMSG truncation is not implemented");
}
packet->files.clear();
}
yep
grand shadow
based
prime juniper
recvMsg will handle that in unix sockets
grand shadow
then you should have everything once netserver
prime juniper
welllll
if I do ssh dennis@<my server ip> I get some weird port parsing memes
sooo
prime juniper
I'll pull the error
grand shadow
so it might not be in mlibc
prime juniper
yeah mlibc is unlikely
oh even worse
unexpected setsockopt call
whoops
then this becomes an issue for later today
I wanna finish my bootstrap PR and go to sleep
wait
layer 0 number 1
grand shadow
ssh from managarm into astral and vice versa soon
prime juniper
number 1 is SO_DEBUG?
the fuck?
layer 0 is AF_UNSPEC?
the fuck?
ah it's also SOL_SOCKET
hmm
I'll debug this later
prime juniper
ssh from managarm into astral and vice versa soon
yes
that would be based as fuck
the Managarm side just needs to not forget to run dhcpcd first (and I did just forget that so I got ENETUNREACH whoops)
robust geyser
im gonna set up a thing to spin up astral vms in a bit
i should be able to spin up ~50 vms at any given time, memory wise
if i give each vm 128MB of memory
idk how well the cpu will handle that but
grand shadow
wtf
I mean
128 mb of memory should be more than enough for bash and stuff
iirc it boots into fvwm with that
robust geyser
i see, i might experiment a bit and give it less even
prime juniper
cries in Managarm footprint
robust geyser
"""micro""" kernel
grand shadow
how much does managarm need
robust geyser
should i just expose a shell over a web interface?
idk how it should work lol
astral shell as a service™️
except its not a service because its free, whatever
grand shadow
how would that even like work
wild marsh
how much does managarm need
iirc the lowest i tried and succeeded to get into weston was 700M
or thereabout
but that was a while ago
robust geyser
how would that even like work
idk, you visit a web page and get an astral shell
or you just ssh into something and that spins up an astral vm
wild marsh
websocket terminal on the client and a server that just bridges an ssh client to that
robust geyser
yeah, basically what i wanted to do
grand shadow
iirc the lowest i tried and succeeded to get into weston was 700M
westom moment
wb kmscon?
wild marsh
wb kmscon?
i think back when i tried we didn't have kmscon yet
grand shadow
or you just ssh into something and that spins up an astral vm
hmmmm there are some vps providers that do something kind of like that
robust geyser
but now that i think about it i could just create a simple ssh server that spins up a vm and then proxies to astral
wild marsh
hm i wonder if copy.sh/v86 gained x86_64 support
robust geyser
prob not
Some features are missing, in particular:
Task gates, far calls in protected mode
Some 16 bit protected mode features
Single stepping (trap flag, debug registers)
Some exceptions, especially floating point and SSE
Multicore
64-bit extensions
wild marsh
yeah i just went to the repo
wild marsh
wb kmscon?
just checked and it seems 482MB or so?, oom at 480MB
grand shadow
hmmmmm
wild marsh
481MB ooms after kmscon is launched but before bash is launched
i guess it might vary a bit from run to run due to scheduling differences
but i guess 512 ± 32MB
i still wonder what takes up the most memory
grand shadow
I mean managarm has a lot of stuff running
wild marsh
i still don't think that should add up to 512M
on the posix side it's 12 processes (incl the ls /proc process ig)
grand shadow
oh LORD
grand shadow
on the posix side it's 12 processes (incl the `ls /proc` process ig)
damn
do you know the usage for each?
the bare minimum for astral right now is 30 mb to drop into a bash shell
nothing else works
wild marsh
no, we don't do that much accounting in the posix server so /proc/<pid>/status just reports N/A for the mem stats
also servers (so drivers, posix-subsystem itself, etc) don't count as posix processes
grand shadow
I mean would the servers even use that much memory?
wild marsh
they shouldn't, most of them don't do anything, and after init-stage1 stuff's launched via udev rules
and init-stage1 only launches servers potentially needed to mount the rootfs
so all the various block device drivers, usb controller drivers, and the usb storage driver
grand shadow
hmmm
I guess you could maybe do some sort of accounting to figure it out if it becomes too big of an issue
wild marsh
yeah
robust geyser
honestly i dont think so
i kinda just want to see astral self host :^)
i mean i know you can build yourself
but not all the tools (yet)
and jinx is also a bit of a problem
but mint mentioned something something containerless mode
so you might be able to make it work? ™️
grand shadow
i kinda just want to see astral self host :^)
hmm I could work on fat next thing
grand shadow
but not all the tools (yet)
yeah gcc build is still funky
I should bump mlibc again and see if I can still not build gcc
or try with different cflags
prime juniper
<@796462670995390524> are there any non stupidly hard ports you wanna see
As in, userspace programs? Most of the big ticket items are already covered. I guess I never ported a webserver so try apache2?
prime juniper
Exactly
grand shadow
here we go again
robust geyser
oh i was gonna download it on astral
wild marsh
yeah
l mao cpp size_t VirtualOperations::getRss() { // Derived classes should track RSS; the generic implementaton does not. // TODO: As soon as all derived classes implement this, we should make it pure virtual. return 0; } (ofc no derived class implements getRss)
grand shadow
oh i was gonna download it on astral
youd need to patch and rerun the autotools which I dont have
grand shadow
l mao ```cpp
size_t VirtualOperations::getRss() {
// Derived classes should ...
TODO: do everything
grand shadow
you will never know if you dont try
I guess
@prime juniper were the perl issues on managarm mlibc related or managarm related
robust geyser
ok we building
damn it is slow
actually, idk, i would have to compare it to linux
grand shadow
also building here
inb4 it works for you
it segfaults like a few hours into the build
due to mlibc rbtree memes
prime juniper
<@261497806307459072> were the perl issues on managarm mlibc related or managarm...
Unknown
It’s just dogshit
prime juniper
l mao ```cpp
size_t VirtualOperations::getRss() {
// Derived classes should ...
Yikes
grand shadow
It’s just dogshit
what exactly happens
prime juniper
what exactly happens
It complains about some scalar refs bullshit while use strict is set
Lemme pull the error for you
re the perl issue
Can't use string ("Getopt::Long::VERSION") as a SCALAR ref while "strict refs" in use at /usr/lib/perl5/5.30.2/vars.pm line 35
The string doesn’t really matter, CPAN gives the same error on a different string
But it’s always scalar ref while strict refs is in use
And this is shipped Perl code , not some third party bs
Outside of this, probably some module finding meme from what I remember of autotools
grand shadow
hmmmm
prime juniper
Idr the error for autotools. But this scalar thing is pure kanker
grand shadow
crazy that python works just fine but perl doesnt
wild marsh
confirmed that python>perl :^)
grand shadow
Still building
I will leave it building overnight and hopefully coming back to it working pls
robust geyser
same here
craggy tide
@wild marsh https://github.com/osdev0/libstdcxx-headers
now we got proper freestanding C++ headers
robust geyser
rip
carmine swallow
now sdl audio works (kinda) except that it has a huge latency for some reason lol
one problem that should be done something about is that the oss header file is installed into /usr/lib/oss/include/sys/soundcard.h instead of /usr/include/sys/soundcard.h, it looks like on a glibc linux system glibc comes with that file so ig that might be why the oss package doesn't put it to /usr/include/sys/soundcard.h
grand shadow
now sdl audio works (kinda) except that it has a huge latency for some reason lo...
Holy shit it actually works?????
Thats amazing tbh I was expecting it to just segfault and die like it did before
But I think that was just some missing stuff in mlibc?
Also what do you think could be causing the delay?
grand shadow
one problem that should be done something about is that the oss header file is i...
And yeah Ill change that I think it is using the linux install script in it so
Unless you already did
carmine swallow
yeah no I didn't
carmine swallow
Also what do you think could be causing the delay?
I think its somehow related to the buffering/queueing of the audio, I changed the buffer size to be bigger and it sometimes is fine but then others it isn't (and also plays the same effect multiple times)
carmine swallow
yeah I don't think unless a thread can not get scheduled in a period longer than 100ms or something like that (though idk if 100ms is a too large buffer time, it might be for time sensitive stuff like games)
grand shadow
My quantum is 100 ms that could be an issue lmao
Idfk why I set it that big
Change it in scheduler.c to 10 ms or something and see if that helps
Actually it shouldnt be an issue if the thread blocks?
Idfk
carmine swallow
yeah looks like that didn't fix it, ig its unlikely that any of those threads run for 100ms without blocking anyway
grand shadow
Hmmmm
Ig you can enable e9 logging and just add some arch_e9_puts("doing foo"); around
Also did the oss utils build and work
carmine swallow
they did build file and I assume at least ossplay should work (though iirc it uses some of the weird ioctls like setsong that I didn't yet handle in any way)
that ioctl doesn't even do anything other than sets the song name that you can then retrieve with an other ioctl lol
grand shadow
Bruh lmao why is that even a thing
carmine swallow
yeah idk
ig you could like have some other program display the currently playing song name somewhere? but idk if anything actually does that
grand shadow
True
proper meteor
ig you could like have some other program display the currently playing song nam...
well my first thought is when you have your session locked while having some media playing in the background. you can have a widget on the unlock screen with the name of the media and a play/pause button
you know, something like that on android
fervent fjord
just get the stream name and executable from the audio bus and interface with OSS/ALSA/pulse/pipewire to manage it
its really common
if you want an example in linux, check gnome or cinnamon
carmine swallow
yeah ig there are actually uses for the name when its used without a separate sound server
flat copper
yeah I don't think unless a thread can not get scheduled in a period longer than...
also music production is extremely time sensitive
wild marsh
well my first thought is when you have your session locked while having some med...
well on desktop linux this is done via mpris over dbus
winged basin
mpris, not to be confused with the late empress of the raj, victoria
grand shadow
I left gcc building with CFLAGS="-O0" CFLAGS_FOR_TARGET="-O0" to see if thatd make a difference
When Im home Ill find out
grand shadow
didn't change anything sadly
grand shadow
init now does what I wanted it to do
waits for poweroff/reboot command,
sends sigterm to all, waits 5 seconds OR until everyone has died
sends sigkill to all, waits until everyone has died
syncs disk
shutsdown
I do need to deal with this console stuff keeping the input
grand shadow
there we go
I moved startwm into /usr/libexec/astral/startwm
and made an alias that only appears if running on /dev/console
robust geyser
damn, very mice
how did you fix bash not wanting to die immediately? just curious
i don’t imagine it was hard but im still curious what the problem was 😅
wild marsh
how did you fix bash not wanting to die immediately? just curious
startwm now execs the script
as per the alias
grand shadow
how did you fix bash not wanting to die immediately? just curious
bash ignores sigterm
so I just worked around that
grand shadow
bash ignores sigterm
When bash is interactive, in the absence of any traps, it ignores SIGTERM (so that kill 0 does not kill an interactive shell), and SIGINT is caught and handled (so that the wait builtin is interruptible). In all cases, bash ignores SIGQUIT. If job control is in effect, bash ignores SIGTTIN, SIGTTOU, and SIGTSTP. ```interesting
grand shadow
either way
I implemented permission checking for kill()
so I think that is all for multiple users
robust geyser
but as a regular user?
grand shadow
ohhh
robust geyser
idk which one you use
robust geyser
unbased
grand shadow
gone are the days
wait holy shit
if qwinci's hda driver works on real hardware
really smooth quake gameplay soon with audio????
robust geyser
imagine if you could boot it off usb…
analog berry
It could probably hit a thousand fps with a proper igpu driver
robust geyser
need to get my shi together 
analog berry
Same
analog berry
Lol
robust geyser
(im cooked)
analog berry
To be fair porting uacpi is a few hours and implementing xhci well is like half a year probably
Xhci and the entire USB stack I mean
grand shadow
probably easier to just install to disk by writing enough nic drivers and downloading the root filesystem
robust geyser
Xhci and the entire USB stack I mean
i did it in less than a week in my kernel lol
analog berry
Well I said well 
fleet osprey
probably easier to just install to disk by writing enough nic drivers and downlo...
I have Realtek 8169 GbE if you're interested in doing a driver for that
I also have a few E1000 (and variants) PCs
analog berry
Well means it works on 99% of real hw and does hubs and stuff
Plus hid and mass storage
If u actually tryhard it full-time maybe you could pull it off in a few months
analog berry
i did hid boot protocol only tbh
Well yeah lol that's what I mean
analog berry
Did u support hubs?
analog berry
Thats where all the pain in the ass comes into play as well
analog berry
but i dont imagine its too hard
Idk managarm still doesn't work with hubs
So I imagine its pretty hard
grand shadow
I have Realtek 8169 GbE if you're interested in doing a driver for that
I also h...
I mean if it'd help test on real hardware it'd be nice
fleet osprey
yup, that's the idea lol
robust geyser
i am pretty sure managarm usb does hubs? idr tho
analog berry
i also did usb ncm
But yeah like if u sorta hack stuff together in a reliable way that might work much faster
robust geyser
i think i saw some code
analog berry
i am pretty sure managarm usb does hubs? idr tho
Does yes, do they work tho? No
fleet osprey
I can say that about so many things with my projects its honestly kinda sad
does code for it exist? yes
does it compile? maybe
does it work when you run it? no
analog berry
i think i saw some code
I saw some of the qookies attempts at debugging xhci and it kinda sucked
Like its pretty hard
But maybe im wrong
fleet osprey
I think debugging USB is just hard in general
I need to do it on the Wii.... and man does it suck
analog berry
Ohh
fleet osprey
never would I wish this message on my greatest enemy
[ 3.805877] ehci-hlwd d040000.usb: fatal error
[ 3.807342] ehci-hlwd d040000.usb: HC died; cleaning up
I mean look
This driver works on other wiis
Right
Its probably a hardware problem
fleet osprey
not with my janky kernel, lol
the issue is when I port the driver to a newer kernel revision
4.20 (nice) vs 4.5
analog berry
69 is when u say nice
fleet osprey
close enough
analog berry
Anyway it works on a newer kernel?
fleet osprey
no, the opposite
analog berry
Just bisect the bug then, that should be easy
fleet osprey
the driver is written for 2.6, ported to 3.15, ported to 4.4, ported to 4.5
works in 4.5, doesn't work in 4.20
analog berry
git bisect
Find the commit that caused the regression
Why is 4.5 the latest one u tried?
grand shadow
Wonder what I should work on now
analog berry
Oh sorry for stealing your thread again
fleet osprey
Why is 4.5 the latest one u tried?
it's not, but let's move this to #lounge-0 or smth, this is the astral thread
lol
grand shadow
Lmao its fine
fleet osprey
same idea at the same time
analog berry
Wonder what I should work on now
Hmm
Well u could unfuck ps2
Make it more stable on real hw
Add support for synaptics touchpad and multiplexed ps2 if ur interested as well
Both relatively easy drivers
grand shadow
I wouldn't have any hardware to test that on tho
both my machines ps2 works for some god unknown reason while everywhere else it just
doesnt
I could look at other implementations and see what I do differently
analog berry
Well you could at least look how Linux does it and see what ur doing wrong and what you're not handling at all etc
analog berry
I could look at other implementations and see what I do differently
Yeah
Or freebsd or anything else
fleet osprey
🎉 nice
grand shadow
bruh
bruh
bruh
bruh
bruh
wait nvm
actually
bruh
wait no
aaaa
idk
I might just throw this all out and rewrite it
grand shadow
I've been having some weird brain fog these past few weeks and its not helping
grand shadow
looking at what ultralegacy does it might be from my timeouts being too restrictive to me not properly sanitizing the input
grand shadow
yeah I understand the fine print now
grand shadow
holy shit the ps2 code seems to have worked first try on qemu lmao
craggy tide
at least you didnt half promise an xhci driver and a usb stack
yeah right :^)
grand shadow
time to see if this will work on the real deal
Expected (no ps2 mouse)
Funnily enough it asks for a resend but there is nothing there
And I forgot to add a check ooopsie
For too many resends
Obligatory acpi for infy
robust geyser
what is that stuff printed to the console lol
grand shadow
I think it tried to print the bash pront to the console and also the init message
I do wonder WHY THE FUCK does uefi give limine a 800x600 framebuffer
I am using an 1080p screen
Doom does work on my desktop tho
Which is nest
Neat
wild marsh
what more could you need
that reminds me that sdl is borked on mammogram so no dsda-doom for me there
grand shadow
And x does seem to work too
Just no mouse
Cuz I dont have one
And the fact its 800x600 grrr
grand shadow
that reminds me that sdl is borked on mammogram so no dsda-doom for me there
Wait what
wild marsh
dsda-doom thinks a window is open but it literally isn't and i can't do anything
i also need to update the port since it's at like v22.5 and the current version is v28.1
grand shadow
Maybe its fixed in a new version
wild marsh
i think it's an sdl issue
iirc tyr-quake also has problems with getting a window open
it used to all work
grand shadow
oh weird
tyr-quake works fine on astral
what version of sdl and tyrquake do you use
the tyrqauek I'm at is 0.71 and sdl is 2.30.2
wild marsh
sdl 2.28.4 and tyr-quake 0.65
hmm sdl doesn't even compile :^)
/var/lib/managarm-buildenv/build/tools/system-gcc/lib/gcc/x86_64-managarm/13.2.0/../../../../x86_64-managarm/bin/ld: CMakeFiles/SDL2.dir/src/video/wayland/SDL_waylanddatamanager.c.o: in function `wl_data_source_destroy':
/var/lib/managarm-buildenv/build/pkg-builds/sdl2/wayland-generated-protocols/wayland-client-protocol.h:2353: undefined reference to `wl_proxy_marshal_flags'
damn
grand shadow
And x does seem to work too
laptop test time
And it works on my laptop + acpi for infy to be happy
Im gonna send an iso here and uh
@fleet osprey you had a bunch of ones that the ps2 didnt work right
I did some changes to the driver can you test them again thx
Im gonna be putting enough in the iso to run X and some fun packages
@sacred gyro
it has these packages
MINIMALPACKAGES=mlibc bash coreutils init distro-files vim nano mount netd shadow sudo neofetch ace-of-penguins xterm fvwm xorg-server xclock xeyes doomgeneric doomgeneric-x11 xneko xf86-input-keyboard xf86-input-mouse xf86-video-fbdev xfontalias xfontmiscmisc xfontmiscethipic
fleet osprey
yeah, will get it soon
parents are being annoying, won't have a chance to check it for a decent bit
grand shadow
Damn
wild marsh
```
/var/lib/managarm-buildenv/build/tools/system-gcc/lib/gcc/x86_64-managarm/13...
perhaps if i actually had sdl 2.28.4 sources and not 2.0.14 it'd work better
craggy tide
I do wonder WHY THE FUCK does uefi give limine a 800x600 framebuffer
have you tried forcing it manually?
wild marsh
dsda-doom thinks a window is open but it literally isn't and i can't do anything
fixed it with the kmsdrm video driver, the x11 driver seems to freeze weston? https://uwu.foundation/3W5iHux.png
the kmsdrm driver has broken logic in it's init function too which was fun
it says that if something fails it's fine but it ends up returning -1 anyway
wild marsh
fixed it with the kmsdrm video driver, the x11 driver seems to freeze weston? ht...
okay yeah SDL_VIDEODRIVER=x11 seems to make weston freeze, and SDL_VIDEODRIVER=wayland leads to egl failing to initialize
fleet osprey
ok checking the new PS/2 code now
PS/2 Keyboard does NOT work on my ThinkPad T14 Gen 1 AMD, probably no mouse either (but I can't start xorg to test)
power button works w/ ACPI poweroff tho 👍
On the Precision T1650, no keyboard emulation support in UEFI, and I can't get it to boot in BIOS (firmware just yells at me, Rufus probably doesn't like me for some reason or another)
And sadly no access to a real PS2 kbd to test the real PS2 ports, nor a low power enough keyboard to use the USB adapter
ACPI shutdown works there too
aand the Optiplex 755 just hangs trying to boot it
I hit enter on the USB Device option in the boot menu, and then nothing visible occurs
remains on the same VGA text mode boot screen into infinity
probably a side effect of trying to flash the ISO to a USB stick
but I really don't want to go dig out a whole ass optical drive and some CD-Rs just for this
and regardless, I am out of time to for today
if you want to provide a BIOS-bootable USB image, I can try it out on there, which'll be interesting
this thing has a known bugged PS2 emulation mode
something like every other key, it'll just eat the IRQ for the keypress
it'll go into the buffer, but not actually fire an IRQ
so the next time you press a key, it'll fire an IRQ, but have 2 keys in the buffer (at least, that's my understanding of it)
DOS and early Win9x actually choke hard on this, on the next IRQ after the bug occurs, it'll only readthe initial (bugged) input, so your inputs are now permanently 1 keypress behind
and it just keeps adding onto itself, e.g. if it triggers again, the OS is now permanently 2 keypresses in the past, etc
so I really do wish I could get Astral running on here, really interested to see how it handles that
craggy tide
@fleet osprey if you could keep in mind the machines one which it did not boot...
but didn't we go through this together already?
analog berry
power button works w/ ACPI poweroff tho 👍
nice
im very happy with uacpi success rate at this point
analog berry
holy shit the ps2 code seems to have worked first try on qemu lmao
are u saying your ps2 code is based on uPS2
grand shadow
have you tried forcing it manually?
I should
grand shadow
PS/2 Keyboard does NOT work on my ThinkPad T14 Gen 1 AMD, probably no mouse eith...
Aaaaaaaa
Did any error messages appear
fleet osprey
<@632725314111275018> if you could keep in mind the machines one which it did no...
Yes, and Yes
the issue being the way that the image is made, it just can't work, iirc from before
I would either need to:
A - boot from Ventoy, which breaks the image
B- use a "real" HDD/USB image, not an ISO
fleet osprey
Did any error messages appear
nope, not on either machine, they both got to a root shell, but PS/2 emulation did not seem to work
grand shadow
are u saying your ps2 code is based on uPS2 <:meme:575445094589661215>
I took some ideas™️
grand shadow
nope, not on either machine, they both got to a root shell, but PS/2 emulation d...
Wtf so it passed everything and still didnt work
The machines I have use real ps2
Maybe I should try checking how my desktop does with emulation
grand shadow
Emulation works on my machine
Bro even ps2 mouse emulation works
The mathewnd curse I guess
carmine swallow
I should also test astral on my laptop, I know that with my own os ps2 sometimes doesn't properly work on it
it just doesn't respond to self test (or at least not in the order that I expect that is first the ack and then the result, idk maybe it does it in the order order or whatever I haven't really verified)
grand shadow
Thatd be neat I really need to get to know whats wrong to get ps2 fixed
obsidian pivot
are u saying your ps2 code is based on uPS2 <:meme:575445094589661215>
If you need to fix a driver you're writing, then looking to ultralegacy always helpa
*helps
fleet osprey
I really need to test it on my optiplex at some point
if that works, I would be very confused
since it must only work with buggy PS2 implementations
lol
grand shadow
I mean it worked with two real controllers already
Idfk what could be wrong since in every possible case itworks on my machine
Even ps2 mouse emulation which Ive heard is almost never right worked
I really am out of ideas
Maybe the initialisation is fine (you said nothing errors out) but the actual interrupt or scanning stuff is bad?
fleet osprey
I'm mainly interested in getting it to work on my optiplex so I can compare against my own impl from an old version of TechflashOS, that I know for a fact worked on this exact machine
no idea if it works on anything else since I haven't touched TFOS in years
analog berry
If you need to fix a driver you're writing, then looking to ultralegacy always h...
Lol
grand shadow
@robust geyser if you have any machines you can try on can you try astral on real hw and see if ps2 works
sacred gyro
I hear you need people to test out PS/2
I’ll test it on my Chromebook
Which iso am I using?
grand shadow
Has a really basic x installation
grand shadow
<@731216127019057273>
@sacred gyro
It edited tge message for some reason
Instead of replying
sacred gyro
Ok, will test in a bit
sacred gyro
ok, so the keyboard didn't want to work....
but the power button, which is on the keyboard, did...
flat copper
ok, so the keyboard didn't want to work....
isnt chromebook keyboard spi or something
errant zephyr
isnt chromebook keyboard spi or something
depends on the model
sacred gyro
yeah
sacred gyro
some of them do have PS/2 because the PS/2 on my OS magically worked on one of my chromebooks
flat copper
depends on the model
looks like some new ones are i2c
analog berry
but the power button, which is on the keyboard, did...
Thats routed differently so thats expected
The power button goes through acpi and not ps2 or spi or whatever
sacred gyro
ye
grand shadow
ok, so the keyboard didn't want to work....
Going insane why is it only on my machines that it works
Did any error happen?
analog berry
Yeah that's the most default ps2 keyboard
sacred gyro
so theoretically it should work then?
analog berry
Yes
sacred gyro
must've somehow missed it the first time
fleet osprey
I don't even think my driver tried to self-test
it just blindly assumed that "yeah it totally exists"
I forget if I even throw any init stuff at it
just blindly assume it's already set up and on IRQ 1
but that may be wrong, I wrote it a very long time ago
....
yeah so my initialization routine was (when it still existed) a bit more simple than yours
but weirdly, it works on my hardware where Astral doesn't
fleet osprey
Maybe the initialisation is fine (you said nothing errors out) but the actual in...
so then, it would seem to be actually the opposite of this?
flat copper
but weirdly, it works on my hardware where Astral doesn't
ps2 is fucked
some machines require init to work properly
some machines work only if you don't init
gritty shore
why do you guys keep talking about ps2, the ps5 launched like 2 years ago already..
wild marsh
who's gonna buy me a ps4 so i can port mammogram to it :^)
fleet osprey
I wonder how Linux does it then
I know for a fact it works on this ThinkPad where Astral doesn't (I'm typing this message on it rn)
I haven't tested it w/o USB (so it forces PS2 emulation) on the Optiplex 755 or Precision T1650 though
gritty shore
who's gonna buy me a ps4 so i can port mammogram to it :^)
put it on the company's credit card
"business expenses: ps4"
lean kettle
who's gonna buy me a ps4 so i can port mammogram to it :^)
if you go to austria id happily sell you my ps4 i dont use lol
grand shadow
Here’s a video of it booting if it helps
Thats the ps2 mouse not being found which is probably right
I do wonder why the keyboard doesnt work
sacred gyro
Yeah, the touchpad in here is i2c
carmine swallow
I really can't get the sdl audio properly synchronized 💀 I even switched it to use double buffering so the queue function puts data into a ring buffer that the irq handler then takes data from if there is data and if there isn't then it just increments the position so it keeps up with the hardware (and ideally it should also memset it to zero so some random data isn't being played again but that's not relevant for this)
grand shadow
I really can't get the sdl audio properly synchronized 💀 I even switched it to ...
Does it take too long to run the irq or is it just hda memes?
carmine swallow
I don't think it should take that long because it just locks a spinlock and then checks the ring buffer + potentially reads a page from there and wakes up a semaphore (though ideally it would be an oneshot event that can't be triggered multiple times but eh)
and there are like 46 irqs per second
(for 48000hz 2channel 16bit)
carmine swallow
if it would take too long for it to run the irq after it has consumed one page of data then it would also break catting a wav file into the device file
it generates an irq after it has played whatever buffer period worth of data you specify
grand shadow
Hmmmm if catting isnt broken then it could just be an issue with sdl
Does sdl write proper data into it or does it take too long
carmine swallow
well I do have one thing that I could try, I could try using whatever sdl specifies as the fragment size (which is basically how much data you can queue without blocking) instead of hardcoded page size
maybe sdl calculates its internal timing based on that or smth
grand shadow
Fragment size?
carmine swallow
its the audio buffer size as far as I understand and there is an oss ioctl to set that (so then you can write that much data before it has finished playing the old data without blocking)
flat copper
its the audio buffer size as far as I understand and there is an oss ioctl to se...
afaik a 64-128ms buffer is the norm to avoid noises
46 irqs per second means 20ms
carmine swallow
afaik a 64-128ms buffer is the norm to avoid noises
hmm so it would be like 10k-24k bytes (with 48000hz 2ch 16bit)
anyway Ill try to see what sdl actually wants to set it as
grand shadow
Yeah maybe sdl just wants it set up in a specific way
carmine swallow
now quake broke somehow so it gets stuck even before it does any audio ioctls it just says "initializing palettes"
grand shadow
The fuck
Maybe some disk corruption? Try regenerating the image
If not some system call logging
craggy tide
Yes, and Yes
the issue being the way that the image is made, it just can't work,...
the question is also whether Limine is on the latest version
craggy tide
Here’s a video of it booting if it helps
yeah judging from this video, the answer is not
it's not updated
@grand shadow please bumpity bump Limine to latest version, i fixed bugs that @fleet osprey reported a while ago
grand shadow
I havent bumped limine in a while and its in my todo list
I will do that and resend techflash an iso
fleet osprey
if you want I can just do it on my local build
fleet osprey
yeah remember, from when I was poking at trying to get procfs
grand shadow
can you also help debug ps2 Ill give you a cookie
fleet osprey
I don't know anything about debugging PS2
If you can give me a list of changes to try I'll try them all of my machines
best I can give is
- did it throw an error
- any debug logs if any were added
- does it work when I try to use it
the answer being no to it throwing an error, and no for it working, on all machines I've been able to test
grand shadow
Change distrotype to minimal in the makefile to make things faster
Add a printf to the ps2 keyboard interrupt and also a
for (;;) {
bool timeout;
uint8_t data = ps2_read_data_timeout(100, &timeout);
if (timeout)
break;
}``` to the end of ps2_init in arch/x86-64/ps2.ck, one sec
gonna try bumping limine first, since I can see if anything breaks here and now
because I'm not at home with my USB drives atm, so I have no way of booting bare-metal on my thinkpad (and ofc, my optiplex is at home as well)
sacred gyro
I'm now confused as to how my PS/2 driver works on my chromebook, but Astral's doesn't...
sacred gyro
idk
grand shadow
The mathewnd curse, ps2 will work on all machines I can test on myself but not on any others
sacred gyro
you could always go look lol
fleet osprey
that seems to be the biggest difference
mine doesn't actually do anything to init the PS2 controller or keyboard, just blindly assumes that both exist, and the keyboard will be firing on IRQ1, Astral's actually does it the "right" way
also, you're fine with bumping to Limine v8? that's 1 major version bump, you're on v7 rn
grand shadow
also, you're fine with bumping to Limine v8? that's 1 major version bump, you'r...
That should be fine just check for any config file changes
fleet osprey
👍
welp, rebuilding clean now because I botched the command
lovely
this'll take a bit
fleet osprey
it's just make DISTROTYPE=minimal, yeah?
grand shadow
it's just `make DISTROTYPE=minimal`, yeah?
You can change it in the makefile so you dont have to pass it every time
tawdry edge
fleet osprey
yeah, it totally did change, will fix
got one config working, no background or anything else working yet
won't have time to finish this sadly, gtg
fleet osprey
looking good so far
all of the options seem to boot w/ no problem
er- well, they start the kernel with no problem
I don't have it set up for NVMe yet
but if that didn't work it wouldn't be a limine issue anyways
PR created with all of these changes now, btw
carmine swallow
on my laptop I get "mm/alloc.c:getcachefromsize:45: cache failed"
after it prints the pci device list
my bet is that its uacpi that's allocating more than 64k for some acpi stuff
carmine swallow
my bet is that its uacpi that's allocating more than 64k for some acpi stuff
yeah I was right, adding a couple more sizes up to 1mb fixed it
fleet osprey
yeah and that works too
probably could've tried before submitting, but I'm also busy :P (tbf I did see it fail to mount /dev/nvme0n1 so I knew the args in limine were working)
carmine swallow
ps2 keyboard doesn't work on my laptop either (at least this boot it doesn't)
it didn't even time out or anything from the looks of that log
grand shadow
It might really be that it has some bogus data in the output buffer that prevents keypresses or some shit
Idk
robust geyser
<@731216127019057273>
does this image have the new ps2 stuff?
grand shadow
Yeah
robust geyser
okay perfect
grand shadow
And a basic install of xorg
robust geyser
yep thats mainly what i wanted
i just didnt know what packages to use
and a full distro iso was 2.4G lol
i accidentally built one
grand shadow
Though dont expect it to work lmao
grand shadow
i just didnt know what packages to use
I sent the packages up here
grand shadow
it has these packages
```MINIMALPACKAGES=mlibc bash coreutils init distro-files ...
Here
robust geyser
ok im gonna test it brb
should i use csm?
or does ps2 emulation work under uefi too
fleet osprey
I already confirmed earlier that NVMe inits on all of my machines, but good to have more test cases
carmine swallow
or does ps2 emulation work under uefi too
it depends, on all my pc's it doesn't
grand shadow
I think for uefi you have to enable it
At least on my machine I had to enable emulation
Cuz it had a real controller
robust geyser
damn i got the same thing as qwinci
carmine swallow
Also nice hda and nvme seem.to initialise in real hw
if I had keyboard working I could also try playing some song on it (though its slightly annoying that the volume is at the max by default iirc and I don't know which one of the output paths is for the headphones)
carmine swallow
damn i got the same thing as qwinci
if you don't need xorg then I can send my iso
grand shadow
damn i got the same thing as qwinci
Ugh I dont like having super big slab sizes because it pre allocates some
robust geyser
i wanted to test it so ill just build my own
grand shadow
And ends up using a ton of memory
robust geyser
just create them on demand
grand shadow
I do that its just that I dont really free them when theyre fully empty yet
I think I have all of the logic I just dont call it
If uacpi passes a memory size I can just make it use vmm_map instead of alloc
To free
carmine swallow
if I had keyboard working I could also try playing some song on it (though its s...
maybe there could actually be some "non standard" symlinks with the device types as their names so then you could link that to /dev/dsp to make that the default
fleet osprey
or you could just log what corresponds to what on boot
robust geyser
ah alloc_init
carmine swallow
<@379250891838193667> where and how did you change the slab stuff?
robust geyser
thank you
i was about to ask
🙏
also i couldnt really do non-uefi boot because for some reason this usb has fucking grub in its mbr
anyway, attempt #2
rip
uacpi 0 processors
robust geyser
god bless
grand shadow
uacpi 0 processors
LMAO
robust geyser
nope, it should be emulated
carmine swallow
uacpi 0 processors
it means that there are 0 legacy processor objects btw
grand shadow
I got something like that on my computer when I didnt have anythingplugged to ps2 but it wasnt emulating
robust geyser
(obligatory acpi test)
carmine swallow
on my laptop it didn't because it uses ec
robust geyser
I got something like that on my computer when I didnt have anythingplugged to ps...
w h a t
grand shadow
Like
My pc has a real controller
And when I didnt have anytjing plugged into the ports
It had the same fails
Which madr me realise I had to turn on emulation in the bios
analog berry
it means that there are 0 legacy processor objects btw
I should really remove this, it confuses people lol
analog berry
If uacpi passes a memory size I can just make it use vmm_map instead of alloc
Yeah it can be enabled optionally
robust geyser
well i have nothing regarding ps2 or emulation in my firmware settings
robust geyser
so i assume i need to use legacy boot
analog berry
They went all out on this aml blob
robust geyser
aka csm
aka i need a disk image instead, i guess
so i can flash it to the drive
and overwrite grub
carmine swallow
why not just dd the iso to the drive?
robust geyser
actually idk
im on windows
and i use rufus
ill try to mount the usb in wsl
unless someone has a better and easier idea
carmine swallow
netboot (assuming you have a second pc) idk how slow would it be with these initrd sizes tho
robust geyser
its ~250MB so
also no 2nd pc
at least thats easily available to me
well its under my desk
but i would need to plug in a display
blah blah blah
grand shadow
and i use rufus
Isnt there an option in rufus or smth
carmine swallow
also now I had ps2 working on my laptop for 2sec after boot but when I started typing stuff then it got stuck
analog berry
I have a lot of laptops it least
robust geyser
fleet osprey
I've tried rufus, it just won't work on BIOS
grand shadow
<@215823874615541761> I might unfuck ps2 myself if u want
Please I have no idea wtf Im doing wrong
fleet osprey
all of my machines either hang, or say that the boot failed
robust geyser
what did you end up doing?
grand shadow
Ill give you a cookie and a big thank you
robust geyser
free cookies lifehack
fleet osprey
what did you end up doing?
give up until I eventually make a make img option that makes a boot HDD/USB image directly
vs make iso
robust geyser
ok well ill make one then
analog berry
Please I have no idea wtf Im doing wrong
sure ill do it some time soon ™️
fleet osprey
btw @grand shadow, since the message kinda got burried and idk if you ever actually saw it:
limine v8 PR - https://github.com/Mathewnd/Astral/pull/9
grand shadow
btw <@215823874615541761>, since the message kinda got burried and idk if you ev...
I got the email Im on the bus Ill look at it when Im home
fleet osprey
ah
grand shadow
sure ill do it some time soon ™️
And tell me what I did wrong if you find out cuz I really am curious
analog berry
im not gonna be using ultra legacy as base probably, ill look at proper production quality implementations
i want this to work everywhere
im kinda past the stage of adhoc crappy drivers
grand shadow
Ps2 was harder to make work on real hardware than like nvme damn
I guess it makes sense
analog berry
it makes sense yeah
same reason why ide is harder than nvme
years and years of legacy crust
robust geyser
bruh hdd.img has no partition table
almost like i forgot
i wanted to bios-install --force-mbr then i realized
bruuuuuuuuuuuuuuh what to do
ill use ventoy i guess
fleet osprey
ventoy breaks it
fleet osprey
initrd won't load
robust geyser
well ill try it anyway
flat copper
same reason why ide is harder than nvme
i mean, nvme will be the same in 20 years
just like ahci did before it
flat copper
then it's all uniformed into ugly scsi on windows
flat copper
ahci being sda yes but is nvme?
analog berry
it really needs to die
fleet osprey
you're gonna want SCSI support if you want to ever support USB storage
analog berry
yes, thats the unfortunate part
carmine swallow
also ufs uses scsi
flat copper
yeah but it doesnt mean that you have to convert every disk format into scsi
analog berry
ahci being sda yes but is nvme?
wdym?
flat copper
wdym?
afaik linux blockdev isnt necessarily scsi
ahci is, and that's why disks are named sdN
but nvme isnt
but ill have to check
analog berry
oh idk, i know there's like a giant scsi thing for USB, IDE, and AHCI
flat copper
yeah that's for sure
analog berry
like part of the kernel
its a very slow subsystem
with too many layers of abstraction
fleet osprey
IDE is actually different last I checked, isn't it like /dev/hd[x]?
analog berry
yeah i think so idk
robust geyser
does work either under csm :(
fleet osprey
USB and AHCI are indeed turned into SCSI tho
robust geyser
maybe my firmware just doesnt do ps2 emulation
flat copper
alot of new machines dont at all
fleet osprey
my main desktop has an option for it in the firmware
robust geyser
i guess i have to finish my xhci driver then
fleet osprey
and has physical ports too
sadly I don't hae a physical PS2 kbd, nor a USB one low-power enough to use with an adapter
I only have fancy gamer RGB keyboard that needs 5V @ 0.8A or something
carmine swallow
i guess i have to finish my xhci driver then
I am interested in testing that with the quirky weird usb 1.1 mouse that I have from 2002
analog berry
i guess i have to finish my xhci driver then
yeah
hopefully u finish it before i unfuck ps2 for astral
flat copper
I am interested in testing that with the quirky weird usb 1.1 mouse that I have ...
i have one too
robust geyser
hopefully u finish it before i unfuck ps2 for astral
hell no
flat copper
i find my 2003 ms mouse too comfy to switch from it :^)
robust geyser
have fun unfucking ps2
flat copper
i have a pic of 3yo me with it lmfao
robust geyser
fleet osprey
does work either under csm :(
yup, that's the ventoy skill issue
it fucks the initrd
analog berry
have fun unfucking ps2
i have some experience with a relatively stable driver for it, but yeah
flat copper
ventoy
robust geyser
then i guess you'll actually have fun ;^)
carmine swallow
i have one too
so far it failed on managarm and my own kernel, it just sends transaction error for the first get descriptor command no matter what I tried (when its attached on boot it sometimes works tho)
I have no idea what actually is wrong
fleet osprey
yeah i kinda like ps2 idk
I also like PS2, specifically because of this :)
analog berry
I also like PS2, specifically because of this :)
lol
carmine swallow
did you try the weirder winxp era usb init sequence
if you mean resetting the device and then reading the first 8 bytes of the device descriptor that's what I tried to do
flat copper
yeah
carmine swallow
but it always fails with the transaction error
I also tried different packet sizes but no luck
flat copper
what does linux do ^:)
carmine swallow
idk
flat copper
shove log messages in linux then figure out what init it does
carmine swallow
at least it works
flat copper
it should count as black box reversing so you dont even need licensing bs
grand shadow
@carmine swallow if you send me an iso with hda and the new ps2 I can try it out on my desktop and laptop
wild marsh
linux has tracing in the xhci driver that you can enable at runtime
grand shadow
Cuz ps2 works there
carmine swallow
<@379250891838193667> if you send me an iso with hda and the new ps2 I can try i...
sure, Ill just put the song back to it because I got rid of it to make it faster to flash
wild marsh
not that it helped me debug the managarm driver :^)
fleet osprey
shove log messages in linux then figure out what init it does
yeah, it doesn't seem to log all that much about actual init by default
techflash in ~ at Michael-ArchT14 ➜ sudo dmesg | grep -iE 'i8042|PS/2|PS2|serio'
[ 1.406297] i8042: PNP: PS/2 Controller [PNP0303:KBD,PNP0f13:MOU] at 0x60,0x64 irq 1,12
[ 1.410161] serio: i8042 KBD port at 0x60,0x64 irq 1
[ 1.410257] serio: i8042 AUX port at 0x60,0x64 irq 12
[ 1.602187] input: AT Translated Set 2 keyboard as /devices/platform/i8042/serio0/input/input5
[ 8.508570] mousedev: PS/2 mouse device common for all mice
[ 9.558025] psmouse serio1: synaptics: queried max coordinates: x [..5678], y [..4694]
[ 9.594049] psmouse serio1: synaptics: queried min coordinates: x [1266..], y [1162..]
[ 9.594069] psmouse serio1: synaptics: Your touchpad (PNP: LEN2064 PNP0f13) says it can support a different bus. If i2c-hid and hid-rmi are not used, you might want to try setting psmouse.synaptics_intertouch to 1 and report this to [email protected].
[ 9.660605] psmouse serio1: synaptics: Touchpad model: 1, fw: 10.32, id: 0x1e2a1, caps: 0xf014a3/0x940300/0x12e800/0x500000, board id: 3471, fw id: 2909640
[ 9.660655] psmouse serio1: synaptics: serio: Synaptics pass-through port at isa0060/serio1/input0
[ 9.721786] input: SynPS/2 Synaptics TouchPad as /devices/platform/i8042/serio1/input/input16
[ 10.267028] psmouse serio2: trackpoint: Elan TrackPoint firmware: 0x12, buttons: 3/3
[ 10.508459] input: TPPS/2 Elan TrackPoint as /devices/platform/i8042/serio1/serio2/input/input22
[ 11.577124] psmouse serio1: TouchPad at isa0060/serio1/input0 lost synchronization, throwing 5 bytes away.
i meant add debug prints to linux
analog berry
oh yeah i can use uacpi now to detect the touchpad and stuff
on astral
so thats nice
fleet osprey
i meant add debug prints to linux
yes, that's what I said lol
it doesn't log that much by default, so you would indeed need to add new debug messages
winged basin
USB and AHCI are indeed turned into SCSI tho
is nvme too?
fleet osprey
no
winged basin
a lot of oses scsify everything
carmine swallow
linux has tracing in the xhci driver that you can enable at runtime
maybe Ill try that
analog berry
nvme is its own thing
fleet osprey
at least, it isn't in Linux
wild marsh
maybe Ill try that
enable in /sys/kernel/tracing/events/ and for the actual trace/sys/kernel/debug/tracing/trace
carmine swallow
<@379250891838193667> if you send me an iso with hda and the new ps2 I can try i...
here, hopefully the generic device class match works for your hda controller, I did also add intel tigerlake smartsound technology id to it manually because that's what I have on my laptop
if there are multiple outputs then each different output should have their own /dev/dspX file you can cat stuff to, though like I said you can't really know which one of them corresponds to what physically rn
carmine swallow
100%
fleet osprey
or headphone-exploding
carmine swallow
it shouldn't be that loud at least on all hardware that I have tested it on, for an example its not the same as max software volume in linux
but better be careful ig
carmine swallow
if you find the correct output then in theory yes
carmine swallow
no it doesn't, on some machines there actually are many different outputs for eg. different kinds of output ports like hdmi and different kind of analog outputs
fleet osprey
💀
grand shadow
Does your hda loop not support interrupble sleep for like ^C
fleet osprey
yikes
carmine swallow
Does your hda loop not support interrupble sleep for like ^C
is there something that I have to do for that to work? because if yes then it likely doesn't
grand shadow
What do you use for sleeping