07:54 <Glanzmann> jannau: For the latest m1n1 the keyboard no longer works for me in u-boot.

07:54 <Glanzmann> I assume the usb-c mini ports are dead.

07:55 <Glanzmann> I tried with your patch and the 1ms patch.

07:58 <Glanzmann> jannau: This is my script that I used to build u-boot: https://pbot.rmdir.de/PlvWBZ7MT39BTsDIin_2aw

07:58 <Glanzmann> I used the tags from the PKGBUILD repository.

07:59 <Glanzmann> Once I boot the kernel everything is fine.

07:59 <Glanzmann> With the edge config option the kernel hangs for me during boot in userland and tells something about deferred probing.

08:00 <Glanzmann> But I wait witht the edge until it is released.

08:06 <_jannau_> that's a side effect of the usb-3 support in the devicetree. u-boot misses at least a dummy drivers atcphy driver

08:14 <mps> that is on minis?

08:16 <mps> on macbook keyboard works with latest m1n1

08:17 <_jannau_> that is a problem for all devices with usb-3 support, in asahi-wip all except m2 but I have the patches for usb-3 on m2

08:18 <_jannau_> the usb-c parts are not working, nothing to do with keyboards specifically

08:18 <mps> hmm, I built latest asahi-wip, latest m1n1 1.1.8 and latest u-boot, lsusb shows usb 3.2 and keyboard works in u-boot

08:19 <mps> I mean on macbook, don't have mini to test

08:20 <_jannau_> this only affects u-boot

09:25 <Glanzmann> jannau: I see, thanks for confirming.

09:32 <Glanzmann> mps: Yes that is with the minis. The old macbooks have keyboard using spi. The new once using something else I forgot.

09:37 <mps> Glanzmann: yes, I understand

10:18 <Glanzmann> jannau: With the lastest u-boot/m1n1 version the 'display=2560x1440 >> /boot/efi/m1n1/boot.bin' results in an unbootable 'boot.bin' is that expected?

10:19 <Glanzmann> I think Ry Darcy also reported that a few weeks back.

10:20 <Glanzmann> Anyway hopefully the dcp will be soon mature enough that this workaround is no longer needed.

10:28 <j`ey> I think Ry's issue was that they didnt gzip uboot

10:31 <Glanzmann> j`ey: At least I gzip u-boot. I wrote a script for ry darcy and send it to him via email. IIRC he replied via email to me and jannau about the 'echo' issue.

10:31 <_jannau_> no, the problem is that the reallocated framebuffer is not reserved. So when the dart-disp0 is initialized the display processor can not access it anymore and crashes

10:33 <Glanzmann> jannau: I see, thank you for the inishgt.

10:33 <Glanzmann> j`ey: I checked the email so for ry darcy it was different. It still booted but no xorg started. For me it crashes.

10:34 <Glanzmann> Ry_Darcy: Hey.

10:38 <Ry_Darcy> Some info: Hallo Thomas. I compiled the asahi-6.1-rc5-7 kernel yesterday on my M1 Mini. Compiles and starts clean. I am using Glanzmann's script. Compiles and restarts (boots) clean. However, I am not using CONFIG_DRM_APPLE in my .config. So, I cannot contribute much here. I will try this evening to incluse CONFIG_DRM_APPLE in the compile and see what happens.

10:38 <Ry_Darcy> include

10:41 <Ry_Darcy> Also, I am not passing "echo display=xxxx" parameter string to boot.bin as the last time I tried to inject 4K resolution using this command resulted in me having to ssh in and copying a virginal boot.bin to get my display back to normal.

10:42 <Ry_Darcy> Not sure what the situation is at this stage.

10:59 <Glanzmann> Ry_Darcy: I see. Let me know who dcp works for you. I'll also try dcp again without the gpu (I have this running currently and everything is fine expect the window resize issue, however most of the time I work in full screen windows and use shortcuts for tiling that I don't notice).

10:59 <Glanzmann> s/who/how/

10:59 <Ry_Darcy> Thomas who=how I presume? :)

10:59 <Glanzmann> Yep.

11:00 <Ry_Darcy> Will report back this evening. I use X11 and XFCE, so nothing special..

11:01 <Glanzmann> Same for me, x11, lightdm and fvwm2, alacritty, firefox, mupdf, ...

11:20 <mps> Glanzmann: how fast is switching virtual screens with fvwm

11:20 <mps> and maximinzing mpv

11:20 <Glanzmann> It is instant.

11:20 <mps> for me with awesome wm these are slow

11:20 <Glanzmann> mps: For mpv, I also start in fullscreen.

11:21 <mps> yes, start in full screen is fine but maximizing for normal size is slow

11:22 <Glanzmann> You're using DCP?

11:22 <mps> yes

11:22 <yuka> sway got really slow with dcp for me

11:22 <yuka> already noticed that the other day when I was testing the gpu drivers

11:23 <mps> also, I didn't built gpu driver

11:23 <Glanzmann> mps: I use DCP only on the mini. On the air I'm still using simpledrm.

11:24 <yuka> I also mean with dcp but software rendering

11:24 <mps> Glanzmann: aha, I use on macbook

11:24 <yuka> but getting dcp enabled was the first step towards testing the gpu driver

11:24 <Glanzmann> mps: I tried building the gpu driver several times. Sometimes the kernel booted but failed to probe the gpu, others it gives me a black screen and no network, sometimes I reach userland but it hangs in initrd telling me about probing stuff and never finishes. However I'll wait until it is released and try again.

11:24 <mps> does mesa have apple drm driver

11:25 <Glanzmann> mps: I still don't understand what I have to do to get a functioning mesa userland for apple gpu.

11:25 <mps> Glanzmann: thanks for info. good to remember, though I still don't need gpu

11:25 <_jannau_> no, mesa has an asahi driver

11:25 <Glanzmann> There was in the irc backlog someone explaining how to to build it, maybe I wrote it down, but I never tried it.

11:26 <_jannau_> apple drm is the display part, the 3D gpu is called asahi

11:26 <mps> _jannau_: by 'asahi driver' what you mean

11:26 <Glanzmann> mps: Neither do I, but I look forward to the day we can watch a movie in fullscreen on m1 with mpv without tiering.

11:26 <mps> ah, ok

11:26 <_jannau_> the driver is called aahi

11:26 <_jannau_> asahi even

11:26 <mps> _jannau_: this is for gpu?

11:27 <_jannau_> yes

11:27 <mps> thanks

11:28 <mps> so I don't need to build mesa from upstream because I don't plan to use gpu now

11:30 <_jannau_> you would need to build mesa from lina's repo (or alyssa's), the uapi parts to speak with lina's gpu kernel driver are not upstream

11:30 <mps> Glanzmann: m_arcan talked about changing refresh rate between 120 and 60 Hz, and I tried but don't see any difference

11:31 <yuka> gpu driver (DRM_ASAHI) worked fine for me from the appropriate kernel/m1n1 branches, _but_ could only get it to work when booting as m1n1 guest

11:32 <_jannau_> I can't reproduce the mouse pointer choppiness with 120Hz on j314/j316 either

11:32 <mps> I will not yet waste time for gpu because I'm not sure how to build kernel with rust on alpine

11:32 <_jannau_> window resize slowness seems to be an rendering issue on X11

11:32 <Glanzmann> jannau: I saw the mouse pointer choppiness on my mini but this was ages ago and some point you fixed it. I have not seen it for ages.

11:34 <mps> but all in all dcp is now very usable. Thank you _jannau_

11:34 <Glanzmann> jannau: So at some point when the gpu is ready, in order to have fast mpv video playback, I need to build the asahi mesa drivers and than tell mpv to use 'gl' as 'vo'?

11:35 <mps> heh, new mpv is fast with drm but on console

11:35 <Glanzmann> jannau: Yes, good job on the dcp, keyboard, trackpoint and god knows what else. :-)

11:36 <Glanzmann> mps: I have not tried that on m1. But I also used mpv on simpledrm in xorg to watch a movies for multiple hours.

11:36 <mps> Glanzmann: yes, mpv with simpledrm is quite fine

11:37 <Glanzmann> yuka: I see, any idea why that is?

11:37 <Glanzmann> yuka: Because I always try the standard boot chain so m1n1, m1n1 stage2, u-boot, grub, kernel/initrd.

11:37 <mps> Glanzmann: I don't complain about mpv (or anything) just telling how things works for me

11:38 <mps> to repeat, everything is ok

11:40 <mps> Glanzmann: btw, because you don't use KDE I have hacky keyboard backlight daemon which dim leds after timeout of activity of input devices, works only with xorg. I you need it

11:41 <Glanzmann> Hrhr, I always turn off my keyboard backlight unless I'm in a complete dark room. But let me see it, maybe I can use it at some point.

11:42 <mps> also I don't use much backlight on kbd, but sometime do. here it is https://tpaste.us/en1b

11:44 <Glanzmann> mps: Thank you.

11:44 <yuka> Glanzmann: no idea what is the issue when booting gpu drivers via the normal boot chain, since there is no way to debug it really

11:44 <Glanzmann> I see.

11:45 <mps> Glanzmann: np, you can tweak it by cli params brightness and timeout

12:22 <Glanzmann> c

15:00 <Ry_Darcy> @Glanzmann, running on linux-asahi-6.1-rc5-9 (Debian, XFCE, x11) with DRM_Apple enabled on an M1 Mini.

15:02 <Ry_Darcy> Of course @jannau also

15:02 <Ry_Darcy> https://paste.debian.net/1260907/

15:03 <Ry_Darcy> https://paste.debian.net/1260908/

15:04 <Glanzmann> Ry_Darcy: Nice, thank you for the feedback.

15:04 <Glanzmann> Will try the same later and report back.

15:05 <Ry_Darcy> I notice a noticeable lag on drawing screens, as has been mentioned already.

15:05 <Glanzmann> I see.

15:08 <Ry_Darcy> https://paste.debian.net/1260909/

19:29 <Glanzmann> jannau: So, after updating to Ventura, I tried the 'edge' config again. It could not find my nvme device. But when I set every config option which is '=m' to '=y' than it boots but crashes or kills usb shortly after boot. Also it does not detect the 2k resolution.

19:29 <Glanzmann> jannau: here is the dmesg: https://pbot.rmdir.de/c-rLnDZFIIgizj068Vz8hA

19:30 <Glanzmann> I used the asahi reference config from the pkgbuild.

19:31 <Glanzmann> I only changed the config.edge options from =m to =y.

19:32 <Glanzmann> HBuildscript: https://pbot.rmdir.de/Yi-Lf7pv0JnxSoEdO_W8ug Config: https://pbot.rmdir.de/b1YpYRx-09Ia7ptCRYYH1w

19:32 <Glanzmann> s/H//

19:34 <Glanzmann> jannau: That was on the mini with a 2k display (Eizo EV2780)

19:35 <jannau> known issue for devices with more than one iommu. not sure if it's a bug in our dart code or in the kernel device probing

19:36 <Glanzmann> I see, thanks for letting me know.