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
iivanov has joined #aarch64-laptops
iivanov has quit [Ping timeout: 480 seconds]
iivanov has joined #aarch64-laptops
iivanov_ has joined #aarch64-laptops
iivanov has quit [Read error: Connection reset by peer]
iivanov_ has quit [Ping timeout: 480 seconds]
<nirik> so I had a bit of time today to play around some more... I didn't get usb boot working. ;( I did find that pxe boot does work and was able to pxe boot to a rescue shell... It might be I am hitting something in rawhide userspace, so tomorrow will try a f40 userspace and see if it gets any futher...
iivanov has joined #aarch64-laptops
iivanov has quit [Ping timeout: 480 seconds]
hexdump0815 has joined #aarch64-laptops
hexdump01 has quit [Ping timeout: 480 seconds]
<colemickens> So I'm not the only one struggling with USB boot at least.
<colemickens> Though apparently it does work.
<colemickens> I dont even know what to try now :/
<colemickens> I guess I was encouraged to enable more built in kernel modules but I'm not even sure what that would be
iivanov has joined #aarch64-laptops
iivanov has quit [Quit: Leaving...]
iivanov has joined #aarch64-laptops
iivanov has quit []
iivanov has joined #aarch64-laptops
<steev> if you aren't using it, i would highly encourage to use johan's 6.11-rc1 and his defconfig
<jhovold> colemickens: also check out my instructions to FarchordSteveCossette[m] from yesterday
<colemickens> so I do have the adsp firmware supposedly available, so that might be my issue
<colemickens> I'll drop the firmware for now and try that. Otherwise I have all those kernel modules and then some. I'll also try both orientations again.
<colemickens> Thank you 🙏
valentine has joined #aarch64-laptops
<valentine> robclark: I pushed a fixup to your MR, I hope I'm not imposing too much
<jhovold> colemickens: make sure you use johan_defconfig as well, otherwise there could be further modules needed
<jhovold> and if you pass 'regulator_ignore_unused' to the kernel and have loaded the keyboard deps you should be able to inspect the system and determine what is missing
<jhovold> given that your distro initramfs provides you a rescue shell (e.g. if it can't find the rootfs)
<jhovold> see the commit message for johan_defonfig for the keyboard deps
iivanov has quit [Remote host closed the connection]
iivanov has joined #aarch64-laptops
iivanov has quit [Ping timeout: 480 seconds]
<FarchordSteveCossette[m]> <jhovold> "and if you pass 'regulator_ignor..." <- The issue I have at least is a chicken & egg kind of problem: I do have a rescue shell, but without keyboard and mouse support (USB or otherwise), I can't use it lol
<FarchordSteveCossette[m]> I'll have to see if adding a bunch of modules helps, I believe I tried that before
<lollaritits[m]> how do i install archlinux with encryption?
<lollaritits[m]> i tried following ironrobins endeavouros terminal and then add encryption but did not work
iivanov has joined #aarch64-laptops
<jhovold> FarchordSteveCossette[m]: I assume you're using my wip branch and johan_defconfig, then you only need to include the modules listed in the commit message for johan_defconfig for keyboard
<jhovold> and note that the keyboard interrupts are not working with mainline so this won't work with, for example, a fedora rc1 kernel
iivanov has quit [Ping timeout: 480 seconds]
iivanov has joined #aarch64-laptops
iivanov has quit [Ping timeout: 480 seconds]
iivanov has joined #aarch64-laptops
iivanov_ has joined #aarch64-laptops
iivanov has quit [Read error: No route to host]
prime has joined #aarch64-laptops
prime has quit [Remote host closed the connection]
prime has joined #aarch64-laptops
prime is now known as Guest1749
Guest1749 has quit [Remote host closed the connection]
prim- has joined #aarch64-laptops
prim- has quit [Remote host closed the connection]
prime_ has joined #aarch64-laptops
prime_ is now known as prime
iivanov_ has quit [Remote host closed the connection]
<FarchordSteveCossette[m]> Here goes nothing, let's see how that looks
iivanov has joined #aarch64-laptops
<FarchordSteveCossette[m]> Nope.... sigh
<FarchordSteveCossette[m]> But let's see if the keyboard at least works in dracut rescue
<jhovold> FarchordSteveCossette[m]: I haven't updated the display deps for x1e yet so you can leave those out, you should be good with efifb as long as pass 'regulator_ignore_unused'
craftyguy has quit [Remote host closed the connection]
<jhovold> and in case you're having trouble including modules in your initramfs you can always build the keyboard deps into the kernel temporarily
craftyguy has joined #aarch64-laptops
craftyguy has quit [Remote host closed the connection]
craftyguy has joined #aarch64-laptops
<FarchordSteveCossette[m]> jhovold: It's in there :)
iivanov has quit [Ping timeout: 480 seconds]
<FarchordSteveCossette[m]> "arm-scmi firmware-scmi: failed to setup channel for protocol: 0x10
<FarchordSteveCossette[m]> arm-smmu 3da0000.iommu: probe with driver arm-smmu failed with error -110"
<FarchordSteveCossette[m]> If I reduce loglevel to 3, I see those errors, and a bunch of thermal zone errors, but idk if that would do anything
<jhovold> the scmi error is a known issue with the cpufreq patches and can be ignored
<jhovold> not sure why the gpu smmu times out
<jhovold> there should be no thermal zone errors
<jhovold> perhaps related to that thermal command you passed from grub
<jhovold> have you enable btrfs, which I think fedora uses?
<FarchordSteveCossette[m]> jhovold: As a driver?
<jhovold> otherwise you may not be able to mount your rootfs
<jhovold> yes
<FarchordSteveCossette[m]> I figured it would, but lemme see
<FarchordSteveCossette[m]> Alright added btrfs to the long list of force_drivers, let's see
<jhovold> not forced drivers, but it needs to be enabled in the kernel config
<FarchordSteveCossette[m]> Nope still the same
<FarchordSteveCossette[m]> I thought it was, but lemme see
<jhovold> it's not enabled in johan_defconfig currently as I only use ext4
<FarchordSteveCossette[m]> Man building sure doesnt take a long time LOL aiite, lemme try to install that
iivanov has joined #aarch64-laptops
iivanov has quit [Ping timeout: 480 seconds]
<Jasper[m]> <FarchordSteveCossette[m]> "Nope.... sigh" <- Lemme check what I do on x13s, I think it's not force_drovers
<Jasper[m]> *drivers
<Jasper[m]> also I'd suggest specifying the fw files, iirc they don't include the entire folder the way you did it
<FarchordSteveCossette[m]> Jasper[m]: There's also the fact that I somehow messed up the config <.< idk why, the menuconfig is.... annoying.
<steev> it can be, but if you have a general idea of what you're after, you can use `/` for searching
<robclark> valentine: thx.. I want to make one more change to the fd patch and then I'll squash your fixup
<FarchordSteveCossette[m]> steev: problem is, I'm trying to make sure it baked in the btrfs driver. I selected it in menuconfig, exited. Then, I did make (In my case, amake -- it's an alias) rpm-pkg and it build. And it _seems_ to have built the rpm, but if I look at the installed kernel, there's no /fs/btrfs in the kernel
<steev> grep BRTFS <your kernel build path/.config> ?
<steev> BTRFS*
<steev> when you did the menuconfig, did you do the amake alias as well? i'm assuming amake has the cross and arm setup
<FarchordSteveCossette[m]> steev: Yes, always.
<FarchordSteveCossette[m]> and yes it does
<FarchordSteveCossette[m]> Sorry sorry I keep forgetting not to reply
<steev> i dunno if rpm-pkg installs the config as well, but check that?
<steev> fwiw, johan seems to set IKCONFIG so you should be able to zgrep BTRFS /proc/config.gz
<FarchordSteveCossette[m]> >.<
<FarchordSteveCossette[m]> just noticed it made a copy of my GOODCONFIG in aarch64/
<FarchordSteveCossette[m]> oooooookay.... annoying.
<FarchordSteveCossette[m]> aiite the GOODCONFIG there does have btrfs, and yet the built rpm doesn'T have it. To heck, I'll skip the rpm packaging thing
iivanov has joined #aarch64-laptops
<gwolf> steev: Thanks a lot! I'm taking note of your answer, will try to compile tomorrow (today I'm quite tired)
<steev> there's no rush :)
<steev> i'm just appreciative that ema wrote that blog :D
<FarchordSteveCossette[m]> Alright jhovold here's what I did, just to make 100% I made it right:... (full message at <https://matrix.org/_matrix/media/v3/download/matrix.org/CrFtvWZVEZaSpjhOEUZTABYG>)
<FarchordSteveCossette[m]> Oh and I also did dracut --kver (whatever) -f to regenerate the initramfs
<steev> do do you need to pass a subvolid or anything? (i don't do btrfs)
<FarchordSteveCossette[m]> <steev> "do do you need to pass a..." <- yeah that's in there too
iivanov has quit [Remote host closed the connection]
Caterpillar has joined #aarch64-laptops
iivanov has joined #aarch64-laptops
<nirik> anoying. The nvme I got doesn't seem to work in the external usb enclosure I have. :( (or it's just broken I guess)
<steev> is it trying to use UAS or something? i don't think that's enabled in johan's defconfig
<steev> specifically i mean CONFIG_USB_UAS
<steev> the usb nvme adapters i use aren't listed as available anymore on amazon :( but i use two different ORICO
Kelsar_ has joined #aarch64-laptops
Kelsar has quit [Ping timeout: 480 seconds]
<FarchordSteveCossette[m]> jhovold: So, I'm booted in to the default fedora kernel right now. Mind you, I still have no laptop keyboard, but I do have USB so I can enter commands. Anything I can do to try to pinpoint this issue?
<FarchordSteveCossette[m]> One thing I wanna try too, we removed the 80-drivers udev rule... does that do anything in the initramfs...?
<nirik> no no, I meant on my x86_64 laptop... I was going to copy a install to the nvme and then swap it into the 7x...
<colemickens> Okay, I removed firmware, I tried both orientations for the USB-C drive, and it still is not seeing the root on USB in initamfs.
<colemickens> this is with abel.vesa's config.
<colemickens> One question I have - what is the USB driver that I would expect to see fr a USB-C USb3 thumbdrive?
<steev> if it's generic enough, usb_storage
<colemickens> I wonder if I somehow need that in addition to 'uas' thten
Kelsar_ has quit [Ping timeout: 480 seconds]
<steev> no idea, but i'd switch to johan's stuff before trying to fight with abel's stuff (and abel has already recommended switching to johan's)
<colemickens> oh
<colemickens> okay. nice.
<FarchordSteveCossette[m]> colemickens ahh youre in the same boat i am! 😂
<colemickens> Yeah, I think so :(
<colemickens> I'm so damn close I can taste it.
Kelsar has joined #aarch64-laptops
Kelsar has quit []
<FarchordSteveCossette[m]> Not me im so lost i am finding presents from past christmases
<colemickens> anyone have a link to johan's linux repo?
<colemickens> here we go again! 😎
<JosDehaes[m]> that's a lot of cores 😁
Kelsar has joined #aarch64-laptops
<colemickens> Still no luck, threw in the regulator_ignore_unused kernel param and usb_storage module but same behavior.
<colemickens> (also removed firmware)
<colemickens> I do get a recovery shell, but not really sure what to look at once I'm there.
<colemickens> * removed firmware) (also, with johan's kernel + johan_defconfig)
<steev> look at the devices in /dev, look at the modules that are loaded, check dmesg output
<JensGlathe[m]> perchance, do you'all build MHI as modules? Last year when I was trying to boot Volterra from USB, it was the final thing to make it work. MHI* as modules, not builtin
<colemickens> I suspect MHI is built in. I don't think I include it in initrd and I do see it trying to load firmware.
<colemickens> (also tried both orientations again)
<JensGlathe[m]> I had to build mhi as modules to make the damn thing boot from anything USB, otherwise it would always fail on mounting root, regardless of fs. I guess there's a connection between bus enumeration and the way mhi is initialized. With mhi loaded as modules, the XHCI controllers are detected earlier, not after trying to mount root.
<steev> jhovold: mani_s ^
<mani_s> steev, how MHI is related to USB?
<steev> mani_s: no idea, that's something jens would need to answer
<steev> though looking at it, the MHI bits seem to be modules, at least for sc8280xp
<mani_s> technically there is no relation between them. MHI uses PCIe
<mani_s> there is no need to build mhi drivers as built-in
<JensGlathe[m]> 6.11 seems to have new tricks up its sleeve 😎 I can drive my 4k display with 4k again, suspend and re-awake. This didn't work reliably (or at all) since 6.6
<steev> it seems like johan's defconfig does set it to M anyway
<mani_s> steev, that's the correct one
<steev> mani_s: i don't have an x1e device :) i'm just grepping around trying to help out the poor nix and fedora users :(
<JensGlathe[m]> re MHI: Maybe MHI using PCIe changes the order of initialization or something, it was just observational. To observe it, though, I had to buy an X13s 😁
<colemickens> idk why I'm seeing mhi log lines then, I don't include the module in my initrd: https://github.com/colemickens/nixos-snapdragon-elite/blob/1c7911aac4227451eb0c71d68c8a1513b3d0df8a/configuration-image.nix#L70-L132
<JensGlathe[m]> so its probably builtin
<colemickens> I'm using johan's defconfig, so presumably it shouldn't be.
<steev> check if it's in there because of a dependency anyway?
<steev> because pcie_qcom is there, and that's gonna need it
<mani_s> MHI is just for modems/WLAN. And it doesn't directly depend on any Qcom drivers
<mani_s> It only depends on CONFIG_PCI
<colemickens> Oh yeah, and I actually have pcie_qcom now from johan's tree I guess.
<JensGlathe[m]> Would anybody try to boot an Ubuntu stick I could prepare from Johan's tree to see if we get it up? No X1E hardware ere, either
<steev> oh, i guess not then?
<steev> well if wlan is in there
<steev> JensGlathe[m]: that's good news on the 4k, i only have a 2k monitor here though :(
iivanov has quit [Remote host closed the connection]
<nirik> I'm happy to try images from folks on the 7x I have, but will likely be later today/this evening... in meetings/etc.
iivanov has joined #aarch64-laptops
<JensGlathe[m]> Need to build it anyway
iivanov has quit [Quit: Leaving...]
<JosDehaes[m]> I've been trying to get the Ubuntu installer to boot on my yoga, but so far was not successful. I see the device tree load and I see a few kernel messages but then the screen goes black. The initrd of the Ubuntu iso only has modules and firmware, so not sure how it could ever boot
<JensGlathe[m]> The x13s 24.04 ISO from Canonical fails to boot on the x13s, a detail not yet (if ever) corrected. https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2060868
<JensGlathe[m]> I use a different image, I dervied from the 24.04 RPi preinstalled Desktop and modified it for Volterra and X13s. It brings up the Ubiquity installer, and sets up a fairly decent system.
<JensGlathe[m]> That's the one I will use to derive for X1E stuff.
<JosDehaes[m]> Thx I can try that base
<FarchordSteveCossette[m]> I..... found it
<FarchordSteveCossette[m]> ....the custom kernel boots now. Well, it boots further at least
<FarchordSteveCossette[m]> nirik: remember when I said to delete the udev 80-drivers rule? It needs to be in the initramfs....
<FarchordSteveCossette[m]> keyboard's working, that's good
<FarchordSteveCossette[m]> Wifi's not working, but meh.... I actually don't care right now
<JensGlathe[m]> what is this udev 80-drivers rule
<FarchordSteveCossette[m]> JensGlathe[m]: /lib/udev/rules.d/80-drivers file
<FarchordSteveCossette[m]> On fedora, for some reason, it makes the computer die
<FarchordSteveCossette[m]> When ran on userspace, at least
<nirik> farcord: good find. You mean it has to be removed from the initramfs? or added back to it?
<FarchordSteveCossette[m]> nirik: added back to it.
<FarchordSteveCossette[m]> And ofc the kernel has to be built with btrfs support
<JosDehaes[m]> Jens Glathe: the strange thing is that the ubuntu arm64 image does boot in qemu. I tried both 24.04 and the dev version
<FarchordSteveCossette[m]> I think I might need to add zfs too, I saw it failed to mount something
<nirik> https://paste.centos.org/view/9c165264 is that file (at least on rawhide)
<JensGlathe[m]> Jos Dehaes: yes why not. What is the qemu host?
<FarchordSteveCossette[m]> nirik: Looks like it yup
<JensGlathe[m]> I have the same kernel running as VM on the EL2 box
<JosDehaes[m]> x86_64
<FarchordSteveCossette[m]> To be clear: It needs to not be in the root filesystem, but it needs to be in the initramfs
<nirik> cool. Something to poke at later when time permits.
<JosDehaes[m]> Jens Glathe: I have no idea how the ubuntu kernel finds init as the initrd only contains modules/firmware and nothing else
<JensGlathe[m]> jglathe@sdbox2:~$ ssh 10.225.42.70
<JensGlathe[m]> Welcome to Ubuntu 24.04 LTS (GNU/Linux 6.8.0-39-generic aarch64)
<FarchordSteveCossette[m]> Lemme try to see, if I add it back in, is it gonna work now that I'm on a custom kernel
<JensGlathe[m]> Jos Dehaes: pobably just mounting root. cloud_init does a lot of things
<FarchordSteveCossette[m]> Oh seems zram is refusing to boot too
<FarchordSteveCossette[m]> Maybe a kernel module?
<FarchordSteveCossette[m]> *mount
<JensGlathe[m]> who cares about zram, out with it. its a systemd unit, link to /dev(null
<JensGlathe[m]> s/(///
<JensGlathe[m]> * on this high-end laptop ofc
<steev> i still use it :( but my high-end laptop only has 16GB of ram
<JensGlathe[m]> I also have it enabled om the X13s
* FarchordSteveCossette[m] has enabled zram, replicated fedora config in that regard. Rebuilds the kernel.
<colemickens> booted johan's kernel/defconfig, removed the firmware (so no adsp), I have the usual kernel params (_ignore_unused), and I'm in emergency console. Nothing jumps out at me from /dev/kmsg, under /dev/block I only have symlinks to /dev/nvme0n1p and loop devices. I don't know what else to check/try. Hotplugging the USB device does not produce new log lines in /dev/kmsg. (uas and usb_storage are loaded)
<pstef> I'm afraid that's not correct ed syntax, BTW
<colemickens> s///, s/nvme0n1p/nvme0n1p_/, s/usb_storage/usb\_storage/
<colemickens> s/_ignore_unused/\_ignore\_unused/, s/nvme0n1p/nvme0n1p_/, s/usb_storage/usb\_storage/
<JensGlathe[m]> I went looking for the T14s btw. my employer sells the 32GB / 1TB variant, regular price €1849. Competition goes down to 1670 or so, that would be the CAMPUS edition (also sold by NBB)
<colemickens> ath12k loads in mhi. I suppose I could remove that so mhi isn't copied and see if that helps per the anecdote shared a bit ago.
<lollaritits[m]> does fedora arm run steam? or any distro?
<FarchordSteveCossette[m]> yeah, that udev rule is causing a mess XD
<FarchordSteveCossette[m]> Backlight control works
<robclark> lollaritits[m]: yes, with fex emu.. but not sure if there is f40 fex-emu RootFS published with new-enough mesa yet
<JosDehaes[m]> Jens Glathe: that rpi image doesn't even have UEFI 😅
<JensGlathe[m]> Didn't care, afterwards it had
<JosDehaes[m]> hehe
<JosDehaes[m]> what did you put in the initrd?
<JensGlathe[m]> the modules and firmwares as specified by Johan / Steev
<JensGlathe[m]> and what I found on armbian
<JensGlathe[m]> the fun part of the rpi image was tht you could boot it on a rpi, if things didn't work. No chroot required, since it was a different boot mechanism
<JensGlathe[m]> ut same arch
<JensGlathe[m]> * but same arch
<nirik> Farchord (Steve Cossette): which kernel are you using now? and for wireless you might need to just modprobe ath12k (and make sure you have the firmware)
<FarchordSteveCossette[m]> nirik: The firmware is what I need to triple-check. But loading ath12k doesn't do much. I won't send logs until I'm sure I'Ve went through everything. I'm using Johan's 6.11-rc1 kernel
<nirik> ok, fair
<FarchordSteveCossette[m]> nirik: if you want, I can zip it up and upload it to my fedorapeople site. It'S not big. It's lacking the dtbs, but tbh I think the ones provided by Fedora are probably fine.
<nirik> if you like. I have to head to a appointment, and will not have time to test anything until later. The dtb should be under /lib/modules/ ? or no?
<steev> you're definitely gonna want to use the dtb that comes with johan's kernel
<steev> s/with/from
<FarchordSteveCossette[m]> steev: Ah, cuz for the life of me, I can't make the dtbs
<FarchordSteveCossette[m]> make dtbs does nothing
<steev> if you built the kernel, it should be part of that process, and running make dtbs afterwards wouldn't do anything because it's done already
<steev> do you have .dtb files in the arch/arm64/boot/dts/qcom directory wherever your built kernel stuff is?
<FarchordSteveCossette[m]> steev: I checked in the target folder where the kernel modules are, nope it's not
<steev> not modules
<FarchordSteveCossette[m]> Yeah the dts files are there
<FarchordSteveCossette[m]> and dtsi
<steev> can you give me tree output? fedora must be doing something weird
<FarchordSteveCossette[m]> The tree of the build folder? sure
<steev> yeah
<steev> and then of the sources too, if you still have it
<FarchordSteveCossette[m]> lol tree file size is 12mb
<FarchordSteveCossette[m]> the pastebin wants nothing to do with it
<steev> tree just arch/arm64
<steev> tree just arch/arm64/boot/dts
<FarchordSteveCossette[m]> ok
<steev> that's 88k
<steev> at least here, so should be roughly thereabouts
<FarchordSteveCossette[m]> https://paste.centos.org/view/d94983f7
<steev> that's source or build?
<FarchordSteveCossette[m]> source
<FarchordSteveCossette[m]> lemme make a tree of the build
<FarchordSteveCossette[m]> https://paste.centos.org/view/32f67de9
<steev> what is in aarch64/
<FarchordSteveCossette[m]> For the record, when I made a rpm, it did bake in the dtbs. But then the modules were all wrong
<steev> build -> /home/farchord/FedoraWork/x1e-kernel/linux/aarch64 this symlink
<steev> is there an arch/arm64/boot/dts directory ?
<FarchordSteveCossette[m]> steev: It's a loop back into the source for some reason
<steev> i'm assuming you also do amake dtbs_install ?
<FarchordSteveCossette[m]> ahh, that's probably what I was missing
<FarchordSteveCossette[m]> yup.... alright now I feel like an idiot
clee_ has joined #aarch64-laptops
<colemickens> hm, I have this "a00000.usb: deferred probe pending: dwc3: failed to initialize core"
<colemickens> that I see in /dev/kmsg. would that be blocking me?
<steev> from usb, yes
<FarchordSteveCossette[m]> colemickens: I think that's just a product of your USB not working
<steev> you could check cat /sys/kernel/debug/devices_deferred
<colemickens> oh my /sys/kernel/debug is empty, do I need a kernel param for that
<colemickens> there's a arm-scmi firmware:scmi: failed to setup channel for protocol:0x10 line above the usb lines, not sure if related
<steev> that's not
clee has quit [Ping timeout: 480 seconds]
<steev> oh, you might need to mount debugfs
<colemickens> okay, I added debug to cmdline for fun too
<colemickens> yeah, it gives me the same lines
<colemickens> 3 of them, presumably for each usb port
<colemickens> "dwc3: failed to initialize core` (a000000.usb, a600000.usb, a800000.usb)
<colemickens> (thats from /sys/kernel/debug/devices_deferred to be clear)
<colemickens> doesn't really tell me /why/ though
<steev> usually a wrong clock or something
<steev> you have clk_ignore_unused in the command line, i'm assuming?
<steev> add in regulator_ignore_unused ?
<colemickens> (note I haven't pushed, but I've switched to johan's kernel and disabled firmware since the last push)
<colemickens> I have that locally, not yet pushed steev
<steev> ah okay
<colemickens> ╰🡒 cat snapdragon.nix | rg regulator
<colemickens> "regulator_ignore_unused"
<steev> any chance you could dmesg output to a file ad post it somewhere?
<colemickens> fwiw I can confirm that I don't have mhi loaded this time too
<colemickens> steev: heh, I wish! I don't have any (immediately obvious/reasonable) way to exfiltrate it
<steev> time for a second usb :P
<colemickens> ooh that'd be neat, a cli app that could take text on standard in and then spew qr codes that could be read from an android app
<colemickens> steev: I don't think _any_ of my USB is working at all.
<colemickens> like hotplugging this in initrd doesn't make it appear, so I don't think a second drive would help?
<steev> hm, that's a good point
<FarchordSteveCossette[m]> colemickens: Here's the tl;dr of how I made mine work, idk what nixos uses for root partitions, but are the fs drivers in the kernel?
<colemickens> yes
<FarchordSteveCossette[m]> steev: I got a bunch of Samsung T7 SSDs right now, love them
<steev> too bulky :D
<FarchordSteveCossette[m]> I also got a Crucial SSD somewhere
<FarchordSteveCossette[m]> Bulky? I guess... I mean it's about as big as a SATA SSD lol
<steev> 32gb is 12 bucks, and that's more than enough for bringup
<colemickens> interestingly it doesn't even have uas/usb_storage loaded, but I guess maybe that's because dwc3 failed?
<steev> colemickens: question
<steev> can you ls your efi partition?
<colemickens> presumably the nvme one? I can sure try...
<steev> wherever nix is installing the files to :D
<colemickens> oddly I can't mount any of the nvme parts, one fails obviously as bitlocker but not sure why the vfat didn't just mount
<colemickens> anyway,lsing the usb is easy, I'll do it from here
<steev> is it loading off the usb?
<JensGlathe[m]> dosfstools mtools
<colemickens> steev: it loads initrd off usb and then fails to move to our stage-2 because it can't find the root on the usb
<colemickens> since it's not seeing the usb devices at all, hence my concern over dwc3
<colemickens> steev: so I'm happy to ls the esp but I think it's sound enough
<steev> sure, i just wanna see the contents
<steev> and then also if you could the snapdragon.nix local you have
<colemickens> tthat's why my esp looks like: https://0x0.st/XOcy.txt
<steev> qcom-x1e80100-lenovo-yoga-slim7x.dtb.efi
<steev> no.
<steev> it should be .dtb
<colemickens> it still works
<steev> your usb not working says otherwise
<colemickens> it doesn't even get the screen/etc working without it
<colemickens> (I know because I built a version where the devicetree line was missing and didn't even get this far)
<steev> i really hate nixos paths
<colemickens> it's like the matrix, eventually you get used to it :P
<colemickens> (but also, no, I understand)
<steev> i'll pass, we have fhs for a reason
<steev> for shits and giggles
<colemickens> heh plenty of both those going on here
<steev> can you copy the dtb to the root EFI as x1e80100-lenovo-yoga-slim7x.dtb
<colemickens> fwiw I've checked /proc/device-tree/compatible and it says "lenovo,yoga7xqcom,x1e80100"
<colemickens> do you want me to change the systemd-boot entry too, even though it seems to be getting loaded?
<steev> for giggles, comment out the devicetree line (does systemd-boot support commenting things out?)
<steev> after doing so
<colemickens> ok will do
<steev> if that goes back to nothing at all working; add dtb=/x1e80100-lenovo-yoga-slim7x.dtb to the kernel command line
<steev> (i'm not sure if you could do dtb=/nixos/qcom-x1e80100-lenovo-yoga-slim7x.dtb.efi but maybe?)
<colemickens> EFI stub: generating empty DTB
<colemickens> and then acts like it did when I didn't specify a dtb
<colemickens> I'll add the kernel dtb param to the entry now
<steev> does the x1e bios have a linux option or nah?
<colemickens> uhhhh oh gosh
<steev> they added one (beta) to the thinkpad bios, but idunno
<colemickens> I don't know I flipped SB and then don't remember anything else
<colemickens> I'll check
<steev> for other devices
<steev> i don't *think* it should matter, but iirc, with the linux option on the thinkpad, just putting the dtb onto the root of the efi partition would have the bios pass it in
<colemickens> yeah, I wondered if that's what we were working towards
<colemickens> so far, with the dtb kernel param line, it's acting like it did when I had systemd-boot loading it.
<steev> okay
<colemickens> 45s til timeout to rec console
<colemickens> I would really hope this doesn't matter but I'm on a Canadian model :|
<steev> doesn't that mean it just says sorry after every audio blip?
<colemickens> and an annoying kb
<steev> oof :(
<colemickens> I have also noticed some dp/edp related driver that kernel panics, but I also assume that shouldn't affect usb.
<steev> that could potentially just be a splat for using panel-simple
<JensGlathe[m]> okay, 6.11 lets the performance cores run hotter. seems like 75°C , that's a lot more
<colemickens> seems like maybe this "failed to initialize core" message comes from a failed dma mappping
<colemickens> oh right, the bios option...
<JensGlathe[m]> build time goes down from 34m to 28m
<robclark> yeah, the edp-panel WARN_ON() can be ignored.. once you get things working enough you can dump edid and figure out the panel module and then we can add the missing entry in panel-edp.c ... or switch the dts if it is an SLS OLED panel
<colemickens> yeah typical sparse lenovo bios, sadly its not that simple, no linux option
<robclark> steev: no linux option on the yoga 7x, fwiw.. the bios options are pretty minimal
<steev> robclark: good to know :) that's why the empty dtb message :(
<robclark> you've gotta pay thinkpad bucks to get extra bios options :-P
<colemickens> I kinda don't get how my dwc3 can just not work when it works for others.
<colemickens> It doesn't rely on some of the firmware does it?
<colemickens> Did I make a different problem by trying to avoid adsp and dumping the rest of the FW? I guess dwc3 could require fw :/ ?
<robclark> for usb boot, idk.. maybe it could depend on probe order
<steev> i can't answer that because i don't know :(
<robclark> on x13s dp over usb-c dependend on cdsp but iirc normal usb functions didn't... seemed like cdsp was controlling the dp/usb routing
<colemickens> well, if someone else wants to try with my image I can wormhole it to them, but otherwise I'm out of ideas / out of my depth :(
<steev> on the x13s, usb comes up way before the remote procs
<steev> colemickens: how about a picture of "dmesg | grep usb"
<FarchordSteveCossette[m]> Alright, so all the firmware is in the filesystem AND in the initramfs... how do I load wifi, graphics and etc?
<JensGlathe[m]> the paths should be in the dts
<FarchordSteveCossette[m]> Oh good point lmc
<FarchordSteveCossette[m]> Yep I am using that dtb
<JensGlathe[m]> and to check what firmwares are loaded I have this in the cmdline "dyndbg=file drivers/base/firmware_loader/main.c +p"
<JensGlathe[m]> and can check with sudo dmesg|grep "Loaded FW"
Caterpillar has quit [Quit: Konversation terminated!]
<FarchordSteveCossette[m]> No results
<JensGlathe[m]> it needs to be booted with this (I assume)
<FarchordSteveCossette[m]> Well the firmware files themselves are in
<FarchordSteveCossette[m]> lemme c what the arch image has
<steev> as long as the wifi firmwares are in the correct place, a modprobe ath12k should give you a device
<steev> maybe ath12k_pci
<FarchordSteveCossette[m]> steev: doesn'T look like it... I just double-checked. I even tried to apply the "robs_hacks" cfg in modules.load.d, oh boy didn't that go bad! XD
<FarchordSteveCossette[m]> im talking about this btw: https://www.irccloud.com/pastebin/LdMlTmkh/
<FarchordSteveCossette[m]> qcom_q6v5_pas causes the whole system to go off the rails lol
<robclark> you could remove the q6v5 stuff for now, it might mean no battery status, but that isn't too critical yet
<FarchordSteveCossette[m]> so, ath12k is loaded, but it's used by nothing
<steev> anything in dmesg?
<robclark> lsmod shows ath12k not used by anything for me too.. I think that is fine.. `dmesg | grep -i ath12k` should tell you if it is missing some fw or calibration data
jhovold has quit [Ping timeout: 480 seconds]
<anonymix007[m]> jhovold: can you pick https://lore.kernel.org/ath12k/20240715023814.20242-1-quic_bqiang@quicinc.com/ to your x1e branch? At least on x86 it's straight up impossible to upload any files without it. WCN785x just becomes unresponsive after a few seconds
<steev> oof, right a he's gone
<steev> blergh, was hoping there would be something more before the dwc3 messages :(
<colemickens> uh, it has a dependency cycle with pmic-glink?
<colemickens> does it actually depend on that because I think it had other error lines
<steev> dmesg | grep pmic
<steev> yeah, those are normal
<steev> the bottom line, i'm pretty sure comes from the panel splat
kalebris has quit [Ping timeout: 480 seconds]
<colemickens> maybe it's got a special button I need to push ;)
<colemickens> I tried whispering nice things to it.
<colemickens> wait I found a google hit for this from a rob clark -_-
<colemickens> they look at that, some more hints: CONFIG_PHY_QCOM_EUSB2_REPEATER and CONFIG_PHY_QCOM_EUSB2
<FarchordSteveCossette[m]> steev robclark
<steev> thats all?
<steev> or did you just probe it
<FarchordSteveCossette[m]> That's it. It'S set to load on start
<steev> typically those messages come right before it loads the firmware
<steev> what about dmesg | grep remoteproc
<FarchordSteveCossette[m]> steev: here's the full dmesg: https://paste.centos.org/view/5b4b4394
<colemickens> so progress
<colemickens> but lol now I'm getting the mid-boot reboot
<colemickens> "phy-qcom-eusb2-repeater"
<colemickens> "phy-qcom-snps-eusb2"
<colemickens> "phy-qcom-snps-femto-v2"
<colemickens> one of them did the trick, and/or in addition to the others I'd added
<colemickens> do we know what udev rule in particular is problematic?
<colemickens> seems like 80-drivers.rules
<FarchordSteveCossette[m]> For me in fedora, /lib/udev/rules.d/80-drivers. But it does need to be in the initramfs, not in the actual main paritition
<colemickens> wow thats ridiculously annoying to implement properly/cleanly :|
<colemickens> can I blacklist the offending module instead?
<FarchordSteveCossette[m]> Unknown
<FarchordSteveCossette[m]> But have you removed the file yet? if not it's in your initramfs
<FarchordSteveCossette[m]> just try to remove it from the filesystem
<colemickens> well for me this wasn't an issue unless I made it out of initramfs
<robclark> interesting that 80-driver.rules works in initrd.. I guess it would be useful to know _what_ it probes that causes things to (I assume) insta-reboot (which typically is the bios getting angry about accessing some hw block that is unclocked/unpowered)
<steev> i bet its the first rule
<steev> ENV{MODALIAS}=="?*", RUN{builtin}+="kmod load"
<FarchordSteveCossette[m]> Yeah I essentially extracted both the good fedora initramfs and the more recent ones, aside from the kernel modules, all I could find that was different was that file....
<robclark> hmm, I deleted that file.. but not super familiar with udev rules syntax so not sure how to translate that into which driver is misbehaving
<FarchordSteveCossette[m]> so bringing it back fixed it. Go figure
<FarchordSteveCossette[m]> robclark: have you checked the dmesg I posted? Not sure if you might have an idea why wifi is just not working lol
<FarchordSteveCossette[m]> And I wish time wouldn't keep resetting to midnight lol
<robclark> could be that probing while some other driver has clks/etc enabled masks the problem?
<robclark> not sure why wifi is not working, I'd expect to see some msgs about loading fw
<FarchordSteveCossette[m]> I did see some, but none that I can see is specifically related to wifi
<steev> that's the only rule i can think of that would cause issues - there's no mtd*ro files (are there?) - pretty sure no parallel ports either, so that one shouldn't be the issue, i2o i also don't think exists, same for memstick and tifm devices
<robclark> so, I had to patch board-2.bin for calibration data.. but there should be some dmesg to give you a hint about what is missing after it loads fw
<colemickens> I'm surprised there's not a proper kernel blacklist mechanism, though the rule y'all suspect means I don't really even know which one I'd want to blacklist
<colemickens> afaict modprobe blacklist won't stop udev from being able to probe it
<robclark> steev: yeah, that is why I wasn't sure wtf that file had to do with anything
<colemickens> unless I'm just ignoran
<FarchordSteveCossette[m]> robclark: Mine's already patched
<robclark> this is what I _expect_ to see:
<robclark> granted, I'm no wifi/ath12k expert
<colemickens> oh well the arch wiki implies that modprobe blacklist will block it
<steev> robclark: could possibly try changing it to something like.... `ENV{MODALIAS}=="?*", RUN+="/usr/bin/udevadm info -q env -p $devpath >> /tmp/udev.log" #RUN{builtin}+="kmod load"`
<robclark> usually working on display/gpu things other people get that low level stuff working first ;-)
<steev> that should instead, print out what MODALIAS expands to to /tmp/udev.log instead of running kmod load
<robclark> steev: yeah, plus maybe something to `sync` and delay to ensure it gets written to disk
<steev> i suppose would probably NOT want it on /tmp these days
<robclark> yeah
<steev> but i was thinking it was initramfs anyway
<robclark> if it is instareboot then linux doesn't get a chance to shutdown cleanly and flush things to disk
<steev> and if it doesn't do the kmod load, then it shouldn't reboot (maybe)
<robclark> hmm, maybe just try your idea w/out the file in initrd so it happens after rootfs is mounted?
<steev> could try it
<steev> i don't have x1e to experiment :D