user982492 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Candygoblen123[m]>
<Glanzmann> "Anyone else who knows macos..." <- tbh it's prob time machine, i use "tmutil thinlocalsnapshots / 999999999999999 4" to clear them
<Candygoblen123[m]>
* it's prob just time machine,
<Candygoblen123[m]>
even one snapshot can take up a lot of space
<kode54>
snapshots are just deltas, but if you do a lot of data modification between snapshots, they can take up a lot of space, yes
<Candygoblen123[m]>
When i installed, i cleared them all, stepped away for about 5 minutes, came back to install Asahi, and had to clear them again because Time machine had created another one
a1eaiactaest has quit [Ping timeout: 480 seconds]
<kode54>
turn off time machine, then
<kode54>
it creates snapshots regularly
<Candygoblen123[m]>
why?
<kode54>
well, don't if you don't have a problem with this
<kode54>
if it's not messing up your space issues, then don't turn it off
dabaiste^ has joined #asahi
<Candygoblen123[m]>
kode54: but yeah, don't tell people to turn off time machine. Usually it's the only thing between them and losing their data
a1eaiactaest has joined #asahi
Tano70 has quit [Quit: Leaving]
<kode54>
aha
<kode54>
but it's not much of a backup if it's only on the one machine
<Candygoblen123[m]>
it's on an external drive
<Candygoblen123[m]>
are you sure you know what you're talking about
<kode54>
no, I never know what I'm talking about
<nicolas17>
I think it creates a snapshot on the local disk
<nicolas17>
then copies that snapshot to the external backup disk
<nicolas17>
that way it gets a consistent copy even if there's programs modifying files while the backup is in progress
<Candygoblen123[m]>
it will create a snapshot when it begins backing up to the drive as well
<kode54>
my only backup is iCloud Documents
<nicolas17>
that's sync, not backup
<kode54>
and I look at massive data loss as an opportunity to start hoarding all over again
<kode54>
if anything, I hoard way too much
<kode54>
all my source code lives in public Git repositories, the rest is old garbage I stopped caring about ages ago
<kode54>
my photos live in iCloud Photos, nearing 30GB, and I hear that's a tiny figure
<kode54>
(I knew someone who already blew through an entire 2TB iCloud account just filling their photos library with raws)
a1eaiactaest has quit [Ping timeout: 480 seconds]
<kode54>
and yes, they absolutely need to carry that 2TB of raws with them wherever they go
<kode54>
guess the 8TB Macs were made for them
a1eaiactaest has joined #asahi
chadmed has joined #asahi
kjm99 has quit []
a1eaiactaest has quit [Ping timeout: 480 seconds]
a1eaiactaest has joined #asahi
user982492 has joined #asahi
ciggi_ has joined #asahi
a1eaiactaest has quit [Ping timeout: 480 seconds]
<kode54>
sorry for ranting
ciggi has quit [Read error: No route to host]
dabaiste^ has quit [Remote host closed the connection]
a1eaiactaest has joined #asahi
bmentink[m] has joined #asahi
<bmentink[m]>
hi guys, I have just done a fresh install on my Macbook air, booted into Arch fine, but can't setup WiFi, when I try iwctl to set it up, it just sits there saying "Waiting for IWD to start"
yuyichao_ has quit [Ping timeout: 480 seconds]
a1eaiactaest has quit [Ping timeout: 480 seconds]
amarioguy has quit [Ping timeout: 480 seconds]
<bmentink[m]>
How do I get network? I can't use ethernet over USB3, as it's not seen ..
<AntoniosPapadakis[m]>
if by Arch you mean Asahi Linux on an M1 [Pro|Max] chip, then WiFi should work. Try fiddling around with WiFi on|off switch. It sometimes gets stuck. If that doesn't work, a reboot should sort out the issue
<AntoniosPapadakis[m]>
the Mac Studio (and by extension the M1 Ultra chip) is not supported
<milek7>
bmentink[m]: you need to start iwd service first
<bmentink[m]>
I can see a wlan0 device ... I have a Macbook Air 2 years old
<bmentink[m]>
how do I start iwd service again?
<milek7>
systemctl start iwd
<bmentink[m]>
ahhh got it thanks
a1eaiactaest has joined #asahi
Aiden[m] has joined #asahi
<Aiden[m]>
hi folks, I wanted to ask what the thoughts of the battery life of M1 airs running asahi's first release are
<Aiden[m]>
I'm willing to stomach potential difficulty in finding supported software given how solid the battery life of said computers seem to be
<ReaganMcFarland[m]>
It’s def not as much as normal macOS
<ReaganMcFarland[m]>
But I think I get like 6-7 hours on my m1 MacBook Pro 2020
<ReaganMcFarland[m]>
I could be off
<Aiden[m]>
that's pretty exceptional for a linux laptop (in my experience)
<ReaganMcFarland[m]>
I get like 16 on macOS
<ReaganMcFarland[m]>
Sometimes more
<ReaganMcFarland[m]>
Depending on my workloads
<Aiden[m]>
is that doing development work?
<ReaganMcFarland[m]>
Vim / rustc
<Aiden[m]>
ah okay
a1eaiactaest has quit [Ping timeout: 480 seconds]
<ReaganMcFarland[m]>
I use a VDI for work - I can get almost 11-12 hours on that with battery
<Aiden[m]>
VDI?
<ReaganMcFarland[m]>
Virtual desktop interface
<Aiden[m]>
oh, through asahi?
<Aiden[m]>
viewing some virtual desktop on a separate machine?
<ReaganMcFarland[m]>
No for macOS but I was just saying how much battery I get on a normal day to day
<Aiden[m]>
ah okay, I see
<ReaganMcFarland[m]>
I would say asahi is considerably less than macOS in terms of battery life
<bmentink[m]>
Can't install xorg, get error retrieving harfbuzz ... any idea's
<ReaganMcFarland[m]>
But if your workloads aren’t super intense it should be fine
<Aiden[m]>
6-7 hours is still very nice
<Aiden[m]>
I get more like 2 on my acer aspire
<ReaganMcFarland[m]>
Lmfao that’s awful
<ReaganMcFarland[m]>
I can’t imagine going back to intel
<ReaganMcFarland[m]>
I miss Linux so much tho - Glad asahi is around :)
<Aiden[m]>
yeah... I will upgrade at some point, but engineering school tuition demands that I wait until my aspire cacks to purchase a new computer
<Aiden[m]>
and yeah, I have tried macos a bit, but really prefer using linux
<bmentink[m]>
Anyone?
<Aiden[m]>
where are you trying to install Xorg from?
<Aiden[m]>
some repository?
<bmentink[m]>
pacman
<Aiden[m]>
or compiling it yourself?
<Aiden[m]>
ah I see
<Aiden[m]>
error message?
<bmentink[m]>
I have done the basic install as I want gnome
<bmentink[m]>
error message is "failed to retrieve 'harfbuzz-4.0.1-1 ... etc
<chadmed>
yeah... whats the etc bit...
<chadmed>
you already told us it failed to get it, why does it say this is the case
<bmentink[m]>
ok, just a second slow typer
<bmentink[m]>
harfbuzz-4.0.1-1-aarch64.pkg.tar.xz from mirror.archlinuxarm.org .. error 404
<kode54>
don't forget to Sy first
<bmentink[m]>
did that
<kode54>
always sync and full upgrade
<kode54>
huh
<kode54>
switch mirrors then?
<bmentink[m]>
did pacman -Syu
<kode54>
try a different mirror
<bmentink[m]>
what mirrors should I be using, not the install ones?
<kode54>
well, for the base mirror list, there's all the alarm mirrors, pick one close to you
<kode54>
rank mirrors them if you feel like spending bandwidth
a1eaiactaest has joined #asahi
<kode54>
*rankmirrors
<bmentink[m]>
ok will try
<kode54>
don't touch the Asahi mirrorlist, because it needs a server that has the Asahi packages
<kode54>
unless there are multiple mirrors there too, I dunno, I haven't looked at it yet
<kode54>
I only set my mirror to the Chicaco, IL mirror
<kode54>
the mirrorlist file mentions a California mirror, but that repo doesn't mirror alarm any more?
<bmentink[m]>
australia mirror did not help ... changed it and did a pacman -syyu
<kode54>
current harfbuzz is 4.1.0-1
<kode54>
at least on mainline arch
<bmentink[m]>
I will try the Chicargo one
<kode54>
I'm not in Asahi right now, so I can't tell you what's in the repo
<chadmed>
if its trying to get 4.0.1 its clearly not synced up
<chadmed>
do pacman -Syyu to force it to update repos
<bmentink[m]>
see above .. I did
<kode54>
annoying issue I have with the Byobu update indicator
<kode54>
if you update your repo listings after installing updates
<kode54>
but the repo index is older than the last update check timestamp, then it won't remove the count
<bmentink[m]>
huh? every mirror I try returns the same error
a1eaiactaest has quit [Ping timeout: 480 seconds]
<bmentink[m]>
So I can't continue the install, until I get some help here ..
<Taro[m]>
im trying to install asahi with the command, but it stops after it downloads and doesnt continue to extracting nor initializing
DragoonAethis has quit [Quit: hej-hej!]
DragoonAethis has joined #asahi
<kode54>
bmentink[m]: none of the mirrors have a 4.0.1, because that's not the current version
<kode54>
your system should be asking for 4.1.0
<bmentink[m]>
well I have just done a fresh install and it is not asking for 4.1.0
<kode54>
and which user are you running pacman with?
<bmentink[m]>
root
<kode54>
weird
<bmentink[m]>
yep :(
<bmentink[m]>
has anyone tried the basic install here ..
yuyichao_ has joined #asahi
<kode54>
I did
<kode54>
but I haven't updated it since 4.1.0 was released today
<kode54>
I've been in macOS for over 24 hours
<kode54>
I've also done evil things in my setup
<bmentink[m]>
would you be able to try an "upgrade" and report back?
<kode54>
I downloaded jemalloc from ABS, patched the PKGBUILD for aarch64 (just needed the arch defined)
<kode54>
I'll do that next time I'm in Asahi
<kode54>
I'm kind of in a video call right now
<bmentink[m]>
ok cool ..
<kode54>
I could do the Discord portion of the call from Linux, since that runs in Firefox
<kode54>
but I normally do the video over Skype, ever since macOS 12.3 came out
<kode54>
12.3 has a bug dealing with webcam formats other than raw YUV
sentient_lilac[m] has joined #asahi
<kode54>
somehow, apps don't see or refuse to touch the MJPEG or H.264 formats now
<kode54>
Discord behaves with this by picking a high res, low frame rate raw format
<kode54>
Skype just picks the highest raw it can do at 30fps
<kode54>
I managed to get Telegram Desktop working with that jemalloc rebuild hack
<kode54>
"hack" because upstream alarm builders don't build on 16k or 64k machines, and jemalloc detects the page size of the building machine, and only supports smaller sizes than it was built with, not larger
<kode54>
this fix works for the few apps that import dynamic jemalloc from the jemalloc package, but not for the huge wealth of apps that static link it
<kode54>
like everything rust
<kode54>
rust needs rustc to bootstrap itself, but the stock rustc is built for 4k pages
<chadmed>
ive been trying to hack the jemalloc that comes with fex to at least allow me to do SOMETHING with 16k pages
<kode54>
in case you may have missed it from a day or two ago
<kode54>
x64box posted a blog entry about getting their 16k page hack working
<enchiladasconpinguino[m]>
so possible DCP bug on M1 Mini: HDMI signal is only negoiated once on boot?
<enchiladasconpinguino[m]>
if I turn the monitor power off, display signal never comes back on. I have to then reboot the system remotely
<enchiladasconpinguino[m]>
can anyone else with Mini connected to HDMI display confirm? is there a workaround?
<enchiladasconpinguino[m]>
also no DPMS control, even though `xset q` show it is enabled. Which is most likely this same bug
<kode54>
yup, that's correct :[
<kode54>
m1n1 sets up the framebuffer on boot
<enchiladasconpinguino[m]>
so initial iBoot HDMI negotiation is all that happens, Linux driver can't do it? is that it?
<chadmed>
yeah box64 has worked on 16k pages for a while
<kode54>
prior to that, it simply inherited the boot framebuffer from the bootloader
<kode54>
no
<enchiladasconpinguino[m]>
oh so m1n1 can but Linux can't ?
<kode54>
it needs to set that up now
<enchiladasconpinguino[m]>
that's better than iBoot
<chadmed>
i think fex is faster for certain floating point codepaths which is why i want it up and running as well
<kode54>
as of macOS 12.0, iBoot doesn't even init a boot framebuffer
<tpw_rules>
i don't think there is DCP yet
<enchiladasconpinguino[m]>
yup
<tpw_rules>
so there's no bug to be had. that's just observed behavior
<kode54>
FEX is getting faster, yes
<kode54>
for instance, they're working on doing 64 bit x87 as scalar neon instead of full softfpu
<enchiladasconpinguino[m]>
well its certainly bug, that you can hang your system that easily .. and have to leave monitor on running
<kode54>
Linux has no DCP support yet
<tpw_rules>
you're lucky you have a monitor that m1n1 can init successfully
<kode54>
I have a 1080p monitor on my HDMI
<kode54>
lucky it didn't have my previous setup
<kode54>
where I had the HDMI hooked up to my 4K monitor
<enchiladasconpinguino[m]>
I haven't tried a newer monitor yet
<enchiladasconpinguino[m]>
same, its an old monitor .. probably why it works
Kevin5 has joined #asahi
<kode54>
m1n1 currently ignores displays over a certain resolution, because of behavior with monitors that are single DPI but lie about their resolution in the EDID
<chadmed>
yeah that x87 optimisation actually matters to me because of goldsrc
<kode54>
for some reason, it was decided that initializing a monitor that high would be detrimental rather than wanted
<enchiladasconpinguino[m]>
and DisplayPort USB-C port isn't working yet either right? almost certainly answer is no, just checking
<kode54>
correct
<kode54>
DisplayPort is not implemented yet
<chadmed>
yeah dp alt mode does not work yet
<enchiladasconpinguino[m]>
yeah figured, v complex
<chadmed>
a few things need to fall into place for that
<kode54>
single display is one of the major reasons I don't like to try to spend my day in it
<kode54>
also lack of audio from the headphone jack
<enchiladasconpinguino[m]>
I guess I can boot it headless without HDMI plugged in at all? reboot for GUI when needed .. or else leave a gentle screensaver going for poor old monitor
<nicolas17>
enchiladasconpinguino[m]: afaik the public kernel has no DCP driver at all
<kode54>
every time I boot it, I have to move my headphone transmitter cable to my bluetooth microphone
<nicolas17>
oh I was scrolled up >.>
<enchiladasconpinguino[m]>
I remember seeing some work on it, but maybe its not merged into alpha release at all hrmm not sure
<kode54>
enchiladasconpinguino[m]: you can use a screen saver like gnome screensaver that just blanks the screen
<kode54>
or something equivalent
<enchiladasconpinguino[m]>
doing that already yeah, black at 100% brightness
<kode54>
oops
<enchiladasconpinguino[m]>
I think slowly changing image is actually healthier for the screen
<kode54>
you could look into installing xscreensaver
<kode54>
and disable all the GL ones
<enchiladasconpinguino[m]>
yup already using that
<nicolas17>
oh yeah there was some DCP development work already, but it's not on the alpha release
<kode54>
one of my favorites for complexity is the old computer ones
<kode54>
a full NTSC simulator
<kode54>
complete with line noise
<enchiladasconpinguino[m]>
actually using XFCE screensaver sorry, not the old jwz one
<kode54>
and then Apple II graphics simulation
<kode54>
oh
<kode54>
I always forget that thing's maintained by jwz
<kode54>
hi, why does the security model need to be implemented by the screen saver itself
<kode54>
shouldn't that be running in a sandbox, and the screen locker take care of protecting the system?
<nicolas17>
the "screen locker" being a separate process is also a problem
<enchiladasconpinguino[m]>
lol situation is slightly better on wayland, at least depending on compositor
<nicolas17>
I have seen bugs where people crash the screen locker with the login screen and fall right back on the user's desktop
<nicolas17>
wayland fixes it yeah
<kode54>
I use Wayfire on my Risen desktop
<kode54>
damnit
<kode54>
Ryzen
<enchiladasconpinguino[m]>
DCP driver isn't working .. because its not in this kernel at all yet, just framebuffer from m1n1
<enchiladasconpinguino[m]>
okay thanks for confirming the screen issue
<kode54>
hell with it, I'm turning off autocorrect
<enchiladasconpinguino[m]>
I thought / hoped it was an alpine specific issue heh
<enchiladasconpinguino[m]>
drm vblank appears to be present in the driver already :)
skipwich has quit []
<chadmed>
and for their next trick, they will implement VRR! (pretty please :-P)
<Glanzmann>
Cy8aer[m]: The arch was wrong. The line needs to be: debootstrap --arch=arm64 testing /mnt http://deb.debian.org/debian
<chadmed>
it will need to come eventually i think, its an enormous battery saving feature
creechy has quit [Ping timeout: 480 seconds]
kov has quit [Quit: Coyote finally caught me]
kov has joined #asahi
skipwich has joined #asahi
skipwich has quit []
skipwich has joined #asahi
creechy has joined #asahi
creechy has quit [Ping timeout: 480 seconds]
<Namidairo>
1600 nit backlight is what the people desire, for outdoor eyeball sizzling
<Namidairo>
although I think the display has thermal limits to stop you cooking it too fast
creechy has joined #asahi
a1eaiactaest has quit [Ping timeout: 480 seconds]
creechy has quit [Ping timeout: 480 seconds]
a1eaiactaest has joined #asahi
marvin24_ has joined #asahi
a1eaiactaest has quit [Ping timeout: 480 seconds]
marvin24 has quit [Ping timeout: 480 seconds]
creechy has joined #asahi
a1eaiactaest has joined #asahi
creechy has quit [Ping timeout: 480 seconds]
a1eaiactaest has quit [Ping timeout: 480 seconds]
creechy has joined #asahi
a1eaiactaest has joined #asahi
KDDLB has quit [Remote host closed the connection]
kode54 has quit [Remote host closed the connection]
KDDLB has joined #asahi
creechy has quit [Ping timeout: 480 seconds]
creechy has joined #asahi
<Ry_Darcy>
@Glanzmann Yesterday evening, tried to to bring up the M1 Mini on a 2K monitor in ZH. Again, a black screen. At this stage, I am convinced that it is the boot chainloading/DCP process that is very particular to which type of Monitor it will talk to.
<Ry_Darcy>
@Glanzmann This afternoon, I am in BE and will try again on the 2K monitor to apply your patch. The 2K monitor in BE worked last weekend. Having done that, I will have to wait until I am in ZH (tomorrow afternoon) to try your patch on my 4K monitor. This is not easy....
<EvadingEvasions[m]>
What are alternatives to firefox which run fine on Asahi?
creechy has quit [Ping timeout: 480 seconds]
<Glanzmann>
Ry_Darcy: I see. Jannau and Alyssa also working on the dcp driver. Maybe we have something we can use soon.
<Glanzmann>
Ry_Darcy: Anyhow, keep me posted.
<Ry_Darcy>
@Glanzmann Will do. Very frustrating nevertheless. Not your fault though. Additional information. Through some magical process/act of god, I have managed to upgrade the M1 Mini to 12.3, fwiw.
<Ry_Darcy>
@Glanzmann I also ask myself, how did Pipcet (who I think worked on his own) manage to get "his" Linux to work with various monitors and resolutions?
doggkruse has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
ptudor_ has joined #asahi
a1eaiactaest has quit [Ping timeout: 480 seconds]
ptudor has quit [Ping timeout: 480 seconds]
creechy has joined #asahi
<j`ey>
Ry_Darcy: he took a copy of the DCP driver and extended it
kode54 has joined #asahi
<Ry_Darcy>
@j'ey He should be working with Asahi then... :)
a1eaiactaest has joined #asahi
<kode54>
verified, it's installing 4.1.0 here
dandiccf[m] has joined #asahi
<Glanzmann>
Ry_Darcy: He did in the beginning but than there was diasgreement and he did his own thing.
<Glanzmann>
kode54: Context?
<kode54>
Glanzmann: bmentink[m] above was having an issue where no matter which mirror they picked, it would try to download harfbuzz 4.0.1-1
<kode54>
the only thing I can figure is that somehow, they have read-only repo metadata on their system
user982492 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
creechy has quit [Ping timeout: 480 seconds]
a1eaiactaest has quit [Ping timeout: 480 seconds]
<Ry_Darcy>
@Glanzmann thanks for the info.
a1eaiactaest has joined #asahi
<marcan>
Ry_Darcy: m1n1 depends on the DCP iBoot interface for early boot bring-up and that interface is very limiting
<marcan>
iBoot had problems with some monitors too
<marcan>
we still should work on trying to make the best of it because without it there is no GRUB/U-Boot menu even if the kernel works with DCP
a1eaiactaest has quit [Ping timeout: 480 seconds]
<marcan>
but there's only so much we can do short of implementing the full DCP interface in m1n1, which, if we end up having to go there, will have to wait because that would *suck*
<chadmed>
we should simply force monitor manufacturers to properly implement their capability advertising :-P
a1eaiactaest has joined #asahi
<Ry_Darcy>
Appreciate all of your herculean efforts. Tis however frustrating..
<chadmed>
do we know how widespread that "fake 4K" problem is? i'd be inclined to just tell people to cop it for the sake of m1n1's simplicity and ensuring that "most" people without that problem get a flicker-free native res boot experience
creechy has joined #asahi
<chadmed>
like if the only other two options are "implement all of DCP" and "assume X*Y res framebuffer" then neither are really that good. the former would just suck soooo hard to do and the latter would still bork compatibility with a bunch of monitors unless you went for like 800x600 as the assumed res
<chadmed>
and you still have to tear all that down once the kernel loads its own dcp driver so you get flickering and black screens and all sorts of nastiness people usually associate with nvidia proprietary drivers (not something we should associate ourselves with)
<Ry_Darcy>
I am lucky if the bloody thing shows up after initial boot at all - whatever manufacturer/screen resolution.
guillaume_g has joined #asahi
<marcan>
chadmed: the "fake 4K" problem is universal and DCP's fault
<marcan>
I get fake 4K on my capture cards
<marcan>
it's not the monitor
<marcan>
it's DCP making up modes
<chadmed>
oh right right i have misread then, i thought it was monitor firmware reporting a 4K mode to the DCP. apologies
<chadmed>
i actually had a monitor years ago htat did this which is probably why i got confused
<marcan>
anyway, now that we have chainloading, the real answer is obvious: add 10 lines of code to m1n1 and 5 lines of code to update-m1n1.sh so that people can do echo 'display=3840x2160@60' >> /etc/m1n1.conf and pick whatever mode they want
<marcan>
chainloading FTW
<marcan>
plus that will force stage2 to re-init the display; we can add a "display=auto" option which will re-probe the display in stage2, which will likely fix flakiness issues for everyone; we can even add a configurable timeout for robustness
<dottedmag>
marcan: I don't even see raw EDID in ioreg under macOS: does DCP expose it at all?
creechy has quit [Ping timeout: 480 seconds]
<marcan>
and then people who don't mind the extra delay can just do that
<marcan>
dottedmag: not via the main interface, it does not
<dottedmag>
ugh
<marcan>
however there is some i2c passthrough stuff that I think has it
<marcan>
in the other endpoints (that the linux driver does not support/use)
<marcan>
but at least the main DCP interface has a *lot* of information per-mode, plus scores
<marcan>
that the iBoot interface doesn't
<dottedmag>
Just curious: have you seen DCP handling tiled monitors? Are they reported as two outputs with some metadata, or DCP merges them internally and handles refresh synchronization internally?
<marcan>
no idea but I assume the latter
<marcan>
would be silly to punt that to macos
<kode54>
the Retina 5k iMac punted it to Windows when they finally supported it at native 5k there
<marcan>
yeah, tell me about it
<marcan>
eDP connected primary 3840x2160+0+0 (normal left inverted right x axis y axis) 597mm x 336mm
<marcan>
3840x2160 60.00*+
<kode54>
it was funny seeing the first setup wizard's fading background fading out of sync in the middle
<dottedmag>
Hah, if this is so, then X / Wayland compositors will work better with tiled monitors under M1 than under x86
<marcan>
still stuck at 4K because Linux doesn't support it
<kode54>
yeah, I hate that
<marcan>
dottedmag: everything graphics will work better under M1 than under x86 because x86 display output hardware kind of sucks, and drivers suck even more
<marcan>
once we have the pieces in place
<kode54>
I eventually sold my heatwell iMac for a fourth of what I paid for it
<marcan>
especially wayland since that has a chance at being as smooth as macOS with its design
<marcan>
OTOH, wayland also has problems with DPI scaling still so I dunno
<marcan>
I tried it earlier on asahi and wasn't impressed
<marcan>
it even seemed slower than xorg
<dottedmag>
marcan: You sure know Wayland is a protocol, not software :-P What compositor have you tried? Kwin?
<marcan>
yes, but also specifically xwayland has DPI problems and this seems to be a longstanding issue
<marcan>
"firefox is blurry as heck" is not a good first impression
<dottedmag>
Yeah, Xwayland is a pain.
<chadmed>
its slower and fractional scaling still causes stuff to become ugly
<chadmed>
like if you log out and log back in the obvious aliasing disappears, but everything kinda just looks smeared
<marcan>
and yeah I noticed it was being "clever" with scaling and just resizing textures instantly when changed on the fly
<jannau>
scaling with kwin_wayland works well with dcp
<marcan>
which is cute and all but if apps don't support on the fly DPI switching you still need to restart them for stuff to become crisp
<marcan>
which is ugh
<jannau>
shouldn't firefox nowadays default to wnative wayland
<marcan>
I dunno, maybe the default app link starts the xorg mode
<chadmed>
yeah but im still liking it better than xorg on this machine
<marcan>
I'm sure this all can be improved
<chadmed>
you need to add MOZ_ENABLE_WAYLAND=1 to /etc/env
<marcan>
but it will take some time
<marcan>
the default xorg experience was much better than the default wayland experience so xorg it was for the release
<chadmed>
firefox does not default to wayland ever on arch without that environment variable, you can make it do so in gentoo by adding the wayland USE flag though
<marcan>
I couldn't even figure out how to set libinput configs by default for wayland, it seems it's punted to the compositor there so I guess it would've been in kdeglobals at that point...
<marcan>
I did eventually figure out where to put those systemwide but by then we were hours from shipping :p
<dottedmag>
Nearly everything is punted to compositor in Wayland. It's "no policy in protocol" at the extreme
<marcan>
yeah
<marcan>
but then libinput should've had config files for this so all compositors can share defaults
<marcan>
s/wayland/wayland ecosystem/ if you will
<marcan>
there still needs to be a way to set a systemwide default keyboard layout for sanity
<dottedmag>
Not much ecosystem, alas. I see the attempts at this, but they are painfully slow.
<marcan>
and this is why wayland still hasn't taken over :-)
<marcan>
anyway, we'll work something out, in time
<dottedmag>
True. Everything is solvable, there is even a protocol for better handling of fractional scaling in works. But it will take enormous amount of time.
<marcan>
I promised "linux works at least as well as it does on every other machine", not "linux works as well as macos at everything", so I'm not going to be holding myself to that standard
<marcan>
but I will take what I can get to make things work well by default on these machines
<marcan>
can't fix all the linux ecosystem's problems single-handed though ;)
<marcan>
for now, let's fix mac keyboard layouts
<marcan>
that one's inexcusable
<marcan>
still can't believe apparently eisu and kana have been swapped on all mac JP keyboard layouts since the dawn of time
<marcan>
it seems nobody maintains any of this seriously
<marcan>
so that's where we can help :)
<dottedmag>
marcan: How do I manually select these keyboard layouts you have mapping for in installer? I'd like to check ru and ua layouts to see if they are any good.
<chadmed>
mac support has been dreadful since forever as far as i know
creechy has joined #asahi
<marcan>
kde settings lets you change the layout (and also add options which the installer does not)
<dottedmag>
Because I feel you'll need en+ua / en+ru instead of plain ua / ru
<Chainfire>
So eh, I allocated 70GB for Asahi initially. I'm running out of space already. What are the steps to enlarge?
<marcan>
uh... good luck, enlarging backwards is always a pain
<marcan>
shrinking macOS more is easy, moving Linux upwards into the space is not so easy
<Chainfire>
so un/reinstall would be easier?
<marcan>
yeah, unfortunately
<marcan>
I wonder if there is even some diskutil magic incantation we can use to make that possible at all
<rkjnsn>
If you can get a live media to boot via U-Boot, gparted makes the actual moving / resizing pretty easy.
<marcan>
yeah, if you use a live linux image or whatever you can do it
<marcan>
honestly we should offer one at some point, "Linux Recovery" or whatever
<marcan>
something bootable from u-boot directly
<marcan>
but that's not really a thing yet for us
<marcan>
even something with busybox and parted and a couple other tools would do the trick
<Chainfire>
Oh, what you said made me think some weirdness was going on, but if I manage to shrink macOS, and was able to run gparted, that would "just work"
coderush has joined #asahi
<marcan>
yes, the problem is you need to do it with your root filesystem unmounted
<marcan>
which means rescue media
<Chainfire>
Yeah, I understand
<Glanzmann>
Chainfire: you can decrease macos, boot debian live, apt-get install -y xinit blackbox xterm gparted
<marcan>
and it'll have to copy all your partition data backwards, so it won't be instant
<Glanzmann>
Chainfire: Than start gparted and move the stub and esp partition to the end of the macos. And than resize your root filesystem. I have done it many times.
<marcan>
but yeah, what Glanzmann said should work
<Glanzmann>
marcan: The nvme on these machines is so fast that something that I would have assumed take 10 minutes was like 45 secs.
<marcan>
if you have an existing install and the DTBs are compatible, I think you can interrupt u-boot and type `run distro_bootcmd` to boot from USB
<marcan>
Glanzmann: heh
<EvadingEvasions[m]>
When is the chromium patch coming? I can't seem to find it.
<Glanzmann>
marcan: Yes, that should work. I always use 'run bootcmd_usb0' but both should work.
<EvadingEvasions[m]>
If that gets patched? What about electron applications?
<marcan>
oh, it's in review already?
<marcan>
electron will have to pull it downstream of course
<chadmed>
electron applications would pick it up when electron pull it in
<EvadingEvasions[m]>
So that means going to latest electron.
<chadmed>
yeah
<EvadingEvasions[m]>
Most applications dare venture beyond minor releases.
creechy has quit [Ping timeout: 480 seconds]
<EvadingEvasions[m]>
I see VS-Code suffering from this issue as well no?
<chadmed>
well maybe reevaluate using electron applications :-P
<marcan>
apple developers managed to port to a whole new architecture, surely electron developers for Linux can update their framework or backport a patch
jonaburg[m] has joined #asahi
<EvadingEvasions[m]>
marcan: Many are not interested in general.
<marcan>
I'm not going to be rushing to offer 4K kernels just because people are lazy
<EvadingEvasions[m]>
Upgrading and testing breaking changes, surely electron will also take time to update chromium.
<marcan>
they can fix their damn software; 16K is better for anyone who doesn't need FEX
<chadmed>
they use electron in the first place because it allows them to do the bare minimum work possible to push a viable gui application out
<EvadingEvasions[m]>
Yet they disregard us off as 'niche' population and 'we don't care'
<chadmed>
theres no chance in hell most electron app devs are going to bother working in patches to support anything but the bare minimum machines
<chadmed>
and as marcan said, it should not be our job to compensate for this
<EvadingEvasions[m]>
Indeed.
<chadmed>
i think until these devs get a grip youll just have to use alternatives
<EvadingEvasions[m]>
Let us hope some kind soul will go out of his way to compile VS-code.
<EvadingEvasions[m]>
Or one has to do it oneself.
<marcan>
4K/16K/64K have been standard on ARM64 since the dawn of time, therefore software that does not support all of them is buggy
<chadmed>
i mean if people really really really love vscode for some reason they can just use it in the web via github now...
<chadmed>
or just use a good text editor
<EvadingEvasions[m]>
It is not some reason good sir, it is very helpful, especially in a web developer's workflow.
<marcan>
kate works fine, you know
<marcan>
:p
<EvadingEvasions[m]>
Alas I am unable to find a good workflow with Kate.
<EvadingEvasions[m]>
Yet I simply enjoy using Linux, with KDE plasma too. I have loved KDE software since the beginning.
<EvadingEvasions[m]>
I have tried incorporating nvim too. Treesitter CLI does not work because of architectural issues.
<marcan>
coderush: ah right, that's the stuff tying it to firmware version measurements right?
<marcan>
I've been meaning to look into how that works too, yeah
<Chainfire>
Glanzmann: how do I make a debian live usb? Just same instructions as for any other system?
<marcan>
of course we need to be very careful going there since that's how you end up with people losing all their data on a keyboard firmware update (*cough* Microsoft Surface *cough*)
<coderush>
marcan: it's only enforced if FileVault is enabled for any APFS volume group, but if so, it measures the HW state, iBoot1 state, and local policy state, and ties user data access to them all
<marcan>
need to make sure to fully understand it and have a good UX flow for it :)
<coderush>
marcan: the diagram on that page is nice enough in explaining the whole thing
<marcan>
coderush: is it always used behind the scenes for these password-derived keys, or is it something we can opt in/out of depending on our specific usage of SEP?
<coderush>
marcan: there's also a limitation on the amount of SKP measurement slots available, so 20 OSes installed in parallel is a limitation for that part too
<Glanzmann>
Chainfire: 'the wifi firmware thing is no longer necessary it picks it up automatically from the esp. I remove that part in the wiki'
<marcan>
I should probably make my Asahi image builder build a live rescue image too, in that style...
<coderush>
marcan: BootPolicy layer always measures and sends the measurements to lower layers, but enforcement of it all is only enabled if FV is on. I also don't think you can do anything to break the measurements, unless you do something really strage.
<coderush>
marcan: i.e. replace iBoot1 or try to manage local policies from Asahi
<marcan>
not like we could do anything with our own local policy anyway
<coderush>
You could run SEP and talk to it the same way libbootpolicy.dylib does.
jakebot6 has joined #asahi
<marcan>
though I do need to figure out how to trigger SFR upgrades, most I've figured out so far is you copy some stuff to a staging location in Preboot (or system recovery?) and put stuff in NVRAM to make it boot into the upgrade ramdisk
<coderush>
But you'll be deemed fuOS and the only thing you will be able to do is to get back to Full Security
jakebot has quit [Read error: Connection reset by peer]
jakebot6 is now known as jakebot
<marcan>
that would let us get rid of macOS to some extent, though we'd still have to direct people into recoveryOS to complete a stub re-provision and upgrade the per-OS firmware
<marcan>
but we have a good flow for that anyway with the USB installer flag hack, just need them to hold down the power button
<coderush>
Actually, there's a flow for a tethered SFR update.
<marcan>
you mean via DFU?
<coderush>
It can be done via non-erase DFU, yes.
<marcan>
I know about revive mode, yes
<chadmed>
it would indeed be nice to yeet macos if we can get by with managing stuff from 1TR, dfu and linux
<marcan>
but it also punts you into boot recovery assistant IIRC?
<marcan>
wipes all BPs or whatever?
<marcan>
IIRC you said telling people to use that was a bad idea
<marcan>
:)
<coderush>
Indeed it puts you into rOS, but it doesn't wipe anything that had been valid before
<coderush>
Just re-measures it after you provide enough creds to authorize the re-measurement
<marcan>
yeah
<marcan>
so old fuOS BPs would still be valid?
<coderush>
Yep
<coderush>
It is a bad idea in sense "please let macOS do it instead"
<marcan>
yeah, but some people don't want macOS
<marcan>
if you think it's okay to tell people to wire up and use idevicerestore, then we might be able to start offering "replace macOS" flows, at least in expert mode and understanding this
<marcan>
ideally recoveryOS would have some way of doing it for us
<marcan>
and then we just ask it to do so and not have to implement too much
<marcan>
since we do drag around a full recoveryOS after all
<coderush>
Well, replace macOS means no FV, at least before SEP FW will be loaded by Asahi and all that.
<marcan>
right
<coderush>
So you'll likely never need to even trip SKP into being enabled if you don't want macOS
<marcan>
yeah
<marcan>
I'm not in a rush to offer that flow either way, but there's always going to be people who ask and that's why I have expert mode in the installer, that implies some level of "here be dragons"
<coderush>
I understand, and admire their dedication. :)
<marcan>
I'm going to add a "delete partition" option to expert mode anyway just because diskutil sucks and we need a better UI for that, and the only thing stopping users from doing a no-macOS install at that point is something to stash around the SEP user identity list from it *before* nuking it so we can transfer it to linux
<marcan>
I've been thinking of just casually dropping that plist into iSCPreboot on Asahi installs, as a fallback and to enable that use case
<marcan>
thinking of it as "last known primary macOS user list"
<coderush>
Still need to get an owner created before, but yeah, that plist is required for all the stuff that can't just ask SEP for the owner list
<marcan>
yeah, that's what I mean, just boot macos once, create the owner, then boot into 1TR and run the installer and nuke it, and have it save just that plist
<coderush>
I feel kind of sad about the current state of this affair, because it seems stupidly complicated for no apparent reason
Chainfire has quit [Remote host closed the connection]
<coderush>
But you know all the reasons already after 1.5 years in those trenches...
<dottedmag>
marcan: I checked ua/ru layouts, and they definitely need to be fixed. I don't know the syntax though (I've been using my own xkb symbols from keycodes up, so I don't know anymore how rules compose), were should I file the bug?
<jonaburg[m]>
the last time I tried to use a Mac was when I was super impressed by the hardware from the 2014 era MacBook Pro (first 'retina' marketing stuff).. almost a decade later and with the m1 I'm equally impressed by the hardware. can't express how much I love the project and where this is going!
<marcan>
if you can include a *very* detailed description of what the best layout/variant available so far is, and exactly what needs changing and how, that would be very helpful
<jonaburg[m]>
marcan: not sure how to send a private message on this client, but please send me a donation ETH address!
<marcan>
jonaburg[m]: sorry, I don't take crypto donations
<marcan>
coderush: :)
<coderush>
Also, please don't forget about the fact that on Macs RU layout has at least 2 popular variants: usual Russian and Russian - PC created for switchers from PC
<marcan>
just software, or are there two physical variants too?
<jonaburg[m]>
marcan: you gotta be a little crazy to do something so cool I guess ;D
<coderush>
I personally use the latter and would really like to have it as an option.
<dottedmag>
marcan: xkeyboard-config already provides proper maps, it's the installer that selects the non-optimal (awfully non-optimal) variant.
<marcan>
you can just file a bug on the asahi-calamares-configs repo and explain exactly what you want and how it should work
<marcan>
and I'll do my best to implement it
<jonaburg[m]>
I kind of screwed up my partitions a bit btw, I wanted to give macOS a bit more space so I actually just removed the Asahi partition.. used to how things are setup in linux, I was shocked to find out volumes do not function as they do over there :D
Chainfire has joined #asahi
<jonaburg[m]>
now I have one tiny 40gb partition with like 850gb unclaimed space in its own volume.. if I remove that "other volume", then it gives an error that I can't expand the initial due to some internal error
<marcan>
jonaburg[m]: is this using the graphical Disk Utility?
<jonaburg[m]>
yes, correct
<marcan>
yeah, don't do that. that thing is broken as all hell.
<marcan>
I should add a flashing red sign for users
<jonaburg[m]>
looks like it.. am I able to repair it at this point via cli?
<marcan>
it does not work unless all you have is a pile of neatly lined up APFS volumes utilizing all available space
<marcan>
yeah, use the `diskutil` cli for everything
creechy has joined #asahi
<coderush>
marcan: it's "simplified to help the less experienced user with their most common operations", end quote.
<marcan>
coderush: and gives outright wrong and broken results if you have a less common set-up
<marcan>
like impossible numbers :)
<jonaburg[m]>
interesting, from my reading I saw I thought they tried to solve the messy block allocation and flexibility by APFS in the first place
<marcan>
yeah, but that's all within a single container
<jonaburg[m]>
but thanks for the advice! ill use diskutil cli
<marcan>
which we can't use because Linux can't use that, and we also don't share a container with macOS for the stub partition because at least it makes wiping easier
<marcan>
I should make a wiki page for diskutil
<marcan>
sec
<marcan>
this is important enough
a1eaiactaest has quit [Ping timeout: 480 seconds]
<coderush>
marcan: anything GUI requires extreme amount of back-and-forth between the devs and the Human Interface people, with the latter doing everything they can to make thee GUI tools easier to use for non-power users. It often leads to software that is more usable for such users, but power users grumble about this to no end, also understandably.
<coderush>
CUI tools are way less regulated, and therefore are as brief or explicit as their devs like them to be.
creechy has quit [Ping timeout: 480 seconds]
<marcan>
right
<Chainfire>
marcan / Glanzmann : u-boot can't see my USB storage device. This machine only has usb-c ports, not sure if that's relevant
<Glanzmann>
Chainfire: For me it works.
<Glanzmann>
Chainfire: So I use a usb-c usb stick in either usb-a to usb-c dongle or in a hub with integrated ethernet and it always picks it up.
<Glanzmann>
Chainfire: Can you paste me the output of parted /dev/sda print
<Glanzmann>
and the outout of 'find .' on the usb stick?
<Glanzmann>
Chainfire: Did you use 'run bootcmd_usb0' in u-boot?
MajorBiscuit has joined #asahi
Major_Biscuit has joined #asahi
<Chainfire>
Glanzmann: yes, seems u-boot doesn't like this drive. Trying a different stick now
coderush has quit []
a1eaiactaest has joined #asahi
<Glanzmann>
Chainfire: It is important that it is the first partition on the drive.
creechy has joined #asahi
<Glanzmann>
Chainfire: I tried with Kingston HyperX 64 GB and sandisk Ultra USB 3.0.
<Chainfire>
Thanks, I just tried an ancient 1GB stick and that works
MajorBiscuit has quit [Ping timeout: 480 seconds]
<Glanzmann>
Chainfire: Perfect. vim /etc/wpa/wpa_supplicant.conf edit ssid and psk; ifup wlp1s0f0; apt update; apt install xinit blackbox gparted xterm; startx
<Glanzmann>
Chainfire: Also don't forget to make the macos container smaller in macos: diskutil apfs resizeContainer disk0s2 <final size of macos>
ktz3[m] has joined #asahi
<ktz3[m]>
Hey guys, is it my client or nobody can fetch prior conversations?
<chadmed>
use the logs
<chadmed>
in the channel description
<ktz3[m]>
yes I'm doing that already thanks, just wondered if I could get them rendered in the client
<chadmed>
nah doesnt work like that unfortunately
<Glanzmann>
ktz3[m]: If you use linux, install irssi and than use the following to conenct to the chat: '/connect -ssl irc.oftc.net 6697' and join the channel '/join #asahi'
<ktz3[m]>
okay, no prob
<ktz3[m]>
oh thanks Glanzmann I got rssi installed but didn't get it to work
<ktz3[m]>
I'll try again
creechy has quit [Ping timeout: 480 seconds]
ktz has joined #asahi
<ktz>
Its working, thanks a lot
<Glanzmann>
ktz: You're welcome.
<Chainfire>
Glanzmann: I now have the unallocated space between two APFS file systems. Can I just move those?
<Rakshit[m]>
I'm unable to change Security Policy for MacOS after installing Asahi
<Rakshit[m]>
Any idea why?
<Rakshit[m]>
Startup Security Utility only shows stub
<Glanzmann>
Chainfire: Yes, move on after the other without resizing.
<Glanzmann>
Press the green checkmark after everyone.
<Glanzmann>
Chainfire: Once you moved root fs you can extend it.
<_jannau_>
marcan: will the cleanup script wipe "external" macos installations if they are not connected?
<marcan>
boot policies? yes
<marcan>
but the boot picker will regenerate those anyway
<marcan>
also that cleanup only works with SIP disabled or from recovery anyway, otherwise it'll fail and do nothing
<_jannau_>
yes, boot policies
<YannickVanier[m]>
Just dropping by to thank the developpers, the install went flawlessly on my 2021 MBP M1 Pro, exeptionnal work by the whole crew. It's a keeper for sure
<mps>
ktz3[m]: I hope to have it on this weekend
<mps>
yesterday I used some parts of my free to build new u-boot and testing different boot methods
<j`ey>
marcan: nice to have a cheatsheet to refer back to, I'm pretty sure I get it, but Ive only actually partitioned once and that was way before we had the installer :)
<mps>
ktz3[m]: and enchiladasconpinguino[m] told that will write someting about install which I will stole ;)
amarioguy has quit [Remote host closed the connection]
<Glanzmann>
marcan: Nice read. Thanks.
<chadmed>
das u-boot has been around since like 2009
<chadmed>
its become basically the universal bootloader for embedded systems
<chadmed>
but it is indeed a nice coincidence!
creechy has quit [Ping timeout: 480 seconds]
<marcan>
AdryzzOLEDEdition[m]: in the wiki page? U-Boot is correct
<j`ey>
iBoot -> youBoot -> weBoot
<marcan>
the boot chain for vanilla Asahi Linux is PMU OTP fuses → SecureROM → iBoot stage 1 → iBoot stage 2 → m1n1 stage 1 → m1n1 stage 2 → u-boot → GRUB core image → +GRUB modules → Linux → initramfs → pivot to root filesystem
<marcan>
(somewhat deliberately making it sound ridiculous :p)
<mps>
Das U-Boot is fullname of u-boot
<j`ey>
marcan: -> kexec into newer kernel
<marcan>
I wonder if kexec works
<marcan>
pretty sure it wouldn't before the AIC changes, but maybe now it does?
<j`ey>
not until PSCI does
<marcan>
ah right, for the core shutdown, since it can't re-spintable itself
the_lanetly_052 has joined #asahi
<marcan>
it does work for single-core though, pipcet was doing that way early on (and I never liked it :p)
<j`ey>
hmm, actually if you boot with maxcpus=1 for the first kernel, maybe it works now
<marcan>
I think that always worked?
<j`ey>
s/works now/worked already/
<marcan>
AIC was one reason why SMP wouldn't work (and still won't if you don't use FastIPIs, but even t8103 does now), but PSCI is another yeah
<jn>
j`ey: ..oO(WiiBoot)
<AdryzzOLEDEdition[m]>
wasnt the bootloader of the wii called iBoot too?
a1eaiactaestCLONE has quit [Quit: WeeChat 3.4.1]
a1eaiactaest has joined #asahi
<j`ey>
AdryzzOLEDEdition[m]: the OS was called IOS, mabe thats what youre thinking of
<jn>
and a third-party experimentation tool was called mini. it is the ancestor of m1n1
<AdryzzOLEDEdition[m]>
j`ey: yeah checked on wiibrew
<chadmed>
argh just the mere mention of IOS has triggered me
<chadmed>
dark memories of putting discs in that would wipe out whatever downgraded IOS versions i had for various stuff
<sorear>
(i have to wonder about kexec in connection to the preboot DCP issues we have)
<AdryzzOLEDEdition[m]>
i was in elementary school when i tried wii homebrew
<AdryzzOLEDEdition[m]>
it was not fun when it was the only console (and sd card for that matter) that i had
<chadmed>
same lol, many fantastic memories
<chadmed>
i didnt have a working dvd player at the time so it was awesome just being able to play dvds in the wii
<j`ey>
I never played games past ps2, so m1 is the first I heard of marcan :D
<chadmed>
well i was a drooling little idiot baby when i was using wii homebrew so naturally i paid no attention to who was actually behind it all
<AdryzzOLEDEdition[m]>
me too
<chadmed>
just a small world we live in i guess
<AdryzzOLEDEdition[m]>
the first time i got interested in development was with the 3ds but that was a mess of its own
<chadmed>
though i do remember when you letterbombed the wii, it would ask you if you intended to pirate games. that definitely feels like something marcan would do
<jn>
(ok, this is getting offtopic)
<Sobek[m]>
Is there like #asahi-offtopic ?
<povik>
take a look
creechy has joined #asahi
<j`ey>
Sobek[m]: yep
realmoot has joined #asahi
<ktz3[m]>
hey jey, I saw you mentioned something about voidlinux in the logs
<ktz3[m]>
do you happen to use it?
<j`ey>
I dont, someone else was using it here
creechy has quit [Ping timeout: 480 seconds]
<j`ey>
maxung[m] is
<ktz3[m]>
maxung any chance to publish some documentation on how its done?
<mps>
povik: repeating my question from last night 'what are VSENSE and ISENSE switches in alsamixer for speakers'
<realmoot>
"Asahi Linux aims to bring you a polished Linux® experience on Apple Silicon Macs."
<realmoot>
That's really funny considering they don't ship a working kernel
realmoot has quit [Quit: Konversation terminated!]
<Glanzmann>
realmoot: If you compile from source it is fixed.
* dottedmag
wonders who are "they"
<Sobek[m]>
What is he referring to, by "don't ship a working kernel" ?
<Glanzmann>
Sobek[m]: He wants a 4kb kernel ...
<dottedmag>
obviously if Emacs from stable Debian does not work, then the kernel is not working ;)
<Sobek[m]>
Some people seem to forget this is still alpha and all the moving parts that have been dealt with yet, and all that remais to be done on the other front.
<Sobek[m]>
Stable Debian, on M1, call me in 2030 perhaps ?
<Glanzmann>
Sobek[m]: On Debian, at lease me is shipping currently a 4kb kernel, so emacs is working on Debian. I tested this morning, wondered why, rembered that I run a 4k kernel.
<Sobek[m]>
Hmm, I hope the discussion with the distros will end up in a state where distro are will to support 16k kernel (or mostly 16k kernel, if eventually a solution is found so that odd balls like FEX can run on the kernel while using 16k pages for everyone else). But sorting out these final bits is way less important than the driver work, isn't it ?
<Sobek[m]>
Glanzmann: I thought emacs was its own OS ? Is there such a thing as the secret Asahemacs cabal ?
<Glanzmann>
Sobek[m]: I agree. But first we need to get everything upstream which is a lot of work.
<Glanzmann>
Sobek[m]: Also emacs is a hard dependency to onboard Linus Torvalds to apple silicon.
<Sobek[m]>
Yeah, just saw the amount of feedback they are getting on the latest patch series sent by Sven. A lot of work there is. And that amount of work deserves praise !
<Sobek[m]>
Glanzmann: Can't he use like, vim or Kate ?
<dottedmag>
Sobek[m]: My ~/.emacs.d can now legally buy alcohol in US, do you really think I'd switch? He uses Emacs (or microemacs?) longer than I do.
creechy has joined #asahi
<Glanzmann>
Sobek[m]: Nope the keystroke are hard wired into his brain. Sobek, I also read the comments to Svens nvme patch series this morning. I mean its all 'just legwork' but someone needs to do it.
klaus has joined #asahi
klaus has quit []
creechy has quit [Ping timeout: 480 seconds]
<povik>
mps: something you shouldn't worry about. refer to tas2770 datasheet if you are curious
<povik>
it's a case of the ASoC subsystem in linux exposing something it shouldn't
jakebot has quit [Read error: Connection reset by peer]
<povik>
the macaudio driver will hide it in the future
jakebot has joined #asahi
<povik>
(it should already but apparently doesn't, something minor there)
<Sobek[m]>
Will the macaudio driver be able to deal with the DSP / prevent speaker kaboom part in the future, or is this impossible for such a driver ?
<mps>
povik: thanks for info
<povik>
no the macaudio driver but a speaker amp driver will limit the maximal gain to a sensible value
<povik>
that's a speaker amp driver in the kernel
<mps>
povik: I thought these are some limiters, but I don't have datasheet so I just tried to guess
<povik>
DSP will be up to userspace
<povik>
whether the amp gain caps will be enough to protect the speakers against evil userspace (if the DSP isn't there) i am not sure
<povik>
mps: no no, these switches are about sending some measurements from the speaker amps back to the SoC
<mps>
povik: aha, thanks. I will not read datasheet, I trust you
<AdryzzOLEDEdition[m]>
there should be clamping in place to make sure that with stuff like pavucontrol setting volume above 100% doesn't have the same result but just clips
<povik>
mps: yeah, it's not much of an interesting read :)
bradfier has quit [Quit: Leaving...]
<mps>
heh, I have garage full of IC datasheets but not much of new ones, most are 15 and up years old
bradfier has joined #asahi
kameks has joined #asahi
<povik>
AdryzzOLEDEdition[m]: there's no way it wouldn't clip already as a result of the samples representation
<ktz3[m]>
hey mps sorry for asking again, could you by any chance fill up an entry on the wiki about installing alpine? I've been looking to hop to either void or alpine since the release but I haven't dared to try something on my own fearing that I might brick the macos installation (not that I care that much now that I think of it)
<ktz3[m]>
Thought I'd ask again before going yolo on my own
<chadmed>
Sobek[m]: no, thats my job ;)
<ktz3[m]>
I mean I've daily driven alpine for a couple of months before switching to void but I can't say I trust myself to do it
<chadmed>
usual caveats apply youre listening to a recording of a recording blah blah blah
<chadmed>
the speakers do not explode if youre careful but i dont think we're ready to trust users to do that just yet
creechy has joined #asahi
jakebot0 has joined #asahi
jakebot has quit [Ping timeout: 480 seconds]
jakebot0 is now known as jakebot
* povik
should finally try the DSP himself today!
<Sobek[m]>
<chadmed> "Sobek: no, thats my job ;)..." <- Has there been any progress in convincing upstream of the usefullness of your feature request (pipewire iirc) ?
<chadmed>
sadly no, seems to be dead in the water
<povik>
i wouldn't say so
creechy has quit [Ping timeout: 480 seconds]
<mps>
ktz3[m]: I will try to create draft this evening (sorry I'm busy with something else for alpine and my $day_job else I would make it right now)
<mps>
enchiladasconpinguino[m]: do you have time to help ktz3[m] to install alpine
<chadmed>
eh even if it is its not a huuuuge problem for us, just sucks for everyone else
<chadmed>
and the fact that there are two audio devices will just have to be some jank we put up with
<chadmed>
id rather that not be the case of course but its hardly the end of the world
<povik>
chadmed: i got the impression that usefulness of hiding devices was acknowledged
<povik>
what do you say to it being a job for the 'session manager'?
<Sobek[m]>
Can someone remind me of the URL for that discussion ?
<povik>
i operate under the assumption that's a different component from pipewire
<povik>
IIRC you said session manager can't solve it
<chadmed>
povik: wireplumber exploded the entire setup and makes things break on these machines so i dont want it to be a wireplumber thing lmao
<chadmed>
of course that could just be related to some other thing
<chadmed>
but i simply cannot get it to work at all on these machines, which is why i say to use pipewire-media-session in the repo
<chadmed>
i think wireplumber is trying to be too clever or something
<chadmed>
fwiw it didnt work even with the physical output so its not something i did
<ktz3[m]>
mps understandable, no worries! thanks for responding.. I got the boot usb handy thanks to your script already, I'm just missing out on what to do after booting from the usb. Any chance describing roughly it in a sentence or two?
<povik>
we got to follow the upstream's view on separation of responsibility
<povik>
when people replace pulseaudio with pipewire they choose one of the session managers?
<mps>
ktz3[m]: ah good. you got to the point to start installation
<enchiladasconpinguino[m]>
happy to help folks with their alpine installs
<chadmed>
yeah i think arch asks you if you want wireplumber or pipewire-media-session
<ktz3[m]>
Almost where I left off before going afk, I just need to grab the static tool and I'm done with it
<mps>
ktz3[m]: use `lsblk -f` to find ESP, mount it to /mnt then untar wifi firmware `cd /lib/firmware && tar xvf /mnt/vendorfw/firmware.tar`
<chadmed>
thats the reason i want pipewire to do the device hiding, because it means pipewire-media-session will continue to work flawlessly for us
<chadmed>
and i have no idea why wireplumber wont work, nor the time or skills to try and debug and fix it
<mps>
ktz3[m]: after this try `/etc/init.d/iwd start`
<ktz3[m]>
I got ethernet all good
apg has joined #asahi
<mps>
ktz3[m]: ah good (actually better :) )
<povik>
sure, AIUI (1) pipewire needs to do device hiding and (2) all else should be done in session managers
<ktz3[m]>
yeah, and then?
<povik>
but that's just from reading the discussion and skimming pipewire's website
<mps>
ktz3[m]: from there you can follow standard alpine install guide but look at part about manually setting disk
<mps>
ktz3[m]: did you read docs.alpinelinux.org
<chadmed>
well yeah if thats the separation of powers they want to go with, that will still work for us
<chadmed>
im not entirely sure how youd go about getting the session manager to set up the virtual sink though...
<ktz3[m]>
yeah this is the part I couldn't figure out in my mind, I've been having setup-disk acting funny sometimes and I'm not totally sure but if that's all there is to it I'm sure I'll do fine
<ktz3[m]>
so basically just extracting the firmware in the ESP and then doing the installer normally? The ESP is supposed to be ready if I'm running arch already or I should be doing that each time?
<mps>
ktz3[m]: if you format root partition and mount it, just run setup disk with some parameters for your case
<mps>
ktz3[m]: setup-alpine (if you don't break it at right step) will repartition drive, so on M1 *don't do this*
<as400[m]>
povik: any progress on t600x ?
<enchiladasconpinguino[m]>
honestly the most fool proof way to install alpine, with the current alpha, is to start with external usb
<mps>
ktz3[m]: I'm waiting enchiladasconpinguino[m] to appear here and ask how he installed alpine
<Sobek[m]>
<chadmed> "well yeah if thats the separatio..." <- And this is where the devs behind the session manager also so say, not our job, pipewire should be the one doing it ?
<chadmed>
probably :D
<mps>
ktz3[m]: iirc first created VM image in qemu and installed linux-asahi kernel, and somehow 'dd'-ed (or with tar) to root partition on M1
<chadmed>
i guess the sad part is that its not exactly something that would be _hard_ to implement, it seems like its just that no one _wants_ to
<Chainfire>
Does anybody know where we're at with GPU? Is (basic) acceleration weeks, months, or years away?
<enchiladasconpinguino[m]>
now successfully done two different installs to usb flash with: 1) script from @mps in chroot 2) standard .iso in qemu
<chadmed>
and by "not hard" i mean "not hard for someone who knows how"
creechy has joined #asahi
<chadmed>
Chainfire: userspace is looking pretty good for an accelerated desktop (~96% GLES2 coverage) but the kernel interface is still some time away i think
<mps>
enchiladasconpinguino[m]: question is how you installed when you booted usb
Major_Biscuit has quit [Ping timeout: 480 seconds]
<enchiladasconpinguino[m]>
I haven't done internal nvme install of alpine yet, we're not quite there yet
<enchiladasconpinguino[m]>
(I mean you can do it manually, its possible .. I just do not recommend)
<Sobek[m]>
@chadmed do you have any pointers on what software does what in the Linux audio chain ?
<j`ey>
enchiladasconpinguino[m]: why dont you recommend it? works fine..
<mps>
enchiladasconpinguino[m]: I think setup-disk should work but I didn't tested on M1 so not 100% sure
<ktz3[m]>
mps: haha, you're the man thanks for helping out by the way
Major_Biscuit has joined #asahi
<enchiladasconpinguino[m]>
if you're an alpine expert already and have proper backups, okay sure ..
<chadmed>
Sobek[m]: in what sense?
<j`ey>
oh ok, well that should probably be the case regardless :P
<povik>
as400[m]: progress on speakers. they can be enabled for experimentation under the risk of damaging them if one isn't careful
<povik>
(the enable switch is in dts)
<povik>
once we deem them safe for general use, we will enable it by default
<Sobek[m]>
chadmed: Would it be possible to write a piece of software on the side that takes care of 2 by loading conf and dealing with pipe wire but sitting outside of the session manager (and that could be installed with any session manager) ?
<mps>
enchiladasconpinguino[m]: ktz3[m]: first time installation for me was rsync-ing rootFS from one of my arm64 chromebooks
<as400[m]>
povik: wow - that's great. Any already existing driver in the kernel that can be used ?
<Sobek[m]>
chadmed: I don’t understand how ALSA, Pipewire, wire plumber and all the other piece of software for together.
<mps>
at that time we used chainloading m1n1+dtbs+kernel with proper kernel cmdline
<chadmed>
Sobek[m]: that was actually my initial design for handling this stuff, but marcan and i decided it would be better to generalise what we're trying to do so that _everyone_ can take advantage of it
<chadmed>
lol _no one_ understands how that stuff works
<povik>
Sobek[m]: it's simple. ALSA is a part of kernel and set of userspace libraries and tools to interact with it
<povik>
anything else is a layer on top
<chadmed>
honestly you just learn it like 5 year old me learnt how windows settings worked
<chadmed>
you play with them until the machine explodes or otherwise stops working and then reverse that change ;)
<Sobek[m]>
And so is it possible for people to oops remove pipe wire and blast sound straight through ALSA and blow up the poor speakers ?
<Chainfire>
isn't that how everybody learnt everything 😂
<povik>
as400[m]: yes, there are drivers in kernel in the asahi branch
<chadmed>
Sobek[m]: very little software actually has a direct alsa pipeline anymore but the possibility is absolutely there and we're not enabling audio until that possibility no longer exists
<ktz3[m]>
@mpv I tried hacking some way to somehow safely hand execution to void by either a chroot or running a wayland session through a container but I didn't get anywhere..
<ktz3[m]>
hit enter by mistake lol
<povik>
as400[m]: if you are asking which drivers we reused that we didn't have to write ourselves: we reused the tas2764 driver with some changes
<Sobek[m]>
Well if you write the soft that eat config files (let’s call it pipewire-dsp) and sets up pipewire, I think you can still lobby for it to be shipped by distros ? It will still be less convenient because people could end up with pipewire without pipewire-dsp but should still be generic enough to be usable by everyone ?
creechy has quit [Ping timeout: 480 seconds]
<as400[m]>
povik: yes - that was my question actually - thx :)
<Sobek[m]>
chadmed: How do you deny that possibility ? Pipewire and friend can take ownership of the output to prevent any other software from using it ?
<chadmed>
i believe the driver will limit the volume. barring that it will just be a UCM2 profile, but thats not a foolproof way to do it either (users can delete them)
<ktz3[m]>
but yeah hopefully I'll be running alpine tomorrow, you'd better be working :D thanks a lot btw mps
<chadmed>
the best and safest way is in the driver with a hard limit so that userspace doesnt even know about higher volumes at all
<povik>
i should do it already just so we don't discuss it anymore :p
<chadmed>
:D
<j`ey>
povik: should be easy, just add: max_volume = clamp(max_volume, 90);
<j`ey>
on line 230
<povik>
pretty much
<chadmed>
itll probably be a per machine thing right
<povik>
yes
<povik>
the '90' value i should probably figure from macos in hypervisor, which is why i have been postponing it
<chadmed>
on this machine at least its around ~65 before a 0dBFS signal will cause clipping
<mps>
ktz3[m]: sorry I don't have time today (or tomorrow) to test complete instalation, will try on weekend. but I expect it should work
<povik>
chadmed: is that with marcan's volume knob fixes?
<povik>
(current linux-asahi)
<chadmed>
yeha current kernel
<Sobek[m]>
Is there any value that would be over conservative but known to be safe that could be used as a default if the exact valkue for the machine isn't known ? (Aside from 0)
<chadmed>
not something we can determine sanely without experimentation
<chadmed>
10% is probably safe, but will you hear anything? no
<chadmed>
alsa kinda sucks in that its volume controls are linear but human hearing is not
<chadmed>
userspace audio servers do a better job of this by having nonlinear volume curves
<povik>
isn't it linear in decibels?
<povik>
or that is the linear you are complaining about :p
<povik>
and human hearing isn't linear even in decibels
<chadmed>
is alsa? i know at least one of them is stupid and wrong
<povik>
ah
<AdryzzOLEDEdition[m]>
decibels arent linear by definition
<povik>
it's a figure of speach
creechy has joined #asahi
<chadmed>
yeah but the problem is the alsa volume slider doesnt work like a fader, so it treats its db scale as linear
<chadmed>
which is why 10% might be inaudible but 20% is loud
<chadmed>
its dumb
<povik>
how does a fader work?
<AdryzzOLEDEdition[m]>
ooh
<povik>
you mean the bars are the same height even if decibel range is different?
<AdryzzOLEDEdition[m]>
i think i got it
<AdryzzOLEDEdition[m]>
it isnt doing a log() of the percentage
<chadmed>
most faders are nonlinear, in that 1cm movement at the bottom will cause a greater change in apparent volume than a 1cm move at the top (log scale)
<AdryzzOLEDEdition[m]>
it treats db as a percentage
<Chainfire>
Guess we have a third Chromium patch submitted now... talk about doing triple work
<j`ey>
Chainfire: link!
<ktz3[m]>
<mps> "ktz3: sorry I don't have time..." <- Its okay, at least I conceptualized a bit what I need to do so I guess I should be fine, lets see.. I want to do it right away tbh but I'm long overdue for bed.. have a great day for now
tomtastic has joined #asahi
a1eaiactaest has quit [Ping timeout: 480 seconds]
<enchiladasconpinguino[m]>
ktz3: ktz3: fwiw this script from mps worked on the first try for me with usb:
<Chainfire>
j`ey: well, as long as one of them gets accepted at some point, that'd be great for Asahi for sure. If it's not mine it's not great for the usefulness of my time though :D
<j`ey>
Chainfire: I wonder if the other patch author is on this channel
<Chainfire>
No idea. I have to run now, but I actually suspect that other patch will not link. Will probably try when I get back home.
<ktz3[m]>
mps: nice, thanks, you made it look simple
<mps>
well alpine moto is: simple, small, secure
<mps>
though I'm not sure any of these words are true ;)
<enchiladasconpinguino[m]>
its all relative ;)
jakebot has quit [Ping timeout: 480 seconds]
jakebot3 is now known as jakebot
<chadmed>
relative to windows? its probably all of those things
chadmed has quit [Remote host closed the connection]
<ktz3[m]>
@mps haha it started stinking to me as well for no particular reason, maybe it mostly has to do with docker but I can't point my finger to something
<ktz3[m]>
I found voidlinux to be the perfect distro to suit an alpine like experience.. mind that I was using alpine at my desktop but yeah it was great nonetheless
<ktz3[m]>
but for now thank god for alpine honestly arch feels like bloatware now lol
creechy has joined #asahi
MajorBiscuit has joined #asahi
Ry_Darcy has quit [Remote host closed the connection]
apg has quit [Read error: Connection reset by peer]
Major_Biscuit has quit [Ping timeout: 480 seconds]
creechy has quit [Ping timeout: 480 seconds]
kameks has quit [Ping timeout: 480 seconds]
<jonaburg[m]>
thanks Marcon for this writeup about disk utility! no idea where you learned this stuff, or if its all by experimentation and inference
<mps>
ktz3[m]: when someone try alpine after that everything else is bloatware
<matthewayers[m]>
Looks like Asahi got a shoutout here (totally wasn’t expecting it) https://youtu.be/AvGuXUJ6xjU
<Sobek[m]>
<mps> "please read and 'help me!' :)" <- This is going to turn into a gentoo install (like twitch trying to install archlinux)
kita99 has joined #asahi
a1eaiactaest has quit [Ping timeout: 480 seconds]
creechy has quit [Ping timeout: 480 seconds]
kelito has quit [Ping timeout: 480 seconds]
<j`ey>
mps: it says qemu-riscv, not qemu-aarch64
yuyichao_ has joined #asahi
MajorBiscuit has quit [Ping timeout: 480 seconds]
MajorBiscuit has joined #asahi
jakimo has joined #asahi
jakimo is now known as help
help is now known as jakimo
<mps>
j`ey: thank you
<mps>
copy-paste writer in action :)
a1eaiactaest has joined #asahi
<jakimo>
Good morning!
creechy has joined #asahi
<jakimo>
I'm curious to know if there's an email update system to be informed as things move along for Asahi.. it's pretty amazing to see this project underway! I've only watched a video of it so far, but I'd like to install soon
<Glanzmann>
jakimo: If you have twitter account you could activate email notifications and subscribe marcan42 or asahi linux ...
<jakimo>
got it, thanks glanzmann
a1eaiactaest has quit [Ping timeout: 480 seconds]
jakimo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<kaprests>
Going forward, is the plan to have numbered releases, or will the alpha be gradually updated sort of like a rolling release?
a1eaiactaest has joined #asahi
<AdryzzOLEDEdition[m]>
rolling release
doggkruse has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
amarioguy has joined #asahi
user982492 has joined #asahi
<dottedmag>
rolling into disappearance
<mps>
marcan: will the filesystem label for ESP stay 'EFI - UEFI' in future? I want to make disk image for alpine linux and create script which will unpack firmware from ESP on boot
LinuxM1 has joined #asahi
<LinuxM1>
someone who uses macbook Air M1 tell me how to turn on the volume? I can't get it to work, if I open "alsamixer" it shows me as "item" jack ADC and as "card" macbook air integrated aud
<mps>
LinuxM1: 'Mixer' control
<LinuxM1>
mps can you explain better the steps I have to take?
<j`ey>
LinuxM1: by volume, you mean the jack right? there;s no speaker support yet
<LinuxM1>
with alsamizer i see only "jack"
a1eaiactaest has quit [Ping timeout: 480 seconds]
<mps>
ah, question is about speakers, they don't work without change dtbs and rebuild
m2231puppy has joined #asahi
<LinuxM1>
to date, is there any procedure to make the audio of my macbook air work? or can I do nothing?
<j`ey>
LinuxM1: do you mean speakers?
<j`ey>
or the headphone port
<LinuxM1>
a don't understand
<j`ey>
LinuxM1: do you want audio from headphones?
<Candygoblen123[m]>
currently, the internal speakers on your macbook air will not work under linux
<j`ey>
or out loud
<LinuxM1>
i want audio from speakers of my macbook air
<LinuxM1>
is is possible or not?
<j`ey>
not possible
<Candygoblen123[m]>
nope
<LinuxM1>
ok not possible
<Candygoblen123[m]>
you have to use the headphone jack
<Candygoblen123[m]>
and even then, sometimes it won't work
<LinuxM1>
I thought I was doing something wrong
a1eaiactaest has joined #asahi
<maxung[m]>
ktz3: I did it very complicated, by installing asahi and then using a debian live stick to delete root partition, extracting void.tar, chrooting and setting everything up. because i had problems with grub and booting void, i ditched it and now boot directly via m1n1 (without u-boot). maybe i will write an install script similar to asahi/debian, not sure yet though
<j`ey>
maxung[m]: you dont need a script, just an image
<mps>
heh, my script to make alpine usb disk could be easily adapted to install alpine on nvme root partition
Major_Biscuit has joined #asahi
amarioguy has quit [Ping timeout: 480 seconds]
a1eaiactaest has quit [Ping timeout: 480 seconds]
MajorBiscuit has quit [Ping timeout: 480 seconds]
<maxung[m]>
<j`ey> "maxung: you dont need a script..." <- but something to create an image right? or just packing my root + efi partition?
<maxung[m]>
mps: do you have a link?
<j`ey>
oh yeah you need a script to actually generate the image
<tpw_rules>
mps: the EFI partuuid is in /proc/device-tree/chosen/asahi,efi-system-partition or so
<tpw_rules>
the filesystem label depends on the name you give your operating system
m2231puppy has quit [Read error: Connection reset by peer]
m2231puppy1 has joined #asahi
shnert[m] has joined #asahi
<mps>
tpw_rules: I don't have this file on my machine
<maxung[m]>
mps: thanks
<shnert[m]>
Good morning everyone!
<tpw_rules>
mps: do you have /proc/device-tree/chosen? did you install your own m1n1.bin?
m2231puppy1 is now known as m2231puppy
a1eaiactaest has joined #asahi
m2231puppy has quit []
<mps>
tpw_rules: I have dir with some files but not asahi,efi-system-partition
<tpw_rules>
did you install your own m1n1.bin on the ESP or use the one the installer made?
<mps>
tpw_rules: no, I installed m1n1.bin from installer, though I created my own m1n1+dtbs+u-boot for ESP/m1n1/
<tpw_rules>
so then you did install your own m1n1.bin on the ESP
<mps>
tpw_rules: right
<tpw_rules>
you need to gzip or xz compress the u-boot you concatenate to m1n1 or the chainloading won't work properly and you won't get that variable
<mps>
tpw_rules: why do ask
guillaume_g has quit []
<mps>
s/do/do you/
<mps>
or to rephrase, what is use case for /proc/device-tree/chosen/asahi,efi-system-partition
<tpw_rules>
[10:36] mps] marcan: will the filesystem label for ESP stay 'EFI - UEFI' in future? I want to make disk image for alpine linux and create script which will unpack firmware from ESP on boot
a1eaiactaest has quit [Ping timeout: 480 seconds]
<tpw_rules>
the answer to that is that the ESP is named 'EFI -' + the name the user gave the OS in the installer
<maxung[m]>
mps: so I guess one would execute the install script from an existing alpine installation right?
<tpw_rules>
the way to unambiguously identify it is to use the UUID in that /proc/device-tree/chosen/asahi,efi-system-partition file
<mps>
marcan: ah, good one text, didn't know it exists. thank you
<marcan>
it's linked in our announcement post...
<mps>
marcan: you are talking to old and uncareful man :)
<mps>
sorry, jk
<mps>
but I missed it
<tpw_rules>
marcan: can you add the bit about compressing stage 2 payloads to make the chainloading work properly? i remember you mentioned you wanted to mandate compressed payloads; that could be good too
<marcan>
"arm64 Linux style kernel images, which should be gzip/xz compressed"
<marcan>
and the subsequent section mentions it again
<tpw_rules>
ah ok
* marcan
writes docs that nobody reads :<
<maz>
the fate of most docs...
<maz>
but someone would shout at you for not writing them otherwise -- it's a loosing game.
<tpw_rules>
well now i know where to look first
<maxung[m]>
I'm wondering, does the m1n1 stage1 installed with kmutil just include "Initializes hardware -> Puts up a pretty logo -> Chainloads another version of itself from a FAT32 partition"?
<tpw_rules>
yes, it doesn't have any payloads by default
<Glanzmann>
marcan: Don't worry I have read all your docs. Everytime I learned something new. I once even had a fight with my wife because she disturbed me multiple times while I was reading one of your more bloated docs. The page turner about the boot chain.
<nicolas17>
'bloated' has a bad connotation :P
kreechy has joined #asahi
Ry_Darcy has joined #asahi
user982492 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
creechy has quit [Ping timeout: 480 seconds]
<Ry_Darcy>
@Glanzmann I have applied your 4K boot.bin and rebooted. As you can see, I am still running on this cheap Benz 2K monitor in BE. I will have to wait until tomorrow evening to test this setup on my 4K monitor in ZH.
<Ry_Darcy>
@Glanzmann Benq not Benz.
<Ry_Darcy>
@Glanzmann - oot@debian:~# ll /boot/efi/m1n1/ total 2928 4 drwxr-xr-x 2 root root 4096 Mar 24 17:55 . 4 drwxr-xr-x 7 root root 4096 Jan 1 1970 .. 1460 -rwxr-xr-x 1 root root 1493310 Mar 23 16:23 1920p.bin 1460 -rwxr-xr-x 1 root root 1492547 Mar 24 17:55 boot.bin root@debian:~#
<Ry_Darcy>
@Glanzmann - when that is legible. :)
amarioguy has joined #asahi
chr has joined #asahi
<WhyNotHugo>
Booting an uefi bundle (systemd-stub+kernel+initrd+cmdline) just results in this and then the system reboots: https://pasteboard.co/UAeo9Q4tVuQK.jpg
<WhyNotHugo>
(previous issues were due to systemd-boot not supporting gziped kernels on aarch64)
<tpw_rules>
the kernel is crashing somehow. try to boot it under the hypervisor. systemd-boot works fine for me
<WhyNotHugo>
tpw_rules: systemd-boot works fine for me without a UEFI bundle, only the bundle fails like this.
<WhyNotHugo>
tpw_rules: Any links on where to start for that?
<WhyNotHugo>
Oh, I need a second mac for that ,right?
<Sobek[m]>
Whoever was trying to make a suitable gentoo install media (+ stage3 I don't remember) how has it been going ? My backup situation has been sorted out, and I'm likely going to try out Asahi plus possibly gentoo this week-end
a1eaiactaest has joined #asahi
starsam80[m] has joined #asahi
r0ni has quit [Ping timeout: 480 seconds]
<Glanzmann>
Ry_Darcy: Let me know, how it turns out to be.
<Glanzmann>
jannau: Congrats on the double uart bring up. `o'
amarioguy has quit [Ping timeout: 480 seconds]
user982492 has joined #asahi
Ry_Darcy has quit [Remote host closed the connection]
Ry_Darcy has joined #asahi
kreechy has quit [Ping timeout: 480 seconds]
kreechy has joined #asahi
alex587[m] has joined #asahi
kelito has joined #asahi
chr has quit [Ping timeout: 480 seconds]
<Cy8aer[m]>
Glanzmann: uuuh got it. This message is written out of Matrix Element, Firefox ESR - (the debian unstable firefox 98.0-2 does not open any window and hangs!!), crypted file system. You'll get some more information how to set this up soon.
<Glanzmann>
Cy8aer[m]: So you have now dm-crypt / lvm / btrfs with the 4k kernel and firefox does not start?
<Glanzmann>
Cy8aer[m]: Ah okay so everything works, just the unstable firefox gives you issues. I always use the esr so I never noticed.
<Glanzmann>
Cy8aer[m]: Good to hear that you succeeded. ;-)
<jannau>
it sounds like the same issue we saw last week with the arch linux arm firefox 98
<Glanzmann>
Ahh.
<Glanzmann>
jannau: But that is on a 4k kernel.
the_lanetly_052 has quit [Ping timeout: 480 seconds]
<Cy8aer[m]>
Yes chromium works
<jannau>
I don't think it is related to the page size
<Cy8aer[m]>
german. Yes. That is a bit nasty because on `<>` is `|` as alternate key.
<Cy8aer[m]>
Ah (grabbing through the issues...)
<geerlingguy>
I didn't see anything in my quick searching... but has anyone tested 10G network performance on the 10G M1 mini's internal port?
<geerlingguy>
On 2.5G network I'm only getting about 1.5 Gbps of throughput (same port and config gets 2.35 Gbps on macOS side, didn't know if driver issue or what). Haven't tested full 10G connection yet though.
amarioguy has joined #asahi
<Cy8aer[m]>
j`ey: Thanks, `echo 1 >/sys/module/hid_apple/parameters/iso_layout` does it for me...
<j`ey>
Cy8aer[m]: great
<Cy8aer[m]>
(Wayland...)
kim0[m] has joined #asahi
<j`ey>
geerlingguy: not many seem to have the 10G model
<geerlingguy>
I can't believe more people wouldn't be willing to run alpha software on the most expensive M1 mini model :D
<jannau>
geerlingguy: not that I know of. we all bought the mini before the 10gb option was available
<nicolas17>
oh it's an option?
<nicolas17>
base model is 1G?
<jannau>
but I see following on the mac studio: "pci 0000:03:00.0: 2.000 Gb/s available PCIe bandwidth, limited by 2.5 GT/s PCIe x1 link at 0000:00:02.0 (capable of 63.012 Gb/s with 16.0 GT/s PCIe x4 link)"
<geerlingguy>
Then I'll keep testing and see what I find. Maybe pinning to a CPU core or something. I noticed if I have the graphical resource monitor up it uses like 26% CPU on four cores just to display itself lol
<geerlingguy>
jannau: oh... I don't have Asahi on the studio, just running on M1. Let me look at pcie devices
<jannau>
that line is from dmesg and sounds if the pcie link is just running with 2.000 Gb/s which would explain poor performance
<geerlingguy>
Sees Aquantia controller, 2.00 Gb/s limited by x1 link
<geerlingguy>
hmm... what's limiting it then?
<jannau>
the studio uses the same nic
<geerlingguy>
It seems like all 10G adapters use that aquantia nic :D
ciggi_ has quit [Read error: Connection reset by peer]
<geerlingguy>
I believe the Thunderbolt adapter from OWC does too (though TB3 doesn't work on Asahi of course)
<jannau>
probably a problem with pcie initialization
artemist has quit [Ping timeout: 480 seconds]
<mini>
all of the mac compatible adapters I'm aware of do - because apple bundle the driver
<geerlingguy>
yeah. I'll document it for now and make a note to look later. Just exploring
kim0 has joined #asahi
<mini>
I have a qnap one that is also aquantia - takes a SFP+ too for extra fun :)
<geerlingguy>
Yeah that was my preferred choice, but it was more expensive than OWC :D
LinuxM1 has quit [Quit: Leaving]
ciggi has joined #asahi
<kim0>
Howdy all. Sorry if this is not directly related! Is there a way to run Linux on M1, in a high security VM .. (full disk encryption, and maybe using secure enclave) ?
LinuxM1 has joined #asahi
<Glanzmann>
About the 10 Gbit/s mini. I just tested on a friends workstation which is running macos: I get 10 Gbit/s in both directions (that is under macos): https://pbot.rmdir.de/DNCF9_McRdUgBVirbloNWQ
<Glanzmann>
iperf3 single stream to another workstation in the same room
<jannau>
that looks better: 'pci 0000:03:00.0: 15.753 Gb/s available PCIe bandwidth, limited by 16.0 GT/s PCIe x1 link at 0000:00:02.0 (capable of 63.012 Gb/s with 16.0 GT/s PCIe x4 link)'
<Glanzmann>
Yep.
<jannau>
we initialize the pcie port as PCIe 1 because apple describes it as "maximum-link-speed = 1" which is clearly wrong
<Glanzmann>
I see.
<WhyNotHugo>
Closing and opening the lid has been very reliable. In the past, new platforms have often had issues like the display not turning on after opening the lid.
* WhyNotHugo
knocks on woof
<WhyNotHugo>
* wood
<jannau>
there is also a "target-link-speed = 4"
<Glanzmann>
WhyNotHugo: Hrhr. Today it failed on me once. But I powered in on, closed the lid and lid was not turning on after I opened it. Probably because a pre m1n1 boot loader turned it off and m1n1 never probed the display. This was with Debian.
<Glanzmann>
WhyNotHugo: But yes, otherwise it is really stable.
<jannau>
that seems to to be the way apple use to describe the port for 1gb and 10gb
<WhyNotHugo>
Glanzmann: Detecting the display at runtime should be possible, just absent on the software side, right?
<Glanzmann>
WhyNotHugo: Yep. It will probably come with the dcp at some point.
<Glanzmann>
jannau: Nice.
<jannau>
yes, apple seems to changed how they announce the link speed for the 10gb nic in 12.3 (or earlier)
<Glanzmann>
jannau: My test was on Macos 11.6.4.
<chr>
Would I be kicked if I asked whether my Asahi Debian partitions could be used with qemu from within Mac OS?
<AdryzzOLEDEdition[m]>
the root can as it is a standard aarch64 root
<AdryzzOLEDEdition[m]>
everything else no
<WhyNotHugo>
What's /dev/nvme0n2 and /dev/nvme0n3?
<geerlingguy>
jannau: ah, so you're thinking it'll run at x4 speed if you set link speed 4 instead? Hopefully that same setup works with the 1G units. I don't have one I could test on
<nicolas17>
huh
<nicolas17>
syscfg is in a namespace too right?
<sven>
isn’t syscfg on the nor flash?
<Glanzmann>
WhyNotHugo: WhyNotHugo:
<Glanzmann>
The others are used for kernel panic logs and stuff like that. That's pretty low level stuff you don't have to care about. This is an NVMe thing, like "low-level partitions". Just don't think too much about it.
<jannau>
geerlingguy: 'pacman -Syu', current version is containerd-1.6.2-1
<lkvrsfld[m]>
Hey geerlingguy
<lkvrsfld[m]>
Love your channel btw
<lkvrsfld[m]>
Docker worked without problem installing with pacman… try to pacman -Rns it, and check the dependencies while reinstalling?
<geerlingguy>
Glanzmann: yeah but my question is specifically regarding Asahi / Asahi desktop ;)
<geerlingguy>
jannau: interesting. It's trying to install 1.6.1
<jannau>
your package index is probably outdated
<geerlingguy>
oooooooh.... didn't realize pacman is like apt
<geerlingguy>
thought it was more like dnf
<geerlingguy>
"btw I don't use Arch [normally]" ;)
<nicolas17>
"like dnf" you mean "updating indexes every time"?
<geerlingguy>
yeah
<geerlingguy>
yay docker installed
<lkvrsfld[m]>
Just disable that service?m, then it will start instantly
<lkvrsfld[m]>
geerlingguy: if docker service takes ages to load, it‘s caused by systemd-networkd-wait-online, that service just never loads, and docker waits 300s (same for other systemd services)
<chr>
j`ey: could you help me compose suitable args for qemu-system-aarch64 ? See http://nybo.no/diskutil/
<lkvrsfld[m]>
> Just disable that service?m, then it will start instantly
<lkvrsfld[m]>
(I had it on asahi and pure arch linux arm)
<lkvrsfld[m]>
> <@prohackerbro:matrix.org> geerlingguy: if docker service takes ages to load, it‘s caused by systemd-networkd-wait-online, that service just never loads, and docker waits 300s (same for other systemd services)
<geerlingguy>
Can't start it though, "devmapper not configured". I'll keep digging
the_lanetly_052 has quit [Ping timeout: 480 seconds]
<j`ey>
chr: I dont know how to do it offhand
<geerlingguy>
jannau++
doggkruse has joined #asahi
<jonaburg[m]>
oh wow geerlingguy! I use many of your sensible roles :D
<jonaburg[m]>
s/sensible/ansible/
kode54 has quit [Quit: Ping timeout (120 seconds)]
<geerlingguy>
It's quite sensible to use them ;)
<geerlingguy>
one of the many things I plan on testing this week...
* nicolas17
is a noob handcrafting ansible stuff with no shared roles
skipwich has quit [Quit: DISCONNECT]
kefu_ has quit []
nepeat has quit [Remote host closed the connection]
kode54 has joined #asahi
kefu_ has joined #asahi
skipwich has joined #asahi
nepeat has joined #asahi
<geerlingguy>
ah... since I updated kernel probably just need a reboot
<lkvrsfld[m]>
geerlingguy: Are you using the mac mini?
<geerlingguy>
yep! Running now. Dang, love Asahi so far.
<geerlingguy>
lkvrsfld[m]: yes 10G M1 mini with 2TB SSD 16GB RAM
<lkvrsfld[m]>
Nice. Didn‘t actually know they brought out a 10G version until reading it here
kode54 has quit []
<geerlingguy>
Yeah I held off on upgrading to M1 until the 10G version came out. And now that I have a Mac Studio, I'm going to play with Asahi on the M1 mini, see if I can get it to be an efficient little ARM server
taowa has quit [Read error: Connection reset by peer]
eichin has quit [Read error: Connection reset by peer]
steev has quit [Read error: Connection reset by peer]
mustache_ride___ has quit [Read error: Connection reset by peer]
peac has quit [Read error: Connection reset by peer]
cptcobalt has quit [Read error: Connection reset by peer]
jkkm has quit [Write error: connection closed]
maennich has quit [Read error: Connection reset by peer]
Chainsaw has quit [Read error: Connection reset by peer]
daniels has quit [Write error: connection closed]
kendfinger has quit [Read error: Connection reset by peer]
philpax_ has quit [Read error: Connection reset by peer]
Vaughn has quit [Read error: Connection reset by peer]
arnd has quit [Read error: Connection reset by peer]
stblassitude has quit [Read error: Connection reset by peer]
Venhal has quit [Read error: Connection reset by peer]
jbowen has quit [Write error: connection closed]
kode54 has joined #asahi
sorear has quit [Read error: Connection reset by peer]
sa1 has quit [Read error: Connection reset by peer]
tardyp has quit [Read error: Connection reset by peer]
enomem has quit [Write error: connection closed]
saintdev_ has quit [Write error: connection closed]
guan has quit [Write error: connection closed]
jabashque has quit [Write error: connection closed]
aeroraptor has quit [Write error: connection closed]
NekomimiScience has quit [Write error: connection closed]
esden has quit [Read error: Connection reset by peer]
ovf has quit [Read error: Connection reset by peer]
austriancoder has quit [Read error: Connection reset by peer]
ZLSA has quit [Write error: connection closed]
mmarchini has quit [Read error: Connection reset by peer]
snek has quit [Read error: Connection reset by peer]
ZLSA has joined #asahi
sa1 has joined #asahi
cptcobalt has joined #asahi
stblassitude has joined #asahi
peac has joined #asahi
daniels has joined #asahi
eichin has joined #asahi
Venhal has joined #asahi
philpax_ has joined #asahi
NekomimiScience has joined #asahi
ovf has joined #asahi
guan has joined #asahi
enomem has joined #asahi
esden has joined #asahi
aeroraptor has joined #asahi
jkkm has joined #asahi
mmarchini has joined #asahi
snek has joined #asahi
jabashque has joined #asahi
tardyp has joined #asahi
austriancoder has joined #asahi
saintdev_ has joined #asahi
sorear has joined #asahi
kendfinger has joined #asahi
mustache_ride___ has joined #asahi
Vaughn has joined #asahi
taowa has joined #asahi
maennich has joined #asahi
steev has joined #asahi
Chainsaw has joined #asahi
arnd has joined #asahi
jbowen has joined #asahi
<lkvrsfld[m]>
Thought you featured the M1 in some video. I may be wrong
<lkvrsfld[m]>
They have potential. Please buy more of them and do a high performance kube cluster with them!😂
kode54 has quit []
kode54 has joined #asahi
jbowen has quit []
aeroraptor has quit [Read error: Connection reset by peer]
saintdev_ has quit [Read error: Connection reset by peer]
sa1 has quit [Read error: Connection reset by peer]
sa1 has joined #asahi
aeroraptor has joined #asahi
jbowen has joined #asahi
saintdev_ has joined #asahi
<geerlingguy>
Heh, definitely a *lot* faster than the Pi
___nick___ has quit [Ping timeout: 480 seconds]
<nicolas17>
[buy me one to replace this desktop PC that I have to keep smacking to silence the fans]
<geerlingguy>
nice thing so far: all the ARM64 Docker images I'm testing on Asahi have worked (just as they did in Docker for Mac)
<AdryzzOLEDEdition[m]>
can't wait for a watercooled mac mini
tardyp has quit [Read error: Connection reset by peer]
kendfinger has quit [Read error: Connection reset by peer]
Chainsaw has quit [Read error: Connection reset by peer]
maennich has quit [Read error: Connection reset by peer]
Vaughn has quit [Read error: Connection reset by peer]
jabashque has quit [Read error: No route to host]
<lkvrsfld[m]>
we should measure disk I/O and compare it to macOS docker. mount have been the biggest problem on macOS so far
maennich has joined #asahi
saintdev_ has quit [Read error: Connection reset by peer]
<nicolas17>
AdryzzOLEDEdition[m]: why? I thought it was silent enough already?
sorear has quit [Read error: Connection reset by peer]
jabashque has joined #asahi
Vaughn has joined #asahi
mustache_ride___ has quit [Read error: Connection reset by peer]
austriancoder has quit [Read error: Connection reset by peer]
mustache_ride___ has joined #asahi
sorear has joined #asahi
saintdev_ has joined #asahi
<mini>
lkvrsfld[m]: there's a recent preview for macOS docker that improves that significantly on 12.3
austriancoder has joined #asahi
kendfinger has joined #asahi
<mini>
(but it's still a bit buggy)
tardyp has joined #asahi
<AdryzzOLEDEdition[m]>
nicolas17: to push it further with freq
Chainsaw has joined #asahi
<nicolas17>
ah
<AdryzzOLEDEdition[m]>
i mean i doubt you can OC these
<lkvrsfld[m]>
mini: cool. we hacked it to work using the same UIDs on the host and in the container )
<nicolas17>
have people gotten the mini to throttle due to heat?
<lkvrsfld[m]>
> <@_oftc_mini:matrix.org> lkvrsfld: there's a recent preview for macOS docker that improves that significantly on 12.3
<lkvrsfld[m]>
* cool. we hacked it to work using the same UIDs on the host and in the container. that was even better than using NFS
<nicolas17>
that is definitely an issue on the macbook air, but I thought the fan on the mini was enough to keep it from throttling
<AdryzzOLEDEdition[m]>
funny how the air doesnt have any air
<nicolas17>
funny how the iMac Pro completely broke Steve Jobs's naming scheme
mustache_ride___ has quit [Read error: Connection reset by peer]
Vaughn has quit [Read error: No route to host]
Vaughn has joined #asahi
tardyp has quit [Read error: No route to host]
kendfinger has quit [Read error: Connection reset by peer]
tardyp has joined #asahi
jbowen has quit [Read error: Connection reset by peer]
saintdev_ has quit [Read error: No route to host]
sorear has quit [Read error: No route to host]
jbowen has joined #asahi
arnd has quit [Read error: Connection reset by peer]
<geerlingguy>
woah... Asahi Linux build within Docker is running in 6m on same M1 that runs in Docker for Mac in 9m — and that's not even using shared volumes!
<jbowen>
<3
<nicolas17>
iXxx for consumer segment, Xxx Pro for professional segment; "iMac Pro" makes no sense in that scheme
<AdryzzOLEDEdition[m]>
and they didnt even add an extra "max" just in case
chr has left #asahi [ERC 5.4.1 (IRC client for GNU Emacs 29.0.50)]
<nicolas17>
we're in year 2020 Max
<nicolas17>
next year is 2020 Ultra
<AdryzzOLEDEdition[m]>
😭2013my beloved
<AdryzzOLEDEdition[m]>
* 😭2013 my beloved
<bmentink[m]>
hi all, using gnome on macbook air, but no sound and when I add a user the password is not accepted by gdm .. any clues
<AdryzzOLEDEdition[m]>
no sound is through the internal speakers?
<nicolas17>
bmentink[m]: headphone jack or internal speakers?
amarioguy has quit [Ping timeout: 480 seconds]
<bmentink[m]>
no sound through speakers or jack
<AdryzzOLEDEdition[m]>
internal speakers dont work yet
<AdryzzOLEDEdition[m]>
for the jack try checking alsamixer
<LinuxM1>
jack dont work too
<nicolas17>
as listed in the alpha release announcement, "M1 only: Headphones jack (might be flaky)"
<nicolas17>
and "Not yet, but coming soon: speakers"
<bmentink[m]>
ok, any ideas on the useradd ?
<nicolas17>
a common suggestion I saw was to switch the audio profile to input only, then back to input+output, and that gets it unstuck
<bmentink[m]>
I tried "useradd -m username -p password"
<AdryzzOLEDEdition[m]>
just a thermal pad will do if you can stand the bottom of your laptop becoming a frying pan
axenntio has quit []
axenntio has joined #asahi
axenntio has quit []
<nicolas17>
which (in early days) proved it's literally the same chip, not "throttled because it's the air" but throttled because it simply got hot
<lkvrsfld[m]>
AdryzzOLEDEdition[m]: Yep
axenntio has joined #asahi
AnhHoangNguyen[m] has joined #asahi
axenntio has quit []
AnhHoangNguyen[m] is now known as aaanh[m]
Tano70 has joined #asahi
Tano70 has quit [Read error: Connection reset by peer]
risk has joined #asahi
<kelito>
Anyone else feels the trackpad like too fast?
<bmentink[m]>
just right for me :)
user982492 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<kelito>
I feel it a little bit diff compared to OSX
<kelito>
maybe the tap
<nicolas17>
does multitouch work?
<kelito>
no idea
<nicolas17>
I mean can you scroll with two fingers etc?
<bmentink[m]>
I can scoll and zoom ok
<kelito>
Yes, I can, so it works
<nicolas17>
neat
Tano70 has joined #asahi
<bmentink[m]>
except zoom does not seem to work in Firefox
<nicolas17>
I think Plasma is working on the 4-finger gestures and such
<mps>
nicolas17: xorg or wayland?
<WhyNotHugo>
Does the magic keyboard's touchId work on Linux?
<nicolas17>
WhyNotHugo: pretty sure we're far from that :)
<nicolas17>
will need a SEP driver etc
<mps>
WhyNotHugo: touchbar?
<nicolas17>
and even once touchid on the MBP works, supporting the external magic keyboard might need extra work on top of that
<lkvrsfld[m]>
multi-finger-gestures (more than two) already work in wayland
<lkvrsfld[m]>
besides the scrollings speed the touchpad driver is great!
<WhyNotHugo>
nicolas17: The TouchId interfacing is via the SEP only?
<mps>
lkvrsfld[m]: in xorg xinput can set accell to 0.7 for example
axenntio has joined #asahi
<brudda[m]>
Talking about the keyboard, is the backlight possible yet?
<nicolas17>
WhyNotHugo: the fingerprint recognition algorithm is in the SEP firmware, and I think the touchid sensor is physically wired to the SEP and the main CPU can't even access the raw fingerprint image
<j`ey>
brudda[m]: no
<lkvrsfld[m]>
mps: ah cool. i only used wayland, saw the discussion of modifying it on wayland is more difficult
<mps>
lkvrsfld[m]: sorry, I don't know how to set it with wayland
<brudda[m]>
j`ey: Okay thanks : )
<lkvrsfld[m]>
mps: no prob, didn't even ask :)
<mps>
does wayland use libinput driver?
<sven>
the fingerprint communicates with the main CPU over SPI iirc
<dottedmag>
mps: Wayland is a protocol, it can't "use" anything
<j`ey>
thats the mesa/bio stuff
<sven>
it’s just paired with the SEP such that the entire communication is encrypted
<dottedmag>
mps: That's similar to saying "Does HTTP use libapr"?
<sven>
and presumably the same protocol works over Bluetooth or usb for that keyboard
<j`ey>
sven: just checked, spi2
<nicolas17>
sven: ah so not "physically wired to SEP alone" but it's encrypted, okay
<sven>
but that’s all still far away
<j`ey>
sven: in a galaxy far far away
<mps>
dottedmag: yes, but people use wayland when meaning sway and similar
<dottedmag>
mps: So every compositor is free to use or not use libinput.
<dottedmag>
so is the question about kwin, mutter or libsway-based compositors?
LinusSexTips[m] has joined #asahi
<mps>
dottedmag: I'm not sure, only tried xorg
<nicolas17>
so it's similar to how the enrolled fingerprint is not *really* stored in the SEP (as Apple claims in marketing), but it's encrypted with a key only the SEP knows, so security-wise it's basically the same
<j`ey>
the SEP does have some storage though
<nicolas17>
yeah I think it has some tiny tamper-proof EEPROM?
m6wiq has quit []
axenntio has quit []
Major_Biscuit has quit []
LinuxM1 has quit [Quit: Leaving]
doublerye[m] has joined #asahi
kelito has quit [Quit: Konversation terminated!]
facekapow[m] has joined #asahi
Ry_Darcy has quit [Remote host closed the connection]
m6wiq has joined #asahi
patrikgy has joined #asahi
Tano70 has quit [Quit: Leaving]
user982492 has joined #asahi
<doublerye[m]>
kode54: can you provide some direction on Telegram Desktop rebuild you did?
<psykose>
i think the default tdesktop build vendors jemalloc, so it would probably just need a rebuild with no changes to work
<psykose>
just guessing though
<tpw_rules>
is there a way to query a jemalloc-enabled binary of whatever for the page size it was built with?
a1eaiactaest has joined #asahi
thk[m] has joined #asahi
a1eaiactaestCLONE has joined #asahi
a1eaiactaestCLONE has quit []
<thk[m]>
Hi, so does anyone have asahi up and running with m1-mini
<geerlingguy>
ooh randomly the built in speaker started working when I opened SuperTuxKart
<nicolas17>
uh
<nicolas17>
geerlingguy: you're on the official asahi-arch build?
<nicolas17>
I thought the built in speaker was disabled in the devicetree
<tpw_rules>
only on the macbooks
<geerlingguy>
(I'm on M1 mini)
<tpw_rules>
and even then i think only on the pros
<tpw_rules>
i don't think there's dangers with the mini
<nicolas17>
oh!
<thk[m]>
just tried installing on mini, display won't wake up?
<tpw_rules>
thk[m]: some HDMI monitors are not compatible because there is only one chance to connect on boot
<tpw_rules>
make sure the display is on and listening on the HDMI port before you turn on the mac
<tpw_rules>
you might be able to configure your display to turn off auto switching and make that easier
<tpw_rules>
if you can't get that to work, you'll have to try another monitor or wait til DCP support is merged
<thk[m]>
tpw_rules: ok, i have one 32" 4k Samsung and 32 "qhd Lenovo, but i have to try again
a1eaiactaest has quit [Ping timeout: 480 seconds]
<nicolas17>
thk[m]: it didn't work ever? or did you turn it off and now won't go back on or something?
<bmentink[m]>
when are DisplayPort adapters via USB3 likely to be supported?
<ChaosPrincess>
soon(tm)
<bmentink[m]>
cool!!
<thk[m]>
nicolas17: well after i choose Asahi and wait it to boot, display goes to sleep without signal
<tpw_rules>
yeah, your display probably just isn't compatible
<tpw_rules>
it should boot into asahi by default now
<nicolas17>
ah so it doesn't ever start, then yeah what tpw said
<tpw_rules>
one of my monitors only works if i turn the mini and monitor on at the same time
<bmentink[m]>
Is ethernet adapters over USB3 being worked on as well?
<AdryzzOLEDEdition[m]>
when USB3 support rolls in all USB3 devices will work out of the box
<bmentink[m]>
Great, looking forward to that day ... :)
<Chainfire>
I use a 5G ethernet adapter on my MBP15. It's currently limited to USB2 speed due to this port not having USB3 support yet, but I see no reason why it wouldn't work once USB3 is up
<thk[m]>
well more monitor testing tomorrow..time to sleep
r0ni has joined #asahi
<Chainfire>
As I understand, macOS on the M1 only runs 64-bit code... can we run it in Asahi (if all dependencies were met) or is this a chip limitation?
<j`ey>
chip
<doublerye[m]>
psykose: used asp to checkout telegram-desktop, it and any other asp checkout only grabs x86-64 arch... not too familiar with this but figure if I could get aarch64 I'd be left to just run makepkg. are you more familiar?
<psykose>
should be fine to just change the arch= line
<dottedmag>
Chainfire: theoretically can be done, but that means implementing drivers for something like QEMU for many M1 peripherals, including GPU. That's a lot of work.
<ChaosPrincess>
qemu-user-mode is your friend
<Chainfire>
dottedmag: even for 32-bit userspace stuff?
<dottedmag>
Well, that means whole Darwin syscall interface, has anyone done that?
<nicolas17>
dottedmag: I think you're mixing up "macos apps on linux" with "32-bit apps" :P
<nicolas17>
there's no such thing as "32-bit ARM macOS apps" even
<dottedmag>
Yeah, I'm definitely missing some context. Sorry.
<j`ey>
Chainfire: the chip doesn't support 32-bit
<nicolas17>
Chainfire: afaik Apple removed support for armv7 (32-bit) instructions on the A11, the M1 chip can only run aarch64 code
<nicolas17>
so you need a CPU emulator just like you would to run x86 code
<Chainfire>
Thanks for the clarification, I thought so.
<nicolas17>
oh finally found something explaining the bits instead of just the field names
<nicolas17>
0001 = AArch64 only
<nicolas17>
0010 = supports both AArch64 and AArch32
<Chainfire>
Then again, if it builds, maybe not a problem. Does anything even use nacl anymore? I thought support was dropped last year in favor of WASM
<nicolas17>
Chainfire: yeah wtf is that, I thought NaCl was abandoned
<j`ey>
yeah thats weird
<nicolas17>
"Although initially Google planned to remove PNaCl in first quarter of 2018, and later in the second quarter of 2019, it is currently planned for removal in June 2022 (together with Chrome Apps)."
<nicolas17>
maybe you can just
<nicolas17>
disable it entirely when compiling
<Chainfire>
oh yeah you can, it is currently disabled for ARM64
<Chainfire>
But if this makes it in, distros are not going to think of disabling it if only the Apple's don't play nice with it. Then again, I don't know if you'd ever encounter an actual issue as long as nacl isn't actually *used*.
<Chainfire>
Weird they're bothering to port at all, so close to removal. Makes my tinfoil hat think something else is going on.
m6wiq has quit []
<r0ni>
so for the past week i had been attempting to cross-compile slackware aarch64 pkgs from a rpi, and well it just wasn't working for me. turns out there's a unofficial aarch64 port already made, only there's no installer ;(
jn has quit [Ping timeout: 480 seconds]
memoryleak_ has joined #asahi
jn has joined #asahi
memoryleak has quit [Ping timeout: 480 seconds]
tych0 has joined #asahi
yuyichao_ has quit [Ping timeout: 480 seconds]
<tpw_rules>
wait, so aarch32 can't just be run o the m1?
<j`ey>
correct
<tpw_rules>
that's kind of disappointing
<tpw_rules>
is that common for aarch64 systems?
<mps>
tpw_rules: didn't tried on m1 but I can run arm32 userspace on some arm64 boards and chromebooks
<j`ey>
it will become more common
<nicolas17>
given that there is a standard cpuid-like bit to say whether the CPU supports 32+64 or only 64, I don't think this is unique to Apple
<nicolas17>
but no idea about "common"
<mps>
hmm, I expected that I could run armv7 in lxc
<tpw_rules>
i wonder how fast emulating 32 bit on 64 bit is then
<r0ni>
i know i tried to run a armv7 binary in the arch install and it def told me "no"
<j`ey>
just recompile :p
<tpw_rules>
i wanted to do an aarch32 build bot at some point
<mps>
oh, m1 can't run arm32 binaries
<nicolas17>
"Armv8-A architecture allows different levels of AArch64 and AArch32 support, for example: 1. AArch64 only designs 2. AArch64 designs that also support AArch32 operating systems and virtual machines 3. AArch64 support with AArch32 at (unprivileged) application level only"