ChanServ changed the topic of #aarch64-laptops to: Linux support for AArch64 Laptops (Chrome OS Trogdor Devices - Asus NovaGo TP370QL - HP Envy x2 - Lenovo Mixx 630 - Lenovo Yoga C630 - Lenovo ThinkPad X13s - and various other snapdragon laptops) - https://oftc.irclog.whitequark.org/aarch64-laptops
<nirik>
yep. Should be.
altacus__ has joined #aarch64-laptops
altacus_ has quit [Ping timeout: 480 seconds]
<steev>
dmesg output?
<steev>
illegal instruction should show up there
<nirik>
I no longer have that boot in front of me. ;) I can get more info from it next time I do.
<nirik>
the nvme drive I swapped in is hanging on boot, so got to sort that first.
<nirik>
but... it's been a long day, so perhaps tomorrow...
<steev>
relatable
<FarchordSteveCossette[m]>
Nirik spends alot of time working for the fedora project, we all appreciate his hard work. Already impressive that he spends all day bashing on a computer just to spend his evenings bashing on a computer! 😁
<robclark>
sigill could be just "disk" corruption
hexdump01 has quit [Ping timeout: 480 seconds]
mcbridematt has joined #aarch64-laptops
<HdkR>
sigill can be a couple of things, including undefined behaviour around writeback instructions
<HdkR>
nirik: Ideally you capture it under gdb and then `disas $pc-64,+128` to get context
<HdkR>
or potentially systemd-coredump
<HdkR>
If the ELF was compiled for the wrong architecture and binfmt_misc wasn't installed, it would atleast just fail on execve :D
kalebris_ has joined #aarch64-laptops
kalebris has quit [Ping timeout: 480 seconds]
kalebris_ is now known as kalebris
xroumegue has quit [Ping timeout: 480 seconds]
iivanov_ has joined #aarch64-laptops
iivanov has quit [Read error: Connection reset by peer]
rfs613- has joined #aarch64-laptops
checkfoc_us has quit []
checkfoc_us has joined #aarch64-laptops
martiert_ has quit [Quit: WeeChat 4.3.5]
rfs613 has quit [Ping timeout: 480 seconds]
iivanov_ has quit [Quit: Leaving...]
martiert_ has joined #aarch64-laptops
iivanov has joined #aarch64-laptops
iivanov has quit []
iivanov has joined #aarch64-laptops
rfs613- has quit [Ping timeout: 480 seconds]
rfs613 has joined #aarch64-laptops
PeterRobinson[m] has joined #aarch64-laptops
iivanov_ has joined #aarch64-laptops
srinik has joined #aarch64-laptops
iivanov has quit [Ping timeout: 480 seconds]
iivanov_ has quit [Remote host closed the connection]
flokli has quit [Quit: WeeChat 4.3.5]
iivanov has joined #aarch64-laptops
iivanov has quit [Remote host closed the connection]
flokli has joined #aarch64-laptops
iivanov has joined #aarch64-laptops
<Molyuu[m]>
<konradybcio> "Molyuu: in real life, do you see..." <- just irregular garbage. i guess it might be caused by panel driver. but after i tried to disabled mdss, it won't boot with simple framebuffer. in fact it never boot with simple-framebuffer, it is weired that screen keeps white with simple framebuffer, and usb controller reset again and again. it only boots with mdss on, and display nothing, or display garbage. no error related
<Molyuu[m]>
This situation makes it difficult for me to determine where the problem is. If it is the Panel Driver, then I only need to modify the Panel Driver, but in fact the system cannot even boot using simple-framebuffer. Even if I use the minimum Device Tree with Vanilla Kernel, the kernel cannot boot using simple-framebuffer.
iivanov has quit [Ping timeout: 480 seconds]
iivanov has joined #aarch64-laptops
iivanov has quit [Remote host closed the connection]
iivanov has joined #aarch64-laptops
hexa- has quit [Quit: WeeChat 4.2.2]
hexa- has joined #aarch64-laptops
iivanov has quit [Remote host closed the connection]
<fomos[m]>
Anyone here having a lenovo yoga slim 7 X?
<fomos[m]>
I shortly tried to boot a live linux distro(pop_os) but was unsuccessful
<fomos[m]>
Any help would be highly appreciated
<robclark>
fomos[m]: several of us do.. but I wouldn't expect distro image to work out of the box, but we aren't far. Right now you would need to build your own kernel. At this point a single dts hack on top of linux-next should work
<nirik>
Yeah, so it's gnome-shell getting the sigill and causing a trap crash for gdm... not much to the core tho: https://paste.centos.org/view/9c6dda13
iivanov has quit [Remote host closed the connection]
iivanov has joined #aarch64-laptops
<FarchordSteveCossette[m]>
<robclark> "fomos: several of us do.. but..." <- Actually, if the stars align properly a distro image works ootb, Fedora's does. But as I said, stars have to align XD
<FarchordSteveCossette[m]>
You essentially need the latest 6.11rc with the right modules baked in lol
<FarchordSteveCossette[m]>
or yes, build your own kernel
iivanov has quit [Remote host closed the connection]
iivanov has joined #aarch64-laptops
Rayyan has joined #aarch64-laptops
<JosDehaes[m]>
How does it find the right dtb then? 🤔 And is all the firmware already upstream?
<robclark>
you still need to copy some of the fw from windows.. and you might still need to patch the ath12k fw
<FarchordSteveCossette[m]>
JosDehaes[m]: the x1e dtbs are in the 6.11 kernel
<FarchordSteveCossette[m]>
But no the firmware files have to be hacked in from windows
<FarchordSteveCossette[m]>
travmurav[m]: Mmmm... we might be able to have them implement it in the arm image installer
<FarchordSteveCossette[m]>
JosDehaes[m]: Oh, I specify it manually. But same thing applies to jhovolt's kernel
<JosDehaes[m]>
FarchordSteveCossette[m]: Yes but I mean how does the bootloader find the dtb
<JosDehaes[m]>
Yes I just understood you booted a distro just like that 😅
* travmurav[m]
wonders what should happen for distros like fedora to ship dtbloader (or equivalent mechanism) in the arm installer images
<travmurav[m]>
I guess last time we discussed it, they wanted to assume ebbr compliance
<travmurav[m]>
that is, prebuilt image available to download
<travmurav[m]>
fwiw we're currently making some changes in pmOS experimental os image so there will be a weekly-ish build of pmOS with linux-next and dtbloader integrated, though it may need broader testing and fixups to actually work well on devices
<fomos[m]>
<travmurav[m]> "fwiw we're currently making some..." <- Would that be available under https://postmarketos.org/install/ ?
<travmurav[m]>
yes-ish, this is a curated list and the image is intentionally hidden
<travmurav[m]>
FarchordSteveCossette[m]: well, that image is /very/ experimental and i.e. if you dd the current one to the usb stick, it won't even be recognized, let alone boot xD
<FarchordSteveCossette[m]>
JosDehaes[m]: What about gluten?
<FarchordSteveCossette[m]>
travmurav[m]: Well yes, booting right now "works" but it's subject to alot of restrictions XD
<JosDehaes[m]>
Certified gluten free and vegan
<travmurav[m]>
I guess if you have a storage with 4k native sector size, you could try xD
<FarchordSteveCossette[m]>
tbh, I might bring my x1e laptop when I travel to VA in september.... the long battery life is very enticing
<travmurav[m]>
(...) but the idea is to try and make that image actually "just working" eventually
<FarchordSteveCossette[m]>
Yep
<travmurav[m]>
thus dtbloader being integrated
<FarchordSteveCossette[m]>
So there's no hope in the laptops actually not requiring a dtb huh?
<FarchordSteveCossette[m]>
i.e. using ACPI
<travmurav[m]>
I'd not count on that
<robclark>
I suspect with some work basic acpi boot should be possible.. I think just no one bothered
<FarchordSteveCossette[m]>
robclark: It would be nice for the end goal to be "Pop the iso on a usb key, plug it in, bam it boots", à la x86_64
ellyq has quit [Remote host closed the connection]
<robclark>
travmurav[m]: I guess the question would be if distro's were willing to integrate dtbloader, or if that must remain a separate step. Fedora arm64 img has a bunch of crap for r-pi so seems like they should be willing to add stuff for qc laptops, but idk
<robclark>
right
<travmurav[m]>
yeah
<travmurav[m]>
tbh I'm a bit confused with fedora rn since I think 39 had iso build and 40 only has img
<robclark>
I think both exist but I never remember how to find the iso
<travmurav[m]>
but well, if they agree to drop an extra .efi there, that would be an easy win for everyone I'd think
<travmurav[m]>
(though still idk how to load drivers in grub, sd-boot its super trivial)
<robclark>
I suppose dgilmore might have some opinions about getting dtbloader into fedora images and isos?
ellyq has joined #aarch64-laptops
travmurav[m] has quit [Ping timeout: 480 seconds]
<dgilmore>
F40 the iso installer failed to build, as it's not release blocking it's not available. I have the last nightly one that built
travmurav[m] has joined #aarch64-laptops
<dgilmore>
robclark: I'm not sure how it can be made to fit in with shim and friends. I also don't know enough about how it determines which one to load.
<dgilmore>
I want to change up the iso creation process to actually include the DTB files in the isos boot files so it can be loaded.
<robclark>
idk if this is still the case but earlier iteration of DtbLoader could be installed as an efi driver, so it would run before shim.. (but travmurav[m] would know the latest).. DtbLoader would use CHID/etc to pick which dtb to load. And perhaps eventually a fwupd plugin could be used to distribute latest dtb
<travmurav[m]>
how it determines <- new dtbloader keeps a map of smbios hwids to canonical dtb name
<robclark>
btw, where is git for latest dtbloader these days?
<travmurav[m]>
I'm squatting the name since the old one is 3 years unmaintained and I think people may as well forget about it
<travmurav[m]>
I'm trying my best to keep up with the new devices and collect hwids from everyone willing to share them, currently I think we're still on-par with linux-next
<robclark>
thx
<robclark>
travmurav[m]: nit, I think the name is "Yoga Slim 7x" (I think there is a 7i which is intel)
<travmurav[m]>
oh I see, will fix rn, thanks
<robclark>
np
<nirik>
I thought there was some kind of loader already? at least theres a /boot/dtb-uname -r/ tree on the image and if you boot without specifiying it still seems to load some dtb?
<steev>
depending on device UEFI firmware, that may or may not work.
<robclark>
I think kernel `make ... dtbs_install` puts dtbs there..
<robclark>
still up to _something_ to figure out which dtb to load.. which is where dtbloader comes in
<travmurav[m]>
I think it's just x13s that has any accommodation to load dtb into system table as of today?
<robclark>
x13s is the only one I'm aware of
<robclark>
so far, at least
<HdkR>
Three months later and you still can't customize the T14s :|
TSIN[m] has joined #aarch64-laptops
<TSIN[m]>
Can vivobook s15 support boot from usb
<TSIN[m]>
?
<TSIN[m]>
s///
<TSIN[m]>
* Can vivobook s15 s5507q support boot from usb?
<TSIN[m]>
* Can vivobook s15 s5507q support boot from usb?
<TSIN[m]>
I didn't find a menu to select boot device from BIOS.
<TSIN[m]>
* Can vivobook s15 s5507q support boot from usb?
<TSIN[m]>
I didn't find a menu entry to select boot device from BIOS.
<robclark>
I expect it can.. on lenovo hit 'f12' when boot splash shows and it will bring up the boot option menu. Idk if vivobook is the same key or not
<bamse>
same key on dell...
<robclark>
I guess you have the xps13?
<bamse>
yes
<TSIN[m]>
I hit f12 many times, still boot into windows😂
<bamse>
and i must say i am impressed about the usb support in uefi...it allows me to operate uefi using my keyboard attached over an usb hub :)
<robclark>
I'm not quite sure what to make of the keyboard and trackpad on the xps13
<robclark>
heheh
<bamse>
robclark: me neither...
<TSIN[m]>
I'm trying to make nvn driver on mesa freedreno.
<bamse>
robclark: that said, i expected having trouble hitting the right keys...but so far that's pretty okay
<bamse>
robclark: that said, i don't have a proper installation on it yet, so too early to really say
<HdkR>
One thing I noticed on the Yoga. Once grub boots, it takes about 30 seconds to enumerate my USB hub attached before it starts booting. Which I never noticed on the X13s
<HdkR>
A bit spicy
<robclark>
somehow in grub, the builtin keyboard on my 7x doesn't work for me (but if I plug in a usb kb that works fine).. maybe it is something about fedora's grub, but I assumed it is just using efi proto's for input
<bamse>
yeah, i too assumed grub would just use efi services for that
<bamse>
i've only used systemd-boot so far, and that works really well
<HdkR>
Interesting, keyboard seems to work fine-ish here. If I end up in EFI shell then it is laggy and drops keystrokes, but that feels like expected behaviour
<robclark>
hmm, I did have sd-boot on my "recovery" usb-c img, and it had same problem
<nirik>
builtin keyboard works for me in grub... but it's pretty slow/laggy. If you try and type fast it drops a lot
<bamse>
one thing i haven't managed to do though is launch efi shell...chainloading that just reboots the machine
<robclark>
huh, neat
<robclark>
I did get shell.efi to load (but it also had the no-kb issue)
<robclark>
idk if lenovo has posted any fw updates for 7x yet
<bamse>
i'll take a swing at getting these test cases added to the boot teams list for the next platform ;)
<robclark>
you'd think there should be some sort of uefi compliance test
<FarchordSteveCossette[m]>
nirik: To be fair, that also happens on my x86_64 laptop too on Fedora XD
<FarchordSteveCossette[m]>
You even see the cursor scanning
<bamse>
robclark: perhaps that would be sufficient :)
<JosDehaes[m]>
I was able to load shellaa64.efi from a USB drive
<JosDehaes[m]>
I think I did have to try several versions of the file
<bamse>
perhaps there was some user error involved here...
<JosDehaes[m]>
And I was able to add boot entries in the UEFI
<bamse>
i'm hoping that merging the bug fixes in tzmem will allow me to modify boot entries from within the os
<JosDehaes[m]>
<robclark> "somehow in grub, the builtin..." <- Huh this has always worked for me
<JosDehaes[m]>
It is very slow though
<robclark>
yeah, somehow my 7x seems to be "unique"
<TSIN[m]>
* Can vivobook s15 s5507q support boot from usb?
<TSIN[m]>
I didn't find a menu entry to select boot device from BIOS.
<TSIN[m]>
Farchord (Steve Cossette):
JasonMontleon[m] has quit [Ping timeout: 480 seconds]
ellyq has quit [Remote host closed the connection]
JasonMontleon[m] has joined #aarch64-laptops
<craftyguy>
ah nice, the updated ath11k fw for the x13s was merged into linux-firmware (the fw that fixes power management for me when connected to some networks)
<bamse>
ohh, i can see my name there...so i guess that answers the question
<steev>
i see it here with my AT&T connection, but not with Spectrum
srinik has quit [Ping timeout: 480 seconds]
weirdtreething_ has joined #aarch64-laptops
weirdtreething has quit [Ping timeout: 480 seconds]
weirdtreething_ is now known as weirdtreething
iivanov has joined #aarch64-laptops
weirdtreething has quit [Ping timeout: 480 seconds]
weirdtreething has joined #aarch64-laptops
<dgilmore>
nirik: there is not /boot/dtb-`uname -r`/ on the install iso. which if you want to encrypt the disk you have to use
<nirik>
hum, well, I was looking at Fedora-KDE-Rawhide-20240813.n.0.aarch64.raw.xz and it has them...
iivanov has quit [Remote host closed the connection]
<dgilmore>
nirik: yeah thats a disk image, they do. but you can not encrypt them. the isos made by lmc do not have them
<nirik>
ah, ok
<nirik>
on the minus side, I think I need some patches to see the nvme (the stock rawhide kernel doesn't seem to detect it at all, which is likely why my nvme boots hang).
<nirik>
on the plus side, suspend/resume just worked fine.
kuruczgy[m] has joined #aarch64-laptops
iivanov has joined #aarch64-laptops
iivanov_ has joined #aarch64-laptops
iivanov has quit [Read error: Connection reset by peer]
<kuruczgy[m]>
Hi! I am experiencing a crash on my yoga 7x during systemd booting, I saw that other people ran into similar issues as well, did this one get figured out? (If I do `echo 1 > /proc/sys/kernel/modules_disabled` after initramfs but before systemd I can boot, so it's definitely some module that's getting loaded.)
<kuruczgy[m]>
Tried module_blacklist=msm,qcom_edac,qcom_q6v5_pas so far, but it's still crashing so it's not one of these.
<steev>
it *should* be qcom_edac
<steev>
robclark seems to know the most tho, so i don't know
<agl>
steev: Here is it 00:28 o'clock. I'am going to bed. Good Night!
iivanov_ has quit [Ping timeout: 480 seconds]
<steev>
night!
<robclark>
well, bamse was the one who tracked that down (thx to ramdump magic).. but yeah edac needs to be disabled
<robclark>
my guess is fw wants to control edac on windows
<kuruczgy[m]>
I tried with and without module_blacklist=qcom_edac just to double check, and it seems that there is a slight difference, without the blacklist the machine reboots, while with it the screen just goes black but there is no reboot.
<kuruczgy[m]>
I am also poking around the kernel log to see what stands out, I get a bunch of messages like:
<kuruczgy[m]>
qnoc-x1e80100 interconnect-1: sync_state() pending due to 30000000.remoteproc
<kuruczgy[m]>
I guess it does make sense that a lot of things are not loaded because only a few modules (and no firmware) are available in initramfs, could this be causing any issues?
<kuruczgy[m]>
What modules do you guys have in initramfs? I have added phy_qcom_qmp_combo, phy_qcom_snps_eusb2, phy_qcom_eusb2_repeater, tcsrcc_x1e80100, i2c_hid, i2c_hid_of, and i2c_qcom_geni so far.
<robclark>
if you are booting from nvme, you shouldn't need anything to exotic to mount rootfs... booting from usb is a bit more involved
<kuruczgy[m]>
Yeah I am booting from USB so that's why I need those modules. But with those it seems to work well, haven't had any issues with usb.
<kuruczgy[m]>
But so IIUC you don't have anything other than the bare minimum in initramfs, and having most modules (and all firmware) only on rootfs should be fine?
<robclark>
in theory.. I guess I still have a bunch of stuff as =y instead of =m since early bringup.. I guess I should switch back to more modules.. FarchordSteveCossette[m] had rawhide distro kernel config which he got working (just replacing the dtb)
<FarchordSteveCossette[m]>
And blacklisting two modules
<FarchordSteveCossette[m]>
Pretty much
<nirik>
rawhide kernel (kde image) works for me from usb... but it doesn't seem to see nvme... likely needs some out of tree patches for that I guess.
<kuruczgy[m]>
robclark: Could you by chance share your kernel + config? I would be interested in trying `=y`ing some stuff based what you have.
<kuruczgy[m]>
Farchord (Steve Cossette): "two modules"? Which one other than qcom_edac?
<FarchordSteveCossette[m]>
<nirik> "rawhide kernel (kde image) works..." <- Have you tried the « throw sh…tuff at the wall and see what sticks » approach? Basically, try modprobing qcom kernel modules and watch to see if your drives show up
<nirik>
<FarchordSteveCossette[m]> "Have you tried the « throw sh…..." <- I haven't. I suppose I could... no idea what one(s) could be responsible there tho...
<FarchordSteveCossette[m]>
nirik: There's a couple qcom modules relating to nvme