00:35 <dunc4n> hey, anyone managed to boot 6.12 on X elite? mine getting stuck at "EFI stub: exiting boot services"

00:36 <dunc4n> full custom .config, also tried to port 6.11 config from ubuntu concept but no go

01:00 <tobhe> I think JensGlathe has built a kernel that works

02:36 <robclark> I'm using jhovold's 6.12 + x1e patches branch with no issue..

05:39 <macc24> ppd: couldn't find any :/

06:32 <dunc4n> <robclark> "I'm using jhovold's 6.12 + x1e..." <- where can I find it?

06:33 <dunc4n> oh nvm!

07:53 <steveej[m]> robclark: will crosvm work with a regularly (what's the technical name for this?) booted linux system? i.e. can i just install and use crosvm and have fast VMs?

09:28 <deathmist> dunc4n: https://github.com/JamiKettunen/cports/tree/x1e/main/linux-stable works for me

10:11 <macc24> steveej: you might need https://patchew.org/linux/20240222-gunyah-v17-0-1e9da6763d38@quicinc.com/

11:38 <dunc4n> <deathmist> "dunc4nn: https://github.com/..."; <- with these patches & .config I get "Failed to update FDT and exit boot services"

11:39 <dunc4n> 64GB T14s, FWIW

11:43 <travmurav[m]> craftyguy: gunyah doesn't exist on 8c3 and, afaiu, broken on x1e firmware

11:47 <maz> I think what ships with X1E is just enough to run Linux, and with a bunch of caveats.

11:49 <travmurav[m]> you're mistaken, x1e firmware was never meant to run linux, and, more funnily, the gunyah stuff there was not meant to be used at all beyond letting uefi survive, since windows loads hyper-v anyway xD

11:49 <travmurav[m]> (/s I guess)

11:51 <travmurav[m]> so like I guess it was literally never tested properly, and this is perhaps half of the reason people here suffer with 64gb devices not working etc etc xD

11:51 <macc24> huh

11:51 <macc24> so linux in el2 would be better if not for remoteproc firmware

12:03 <dunc4n> iirc there was supposed to be an option for EL2 boot, is it disabled by OEMs or did we get qualconned again

12:04 <travmurav[m]> afaiu it exists on crd and it's broken

12:04 <travmurav[m]> no idea if there is anything in the hyp side to use of it's completely cut out from retail releases

12:05 <deathmist> dunc4n: hm I don't have any T14s pieces in that tree from ubuntu where it supposedly works with 6.11 for you right?

12:05 <travmurav[m]> and then again it would probably be same story as with slbounce, minus signed windows exe

12:06 <dunc4n> deathmist: yeah the concept image boots just fine but nothing else so far

12:08 <tobhe_> because I also have a 64GB T14s ;)

12:39 <deathmist> dunc4n: to be clear that tree has so far only been tested on the HP OmniBook which I have, also how are you booting it, through ubuntu x1e grub with 6.11 devicetree specified? that's how I kickstarted bringup to a new distro: it stayed booted after picking manually added grub entry, ctrl+alt+delete rebooted from a blind tty and ssh via usb ethernet

12:40 <deathmist> I could add in the ubuntu diff on top if you still want to try that tree, tho it's mostly just jhovold's x1e 6.12 one with ubuntu diff on top for hp omnibook and lenovo slim7x (which it remains untested on for now, I'm in contact with someone who wants to try that later once I figure out isos)

12:47 <deathmist> ig knowing more about how you're installing all the kernel bits in general would be useful, once they're properly installed you can boot it through whatever you like really if you know how to get it picking the right dtb (different kernels ship their own with potentially different topology from extra patches)

12:49 <jhovold> dunc4n: just use my wip branch, you need to some workarounds to boot the 64 GB variant:

12:50 <jhovold> https://github.com/jhovold/linux/wiki/T14s#uefi-firmware

12:51 <jhovold> and ExitBootServices() often fails on my 32 GB t14s as well

12:51 <jhovold> starting and exiting an EFI shell is the only workaround I've found for that so far

12:51 <deathmist> ah more device-specific firmware quirks, all of this sounds real fun even compared to androids I've booted mainline on before

12:54 <deathmist> then again those in comparison already boot some linux kernel by default really so it can't be *that* bad

13:16 <dunc4n> <deathmist> "dunc4nn: to be clear that tree..." <- booting as EFI stub via systemd-boot, patched gentoo-sources with patches you linked (also tried with vanilla sources), nothing special there, unless some specific bootloader hacks are needed and only doable with grub?

13:17 <dunc4n> <jhovold> "dunc4nn: just use my wip branch,..." <- i'll try this later today, thanks

13:17 <dunc4n> oh and about devicetree, tried to bundle it in the UKI and also load from ESP, no differencee

13:20 <deathmist> here's ubuntu stuff on top of that tree as a diff fwiw once you deal with the seeming firmware quirks https://paste.c-net.org/bojag6uehs4s

14:47 <robclark> steveej[m]: you'd need linux patches.. and IME crosvm cmdline is perhaps even harder than qemu to operate

16:54 <steveej[m]> robclark: ah got it, and agreed. qemu with gunyah support sounds like what i'm after

17:55 <rsalveti> jhovold: also having some weird failures on exitbootservices, added some comments to https://bugs.launchpad.net/ubuntu-concept/+bug/2086540, but looks like a memory corruption issue of some sort

22:33 <kuruczgy[m]> Just tried DP altmode (x1e) with another dongle and also a different monitor, neither worked, it's the usual link training failure. Is someone working on figuring out the cause for the link training errors? (IIRC it was mentioned here that debugging this might require some measurement equipment, making it quite difficult.)