Am Thu, 31 Oct 2024 10:41:45 +0000 schrieb Mithil Bavishi <bavishimithil@xxxxxxxxx>: > > well, that takes time, I wanted to start that on the right thing. > > Yes indeed, I'll be more careful the next time, again sorry for the > trouble, I am not used to the process of mailing lists and may have > done some mistakes there as well. > probably you did not notice an error in make dtbs and the old devicetree on the device was still there and was used. > > 1. make dtbs shows warnings > > > 2. make CHECK_DTBS=y ti/omap/omap4-samsung-espresso7.dtb is too > > noisy (probably same for espresso10). > > > a lot comes from the dtsi files, so you need to ignore a lot, > > probably either strip down the new dts to almost nothing besides > > dtsi includes to determine the background noise or take a similar > > device, redirect output and errors, diff that output with the full > > devicetree. I am trying to clean that dtsi warning mess up, > > linux-next shows a lot less warnings but that takes time. > > Oh, I was not aware of such tool, ran it and yeah there are a ton of > warnings, where can I ask for assitance if I need it while fixing > them. > I hope you understood that warnings like this: arch/arm/boot/dts/ti/omap/omap4-samsung-espresso7.dtb: /ocp/interconnect@4a000000/segment@0/target-module@4000/cm1@0/clocks/pad_clks_ck@108: failed to match any schema with compatible: ['ti,gate-clock'] are not the ones you need to fix, so just the diff between old and new. > > One of the warnings that should be fixed: > > dts/ti/omap/omap4-samsung-espresso7.dtb: lvds-encoder: compatible: > > 'oneOf' conditional failed, one must be fixed: ['lvds-encoder'] is > > too short 'lvds-encoder' is not one of ['ti,ds90c185', > > 'ti,ds90c187', 'ti,sn75lvds83'] 'lvds-encoder' is not one of > > ['ti,ds90cf364a', 'ti,ds90cf384a', 'ti,sn65lvds94'] 'lvds-encoder' > > is not one of ['thine,thc63lvdm83d'] from schema $id: > > http://devicetree.org/schemas/display/bridge/lvds-codec.yaml > > > > Ah right, I have to add the encoder (doestek, dtc34lm85am) in > bindings and in vendor, this patchset may grow too big I assume. > You can do the binding stuff separately, lvds-encoder/panel should be trivial, so one patch for vendor prefix if needed and one for the binding. Regards, Andreas
