04:19 <Daanct12> anyone here working on a dmc driver for rk35xx?

09:03 <mmind00> minute: you're adding another step in your device-chain (dsi -> bridge -> panel-whatever), so this might affect probe ordering, hence some clock or regulator getting auto-disabled in the meantime?

09:05 <minute> mmind00: thanks, that's a good idea. thinking what's the best strategy to debug this...

09:08 <mmind00> minute: to get a feeling if I'm on the right track, I mostly go the brute-force variant in these cases ... commenting out the actual disable parts in regulator_disable and clk_disable ;-)

09:09 <minute> mmind00: ah, good point :D will try

09:15 <robmur01> although the "{clk,pd,regulator}_ignore_unused" arguments are handy if your bootloader is conducive to quick command-line edits

09:25 <CounterPillow> Danct12: nope, but I assume that's used for doing the same kinda memory bandwidth measurement perf stuff as rk3399 has?

09:25 <minute> robmur01: i believe i already have clk_ignore_unused but not the others

09:29 <CounterPillow> btw I'm currently working on tracking down a PCIe3x4 regression on RK3588 with v6.15-rc that's causing a specific M.2 NVMe drive of mine to fail allocating bar space for most boots depending on how quickly it initialises, so that one's fun. If anyone has noticed similar, please let me know.

09:57 <diederik> Since 6.14 the screen on my PT2 stopped showing any output (thus stays blank). I don't know (yet) why, but I suspect it also has something to do with how quickly 'things' initialize

09:59 <diederik> The ``clk: Disabling unused clocks`` and ``PM: genpd: Disabling unused power domains`` lines occur just after 2 seconds while it was at 1.7 secs on 6.13. Maybe be a coincidence and/or irrelevant. Maybe not

13:53 <minute> ah m))) it's actually not freezing anymore, just yanking the console away from serial to invisible framebuffer console

14:01 <minute> ok, now need to solve this wrong calculation > rockchip-drm display-subsystem: [drm] *ERROR* MIPI dclk out of range, dclk_out_rate: 40500000 KHZ. iirc there was a patch for that already, weird that it's not in my stack now

14:01 <minute> 40 ghz :3

14:09 <minute> yeah https://lore.kernel.org/lkml/20241115151131.416830-1-heiko@sntech.de/

15:01 <minute> ok wow i had the enable gpio accidentally set to the same pin as the i2c clk... well that explains it

15:01 <minute> interesting that that's possible

15:09 <CounterPillow> pinctrl will tell you about conflicts but if you don't assign pins to devices through pinctrl it won't know about it

15:33 <minute> CounterPillow: alright

15:48 <minute> hmm, looks like hdmi0 can still make the system freeze at boot, at least on 6.13 (nothing connected to hdmi, just having the drivers in dtb). freezes around every second boot

15:49 <minute> with hdmi0 commented out it boots reliably and brings up dsi every time. i also have {clk,pmdomain,regulator}_ignore_unused in place

16:03 <minute> ah, possibly floating signals are the problem, esp hpd