07:44 <jhovold> clover[m]: basically as a service to distros so that they don't have to figure out the display dependencies themselves, and to align with the arm64 defconfig

07:45 <jhovold> for a distro, what can be built as a module should be built as a module, but if you manage your own config for a particular machine you can chose to compile things in instead

07:46 <jhovold> choose*

07:47 <javierm> jhovold: I think your answer was for craftyguy ?

07:48 <jhovold> during bringup these clock controller were compiled in and when I hesitated a bit about addressing this, but a new rc1 is probably the best time to change this

07:48 <jhovold> javierm: indeed, thanks for spotting that

07:49 <javierm> jhovold: as a distro developer, I wondered what was the question so I scrolled back :)

07:50 <jhovold> s/when/then/

07:52 <javierm> jhovold: we also do the same in fedora, at the beginning we had most clk drivers as built-in but then figured out what could be built as a module

07:52 <javierm> jhovold: e.g: https://gitlab.com/martinezjavier/kernel-ark/-/commit/8b982129a805b47ab650be71047d8dcc03eb9623

07:54 <steev> i'll try to find some time this weekend to clean things up a bit; there is definitely some divergence and things still being built in

08:04 <travmurav[m]> javierm: I'm curious, but say I want certain config options to be set by fedora (or any other "mainstream" distro for that matter) what would be the best way to get those in? Is making sure upstream defconfig sets them enough?

08:04 <jhovold> i just pushed out a patch from Johannes to the 6.7-rc1 wip branch that fixes the wifi lockdep assert, which mean there are now no (new) known issues with that branch

08:09 <javierm> travmurav[m]: you can propose a config change to the fedora kenrel git repository: https://cki-project.gitlab.io/kernel-ark/#contributor-guide

08:10 <javierm> travmurav[m]: for example the commit I shared before touches only the config

08:14 <travmurav[m]> javierm: ah I see, thanks!

08:15 <travmurav[m]> just thinking that I'm slowly getting to the point where I have most things working upstream with 7c aspire1, so it would soon be worth exploring making "mainstream distros" work there properly

08:15 <javierm> travmurav[m]: awesome

08:16 <javierm> the fedora kernel basically tracks the latest stable kernel version, so is mostly about config options and upstream backports to fix bugs, etc

08:16 <javierm> travmurav[m]: every time that a new Kconfig symbol is found during a rebase, it is put in https://gitlab.com/cki-project/kernel-ark/-/tree/os-build/redhat/configs/pending-fedora/generic

08:17 <javierm> so should be mostly about moving from pending to redhat/configs/fedora/

08:17 <travmurav[m]> I guess my main problems there are 1. dtb provisioning since i think fedora installer images expect the dtb to be provided by firmware? (for this I'm thinking to adapt/write an efi driver that the user could install along with their out-of-band dtb)

08:18 <javierm> likely redhat/configs/fedora/generic unless the config option should vary between arch

08:18 <travmurav[m]> javierm: Oh, cool, I immediately see (unrelated to this) symbols I've recently added

08:18 <javierm> travmurav[m]: yeah, ideally Fedora expects the DTB to be provided by the firmware, so that all the machines are at least SystemReady-IR-ish

08:19 <javierm> but there are exceptions. The https://pagure.io/arm-image-installer is used to flash these

08:20 <travmurav[m]> and 2. is the installer - the cheap woa devices like aspire1 have pre-partitioned emmc with like 50 partitions for the bootloaders and uefi, followed by the windows partitions. If the "automatic install" decides to clear the emmc and partition from scratch, the device is a hard brick

08:20 <javierm> travmurav[m]: I see...

08:20 <travmurav[m]> the 2 is my biggest concern so far

08:21 <javierm> scary that don't have a recovery option like the chromebooks

08:21 <javierm> travmurav[m]: if you have any questions about the fedora dev process or need any assitance with that, don't hesitate to ask :)

08:22 <travmurav[m]> well, the recovery option is "there" - an EDL payload, but it's signed by the OEM and often is impossible to get

08:23 <javierm> travmurav[m]: but yeah, the installation part for non-standard devices is a PITA, for Chromebooks we are using https://github.com/eballetbo/chromebooks/

08:24 <javierm> and have some kernel-install script plugins to update the kernel (since needs to convert in a FIT, write to a kernel partition, etc)

08:24 <travmurav[m]> javierm: well, the scary part that it /is/ /almost/ standard - you get a normal looking uefi, with ESP partition and just a billion random other partitions that you /can/ but /must not/ touch...

08:25 <javierm> travmurav[m]: right... it's almost an invitation to brick the machine :/

08:25 <travmurav[m]> yeah...

08:26 <travmurav[m]> Thankfully UFS devices have dedicated LUNs (so just don't touch anything but sda to be safe) and NVME devices as well as chromebooks have the bootloader on the dedicated spi flash, so it's only the cheapest eMMC ones that are a big problem

08:44 <javierm> travmurav[m]: I see

15:23 <alpernebbi> javierm: I have https://github.com/alpernebbi/depthcharge-tools for chromebook boot stuff, would like to get it in fedora & installer and one day, but too busy to work on that so just telling you about it for now

15:28 <alpernebbi> it's used to support chromebooks in postmarketOS, and I've been integrating it to debian-installer and trying to make prebuilt installer images

16:46 <craftyguy> jhovold: thanks, ya I work on a distro and having modules makes the most sense. I was just curious if there was some other reason since everything else in your defconfig is builtin :)

16:47 <javierm> alpernebbi: cool, thanks for the info

23:43 <clover[m]> anyone else have an issue with chromium / brave under wayland where if you click off the window you have to click like 10 times to get the window to respond again?

23:43 <clover[m]> i wonder if its trying to save RAM when idling or something