marcan changed the topic of #asahi to: Asahi Linux: porting Linux to Apple Silicon macs | Not ready for end users / self contained install yet. Soon. | General project discussion | GitHub: https://alx.sh/g | Wiki: https://alx.sh/w | Topics: #asahi-dev #asahi-re #asahi-gpu #asahi-stream #asahi-offtopic | Keep things on topic | Logs: https://alx.sh/l/asahi
Dcow_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Dcow_ has joined #asahi
yuyichao has quit [Ping timeout: 480 seconds]
Dcow_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
aleasto has quit [Quit: Konversation terminated!]
<ar>
tpw_rules: interestingly, my 16core 2GHz a72 router (nxp lx2160a) also took 11 minutes (real 11m17.078s/user 80m58.745s/sys 11m22.666s, according to time) >Building the image requires downloading of a large amount of data and compilation of a number of packages, including the kernel. On my six core Xeon laptop, building it took about 11 minutes (90 CPU minutes).
<ar>
(time nix-build -A installer-bootstrap -o installer -j16)
<tpw_rules>
building the kernel itself is about 30% faster on the mac mini than that laptop
<tpw_rules>
a much greater portion of the build time should be the kernel if you aren't cross-compiling though
<tpw_rules>
do you have an m1 mac or are you just interested in benchmarking?
Dcow_ has joined #asahi
<ar>
just interested in benchmarking
<tpw_rules>
i'd be interested if you could just run the kernel itself.
<ar>
i'm not eager to reboot my router as it's, well, my router
<tpw_rules>
sorry, i mean run the build for
<tpw_rules>
i highly doubt that router will boot it due to page sizd
<tpw_rules>
if you need i can look up the exact derivation name
<ar>
yup
<ar>
gonna need that
<Glanzmann>
threerik[m]: /clear
<tpw_rules>
ar: time nix-build --check /nix/store/a6rqww6p083p59wa4bcn5xhlffds75q2-linux-5.16.0-rc5-asahi-next-20211214.drv
<ar>
tpw_rules: what if i already did a nix-collect-garbage -d? ;)
<tpw_rules>
uhhhh, build the installer again?
<ar>
ok
<tpw_rules>
there's no attribute path that i know off the top of my head for the kernel, sorry
<ar>
anyway, gonna report back in the morning; going to get some sleep now
<tpw_rules>
okay, thank you for the input
<tpw_rules>
on the m1 mac mini that command is real 4m32.986s/user 25m5.629s/sys 1m31.528s . that's with all the files on the internal nvme, and i am led to believe the CPUs are not quite running at full power. it's about 10 seconds faster if /tmp is on RAM
Dcow_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
yuyichao has joined #asahi
Dcow_ has joined #asahi
Dcow_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
chadmed has joined #asahi
<amw>
Glanzmann: I have gotten the u-boot nvme booting my linux via grub - but had to copy over EFI/BOOT/grubaa64.efi to bootaa64.efi
<amw>
Otherwise u-boot fails with "Could not read \EFI\: Invalid Parameter" error
<amw>
This is debian bullseye
<amw>
Glanzmann: Did you need to do that - is it in your notes?
<tpw_rules>
amw: that's what the dpkg-reconfigure step is for
<tpw_rules>
teach grub to pass the --removable flag and it will put the .efi in the right place
<amw>
tpw_rules: Ok - thanks I missed that flag in my notes so I expected I didn't do it
kov has quit [Quit: Coyote finally caught me]
kov has joined #asahi
c10l1 has joined #asahi
c10l has quit [Read error: Network is unreachable]
phiologe has joined #asahi
PhilippvK has quit [Ping timeout: 480 seconds]
c10l1 is now known as c10l
marvin24_ has joined #asahi
marvin24 has quit [Ping timeout: 480 seconds]
<ar>
tpw_rules: real 8m58.794s, user 79m28.031s, sys 10m8.486s; for the kernel itself
Guest10325 has joined #asahi
Guest10325 has left #asahi [#asahi]
Guest10325 has joined #asahi
Guest10325 has left #asahi [#asahi]
akemin_dayo has quit [Ping timeout: 480 seconds]
<Glanzmann>
amw: That is being expected.
<Glanzmann>
The problem is that uboot has a very basic efi implementation and expects it to be here EFI/BOOT/grubaa64.efi
<Glanzmann>
jannau: A few days I complained about my mouse jumping arround. I can't reproduce the issue with your kernel tree. Just that you know, you did a very good job on your kernel tree. Whatever you did. (USB hot plug works, trackpad is very good, wifi take 5 applys on top of it with one minor rej).
<Glanzmann>
Only thing thta is missing for me is that nvme does not come up with u-boot. I tried two things to fix that: Copy the dts from your tree over to u-boot and recompile and revert u-boots shutdown nvme patch, but no luck.
<Glanzmann>
jannau: I just did a typing speed test on 10fastfingers.com with 84 wpm, the mouse did not move.
the_lanetly_052 has joined #asahi
jeffmiw has quit [Ping timeout: 480 seconds]
<amw>
Glanzmann: I did the "dpkg-reconfigure grub-efi-arm64" and answered yes to the force install to removable partition and it put the smaller 864K shim as /boot/efi/EFI/BOOT/BOOTAA64.EFI
the_lanetly_052 has quit [Ping timeout: 480 seconds]
<Glanzmann>
amw: I see, but does it boot for you? I tested with the symlink and also with the --removable and both booted for me.
dcol542 has joined #asahi
<dcol542>
Hey all. So I managed to get wifi working last night however the adapter only appears 1050s after boot (seems pretty consistently 1050 seconds). Devices show in `lspci -k` but no adapter showing in `ip link show` nor `iwctl device list`. I took a dmesg log from when it started working: https://pbot.rmdir.de/rqqQedM_7ry_lL5C74uopA
<jannau>
dcol542: disable CONFIG_FW_LOADER_USER_HELPER_FALLBACK and possibly CONFIG_FW_LOADER_USER_HELPER
<Glanzmann>
jannau: I'll test now. Thank you.
<dcol542>
jannau: thanks, will give it a go now. Do I need to re-install modules after this?
<jannau>
Glanzmann: I'm not aware any kernel changes which would change the trackpad behavior, is it possible that there are user space changes on your system which makes disabling the trackpad while typing work?
<Glanzmann>
jannau: Of course that could be, I simply try to reproduce and let you know.
Major_Biscuit has joined #asahi
<dcol542>
jannau: thanks, disabling the first one worked!
Major_Biscuit has quit []
MajorBiscuit has joined #asahi
<amw>
Glanzmann: Yep - after the cp command it booted perfectly.
<mps>
Glanzmann: kettenis: last night I talked on irc with an alpine developer who understand grub a lot better than and he explained what was my issue with grub can't find /boot/grub on proper partition
<mps>
fact is, grub embeds partition number in bootaa64.efi during grub-install, so it can't be simply copied from somewhere
<mps>
after I run 'grub-install --target=arm64-efi --efi-directory=/boot/efi' everything is fine
<Glanzmann>
mps: I see, good to know.
<mps>
and ofc, --removable flag is good idea to add but above invocation was enough
<mps>
grub is big mess, friend told me that debian grub pkg have over *200* patches
<mps>
Glanzmann: did you noticed my msg about boot from usb with u-boot
tertu has quit [Ping timeout: 480 seconds]
<mps>
jannau: you didn't comment my question about keyboard patch which add function keys with Fn+number_keys
<mps>
it is quite usable to me on mb pro
<jannau>
I guess it's an acceptable workaround until we have basic touchbar support. any reason to exclude '-' and '=' for F11/F12?
<mps>
no, just didn't thought about '-' and '=', good catch :)
<mps>
and yes, it is workaround not a 'good/smart' solution
<jannau>
has anyone a 16-inch macbook pro? I'm interested in its keyboard/trackpad product id. reported in macos by `ioreg -l | grep -A5 "Apple Internal Keyboard / Trackpad" | grep ProductID`
<Glanzmann>
mps: Yes, I saw it, thanks.
<Glanzmann>
jannau: as400[m]: I think you have a macbook pro 16-inch, don't you?
eroux has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<mps>
on usb disk it gpt2 and grub is on sda2 partition
<Glanzmann>
jannau: About the mouse jumping. it is the kernel. If you care, I can do you a video.
<Glanzmann>
To show you the effects. With your kernel the mouse holds still while I'm typing. With the asahi take 5 wifi it jumps all over the place when I'm typing.
<kettenis>
mps: the grubaa64.efi from the debian installer ISO doesn't have such a string
<Glanzmann>
mps: Which grub version do you have? I have: 2.04
dcol542 has quit [Remote host closed the connection]
<mps>
kettenis: I have alpine grub 2.06
<jannau>
Glanzmann: no video necessary, I'm not aware of any kernel change I would make responsible for this but I integrated a later dev state of the HID patches in my patchset
<mps>
maybe debian patched this out somehow
<Glanzmann>
jannau: Yep, it might be that.
<Glanzmann>
mps: Checking now.
<mps>
yes. looks like debian bootaa64.efi doesn't have this
<mps>
good that we found what is the issue with different grub packages
Major_Biscuit has joined #asahi
MajorBiscuit has quit [Ping timeout: 480 seconds]
<sven>
ugh
Genbuchan[m] has joined #asahi
aleasto has joined #asahi
<jannau>
lol, Dell's newest xps13 has a 1st gen touchbar without physical Esc button. I wonder if they add support in Ubuntu
<jannau>
that might make it easier to support the macbook pro 13" touchbar
<sven>
:D
<ChaosPrincess>
i thought that is not a toucscreen like in macos, just a bunch of fixed-function capacitive buttons
<ChaosPrincess>
they will probably just connect it to the kb controller and not do snowflake stuff like apple
<jannau>
they switch at least the displayed key depending on Fn between the media keys and Fx
<jannau>
but yes, it doesn't look like there is display behind it like apple's touchbar
<opticron>
looks like what lenovo did on the 4th gen x1 carbon
<opticron>
they got sooooo much blowback
<opticron>
I'm assuming dell will with this, too
<sven>
i'm just glad apple got rid of the touchbar as well
<ChaosPrincess>
that dell laptop looks like 'lets take the worst parts of the macbook and add them to out product'
<opticron>
I have mixed feelings about the touch bar. I could be ok with TB+ESC, but I'd really need the OS/apps to integrate it which I feel like just won't happen
lanodan_ has joined #asahi
<mps>
now I think I made mistake to select mb pro instead of air and reason is touchbar, but pro looked somewhat better
lanodan has quit [Ping timeout: 480 seconds]
<mini>
is it still doing the Nostril Cam(tm)? :)
<mps>
I'm using it with linux, just for some short (about month) used macos to run linux in qemu, nothing else on macos
<sven>
opticron: yeah, i sounded interesting at first but then a few years passed and nothing useful came out of it for me
<sven>
*it
<Glanzmann>
landscape15[m]: Nope, it's not.
<j`ey>
(if you use u-boot)
<Glanzmann>
landscape15[m]: But, you won't have usb-c hot plug and you will not have the new anti jump trackpad code in the current rc kernel.
<Glanzmann>
mps: I see. So when I run the air with macos it is colder than my hands. When I run Linux it is hoter than my hands.
<mps>
if I plug/unplug devices to external hub they always works
* jannau
too. there should be no difference regarding usb-c hotplug. the same commit is in my patchset
<sven>
yeah
<sven>
as long as the i2c nodes, the usb-pd and the dwc3 nodes in the device tree are present and you use that commit usb hotplug should work
<Glanzmann>
sven: For me it does not. For mps it also does not work.
<sven>
then i'm pretty sure you're doing something wrong
<Glanzmann>
Might be. I can send you my kernel config. And tree is there, maybe you can reproduce.
<mps>
sven: sven: CONFIG_USB_DWC3_DUAL_ROLE is not set ?
<sven>
could be
<mps>
thanks, will try to build with this enabled after pause
<sven>
i'm kinda surprised that isn't enabled fwiw
<mps>
sven: yes, that is
<mps>
works
<mps>
probably need some more tests and maybe tweaks
<Glanzmann>
mps: Nice. :-)
chadmed has quit [Remote host closed the connection]
<Glanzmann>
mps: I look forward to marcan rebasing the asahi tree on the latest rc. Than we will have u-boot, wifi, usb-c.
<sven>
CONFIG_USB_GADGET is in arm64 defconfig which should enable CONFIG_USB_DWC3_DUAL_ROLE
<sven>
so i'm really not sure how that happened
<j`ey>
Glanzmann: are you sure youre using the same config between janna's kernel and the wifi/take5?
<Glanzmann>
j`ey: No, I'm not. good point. I'll try to use jannau's config on take5 and see if the trackpad still jumps around.
<jannau>
defconfig is not nice though since we have to at least change the page size for PCIe support
<j`ey>
Glanzmann: and USB hotplug, since it seems maybe youre missing CONFIG_USB_DWC3_DUAL_ROLE too?
<Glanzmann>
j`ey: Yes, I was missing it, too. Because I used mps config. :-)
<Glanzmann>
Compiling.
<jannau>
Glanzmann: make sure SPI_HID_APPLE and HID_APPLEFT are enabled otherwise you will be end up without keyboard and/or trackpad
<landscape15[m]>
Glanzmann: thanks. Also, after installing a module the kernel should be compiled with ```make Image modules dtbs``` right?
<jannau>
the config names changed between in my newer input device series
<Glanzmann>
jannau: Thanks for the heads up. Both were disabled.
<Glanzmann>
landscape15[m]: Are you using Debian or another Distro? If you're Using Debian I always build a kernel package: make -j $(( 2* `nproc`)) V=0 bindeb-pkg; For another Distro, what you said. And don't forget to do 'make install' or at least 'make modules_install' and copy the kernel where it belongs to.
<jannau>
would it make sense to have a apple_defconfig in linux-asahi? not intended to submitted upstream
<sven>
jannau: yeah.. i'll also have to get back to that series that will allow PCIe with 4k pages
<Glanzmann>
jannau: I see. I wondered why t was disabled.
<Glanzmann>
jannau: You have my vote for it, since I screw up every other kernel config ...
eroux has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<mps>
I see this in your log 'xorg-server 2:1.20.13-3', somewhat outdated
pent1ckel has quit [Ping timeout: 480 seconds]
<mps>
could you upgrade to 21.1.X series
___nick___ has joined #asahi
___nick___ has quit []
___nick___ has joined #asahi
yuyichao has quit [Ping timeout: 480 seconds]
<Glanzmann>
mps: I could. But to be honest, I'll try to get back to Debian stable. Currently I'm on testing.
<Glanzmann>
mps: What is strange with jannaus kernel it is not jumping with the asahi tree it is. So either it is a config option or something went into rc4,rc5, or rc6 that makes it behave better.
<Glanzmann>
jannau: Perfect. I just checked, my Debian testing as that exact file. I just tried debian stable but there is libinput so old. That I can't move the trackpad, but only press the left key. So I'll stay on testing.
<povik>
i don't understand the mca1 pwrstate issue that mps has reported happens on macbook air
<povik>
> apple-pmgr-pwrstate 23b700000.power-management:power-controller@2c0: PS mca1: Failed to reach power state 0xf (now: 0x24f)
<povik>
marcan: how can 0xf be a valid target pstate?
<povik>
maybe i am reading old pwrstate driver code
<povik>
oh wait my bad
<povik>
i miscounted the number of bits in 0xf...
<mps>
povik: this is on pro, j293 model, but maybe is it same on air
<povik>
ok, should not matter
<mps>
povik: if you create fix please post it, I will test
<povik>
mps: can you post output of tools/dump_pmgr.py on fresh boot of m1n1?
<povik>
i would like to understand how this happens, should be able to reproduce on mini
<mps>
povik: let me see, I think I did already posted it here, have to look where is the file, give 10 minutes
<povik>
sure, no hurry, at your convenience
<Glanzmann>
povik: Do you need that also for the air?
<tpw_rules>
does the m1 have performance counters?
<maz>
tpw_rules: it does.
<maz>
tpw_rules: just not the architected version.
<tpw_rules>
i don't know what that means
<maz>
tpw_rules: M1 doesn't implement the ARMv8 PMU. instead, it has a home grown PMU that doesn't work like the standard one, and doesn't count the same things.
<tpw_rules>
so architected == in the ARMv8 architecture
<maz>
yes
<tpw_rules>
is it something that eventually can be wired into `perf`? (or maybe has and i haven't turned on the appropriate kernel options?)
<maz>
tpw_rules: yes. I have a driver on the list that does exactly that. the main issue is that the documentation for the various events is extremely sparse.
<maz>
what we don't (and won't) have is direct access to the counter from userspace (that'd be a new ABI, and I'm not going there), nor PMUs being exposed to KVM guests (for the same reasons).
the_lanetly_052 has joined #asahi
loki_val has joined #asahi
crabbedhaloablut has quit [Ping timeout: 480 seconds]
malvo has quit [Remote host closed the connection]
malvo has joined #asahi
Major_Biscuit has quit [Ping timeout: 480 seconds]
jkkm has joined #asahi
Major_Biscuit has joined #asahi
Major_Biscuit has quit [Ping timeout: 480 seconds]
<tpw_rules>
hm, what speed do the power and efficiency cores run at currently?
<jannau>
both at ~2 GHz
<maz>
and with a tiny m1n1 hack, you get 3GHz on the p-cores.
<tpw_rules>
what is that hack
the_lanetly_052 has quit [Ping timeout: 480 seconds]
<tpw_rules>
doing some rudimentary benchmarking and i'm curious now
<tpw_rules>
sounds easy enough to me, thanks
<tpw_rules>
any penalty aside from shorter battery life on the laptops
<maz>
dunno, I don't do laptops.
<maz>
but given that my mini draws less than a 20th of my normal box for half the speed, I see an immediate benefit.
<maz>
(it also has only a 20th of the memory, but hey...)
<tpw_rules>
what is your normal bo
<tpw_rules>
x
<maz>
HiSi D05, 64xA72, 256GB of RAM. quite an odd/old machine.
<tpw_rules>
ah, it's an aarch64
<maz>
yeah, I don't do x86 either.
<maz>
(except for $work)
<tpw_rules>
valid
<kettenis>
maz: thanks, didn't get around to make that change muself
<kettenis>
shaves another 20 seconds off my kernel builds ;)
<maz>
kettenis: that's courtesy of marcan (I really wouldn't have found that myself!).
<j`ey>
maz: im surprised the commit message wasnt 'speed-freak boogie' :P
<maz>
j`ey: I don't *only* communicate using FZ songs!
<j`ey>
:-)
<landscape15[m]>
<Glanzmann> "landscape15: Are you using..." <- Thanks a lot. So first I should do ```make modules_install``` before or after compiling the kernel?
<Glanzmann>
after, it install the modules you have compiled into your system so that modprobe, insmod and friends can load them on demand.
<Glanzmann>
installs*
<as400[m]>
Glanzmann: I have 14-inch mbp.
<as400[m]>
jannau: if it's of any help i can check whatever you want. But on 14-inch.
<jannau>
thanks, I can check 14-inch myself. it's not that important
<as400[m]>
jannau: ok. As to the apple_defconfig you also have my voice.
<as400[m]>
I mean in linux-asahi
<Glanzmann>
I need to maintain a list of devices everyone has, I'm constantly get confused.
<as400[m]>
Glanzmann: heh, wait for marcan releasing installer :) You will have to create database for keeping such a list.
<mps>
today I created git repo for keeping config changes :)
<Glanzmann>
mps: I should have done that a long time ago. Also I should backup my linux m1 installation, instead I wipe the system every other month and don't recall how I got the trackpad working ...
joske has joined #asahi
<joske>
does anyone have a patch for vscode to work with 16k pages? I recall someone had one for chromium?
darkapex1 has joined #asahi
<mps>
Glanzmann: I just looked at jannau config and merged some of them
darkapex has quit [Ping timeout: 480 seconds]
___nick___ has quit [Ping timeout: 480 seconds]
joske has quit [Remote host closed the connection]
jeffmiw has joined #asahi
roxfan has quit [Read error: Connection reset by peer]
roxfan has joined #asahi
Nspace has quit []
Nspace has joined #asahi
jochum_ has quit [Remote host closed the connection]
jochum has joined #asahi
<Glanzmann>
mps: I see, maybe, I have to retry then. Can you post your git repository link or config somewhere?
<mps>
Glanzmann: I don't have public git repo, latest config is here https://tpaste.us/VELa
<Glanzmann>
I see, no issue.
<Glanzmann>
mps: Thnak you. And have a good night sleep.
<Glanzmann>
mps: So I always download with a browser, than upload it again and than upload it to somewhere else.
hellstabber[m] has joined #asahi
hellstabber[m] is now known as Eren[m]
yuyichao has quit [Ping timeout: 480 seconds]
<jannau>
https://github.com/jannau/linux/tree/asahi-defconfig - not yet tested but should enable 16k page size and drivers for all supported HW, based on the arm64 defconfig so there are still tons of drivers which are not useful
<jannau>
also requires installed modules for non essential HW