14:52 <konradybcio> sboyd: i'm wondering why CLK_OPS_PARENT_ENABLE is optional.. just hit another bug caused by the lack of it and I'm now unsure how this isn't the default.. some clock controllers implement auto-enable-parent-on-child-enable, but surely this can't be the expected standard?

14:53 <konradybcio> setting it for e.g. all qcom clocks isn't feasible, as clk_init_data which houses flags is const

21:42 <z3ntu1> lumag: any feedback from you on writeback now? ;)

21:42 <z3ntu1> or rather idea what could be wrong?

22:14 <lumag> z3ntu1, do you have any compositor running at the same time?

22:14 <lumag> [ 70.842815] msm_dpu ae01000.display-controller: [drm:drm_mode_getconnector] User-space requested a forced probe on [CONNECTOR:38:Writeback-1] but is not the DRM master, demoting to read-only probe

22:14 <lumag> Or maybe you run that from the non-root userS?

22:16 <z3ntu1> lumag: yes phosh UI is running

22:17 <z3ntu1> The UI is running as non-root, I always ran modetest with sudo I think

22:17 <lumag> z3ntu1, I think trying the modetest -ac from when you are switched to non-UI vc might help.

22:23 <lumag> z3ntu1, at the same time I wonder why total_modes_count += connectors[i]->funcs->fill_modes(connectors[i], width, height); doesn't work

22:34 <lumag> z3ntu1, I see... drm_client_for_each_connector_iter() explicitly skips writeback, that's why WB is not probed during drm_client_modeset_probe(). So yep, starting modetest -ac as a master (root, no other masters) should be the way to go.

23:44 <Adrian[m]> <Marijn[m]> "Doublecheck the init sequence..." <- I double checked the init sequence and pinctrl. I've also added a `regulator_bulk_enable` call in the panel driver to enable the supply regulators and a call to `mipi_dsi_dcs_set_display_brightness` at the end of the prepare (tried a couple of values), but none of that changes anything.

23:44 <Adrian[m]> Behaviour is still the same: No error logs, but when I try to read `actual_brightness` I see the encoder errors in dmesg.