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)
hexdump01 has joined #aarch64-laptops
hexdump0815 has quit [Ping timeout: 480 seconds]
<steev> bamse: before i test them.. those geni patches... i shouldn't use on the flex if i want a (somewhat) working system?
<clover[m]> New error messages yey
<steev> what does your.... whatever your bootloader uses... look like
<steev> and the files are at that path?
<clover[m]> is there a way to get into a shelll at the boot menu to check?
<steev> i have no idea how systemd-boot works
<steev> it's a live cd though right?
<steev> you should be able to look at the efi partition and see the paths
<steev> since its just vfat
<clover[m]> I copy the files here
<clover[m]> not sure if that's correct. couldn't figure out mcopy from docs so i just tried matching the existing syntax
<steev> weird that is says unsupported
<clover[m]> maybe i need Image.gz?
<clover[m]> like does it expect it to be compressed
<steev> i have no idea what systemd boot requires, no other system requires it
<bamse> steev: they won't break anything, just make things better
<bamse> steev: the problem i have on flex 5g, is that on linux-next as soon as we try to do a DMA operation in the i2c driver we screw up the busses or something...still trying to figure out why that is
falk689 has quit [Remote host closed the connection]
falk689 has joined #aarch64-laptops
<steev> oh, yeah i'm still on 5.19
<steev> and also... something linus pulled in recently broke audio
<steev> not sure what though since nothing sticks out, but the audio device is probe deferred
<steev> soc@0:sound msm-snd-sdm845: MultiMedia1: error getting cpu dai name
<steev> srinik: ^
<clover[m]> <steev> "you should be able to look at..." <- good point
<steev> the other weirdness is, whatever it is has broken wifi on the flex :(
<steev> no devices deferred there though, and the module is loaded, just not getting a device
<steev> clover[m]: hm, okay, i guess you could try pointing it at the .gz
<clover[m]> I tried it, still unsupported T_T
<steev> weird... and it's a vfat partition?
<steev> technically modern systems should support EFI being ext4, but vfat is usually what it is
<steev> clover[m]: what does "file /mnt/arch/boot/aarch64/*" say?
falk689 has quit [Remote host closed the connection]
falk689 has joined #aarch64-laptops
jhovold has joined #aarch64-laptops
<clover[m]> it's some kinda fat. here's the command used to make the image file before copying over the kernel, initramfs, and dtb:
<clover[m]> `mkfs.fat -C -n ARCHISO_EFI "${work_dir}/efiboot.img" "${imgsize}"`
<clover[m]> MS-DOS FAT filesystem...
<clover[m]> i wonder if i need to disable 'fast boot' mode or something
<clover[m]> i noticed in my kernel config file i have
<clover[m]> `# CONFIG_EFI_ARMSTUB_DTB_LOADER is not set`
<clover[m]> now im wondering if that's why dtb isn't loading
<qzed> I don't have that set either and it works with grub
<steev> you might need to enable that
<steev> qzed: he's using systemd-boot which does efi stub stuffs
<qzed> I have no clue how systemd-boot works, so you probably know more than I do
<steev> it just, afaik, uses efistub
<steev> i have that option =y here on my x13s
<steev> not sure why it's disabled there
<clover[m]> can you send me your kernel config steev pretty please
<steev> clover[m]: it should be the laptop_defconfig
<steev> but
<steev> which, apparently doesn't have https in 2022...
<clover[m]> Ty ty
jhovold has quit [Ping timeout: 480 seconds]
<clover[m]> More super helpful output... grr.
<clover[m]> Time to hit the gym
<qzed> I think that's the same output I get without efi=novamap
<steev> at that point it should be jumping to the kernel - you might want to add earlycon=efifb
falk689_ has joined #aarch64-laptops
falk689 has quit [Remote host closed the connection]
<qzed> huh, I didn't know that uefi also provides a rtc interface...
<qzed> in true fashion for our devices, it seems to be unusable though
<steev> there is something about qcom's rtc... i can't remember what
<steev> i found it somewhere in their forums
<qzed> afaik it's not accessible from the unsecure world or something
<qzed> yeah, you've posted that before
<steev> yeah, something along those lines
<qzed> I was able to read from that though... but with the uefi interface not even that is working
<steev> bamse: do you happen to know if 300MHz was left off of the sc8280xp dtsi for a reason? is it broken on !thinkpad ?
<szclsya[m]> steev: I'm currently using your kernel tree and is stuck at no filesystem to mount as root, do you have similar trouble?
<szclsya[m]> (my root fs is currently on a usb drive, probably USB stack issue?
<steev> szclsya[m]: if you aren't using an initramfs, you need to build usb in to the kernel
<steev> if you are... try the other usb port
<qzed> i.e. figure out what kind of magic incantation needs to go into the config, or alternatively what kind of magic incantation needs to go into initcpio.conf (or the equivalent conf)
<qzed> you should be able to trace everything from the '.compatible's in the dts
<szclsya[m]> Hmm... I'm working upon the laptop_defconfig in the tree and seems that it has the USB stuff built in already
<szclsya[m]> Lemme check qualcomm specific drivers
<steev> no, definitely doesn't have it built in
<steev> you might need the phy qcom usb stuff
<steev> at least, the phys aren't
<steev> or maybe it doesn't need them
<szclsya[m]> Just checked my .config, CONFIG_PHY_QCOM* are already on
<szclsya[m]> I'm using ext4 which is also built into the kernel, hmm...
<steev> maybe i should attempt to pull in qzed's usb stuff, but i need to do some other things away from the computer
<szclsya[m]> Wait, just realized I didn't build various TYPEC stuff into the kernel
<szclsya[m]> Lemme do a rebuild real quick
<steev> or throw them into the initramfs, then you wouldn't need to rebuild the kernel
<steev> there are however, comments in the dts to define usb-c connector properly, so it may not be fully working just yet regardless
<szclsya[m]> Right... Should take a look at how to do initramfs manually (I just relied on dracut and mkinitcpio before
<steev> what distro are you using?
<szclsya[m]> I'm using Arch (btw) on my x86 thinkpads
<bamse> steev: in the cpu opp-table?
<steev> bamse: yeah
<steev> i added it in here, and all seems to be okay on the thinkpad at least
<steev> szclsya[m]: i mean on whatever device you're using my kernel on
<szclsya[m]> I'm just manually building the kernel right now
<steev> but which device? lol, i have many trees
<steev> well, branches
<szclsya[m]> Ah sorry, the X13s
<bamse> steev: i believe the opp-table is a dump of the frequencies on my first x13s
<bamse> steev: if there's things missing now that we've hit production, please do send me patches
<steev> bamse: will do :)
<clover[m]> <steev> "at that point it should be..." <- testing
<szclsya[m]> Huh the behavior changes. It panicked before, but now it just gives me a blank screen after some messages on screen without resetting. I guess it might be running as I didn't specify panic=0
<szclsya[m]> Should I extract the firmware first?
<steev> szclsya[m]: same thing i told clover, earlycon=efifb :)
<steev> should boot and be usable without the firmware (minus the charging)
<szclsya[m]> Been using it all along
<steev> you might wanna get with clover though, they're using the mkarchiso thingiemabobber
<steev> i haven't used arch in a long time, minus their wiki, which is stellar
<steev> though, they're doing an install it seems, not just running off usb
<szclsya[m]> Ah I'm getting somewhere. VFS Cannot open root device error -6
<clover[m]> woot!
<steev> clover[m]: that helped?
<clover[m]> idk yet, just happy for Leo :)
<steev> ah :D
<steev> blah, of course rc7 comes out
<steev> but i do need to step away, i need food, just realized i haven't eaten yet today, will be back later
<szclsya[m]> Take care!
<steev> good luck, i'd try hard coding the rootdevice and also, if you don't have it, add rootwait, as sometimes it can take usb a bit to settle
<steev> HOWEVER that will have the side effect of waiting forever, so you may want to give it like, say 30 seconds
<clover[m]> Stuff and things happened!!
<szclsya[m]> ayy usb is working
<szclsya[m]> (on clover
<clover[m]> No prompt 😢
<clover[m]> But progress
<clover[m]> <szclsya[m]> "(on clover..." <- Leo if you wanna try the latest artifact is pushed to my fileshare maybe you can at least get a serial console working now.
falk689_ has quit [Remote host closed the connection]
falk689 has joined #aarch64-laptops
<szclsya[m]> <clover[m]> "Leo if you wanna try the..." <- archiso?
<clover[m]> still confused why systemd-boot option `devicetree` says 'Unsupported', but kernel option `dtb=` works
<szclsya[m]> Downloading. Also how did you build it? I might want to tweak the kernel
<bamse> clover[m]: does the kernel actually pick up the devicetree with dtb= ?
<bamse> clover[m]: when i tried that on my machine the kernel would claim that secure boot was enabled and it ignored it...and hence booted with acpi anyways
<szclsya[m]> thx
<clover[m]> bamse the works you are saying i don't understand T_T
<clover[m]> words*
<steev> clover[m]: record the boot with a video, then watch it back and pause it very near the beginning and it should say ACPI: Interpreter disabled. if it's not using acpi (i believe)
<steev> welcome to debugging without serial console :P
<clover[m]> Will try later. Maybe Leo can capture with serial. Out getting sammich
<steev> szclsya[m]: you have serial?
<bamse> clover[m]: dtb=foo.dtb is supposed to make the kernel reach into uefi and load foo.dtb...what i'm saying is that on my device it refuses to do that...and therefor boots with acpi instead
<szclsya[m]> I have the devices but hasn't try them yet
<szclsya[m]> But I believe my device tree should be working, use dtb=/path/to/device/tree in cmdline with systemd-boot
<steev> that seems like a bug if systemd-boot isn't using devicetree option properly
<szclsya[m]> Yeah, I had the same issue with sd-boot's devicetree config line
<steev> supposedly https://github.com/systemd/systemd/pull/19417 is supposed to fix it, but that's merged
<steev> and i'd assume systemd had a release since then
<szclsya[m]> I can see the kernel log now with rootwait
<szclsya[m]> Weird, the kernel is able to load USB 2.0 and 1.1 drivers and usb-storage driver, but it should show a message recognizing my usb drive, but it doesn't
<szclsya[m]> Probably because it didn't load USB 3 drivers? Since my drive is a USB 3 one
<qzed> I'd guess it should still work via usb2...
<qzed> do you have CONFIG_USB_STORAGE and CONFIG_USB_UAS built in?
<steev> storage should be, uas is a module, in the defconfig, at least
<szclsya[m]> I changed them to built in, no dice
<szclsya[m]> https://fars.ee/nsiy The kernel .config file I'm currently working on
<steev> what does your command line look like?
<szclsya[m]> dtb=/x13s.dtb efi=novamap root=UUID=[REDACTED] rw clk_ignore_unused pd_ignore_unused rootwait
<steev> and when it's booting, (earlycon=efifb) - does it show that it's using acpi?
<clover[m]> back, ok recording my boot
<clover[m]> Irc peeps probably won't see the vid
<steev> we get the link to it
<steev> i didn't realize \ for the path works with the kernel
<steev> that's cool
<clover[m]> from systemd-boot notes:
<clover[m]> >Note that all paths used in the configuration snippets use a Unix-style "/" as path separator. This needs to be converted to an EFI-style "\" separator in EFI boot loaders.
<clover[m]> lol back slash didn't go through
<szclsya[m]> hmm does it count as using ACPI?
<steev> i believe it is
<steev> because i don't have the efi: ACPI 2.0... line
<clover[m]> what does this mean
<szclsya[m]> Hmm...
<steev> pretty sure you are both still getting acpi boot
<steev> https://systemd.io/BOOT_LOADER_SPECIFICATION/ the specification shows the paths written with /
<steev> but szclsya[m]'s dtb line is using / too so
<steev> clover[m]: mind editing your entry to also add "architecture aa64" ?
<steev> with the devicetree line added back, not using dtb=
<clover[m]> sure
<qzed> whelp, that explains one thing...
<steev> yes, but aa64 not aarch64
<steev> qzed: interesting
<steev> kinda a shame that both arm and mips support appended dtb
<steev> but not arm64
<steev> otoh.. you guys could just install grub, and let grub do its thing
<clover[m]> testing
<steev> although, i'd just edit the entry itself to test instead of building an entire iso to do it
<clover[m]> it's fine, i have the 'working' entry saved. aka the one that is doing the acpi boot
<clover[m]> takes 5 minutes to build an iso with my github action
<steev> btw, you won't ever need b43-fwcutter or sof-firmware
<clover[m]> talking to me or qzed?
<steev> you, i was looking at the packages you install in your iso :)
<clover[m]> oh ok, cool
<steev> i doubt that will help at all (the architecture line) but figured it's worth a shot
<clover[m]> Didn't work. But with dtb I got a useful log I think
<qzed> more confirmation to what we already know...
<steev> clover[m]: try in the other part?
<steev> s/part/usb port/
<clover[m]> you want me to try the power port or non-power port
<steev> whichever one you're currently in... use the other
<clover[m]> lmao ok
<steev> there are only 2, and i can't tell which one you're currently using :P
<clover[m]> Behavior is indifferent to whichever port I'm using
<steev> that's very weird, because usb definitely works on the debian side of things, that's how i did my installation