05:53 <winter> tpw_rules: Had to uninstall NixOS due to storage space concerns, I'll eventually install it again and try to debug stuff. Random question when you get a sec, if you don't mind: I remember there being some compilation error when setting `boot.systemd.initrd.enable = true`. Do you happen to mind quickly trying to build your configuration with that

05:53 <winter> enabled and confirm? I think it had something to do with like, hardware keys? Don't recall exactly, but want to make sure that it's expected. Thanks.

05:55 <winter> Sorry, the option is `boot.initrd.systemd.enable`.

09:46 <mps> ncopa: I'm preparing upgrade for linux-asahi for alpine, should we change CONFIG_DRM_APPLE to =m? I think this will be better option for now instead of =y

09:47 <mps> also, will disabel CMA

09:54 <ncopa> sounds good to me

09:55 <ncopa> what puzzles me a bit is that there was no attempt to also include the 6.1.1

09:56 <ncopa> i wonder if we should create a patch on top of 6.1.1, similar to what we do with linux-rpi

10:04 <ncopa> asahi-6.1-3 was done from asahi-wip branch

10:32 <mps> yes

10:33 <mps> but I'm not sure that we have to rebase these patches on top of 6.1.1

10:33 <ncopa> i already have an asahi-6.1.1-3-alpine.patch

10:33 <mps> I don't see any serious reason in 6.1.1 upgrade for this

10:34 <mps> imo, it will be waste of our time

10:34 <mps> but do whatever you like to do

10:38 <ncopa> mps: so will you tag the 6.1-3 release or do you want me to do it?

10:40 <mps> I'm right now testing build but if you want to upgrade please do, just disable CMA then

10:41 <mps> if you take upgrade I will test some other options combo locally

10:42 <mps> looks like current alpine mesa works fine with DRM_APPLE so maybe we shouldn't build it as module

10:42 <mps> I have split mind about this

10:42 <ncopa> i use it with the notch thingy. works fine for me

10:43 <mps> simpledrm is more stable but DRM_APPLE have brightness

10:43 <mps> ok, then we should keep as =y

10:45 <mps> (go forward and don't look back :) )

10:57 <mps> ncopa: pushed upgrade to builder

11:08 <ncopa> thanks

11:08 <ncopa> i think uname shows wrong version though

11:13 <mps> uhm

11:15 <mps> ncopa: I have spare m1 mbp which keeps dust mostly and on it I could test your install process, do you have any url with guide for that

11:17 <ncopa> mps: not yet

11:19 <mps> aha, ok

11:20 <ncopa> mps: do we have any plans to do multiple buidls of asahi kernel with different configs? I think maybe not

11:20 <mps> ncopa: I contemplated this but didn't come to any conclusion

11:21 <ncopa> the current APKBUILD uses a list of _flavors which opens up for this, but does not use out-of-tree builds, which closes this possibility again

11:22 <ncopa> I'm gonna simplify it

11:22 <mps> for now I think we don't need it except if we want gpu driver, and then we will have to add mesa-asahi pkg to aports

11:23 <mps> yes, I left _flavor just in case we need it or someone wants to use multi flavor locally

11:23 <ncopa> they cannot build multi flavor anyway

11:23 <mps> why?

11:24 <ncopa> because build is done in-tree

11:24 <mps> ah yes, forgot that

11:25 <mps> then you could simplify it

11:25 <mps> never like _flavor in alpine kernels APKBUILD ;-)

11:26 <mps> s/like/liked/

11:27 <mps> ncopa: btw, asahi distro build two version of linux-asahi from same PKGBUILD, without _flavor

11:28 <ncopa> good for them

11:29 <ncopa> they do have two flavors: "base" and "edge"

11:30 <mps> it is tempting for me to have kernel flavor with gpu in alpine

11:52 <ncopa> looks like that is what asahi edge kernel adds

11:55 <ncopa> alright.. so we do want different kernel configs. I'll revert the last hour of work i did

11:57 <mps> ncopa: if you want different kernels I'm not against

13:00 <ncopa> mps: are you sure apple silicon does not use CMA?

13:01 <mps> ncopa: well, jannau confirmed that and also gregkh told me that it is used in rare cases with some SoCs and that it should be enabled on machines where we are sure it is used

13:02 <ncopa> ok

13:02 <mps> I opened bug report on alpine for linux-edge about this

13:03 <ncopa> i was not sure if m1 chip was a rare case of SoC or not

13:03 <mps> looks like it is not

13:03 <jannau> ncopa: yes, every HW block which might require physically continuous memory is behind an iommu

13:04 <ncopa> thanks

13:04 <mps> looks like CMA for SoCs which don't have iommu

13:04 <mps> fo

13:04 <mps> for*

13:05 <mps> uhm

13:05 <jannau> not sure how rare it is with older SoCs

13:05 <mps> mostly arm32

13:06 <mps> hmm, better say 'some arm32'