Turl changed the topic of #linux-sunxi to: Allwinner/sunxi /development discussion - did you try looking at our wiki? https://linux-sunxi.org - Don't ask to ask. Just ask and wait! - https://github.com/linux-sunxi/ - Logs at http://irclog.whitequark.org/linux-sunxi
cptG_ has quit [Ping timeout: 250 seconds]
arossdotme has joined #linux-sunxi
dr1337 has joined #linux-sunxi
Guest76699 is now known as xenoxaos
nikre has quit [Quit: Leaving]
Colani1210 has joined #linux-sunxi
Colani1200 has quit [Ping timeout: 260 seconds]
terra854 has joined #linux-sunxi
chomwitt has quit [Ping timeout: 244 seconds]
apritzel has quit [Ping timeout: 250 seconds]
chong has joined #linux-sunxi
kaspter has joined #linux-sunxi
ninolein has quit [Ping timeout: 260 seconds]
ninolein has joined #linux-sunxi
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
cnxsoft has joined #linux-sunxi
station has joined #linux-sunxi
snblitz has joined #linux-sunxi
snblitz2 has joined #linux-sunxi
<snblitz2> I am running 4.9.0-rc1-44721-g144926d how does one enable the otg usb port in host mode?
<snblitz2> on a orangepipc
bugzc has joined #linux-sunxi
fvogt_vps has quit [Remote host closed the connection]
snblitz2 has quit [Ping timeout: 265 seconds]
snblitz has quit [Ping timeout: 260 seconds]
robogoat has quit [Ping timeout: 256 seconds]
akaizen has joined #linux-sunxi
Axl_z has joined #linux-sunxi
pg12 has quit [Ping timeout: 260 seconds]
pg12 has joined #linux-sunxi
dr1337_ has joined #linux-sunxi
yann-kaelig has quit [Quit: Leaving]
cnxsoft has quit [Ping timeout: 265 seconds]
cnxsoft has joined #linux-sunxi
IgorPec has joined #linux-sunxi
Putti has quit [Quit: Leaving]
cnxsoft1 has joined #linux-sunxi
cnxsoft has quit [Ping timeout: 265 seconds]
cnxsoft1 is now known as cnxsoft
snblitz2 has joined #linux-sunxi
cptG has quit [Ping timeout: 265 seconds]
cptG has joined #linux-sunxi
TheSeven has quit [Ping timeout: 260 seconds]
[7] has joined #linux-sunxi
f0xx has joined #linux-sunxi
kaspter has quit [Remote host closed the connection]
kaspter has joined #linux-sunxi
kaspter has quit [Read error: Connection reset by peer]
jernej has joined #linux-sunxi
kaspter has joined #linux-sunxi
kaspter has quit [Remote host closed the connection]
kaspter has joined #linux-sunxi
JohnDoe_71Rus has joined #linux-sunxi
IgorPec has quit [Ping timeout: 260 seconds]
rkkang has quit [Quit: Leaving]
jernej has quit [Ping timeout: 256 seconds]
Gerwin_J has joined #linux-sunxi
kaspter has quit [Ping timeout: 245 seconds]
snblitz2 has quit [Ping timeout: 256 seconds]
reinforce has joined #linux-sunxi
cnxsoft1 has joined #linux-sunxi
cnxsoft has quit [Ping timeout: 256 seconds]
cnxsoft1 is now known as cnxsoft
kaspter has joined #linux-sunxi
sunxi_fan1 has quit [Ping timeout: 250 seconds]
mosajjal has joined #linux-sunxi
chong has quit [Read error: Connection reset by peer]
mosajjal has quit [Ping timeout: 256 seconds]
mosajjal has joined #linux-sunxi
mosajjal has quit [Ping timeout: 256 seconds]
mosajjal has joined #linux-sunxi
bugzc has quit [Ping timeout: 276 seconds]
mosajjal has quit [Read error: Connection timed out]
mosajjal has joined #linux-sunxi
leviathanch has joined #linux-sunxi
massi has joined #linux-sunxi
Gerwin_J has quit [Quit: Gerwin_J]
fvogt_vps has joined #linux-sunxi
mosajjal has quit [Ping timeout: 260 seconds]
apritzel has joined #linux-sunxi
apritzel has quit [Ping timeout: 244 seconds]
cnxsoft has quit [Remote host closed the connection]
cnxsoft has joined #linux-sunxi
apritzel has joined #linux-sunxi
florianH has joined #linux-sunxi
premoboss has joined #linux-sunxi
apritzel has quit [Ping timeout: 245 seconds]
mosajjal has joined #linux-sunxi
matthias_bgg has joined #linux-sunxi
gianMOD has joined #linux-sunxi
snblitz2 has joined #linux-sunxi
speakman has quit [Ping timeout: 260 seconds]
speakman has joined #linux-sunxi
<oliv3r> a small update; 456 MHz with ambient temperature of around 22 C; all boards still running
<oliv3r> 480 they failed the long test, 504 MHz they failed even a 30 minute test
leviathanch has quit [Remote host closed the connection]
<oliv3r> i don't recall who asked, but temperature wise, we do try to minimize this already. We have a bottom plate made of thin aluminium which can transfer heat and use a thermal transfer pad. on top we have a heatsink installed. Airflow is not the greatest, but it helps a little bit.
whaf1 has joined #linux-sunxi
IgorPec has joined #linux-sunxi
whaf has quit [Ping timeout: 256 seconds]
<KotCzarny> yes, but the room in which the whole thing will be, will get hotter
<speakman> Anyone successfully been using dual channel LVDS on A20?
whaf has joined #linux-sunxi
<KotCzarny> and ambient temp is very influential
kaspter has quit [Ping timeout: 256 seconds]
<oliv3r> KotCzarny: our official operating temperature is upto a maximum of 35C ambient
whaf1 has quit [Ping timeout: 244 seconds]
<KotCzarny> its not what im suggesting
<KotCzarny> im saying the room will get hotter inevitably
<oliv3r> but yeah
<oliv3r> the temperature of the machine, especially under, will be warmer
<KotCzarny> you should make a test how much, by starting 1h job and see how much
<KotCzarny> in a room without much air flow (no ac?)
<oliv3r> yeah we want/need a climat control chamber
<oliv3r> and set its temperature to 35
<oliv3r> but we did do a test like that :)
<oliv3r> our entire machine had to be tested
<oliv3r> but without lima-memtester and on a good batch of boards, that works quite well
f0xx has quit [Quit: terminated!]
<KotCzarny> well, its not about temperature control, but simulating user's env
f0xx has joined #linux-sunxi
<oliv3r> aye; hence climate controled chamber
<KotCzarny> ie. room 3m x 3m, heater, no airflow
<oliv3r> it does exactly that
<KotCzarny> climate control sets constant temp, right?
<oliv3r> it simulates environments
<oliv3r> you can, but you can also let it vary
<oliv3r> to see behaviour on temperature variations
<oliv3r> but worst case (95% humidity, 40C) is I think what we tested
<oliv3r> which are pretty harsh settings
<KotCzarny> ;)
<oliv3r> 99% humidity even maybe
<KotCzarny> its not that worst case, seen worse in nyc
<KotCzarny> ;)
<oliv3r> lol
<oliv3r> well our max. temp that we say our product works at, is 35C
<oliv3r> so by testing 40C we are already exceeding our maximum operating temperature
<oliv3r> users will do it anyway
<oliv3r> and the electronics is not at question here
<oliv3r> but the materials tend to degrade to fast in these high temperatures
<oliv3r> especially PVA gets too soft to print at 40+
<oliv3r> think coconut oil ;0
<KotCzarny> on the plus side, if it gets too cold it hurts the print quality too
Axl_z has quit [Read error: Connection reset by peer]
apritzel has joined #linux-sunxi
chomwitt has joined #linux-sunxi
<oliv3r> yeah
gianMOD has quit [Remote host closed the connection]
tkaiser has joined #linux-sunxi
whaf1 has joined #linux-sunxi
<tkaiser> Can anyone able to speak Chinese read here when first R40 hardware will be available: http://forum.banana-pi.org/t/the-2nd-allwinner-bpi-campus-maker-tournament-banana-pi-allwinner/2402
<oliv3r> tkaiser: ah you are here
<oliv3r> not that its needed anymore :p
<tkaiser> Nope, not really ;)
<oliv3r> i found the secret combo to make mainline u-boot work with 3.4
<oliv3r> :)
<tkaiser> ?
<KotCzarny> its supported since ages
whaf has quit [Ping timeout: 256 seconds]
<KotCzarny> you should read wiki more
<KotCzarny> :)
whaf has joined #linux-sunxi
mosajjal has quit [Ping timeout: 252 seconds]
whaf1 has quit [Ping timeout: 260 seconds]
gianMOD has joined #linux-sunxi
<oliv3r> tkaiser: u-boot, your dl link did not work anymore
<oliv3r> KotCzarny: yeah but i need some config parameters
<KotCzarny> all in wiki
<oliv3r> KotCzarny: and yes i should; its hard to imagegine that me and libv had the top 'commits' on the wiki 3 years ago
<KotCzarny> internet is a great memory
<oliv3r> ironically, i have a new page open right now on my other monitor :)
<oliv3r> where i'm starting to document my memory testing process
<oliv3r> i wonder if my book is mentioned on the wiki at all!
<oliv3r> omg no mention of the book on the wiki!
<oliv3r> bad advertisement for myself :p
<oliv3r> no wonder sales haven't been through the roof
<wens> lets see
<KotCzarny> who reads books nowadays?
<KotCzarny> MTD driver for sunxi-3.4 kernel requires modified u-boot. It is not stable and therefore not suitable for production environment.
<KotCzarny> he he
<oliv3r> KotCzarny: me and boris did a lot of early work on the mtd stuff
<oliv3r> well boris mostly
<KotCzarny> i guess it might be outdated
<oliv3r> and i found it to be very unstable
<oliv3r> lots of corruption
<wens> tkaiser: says nothing about general availability
<wens> some doc mentioned device tree... maybe the r40 will come with 3.10, and their version of DT (read FEX)
dr1337 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
snblitz2 has quit [Ping timeout: 250 seconds]
whaf has quit [Ping timeout: 250 seconds]
whaf has joined #linux-sunxi
<plaes> tkaiser: do you now have access to "move" feature in wiki?
fdcx_ has quit [Quit: Leaving]
kaspter has joined #linux-sunxi
lamer14776464357 has joined #linux-sunxi
lamer14776464357 has quit [Client Quit]
lamer14776465769 has joined #linux-sunxi
lamer14776465769 has quit [Client Quit]
tkaiser has quit [Ping timeout: 250 seconds]
tkaiser has joined #linux-sunxi
<tkaiser> wens: Thanks, R40 is 3.10.65, I looked through one of the SinoVoip OS images already: http://forum.armbian.com/index.php/topic/2169-allwinner-r40-some-already-available-information/
<oliv3r> can anybody check http://linux-sunxi.org/File:Lime2_dram_freq_testsetup_01.jpg and see if they get the recent file WITH the annotations added?
<KotCzarny> red texts?
<tkaiser> plaes: Will check next time and poke you if it doesn't work ;)
<KotCzarny> yes, looks it shows them
<oliv3r> ohh, the R40 (A40) is the pincompatible A20/A10?
<oliv3r> KotCzarny: tnx
<KotCzarny> pincompatible but not dropin compatible
<tkaiser> oliv3r: Not pin compatible
<oliv3r> tkaiser: that sucks
<oliv3r> i thought the whole a10 -> a20 transition was great
<tkaiser> oliv3r: According to Olimex/Tsvetan Allwinner still wants to provide a pin-compatible quad-core successor.
<tkaiser> But R40 is not
<oliv3r> so there will be an 'A40' eventually :p
<oliv3r> tkaiser: i just thought so as you wrote: 'quad-core A20 successor'
<tkaiser> oliv3r: That's an Allwinner quote
<oliv3r> tkaiser: ah sorry :)
<oliv3r> well obviously not really
<KotCzarny> or just badly photoshopped
<oliv3r> but a13 upgrade? as it's a gul-wing design
<oliv3r> lol more probable
<mripard> oliv3r: it's a chip
<mripard> so, definitely photoshopped :)
<oliv3r> or is it a C.H.I.P. !
<oliv3r> lol
<oliv3r> yeah
<tkaiser> oliv3r: Badly photoshopped, just zoom in and laugh
<oliv3r> oh wow indeed
<tkaiser> mripard: Do you know whether GR8 is still 55nm process?
<oliv3r> a bit supprising that it's vertically printed however
gianMOD has quit [Remote host closed the connection]
gianMOD has joined #linux-sunxi
gianMOD has quit [Remote host closed the connection]
gianMOD_ has joined #linux-sunxi
premoboss has quit [Remote host closed the connection]
apritzel has quit [Ping timeout: 250 seconds]
popolon has joined #linux-sunxi
<oliv3r> tkaiser: what was the cpu bench again?
<tkaiser> olive3r: Sorry? What's the purpose? Burning the cpu or benchmarking?
premoboss has joined #linux-sunxi
Putti has joined #linux-sunxi
<KotCzarny> i think he asks about armbian monitor
<KotCzarny> or not
<KotCzarny> ssvb: where does the lima-textured-cube draw? on the fbdev or something else? because when i try to dump the contents of fbdev its regular login prompt and no sign of cube
<KotCzarny> on a sidenote, dmesg is filled with: http://pastebin.com/raw/TviRZryR
terra854 has quit [Quit: Connection closed for inactivity]
tkaiser has quit [Quit: jIRCii - http://www.oldschoolirc.com]
mosajjal has joined #linux-sunxi
gianMOD_ has quit [Remote host closed the connection]
IgorPec has quit [Ping timeout: 260 seconds]
paulk-collins has quit [Quit: Leaving]
pekka10 has joined #linux-sunxi
<oliv3r> tkaiser testing wether the voltage is high enough; so stability
cnxsoft has quit [Quit: cnxsoft]
<oliv3r> hah! the sunxi service is still called maxima :D
IgorPec has joined #linux-sunxi
<oliv3r> ssvb: would it be possible to do a static release of https://github.com/ssvb/a10-dram-tools or atleast a10-meminfo? (I have compiled it and it works fine, but I want to link to it from a wikipage)
tkaiser has joined #linux-sunxi
fdcx has quit [Ping timeout: 260 seconds]
<tkaiser> oliv3r: Search for 'linpack' here: https://github.com/longsleep/build-pine64-image/pull/3 (the unoptimized linpack versions you get via apt-get won't do it)
<oliv3r> tkaiser: ideally i'd want statically linked binaries so i can put it on the armbian image :)
<oliv3r> but let me look there
<tkaiser> oliv3r: Good luck, building this version is PITA ;)
<tkaiser> oliv3r: I asked in Armbian forum for volunteers building the whole stuff as .debs a few months ago (to provide the ability to check dvfs settings individually and automated on each device). But no one answered and I won't do this due to lack of time.
<oliv3r> tkaiser: so, your telling me to use linpack, but i can't use the packages, so i need to manually build it, but there are no binaries as of yet?
<tkaiser> Just follow ssvb's first link there. The person reporting undervoltage problems with RPi 3 explains how Linpack detected that and he links to a 'tutorial' how to build this stuff.
<tkaiser> You've to first build OpenBLAS and then using this the Linpack variant with NEON optimizations.
<oliv3r> fair nuff
<tkaiser> The recipe to success is to blindly follow his tutorial even if it doesn't seem to make sense sometimes :)
<tkaiser> And with A20 it might take some time to build...
<oliv3r> oh native compile
<oliv3r> well it's a form of stress test, so i don't mind i suppose :)
<oliv3r> the weekend is long anyway :)
<oliv3r> so let it compile i'd say
<tkaiser> Yes, and the benefits are nice. You might then decrease the voltage of dvfs operating points by 20 mV in an automated fashion and don't run into freezes/deadlocks but Linpack will happily report 'invalid results'.
<oliv3r> i would expect this to be ssvb's work :p http://linux-sunxi.org/Hardware_Reliability_Tests
fdcx has joined #linux-sunxi
montjoie has joined #linux-sunxi
<tkaiser> Sure, but when we played with A64 back in March I found Linpack more easy to detect undervoltage situations.
<tkaiser> oliv3r: Nope, just use your browser's search function and search for 'linpack' ;)
<oliv3r> duh
<tkaiser> In this github issue page
<oliv3r> that's the one
<tkaiser> yep
<KotCzarny> ssvb: hrm, you only run memtester on one core?
<oliv3r> for lima-memtester? shouldn't matter for memory bw as you do that with the lima part
<oliv3r> as for cpu/temperature load, yeah you could couple it with cpuburn i suppose
<KotCzarny> i've hacked lima-memtester to show error count/freq/temp on the console
<tkaiser> KotCzarny: Why should memtester run with multiple instances?
<KotCzarny> tkaiser, im curious if soc can be responsible for memory errors too
gianMOD has joined #linux-sunxi
<tkaiser> KotCzarny: SoC is responsible, that's why there's this GPU stress test ;)
<KotCzarny> and if soc is responsible, then temperature of the soc might matter too, or not?
kaspter has quit [Remote host closed the connection]
kaspter has joined #linux-sunxi
<KotCzarny> if anyone could check if it works (i dont have hdmi cable atm): https://transfer.sh/aNThC/lima-memtester-verbose
<KotCzarny> another thing is that i've muted those pp/gp job errors 0x00800000 (_MALI_UK_JOB_STATUS_END_UNKNOWN_ERR)
<KotCzarny> so at least console is readable now
<tkaiser> KotCzarny: Old boring Mali400 clocked at 600 MHz generates more heat than the 4 A7 cores in H3.
naobsd has quit [Ping timeout: 276 seconds]
<KotCzarny> still, when tested on my opi+2e i didint reach 60C, while having 4 cores at max it easily goes over 70C quite fast
<KotCzarny> but as i said, my hdmi cable is missing and i cant confirm the cube spins
<tkaiser> KotCzarny: You're the one always baking your own kernels, true? I was talking about Mali400 running at 600 MHz and not 252 ;)
<mripard> tkaiser: no idea
<KotCzarny> ahm, right, forgot about it
<KotCzarny> lets 'fix' the fex
<tkaiser> mripard: Thanks anyway :) Was just curious since Cortex-A8 seems not be the best choice for an IoT node too
mnemoc has joined #linux-sunxi
apritzel has joined #linux-sunxi
Nacho has quit [Ping timeout: 260 seconds]
fest has quit [*.net *.split]
menomc has quit [*.net *.split]
diego71_ has quit [*.net *.split]
as0d70apf has quit [*.net *.split]
ganbold has quit [*.net *.split]
mozzwald_ has quit [*.net *.split]
apritzel has quit [Ping timeout: 260 seconds]
<KotCzarny> tkaiser, nope. same story. temperature reported is influenced mostly by cpu freq/voltage/load
<KotCzarny> not going over 60 in any case
<tkaiser> KotCzarny: how do you measure GPU clockspeed?
<KotCzarny> i dont, just set it in fex
<tkaiser> KotCzarny: Why not writing it in a book instead? ;) Obviously what you did has no effect. But we don't need to argue here, not interested since it's known what's happening
<KotCzarny> well, kernel was sourced from armbian, and your pointers on mali clock
<tkaiser> Anyway, taking pieces out of something might not be the same as using something obviously ;)
<KotCzarny> as for the mali speed, if allwinner didnt castrate mali driver, it would be easily checkable/settable via /sys/
premoboss has quit [Quit: Sto andando via]
gianMOD_ has joined #linux-sunxi
fest has joined #linux-sunxi
mozzwald has joined #linux-sunxi
diego71 has joined #linux-sunxi
gianMOD has quit [Ping timeout: 250 seconds]
ganbold has joined #linux-sunxi
gianMOD_ has quit [Remote host closed the connection]
vishnup has joined #linux-sunxi
pg12 has quit [Ping timeout: 245 seconds]
vishnup has quit [Ping timeout: 244 seconds]
pg12 has joined #linux-sunxi
matthias_bgg has quit [Quit: Leaving]
<KotCzarny> tkaiser, btw. your assumption that mali400 does 600mhz is also wrong (its overclocking according to http://www.arm.com/products/multimedia/mali-gpu/ultra-low-power/mali-400.php )
mossroy has joined #linux-sunxi
vishnup has joined #linux-sunxi
Putti has quit [Ping timeout: 252 seconds]
<tkaiser> KotCzarny: Who cares? And regarding clock changes it should be CONFIG_MALI400_DEBUG (then dmesg should tell when throttling occurs). But I've no idea how your kernel build looks like so...
matthias_bgg has joined #linux-sunxi
<ptx0> you can run your own thermocouple you know
JohnDoe_71Rus has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org]
[Awaxx] has quit [Remote host closed the connection]
[Awaxx] has joined #linux-sunxi
vishnup has quit [Ping timeout: 260 seconds]
jernej has joined #linux-sunxi
<tkaiser> ptx0: All modern SoCs have this stuff built-in for one single reason: Measuring temperatures and preventing overheating/damage. And the sensor is there where it should measure (expection: RPi 3)
premoboss has joined #linux-sunxi
yann-kaelig has joined #linux-sunxi
cnxsoft has joined #linux-sunxi
<ssvb> oliv3r: about "456 MHz with ambient temperature of around 22 C; all boards still running, 480 they failed the long test, 504 MHz they failed even a 30 minute test"
<ssvb> oliv3r: it just means that most likely 432MHz is a reasonable choice, and 408 if you really want to play it safe
vishnup has joined #linux-sunxi
<ssvb> oliv3r: still I'm somewhat surprised and find it a bit suspicious that all your boards have the same DRAM clock speed limit
<oliv3r> well close batches?
<oliv3r> i do notice a difference between 'when' they crash
<oliv3r> but yeah it seems they all act reasonable the same
fl_0 has quit [Ping timeout: 245 seconds]
<oliv3r> then again, it shouldn't be that supprising, it's the same board :)
<ssvb> I would have expected at least a little bit of variation between different boards, something like https://linux-sunxi.org/Orange_Pi_PC#DRAM_clock_speed_limit
<oliv3r> yeah I agree
<oliv3r> and there is
<ssvb> BTW, you can try to add your test results to the Lime2 wiki page, or at least somewhere
<oliv3r> ah yeah
<oliv3r> well i'm shuveling stuff around already
<oliv3r> cause it really goes with Hardware_Reliability_Test
<oliv3r> and the results specifically are for the lime2 in this case
<KotCzarny> umkay, caved in and went to store for hdmi cable, lets see the love cube
station has quit [Remote host closed the connection]
<KotCzarny> no cube. oh well
<oliv3r> aww :(
my123 has quit [Ping timeout: 260 seconds]
<KotCzarny> that answers few questions
<tkaiser> KotCzarny: you're always baking your kernels yourself? ;) 'optimize for size' would be responsible for no HDMI
<KotCzarny> nah, hdmi is on
<KotCzarny> just mali fails to run apparently
<KotCzarny> tkaiser, btw. did you see love cube on opi+2e ?
<tkaiser> Of course not, Mali on H3 with 2 GB doesn't work
fl_0 has joined #linux-sunxi
<KotCzarny> shucks
<tkaiser> We talked about that few months ago
<KotCzarny> yeah, my memory is spotty
<KotCzarny> starting to remember the whole talk
<KotCzarny> pity no one tried to check/fix it
apritzel has joined #linux-sunxi
<KotCzarny> i wonder if android kernel from emmc would run with plain linux
<oliv3r> could be possilbe
<KotCzarny> or if anyone used mali with android on opi+2e
<tkaiser> KotCzarny: Mikhail from Armbian team tried to but to no avail. And 'Android kernel' is what you're using
<oliv3r> ssvb: but no worries, i have some old batches (with nand) that i will test after this series...
my123 has joined #linux-sunxi
my123 has quit [Changing host]
my123 has joined #linux-sunxi
<jernej> KotCzarny: mali framebuffer r4p0 driver works on opi+2e
<KotCzarny> tkaiser, yeah, but compiled with different options. or asking another way, anyone tried android shipped with opi+2e and some 3d apps on it?
<jernej> but I guess that this version doesn't help you
cnxsoft has quit [Remote host closed the connection]
<KotCzarny> jernej, care to elaborate?
<jernej> about what?
<KotCzarny> if it runs, why it doesnt help ?
<jernej> ah, I thought you need X11 version
<KotCzarny> nah, for now just for limamemtester
<KotCzarny> dont need for x
<jernej> then it should work
<tkaiser> jernej: Tested with 2 GB DRAM?
<jernej> tkaiser: yes, ordinary opi+2e version
<jernej> or do you mean with some kind of software limitation?
<jernej> I mean, artificialy lowered amount of RAM?
<jernej> I have this board here, at hand
<KotCzarny> artificall lowering doesnt help
<oliv3r> how do I easily link to a section on the wiki? e.g. the url#section
<KotCzarny> tried it with uboot/linux
<jernej> let me test again, but I'm pretty sure it works
<KotCzarny> oliv3r: [[something#section]]
<oliv3r> ah double quotes
<oliv3r> well the url should also work
<KotCzarny> (or one [], dont reme,mber)
<oliv3r> so how do I easily and reliably get the section part
<ssvb> jernej: the lima-memtester tool is using the open source lima driver code only because we can't redistribute the proprietary blob
<jernej> yeah, I know
<KotCzarny> ssvb, but it could run on normal linux, right?
<jernej> I just want to say that there is a version, which works
<oliv3r> KotCzarny: it's [] but that's not my issue, it's properly obtaining the source url to put into the [] :)
<ssvb> and the lima userland code is currently not compatible with the r4p0 mali kernel driver
<KotCzarny> oliv3r: assume it doesnt change, unfortunatelly you cant guess otherwise (or i dont know)
<oliv3r> erm [[]]
<oliv3r> the contents table of course!
<KotCzarny> toc can change too
<oliv3r> yeah but i can get the url from there :)
<oliv3r> so atleast i can make the initial reference
<oliv3r> ssvb: also, i'm going to raise the ambient temperature of my 456 lime2 to see if that influences things
<tkaiser> jernej: Ok, so it's just a limitation of lima-memtester (which is the *only* use case I ever had for Mali400)
tkaiser has quit [Quit: jIRCii - http://www.oldschoolirc.com]
<ssvb> oliv3r: thanks! this would be very interesting
<jernej> ssvb: so mali drivers on linux-sunxi github also have license issues?
<ssvb> oliv3r: and please just check if the dram_zq value (as reported by a10-meminfo) changes between runs at different temperatures and whether it's the same or different on different boards
<ssvb> jernej: we don't have any mali blob with a reasonably permissive EULA notice yet
jstein_ has joined #linux-sunxi
jstein_ is now known as jstein
petr has quit [Ping timeout: 244 seconds]
petr has joined #linux-sunxi
<slapin> hi, all!
<slapin> it looks like ARM allows redistribution of Mali drivers since Jan 2016, can anybody confirm?
<jernej> slapin: Yes, but from what I understand, this applies only to blobs which came with such license.
<jernej> but I'm no expert in licenses
premoboss has quit [Ping timeout: 260 seconds]
<ssvb> slapin: yes, ARM has changed their EULA to a more permissive one and the mali blobs with the new EULA are slowly showing up
<ssvb> slapin: the userland blob may be built in different configurations (32-bit/64-bit, Mali-400/Mali-450, fbdev/x11/wayland, UMP/DMA-BUF)
<ssvb> slapin: it's only a matter of eventually receiving a suitable blob with the new EULA
leviathanch has joined #linux-sunxi
<oliv3r> ssvb: 50 degree's ambient nice enough? :)
<oliv3r> ssvb: i've put the full output of a10-meminfo after each boot in my document
<oliv3r> see the ods
<ssvb> oliv3r: are you telling me that now I need to install openoffice for that? ;-)
<oliv3r> lol
<oliv3r> it was easiest initially for me :p
<oliv3r> and no, install libreoffice :D
reinforce has quit [Quit: Leaving.]
critch has joined #linux-sunxi
<critch> anyone have a good pointer to the fel mode via sd card boot image referenced in http://linux-sunxi.org/FEL
<KotCzarny> what's a pointer?
<critch> As in link, help, information on the boot image.
<KotCzarny> Included in our sunxi-tools repository
<critch> I must be overlooking it
<KotCzarny> might require building?
<KotCzarny> sunxi-tools/bin/fel-sdboot.sunxi
<KotCzarny> according to the page you linked
<critch> Thanks, that might be what I was missing
<KotCzarny> you should get some sleep
<KotCzarny> or just a 5 minutes of looking at the forest through the window
<KotCzarny> :)
<ssvb> critch: https://linux-sunxi.org/FEL#Through_a_special_SD_card_image provides all the necessary information
<critch> beginning of my day. playing with a new a64 tablet
<critch> yes, should have read the command lines. I was expecting more explination in the write up.
<ssvb> critch: I have just added a wget command, so you even don't need to go to the sunxi-tools repository to fetch this particular file
<critch> I already downloaded and built all the tools. Should be handy anyways.
<critch> thank you
<KotCzarny> ssvb, is this file prerebuilt or some binary blob?
<critch> KotCzarny: looks like just enough of a boot image to not boot properly so the CPU goes into fel mode.
<slapin> ssvb: fbdev/x11/wayland is not a problem, it looks like more a kernel part than userspace, buut UMP/DMA-BUF is important bit, so we might eventually need the blob for that...
<ssvb> KotCzarny: the sources of this blob are available in the sunxi-tools repository, the compiled blob is provided there for the sake of convenience
<slapin> ssvb: as running UMP these days looks not effective having shiny new effective API in kernel...
<KotCzarny> uhum
<ssvb> slapin: having UMP blobs should make the xf86-video-fbturbo DDX usable with only some minor tweaks
<ssvb> slapin: switching to DMA-BUF needs a bit more work
<slapin> ssvb: this work is well worth it
<ssvb> we'll see, right now I'm still waiting for any legally redistributable mali blobs
<slapin> btw, regarding A64 - was anybody able to boot pine64 android with stock u-boot?
<ssvb> stock u-boot?
<slapin> ssvb: mainline one
<ssvb> no, you can't do this
<slapin> ssvb: why?
<ssvb> because the whole boot process has become more complicated with the addition of ATF
<slapin> ssvb: can't ATF be attached to u-boot? is there ATF source code for A64?
<ssvb> you can find some notes here - https://linux-sunxi.org/Pine64#U-Boot
<ssvb> of course it is technically possible to make something that is compatible with the Allwinner's Android software stack
<ssvb> but certain design decisions made by Allwinner are quite questionable, you can ask apritzel for more details
<critch> Another of those, I need to read the whole instructions bits. I missed the seek at the end of the dd line. Makes all the difference as to if the sdcard trick works or not.
<apritzel> slapin: there is ATF source in the A64 SDK
<apritzel> slapin: but it's quite broken, I fixed up most of it
<ssvb> apritzel: my understanding is that slapin wants the mainline U-Boot to be compatible with the Allwinner's Android kernel
<apritzel> longsleep took most of it in this ATF tree, till the point when I made changes which deviate from the original AW boot process
<apritzel> ssvb: I got this
tsuggs has joined #linux-sunxi
<apritzel> but I really don't want to go with Allwinner's boot process design choices
<apritzel> I think in the long run we may move the Android kernel to something more upstream like
<apritzel> either by using an upstream AOSP kernel
<apritzel> or by changing the 3.10 based kernel to go with the new firmware design
<apritzel> that's seem more worthwhile to me than tuning the upstream kernel to AW's firmware
<slapin> well, I want to be able to boot with something I can build myself for now (and not lose youtube feature), and I don't like magic they put in u-boot and all this boot0 stuff.
<silviop> I have sd-breakout on A33 tablet, there is a possibility to write an FEL UART0 Hello word ? Without initialize ram , i only need to know if 0x2000 code could be executed and return to FEL , only test
<slapin> apritzel: as I understand the best working base is Pine64 android now, am I right?
<silviop> I suppose that UART0 is initialized because boot0 output his messages
<apritzel> slapin: depends on the features you need: if this involves WiFi, graphics, sound: then yes
Leepty has quit [Remote host closed the connection]
<apritzel> slapin: if you are OK with Ethernet, USB, serial, MMC, then the almost-upstream kernel works quite well for me
<apritzel> slapin: but you can build the whole chain for Android? Or do you care about boot0?
<slapin> ssvb, apritzel: I don't need wifi, but I definitely need video acceleration and 3D.
<slapin> apritzel: I want to start by getting rid of boot0 and have clean ATF/u-boot without strange stuff
<apritzel> video isn't there at all, but I see a path to get there
<silviop> I read it, but no one write helloworld for tablet that use sdcardbreakout (A13 or A20)
<silviop> Is not A33 only problem
<apritzel> slapin: I have a working SPL port, which replaces boot0
<slapin> apritzel: and everything documented...
<ssvb> slapin: video decoding acceleration works in GNU/Linux if you use longsleep's U-Boot & kernel tweaks for the Allwinner BSP sources, but not for youtube
<apritzel> I didn't think about how this would fit in with the Android kernel yet
<apritzel> slapin: but it should be possible to use that to-be-upstreamed SPL support for the Android kernel as well
<ssvb> slapin: and the EULA notice is still missing for the mali blobs, even though some X11 blobs had been provided
<slapin> apritzel: what is major requirement to boot android kernel? as I understand, we can change it
<slapin> ssvb: can you link me there?
paulk-aldrin has joined #linux-sunxi
<apritzel> slapin: you can change and recompile U-Boot, Android kernel and ATF thanks to longsleep's work on the Allwinner kernel base
<slapin> ssvb: as I understand all newer blobs are permissive...
<apritzel> slapin: boot0 is the only thing without a source
<ssvb> slapin: video acceleration via libvdpau-sunxi - http://forum.pine64.org/showthread.php?tid=424
<apritzel> slapin: but you can patch it (I did it to move U-Boot below 1MB instead of the 19MB in the original boot0)
jernej has quit [Quit: Konversation terminated!]
<slapin> ssvb: well, last time I tried to make it work in browser I failed :(
<ssvb> slapin: the mali blobs story - http://forum.pine64.org/showthread.php?tid=587
<critch> Thanks ssvb, KotCzarny. I was able to flash my custom firmware to our new tablet now. So now I get to go write the instructions for our customer to do the same.
<slapin> apritzel: I'd prefer standard SPL though, I think original boot0 can be disassembled as it is probably quite small...
<ssvb> slapin: just work harder on it ;-)
<KotCzarny> hrm, what's up with green display over hdmi?
<KotCzarny> i see old messages about yuv vs rgb, but why suddenly my display went green?
<slapin> ssvb: it is not fun at all
<slapin> ssvb: anyway, is mali really needed for 2d accel? doesn't we have some 2D hardware?
<ssvb> slapin: then find somebody who can do your boring work for $$$$
<slapin> *don't we?
<KotCzarny> slapin: you are reading wrong forums
<ssvb> mali is not very good for 2D
<KotCzarny> pine64.org is full of misinformation
<slapin> ssvb: it os not boring it is overwhelming
<slapin> ssvb: neither SGX or any other embedded GLES thing.
<apritzel> slapin: as I said: I think we can tweak the upcoming SPL to load the AW based U-Boot, ATF and kernel
<apritzel> to become blob-free on this side as well
<slapin> apritzel: I think it would be great to have some sane u-boot too...
<apritzel> slapin: that is where it starts got get complicated
<apritzel> as AW's U-Boot is quite hackish
<apritzel> as I mentioned before it would be easier to change the Android kernel to not use AW's broken firmware interfaces
<apritzel> then we could move to one firmware to rule them all
<ssvb> slapin: as for the 2D hardware, A64 has the 'transform' unit instead of the G2D (Mixer processor), this 'transform' thing is a simple hardware block which can do basic color formats conversion and rotation
<apritzel> ssvb: rotation in 90 degree steps or more freely?
<ssvb> 90 degree steps
<apritzel> no scaling?
<ssvb> I don't remember, need to look up the sources
<ssvb> the biggest change is that it does not have the ROP (raster operations) unit anymore
<apritzel> wow, compiling all modules from the Ubuntu kernel .config takes a while ...
premoboss has joined #linux-sunxi
<ssvb> silviop: nobody with the right skills just has the A23/A33 hardware
<ssvb> silviop: but as NiteHawk already explained, adding support for A23/A33 to the uart0-helloworld-sdboot example application is reasonably simple and straightforward
<ssvb> silviop: you can do it yourself as a learning exercise :-)
|Jeroen| has joined #linux-sunxi
<longsleep> apritzel: get a faster compile host and use distcc, Ubuntu kernels include basically _everything_ ...
<apritzel> longsleep: yeah, I need to connect that old PhenomII X6 under my desk
<apritzel> longsleep: but hopefully I don't need to compile the modules everyday
<longsleep> urg phenomII x6 - not very efficient
<apritzel> longsleep: indeed
<apritzel> that's why it's turned off most of the time
<apritzel> but it's way faster for compilation than my X230
<longsleep> i nowadays compile on a bunch of very cheap vps servers with xeon cpus and ssd backed disks
<slapin> ssvb: too bad, then GUIs will be ineffective on A64...
<ssvb> slapin: software rendering is generally very effective for a simple X11 desktop
<ssvb> slapin: GPU has exactly nothing to offer to improve the simple 2D performance, because you are memory bandwidth limited anyway
<slapin> apritzel: do I uderstand it right - the hardware starts in 640bit mode, then boots SPL/boot0 which start ATF (all 64 bit) then ATF starts u-boot in 32-bit svc mode (insecure world) and from now on everything works via svc calls?
netlynx has joined #linux-sunxi
<slapin> ssvb: hardware rasterized lines do a huge boost here on my graph apps,,,
<apritzel> slapin: no, the SoC starts in 32-bit mode, BROM, boot0 and U-Boot are all 32-bit
<ssvb> slapin: which apps exactly?
<apritzel> only ATF is 64-bit (because there was no 32-bit support back at the time)
<ssvb> slapin: based on your description, it does not seem like a very typical workload
<apritzel> boot0 briefly switches to 64-bit to start ATF, this then drops back to 32-bit into U-Boot
<apritzel> slapin: U-Boot then just before launching the kernel calls ATF to switch back to 64-bit
leviathanch has quit [Remote host closed the connection]
<apritzel> slapin: but this is not the real issue, it's what they do in ATF and U-Boot and that you need a SCP binary for that firmware toolchain to work
<slapin> apritzel: is kernel 64-bit then?
<apritzel> slapin: yes
gianMOD has joined #linux-sunxi
<apritzel> SCP is another thing I forgot to mention before: it runs on the OpenRISC core and is a total blob
<apritzel> beside missing source it does weird things (like turning cores off when temperature is too high and not bringing them back)
reinforce has joined #linux-sunxi
<slapin> apritzel: well, my board never gets hotter than 38C
<slapin> SCP should be easy enough to hack, need to disassemble this stuff...
<ssvb> slapin: this only means that you have never run anything heavy on it ;-)
<apritzel> AFAIK you don't need anything particularly "heavy", just normal Linux load is sufficient ;-)
<apritzel> Android seems to have a different load characteristic
<apritzel> slapin: or maybe you have lost some cores already? ;-)
<slapin> ssvb: GTA games should be heavy enough... :) but it is 29C on them, on non-fulscreen video it gets up to 38C, on full-screen videos it gets to 32-36C
<slapin> ssvb: and it works 24/7
<slapin> apritzel: how can android have different load characteristics? :)
<apritzel> slapin: because they schedule differently and the workloads are much different
<slapin> apritzel: anyway you intrigued me with SCP a bit, I worked a little bit witj OpenRISC, I think I can hack on it a bit, where is the firmware?
ElBarto has quit [Ping timeout: 252 seconds]
<apritzel> hidden in an AW provided image, or find it somewhere in longsleep's build system
Nacho__ has joined #linux-sunxi
<apritzel> slapin: anyway, the point is not to hack the SCP, but whether it is really useful as it is
<apritzel> in the long run it seems worthwhile to have something running on the OpenRISC core, but atm there is more important stuff to do
scream has joined #linux-sunxi
<slapin> apritzel: well, if it loses cores, it is important stuff to do though...
leviathanch has joined #linux-sunxi
apritzel has quit [Ping timeout: 256 seconds]
<pcduino> does anyone know why
<pcduino> does anyone know why I could be getting 104857600 bytes (105 MB) copied, 3.45724 s, 30.3 MB/s on a A20 SoC? It says it supports DATA 1.5Gb/s
<KotCzarny> because hdparm is not a benchmark, because something east cpu cycles ?
<pcduino> I am using a 6Gb/s rated SSD
<KotCzarny> *eats
<KotCzarny> because you are using braindead governor?
<pcduino> so even though it can handle up to that speed I am limited by the cpu cycles :/
<pcduino> the cpu usage does not exceed 24%
<KotCzarny> try setting performance governor for a start
<KotCzarny> and disable any interrupt eaters too
<KotCzarny> disabling display/hdmi might help
<pcduino> there is no screen attached
<pcduino> the board is pretty much idle
<KotCzarny> doesnt matter if you have display code enabled
<pcduino> okay I need to know in extreme anyway so if it can only run at 30 MB/s then so be it
<slapin> pcduino: 30MB/s is very good speed for these things though
<pcduino> oh yes I know
<KotCzarny> ah, right which soc?
<pcduino> the board I am replacing was 7 MB/s
<KotCzarny> a20 should be able to do more
<pcduino> yes A20
<pcduino> I think it should be able to do 180 MB/s
* pcduino wishful thinking I know :)
tkaiser has joined #linux-sunxi
<KotCzarny> Timing buffered disk reads: 220 MB in 3.01 seconds = 72.98 MB/sec
<KotCzarny> my old banana
<KotCzarny> but its wd drive
<KotCzarny> (wd green)
<pcduino> am using dd if=/dev/zero of=/root/testfile bs=100M count=1 oflag=direct
<pcduino> the wd won't go fast enough
<tkaiser> pcduino: With A20 you might get ~40 MB/s write and +200 MB/s read. And both depend on memory bandwidth
<pcduino> yes noticed it used up the memory
<KotCzarny> still, 30MB on write isnt bad
<tkaiser> bandwidth not useage ;)
<tkaiser> pcduino: Which kernel and which settings are you using?
iaglium_ has quit [Ping timeout: 250 seconds]
<pcduino> 4.8.1-pcduino3b-test-dirty :P
<tkaiser> pcduino: Then you use most probably 960 MHz cpufreq maximum. Which scheduler?
<slapin> apritzel: as I understand u-boot can work as TF, isn't it intended goal to move everything into u-boot instead of ATF+u-boot ?
<KotCzarny> tkaiser, do you have a clue why hdmi went green?
<pcduino> 960000 kHz operating points?
<tkaiser> KotCzarny: you bought a cheap cable? ;)
<slapin> apritzel,ssvb: also does it boot without SCP firmware?
<KotCzarny> hehe, nah, it was ok at first, but then after reboot it went green
<tkaiser> pcduino: anyway with ML kernel and those settings (408 DRAM clockspeed, 960 MHz cpufreq) you're limited to ~35MB/s write.
<ssvb> slapin: the mainline kernel? yes
<ssvb> slapin: the Allwinner's BSP kernel? probably not
<pcduino> was running a very large system on an A8 and microsd but was to slow so moved it to a intel server but now want to see if this A7 with an ssd can handle it
maz has quit [Quit: Leaving]
steev has quit [Read error: Connection reset by peer]
<tkaiser> pcduino: I did some overclocking tests long ago but that's not recommended as data corruption might occur: http://forum.lemaker.org/forum.php?mod=viewthread&tid=12167&extra=page%3D1
ninolein has quit [Quit: No Ping reply in 180 seconds.]
<pcduino> tkaiser: I need to stay safe
<slapin> KotCzarny: my tv likes allwinner devices, but 1 of LCD displays do not like HDMI output of device, on some of devices it is possible to change HDMI output mode, when no audio frames present no green, same thing happens for HDMI-DVI converters.
<pcduino> this is for a product
nixdork has quit [Quit: EliteBNC free bnc service - http://elitebnc.org/]
scream has quit [Ping timeout: 256 seconds]
ninolein has joined #linux-sunxi
BurtyB has quit [Ping timeout: 256 seconds]
<pcduino> even still 32 MB/s write a lot better than 7 MB/s :)
<ssvb> slapin: just as a clarification, I'm talking about the Allwinner's SCP blob
steev has joined #linux-sunxi
foodev has quit [Ping timeout: 256 seconds]
Turl has quit [Ping timeout: 256 seconds]
<pcduino> and I would imagine the read is an equal multiplier
<ssvb> slapin: SCP itself is not a bad thing and will be probably supported in the mainline software stack too (of course, as an open source implementation)
foodev has joined #linux-sunxi
<tkaiser> pcduino: A 'cheap' trick with mainline kernel is to use btrfs with LZO compression. If your data is compressible IO bandwidth increases magically. With /dev/zero you get even bottlenecked by CPU easily ;)
<pcduino> ah okay thanks
<pcduino> I will run a full system test using a 150GB database :)
Putti has joined #linux-sunxi
<KotCzarny> hmm, this box had 40 days of uptime, and i remember doing tests on different monitor, so it might have negotiated different colorspace, hum
scream has joined #linux-sunxi
BurtyB has joined #linux-sunxi
nixdork has joined #linux-sunxi
Turl has joined #linux-sunxi
<tkaiser> pcduino: If latency and random IO is important than A20 can shine with databases, here SATA really makes a difference even if sequential write speeds are slow. The smaller the data chunks the more you benefit from SATA (even if it's just A20) compared to USB.
<TheLinuxBug> ^^ this
<slapin> does A80 have SATA?
<KotCzarny> nope
<KotCzarny> only a10/a20 have native sata
<slapin> too bad
<Wizzup> we need more A20's
<KotCzarny> and maybe some future a20 successor
<slapin> well, A10 is good work horse, but it is not that fast...
<pcduino> tkaiser: thanks for your help
<Wizzup> has anyone used the smart card readers on the a20 ?
<slapin> also torrents are very good on SATA vs USB... :)
<slapin> Wizzup: I used USB one, no difference comparing to PC
<Wizzup> I mean, doesn't the SoC contain one?
<Wizzup> I'm sure usb ones work :)
<slapin> Wizzup: dunno
<tkaiser> pcduino: But be careful with btrfs' transparent file compression and especially *huge* databases. This might have a performance impact too. Anyway: Never ever use crap tools like dd or hdparm for benchmarking especially when your workload is so different.
mzki has joined #linux-sunxi
* slapin use old SSD on A20 SATA, performance is about 45MB/s works quite well... also it si possible to hack it to support SATA/eSATA splitters and have a pack of disks connected...
<KotCzarny> splitters? or port multipliers?
<slapin> but I prefer good old dreamplug for NAS.
<slapin> KotCzarny: same thing
<tkaiser> slapin: Have a look at Marvell ESPRESSOBin
<pcduino> tkaiser: I will just leave it as it is for now and benchmark it with our system
leviathanch has quit [Remote host closed the connection]
<tkaiser> pcduino: sure, benchmark is necessary anyway. And BTW: If I were you I would also look at the ESPRESSOBin ;)
premoboss has quit [Remote host closed the connection]
<slapin> tkaiser: looks cool, will bake I think...
vishnup has quit [Ping timeout: 250 seconds]
y0g1 has joined #linux-sunxi
vishnup has joined #linux-sunxi
|Jeroen| has quit [Quit: dada]
matthias_bgg has quit [Quit: Leaving]
<TheLinuxBug> WAT
<TheLinuxBug> I want one of those
<TheLinuxBug> looks awesome
<TheLinuxBug> but
<TheLinuxBug> March
<TheLinuxBug> :(
<KotCzarny> get early bird one
<TheLinuxBug> yeah not available anymore
<TheLinuxBug> why you guys keep me in the dark so long :Z
<KotCzarny> because you werent on irc
<TheLinuxBug> I was toooooo
<TheLinuxBug> im always here
<KotCzarny> someone posted it few days before it even launched on kickstarter
<TheLinuxBug> pft where was I when that happened
<TheLinuxBug> btw whats the SATA performance supposed to be like on those?
* BurtyB looked when postage was crazy and decided against it
<TheLinuxBug> any idea?
<KotCzarny> TheLinuxBug: they say bandwidth is more than one terabyte
<KotCzarny> so should easily drive your ssds
<KotCzarny> but what is the reality we will see in a few months
<TheLinuxBug> I want to use it to replace the A20 for my NAS solution
<TheLinuxBug> sounds like would get a bit better performance
<TheLinuxBug> but now looks like have to spend 70-80$ to get it
<KotCzarny> but at least it will be 1MB model
|Jeroen| has joined #linux-sunxi
<tkaiser> KotCzarny: Which is good since 1MB is more than 640K ;)
<KotCzarny> hehe, right, 1GB
<TheLinuxBug> well
<tkaiser> TheLinuxBug: Early bird is still available but with lower shipping costs. They made a mistake in the beginning and can't change shipping. So they stopped the older pledges (and do a refund there or something like that) and have new variants now.
<y0g1> hello
<TheLinuxBug> was looking at a Udoo for this use case
<TheLinuxBug> but it seems they are about twice the cost of that board
<TheLinuxBug> so maybe
leviathanch has joined #linux-sunxi
<y0g1> does somebody use 4.8.X kernel with cubietruck and mmc card?
<TheLinuxBug> tkaiser: all the open pledges I see all have delivery in March 2017
leviathanch has quit [Read error: Connection reset by peer]
<TheLinuxBug> and none are 'early bird'
<TheLinuxBug> if they were I would likely pledge
<tkaiser> TheLinuxBug: In Armbian forum, Free section I tested through all currently available SATA implementations. Well at least A20, i.MX6 and Marvell Aramada 38x. And it's the third pledge if you want to play early adopter.
<TheLinuxBug> ohh wait
<y0g1> after last update i lost access to mmc. It boots (system root is on sata drive) but after booting I can't access /boot from mmc
<TheLinuxBug> yeah I see that now
<TheLinuxBug> missed it first time through
leviathanch has joined #linux-sunxi
<TheLinuxBug> tkaiser: I assume the Armada has much better performance than the rest
<KotCzarny> TheLinuxBug: in short, if you have nas box, its most likely to have marvel soc
<TheLinuxBug> yeah well generally thast true
<tkaiser> y0gi: maybe mmc enumeration changed? In Armbian we switched to UUID a few days ago since device node referencing by name/number sucked too much
<TheLinuxBug> hmm
<y0g1> dmesg |grep mmc says: smc 1 err, cmd 8, RTO !!, no support for card's volts, error -22 whilst initialising SDIO card and smc 1 err, cmd 55, RTO !!
<TheLinuxBug> early bird is only 512Mb ram
<TheLinuxBug> thast a bit depressing
<KotCzarny> yup
<KotCzarny> enough for nas box though
<y0g1> tkaiser: /proc/partitions doesn't show mmc partitions
<KotCzarny> y0g1: try different card?
<TheLinuxBug> I guess
<KotCzarny> or just newer kernel
<TheLinuxBug> though I make use of the extra memory on the A20
<TheLinuxBug> and have a regular cache of about 400mb
<TheLinuxBug> but maybe the SATA performance will offset the need
<y0g1> KotCzarny: this card is ok, old kernel worked ok, but with newer there is problem
<y0g1> do you use 4.8.3 or 4.8.4 kernel with cubietruck and mmc?
<y0g1> I don't remember if I changed somthing in .config so maybe there is something wrong.
IgorPec has quit [Read error: Connection reset by peer]
IgorPec has joined #linux-sunxi
<slapin> heh.. Quatar + $30, Russia + $50... maybe next time...
camh has quit [Ping timeout: 250 seconds]
<KotCzarny> you are looking wrong, it should be 23usd everywhere
p_rossak has quit [Remote host closed the connection]
<KotCzarny> they've updated their shipping carrier
camh has joined #linux-sunxi
<slapin> KotCzarny: thanks a lot!
gianMOD has quit [Ping timeout: 252 seconds]
p_rossak has joined #linux-sunxi
apritzel has joined #linux-sunxi
<apritzel> slapin: as ssvb said: the AW code (U-Boot, ATF and kernel) needs the SCP
<apritzel> slapin: but I don't really see the need, so I removed it from ATF
<apritzel> slapin: and upstream kernel and upstream U-Boot didn't have it in the first place
<KotCzarny> tkaiser, btw. cable is ok, because when booting android from emmc colors are ok
<apritzel> y0g1: did you update your .dtb?
<apritzel> or was that sun7i-a20-mmc change in 4.9-rc1 only?
p_rossak has quit [Remote host closed the connection]
p_rossak has joined #linux-sunxi
<y0g1> apritzel: yes
<tkaiser> KotCzarny: Hmm... no idea (headless guy). Did you check color range (h3disp -c)? And in case there's a DVI adapter/cable involved then h3disp -d
<KotCzarny> yes, tried all three values
<KotCzarny> luckily i managed to change language of android to english
<KotCzarny> got sidetracked with it now :>
massi has quit [Remote host closed the connection]
vishnup has quit [Ping timeout: 256 seconds]
<KotCzarny> wow, it works
<KotCzarny> playing videos is smooth
vishnup has joined #linux-sunxi
y0g1 has quit [Quit: WeeChat 1.6]
mzki has quit [Ping timeout: 260 seconds]
apritzel has quit [Ping timeout: 252 seconds]
<KotCzarny> it's the first time i see smooth fullhd vids on allwinner
<KotCzarny> and the best part nfs browser worked flawlessly
<tkaiser> KotCzarny: Ever heard of Cedrus?
<KotCzarny> yeah, never had it setup though
<KotCzarny> and tbh, people were bickering at android packaged with opi+2e, but for video play it works, just works
jbrown has quit [Quit: Leaving]
<tkaiser> KotCzarny: well, does the same with Linux from OPi Plus 2E's day one ;)
<KotCzarny> with smooth ui and rest?
<tkaiser> KotCzarny: Whatever, mpv works and that was the first moment I realized that my display has speakers. Scared to death when hearing sound from the movie I tested back in March.
<KotCzarny> hehe
mzki has joined #linux-sunxi
vishnup has quit [Ping timeout: 260 seconds]
<KotCzarny> oth the only thing that works is file manager (but its quite nice, i wonder if i can snatch it for my droid device)
vishnup has joined #linux-sunxi
<KotCzarny> *otoh
afaerber has quit [Quit: Ex-Chat]
station has joined #linux-sunxi
mzki has quit [Ping timeout: 245 seconds]
station has quit [Read error: Connection reset by peer]
station has joined #linux-sunxi
<TheLinuxBug> <KotCzarny> and tbh, people were bickering at android packaged with opi+2e, but for video play it works, just works -- Play full 1080p and watch for about 20-30 minutes until you hit the DRAM issues where it just fully freezes
<TheLinuxBug> also
<TheLinuxBug> KotCzarny: use smplayer
<KotCzarny> yeah, but first impression was positive
<TheLinuxBug> in linux which is a GUI interface for mpv
<KotCzarny> now, how do i turn off the android via keyboard?
<TheLinuxBug> yeah mine was too until I sat watching videos with it for an hour and ran into that issue
<TheLinuxBug> you can't
<TheLinuxBug> you have to push the power button
<TheLinuxBug> as your not root you can't just download a shutdown ap
<TheLinuxBug> app*
<KotCzarny> heh, no power button on the kb
tkaiser has quit [Ping timeout: 245 seconds]
<TheLinuxBug> no I mean physical power button on board
<TheLinuxBug> hold it down for 2-3 second
<TheLinuxBug> will give you the power off dialog
<KotCzarny> there is power button on the board?
<TheLinuxBug> OPi Plus 2e? yes
<KotCzarny> wow
<KotCzarny> the one near power plug worked
vishnup has quit [Ping timeout: 260 seconds]
<TheLinuxBug> ya
<KotCzarny> wow, it even works for turning device back ON
<TheLinuxBug> for how to setup smplayer in Armbian
<KotCzarny> wow2, now my linux shows proper colors
<TheLinuxBug> gives you full gui with hw accel
<TheLinuxBug> and 1080p actually runs very well
<KotCzarny> thx, will probably do it
<TheLinuxBug> hehehe, these are the things I spent time on when I first got that bird
<TheLinuxBug> board*
<KotCzarny> btw. did the hang went away using linux?
<TheLinuxBug> yes
<TheLinuxBug> with Armbian the DRAM issues is corrected
<KotCzarny> good
<KotCzarny> offtopic, 1280x1024 mode works good after patching
<mdsrv> X server lags on bananapi
<mdsrv> on armbian at least
<slapin> is it possible to run VGA on H3?
<KotCzarny> via vga--hdmi adapter
<KotCzarny> which im using now
<KotCzarny> and on 1280x1024 monitor
<slapin> I have unused VGA display, wanna use it
<slapin> yeah, 1280x1024 too
<slapin> older LCD display with strange colors
<slapin> but still no broken pixels
<KotCzarny> tbh 1280x1024 is almost full hd vertically
<KotCzarny> which means you get quite a lot of screen estate and it doesnt take whole desk
<slapin> KotCzarny: well, I use various displays, but slowely tend to get bigger ones, now I use 2 "4K" displays (wtf they call it 4K?) and have 2560x1440 one as spare and have a few VGA-only old displays for toys.
<slapin> KotCzarny: also some apps do work too well with "normal" resolutions than on wide ones
<slapin> anyway, how to fight cpufreq switching off cores on pine64?
vagrantc has joined #linux-sunxi
<slapin> the cpus are immediately shutdown if I enable one with online...
<KotCzarny> add some cooling?
<slapin> KotCzarny: nahm they are cool
<ssvb> slapin: there is a budget cooling table, which describes how many cores can be online and their clock frequency limit at each temperature level
<slapin> ssvb: I have 32C now
<ssvb> slapin: it can be tuned to either prefer killing cores or restricting the clock speed
<slapin> ssvb: in /sys/devices/virtual/thermal/thermal_zone0/temp
mzki has joined #linux-sunxi
<ssvb> yes, because cpufreq has solved the overheating problem by killing cores, and *now* you are cool
<ssvb> bring more cpu cores online, load them with work and you get an overheating SoC again
<slapin> ssvb: well, the load of the device is quite low today, not playing games :) but I recorded temperature in log, remembering orangepi, 38C is max
<slapin> ssvb: which temp it uses as reference for killing cores?
scream has quit [Remote host closed the connection]
<ssvb> which SoC and which kernel are you using now?
<slapin> ssvb: I use pine64 2GB version, stock with android
vagrantc has quit [Ping timeout: 265 seconds]
<ssvb> there is a dtb blob, which you can try to find and decompile
<slapin> ssvb: android is quite modified, but not regarding bootloaders/kernel (which I don't know yet how to build safely to not break anything)
<slapin> ssvb: well, android is mystery to me regarding what is stored where
<slapin> ssvb: especially the one which I did not build myself :(
<ssvb> slapin: anyway, just look for the 'cpu_budget_cool' node in this dtb
<ssvb> slapin: and you might find some useful information here - https://github.com/longsleep/build-pine64-image/pull/3
<slapin> ssvb: I feel very uncomfortable about that, I need to rebuild damn thing so I know I can continue from there...
yann-kaelig has quit [Ping timeout: 244 seconds]
yann-kaelig has joined #linux-sunxi
JohnDoe_71Rus has joined #linux-sunxi
<ssvb> slapin: well, good luck anyway
<slapin> ssvb: do you know where is kernel stored on pine64 android?
<ssvb> maybe the dtb blob is even embedded into U-Boot in the original Allwinner's BSP, I don't remember the exact details anymore
<ssvb> longsleep might possibly help and explain something to you about all of this
IgorPec has quit [Ping timeout: 260 seconds]
jernej has joined #linux-sunxi
sunxi_fan1 has joined #linux-sunxi
gianMOD has joined #linux-sunxi
vagrantc has joined #linux-sunxi
<slapin> ssvb: yeah, it looks like that
mzki has quit [Ping timeout: 256 seconds]
gianMOD has quit [Remote host closed the connection]
whaf has quit [Read error: Connection reset by peer]
gianMOD has joined #linux-sunxi
BenG83 has joined #linux-sunxi
whaf has joined #linux-sunxi
yann-kaelig has quit [Quit: Leaving]
whaf has quit [Ping timeout: 252 seconds]
mossroy has quit [Quit: Leaving]
ptx0 has quit [Ping timeout: 252 seconds]
<longsleep> slapin: No idea about android, but what ssvb said is true, U-Boot has an embedded dtd, its the same which is used by the Kernel in my Linux builds
mosajjal has quit [Ping timeout: 265 seconds]
paulk-aldrin has quit [Remote host closed the connection]
jernej has quit [Ping timeout: 244 seconds]
<slapin> longsleep: have you tried booting android using your u-boot/ATF/kernel?
<longsleep> slapin: no - i do not care about android, but some guy builds android images with at least some of my gear - its all in the pine64 forum somewhere
<longsleep> slapin: also i doub that one can boot android with my modified u-boot
<longsleep> slapin: after all i modified boota and didnt remove those mods after i backported booti
paulk-aldrin has joined #linux-sunxi
<slapin> longsleep: I mean androud using your kernel, which should be compatible with your u-boot, isn't it?
<longsleep> slapin: no idea, lookup the differences between boota and booti (u-boot docs)
Mr__Anderson has joined #linux-sunxi
<ssvb> slapin: maybe try to find other android people in the pine64 forum?
whaf has joined #linux-sunxi
<slapin> ssvb: nah, I'm not big fan of droid stuff, it is just that I have to use it currently on the device to support features I need and to learn what to do to make it less evil
<slapin> ssvb: and nobody there explain about booting stuff, so all I can get I get here
<longsleep> slapin: booting stuff is unrelated to android in any case, its all handled by u-boot (heavily hacked by allwinner)
<slapin> ssvb: currently I try to get more control to find-out what I can do in non-destructive way havin only so much time
<longsleep> what is your goal?
<longsleep> there is no non-destructive way, the allwinner software chain and modifications are beyond repair
leviathanch has quit [Read error: Connection reset by peer]
<slapin> longsleep: my final goal is to have u-boot as E1S, have kvm as E2NS or secure too, run normal Linux as hipervisor and android in vm
<slapin> longsleep: I mean to have it still working :)
tllim has joined #linux-sunxi
<slapin> longsleep: so no ATF, simple clean boot chain, and control over system
<longsleep> tllim: just the one who i need, the patch you sent me for the gmac, do you know what version of the gmac driver is it based on?
tllim has quit [Ping timeout: 260 seconds]
<longsleep> :-/
tllim has joined #linux-sunxi
<tllim> longsleep: I don't have the quick answer, I will check with engineer and PM you
f0xx has quit [Ping timeout: 250 seconds]
tllim has quit [Ping timeout: 260 seconds]
gianMOD has quit [Remote host closed the connection]
|Jeroen| has quit [Quit: dada]
paulk-aldrin has quit [Remote host closed the connection]
whaf1 has joined #linux-sunxi
whaf has quit [Ping timeout: 250 seconds]
JohnDoe_71Rus has quit [Quit: KVIrc KVIrc Aria 4.9.2, revision: git-7004-g1ef3cc7, build type: debug, sources date: 20160102, built on: 2016-10-28 16:02:21 UTC git-7004-g1ef3cc7 http://www.kvirc.net/]
reinforce has quit [Quit: Leaving.]
gianMOD has joined #linux-sunxi
gianMOD has quit [Remote host closed the connection]
Mr__Anderson has quit [Ping timeout: 252 seconds]
afaerber has joined #linux-sunxi
apritzel has joined #linux-sunxi
vagrantc has quit [Quit: leaving]
Putti has quit [Ping timeout: 260 seconds]
netlynx has quit [Quit: Ex-Chat]
ptx0 has joined #linux-sunxi
Nyuutwo has quit [Ping timeout: 245 seconds]
Nyuutwo has joined #linux-sunxi
p_rossak has quit [Ping timeout: 260 seconds]
p_rossak has joined #linux-sunxi
jstein has quit [Read error: Connection reset by peer]
BenG83 has quit [Ping timeout: 260 seconds]
apritzel has quit [Ping timeout: 252 seconds]
cptG_ has joined #linux-sunxi