ChanServ changed the topic of #aarch64-laptops to: Linux support for AArch64 Laptops (Asus NovaGo TP370QL - HP Envy x2 - Lenovo Mixx 630 - Lenovo Yoga C630)
baozich has quit [Ping timeout: 480 seconds]
<jenneron[m]>
<Jasper[m]1> "From the things I did with pmos:..." <- no
<steev>
it should kind of go without saying... if you're doing **full disk encryption** that you cannot load things from that encrypted disk **before you unlock it** so, you need everything in the initramfs that you need to use the system before hand, display, firmwares, input devices, so on
derzahl has quit [Read error: No route to host]
<jenneron[m]>
yeah
<jenneron[m]>
pmOS can be a good reference for it, I think I got FDE to work on galaxy book s
<jenneron[m]>
and I did almost the same on flex 5g with debian
<clover[m]>
Nice. Always respect the FDE people. Can't really do it full booting windows I don't think?
<clover[m]>
Dual booting*
<steev>
sure you can
hexdump0815 has joined #aarch64-laptops
hexdump01 has quit [Ping timeout: 480 seconds]
<Jasper[m]1>
<jenneron[m]> "no" <- Panel and input no?
hexdump0815 has quit [Quit: WeeChat 1.9.1]
hexdump0815 has joined #aarch64-laptops
<hexdump0815>
ungeskriptet: my first try to get ufs working with a self built kernel based on your sources did not work
<hexdump0815>
ungeskriptet: which pmaports tree did you use to build it - then i can see from the APKBUILD file how you built it
<jhovold>
Jasper[m]1: I've listed the modules needed for keyboard and display in the commit message for "johan_defconfig: add custom config"
<jhovold>
Been a while since I tested it myself so it is possible that list is no longer complete, also depends on the config you use, which could theoretically have more dependencies built as modules than with johan_defconfig
<jhovold>
ideally we should not need all of this in the initramfs, but there's work to be done (e.g. in regulator core) before that can happen
<ungeskriptet>
hexdump0815: I used the device-samsung-w767 package as a base and changed/renamed the device specific values
<ungeskriptet>
I can upload it later
<jhovold>
clover[m]: the speaker volume is low currently as the audio configuration is not entirely correct and relies on digital gain where it should not. If we were to crank the volume the distortion becomes unbearable. srinik is working on fixing this properly, which should allow increasing the default volume level
<ungeskriptet>
hexdump0815: Do the UFS partitions show up in dmesg?
<konradybcio>
6.5-rc1 is broken for me.. if i run watch on /proc/irq and start a stress-ng stresstest, i get a segfault on the watch and a hard system stall after a couple of minutes
<konradybcio>
sorry /proc/interrupts obviously
<jhovold>
hmmm. i just verified that by simply catting /proc/interrupts, something's broken alright
<konradybcio>
For me it took a while, maybe until the first update.. to be more precise i did a watch -n1 cat with a grep..
<konradybcio>
Not by any linux machine rn though
<jhovold>
que?
<maz>
/proc/interrupt exploding makes me suspect the new mapple-tree based backing store for the irqdescs...
<maz>
probably worth having a look (though nothing seem to explode on a M2...)
<jhovold>
yeah, i just reverted 721255b9826b ("genirq: Use a maple tree for interrupt descriptor management")
<jhovold>
can't seem to reproduce it without, but then again, does not happen on every access with either it seems
<maz>
what system is that on?
<jhovold>
the x13s
<maz>
don't have that in my zoo -- I'll try to reproduce it on another system.
<konradybcio>
jhovold rn = right now, I meant that I can't investigate further for the time being
<jhovold>
konradybcio: got it
<jhovold>
maz: just hit it again with a simple cat, had had the machine idling for a few minutes before trying to reproduce it this time
<maz>
jhovold: with the suspected patch?
<jhovold>
yes, with the maple tree patch
<maz>
can you dump the trace somewhere?
<jhovold>
maz: i just sent it to you in a mail with the list on cc
<clover[m]>
Jasper: more info from the other luks guy:
<clover[m]>
In the initramfs, I have to enter the passphrase to decrypt the filesystem (luks). If kms is enabled, I don't see the prompt and the console is back on when I have the login prompt. If I disable kms, I see the passphrase prompt. However, if I wait for like 20 seconds to enter the passphrase, the console goes off and I have the same behavior as with kms. If I don't wait and enter the passphrase without waiting, the console never goes
<clover[m]>
off.
<jhovold>
clover[m]: known issue, regulator core disables the seemingly unused display regulators atfer a 30 s timeout
<jhovold>
until we've reworked that, you need to have all the display driver deps in your initramfs if you want more than 30 seconds of display as I mention in the commit message for johan_defconfig
<jhovold>
Jasper[m]1: ^
<Jasper[m]1>
Thank you, will check it out
Lucanis has quit [Ping timeout: 480 seconds]
hexdump0815 has quit [Quit: WeeChat 1.9.1]
hexdump0815 has joined #aarch64-laptops
<hexdump0815>
ungeskriptet: thanks a lot for the repo link - i'll play around with that maybe later tonight
<hexdump0815>
ungeskriptet: i think ufs was detected with your kernel as my usb drive was discovered as /dev/sdg (usually /dev/sda) and it looks like the other ufs partitions were discovered too
<hexdump0815>
ungeskriptet: the problem just was that with your kernel no keyboard was working, no usb etc. so i could not really interact with the system except taking pictures of the bootup :)
kitsune-tsuki has joined #aarch64-laptops
<hexdump0815>
ungeskriptet: so my plan now is to be able to rebuild your kernel so that a self built version will discover ufs too and then fold in all other of my options and dts entries to kdb, usb, gpu etc. working as well
<hexdump0815>
(maybe for jenneron[m]) in case we get ufs working: is it in theory possible to completely wipe the ufs and only put linux onto it or does the bios/boot somehow rely on some stuff to be at certain places? (i have no real experience with ufs)
<hexdump0815>
i'm not planning this, but it would be interesting to know if it is possible or not
<travmurav[m]>
iirc there were multiple different LUNs on kitsune-tsuki 's laptop
<travmurav[m]>
so presumably you could wipe the OS LUN but not any other
<travmurav[m]>
especially for the samsung laptop I suppose (hardware) secure-boot is on and thus no EDL to recover from bricking it
Lucanis has joined #aarch64-laptops
<travmurav[m]>
(that is, the entire boot chain, uefi and modem data is on multiple ufs LUNs and they are all essential to boot)
<hexdump0815>
travmurav[m]: thx - good to know - i plan to resize win11 maybe and free up some space for linux - but first we need to get ufs working reliable :)
<travmurav[m]>
I guess this is already a bit better compared to emmc where everything is in one big pile, but I think these LUNs wasted quite a lot of space too...
<hexdump0815>
travmurav[m]: out of curiosity: how is that on your aspire 1 with emmc - similar or could you wipe it completely in theory?
<travmurav[m]>
hexdump0815: worse - the secure-boot is on so if the boot chain is gone, it's hard brick. + everything is on the same GPT so messing with partitions is a bit more risky too
<travmurav[m]>
if it was SB-off, I could (theoretically) wipe it and then re-provision the same bootloader partitions to get it back up tho
<hexdump0815>
travmurav[m]: oh - not nice ... so you resized your win with some partitioning tool on win and created a new partition for linux?
<travmurav[m]>
(well, or if I had the signed EDL payload from the vendor, but doubt I can get it)
<travmurav[m]>
hexdump0815: I shrunk the windows FS using windows tools (so it does it best to not beak the ntfs) then created a second ESP and a data partition to install my OS
<travmurav[m]>
so I did add new partitions using gparted (iirc) trying to be careful and not touch anything else
<travmurav[m]>
there are like 50 partitions on this thing
<hexdump0815>
looks like arm chromebooks are so much nicer than woa devices: you have a bootloader which can even boot a linux kernel in flash and can wipe the regular storage without any problem
<travmurav[m]>
it's also delightful to see all of them in the windows disk manager and know that someone without the proper knowledge could "free some space" and hard brick their device
<hexdump0815>
i guess x13s is a bit better as it is using nvme then maybe?
<travmurav[m]>
yeah iirc they have a separate qspi flash for the boot chain
<travmurav[m]>
(both cros and x13s I think)
<travmurav[m]>
so the "important" boot chain is separated from the user
<travmurav[m]>
contrary to the "phone" style where the user was not supposed to access the raw disk
<Jasper[m]1>
<travmurav[m]> "yeah iirc they have a separate..." <- Yep
<hexdump0815>
so woa devices with emmc or ufs are essentially large form factor mobile devices with some kind of mobile windows on them ;)
<travmurav[m]>
hexdump0815: actually, you say your usb stick is now sdg? so the 6 other ones (a-f) map exactly to 6 LUNs that were dumped on that other laptop with UFS (via edl)
<travmurav[m]>
hexdump0815: the 7c is more or less a re-label of a mobile snap 720g at least in my eyes
<travmurav[m]>
(1st, 2nd gen that is)
<travmurav[m]>
so yes, it's pretty much a mobile design slapped into a laptop chassis
<travmurav[m]>
which is not even a bad thing, those platforms got pretty powerful and efficient
<travmurav[m]>
(with 8cx based designs like x13s I think there is a bit more laptop-ness in places, at least with the IO choices)
<konradybcio>
8280 is less of a phone chip.. i'd say advanced tablet ;)
<kitsune-tsuki>
hexdump0815: hi!
<kitsune-tsuki>
travmurav[m]: hi!
<travmurav[m]>
o/
<kitsune-tsuki>
hexdump0815: most of the LUNs are UEFI and various other firmvare
<travmurav[m]>
ungeskriptet was working on UFS for snap 720g and it should be the same IP block as on the 7c1 so he needed someone with UFS on 7c to test the series
<hexdump0815>
i'm happy to test it, especially if ufs support for sc7180 comes out of it :)
<kitsune-tsuki>
UFS and touchpad are the only two things that prevents me from using Linux on my TCL laptop normally
<konradybcio>
let's knock one of them off the list then
<konradybcio>
kitsune-tsuki: what's your laptop model?
<kitsune-tsuki>
throught I would also like to reclaim 4GB that is spent on UEFI for some reason -- but that mean bringing up U-Boot for that platform...
<travmurav[m]>
jenneron: ah so it's the same thing with LUNs on ufs 8cx, great
<jenneron[m]>
pmbootstrap install --sdcard /dev/sdX, where is X is the correct lun should work
<jenneron[m]>
it might be good for kernel to expose something like /dev/ufs0lunX instead of /dev/sdX
<jenneron[m]>
I'm not sure if the order of sdX is random
* travmurav[m]
thinks he would never get over the fear of automatic os installers on arm
<travmurav[m]>
Ignorance is bliss
<kitsune-tsuki>
I really enjoyed my laptop, even on windows
<kitsune-tsuki>
28 hours of reading and writing on one charge
<kitsune-tsuki>
OFC with wifi and wwan offline an on minimal brightness
<kitsune-tsuki>
I can only hope that linux could come close to that
<travmurav[m]>
Probably not too far given I get roughly 12 hours of screen with wifi and mid brightness, or roughly 10 days of suspend from the full battery
<hexdump0815>
ungeskriptet: i'm still not sure how exactly you built your kernel - the pmaports repo seems to still reference the old v5.19 kernel and the kernel config you mentioned yesterday is just a config fragment
<hexdump0815>
ungeskriptet: did you merge that with defconfig?
<ungeskriptet>
I used envkernel.sh from pmbootstrap to compile the kernel
<ungeskriptet>
The sm7125 kernel there is just for packaging
<konradybcio>
fwiw you should only need to build the kernel once and then you can rebult the dt if you don't change anything on the drivers side
<ungeskriptet>
You can merge the fragment to the defconfig with `make defconfig sc7180.config`
<hexdump0815>
that does not work in the regular kernel tree, but i use "scripts/kconfig/merge_config.sh -m arch/arm64/configs/defconfig arch/arm64/configs/sc7180.config" now which should merge them accordingly
<hexdump0815>
oh - i guess just using your sc7180.config as .config and a make olddefconfig afterwards should do the same
<hexdump0815>
ungeskriptet: built another kernel from your tree, using defconfig and your config snippet merged and using your dtb and that one does not see ufs
<hexdump0815>
ungeskriptet: can you please put the .config used for your sc7180 kernel build up somewhere?
<hexdump0815>
ungeskriptet: do you maybe have uncommited changes in your local tree?
<Jasper[m]1>
<jhovold> "Jasper: I've listed the modules..." <- Including these gives me a password prompt :D
<Jasper[m]1>
Only have one error related to adreno firmware I think. It's in the screenshot
<clover[m]>
ok, i will add to my wiki that if you want FDE you will need to add those to mkinitcpio
<Jasper[m]1>
Should probably also replace the link to leon's repo in the readme for archiso with the one from your github
<clover[m]>
i dont see that link anywhere
<Jasper[m]1>
Nvm, was looking at upstream (compared to your x13s-archiso repo)
<Jasper[m]1>
My bad
<clover[m]>
ok cool
<Jasper[m]1>
Crap I forgot to install networking stuff
<clover[m]>
Time to fire up the iso again
<Jasper[m]1>
clover[m]: Base apparently includes systemd-networkd and systemd-resolved so I got it to work in the end
<Jasper[m]1>
On to install more stuff
<clover[m]>
Happy arching
<Jasper[m]1>
clover[m]: Thanks! Got any more tips of packages to install to get hw working more comfortably?
<Jasper[m]1>
pd-mapper and qmic are in aur I saw
<clover[m]>
BT needs a systemd override to set the bt address each boot
<Jasper[m]1>
Also, is efi=novamap,noruntime still needed or just noruntime?
<Jasper[m]1>
clover[m]: Is that on your git somewhere?
<clover[m]>
Don't remember and I'm out of pocket rn
<clover[m]>
Yeah on the wiki
<Jasper[m]1>
Rip, no problem.
<Jasper[m]1>
clover: okay last question, I might be missing something, but `sudo pacman -S mesa-a690 alsa-ucm-conf-x13s` do not work
<Jasper[m]1>
checking your packages repo shows alsa-ucm-conf-x13s-git and regular mesa. Do I just install those then?
<konradybcio>
well you can try and undo it if it breaks stuff :p
<Jasper[m]1>
That's true
<Jasper[m]1>
But turns out it can't find that either
<Jasper[m]1>
Well the alsa config thing
<clover[m]>
Put my repo higher in pacman.conf and just do pacman -S mesa
<clover[m]>
It should take priority that way
<clover[m]>
Jasper[m]1: Yes
<clover[m]>
If you want louder speakers and you want Vulkan fredreno stuff yeah
<Jasper[m]1>
Yes!
<clover[m]>
you can get those with my repo and doing sudo pacman -S vulkan-mesa-layers vulkan-freedreno mesa
<Jasper[m]1>
Got them, thanks! That's about it, the rest is on the wiki. Thank you VERY much so far and will report back if I run into interesting stuff
<clover[m]>
Np glad you got it working finally
<Jasper[m]1>
Now to wait for plasma and kde-applications to install
<Jasper[m]1>
which takes kinda long
<clover[m]>
Why no sway 😕
<Jasper[m]1>
Any idea what aur helper is currently the one to use?
<Jasper[m]1>
clover[m]: I've used i3 for a while and got annoyed by setting it up
<Jasper[m]1>
I'm not that into ricing my desktop
<clover[m]>
I use yay
<clover[m]>
You will have to install it manually then then are good to go
<clover[m]>
Another reason why I like EndeavorOS, they package yay for you
<Jasper[m]1>
clover[m]: Can it self update?
<clover[m]>
Nope, eos comes with an update notifier but otherwise you basically get used to running yay -Syu every day out of habit
<clover[m]>
Oh yeah it does self update once you install manually from aur
<clover[m]>
Git clone it, CD inside and then makepkg
<Jasper[m]1>
clover[m]: Yeah that's what I mean
<Jasper[m]1>
Cool, I'll use that then.
<clover[m]>
Then pacman -U <package>
<Jasper[m]1>
Hope I can speed up downloads in the meantime, there's a bit of a storm coming
<clover[m]>
Uncomment parallel downloaded in pacman.conf
<clover[m]>
Protip right there
<clover[m]>
s/downloaded/downloads/
<Jasper[m]1>
Done that, gonna set the mirror a bit closer aswell
<Jasper[m]1>
Nevermind, there's none closer to me. Forgot alarm has a different set
<Jasper[m]1>
@clover pd-mapper is not cloning for some reason hmmm
<clover[m]>
see my comment on AUR
<clover[m]>
its super annoying
<clover[m]>
should probably flag that package as out of date or something
<Jasper[m]1>
Ah that's what it was about
<Jasper[m]1>
Yeah you can probably edit it in some way no?
<clover[m]>
unless you are a maintainer you can't edit
<Jasper[m]1>
Time to complain in irc/ML
<konradybcio>
Jasper: if you have a spare low power computer, you can set up your own mirror
<konradybcio>
and you totally do :D
<Jasper[m]1>
konradybcio: hurr durr
<Jasper[m]1>
I would, but I don't have an extremely good internet connection
<konradybcio>
I mean after the initial sync there's not that much data flowing
<Jasper[m]1>
konradybcio: Depends on how many people start doenloading
<Jasper[m]1>
*downloading
<Jasper[m]1>
Private mirror could work though
<Jasper[m]1>
Is there anything on video acceleration btw? Missed that on the wiki I think
<konradybcio>
not very many people if you keep it local
<konradybcio>
yes we have venus working on Johan's branch
<Jasper[m]1>
Cool, I'll wait for that to be pushed then
<Jasper[m]1>
Has anyone here figured out if the wwan slot can be used as another nvme slot? Tends to be possible with these lenovo's.
<konradybcio>
check the m.2 key first
<HdkR>
The keying is not SSD keying
<Jasper[m]1>
Looks like M or B
<Jasper[m]1>
HdkR: Ah, sad.
<HdkR>
The WWAN slot is also vendor locked, so you can't really stick much in it.
<Jasper[m]1>
I know about the regulation stuff, but since I don't care for wwan and Lenovo tends to make it possible, it would be worth asking
<Jasper[m]1>
But yeah the keying is wrong
<HdkR>
I can't even remember which keying I slapped in to it now