<scelestic>
i cannot get thermal working on my pi plus 2
florianH has joined #linux-sunxi
matthias_bgg_ has quit [Quit: Leaving]
matthias_bgg has joined #linux-sunxi
<scelestic>
since it has the same SoC as an orange pi pc shouldnt those dts settings apply for the plus 2 as well?
yann has joined #linux-sunxi
<ErwinH>
What kernel do you use?
<jonkerj>
ErwinH: are you using megi's 4.9 branch?
<ErwinH>
Jup
<jonkerj>
and are you able to test if NFS mounts work?
<jonkerj>
I am using megi's 4.8 branch, since 4.9 cause oopses when mounting NFS :-(
<scelestic>
ErwinH: using megi's 4.9 branch as wel
<ErwinH>
Think I left NFS out of my images, but I'll recompile in a bit, currently working on a OPi Zero image.
<jonkerj>
I am not really sure if I am undervolting (which makes things wonky) or hitting a bug
<jonkerj>
but the former is of course easy to check
camh has quit [Ping timeout: 246 seconds]
<ErwinH>
jonkerj: after installing nfs-common nfs is running perfectly :)
<ErwinH>
scelestic: what part of the thermal part isn't working? can you read the temp of the cpu?
camh has joined #linux-sunxi
<scelestic>
ErwinH: no, it errors on boot about missing something
<scelestic>
ErwinH: i just compared the dts files (orange pi pc and plus) and it seems it's missing some parts
<scelestic>
when comparing the fex files it seems everything is at the same values the only thing i was not able to find yet is what i2c address the regulator should be
muvlon has quit [Ping timeout: 260 seconds]
<scelestic>
my assumption would be at the same adderss, 0x65
<scelestic>
so i prepared a new dts but i dont have the machine near me so i rather not try it yet :)
ganbold has joined #linux-sunxi
paulk-collins has quit [Quit: Leaving]
<ErwinH>
I'd say, step by step, start with the ths patches and see if you can get a thermal readout. You don't need the i2c address to get the temperature of the cpu.
<jonkerj>
ErwinH: ok, let me check again, since megi had updated his repo today
<scelestic>
ErwinH: i have those applied but while booting i see "thermal thermal_zone0: failed to read out thermal zone (-16)"
* ErwinH
too :)
<jonkerj>
hmm, NFS works for me too
<ErwinH>
But what does : cat /sys/class/thermal/thermal_zone0/temp say?
<jonkerj>
I'd swear that it oopsed a week ago
<ErwinH>
Or, what does armbianmonitor -m say?
<scelestic>
ErwinH: oh wow, i looked around for temperatures but obviously i missed them , that one reports 13680
<ErwinH>
13 degrees?
<scelestic>
that's what it said :)
muvlon has joined #linux-sunxi
<scelestic>
environment temperature should be around 17 so that value is too low
<ErwinH>
The -16 error code is Device or resource busy.
<ErwinH>
And if you stress it a little?
<scelestic>
oh right, 4 cores burning and it's over 21 now
<scelestic>
23737
<scelestic>
and quickly back to 16 after killing burncpu
fkluknav has joined #linux-sunxi
<ErwinH>
Next thing is the thumb check :)
<scelestic>
i forgot to bring my pi to work today :(
<scelestic>
but i think i have to testing to do, the dts of orange pi plus seems rather empty and if most things are in the same place (as the fex files seem to confirm) all those things should be rather easy to add to the plus as well?
muvlon has quit [Ping timeout: 246 seconds]
<ErwinH>
That's correct. Megi's branch only "supports" the pc and one, the rest is up to the users.
<jonkerj>
I've got three patches to make megi's branch working on orange pi plus
<jonkerj>
I will mail him right away, maybe he wants to include them
muvlon has joined #linux-sunxi
<scelestic>
cool
<scelestic>
jonkerj: mind putting them on pastebin as well? i have some spare time at work :)
<scelestic>
oh yeah that seems to be working as well
<scelestic>
120000 12mhz?
<ErwinH>
120mhz
<scelestic>
oh 120
<scelestic>
cool
<ErwinH>
Or just use armbianmonitor -m
<scelestic>
yeah i was looking for that but i'm running stock debian for the moment
<KotCzarny>
armbian is almost stock debian
<ErwinH>
You could recompile using very low trip points :) That way you can test if throttling works as well :)
<scelestic>
good point but i'm kinda curious why the readout is this low
<KotCzarny>
would be nice to be able to edit trip points from userland
<ErwinH>
Nah, kind of useless...
<KotCzarny>
one kernel for all boards
<KotCzarny>
:)
<jonkerj>
trip points are described in DT, so you can have one kernel for all boards
* jonkerj
uses 1 kernel for several H3-based boards, with different DT's
tkaiser has joined #linux-sunxi
muvlon has joined #linux-sunxi
<tkaiser>
KotCzarny: You can simply use dtc tool to adjust individual trip points.
<KotCzarny>
i was thinking about having it autodetected on boot (in linux, not uboot)
<KotCzarny>
frigging arm complicating things more than necessary
<tkaiser>
For anyone playing around with optimizing these settings (they need some attention/love at least on the boards that switch only between 1.1V/1.3V): I added support to display VDD_CPUX voltage to RPi-Monitor some months ago.
<tkaiser>
But I fear that will only work without hassles within Armbian. There it's simply an 'armbianmonitor -r', wait up to 5 minutes and then visit the URL that will be displayed (rpimonitord will then run on port 8888)
<tkaiser>
installing a helper script that parses DT settings since on H3 boards voltage can not be read out)
tkaiser has quit [Client Quit]
Worf has quit [Quit: Konversation terminated!]
yann-kaelig has joined #linux-sunxi
Da_Coynul has joined #linux-sunxi
<jonkerj>
I'm not very fond of the way 'armbianmonitor' messes around with my system
<jonkerj>
it contains a base64 blob which is extracted on my system
<jonkerj>
'jay'
<dgp>
jay is European for yay?
<jonkerj>
something like that (:
<KotCzarny>
naaah :P
Pepe has joined #linux-sunxi
paulk-minnie has quit [Quit: Leaving]
Da_Coynul has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<scelestic>
jonkerj: that's the reason i didnt want to install it myself
<scelestic>
probably a dutch thing?
<scelestic>
(the jay part)
tkaiser has joined #linux-sunxi
Da_Coynul has joined #linux-sunxi
<tkaiser>
jonkerj: Yes, the base64 stuff is the more interesting part (and adding Armbian repo is 'necessary' since there we have a armhf rpimonitor .deb lying around since original source isn't available any more)
<tkaiser>
jonkerj: You can pipe the base64 stuff into a .tgz and have a look what's inside. It is on my TODO to improve this though (since ages ;) )
<jonkerj>
so armbian ships rpimonitor and a tool that "fixes" rpimonitor?
<jonkerj>
why not ship two versions of rpimonitor or only ship the fixed one?
bmeneg has quit [Ping timeout: 260 seconds]
<jonkerj>
ahwell, it's not very sunxi-kernel related, so let's not go into details too much here
<tkaiser>
We installed it from Xavier Berger's repo but that's not available any more. And the base64 stuff contains basically two sets of templates and a helper daemon to either parse fex or DT settings on sun8i
<tkaiser>
To provide VDD_CPUX monitoring. Without this I find it pretty hard to develop good settings.
<tkaiser>
IMO we need a few different DT settings for different boards: All the larger Oranges can share one, OPi One/Lite and most probably Zero another (the NanoPis might fit here too). And then there's overheating Banana Pi M2+ that maybe needs own settings.
<tkaiser>
But since we can now rely on schedutil cpufreq governor one set of cpufreq settings for almost all boards might be sufficient. But the problem with different thermal/dvfs settings remains
<tkaiser>
And currently megi's settings show a pretty low performance compared to legacy kernel as soon as throttling starts.
<scelestic>
if i calculate the temperature according the H3 datasheet i end up around 182.somethign, assuming that means 18.2C, sounds like a better value (my home is at about 16/17C right now)
<tkaiser>
On those boards that switch between 1.1V and 1.3V only
<tkaiser>
scelestic: There's something wrong. But your thumb on H3 and confirm ;)
<tkaiser>
s/But/Put/
<scelestic>
it felt fairly cool, dont think 18C but also not 10C
<scelestic>
last i used my IR thingy it measured around 30C but that was running 3.4 sunxi
raknaz has joined #linux-sunxi
<scelestic>
currently reading 10894 from /sys/class/thermal/thermal_zone0/temp
<tkaiser>
scelestic: Below ambient temperature?
<scelestic>
yes
<tkaiser>
scelestic: Then it's wrong anyway.
<scelestic>
it does react to the cpu warming up
raknaz has quit [Client Quit]
<scelestic>
seen it around 23000 with 4 cores loaded
<dgp>
scelestic: Mine is like that too. Temp seems off and the curve doesn't seem to be right. Like it starts to go up when it is getting hot but nowhere as far as I think it should
<KotCzarny>
measure with ir and fix the formula?
<KotCzarny>
then let someone else test it too
<dgp>
Waiting for the ir thermometer to come from banggood :D
<KotCzarny>
borrow from cute neighbour? :P
<dgp>
all my neighbours are country bumpkins I don't think they'll have anything like that
tkaiser has quit [Ping timeout: 265 seconds]
<KotCzarny>
what's a bumpkin?
<dgp>
simple people
<dgp>
most of them don't have computers or smart phones
<KotCzarny>
they get sick, and nowadays most thermometers are those ir things
<scelestic>
i can do some measurements this weekend, for now i'm digging in the 3.4 sources, there seems to be a few calculations that can be done but i have not found anything that seems 'right'
<dgp>
KotCzarny: we still use the contact ones you stick under your arm
* dgp
gets given one each time a kid happens...
alexxy has quit [Quit: No Ping reply in 180 seconds.]
alexxy has joined #linux-sunxi
muvlon has quit [Quit: Leaving]
dizz74 has quit [Quit: Page closed]
scream has quit [Remote host closed the connection]
* scelestic
is trying to figure out how to get the raw value somewhere in /sys
<jonkerj>
scelestic: my patches are included in megi's branch as of now
afaerber has quit [Quit: Ex-Chat]
avph has quit [Ping timeout: 260 seconds]
<scelestic>
jonkerj: thanks!
silviop has joined #linux-sunxi
avph has joined #linux-sunxi
lkcl has quit [Read error: Connection reset by peer]
Da_Coynul has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
afaerber has joined #linux-sunxi
<jonkerj>
but I agree with KotCzarny that userspace/runtime altering of operating points has its uses
<jonkerj>
ie, tuning the voltages without going through many reboots
<scelestic>
on that same regard i noticed that dmesg mentions "Calibrating delay loop (skipped), value calculated using timer frequency.. 48.00 BogoMIPS (lpj=240000)" lpj was 10 times higher as was the bogomips before i switched to mainline uboot/kernel
<scelestic>
3.4 dmesg "[ 0.000821] Calibrating delay loop (skipped), value calculated using timer frequency.. 4800.00 BogoMIPS (lpj=24000000)"
<ErwinH>
tkwiser: My SoC temperatures seem quite accurate... 37C while my IR meter says 42C, but when I increase the load and the temperature goes up, the difference becomes smaller.
<ErwinH>
On the One and the Zero.
<miasma>
tkaiser: thanks for the comments. i'll update the power article later this weekend
<tkaiser>
ErwinH: Those numbers look reasonable, but below ambient temperature is simply not possible. BTW: It gets a lot easier when running continuous monitoring since then stuff like the 'thumb test' make things immediately visible :)
lkcl has joined #linux-sunxi
<tkaiser>
miasma: Thank you :) I already linked to the new page :)
<miasma>
i noticed :)
<ErwinH>
I noticed the temperature drop, once I put my finger on the chip and apply some force.
<ErwinH>
But at least throttling is working again :)
<longsleep>
That pinebook think is actually interesting, assuming it has a reasonable display, keyboard and battery - anyone found some detail tech specs?
<ErwinH>
Small thing btw: I didn't notice a temperature difference in the readout before the last commit and after, to be more precise, when applying only the 2 commits for the THS, the temperature was the same as well.
<scelestic>
i'd be interesting in that pinebook as well but it has to have mainline support though
<jelle>
it won't have that from the launch
<scelestic>
yeah i read that as well
<scelestic>
lets hope everything will eventually get there
<dgp>
near mainline won't happen until the kernel patches for android $newest - 3 require it
tkaiser has quit [Ping timeout: 250 seconds]
JohnDoe_71Rus has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org]
tkaiser has joined #linux-sunxi
lamer14800799490 has joined #linux-sunxi
lamer14800799490 has quit [Client Quit]
<silviop>
PB06 is < 2 6 ...> in device tree syntax or 1 6 ?
<mripard>
1
tkaiser has quit [Ping timeout: 248 seconds]
<dgp>
I think it's bank number and then pin, with A being bank 0 .. you can check the names in the pinctrl nodes of debugfs though to make sure
tkaiser has joined #linux-sunxi
<tkaiser>
jelle: What's missing?
<jelle>
there is some information on the wiki I think
<tkaiser>
jelle: For an A64 laptop HDMI and PMIC is missing but what else?
<jelle>
tkaiser: huh?
<tkaiser>
jelle: I thought you were talking about Pinebook and mainline kernel?
<jelle>
is there mainline boot with the pine now?
<silviop>
thank
<tkaiser>
jelle: Sure, since a long time. I think mainline booted prior to BSP kernel there ;)
bmeneg has joined #linux-sunxi
<silviop>
there is a mir3da 3axis sensor mainline kernel driver ?
Ntemis has joined #linux-sunxi
<jemk>
tb
<jelle>
silviop: look into the iio tree I'd say
chomwitt1 has joined #linux-sunxi
DullTube has quit [Quit: Leaving]
<longsleep>
tkaiser: A64 mainline u-boot still requires to be built with the dram blob no?
<luis_>
Guys, just wanted to drop a line to thank you for the superb work at linux-sunxi.org. I have to deal with Allwinner tablets at my job these weeks, and your site has been an invaluable resource
<tkaiser>
miasma: Connect it and do a dmesg | egrep -i "uas|usb"
<tkaiser>
miasma: Unfortunately there are a few UAS capable ASMedia chips around that are blacklisted since ASMedia re-used device-ids for different chips
<longsleep>
tkaiser: whats the problem with ZFS? the dkms module should build just fine on xenial
<tkaiser>
longsleep: Yeah, but that's failing in my setup (Armbian, also Xenial)
<miasma>
tkaiser: so if uas loads, it's enabled?
<miasma>
the module is there
<longsleep>
tkaiser: for your 4.9 performance test, did you run Pine64 at 816MHz?
<tkaiser>
miasma: Yes but of course you need support for it in kernel or as a module. Whether it's used or not depends on the module, see footnotes in USB/UAS wiki page
<apritzel>
tkaiser is a dirty overclocker, I think he used 864 MHz
<apritzel>
longsleep: ^^^
<longsleep>
tkaiser: mhm maybe i broke something in the latest Kernels, havent rebuilt/updated the zfs cluster since a while
<tkaiser>
longsleep: No, am an overclocker: 864 MHz
<apritzel>
I can give the THS based emergency shutdown in ATF a shot (turn off the board if temp > 110C)
<longsleep>
well, thats still way below the max speed - so it might even be faster once it learns to not fry itself
<miasma>
tkaiser: i read the page but thought that there might be some flag set somewhere. it definitely loads the module, but i wasn't sure if it's in use
<apritzel>
this should give people more freedom to boost frequencies
<tkaiser>
miasma: You really need to connect the enclosure to be sure since (check dmesg afterwards).
<KotCzarny>
miasma: lsusb -t, it will show which driver is in use (massstorage or uasp)
<KotCzarny>
but usually you can just check the chip model
<KotCzarny>
there are only few that do
<KotCzarny>
notably ns1066 doesnt, ns1068 does. similarly jmp567 and 578 do
<miasma>
ok, thanks. so Class=Mass Storage, Driver=usb-storage, 480M
<KotCzarny>
as for asmedia chips, they are buggy and firmware upgradeable
<miasma>
too bad the aliexpress sellers don't list the usb id or vendor
<KotCzarny>
they do sometimes
<KotCzarny>
search for chip model
<miasma>
but it might increate the price :)
<KotCzarny>
anyway, i bought me jmp578 based one and waiting for arrival, (under 5usd)
<KotCzarny>
(no one tested it here, yet, so its a bit of experiment)
<KotCzarny>
another thing to watch for is trim support, not all do
tkaiser has quit [Ping timeout: 250 seconds]
tkaiser has joined #linux-sunxi
apritzel has left #linux-sunxi [#linux-sunxi]
Mr__Anderson has joined #linux-sunxi
Ntemis has quit [Remote host closed the connection]
<igraltist>
hi
<tuxillo>
hi
<igraltist>
on my cubietruck i try to add ext4 fs encryption
<igraltist>
but when i do e4encrypt set_policy xxx my_path i get permission denied
<igraltist>
i log file i see: sun4i-ss 1c15000.crypto-engine: ERROR: Empty IV
IgorPec has joined #linux-sunxi
silviop has quit [Ping timeout: 265 seconds]
<KotCzarny>
montjoie: ^
IgorPec has quit [Ping timeout: 246 seconds]
<igraltist>
yes :D
|Jeroen| has quit [Quit: dada]
<montjoie>
someone reported some month ago, cannot reproduce it for the moment
<montjoie>
igraltist: which kernel ?
<montjoie>
igraltist: could you share full command list to error
jernej has joined #linux-sunxi
lennyraposo has quit [Remote host closed the connection]
Andy-D has joined #linux-sunxi
diego_r has quit [Quit: Konversation terminated!]
lennyraposo has joined #linux-sunxi
Leepty has quit [Remote host closed the connection]
<igraltist>
montjoie: Linux jaschtschik 4.4.34-rsbac-2+ #2 SMP PREEMPT Thu Nov 24 04:21:51 CET 2016 armv7l Allwinner sun7i (A20) Family GNU/Linux
<igraltist>
montjoie: and just e4encrypt set_policy
<tuxillo>
hmm orange pi PC H3 for 16eur
<tuxillo>
it has a H3 SoC
<tuxillo>
ah, 32bit
<montjoie>
igraltist: do you have tried with 4.9 ?
IgorPec has joined #linux-sunxi
Andy-D has quit [Ping timeout: 244 seconds]
tkaiser has quit [Ping timeout: 258 seconds]
<igraltist>
montjoie: i stuck on lt
deskwizard has quit [Read error: Connection reset by peer]
<igraltist>
montjoie: would be enough just replace the drivers/crypto/sunxi-ss with 4.9 files