marcan changed the topic of #asahi to: Asahi Linux: porting Linux to Apple Silicon macs | "Does XXX work yet?": https://alx.sh/fs | GitHub: https://alx.sh/g | Wiki: https://alx.sh/w | Topics: #asahi-dev #asahi-re #asahi-gpu #asahi-alt #asahi-stream #asahi-offtopic | Keep things on topic | Logs: https://alx.sh/l/asahi
Zopolis4_ has quit []
A_L_I_C_E has quit [Remote host closed the connection]
<ktz>
the errors in the tail are produced by setting wifi into ap mode which works and then trying to start an ap
<ktz>
device wlan0 set-property Mode ap
<ktz>
ap wlan0 start net passphrase
MajorBiscuit has quit [Quit: WeeChat 3.6]
<ktz>
with those commands, prepended with iwctl of course
<jannau>
ktz: I think it's suspected/known broken. not a priority item to fix. if you need ap mode I'd recommend getting usb wifi adapter
giskard has joined #asahi
<ktz>
I'm mostly trying to make sure my machine is properly configured at this point, eg the builtin keyboard only functions in a tty console and the touchpad doesn't work at all, I think it could be due to this
<ktz>
[ 0.857703] apple-dcp 231c00000.dcp: Could not read 'apple,firmware-compat': -22
<ktz>
I'm not running alarm tho so most likely that's a discussion for #asahi-alt
<jannau>
the dcp error means you're using an old 2nd stage m1n1
bps has joined #asahi
<ktz>
second stage means initrd?
atodorov has joined #asahi
<jannau>
that shouldn't have anything to do with keyboard/touchpad though. the error looks like the problem we had with the outdated libinput version
<ktz>
ok thanks for the hints on m1n1 and libinput, I'll try to take it from there
<jannau>
there is a second m1n1 on the ESP in asahi/boot.bin. that one is used to keep the devicetree updated with kernel updates and allows to distribute m1n1 updates
<mps>
ktz: second stage m1n1 means m1n1+dtbs+u-boot on ESP in m1n1/boot.bin file
<mps>
oh, jannau is faster
<mps>
ktz: iirc you are using alpine, upgrade m1n1 and u-boot-asahi and linux-asahi pkgs, that should upgrade second stage m1n1
<ktz>
apparently this is indeed dated on my distribution, I think it happened a couple of times having it being overwritten by an alarm update which caused my distro being super slow, I had to chroot from alarm and replace m1n1 with the .old version the update adds
<ktz>
mps: I didn't even have those packages installed... how comes? I've never installed them manually on the one year+ I'm running alpinelinux
ktz has quit [Remote host closed the connection]
<mps>
you should upgrade, they should be installed automatically
<mps>
apk add m1n1 u-boot-asahi, will do 'trick'
<mps>
also there is '/usr/sbin/update-u-boot-asahi' which could be run manually to update second stage m1n1
<mps>
and this is for #asahi-alt ofc
ktz has joined #asahi
jn has quit [Read error: No route to host]
jn has joined #asahi
<marcan>
ayke_: it is likely that we will stop supporting Fn key combinations on Xorg in the future (other than the edit keys for minimal functionality), so no more media/system keys.
<marcan>
(as soon as I figure out how to make at least the F-keys do the right thing on XWayland, which I don't want to break)
<marcan>
right now the Fn mess is handled in the keyboard driver and that is just thoroughly wrong, we need to chagne that. But Xorg can't see the Fn key and never will, because it's more than 8 bits and there is no way to fix that in the legacy protocol.
<pipppero>
fn + media key is more than 8 bits?
<marcan>
the Fn key is more than 8 bits
<marcan>
Xorg can't see it at all
<pipppero>
oh wow
<maz>
marcan: can't this be dealt with in userspace with something like keyd?
<marcan>
turns out there are more than 256 possible key codes in the universe, who'dve thought
<marcan>
maz: you could play games with uinput if you want, but that's just overcomplicating what should be a simple xkb mapping
<pipppero>
not in the 80's!
<marcan>
I mean I'm certainly not going to stop people from making workarounds for xorg, but it's not something anyone is likely to ship by default
<pipppero>
...or the beginning of the 2000s
<marcan>
my plan for the keyboard driver is to *only* implement the Fn mappings for the edit keys (home/end/pgup/dn/ins/del) so that at least those work on the TTY/Xorg, and leave the F keys as F keys by default (not media keys)
<maz>
marcan: I'm all for keeping the kernel driver simple and rely on userspace to deal with the outcome (that's what i do for chromebooks).
<marcan>
that will give you usable TTY/xorg
<marcan>
then if you want media keys and extended Fn key chords or customizable F mappings, either use Wayland where all this will work, or roll your own emulation with uinput
kaprests has quit [Ping timeout: 480 seconds]
<marcan>
I'm just not interested in such emulation just for the sake of xorg so it won't be me writing it :p
<marcan>
on Wayland (hopefully) this can all just be dealt with XKB as normal, which means you can also configure it all through existing UIs, and it can all be made nice and user friendly
<chadmed>
we need to shepherd people away from that tangled mess, especially now that its looking increasingly likely that it will go unmaintained in the next year or so
<marcan>
yup
<pipppero>
xorg going unmaintained? :o
<marcan>
the main question right now is how to make at least the F keys work in XWayland clients when those keys are mapped as media-keys-first, since that has to work (I don't intend to break XWayland for at least normal use cases of apps, not desktop integration components)
<marcan>
pipppero: xorg is already basically unmaintained
<marcan>
KDE will drop support in the next version
ktz has quit [Remote host closed the connection]
<chadmed>
theyre struggling to find people interested in working on it, the number of commits has dropped precipitously recently
<marcan>
nobody wants to work on it
<chadmed>
its just a total dead end
<pipppero>
gosh i really am out of the linux on the desktop world nowadays
<marcan>
the issue with XWayland being that... Xorg supports up to 8 modifiers and they are all assigned
<marcan>
so we need to figure out how to either shoehorn Fn in there or remap in the XWayland server
kaprests has joined #asahi
<marcan>
but once F1-F12 work as intended on XWayland that's good enough for me, I expect very few people to care about XWayland clients receiving media keys and don't particularly intend to support it
<marcan>
since almost always these days those would be handled by the desktop environment which better be native Wayland
<chadmed>
to me xwayland is for games linked against old versions of sdl and not a lot else
<ar>
wayland has a major showstopper ;) - oneko doesn't work correctly with it ;)
<marcan>
chadmed: pretty much
<marcan>
well, for me also audio stuff
<marcan>
because lol old audio plugins
<marcan>
the old embed stuff was based on XEmbed IIRC
<chadmed>
pretty sure all the lv2 plugins show their uis via xwayland too which... :/
<marcan>
yeah
<marcan>
I actually had stuff crash with carla on native wayland, works fine with Xwayland so I use that
<marcan>
but whatever, that's what xwayland is for
<marcan>
those use cases don't care about fancy keys
<chadmed>
omg i need to try something right now
<ar>
but seriously, my biggest problem with wayland, is that even xf86-input-evdev works better with some input devices. libinput maintainer sticks his head in the sand and pretends that they don't exist and treating them poorly is fine…
<marcan>
sounds like instead of complaining you should fix it like we did for apple stuff :p
<marcan>
I've had no problems getting libinput patches upstreamed
<ChaosPrincess>
the fn key situation with xorg is so broken, that back in the day i had a patch to that driver to convert fn into f13 and then some other cursed configs to convert it back into Mod-4
<ar>
marcan: the problem is, he also refuses to accept patches that would allow recreating the correct behavior because, omg, they would also require a configuration knob or two
<marcan>
I don't have the context so I can't comment
<ar>
and another usecase, is that on xorg, whenever I want to play around with CAD on my laptop, which requires the middle mouse button to act normally instead of making it work as a scroll, I could just: xinput set-prop 'ETPS/2 Elantech TrackPoint' 'libinput Scroll Method Enabled' 0 0 0
<ar>
and plasma-wayland doesn't expose any such configuration knobs. and, what's worse, settings such as these are compositor-specific
<ar>
so fixing stuff like that for, for example, plasma-wayland, would leave gnome and sway broken
<ar>
obviously, neither of these is a problem for asahi, because apple laptops only ship touchpads. but still, libinput behavior is annoying at times
<chadmed>
"re-bind addtional mouse buttons"
<chadmed>
its right there actually
bpye has joined #asahi
<ar>
chadmed: wdym "it's right there"? in plasma systemsettings → input devices → mouse i see no such option
<ar>
and searching for "additional" there returns nothing
<dottedmag>
marcan: I had a look at >255 keycodes in X before, and haven't found a good solution. One that might work requires work in every compositor: Xwayland would get a modified keymap with Fn keycode moved below 256, and compositor would translate it :-/
<ar>
chadmed: ah. that only allows rebinding mouse buttons that are above 3rd
<ar>
(it allows me to rebind stuff on my desktop with a 2-scroll, 6-button mouse)
<chadmed>
ahhhh makes sense then :/
atodorov has quit [Ping timeout: 480 seconds]
<ar>
(and don't get me wrong - i do like at least the display side of things in wayland. it's just that the input device side tries to shoehorn square pegs into round holes by cutting corners off)
<dottedmag>
ar: Wayland is not a problem. Existing compositors and their exclusive use of libinput is.
<ar>
dottedmag: wayland is the problem, because now every wm/de is its own entire windowing server, and instead of implementing proper input device support once with all the proper configuration knobs exposed, everyone is forced to reinvent the wheel
<chadmed>
i mean wlroots exists
<chadmed>
so yeah its a compositors-not-using-it issue
<psykose>
i don't know of a wm i've used that didn't just expose every libinput setting directly
* mps
listen for more than 30 years that the unix will die next year, same for X
<ar>
mps: well, at least the commercial unixes are, for practical intents and purposes, dead
<mps>
ar: right but unix (linux, *BSD ...) is very alive
<ar>
yup
atodorov has joined #asahi
<mps>
“It's tough to make predictions, especially about the future.” - Yogi Berra
<mps>
ar: aha, I even saw photos of future mobile phones in very old newspapers, even skype was predicted about 40-50 years ago and it was named exactly 'skype'
<chadmed>
unix exists as a philosophy and general approach to operating system design still sure but saying unix-proper still exists because of linux and the BSDs is like saying ARC still exists because NTLDR emulates it
<chadmed>
X will fall out of general use and thats okay
<chadmed>
we got used to all of its quirks, we will get used to waylands
<chadmed>
the nice thing about wayland is that there is a whole mess of folks willing to fix its quirks
<chadmed>
this mess of folks never materialised in all of X's history
<mps>
for me scroll in firefox with sway on m1 is very weird
<mps>
and also, my beloved awesomewm doesn't work on wayland
Nae has quit [Quit: Connection closed for inactivity]
cylm has joined #asahi
bpye has quit [Ping timeout: 480 seconds]
bps has quit [Ping timeout: 480 seconds]
<tobhe>
also wayland doesn't really work on anything but linux
<ar>
tobhe: tbh, *bsds have long ago stopped being daily-driveable on desktops/laptops, imho
<psykose>
freebsd supports wayland
<psykose>
although i'm not sure why it matters if literally every platform supports something or not
<dottedmag>
That's the same as "freebsd supports HTTP", right?
<dottedmag>
Though there are a few quirks in protocol that are kinda linux-ish
<psykose>
it's the same as "you can run sway on freebsd"
<dottedmag>
yeah, this is more it
<psykose>
personally though i'm very tired of the whole "someone made something" -> "your thing doesn't work on AIX so it's garbage"
cylm_ has joined #asahi
<psykose>
can't use anything without someone complaining it doesn't build with gcc3 on minix or whatever the fuck
<dottedmag>
Yeah, requiring portability by default is quite tiring.
cylm has quit [Ping timeout: 480 seconds]
ktz_ has joined #asahi
<chadmed>
i mean in this specific case too the hardware simply isnt designed to fit x11's assumptions about display hardware
<ktz_>
hi marcan: have you ever got wifi function in ap mode?
<chadmed>
its really just serendipitous that x11 is independently dying a slow death
<ktz_>
wish there was a proper wm available in wayland like xmonad or awesome.. having to deal with sway aka i3 is painful
<ktz_>
river could be an alternative but its performance is particularly bad compared to sway, at least on my machine
<ar>
ktz_: dwl or hyprland?
<j`ey_>
ktz_: what makes sway not proper?
atodorov has quit [Ping timeout: 480 seconds]
<ktz_>
lacking proper layout support and having scripting it via swaymsg ipc mostly
Zopolis4_ has joined #asahi
<ktz_>
ar: I'll check dwl for some reason I didn't look into it
bpye has joined #asahi
<ktz_>
ar: maybe hyprland, I remember checking on it twice but can't recall what didn't stick about it
<ktz_>
but overall I've got used to it, almost got everything I wanted sorted, except the layouts which is a big part j`ey_:
<j`ey_>
dwl could be neat though, maybe I'll give that a try at some point, used dwm for quite a while
<marcan>
dottedmag: the bigger problem for XWayland isn't the keycode right now, it's the lack of a free modifier number to use
<marcan>
Wayland has 32 modifiers, X11 has 8
<marcan>
ktz_: wifi AP mode is known not to work, it's on my long tail list of stuff
<marcan>
I appreciate any assistance with the broadcom driver because honestly I'm a bit tired of just how neglected it is and that I keep having to do broadcom's work for them and that their sole engineer putting some time into Linux is rarely very helpful on the ML
<bluetail4>
ML?
<chadmed>
mailing list
<bluetail4>
thanks
Zopolis4_ has quit []
possiblemeatball has joined #asahi
minecrell has quit [Read error: Connection timed out]
minecrell has joined #asahi
Raito_Bezarius has quit []
Luj has quit [Ping timeout: 480 seconds]
zerdox has joined #asahi
<kit_ty_kate>
mps: are you using MOZ_ENABLE_WAYLAND=1 ?
<mps>
kit_ty_kate: no, is this solution for it
<zerdox>
hi everyone! i've compiled and installed asahi kernel with rust, drm asahi and drm apple so i can get working graphics. but when i boot into it, while initializing (even before i get login prompt) my screen turns off, no backlight. system is running ok and i can get into it with ssh or by setting nomodeset option. so how can i get and see logs where info about this modules will be covered? dmesg and asahi-diagnose looks ok. i have also checked backlight level
<zerdox>
brightness level is ok
<kit_ty_kate>
mps: for the scrolling weirdness yeah i believe, it also smooths fonts used
<mps>
kit_ty_kate: yes, this fixes scroll. thanks
jacksonchen666 has joined #asahi
<marcan>
zerdox: make sure to set CONFIG_BACKLIGHT_GPIO=n
<marcan>
and CONFIG_DRM_SIMPLEDRM_BACKLIGHT=n
<zerdox>
sure. i'll check it rn. thanks
<marcan>
the GPIO backlight hack is mutually exclusive with drm-apple
<marcan>
(we will eventually remove it from the device trees once drm-apple is in the default kernels, but this is the hack for now)
<zerdox>
is there any starting kernel config for asahi? or list of required options
<zerdox>
marcan: thanks. it does the trick. could you also let me know which opengl version do we use? do i need to compile hyprland for opengl es 2 or 3?
zerdox has quit [Read error: Connection reset by peer]
<marcan>
only gles2 is officially advertised
<chadmed>
zerdox: the kernel config you have installed rn is the default linux-asahi config (other than the options you have now enabled/disabled)
Raito_Bezarius has joined #asahi
j`ey_ is now known as j`ey
Luj has joined #asahi
delsol has joined #asahi
delsol has quit [Remote host closed the connection]
delsol has joined #asahi
delsol has quit [Ping timeout: 480 seconds]
delsol has joined #asahi
gladiac has quit [Quit: k thx bye]
delsol has quit [Remote host closed the connection]
jacksonchen666 has joined #asahi
leviz has joined #asahi
leviz has quit []
leviz has joined #asahi
leviz has quit []
gladiac has joined #asahi
rvalue has quit [Remote host closed the connection]
rvalue has joined #asahi
possiblemeatball has quit [Quit: Quit]
possiblemeatball has joined #asahi
cylm_ has quit [Quit: WeeChat 3.6]
jn has quit [Remote host closed the connection]
jn has joined #asahi
MajorBiscuit has joined #asahi
MajorBiscuit has quit []
MajorBiscuit has joined #asahi
jacksonchen666 has quit [Ping timeout: 480 seconds]
jacksonchen666 has joined #asahi
MajorBiscuit has quit [Ping timeout: 480 seconds]
bps has quit [Ping timeout: 480 seconds]
raveling has quit [Ping timeout: 480 seconds]
<ayke_>
marcan: will the F keys remain functional in TTYs?
delsol has joined #asahi
jacksonchen666 has quit [Ping timeout: 480 seconds]
raveling has joined #asahi
Misthios has quit [Quit: Misthios]
Misthios has joined #asahi
raveling has quit [Ping timeout: 480 seconds]
<jannau>
ayke_: yes, the F keys will be default binding and the media keys require fn
A_L_I_C_E has quit [Remote host closed the connection]
jacksonchen666 has joined #asahi
raveling has joined #asahi
MajorBiscuit has joined #asahi
jacksonchen666 has quit [Quit: WeeChat 3.8]
farghal has joined #asahi
lawrence8 has joined #asahi
raveling has quit [Ping timeout: 480 seconds]
Misthios has quit [Quit: Misthios]
Misthios has joined #asahi
<marcan>
more like it'll probably be media keys by default in the XKB keymap (with a default-off option to switch to F keys by default) but the physical keycodes are still Fx at the driver level and will work as such on the TTY
<mps>
iiuc that is how media keys works on chromebooks in linux
A_L_I_C_E has joined #asahi
bakk has joined #asahi
<mps>
i.e. as 'normal' F keys
<marcan>
the hardware sends F keys at the HID level
lawrence has quit [Ping timeout: 480 seconds]
lawrence8 is now known as lawrence
A_L_I_C_E has quit [Read error: Connection reset by peer]
A_L_I_C_E has joined #asahi
raveling has joined #asahi
<bakk>
Is there anything that would prevent DisplayLink from working? I decided to try DisplayLink with my dock (according to the Arch wiki) but can't get an output. Maybe someone has tried already even? Both evdi and udl seem to load fine and the displaylink service runs. I don't see anything particularly suspicious in dmesg. I'm not seeing a new frame buffer device in /dev though. Could of course be that displaylink is a mess in general on Linux, but I
<bakk>
don't know. Would be cool to get it working, despite its flaws.
<mps>
fn+NumberKeys to emulate F keys will stay on machines with touchbar?
raveling has quit [Ping timeout: 480 seconds]
leviz has joined #asahi
leviz has quit []
farghal1 has joined #asahi
farghal has quit [Read error: Connection reset by peer]
farghal1 is now known as farghal
A_L_I_C_E has quit [Quit: Quit]
A_L_I_C_E has joined #asahi
farghal has quit [Remote host closed the connection]
farghal has joined #asahi
farghal has quit []
raveling has joined #asahi
farghal has joined #asahi
farghal has quit [Quit: farghal]
raveling has quit [Ping timeout: 480 seconds]
farghal has joined #asahi
possiblemeatball has quit [Quit: Quit]
kettenis has quit [Ping timeout: 480 seconds]
kettenis has joined #asahi
farghal has quit [Ping timeout: 480 seconds]
raveling has joined #asahi
<marcan>
mps: until the touchbar is supported at least
<mps>
marcan: nice to hear. thanks
avi has joined #asahi
<marcan>
after that the default might change not to do that, but by then there should be some uinput thing taking care of the touchbar
<ChaosPrincess>
marcan: nah, we keeping f-keys emulation even w/ touchbar, cause we still need them to work in non-ui consoles
avi has quit []
MajorBiscuit has quit [Quit: WeeChat 3.6]
<jannau>
the touchbar daemon should work for non-ui consoles as well
raveling has quit [Ping timeout: 480 seconds]
<ChaosPrincess>
has anyone written a touchbar daemon yet?
<jannau>
no, I'd start with https://crates.io/crates/evdev-rs, it should grab the touchbar touchscreen exclusively and use uinput to generate key events
<jannau>
that would cover the dumb F key only touchbar daemon
<jannau>
and of course display a static image of F keys
A_L_I_C_E has quit [Read error: Connection reset by peer]