07:54 <MichaelLong> hm after the last update of my gentoo system its u-boot installation does not enumerate any USB-controllers. Unsurprisingly this leads to issues e.g. I cannot use the keyboard in grub to select another kernel, my default kernel boots but I cannot login nor can I ssh into it via ethernet (wifi probably only connectes after a successful KDE login). What could that be? My Fedora installation works fine.

07:57 <jannau> leio: anything before that spam? does replugging the hdmi connector fixes it? I'll see what can be done to stop the spam

08:01 <jannau> MichaelLong: do you remember what was updated? I assume this is a desktop system. When was the last update before that?

08:02 <jannau> yopu could try to put the m1n1/boot.bin from your fedora esp onto your gentoo esp

08:02 <MichaelLong> jannau, correct it is a desktop system. A lot of stuff was updates. Now I try to chroot from my fedora installation into the gentoo one and check emerge-log

08:02 <MichaelLong> jannau, that would my next question, could I somehow "transplant" the u-boot from fedora to gentoo

08:02 <jannau> when was the last time you ran update-m1n1?

08:03 <MichaelLong> at the time of that update, on 1st of May

08:03 <jannau> MichaelLong: copying m1n1/boot.bin will transplant u-boot

08:04 <MichaelLong> I noticed a change in u-boot, the font is not much bigger than before so there must have been a change

08:04 <jannau> which version does u-boot report?

08:04 <MichaelLong> jannau, I'll check that, missed to note hat

08:09 <MichaelLong> the gentoo u-boot version is 2023.07.02

08:09 <MichaelLong> trying now the transplant :)

08:13 <MichaelLong> it is kinda weird, this non working u-boot file is about 1.8mb in size, there's also a boot.bin.old (which I could try) which is about 3.12mb in size which is basically very similar to the size of the fedora's u-boot bin-file.

08:15 <leio> jannau: I tried one replug, that didn't change anything. Also I then shut down the machine for the night, specifically checked that the power LED DID turn off, but in the morning now the machine was hot and been running on some kernel oops spin or whatever again.

08:17 <leio> jannau: there's more interesting spam in journal I didn't check before, including fun like "pipewire[1854]: spa.alsa: 'hw:1': playback open failed: No space left on device" whereas I have 7GB free on the most full partition.

08:19 <jannau> MichaelLong: that sounds like it's missing dtbs. if you did a kernel update you probably did not ran `make dtbs_install`. dtb path was changed a few weeks ago and we haven't changed back after I fixed asahi-kernel

08:19 <MichaelLong> jannau, thanks for the hint with the boot.bin file, would not be able to find it that quickly myself. For now I'm backup business again :D

08:21 <leio> jannau: https://bpa.st/2VZA

08:21 <jannau> leio: the ENOSPACE is most definitely coming from the audio device

08:21 <leio> (my journal doesn't seem to include dmesg?)

08:22 <jannau> MichaelLong: run `make dtbs_install` from you kernel build dir and run update-m1n1 again. that should you bring back on track

08:23 <MichaelLong> jannau, I'll try that

08:25 <MichaelLong> looks better, let me reboot

08:30 <MichaelLong> jannau, that worked. I've adjusted my kernel-compile script to add that step. Thx again. Now back to the original issue I wanted to fix: the system is running on llvmpipe

08:31 <jannau> what does `make rustavailable` in the kernel dir report?

08:38 <MichaelLong> Ok I really need to sort this out first. At the last update kernel 6.8.8_p2 was installed, when going into its build dir the make rustavailable is giving me its usual output (although slightly too new). However this kernel hangs for me when SDDM is supposed to start - im only seeing a non blinking cursor in the upper left corner. The former (now running kernel) 6.6.0_p14 tells me no rule to make...

08:39 <MichaelLong> so I either check the "old" kernel about the rustavailable basics, or I look into the 6.8.8-issue

08:45 <chadmed> MichaelLong: when you updated the kernel, did you copy your old kernel config into your new kernel build dir first?

08:45 <MichaelLong> chadmed, sure

08:46 <chadmed> if you go into menuconfig, are DRM_ASAHI and RUST selected?

08:47 <MichaelLong> chadmed, checked both it is on.

08:48 <MichaelLong> hm this is weird now 6.8.8 boots fine SDDM comes up and I can log into KDE but it does not really load the desktop, getting only a black screen. dmesg has no outputs regarding graphics

08:48 <chadmed> do you recall what version of mesa was installed when you updated?

08:49 <MichaelLong> chadmed, 24.1.0_pre20240228-r1

08:50 <chadmed> everything seems fine then. what device do you have

08:51 <MichaelLong> M1 Pro

08:52 <chadmed> can you be more specific please

08:53 <jannau> MichaelLong: did you enable APPLE_SIO for 6.8.8?

08:54 <chadmed> is that still borked on M1? its fine on my m2 max studio now

08:56 <MichaelLong> chadmed, sorry give me a moment please, jannau yes

09:01 <MichaelLong> chadmed, https://pastebin.com/qktS5Dpu

09:02 <chadmed> hm yeah okay i tested on my j314 and it was fine

09:02 <chadmed> maybe try disable APPLE_SIO?

09:03 <MichaelLong> I can give it a shot

09:08 <jannau> works here on all my systems but unexpected configs are morte liekly to result in breakage. probably nothing wrong with the kernel then

09:20 <MichaelLong> hm interesting re-compiled the 6.8.8-kernel, now it is working as expected including having apple_sio loaded. Not sure what changed though :/

09:22 <MichaelLong> thx again for the help so far!

09:22 <chadmed> wind direction changed

15:33 <faddat> hey, anyone have recommendations on "the best way" to have nix do stuff at boot time?

15:34 <faddat> specifically join networks. I figured out the distro (and big thanks to @PaulFerster) -- it is to enable "squad computing"

16:29 <leio> chadmed: that LLVM_MAX_SLOT bump in mesa doesn't actually get it to link with llvm-18; still getting 17. It's all completely different in main tree version, it all could use a full sync

16:29 <leio> (and I need to deal with gnome instead still)

19:59 <jannau> leio: youi can check with `cut -d' ' -f 1,2,18,19 /proc/$(pidof kwin_wayland)/task/*/stat` and s/kwin_wayland/mutter/

20:01 <leio> -21

20:01 <leio> 1372 leio -21 0 8089M 692M 534M S 0.6 0.4 0:22.09 │ │ ├─ KMS thread

20:01 <leio> PRI NI VIRT RES SHR S CPU% MEM% TIME+ Command

20:01 <leio> oops, lost the block select, first are PID and user

20:05 <leio> I suspect the custom thread name containing space is messing with the stat one

20:05 <leio> 1372 (KMS 3024 -21

20:05 <leio> yeah, 1,2,3,19,20 is better: 1372 (KMS thread) -21 0

20:51 <leio> with just the bit flag added manually on top of 6.8.8-2, I'm having a harder time to get glxgears below 50 fps now; es2gears_wayland sometimes goes 30-50fps even without touching it at its original size though, but probably unrelated and I didn't test it without resizing it much before.

20:57 <jannau> that sounds bad. I guess I need to test with mutter. with the fixed kwin and kernel I can't get below 59fps anymore

20:57 <leio> I do need to still get to gnome-shell 46, might help.

21:00 <jannau> I've added a package.use with 'kde-plasma/kwin caps' for the overlay but I think at lease kwin 6 should get IUSE="+caps"

21:00 <leio> please file a bug, asturm will hopefully get on it then

21:06 <leio> I guess CC to asahi@g.o might be appropriate too

21:17 <jannau> https://bugs.gentoo.org/931215

21:23 <leio> wrangled that, thanks. Good night