<ukleinek>
xdarklight: you're welcome. But I admit my main motivation was to lure Thierry into continuing the discussion about improving the core :-)
return0e has joined #linux-amlogic
ldevulder has quit [Quit: Leaving]
<ukleinek>
xdarklight: do you care to check the resulting waveforms for old and new code with the parameters I found?
Barada has quit [Read error: Connection reset by peer]
Barada has joined #linux-amlogic
ldevulder has joined #linux-amlogic
Darkmatter66 has joined #linux-amlogic
sputnik_ has quit [Read error: Connection reset by peer]
sputnik_ has joined #linux-amlogic
vagrantc has quit [Quit: leaving]
Darkmatter66 has quit [Ping timeout: 258 seconds]
sputnik_ has quit [Read error: Connection reset by peer]
Darkmatter66 has joined #linux-amlogic
sputnik_ has joined #linux-amlogic
hexdump0815 has joined #linux-amlogic
<hexdump0815>
xdarklight: sorry - some stupid questions regarding the Tested-by - should i keep all the linux lists on cc (amlogic, kernel and kernel-arm) or drop any of them? do i have to be subscribed to the lists to keep them on cc? last time i submitted something for the linux kernel was about 20 years ago i think :)
<hexdump0815>
xdarklight: looks like my m8s is maybe different from yours - it seems to frequency scale on android from 96 to 1608 mhz ... will have to look closer at the dtb for more details maybe ...
afaerber has quit [Quit: Leaving]
return0e has quit [Ping timeout: 272 seconds]
return0e has joined #linux-amlogic
<ukleinek>
hexdump0815: reply to all is usually fine
sputnik__ has joined #linux-amlogic
sputnik_ has quit [Ping timeout: 248 seconds]
afaerber has joined #linux-amlogic
sputnik_ has joined #linux-amlogic
sputnik__ has quit [Ping timeout: 246 seconds]
sputnik__ has joined #linux-amlogic
sputnik_ has quit [Ping timeout: 258 seconds]
<hexdump0815>
ukleinek: thx for the info - do i have to be on all of the lists to reply to them - in times of spam protection i guess yes?
sputnik__ has quit [Remote host closed the connection]
sputnik_ has joined #linux-amlogic
Darkmatter66 has quit [Ping timeout: 258 seconds]
bengal has quit [Ping timeout: 258 seconds]
bengal has joined #linux-amlogic
Elpaulo has quit [Quit: Elpaulo]
sputnik__ has joined #linux-amlogic
sputnik_ has quit [Read error: Connection reset by peer]
sputnik__ has quit [Read error: Connection reset by peer]
sputnik_ has joined #linux-amlogic
sputnik_ has quit [Read error: Connection reset by peer]
sputnik__ has joined #linux-amlogic
Elpaulo has joined #linux-amlogic
vagrantc has joined #linux-amlogic
sputnik__ has quit [Remote host closed the connection]
sputnik__ has joined #linux-amlogic
<ukleinek>
hexdump0815: If not I think you just have to hold out that your mail ends in a moderator's queue
sputnik__ has quit [Ping timeout: 258 seconds]
_whitelogger has joined #linux-amlogic
<hexdump0815>
xdarklight: tested-by is out now - sorry for the extra round, had forgotten to switch to plain text the first time ... next time it will be a bit more fluent :)
tlwoerner_ is now known as tlwoerner
<hexdump0815>
xdarklight: my m8s boots fine with your m8s dtb and everything looks well (with fixed 1.2 ghz) - so in case you want a tested-by for this one as well, please let me know - should be easier now :)
<hexdump0815>
but i guess it should be possible to get frequency scaling working for this one too as it dies that on android too - i tried your mxiii and mxiii-plus dts, but they do not seem to match well and gave me an empty frequency scaling die in /sys
<hexdump0815>
i have dumped the android dtb: https://pastebin.com/raw/AyCGht3W - i'm not very familiar with that, but the cs_voltage_table in meson_vcck_dvfs_driver looks to me like it can do multiple voltages - any idea how to proceed from there?
Barada has quit [Remote host closed the connection]
return0e_ has joined #linux-amlogic
return0e has quit [Ping timeout: 252 seconds]
hexdump0815 has quit [Ping timeout: 256 seconds]
Darkmatter66 has joined #linux-amlogic
warpme_ has joined #linux-amlogic
warpme_ has quit [Quit: warpme_]
warpme_ has joined #linux-amlogic
warpme_ has quit [Quit: warpme_]
<xdarklight>
ukleinek: I don't have fine-grained control over the input clock so I can't use 99990001Hz as input. on GXL I can use 24MHz, 500MHz and 666.67MHz (and the HDMI PLL frequency, but will just make things harder as the clock may change at runtime)
<xdarklight>
but I can test with any of these input clocks if you have a matching period :) (assuming that the 24MHz sampling rate of my "scope" is still enough)
<xdarklight>
narmstrong: thank you for all the Reviewed-by :)
warpme_ has joined #linux-amlogic
<ukleinek>
xdarklight: I thought about calculating the results than measuring.
chewitt has joined #linux-amlogic
<chewitt>
xdarklight: what VIM2 board do you have? .. basic or pro?
<xdarklight>
chewitt: Basic (2GiB RAM)
<chewitt>
same as me then
<xdarklight>
weird
<chewitt>
I need to check my patch-set and make sure @na
<chewitt>
neil didn't tweak that one :)
<chewitt>
he cleaned a couple of others
<ukleinek>
xdarklight: not at all. If you chose both randomly the chance for both of you picking the same board is 50% :-)
<chewitt>
I chose whatever Khadas send me :)
<xdarklight>
ukleinek: there's more differences now because there are different board revisions (v1.1 and v1.4) ;)
<xdarklight>
my "weird" comment was actually that it's strange that it works on one example but not on the other
<chewitt>
I have a 1.4 pro board kicking around somewhere
* ukleinek
is always amazed about the constructive reponse to unconstructive non-sense talking :-)
<xdarklight>
:)
<xdarklight>
chewitt: mine's actually a v1.2
<chewitt>
I wanted a pro one for poking at brcmfmac4359-sdio suppport .. then realised VIM3 has the same chip
<xdarklight>
you already have a VIM3? :o
<chewitt>
one of the early samples
<ukleinek>
chewitt: what do you do that khadas sends you their boards?
<chewitt>
seems to be decent .. currently watching the indy500 replay on it :)
<chewitt>
I'm one of the people irresponsible for LibreELEC
<ukleinek>
chewitt: ah, I see
* ukleinek
considered getting a vim2, too, but my employer suggested an imx8 board instead to play around with kodi
<chewitt>
what kind of playing around with Kodi is of interest?
<chewitt>
I'm also part of Team Kodi :)
<chewitt>
we like people playing around
<ukleinek>
chewitt: well, the initial plan was just to have a nice arm64 platform to play around with for Debian and Kodi (maybe in form of libreelec)
<chewitt>
imx8 will be pushing a few boundaries on gbm/v4l2 support in Kodi
<ukleinek>
and maybe in the end simplify our TV experience
<ukleinek>
chewitt: is that good or bad?
<chewitt>
probably depends on who's perspective you look from :)
<chewitt>
whose
<ukleinek>
chewitt: I have competent colleagues I guess.
<chewitt>
one of my colleagues "lrusak" has been working on imx6 support (which is working-ish) and imx8
<ukleinek>
chewitt: I think I mostly have a language problem I think. "pushing a few boundaries" means there is some work to do, I guess?
<chewitt>
it should/can work, but is not one of the primary targets for development at the moment
<xdarklight>
ukleinek: which edge case did you have in mind with the parent clock of 99990001Hz? since I have my Khadas VIM2 on my desk anyways I can easily measure with an 24MHz input clock assuming I choose a "problematic" period
sputnik_ has joined #linux-amlogic
<xdarklight>
I chose something big (16777215ns) and there no measurable difference between the old and the new code (pre_div is identical, "cnt" is different by 2)
afaerber has quit [Quit: Leaving]
khilman has joined #linux-amlogic
<ukleinek>
xdarklight: it was just the first combination of parameters I found that results in a different pre_div
narmstrong has joined #linux-amlogic
steev_ has joined #linux-amlogic
<xdarklight>
ukleinek: I tried with period = 2730624, old pre_div=1, old cnt=32768, new pre_div=0, new cnt=65534
Darkmatter66 has quit [Ping timeout: 245 seconds]
<xdarklight>
old gives me a measured period of 2732ms, new measured: 2731ms
Darkmatter66 has joined #linux-amlogic
<xdarklight>
the old one is off by 1376ns, the new one off by 376ns
<xdarklight>
I'll add that case to the patch description as I'll have to re-send it anyways. let me know if you want me to test another case :)
<xdarklight>
I made a mistake while measuring the old values (I set the wrong duty cycle...). turns out that I cannot measure a difference between old and new (both give me 2731ms now)