07:59 <chadmed> stintel: are you using the overlay/gentoosupport stuff?

08:01 <stintel> yes

08:01 <chadmed> oh nvm sorry just saw that you already tried with dracut

08:01 <chadmed> thats interesting

08:01 <stintel> not all of it though, I need patches asahi-wip kernel as the MBA 15 is not supported yet

08:01 <chadmed> because i had similar problems with genkernel and kmod not being built with zstd support

08:02 <stintel> s/patches/patched/

08:02 <chadmed> and the .config youre using is the one from the arch pkgbuild?

08:02 <stintel> yeah

08:03 <stintel> I'm already mighty impressed with the performance of this thing but using a browser with nomodeset is quite painful :)

08:03 <chadmed> no freaky KRUSTFLAGS or KCFLAGS?

08:04 <stintel> it freezes even without RUST enabled in the kernel at all

08:06 <chadmed> hm ok, did you try with simpledrm and no dcp driver

08:06 <j`ey> I dont think you want dcp without GPU

08:06 <j`ey> I heard that's worse than simpledrm?

08:07 <chadmed> because where its freezing looks kinda like where dcp tries to take over the fb

08:07 <chadmed> if it works with simpledrm and no nomodeset flag then we know its dcp something

08:07 <jannau> yes, it works without dcp, that's basically what nomodeset does

08:08 <chadmed> yeah ok

08:08 <stintel> what's the config symbol for DCP? not finding DCP

08:08 <chadmed> DRM_APPLE

08:08 <stintel> ah ok

08:08 <jannau> it's not dcp, it seems to be an issue of probing dcp's dart

08:09 <stintel> so yeah that's what I'm doing - simpledrm with nomodeset so drm_apple isn't loaded

08:09 <jannau> see https://www.linux-ipv6.be/1691543656845.jpg

08:11 <chadmed> ohh ok

08:11 <chadmed> sorry for the repitition

08:11 <chadmed> saw gentoo and just wanted to rule out any userspace weirdness for myself

08:11 <stintel> this particular trace was related to order of loading modules, due to only including few modules in the initrd (this was before I added all modules that are in /usr/lib/dracut/modules.d/99asahi-firmware/module-setup.sh in genkernel, and later I switched to dracut)

08:11 <jannau> the 'sysfs: cannot create duplicate filename ".*iommu.*dcp'

08:13 <jannau> stintel: can you try to the on device kernel via usb/hv. That way you will get the full kernel log as text

08:13 <stintel> so I'm on asahi-wip kernel with mba 15 dts patched in, u-boot asahi-releng + same DTS, and m1n1, I don't need a different m1n1 branch do I?

08:14 <jannau> no

08:16 <jannau> you can boot your system via hv if you chainload a m1n1.bin which has parameters appended like

08:18 <stintel> https://github.com/AsahiLinux/docs/wiki/m1n1%3AUser-Guide#running-a-linux-kernel-as-a-m1n1-hypervisor-guest-with-virtual-uart ?

08:18 <jannau> cat build/m1n1.bin <(echo -n "chosen.asahi,efi-system-partition=8b03438a-ad0e-c14a-900c-fc7e535882e9\nchainload=8b03438a-ad0e-c14a-900c-fc7e535882e9;m1n1/boot.bin\n\0\0\0\0") > build/m1n1_system.bin

08:19 <jannau> and then tools/run_guest.py ../build/m1n1_system.bin

08:19 <stintel> ok will give that a try

08:19 <jannau> replace the 8b03438a-ad0e-c14a-900c-fc7e535882e9 with the PART_UUID of your EFI partition

08:42 <stintel> having some trouble getting that to work: https://gist.github.com/stintel/f05523ae2ff7d95fcd12234c34157f93

08:46 <_jannau__> missing `-r`in the command line, sorry

08:49 <stintel> that seems to lock up: https://gist.github.com/stintel/39f8ad15c1a2bc2cba9d60deaeaf7618

08:49 <stintel> probably doing something wrong

08:51 <_jannau__> that seems like appending the variables did not work

08:53 <stintel> they seem to be there in the m1n1_system.bin - see comment on the last gist

08:58 <_jannau__> can you check why the variables are not detected. see check_var() in src/payload.c

08:59 <_jannau__> the unknown magic matches "chos"

09:02 <stintel> if (!end) matches

09:02 <stintel> s/matches/evaluates to true/

09:03 <_jannau__> missing -e in the echo command

09:07 <stintel> hmm now if (!val) evaluates to true

09:13 <stintel> hmm wait

09:15 <stintel> https://gist.github.com/stintel/2a14c29e180136a25e2fc1f9648110a9

09:15 <stintel> so it finds both variables but then: Chainloading files not supported in this build!

09:15 <j`ey> 'Chainloading files not supported in this build!

09:15 <j`ey> need to build with rust

09:17 <stintel> alright

09:17 <_jannau__> that, or just grab the m1n1 + u-boot + dtb bundle and append just 'chosen.asahi,efi-system-partition=$P{PARTUUID}\n\0\0\0\0'

09:17 <_jannau__> 'chosen.asahi,efi-system-partition=${PARTUUID}\n\0\0\0\0'

09:21 <stintel> that worked. seems it's just the display that locks up: I believe this is where the useful info starts: https://gist.github.com/stintel/bf4dc23519e22a7b9daa014a4e6cfec7#file-gistfile1-txt-L432

09:25 <stintel> is this because disp0_dart is status disabled ?

09:26 <_jannau__> stintel: can you see why disp0_dart is disabled? it shouldn't

09:27 <stintel> _jannau__: it is disabled in t8112.dtsi and never enabled afaict

09:30 <_jannau__> right. it should be enabled by m1n1 see https://github.com/AsahiLinux/m1n1/blob/8e3016e323dccbbf16aa2d297d4a5e38bb38287d/src/kboot.c#L1406

09:33 <_jannau__> can I get the output of m1n1 which executes u-boot as payload

09:36 <stintel> _jannau__: this? https://gist.github.com/stintel/c7b115307bb6629151a099be6cf06a1d

09:37 <_jannau__> no, the one in the hypervisor

09:39 <stintel> from the serial console? https://gist.github.com/stintel/6e2de803614853c619c75e93018218b0

09:52 <stintel> /sys/firmware/devicetree/base/soc/iommu@231304000 # cat status

09:52 <stintel> okay

09:52 <stintel> so it actually is enabled

09:53 <_jannau__> that looks sane. please paste the output of `dtc -I fs -O dts -o - /proc/device-tree | sed -n '/^\t\tdcp@/,/^\t\t};/p'`

09:54 <_jannau__> still looks if you have an old devicetree in the m1n1 + ... bundle

10:26 <stintel> https://gist.github.com/stintel/6e19e5ad9ebdfb69094a1a1b1f92b928

10:27 <stintel> interesting, there the dcp does not have the piodma node

10:27 <stintel> could this be because of the m1n1 installed by the installer?

10:29 <stintel> ahh I think I see what's going on. the t8112.dtsi in u-boot is outdated

10:30 <_jannau__> no, that is the dtb from m1n1/boot.bin or the one youused with run_guest.py

10:30 <_jannau__> yes, for now always use the dtb from the same kernel you're booting

10:33 <stintel> sweet

10:33 <stintel> than you so much for helping me get that sorted

17:33 <stintel> man this thing runs like a dream

17:35 <stintel> _jannau__: thanks again. appreciate your patience :)

17:36 <stintel> and all the instructions are in my irclogs, so if I hit some problems in the future I know at least a few things to try