12:51 <pak0st> hmm, how should I get /dev/dri/card0 to show up? Wouldn't it be enough to install the `soc-qcom-sdm845-nonfree-firmware` package to get the andreno firmware stuff up and running? (can't find complaints in dmesg apart from not finding the said /dev/dri entry)

17:00 <f_> pak0st: XY Problem?

17:27 <pak0st> f_ : I'm trying to enable the GPU on the LG G7 but I'm not seeing any errors about it in dmesg. I thought to start with the firmware but maybe there is something else to be enabled before that?

17:28 <f_> SDM845?

17:28 <f_> AFAIK it needs firmware

17:30 <pak0st> that's what I thought as well - installed the aforementioned qcom non-free package but it doesn't look like I'm successful (/dev/dri is empty)

17:32 <f_> hm

17:36 <aka_[m]> something in logs?

17:36 <flamingradian[m]> is gpu enabled on your device in dts?

17:36 <f_> pak0st: dmesg?

17:36 <flamingradian[m]> (also check /sys/kernel/debug/devices_deferred)

17:37 <aka_[m]> is root adreno node enabled

17:41 <pak0st> I've added the gpu block to judyln dts: https://gitlab.com/sdm845-mainline/linux/-/blob/sdm845/6.7-dev/arch/arm64/boot/dts/qcom/sdm845-oneplus-common.dtsi?ref_type=heads#L387

17:41 <pak0st> will check /sys/kernel/debug/devices_deferred in a few

17:42 <pak0st> aka_ : root adreno node? not sure what that is (so most likely not?)

17:42 <aka_[m]> &adreno {}

17:42 <aka_[m]> in device dts

17:42 <aka_[m]> adreno prob is disabled by default

17:47 <pak0st> mmm... looks like there are 2 bounces to reach judyln - LG Common: https://gitlab.com/sdm845-mainline/linux/-/blob/sdm845/6.7-dev/arch/arm64/boot/dts/qcom/sdm845-lg-common.dtsi?ref_type=heads#L12 aaand adreno is defined in sdm845: https://gitlab.com/sdm845-mainline/linux/-/blob/sdm845/6.7-dev/arch/arm64/boot/dts/qcom/sdm845.dtsi?ref_type=heads#L4771

17:47 <pak0st> I don't think the &adreno block is the problem

17:48 <aka_[m]> indeed active

17:49 <pak0st> uuuhh.. if I don't see /sys/kernel/debug/ maybe I'm missing the required debug flags to get any output in dmesg 🤔

17:55 <pak0st> dmesg has a few references to firmware blobs but not which ones (could just as easily be wifi or smth else): https://dpaste.com/9S4UVSFHG

17:58 <f_> pak0st: you can do `mount -t debugfs debug /sys/kernel/debug` to have it.

17:59 <pak0st> thank you!

17:59 <pak0st> deviced_deferred is empty

17:59 <f_> Also these references to firmware have nothing to do with adreno.

18:00 <f_> >[Firmware Bug]: Kernel image misaligned at boot, please fix your bootloader!

18:00 <f_> >psci: [Firmware Bug]: failed to set PC mode: -3

18:00 <f_> You can thank LG for misaligning kernels at boot.

18:03 <pak0st> I think I've misaligned my kernel on the pc tower as well - I'm not bothered by that :) (much)

18:04 <pak0st> blah, I guess I would have to use gallium to check the panel-simple addition?

18:14 <pak0st> oh, that's interesting: https://dpaste.com/3MRAC3Q7M am I not installing the custom kernel as expected?

18:17 <pak0st> I'm using `pmbootstrap build --envkernel linux-postmarketos-qcom-sdm845 && pmbootstrap sideload linux-postmarketos-qcom-sdm845 --install-key` but the above makes me think that I haven't loaded the custom one as expected

18:21 <pak0st> something to look tomorrow I guess