ChanServ changed the topic of #linux-msm to:
marvin24 has joined #linux-msm
marvin24_ has quit [Ping timeout: 480 seconds]
marvin24_ has joined #linux-msm
marvin24 has quit [Ping timeout: 480 seconds]
<djakov> konradybcio: that sounds hard... someone should request access to hundreds of platforms, get approval, find this info etc.
<djakov> i believe that this applies for the platforms you mentioned. and also for 100% of the newer rpmh based platforms.
<djakov> in general, noc and qos registers are retained accross power collapse. they are lost only if you power-off the chip and go in hybernate
<aka_[m]> Or find your way straight to FTP server with confidential data and fish what is needed.
<aka_[m]> However that's not legal-friendly solution
pespin has joined #linux-msm
<krzk> bamse: any schedule on them? or comments?
srinik has quit [Killed (NickServ (Too many failed password attempts.))]
srinik has joined #linux-msm
<aka_[m]> Can anyone tell me from where do venus-X.Y comes in in firmware packages?
<aka_[m]> its for sure not VENUS_WRAPPER_HW_VERSION
<aka_[m]> konradybcio: is slight exceed on column size very bad?... (full message at <https://matrix.org/_matrix/media/v3/download/matrix.org/DwBTNtBtrcbVkVOwZLyUJDXr>)
<aka_[m]> ⚰️
<mal> you could add the last "tx-rings-empty" below "tx-enable" a bit like is done for interrupts
pespin has quit [Remote host closed the connection]
<aka_[m]> got same idea from Tooniis
<aka_[m]> tho i still feel weird with gedit showing me lines are bit off but both are starting from column 73
<aka_[m]> tabbing to 8 in settings
<aka_[m]> konradybcio: would you mind to give any insight on wakegic/mpm?
<aka_[m]> i see pinctrl does have some mappings
<aka_[m]> but also node on sm6375 have some mappings too
<aka_[m]> this is probably like tlmm->mpm pin mapping atleast if there is X Y where X is not 0xff
<aka_[m]> konradybcio: where did you get mappings for 6375mpm?
<aka_[m]> i found these there https://github.com/xiaomi-sm6375-devs/android_kernel_xiaomi_sm6375/blob/lineage-20/drivers/irqchip/qcom-mpm-holi.c but last two are different from what you added
<aka_[m]> ok i get where ones from dt come but what about TLMM mpm/pdc no idea from where i can get these
<aka_[m]> ok im blind
<aka_[m]> lol
<aka_[m]> qcom,gpio-map inside same node of msm8953-pm.dtsi
alexeymin has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]
alexeymin has joined #linux-msm
<aka_[m]> bamse: your scripts went nuts or that huge pull was intended?
<steev> i hope it was intended :D
<konradybcio> aka_: QC_IMAGE_VERSION_STRING
<konradybcio> aka_: the "mpm gpios" are lines that are wired up to the RPM subsystem, such that the vast majority of the SoC can be powered off, but e.g. usb or tsens can wake it up from total power collapse
<aka_[m]> So device sleep and I connext usb, then interrupt is triggered and rpm wakes device up?
<konradybcio> yes
<konradybcio> it wakes up the AP and then your device works like usual
<aka_[m]> And now its all good because we don't really deep sleep right?
<konradybcio> well it's there to allow "deep sleep"
<konradybcio> qualcomm socs can gate all the clocks and all the power lines coming to the AP and peripherals
<konradybcio> RPM is always on as long as the device is powered on
<aka_[m]> konradybcio: and what about other question? The pins in pinctrl comes from ds binding right?
<konradybcio> and powered on == the battery is connected and it hasn't been powered off for good
<konradybcio> yes from there
<aka_[m]> Ita confusing
<konradybcio> they shuffled it between dt and .c a few times
<aka_[m]> PDC is some new kind of mpm?
<konradybcio> yes PDC is essentially MPM for this purpose
<aka_[m]> Or rpmh only
<konradybcio> rpmh only yes
<konradybcio> they remade the entire (low) power setup with the introduction of rpmh
<aka_[m]> konradybcio: as you are working on multiple new socs, any idea if older misses something other than perf which makes them run uis like phosh/gnome with frame rate looking like 30 rather than 60?
<konradybcio> define "new"
<aka_[m]> Anything with "dpu"
<konradybcio> anything with DPU>=5 needs changes to the cmd mode handling which Marijn is working on
<konradybcio> video mode should be fine
<aka_[m]> Well pleb devices like Mido or leeco so vid mode
<aka_[m]> Maybe just a5xx is sad and that's all
<konradybcio> and what socs are affected?
<aka_[m]> 8953,8976 aren't really snappy with cpufreq in place
<aka_[m]> Maybe interconnects
<Marijn[m]> kmscube?
<aka_[m]> No just uis
<aka_[m]> For kmscube we aren't supposed to launch it inside de?
<aka_[m]> Or can we for testing
<konradybcio> i don't think it will even let you
<aka_[m]> So what we have to do?
<konradybcio> kill de, run kmscube
<aka_[m]> Kill de And login manager.
<aka_[m]> Ah right
<Marijn[m]> ^
<Marijn[m]> I want you to test if it's mdp5 or the GPU
<Marijn[m]> IIRC mdp5 was okay, it's the GPU
<Marijn[m]> (kmscube uses the GPU but only very slightly, if it doesn't reach 60fps defer to modetest)
Daaanct12 has joined #linux-msm
Daanct12 has quit [Ping timeout: 480 seconds]