12:41 <z3ntu1> Is there some upstream solution for qcom,spmi-adc-tm5 that the hardware can handle less (I guess) trip points than would be needed based on what's connected? Downstream msm-4.19 seems to split this stuff in qcom,adc-tm5 and qcom,adc-tm5-iio where the latter doesn't seem to register any trip points in the hardware, but that way you can register e.g. 6 thermal zones on pm6150l which has channels_available/ADC_TM5_NUM_BTM=4

12:41 <z3ntu1> The downstream commit adding this support is https://git.codelinaro.org/clo/la/kernel/msm-4.14/-/commit/d72e247067a61cf0e76977af7bb83e3ddbfe8e06

12:42 <z3ntu1> Or in other words, on sm7225-fairphone-fp4 with pm6150l there's 6 thermal zones that should be registered but the hardware of pm6150l_adc_tm only seems to support 4, so the driver errors out with "Invalid channel 4" when it tries to register the last two

15:47 <z3ntu1> Actually it seems like I might be able to use generic-adc-thermal for that, qcom,adc-tm5-iio seems to do barely anything more

16:42 <aka_[m]> anyone good with bindings?

16:42 <aka_[m]> i don't understand it at all and mdp5 cries on me

16:48 <aka_[m]> https://pastebin.com/1YQzL2e4

16:51 <konradybcio> 'lut' would have to precede 'tbu'

16:52 <konradybcio> in your case there's no 'lut'

16:55 <aka_[m]> uh

16:55 <aka_[m]> that sounds complicated

16:56 <aka_[m]> but shouldn't this trigger if it had different than oneOf

16:57 <aka_[m]> like idk

16:57 <aka_[m]> allOf:

16:57 <aka_[m]> uh minItems is like you have to have atleast 4 of these?

16:58 <aka_[m]> but why have order static

16:59 <aka_[m]> guess i have no idea how to do it

17:02 <konradybcio> yes order is enforced under items:

17:02 <konradybcio> and minItems: 4 means "at least 4 first entries, or more, with the order preserved"

17:02 <aka_[m]> oh

17:02 <aka_[m]> i expected it to be like any of 4

17:02 <konradybcio> yes it's not obvious at all

17:04 <aka_[m]> i have so many bindings changes

17:04 <aka_[m]> i should start sending this shit, it aint normal to sit atleast half of year

17:04 <aka_[m]> its been like 3years since i touched this

17:18 <konradybcio> bits rot quickly

19:32 <aka_[m]> konradybcio: mind taking a look?... (full message at <https://matrix.org/_matrix/media/v3/download/matrix.org/RhiWVqllCthmEiXiFHqvBUeg>)

19:33 <konradybcio> 4a0e897abf481faf8dd3a5586c3a88cb1142db2f is certainly interesting..

19:33 <konradybcio> fix up the unnecessary change though

19:33 <aka_[m]> msm8936 /t i guess

19:33 <aka_[m]> separate

19:34 <aka_[m]> konradybcio: so 8936 one?

19:34 <aka_[m]> its formating

19:34 <konradybcio> yes

19:34 <aka_[m]> probably goes into separate commit

19:34 <aka_[m]> without fixes cuz it aint fix

19:35 <aka_[m]> i guess i will send tomorrow, im on first shift now in job so can sit few hours after getting home

19:35 <konradybcio> 80b851ca7b00a2541f2071704c1731eb786203d9 property order is totally random, plus there are a couple of double spaces

19:37 <konradybcio> for adreno, check if it works without the tbu clocks

19:37 <aka_[m]> konradybcio: follow sample of qcom_iommu.yaml?

19:37 <aka_[m]> konradybcio: should

19:37 <konradybcio> the examples in dt-bindings are.. unfortunate most of the time

19:37 <konradybcio> which defeats their purpose

19:37 <aka_[m]> tbu clocks are used inside kgsl_iommu when invalidating pagetable

19:38 <aka_[m]> which probably goes somewhere in "per-process-pagetable" flavour

19:38 <aka_[m]> which we ain't supporting

19:38 <aka_[m]> albeit alwayson is required on a510/a506

19:38 <aka_[m]> it ain't working without

19:39 <aka_[m]> 8939 also have these tbus specced in ds dts

19:39 <aka_[m]> and 8917

19:40 <aka_[m]> i guess for iommus either it was coppied from 8953 or from ancient 8976

19:41 <aka_[m]> you might like those tho:

19:41 <aka_[m]> https://github.com/Kiciuk/linux/commits/next/20231027/features/icc

19:41 <aka_[m]> i haven't tested them and Danct12 had few nodes which require -1 on ID but i could try getting this after being done with this bigger series

20:20 <aka_[m]> oot, whats fastest way of testing kernel vs USB device? virtualmachine and passing it?

20:39 <aka_[m]> https://patchwork.kernel.org/project/linux-input/list/?series=799892

20:39 <aka_[m]> oh thats nice

20:39 <aka_[m]> i guess that could be used on phones where you can have multiple possible touch panels