09:09 <SSJ_GZ> Hi all! How well does C630 Suspend/ Resume work nowadays? Last I heard was from a talk in March 2021, when Suspend used approx the same power as non-Suspend :/

17:55 <steev> SSJ_GZ: there are still some patches outstanding from abelvesa (i believe it is) but the power usage should go down a bit once all that stuff is in

17:55 <SSJ_GZ> steev: Thanks! Is there anywhere I can follow this in detail?

17:56 <steev> i don't know the easiest way to do so, i just look at what is submitted to the linux-arm-msm patchwork (and i turn off all the filters) on the website and look for things that might be interesting

17:57 <steev> though when i go to test them, i apply them with b4

17:57 <steev> used to use patchwork's cli, but it left a bit to be desired for my purposes/needs/skill level

17:58 <SSJ_GZ> Many thanks - I can also search the mainline linux git logs for their email address :)

18:09 <steev> you can, but he's also in here :)

18:09 <steev> not super active, but he does reply

18:11 <SSJ_GZ> Oh, so I see :)

18:12 <SSJ_GZ> Actually, I have a general question for all of you: I don't have hard data on this, but it *feels* like most ARM devices get working power management *after* they get GPU support.

18:13 <SSJ_GZ> As a laymen, I thought that GPU would be much harder than power management, but it seems like I'm missing something.

18:13 <SSJ_GZ> Is that observation correct, and if so, why should it be so?

18:14 <SSJ_GZ> Lower priority? More "fiddly"/ requires more quirks? Power management subsystem less advanced than DRM+Mesa?

18:15 <steev> i'm just a rando user who is dangerous enough to know how to apply patches (mostly), so i can't say there, but at a guess, more fiddly

18:17 <steev> bamse: https://paste.debian.net/1254124/ line 19 (this is on the Thinkpad X13s) - i guess maybe that's where our mac address and stuff would be (haven't yet tested running it on the c630 to see what it spits out)

18:19 <robclark> so in a devicetree world, it is pretty unlikely that something like a GPU would work at all without power mgmt.. clk and power domain ctrl are one of the very first steps in bringing up a new SoC. But that does not mean that power mgmt is fully debugged (ie. tracking down things that leave clocks on blocking suspend, and things like that) ;-)

18:23 <steev> woo, the c630's ucm configs are merged in :D

18:23 <steev> https://github.com/alsa-project/alsa-ucm-conf/pull/88

18:24 <SSJ_GZ> robclark: Thanks for the insights :)

18:25 <SSJ_GZ> Sounds like getting good, low-power Suspend is similar to playing whack-a-mole :)

18:27 <steev> kind of, yeah

18:30 <clover[m]> mali gpu on pinebook pro was reverse engineered very quickly compared to trusted firmware bug fix for suspend-to-ram, and its still buggy

18:36 <SSJ_GZ> The Pinebook Pro was one of the devices I had in mind when I asked :)

18:44 <bamse> SSJ_GZ: in the qcom case robclark had the functional dependencies for doing gpu work quite early...while it's not until recently that we have enough functionality in place for really working on power management

18:46 <SSJ_GZ> bamse: Thanks! Will the PM functionality you have now help with future qcom socs, do you think?

18:50 <bamse> SSJ_GZ: most of the problem are completely generic and applies to all qcom platforms (and probably many other soc vendors' platforms as well)

18:51 <bamse> SSJ_GZ: some are specific to a particular platform, but the plumbing necessary will be reusable

18:51 <bamse> plumbing and understanding

18:51 <SSJ_GZ> Sounds very promising :)

20:00 <clover[m]> New toy for x13s came in 😁

20:04 <steev> nice

20:08 <szclsya[m]> spiritual successor of the w701ds!

20:11 <bamse> cool

20:12 <clover[m]> it didnt seem to work :(

20:12 <bamse> with the x13s?

20:13 <clover[m]> um steev do we know if we have DP 1.2 Alt mode?

20:13 <bamse> did you try flipping the usb-c cable around?

20:13 <bamse> clover[m]: we do

20:13 <clover[m]> ah the flip

20:13 <clover[m]> gets me every time

20:14 <bamse> there's a patch for that...

20:14 <steev> there is?

20:14 <bamse> steev: go fish ;)

20:15 <steev> oh, is it part of v2?

20:15 <bamse> there are 3 different implementations for that, don't think anyone of them are on the lists

20:15 <steev> oooh

20:15 <steev> oh

20:15 <bamse> the most complete was in my 8350 branch, but it did flip properly...but muxing 2+2 -> 4 -> 2+2 lane usb/dp would loose usb

20:16 <steev> got ya

20:17 <steev> i'll try to find some time this weekend

20:17 <steev> i've been poking at the radxa zero and the waveshare 7.9" display again lately, and got it kinda sorta working, just with a nice pink background instead of black

20:17 <steev> works fine when plugged into the thinkpad :D

20:18 <clover[m]> this screen is adorable

20:18 <steev> so the issue is somewhere in the meson driver, i just.. haven't figured out where to look yet

20:18 <steev> https://usercontent.irccloud-cdn.com/file/xj9QaZ2Y/1663316182.JPG

20:19 <bamse> steev: odd

20:19 <steev> clover[m]: oh that is nice

20:19 <bamse> clover[m]: nice!

20:19 <steev> the person who put me on to figuring out that the driver was marking modes where h < 640 or v < 480 were bad, thinks it might be RGB vs YCBrb or whatever

20:20 <steev> it's 400x1280

20:21 <clover[m]> steev: you say that every weekend ;)

20:21 <steev> i know :(

20:21 <steev> in my defense, people ping me and ask me to test stuff for them, and i wanna get that done so they can post things to the list and get them out there for everyone to enjoy

20:29 <steev> or, like currently, i'm trying to mash up the c630, flex5g and thinkpad into one thing

20:30 <clover[m]> one thing to rule them all

20:34 <steev> pretty much, yeah :D

21:26 <steev> bamse: so the v2 of the dp/edp patches, that's the ones that used to rely on MDP_TOP ?

21:26 <steev> oh wait, no i'm an idiot, that's dpu

21:26 <bamse> steev: no, that's the dp-side of things

21:26 <bamse> you're not an idiot, but yes

21:27 <steev> found the ones :)

21:27 <bamse> should bump those as well...

21:31 <clover[m]> <szclsya[m]> "spiritual successor of the..." <- That thing looks crazy

21:34 <szclsya[m]> impractical but impressive nonetheless eh

21:37 <clover[m]> Kind of like an arm compute device :P

21:39 <clover[m]> <bamse> "steev: no, that's the dp-side of..." <- Does that happen to address the fail on suspend while connected to external monitor?