ChanServ changed the topic of #aarch64-laptops to: Linux support for AArch64 Laptops (Asus NovaGo TP370QL - HP Envy x2 - Lenovo Mixx 630 - Lenovo Yoga C630)
alpernebbi has quit [Ping timeout: 480 seconds]
alpernebbi has joined #aarch64-laptops
hexdump0815 has joined #aarch64-laptops
hexdump01 has quit [Ping timeout: 480 seconds]
<steev> bamse: so, the wifi still dies (it still shows an ip address, just that name resolution stops happening and no traffic happens, BUT with that patch applied, i can do exactly what is in the commit message, to unbind and then bind the wifi device and it starts working again, without a reboot, so that's good
iivanov__ has quit []
<steev> bamse: one other thing i've noticed... cpu0 is "stuck" at about 90% usage but... as far as i can tell, nothing is actually using it?
<steev> maybe that's why i always see the little cores at 1.76GHz and none of the others?
jhovold has joined #aarch64-laptops
matthias_bgg has quit [Ping timeout: 480 seconds]
matthias_bgg has joined #aarch64-laptops
<qzed> bamse: in your flex 5g dsdt: why is VREG_S4A separate? is it not controlled by the PMC8180_A?
<bamse> qzed: it's separate because it's so always-on that it's not exposed in the rpmh interface
<bamse> steev: okay, better than before then (re wifi)
<bamse> steev: and i would suspect that your 90% usage is the reason for why you got more of the mouse cursor lagging...wonder what differs between our setups
<bamse> steev: but yeah, if you have something that's hogging cpu0 at 90%, then it's likely going to stay at 1.76ghz...and if that load doesn't migrate i think it's quite reasonable to expect that to be an interrupt-related issue
<qzed> bamse: got it, thanks
<qzed> bamse: also is VPH_PWR flexible? according to stuff I have it seems to be 3.3V on the SPX
<bamse> qzed: yes, it's flexible...and i think it should be more like 3.8V...
<bamse> qzed: you should be able to measure it using the adc in the pmic
<qzed> oh I'll have to try that, could you give me a pointer on how to do that?
<bamse> the vadc_vph_pwr node in the pm660.dtsi seems plausible
<qzed> I'm working of a schematic I got sent which says 12A@3.3V but that looks like it might be development/pre-production stuff
<bamse> it might be that they just picked 3.3V because i think that's a possible voltage
<qzed> possible
<qzed> thanks! I'll try to read the voltage once I have a somewhat working DT again
<bamse> nice
<bamse> i was hoping to get the flex 5g dts out there for 5.19, but got caught up in various other topics...but i think the drivers are in good shape now at least
<bamse> still working on the displayport integration though, so that will have to wait a little bit longer
<qzed> yeah, looking pretty good driver-wise I'd say
<qzed> btw. do you know anything about a third (and fourth) USB interface?
<qzed> I'll have to try and get that working as well at some point
<bamse> yes, i actually got that working (not upstreamable yet) on the sc8280xp this week...
<qzed> but voltage regs first
<qzed> ah nice! thanks, i'll need to check that out
<bamse> we need to figure out how to make the dwc3 driver support more than two phys
<qzed> I see, okay
<bamse> https://github.com/andersson/kernel/commit/97752ef4cc5e00b1ebc96efebb8b689dcb6615b3 <- something slightly more sophisticated than this ;)
<qzed> heh, at some point I also need to figure out how to make the qup/geni serial driver work on more than one interface...
<qzed> got a similar hack for it
<bamse> it breaks down if you enable > 1 geni serial instance?
<qzed> not sure how the situation is at the moment (still broke somehow but not sure if it's the same way as on old v5.13)
<qzed> give me a sec and I'll post a link for the workaround
<bamse> could it be that you don't have aliases { serialN...} defined for all of your instances?
<qzed> was setting "line" to zero because somehow it didn't get that (or I might just have set up the DT wrong, that's also very much possible)
<qzed> I think I might just have set hsuart wrong...
<qzed> ahh, that I missed
<bamse> you need a unique serialN for each uart...or strange things will happen
<qzed> got it, thanks again!
<bamse> ohh, seems that has been improved since i faught that last time...now you'll get line = -ENODEV and get_port_form_line() will return an error
<bamse> i.e. exactly the problem you worked around
<bamse> last time i tried all serial ports that wasn't mentioned got line 0, so all of them was messing with the same data structure
<qzed> ah yeah no I got the ENODEV error
<bamse> lucky you :)
<bamse> add the aliases and you'll be good
<qzed> thanks!
<qzed> all falling back to the same sounds like fun to debug...
<bamse> it resulted in some weird side effect a little bit later
<steev> bamse: the "cursor lag" i only see during wifi usage when e.g. downloading - i thought you see that as well
<qzed> bamse: okay, another question: in the flex 5g and primus dt vdd-l7-l11-supply and vdd-l9-l10-supply are missing. Are they supposed to?
<qzed> *for the pmc8180c
<krzk> steev: thanks for tests!
<steev> krzk: ah, you are on irc :) yes, thank you for that patch, that helps immensely on the flex5g
<steev> one thing to note
<steev> on the c630, unbind->bind does not get wifi working again, and i'm not sure why, but at least it isn't crashing like previously
<steev> crashing when unbind->bind - i've never actually seen it crash wit that message, only saw it on the flex5g (i mentioned it to bjorn a few weeks back when doing some testing of his branch)