11:27 <z3ntu> srinik: Do you (or someone else) have an idea what could be wrong with soundwire<->wcd9385 communication if the soundwire slaves (wcd_tx & wcd_rx) appear ("qcom-soundwire 3230000.soundwire: SWR new slave attached", "soundwire sdw-master-1: SDW Slave Addr: 230217010d00" etc) but then the first sdw_update_no_pm call fails here "SDW_SCP_INTMASK1 write failed:-5" (the error there is "swrm_wait_for_rd_fifo_avail err read underflow")?

11:27 <z3ntu> This is on sc7280.dtsi-based qcm6490, where I needed to do some changes to the dtsi for using e.g. q6afecc clocks, so different setup to ChromeOS firmware there, but I'm half confident my changes are good there based on looking at other dtsi in mainline and checking downstream dtsi

11:28 <z3ntu> I'm not sure yet about qcom,tx-port-mapping & qcom,rx-port-mapping but I tried a few values other devices are using (:D) and no change so not even sure this is relevant currently

12:11 <krzk> z3ntu: some clock or reset could be missing. Also maybe device did not attach?

12:11 <krzk> z3ntu: grep . /sys/bus/soundwire/devices/*/status

12:12 <krzk> like device did not go out of reset (reset-gpio)

12:12 <krzk> z3ntu: another is known problem of accessing WCD device resgiters before attaching

12:13 <krzk> z3ntu: I fixed this for wcd938x, so either you run old code or fix is somehow incomplete (but worked for me)

12:14 <krzk> z3ntu: so just be sure - we are talking here about one of the latest linux-next, so from November, right?

12:53 <luka177[m]> <konradybcio> "luka177: what does it show if..." <- i did add it that resulted in error like:... (full message at <https://matrix.org/_matrix/media/v3/download/matrix.org/chKutkuGksazxcTQvUFzXzCX>)

13:09 <luka177[m]> can i call it after panel prepare?

15:05 <z3ntu> krzk: I'm testing on 6.6. The reset gpio should be okay, when I set reset-gpios on the wcd to a different gpio, the device never appears, so no "SWR new slave attached"

15:05 <z3ntu> I can re-test on linux-next, I wasn't aware there were significant fixes lately

15:07 <krzk> z3ntu: there are several fixes every month :) but v6.6 has my fix for wcd938x

15:09 <krzk> z3ntu: maybe my fix is incomplete for some cases, like runtime suspend. You can add debug statements to wcd938x drivers to print their state (attached/unattached) when the issue happens.

15:10 <krzk> z3ntu: because I still would point this as most probable cause...

15:16 <z3ntu> okay, I'll try your suggestions, thanks!

15:17 <z3ntu> one more thing, is there anywhere anything regarding how to get qcom,tx-port-mapping & qcom,rx-port-mapping for the device based on either schematics or downstream dt/driver? The only thing remotely like that I've found is https://lore.kernel.org/linux-arm-msm/c017342d-711a-f10f-4154-3ff17ec100d5@linaro.org/ but it's not very clear either ^^

15:17 <z3ntu> doesn't help that downstream (msm-5.4/4.19) soundwire is completely different to mainline implementation

15:20 <krzk> z3ntu: eh, I don't know any method other than downstream sources or prioprietary Windows sources. And you have none of these.

15:20 <krzk> z3ntu: unless you have downstream sources?

15:20 <z3ntu> i do have

15:21 <luka177[m]> <luka177[m]> "can i call it after panel..." <- konradybcio: ok did that, after boot i see corrupted DSC blocks for few seconds then black screen

15:21 <z3ntu> downstream sources, not windows of course :D

15:22 <krzk> then downstream audio sources should have them in the files mentioned on above LKML discussion - look for xxx-port-config.h files

15:23 <flto> luka177[m]: the DSI TPG is mostly useless with DSC, DSI does not have HW to DSC compress the test pattern (test pattern is sent to panel as-is, which is garbage for DSC)

15:32 <z3ntu> krzk: kk, will take a look, and if I figure it out write it down somewhere!

15:33 <z3ntu> I'd hope by msm-5.15 (or msm-6.1 if that ever happens?) this stuff is better

15:35 <krzk> z3ntu: 5.15 is the same :/

15:35 <krzk> 6.1 I did not see, but 99% is the same as well

15:35 <krzk> welcome to my world :(

15:36 <luka177[m]> <flto> "luka177: the DSI TPG is mostly..." <- Ouch, thanks didnt knew that

15:37 <aka_[m]> there is msm-6.1 already?

15:37 <z3ntu> "already"

15:37 <aka_[m]> its funny qcom even on 5.15 do import ds code instead of reworking already upstreamed one

17:03 <aka_[m]> srinik: hai, are you there

21:40 <luka177[m]> When i stop X11/modeset for very short time i can see that. arond 2/3 of the left screen half have some other pattern for some short time