10:08 <konradybcio> lumag: sm6350 0x0ae05a00 = mdp_base + 0x4000 + 0xa00 => 0x3000

10:12 <konradybcio> so that is qseed4 after all..

10:17 <aka_[m]> konradybcio: how good to have access to internal docs right

10:17 <aka_[m]> even downstream call it qseed3lite

10:18 <konradybcio> i mean, reading a register requires /dev/mem, not internal docs

10:18 <aka_[m]> but 0x3 being qseed4

10:18 <aka_[m]> thats weird

10:18 <konradybcio> check 4.19 techpack

10:18 <konradybcio> it's defined there

10:18 <konradybcio> 3lite is 0x2004

10:18 <aka_[m]> meh

10:19 <aka_[m]> konradybcio: i still remember downstream versioning of SAW2

10:19 <aka_[m]> 0x3 in register being 2.3 but having same layout like 3

10:20 <aka_[m]> and 2.1 being different

10:20 <aka_[m]> ehhh

10:20 <aka_[m]> well that could work if we assume saw2 is "IP+Major" then 0x3 from register would place it as SAW2.3

10:22 <aka_[m]> still why SAW4 on MSM8953 is 0x4xx1xx00

10:50 <lumag> aka_[m], konradybcio yes. I was very puzzled with these version too, as you can see from the git history.

10:50 <konradybcio> i will check 6375 in a couple minutes too

10:51 <lumag> I think that might end up with 7150/6150/6125 for v3lite

10:52 <lumag> unless 8150/c8180x are v3lite too

10:52 <konradybcio> is there any good reason we hardcode it instead of reading the registers?

10:56 <lumag> No, patches are welcome. I think we should declare QSEED3 (as opposed to QSEED2) and then determine 3/3lite/4 from register read

11:00 <konradybcio> lumag: 6375 reads 0x00003000 as well

11:00 <lumag> konradybcio, thanks!

11:22 <lumag> konradybcio, I double-checked downstream. It has special quirks for qseed3, rev 1.2 (sdm660, maybe msm8998?), but doesn't have special name for that.

11:23 <konradybcio> "fun"..

11:23 <aka_[m]> that sounds so bad

11:23 <aka_[m]> still 660 is running on MDP right

11:24 <lumag> So, I think, we better handle this as a QSEED3 plus version checks

11:24 <lumag> aka_[m], yes. Nobody had enough action points to finish and polish conversion from mdp5 to dpu1. Despite me having the actual board on my desk

11:25 <lumag> The fun thing is that we agreed with robclark some time ago that together with the next conversion we will also introduce command line argument to toggle between mdp5 and dpu1. So it's not just about the catalog entry

11:35 <konradybcio> lumag: well.. i did 2y ago :P

11:35 <konradybcio> i think i should have it on some branch somewhere

11:35 <lumag> Yes, I even saw that once

11:36 <lumag> But I don't remember why nobody pushed it out.

11:36 <lumag> That's why I said 'finish and polish'.

11:36 <konradybcio> i was busy with the 70 patches long 660 series, probably

11:36 <konradybcio> :D

11:38 <lumag> Most likely

17:28 <Leandro[m]> <lumag> "Leandro, I'd say, start really..." <- i hear uart my brain goes bwom

17:28 <Leandro[m]> s/bwom/bwomp/

17:31 <vknecht[m]> don't sweat on uart, if you're able to get usb working without it you're good to go too :-)

18:30 <aka_[m]> lumag: have you tested 660 lately?

18:30 <aka_[m]> This bug with rings probably affects all adreno 5xx right

19:59 <gpiccoli> Folks, being facing an issue here..I have a patch ready but wants to validate first. I tried to build the dt blog with skales-dtbtool, it fails..I've narrowed down to an issue with a new dts missing qcom,msm-id. Adding that, it works. Is it a correct fix?

19:59 <gpiccoli> it's msm8996-based

19:59 <gpiccoli> tnx in advance!

20:02 <konradybcio> does this concern a device that's currently supported upstream? gpiccoli

20:02 <gpiccoli> konradybcio, I think it's supported upstream - oneplus3t <- it's on tree

20:03 <gpiccoli> the only thing is that I'm far from expert on DT, and seeing "deprecated:yes" on the qcom,msm-id makes me a little dubious on adding it...

20:03 <gpiccoli> heh

20:04 <gpiccoli> but with my patch dtbtool works fine!

20:04 <konradybcio> looks like OP3 sets it, but OP3T doesn't

20:04 <konradybcio> so it was probably overlooked

20:04 <gpiccoli> yeah, exactly

20:04 <gpiccoli> cool! so gonna send that, thanks! I'll CC you on that konradybcio , appreciate the help =)