01:54 <_mike> hi

02:14 <ellyq> hello there

02:46 <_mike> hi ellyq

06:47 <_mike> yay

06:48 <_mike> i got qemu-user-static going

06:48 <_mike> now arch-chroot to my x86 bootstrap works

07:53 <_mike> nice now i have wine working

08:05 <HdkR> :blobsweat:

08:06 <HdkR> The performance of wine running with qemu has got to be amazing

08:11 <_mike> i think i need vulkan-virtio built into my kernel

08:12 <_mike> HdkR: all i need it for is mikrotik winbox but i'm trying steam while i'm at it

08:15 <HdkR> Watch out, Steam's integrated CEF has a time based watchdog that restarts the process if it doesn't ping every 30 seconds or something. So slow environments are rough

08:20 <_mike> oh there is a vulkan-virtio package!

09:06 <SpieringsAE> macc24: this machine does not have acpi, it uses U-boot -> uefi boot its not my snapdragon machine

09:07 <SpieringsAE> the vendor dtb also does not describe it properly unfortunately

09:08 <travmurav[m]> what's that fun device?

09:08 <SpieringsAE> travmurav: mine?

09:08 <travmurav[m]> yes, the one with hdmi muxing

09:08 <SpieringsAE> its the deepcomputing fml13v01, its the one made for the framework 13 chassis

09:09 <travmurav[m]> ah the risc-v motherboard?

09:09 <travmurav[m]> interesting

09:09 <travmurav[m]> hm

09:09 <SpieringsAE> yep

09:09 <travmurav[m]> does the mux switch transparently to the OS or linux has to swithc when it detects type-c alt mode device?

09:10 <SpieringsAE> it seems to do it automatically, I dont really understand the logic it uses for it. there is no description of it in the dtb so I don't see how linux would do it

09:10 <travmurav[m]> well in that case I guess best you can do is to just leave the "port" dangling in the dt and write a comment that platform handles that "somehow"

09:11 <travmurav[m]> I guess if you have access to schematics it would be easeier to explain "how"

09:11 <SpieringsAE> it uses the asw3642 as a switch

09:11 <SpieringsAE> I want to attach the backlight node to the actual panel though, which is currently not the case, ah well

09:11 <travmurav[m]> mhm

09:12 <kuruczgy[m]> _mike: if you want the best perf you might want to eventually look into drm native context, that should be significantly more performant than venus. (I got it working, but it needs a patched qemu and a guest mesa with some extra features.)

09:12 <travmurav[m]> well I guess the dumb way would be to descibe just the internal panel then, does it work that way?

09:12 <travmurav[m]> if there is some way to poke platform mcu/gpios to force alt mode on/off then perhaps could later write a custom bridge mux for that

09:13 <SpieringsAE> Yeah I'm hoping there is some sort of feedback switch from that asw3642

09:13 <kuruczgy[m]> _mike: also virtual cpufreq for the vm (I haven't done that yet, so right now it's just all cores running at 100% cpu. not sure why this is an issue on arm and not on x86...)

09:13 <travmurav[m]> but my personal opinion is that if we can do something "today" that "apparently works properly", it's fine to do that and fix "theoretically correct" stuff later

09:26 <vanveldt> 6

09:37 <_mike> kuruczgy[m]: ooh can you point me to some info on drm native context?

09:40 <kuruczgy> _mike: if you just want some background on the tech, I think robclark's original talk on it is a good intro, I think this is it: https://www.youtube.com/watch?v=9sFP_yddLLQ

09:41 <kuruczgy> for the setup, my Nix expressions could be helpful to read, e.g. for qemu: https://codeberg.org/kuruczgy/aarch64-gaming/src/branch/main/overlays/qemu.nix

09:41 <_mike> thx

09:43 <kuruczgy> basically host virglrenderer and guest mesa need an extra compile flag, and you need to compile qemu from that specific branch plus add the resource uuid patch

09:43 <kuruczgy> but if you don't want the trouble just check back in half a year, the qemu stuff might be upstream by then

13:09 <icecream95> Hmm, so OpenMW works pretty well on x1e except that the sun is too bright: Freedreno (and Turnip) use early-z for occlusion query draws with discard/alpha-test if depth write is masked, meaning samples are counted which are later discarded in the FS, therefore the game thinks the sun is a lot more visible than it really is...

13:27 <HdkR> icecream95: That's quite a bad bug

13:29 <HdkR> Probably should be reported in the mesa bug tracker

22:26 <_mike> kuruczgy: is that patch included in qemu-9.2.0?

22:28 <_mike> and can i just add -C msm drm-renderers to meson?

22:29 <_mike> err

22:29 <_mike> i think i need -Doption=msm

22:29 <_mike> etc

22:30 <_mike> meson -Doption=msm,drm-renderers install -C build-static --destdir "$pkgdir"

22:30 <_mike> i'm going to try that

22:31 <kuruczgy> _mike: I don't know, but I am pretty sure not everything is upstream yet. You might want to ask digetx (not sure if they are in this channel, maybe try asking over in #freedreno)

22:32 <_mike> kuruczgy: ok, thanks heaps

22:33 <kuruczgy> _mike: more like `-Ddrm-renderers=msm`, I think that's the proper syntax

22:33 <_mike> kuruczgy: thanks!

22:33 <kuruczgy> (but this is for virglrenderer, not qemu)

22:37 <_mike> oh ok i get it

22:37 <_mike> virglrenderer for arch uses venus

22:37 <_mike> i'll try that first

22:52 <_mike> kuruczgy: fwiw it's definately a qemu patch, not virtglrenderer

22:52 <kuruczgy> _mike: the patch is for qemu, the flag is for virglrenderer

22:52 <_mike> oh right

22:53 <_mike> thanks for clarifying

22:56 <_mike> kuruczgy: i've never used nix

22:57 <kuruczgy> _mike: yeah sry it's the only form that I have this documented in, and there was at least someone in the a

22:57 <_mike> (;

22:58 <kuruczgy> past who had success just following my nix expressions and "doing it manually", so I figured it's better than nothing

22:59 <_mike> i'm recompiling qemu now i put the patch in prepare() so i assume it's applied

23:44 <_mike> kuruczgy: i'm still getting msm: driver missing in the chroot

23:45 <_mike> do i need to install the virglrenderer in the chroot?