alexxei has quit [Read error: Connection reset by peer]
perr has quit [Quit: Leaving]
alexxei has joined #linux-sunxi
PietreLinux has joined #linux-sunxi
apritzel has quit [Quit: Leaving.]
Colani1200 has joined #linux-sunxi
Colani1210 has quit [Ping timeout: 245 seconds]
<wens>
ssvb: i remember the 'ondemand' governor didn't work so well with some generation of amd cpus
<wens>
ondemand really only works when the cpu clock rate can be ramped up really quick
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
ninolein has quit [Ping timeout: 245 seconds]
ninolein has joined #linux-sunxi
<PietreLinux>
I want to communicate that sunxi-spain is already underway, we are translating sunxi.org full in Spanish, I have people who will update based on the original sunxi. a cordial greeting
<PietreLinux>
Of course this website will be hosted on a public site funded by my
arossdotme has joined #linux-sunxi
vagrantc has quit [Quit: leaving]
cnxsoft has joined #linux-sunxi
terra854 has quit [Quit: Connection closed for inactivity]
PietreLinux has quit [Quit: Page closed]
VargaD has quit [Ping timeout: 240 seconds]
VargaD has joined #linux-sunxi
<TheLinuxBug>
KotCzarny: not sure why you guys crap on Android on Allwinner so much, it is quite possible to run a stable Android install on A20 and A10, have done it on both. A10 handles SD fine, A20 can do 720p, just don't expect 1080p on either and you'll be OK. A nice SDcard and it works pretty nice. Also, past the poorly made Android image from Allwinner SDK on OPi2+ which blocks root and uses shit DRAM settings, Android runs pretty stable on it.
<TheLinuxBug>
Orange Pi PC also runs it pretty decently from SDcard and it actually has some better working versions out there which give good DRAM settings and allow 1080p video
pg12 has quit [Ping timeout: 240 seconds]
<TheLinuxBug>
now if you want to play a bunch of games, not sure about that, that wasn't really my use case when testing.
pg12 has joined #linux-sunxi
<TheLinuxBug>
Played youtube fine though in all cases
<TheLinuxBug>
One thing though worth noting is you don't really get to choose your version and mosty don't have new releases out like Android 6 from my review.
<TheLinuxBug>
So you have to be okay with older version like 4.2.2 or 4.4.4
<TheLinuxBug>
interesting about PietreLinux as I thought it was communicated that it wasn't wanted for linux-sunxi to be in any other language than English... but yet someone is translating it into Spanish... eesh
IgorPec has joined #linux-sunxi
whaf has quit [Remote host closed the connection]
whaf has joined #linux-sunxi
[7] has quit [Ping timeout: 240 seconds]
TheSeven has joined #linux-sunxi
JohnDoe_71Rus has joined #linux-sunxi
dlan has quit [Ping timeout: 260 seconds]
whaf has quit [Quit: Leaving.]
dlan has joined #linux-sunxi
reinforce has joined #linux-sunxi
merbanan has quit [Ping timeout: 260 seconds]
premoboss has joined #linux-sunxi
iamfrankenstein has joined #linux-sunxi
merbanan has joined #linux-sunxi
scream has quit [Remote host closed the connection]
terra854 has joined #linux-sunxi
futabachannel has quit [Quit: Leaving...]
stk is now known as speculae
perr has joined #linux-sunxi
perr has joined #linux-sunxi
perr has quit [Changing host]
bugzc_ has joined #linux-sunxi
bugzc has quit [Ping timeout: 260 seconds]
gzamboni has joined #linux-sunxi
f0xx has joined #linux-sunxi
IgorPec has quit [Ping timeout: 260 seconds]
leviathanch has joined #linux-sunxi
speculae is now known as stk
IgorPec has joined #linux-sunxi
yann has quit [Ping timeout: 260 seconds]
medvid has quit [Ping timeout: 250 seconds]
sunxi_fan has quit [Quit: Leaving.]
sunxi_fan has joined #linux-sunxi
premoboss has quit [Ping timeout: 260 seconds]
medvid has joined #linux-sunxi
codekipper has joined #linux-sunxi
perr has quit [Quit: Leaving]
<codekipper>
wens: my OPi2 is now singing thanks to you.
<apritzel>
it contains a libdram based DRAM init for now
<jelle>
I hope my OPi PC 2 arrives soon :)
<apritzel>
MoeIcenowy: ssvb: if you can give it a try, works best via FEL at the moment
<apritzel>
jelle: ah, another tester ;-)
<jelle>
apritzel: sure
Putti has quit [Ping timeout: 256 seconds]
<apritzel>
H5 users: pls let me (or the logs ;-) know how this works...
<jelle>
apritzel: doubt I have it soon, ordered 10 days ago
<zoobab>
H2 ordered yesterday
<apritzel>
for using it one has to build a 32-bit SPL with sun50i_h5_spl32_defconfig and then the 64-bit U-Boot proper with orangepi_pc2_defconfig, and put the bits into FEL together with ATF
<jelle>
hmm sounds complicated, but I bought it to try out 64 bit arm
<apritzel>
wens: I made a change in there which affects sun9i as well (CONFIG_SUNXI_HIGH_SRAM), so if you could test it I'd be grateful
<apritzel>
tty l8er
apritzel has quit [Quit: Leaving.]
Ctk is now known as stk
whaf has joined #linux-sunxi
<wens>
codekipper: didn't you say it wouldn't boot anymore?
<wens>
oh great... one of my cpus on the a80 is booted in secure mode :/
Uninstall has quit [Remote host closed the connection]
Uninstall has joined #linux-sunxi
<codekipper>
wens: I did and I'm sure I tried again a couple of days later....tried with the armbian image and it booted. Then had another go with my netboot card and it worked. I'll push the patch when I see your stuff go in.
<codekipper>
I had a look at i2s and spdif on the H3 yesterday...there is something up with the clocking...a 4 min track completes in 2secs..is there any patches that I've missed(currently sat ontop of your 23-h3-audio branch)
<wens>
not that i know of
<codekipper>
is it something that you can reproduce?
<wens>
uh i don't have spdif codecs
<codekipper>
you don't need to have it connected to anything.....
<wens>
oh right
<wens>
just play, huh?
<codekipper>
just to reproduce the issue.
<wens>
i dont think i have your repository configured
<codekipper>
I can through together some patches...
<codekipper>
I've not uploaded anything for a while....which board do you have?
<codekipper>
better check that the pins isn't been used for anything important
<wens>
opi pc
<codekipper>
OK...I'll ping you some patches in a mo
massi has joined #linux-sunxi
yann has joined #linux-sunxi
Leepty has joined #linux-sunxi
perr has joined #linux-sunxi
Mr__Anderson has joined #linux-sunxi
alexxei has quit [Ping timeout: 258 seconds]
Putti has joined #linux-sunxi
Mr__Anderson has quit [Quit: Leaving.]
premoboss has joined #linux-sunxi
paulk-minnie has joined #linux-sunxi
premoboss has quit [Ping timeout: 268 seconds]
<KotCzarny>
TheLinuxBug: i have only tried (by simple mistake of failing to boot sdcard ;) android bundled with opi+2e. the only thing that was working was file browser (which is quite nice actually) and video player. all other aps either crashed or were unavailable. im not talking about self-built android, im talking what is bundled/provided by vendors ;)
<KotCzarny>
TheLinuxBug: and as im a complete noob when it comes to build android from scratch, i would yet have to see fool-proof step by step guide how to create it
dh1tw has joined #linux-sunxi
stk is now known as stk_py
BigFellow has joined #linux-sunxi
matthias_bgg has joined #linux-sunxi
florianH has joined #linux-sunxi
<montjoie>
strange A64 is getting DMA timeout for PRNG/SHA384/SHA512
Putti has quit [Ping timeout: 260 seconds]
fkluknav has joined #linux-sunxi
perr has quit [Quit: Leaving]
premoboss has joined #linux-sunxi
Andy-D__ has quit [Ping timeout: 246 seconds]
lemonzest has joined #linux-sunxi
alexxei has joined #linux-sunxi
perr has joined #linux-sunxi
andi_sp has joined #linux-sunxi
andi_sp has quit [Client Quit]
premoboss has quit [Ping timeout: 240 seconds]
lastebil has quit [Ping timeout: 244 seconds]
f0xx has quit [Ping timeout: 246 seconds]
reinforce has quit [Quit: Leaving.]
fvogt has joined #linux-sunxi
tkaiser has joined #linux-sunxi
<tkaiser>
KotCzarny: There you go: http://filez.zoobab.com/allwinner/h2/201609022/ -- do a find for sys_config.fex to adopt sane throttling/dvfs settings and don't forget to upgrade kernel to 3.4.113 before! ;)
mzki has joined #linux-sunxi
Keziolio has quit [Remote host closed the connection]
dr1337__ has joined #linux-sunxi
<dr1337_>
Does anyone know how I can apply snd_soc_dapm_force_enable_pin to a line out pin for the a23 codec?
lkcl has quit [Ping timeout: 246 seconds]
lkcl has joined #linux-sunxi
yann-kaelig has joined #linux-sunxi
paulk-minnie has quit [Ping timeout: 248 seconds]
bugzc_ has quit [Read error: Connection reset by peer]
bugzc has joined #linux-sunxi
paulk-collins has joined #linux-sunxi
<agraf>
ssvb: do you know if there's a nand reading fel tool?
avph has quit [Ping timeout: 260 seconds]
avph has joined #linux-sunxi
leviathanch has quit [Remote host closed the connection]
Andy-D has joined #linux-sunxi
petr has quit [Remote host closed the connection]
ksyz has joined #linux-sunxi
alexxei has quit [Ping timeout: 256 seconds]
whaf has quit [Quit: Leaving.]
paulk-minnie has joined #linux-sunxi
petr has joined #linux-sunxi
Putti has joined #linux-sunxi
<KotCzarny>
tkaiser: thx, will look at it, though i wanted some guide too ;)
<montjoie>
lol found why sha384/512 doesnt work on A64, simply it do not support it
<KotCzarny>
montjoie: is it advertised anywhere that it is?
<montjoie>
no
<montjoie>
just it is the same Crypto Engine
<montjoie>
dont know why they removed it
<KotCzarny>
maybe it was buggy
<KotCzarny>
or maybe they needed silicon space
<montjoie>
now I need to find why PRNG do not work
reinforce has joined #linux-sunxi
perr has quit [Quit: Leaving]
<montjoie>
KotCzarny: A64 have HMAC-sha256, so it was not for gaining space
<KotCzarny>
tkaiser: quick question, is that h2.tar.gz just a packed version of those android/lichee dirs?
<KotCzarny>
dont know if i have to download everything
zzeroo has joined #linux-sunxi
f0xx has joined #linux-sunxi
bugzc has quit [Ping timeout: 260 seconds]
<IgorPec>
kotczarny: nothing useful there
<KotCzarny>
igorpec: i have mild interest in working android, preferably with easy guide how to build it
<IgorPec>
ahaa, Android ... no idea
dr1337__ has quit [Remote host closed the connection]
reinforce has quit [Quit: Leaving.]
Worf has joined #linux-sunxi
<tkaiser>
KozCzarny: The archive contains everything. Lost my local copy of both H2/H5 'SDKs' yesterday when doing UASP test ;)
<KotCzarny>
lol
<KotCzarny>
which chip/what happened?
reinforce has joined #linux-sunxi
<tkaiser>
KotCzarny: Only to confirm H2+ is UASP capable. Took the SSD lying next to me, mkfs.btrfs complained about FS already existing, used -f first and a second later: 'Ah, the BSP stuff is on this disk'
<KotCzarny>
:)
<KotCzarny>
not a great loss apparently
<tkaiser>
Nope, only interesting piece in H2 SDK is xradio_wlan driver.
<tkaiser>
agraf: Is SuSE hacking Nintendo? ;)
<agraf>
tkaiser: no, just a few friends ;) not everything i do is suse related :)
afaerber has quit [Quit: Ex-Chat]
plm has joined #linux-sunxi
<plm>
Hi all
BigFellow has quit [Quit: BigFellow]
alexxy has quit [Ping timeout: 244 seconds]
myy has joined #linux-sunxi
premoboss has joined #linux-sunxi
<KotCzarny>
zoobab: that h2.tar.gz is really tar.bz2
<tkaiser>
KotCzarny: You know that there exists a project for A64 to bring Android 7.0 to Pine64? Is based on community efforts, eg. longsleep's BSP kernel fork: https://github.com/ayufan-pine64/
JohnDoe_71Rus has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org]
<KotCzarny>
tkaiser: i dont own pin64, only h3 based devices
<tkaiser>
KotCzarny: Then good luck with Android 4.4 -- that should serve as a hint regarding H5 soon ;)
<longsleep>
montjoie: did you make some speed tests of the A64 crypto stuff? I disabled their kernel drivers for my Pine64 tree if i remember correctly
<KotCzarny>
tkaiser, noob question, simply replacing kernel/uboot would make that pine64 android run on h3 ?
<montjoie>
longsleep: last bench said perf/10
<tkaiser>
KotCzarny: I doubt it, but to be honest: Zero interest in Android and almost zero clue as well
<montjoie>
longsleep: with generic/arm linux implementation
<tkaiser>
KotCzarny: But if A64 and H5 are pretty close then OPi 3 might be an little bit more interesting device for Android experiments (OPi 3 Plus with eMMC to be more precise)
whaf has joined #linux-sunxi
alexxy has joined #linux-sunxi
Antitrack2 has joined #linux-sunxi
<Antitrack2>
has anyone successfully installed the bpi-m3 so it boots from sata instead of from sd-card?
<tkaiser>
Antitrack2: Me, but it's a really bad idea since this is not SATA but ultra slow USB
<KotCzarny>
but it has SATA written on it, so it must be goood
<KotCzarny>
so the question is, 'can any allwinner board boot from usb'
<Antitrack2>
tkaiser: I have to do it, my boss forces me to do it since he ruined sdcards compiling freepascal code on the sd card, so we both know its bad, but I neeeeeed to do it - help plz
<KotCzarny>
Antitrack2: you still have to boot uboot from any supported device (sdcard)
<Antitrack2>
I have already read all the forum stuff how baaaad the usb2sata bridge is on the bpi-m3 , but .....
<Antitrack2>
I know
<tkaiser>
Antitrack2: If you use any of the not so crappy OS images from SinoVoip you can adjust the path to the rootfs in uEnv.txt
<Antitrack2>
I d love to fix the proper files on the sdcard so the rest of the booting goes to "pseudo-sata", but all the docs on the net did not work
<KotCzarny>
so the only possible way is to install uboot/kernel on sdcard (and nothing else) and make it boot to 'sata'
<Antitrack2>
ok plz tell me which image to use
<Antitrack2>
the img i have does not have anything in the uEnv.txt
<KotCzarny>
also, who in the right mind uses pascal today
<Antitrack2>
except the hw addr of eth
<tkaiser>
Antitrack2: LOL, are cou kidding? They release 'beta images, only for test' since a year or so.
<Antitrack2>
my boss loves pascal :D not me
<Antitrack2>
Im not kidding
<KotCzarny>
i know, the older we get, the harder to learn new tricks
<Antitrack2>
Linux bpi-iot-ros-ai 3.4.39-BPI-M3-Kernel #1 SMP PREEMPT Tue May 3 13:47:01 UTC 2016 armv7l armv7l armv7l GNU/Linux
<KotCzarny>
ok, first things first, do you know anything about uboot and/or linux?
<Antitrack2>
a lot about linux but that uboot gives me a headache, I modded all the files, copied everything to sata, edited the /etc/fstab, etc.etc. and it still boots from mmc....
<tkaiser>
Antitrack2: For the newer ones somewhere in this famous bananapi.org forum, just ask aunt google: sata uenv.txt /dev/sda1 site:forum.banana-pi.org
<KotCzarny>
you were apparently already having the thing working from sdcard, so my advice is: mount /home from usb, and leave os as it is
<Antitrack2>
hmmm
<Antitrack2>
but my boss doesnt want to
<jelle>
seems that the sunxi wiki contains a lot more info the amlogic's mainline effort
<KotCzarny>
he doesnt have to know. you can say its the only way
<KotCzarny>
;)
<Antitrack2>
"grab the bsp" -> bsp ?
<Antitrack2>
I am at that page now
<KotCzarny>
think 'sdk'
<tkaiser>
Antitrack2: There's eMMC on this board, move the rootfs to eMMC and put the home on pseudo-SATA. And if your boss complains then... grab a new boss
<Antitrack2>
not so easy. what boss says I have to do. Yes, stupid. But what can I do.
<Antitrack2>
grab the BSP and adjust sunxi-pack/chips/sun8iw6p1/configs/*/env.cfg -> BSP? where is this file
<KotCzarny>
tkaiser, can bpi-m3 use mainline uboot simply?
<KotCzarny>
or its not yet supported?
<Antitrack2>
what IS this "uboot"
<KotCzarny>
Antitrack2: think of it as a bios+grub
<Antitrack2>
ok
<tkaiser>
KotCzarny: Sure, but that's not necessary, works even with old crappy u-boot 2011.09 since those SinoVoip people learned that u-boot knows boot scripts in the meantime.
<Antitrack2>
so where to find this "BSP" and what is it
<KotCzarny>
tkaiser: but its easier and cleaner to just install new uboot instead of getting all dirty touching bsp uboot
<tkaiser>
Antitrack2: It's linked from the wiki page. Time to stop here.
<tkaiser>
KotCzarny: I don't care. This device is crap anyway. If you do 2 seconds research you will never buy it.
<tkaiser>
It's just an entry in uEnv.txt so what the heck?!
<KotCzarny>
i know. been there. i wanted to buy it because of specs. then seen linux-sunxi. then went with bpi-r1 ;)
<KotCzarny>
generally allows you to create whole image for particular device
<Antitrack2>
thxx
<Antitrack2>
that looks helpful
<Antitrack2>
ok
cnxsoft has quit [Ping timeout: 244 seconds]
cnxsoft1 is now known as cnxsoft
<Antitrack2>
the link to kaiser edv goes 403
<KotCzarny>
and this is where it gets messy, as vendor bsps in the world of allwinner are very badly coded
<Antitrack2>
you may wanna fix that
<Antitrack2>
ok
<KotCzarny>
and linux-sunxi is about saying 'f*ck the legacy, broken stuff and rewrite it properly'
<KotCzarny>
unfortunatelly a80/a83t have the poorest support, almost close to nothing
<tkaiser>
Antitrack2: thanks for the reminder, will now remove the whole directory :)
<Antitrack2>
ok
matthias_bgg has joined #linux-sunxi
<Antitrack2>
my boss will be pissed having bought several bpi-m3
<Antitrack2>
:(
<KotCzarny>
Antitrack2: i pity the fool
<Antitrack2>
me too .... :)))
<tkaiser>
Antitrack2: See above, it's uEnv.txt. Slow down, read carefully, adjust stuff and then realize that this device is crap anyway :)
<Antitrack2>
he has phd (dipl.ing dr.techn) fooled by M3 :D
<KotCzarny>
being smart in one field doesnt make you life smart
<Antitrack2>
the native uEnv.txt does not contain anything but the hw address of eth0
<KotCzarny>
which in this case would require finding board support state first and possible gotchas
<KotCzarny>
Antitrack2: you can add things to uenv
<Antitrack2>
how ?
<KotCzarny>
text editor?
<tkaiser>
Antitrack2: Slow down, read carefully, they described somewhere in their fabulous forum where these files reside. They're not there where you expect them
<Antitrack2>
ok
<Antitrack2>
but first i need bsp anyway. right ?
<tkaiser>
Nope. Over and out :)
<KotCzarny>
but anyway, you want this: boot from (sdcard or emmc?) and run from 'sata' ?
<nove>
well, some people equal "asking for the license violations to be resolved" to mean "legal enforcement treats" when this is only possible to do by the copyright holders
popolon has joined #linux-sunxi
* nove
is happy to not have copyright in the matter
reinforce has quit [Quit: Leaving.]
Worf has quit [Quit: Konversation terminated!]
<wens>
nove: you mean offended?
<oliv3r>
hi!
<KotCzarny>
nove, but more seriously, if it's only up to good will, only argument would be people caring and not buying violator's hw
<nove>
wens: thanks, yes i meand offended
<KotCzarny>
sad state of things is that people dont care as long it works. somehow.
<ksyz>
KotCzarny if they did care, we would not have rpi.
<tkaiser>
longsleep: Yes. And those guys doing software stuff also limited max cpufreq to 1008 MHz here :)
<longsleep>
tkaiser: well, the log from KotCzarny does look ok to me, so tuning the details and patching up the Kernel to .104 should be a nobrainer
<terra854>
tkaiser: What is the max frequency the H5 in the OPi Pc 2 can reach?
gumblex has quit [Remote host closed the connection]
<longsleep>
the only thing i could complain about is that they boot with bootm and not with booti, which requires the kernel to be wrapped, but thats details
<longsleep>
didnt check if their u-boot tree already has booti support
<tkaiser>
terra854: According to the SDK 1200 MHz, I guess that's just Allwinner getting more cautios. In this image they limit it to 1008 MHz. I just check the dvfs operating points and if that looks good will reboot with 1200 MHz max :)
<tkaiser>
Ah, no dvfs avaliable?
<longsleep>
tkaiser: well you have the table in the device tree you posted
<longsleep>
they even put the device tree in a subfolder, nice
<longsleep>
looks like progress to me :)
<tkaiser>
longsleep: Found it.
<tkaiser>
longsleep: And think I found also the reason for 1008 MHz max. All dvfs operating points use 1.1V so they didn't figure out how voltage regulation works without PMIC? :)
<longsleep>
tkaiser: seems like all your messages to them had them look at some of the stuff we did for Pine64
<KotCzarny>
so, its a pine64 kernel running on opipc2 ?
<longsleep>
KotCzarny: no its their Kernel, no idea how close it is to the Pine64 one but the base linux version is the same (BSP)
<tkaiser>
longsleep: I never had direct communication with Stephen/Xunlong. I posted stuff in a forum and months later got an reply via aliexpress for example. But really looks like progress :)
JohnDoe_71Rus has joined #linux-sunxi
<tkaiser>
But currently on all their new OS images voltage regulation seems to not work properly. The H2+ images define settings wrong so H2+ runs with 1.3V all the time, here they defined only 1.1V
IgorPec has joined #linux-sunxi
<tkaiser>
Hehe, ready to destroy the device: state0 = <0x124F80 0x4>; (at least a quick undervoltage test :) )
matthias_bgg has joined #linux-sunxi
<terra854>
1.2 Mhz???
<tkaiser>
terra854: Allwinner style, sometimes you have to add 000 to values, sometimes not. And it doesn't work since this setting has to be set already in u-boot. Still at 1008 MHz
<longsleep>
if you ask me then there is no point to even support hdmi on a board with only 1GB or RAM
<tkaiser>
longsleep: I spotted an 'axpdummy' driver in the H5 BSP. Maybe this is supposed to do the translation?
<terra854>
" A new and more permissive version was introduced in January 2016 and all Midgard user-space drivers starting with r6p0 are now distributed under the new terms. The main changes are to allow redistribution of the binaries under the same EULA, commercial use and benchmarking."
<tkaiser>
longsleep: OPi PC 2 is here as an excercise. OPi 3 with 2 GB DRAM is already scheduled ;)
<terra854>
So in simple terms, we can redistribute them?
<longsleep>
terra854: if you had them in the first place und this licesen yes
<tkaiser>
terra854: Please ask Lenny or the old stubborn beardy man over in pine64 forum.
<longsleep>
tkaiser: 3GB would be better :P
<rellla>
terra854: seems so, yes.
<terra854>
tkaiser: Lenny is AWOL for 2 months now:(
<tkaiser>
terra854: And the Mali hype is still there?
<terra854>
No. In fact, i don't think anyone is talking about it
<tkaiser>
Nice, a shutdown sends the board in 'suspend to RAM' state. It reboots to u-boot and waits for power key or IR
<terra854>
Not sure if the Pine is capable of that
<tkaiser>
But pin mappings obviously don't match :)
<tkaiser>
terra854: It is
<tkaiser>
But most likely longsleep was busy with an axe in his hand when starting to deal with u-boot back then ;)
premoboss has quit [Ping timeout: 240 seconds]
<terra854>
Well, if that is implemented, that will be plus
<terra854>
* a plus
<tkaiser>
Also nice, same wrong colors as with Pine64 when using a DVI display and same wrong led definition as with H3 (red led as 'powered on' sign and green as 'off').
<longsleep>
well, its the same BSP base i would say
<tkaiser>
yep
<longsleep>
so same issues in the display driver and all
<tkaiser>
yep ;)
<terra854>
3.10.65?
whaf has quit [Quit: Leaving.]
<tkaiser>
terra854: What else, there's only 3.4.39 or 3.10.65 in Allwinner land
<longsleep>
terra854: that is just a version and has no real meaning
<tkaiser>
IIRC those were some Linaro kernel LTS versions?
<longsleep>
terra854: there can be any number of different patches in whatever bsp download all named 3.10.65
<longsleep>
tkaiser: yes, the pine64 bsp one is based on the linaro android tree
<longsleep>
and the hdmi/disp2 driver needs the android patches because it uses ion
ErwinH has joined #linux-sunxi
<terra854>
Well, the link to the OPi PC 2 SDK is dead
<tkaiser>
longsleep: BTW: My DVI display shows black as purple (as expected) but no contents. So maybe really everything the same as back then?
<longsleep>
tkaiser: no contents? what was that again?
<ErwinH>
Hi there! I had a look at the schematics of the OPi Zero and it looks like the signal voltage for the SD-card is 1.8V meaning the performance of the SD-cards could become a lot better then the previous versions, up to 104MB/s...
<beeble>
tkaiser: sounds like it doesent use RGB444
<beeble>
tkaiser: 3.4/3.10?
<beeble>
if yes, then the edid parsing is broken
<tkaiser>
longsleep: Don't remember, I connected a display only twice when you sat in beergarden and I was on my way to check the DVI fix
<tkaiser>
beeble: Yeah, the same with BSP kernel for H3, A83T, A64 and now H5 :)
<longsleep>
tkaiser: dvi fix was using the drm driver - you mean that one?
<longsleep>
tkaiser: do they use the drm driver in the h5 bsp?
<beeble>
so forcing RGB444 output for dvi output should be fine
* terra854
is wondering what the hell is Allwinner doing using outdated linux kernels...
<tkaiser>
longsleep: Nope, the hdmi_cts_compatibility = 1 and hdcp_enable = 0 stuff
<longsleep>
tkaiser: ah
<tkaiser>
beeble: this is exactly what this setting does
<KotCzarny>
but that only fixes lack of signal
<KotCzarny>
purple is probably longsleep's fix
<tkaiser>
KotCzarny: Nope, that fixes colors
<terra854>
They could have at least used a 4.x kernel for the BSP
<tkaiser>
You can also fix it at the display if that is capable to accept different input color spaces (mine does)
<beeble>
have to find our patches for a31, same issue there with 3.4
<tkaiser>
beeble: A31 was 3.3?
<beeble>
original yes, we rebased a lot to 3.4 for android 5.1.1
<tkaiser>
beeble: ah, ok.
<longsleep>
is there any asian arm SoC vendor with a reasonably recent Kernel?
<terra854>
Samsung... i guess
<tkaiser>
terra854: The last Samsung/Nexell SoC I played with was on 3.4.39 ;)
<terra854>
Exynos you mean?
<montjoie>
my phone got a 3.4.1:)
<tkaiser>
terra854: The SoCs NanoPi M2/M3 use, can't remember the name
<longsleep>
well galaxy s7 has 3.18 afaik
<longsleep>
i mean look at qualcom and all their crappy out of tree drivers for ancient kernels
<terra854>
tkaiser; S5P4418
<longsleep>
google just paid $120K for a exploit on their Pixel phone
<terra854>
Qualcomm Snapdragon 821
<zoobab>
@KotCzarny thanks fixing it
<longsleep>
terra854: Pixel phone also has 3.18
<terra854>
Just wondering though, what is the difference between the Linaro kernel and the normal Linux kernel?
tkaiser has quit [Ping timeout: 256 seconds]
<longsleep>
terra854: different patches / additional patches - there is no "normal" Kernel. There are trees and all are normal - they are managed by different people. You probably mean with "normal" the tree from Linus Torvalds
tkaiser has joined #linux-sunxi
leviathanch has joined #linux-sunxi
<terra854>
Yeah. I was wondering about what is so special about the Linaro kernel
<longsleep>
tkaiser: ok, then either the defaults have changed or gcc 4.9 is just shit for arm64
<KotCzarny>
i wonder if just recompilation of iperf3 could yield different results
<KotCzarny>
or even trying to run iperf3 bianry from xenial on jessie
<KotCzarny>
*binary
<tkaiser>
KotCzarny: It's really just that iperf is CPU bound on Jessie while not on Xenial. Since it uses only one CPU core in one direction. But it's only important to take notice to be able to interpret results.
<KotCzarny>
or switching kernels. just to pinpoint which is the bottleneck
<tkaiser>
BS, we don't want to improve benchmark scores but performance.
<tkaiser>
And that's most probably adjusting TX delay
jernej has joined #linux-sunxi
<longsleep>
mhm, if you use only one cpu core you think it could be faster than the 380 Mbits/sec you got?
KB3VGW has joined #linux-sunxi
<tkaiser>
longsleep: LOL, just spotted in htop: /usr/local/sbin/OrangePi_corekeeper.sh
<longsleep>
tkaiser: oh yeah so they copied it all
<longsleep>
they had better copied the in kernel core keeper from armbian
fkluknav has joined #linux-sunxi
<longsleep>
but better than nothing (if it works) :)
fvogt has quit [Quit: Konversation terminated!]
<tkaiser>
longsleep: I repeated the test and let it run for 120 seconds: Now 460 MBits/sec on average. So the first test was too short and maybe something with IRQ distribution went wrong (that was something I observed with Pine64 as well. When iperf3 sat on the same CPU core than the IRQ handler performance degraded).
<tkaiser>
s/IRQ/Ethernet IRQ/
<KotCzarny>
move gmac to different core then?
<longsleep>
tkaiser: so single core 460 MBits/sec thats not bad
<tkaiser>
I would say this image is already in a nice state. DVFS and GbE TX delay being the two things that need some attention.
<tkaiser>
Ahh, I forgot 'the Mali'!!1!!11!!
<longsleep>
yes agreed, probably the best image they ever released
<longsleep>
tkaiser: do they have the a tune script in /usr/local/sbin as well?
<nove>
with this differences between marketing material and the datasheets nothing should be trusted, the marking also claims vp9 which is not in the datasheet
<nove>
and could be the same case of A80 which the support was by software
<KotCzarny>
unless you are after number crunching, then ibm power or x86
<beeble>
plm: ls1046a
premoboss has quit [Ping timeout: 256 seconds]
<zoobab>
@KB3VGW you talked to wigyori?
<KB3VGW>
yes
<KB3VGW>
?
<zoobab>
if you want openwrt on there
<plm>
beeble: sorry, the "ls1046a"?
<zoobab>
with which kernel version?
<KB3VGW>
yes he said he would work on it
<KB3VGW>
he said he would try 4.8 first
<KB3VGW>
I have to grab i2c-tools and make the changes to twi-tools
<beeble>
plm: google first hit. but freescale^Wnxp^Wqualcomm
<KB3VGW>
so we have a tool for the twi interface
<beeble>
why do you need a twi tool?
<beeble>
what do you want to do that is not possible with the i2c tool?
<KB3VGW>
i2s-tools currently does not see the twi intrerface
<KB3VGW>
so it needs the mod
<KB3VGW>
I will double check but I just loaded up the driver and i2ctools
<KB3VGW>
and its not seeing the twi interfaces
<beeble>
there is somethign else wrong
<KB3VGW>
ok
<beeble>
the drivers will bind as i2c device
<beeble>
and you can use the i2c-tools
<KB3VGW>
ok
<KB3VGW>
its the 4.7 kernel and armbian
<beeble>
does your dts contain the twi entries?
<KB3VGW>
checking that next
cptG has joined #linux-sunxi
BigFellow has quit [Quit: BigFellow]
cptG_ has quit [Ping timeout: 268 seconds]
iamfrankenstein has quit [Ping timeout: 246 seconds]
<KB3VGW>
ok I have to recall how to decompile the dtb
iamfrankenstein has joined #linux-sunxi
<Tartarus>
fdtdump
netlynx has joined #linux-sunxi
netlynx has quit [Changing host]
netlynx has joined #linux-sunxi
<KB3VGW>
nope it seems very basic
bugzc has joined #linux-sunxi
<tkaiser>
dtc
<KB3VGW>
dtb
<beeble>
dtc -I dtb -O dts <dtbfile>
<KotCzarny>
there is a tool to compile/decompile in kernel scripts dir
<tkaiser>
It's part of 'device-tree-compiler' and that's default in Armbian
<plm>
beeble: yes, I think that the ls1046a is better than phytium :)
<beeble>
plm: better depends on the usecase, as always
paulk-minnie has quit [Ping timeout: 265 seconds]
<KB3VGW>
there is a bug in the pine64 armbian
<KB3VGW>
I ran that and ssh crashed
<KotCzarny>
lol
<tkaiser>
KB3VGW: Honestly, this version is a preview based on an outdated kernel tree from MoeIcenowy's repo a few months back. And Armbian also is not a distro but a build system. So you have either Debian or Ubuntu packages.
<plm>
beeble: has 10GbE ports :)
<KB3VGW>
4.7 kernel is still using i2s
<KB3VGW>
in the dtb
<KB3VGW>
I ment i2c
<KB3VGW>
sorry
<KB3VGW>
so the armbian vanilla kernel is got issues at 4.7
<KB3VGW>
it has no lib/modules/4.7-*/kernel/driver/i2c driver
<KB3VGW>
ok
<KotCzarny>
KB3VGW: read what tkaiser said, it was just experimental image
<tkaiser>
KB3VGW: No idea, if you want to use this image for anything other then testing and reporting back (useless in the meantime) then you're doing something wrong
<tkaiser>
Not even USB is working there.
<KB3VGW>
just testing it
<KB3VGW>
ok
<KB3VGW>
4.9 is more workign
<KB3VGW>
just testing to see whats working where so we can get openwrt going asap
<beeble>
plm: but don't expect it to be cheap
<KB3VGW>
this is debian
<tkaiser>
KB3VGW: Just for your info: I just wiped out Debian from an SD card and am running now Xenial on Orange Pi PC 2 ;)
<KB3VGW>
Xenial?
<tkaiser>
Ubuntu 16.04
<KB3VGW>
ok
<KB3VGW>
I did not se a server ver of xenial
<KB3VGW>
we need a server ver
<plm>
beeble: do you think that phytium is very cheaper than ls1046a?
<KB3VGW>
and that means I have to learn how they pkg compaird to how normal debianpkgs pkg
paulk-minnie has joined #linux-sunxi
<KotCzarny>
doesnt openwrt have sdk/buildroot ?
<KB3VGW>
we have been doing debian for so long
<KB3VGW>
I am just getting started in openwrt
<KB3VGW>
wigori is helping to get te board we have going
<KB3VGW>
its not even in the supported boards yet
<KotCzarny>
because really even having working kernel is wip
<KB3VGW>
well I was tols to test 4.9 and the pine64
<beeble>
plm: if you compare performance for price, i think it will be a lot cheaper
jstein_ has joined #linux-sunxi
<beeble>
plm: also phytium is a number crunching cpu. ls1046 is a network cpu
jstein_ is now known as jstein
enrico__ has quit [Quit: Bye]
<plm>
beeble: Yes, I think that I will go of phytium :) The problem with phytium (like as ohter chinese company) is a poor documentation compared now with NXP ls1046a. In the site of phytium has just the very basic
<beeble>
plm: you can also try a armada 8k
<beeble>
they have good documantion, but you have to convince them to sell you the part :)
<beeble>
but same goes with phytium
<plm>
beeble: why marvell will not sells armada 8k to me?
<tkaiser>
longsleep: In case you try to use a Pine64 rootfs on OPi PC 2: Loading 8723bs panics :)
premoboss has joined #linux-sunxi
<beeble>
plm: because they are interessted in big established customers
<plm>
beeble: ahh ok. And the phytium not?
fkluknav has quit [Remote host closed the connection]
<beeble>
they are not used to sell outside their own country and the language barrier is even higher
paulk-minnie has quit [Quit: Leaving]
terra854 has quit [Quit: Connection closed for inactivity]
<plm>
beeble: ahh, but a big company like as phytium there are people talking english
<vagrantc>
main problem with the Cello is it doesn't exist
<vagrantc>
or at least, "shipping Q2 2016" is about as far as it exists
<vagrantc>
without the actual shipping part
<tkaiser>
vagrantc: Is LeMaker still existing? Their only asset was the Banana community and this is gone now
* vagrantc
shrugs
<plm>
tkaiser: the LS2088A is better than 88F3720
<jelle>
tkaiser: and it shall not be missed ;-)
<KB3VGW>
ok bbl lunch
KB3VGW has left #linux-sunxi ["Leaving"]
<tkaiser>
plm: Who cares? I don't buy a
<tkaiser>
'server' for 300 bucks when I can either do it right (buy one piece of big iron and let 100 servers run there) or start to do it in a decentral fashion
IgorPec has quit [Ping timeout: 260 seconds]
<plm>
tkaiser: well, but is cheaper one that 300 bucks
<tkaiser>
plm: I still don
<tkaiser>
't understand what *kind* of server you're talking about.
<KotCzarny>
and all that in the context of sunxi
<KotCzarny>
which is cheapest possible solution (and you get what you pay for)
my123 has quit [Ping timeout: 252 seconds]
<rellla>
reading logs makes me wonder about the channel topic anyway :p
<premoboss>
wanna many cores? wait few months, parallella.org near to release a 1024 cores on single chip.
<tkaiser>
KotCzarny: Well, 'sunxi context' is to take a $20 device with 4 USB2 ports, do some weirdo mdraid+btrfs stuff and get performance of an entry level NAS. That's it. Or longsleep's use case: Pine64 cluster testing OS images.
<beeble>
premoboss: yeah, then i can wait on mir cores till the caches are filled by dram :)
<KotCzarny>
i think first thing people should be asked after joining the channel is 'what is your use case'
matthias_bgg has quit [Quit: Leaving]
<premoboss>
that chip is dedicated for number-crunching activities, if you are looking huge ram size, look elsewhere.
<tkaiser>
O, I used an Armbian desktop image for Pine64 by accident with OPi PC 2. Just switched display inputs and realize that there's a full desktop. Now to test video with H5 ;)
<beeble>
premoboss: if your number crunching algorithm fits into 32k
<beeble>
premoboss: not saying there is no usecase for that
<premoboss>
they have 64KB each core. but yes, if algoritm does not fit local ram, move data up and down along the main ram , became a issue.
<beeble>
ah did they increased the size? i had only the number of the 16 core version in my head
<premoboss>
yes. 64 KB eachc core and.... voved from 32 bit to 64 bit architecture.
<beeble>
i had a transputer once
IgorPec has joined #linux-sunxi
<beeble>
it had wheels :)
<premoboss>
16 cores goes ok. 64 core was done il few number cos found restriction, 1024 was founded by darpa (many $$$) so parallella did a big step.
<premoboss>
gold wheels :-)
<beeble>
it was for free. but it was the around 2000 and people just wanted to get rid of it
<beeble>
common internet speed is nowadays about 10 times faster then the cpu interconnect then
my123 has joined #linux-sunxi
<KotCzarny>
internet is a computer!
<KotCzarny>
multimilion core
<KotCzarny>
os is crap though
BigFellow has joined #linux-sunxi
<KotCzarny>
buggy, incosistent, full of bad softwares
<beeble>
but nice pictures of cats
<KotCzarny>
that's part of bad softwares
<tkaiser>
Tried to play a video with mpv on Orange Pi PC 2. Only garbage in a window but this on console: sunxi-ion ion.26: Fail to allocate buffer
<KotCzarny>
eating cycles over nothing
<beeble>
tkaiser: increase your cma size (kernel commandline)
<longsleep>
tkaiser: you could blacklist that module if it prevents boot
<tkaiser>
Tried one h.264 after increasing CMA size. Video gets displayed but no HW acceleration. And in console: sunxi_ion_ioctl(99) err: cmd not support!
<longsleep>
tkaiser: do you have vdpau and all?
<tkaiser>
longsleep: Just removed the module from /etc/modules. But there are still issues. See the bottom of http://sprunge.us/LWbX
<tkaiser>
longsleep: Mikhail prepared yesterday a desktop image for Pine64 and i simply took the rootfs. cat /etc/profile.d/90-vdpau.sh
<tkaiser>
export VDPAU_OSD=1
<tkaiser>
cat /etc/profile.d/99-vdpau.sh
<tkaiser>
export VDPAU_DRIVER=sunxi
<longsleep>
ok
iamfrankenstein1 has joined #linux-sunxi
<tkaiser>
Anyway: time to stop here. I didn't want to waste time with display stuff :)
<tkaiser>
But browsing in firefox went surprisingly well ;)
<longsleep>
tkaiser: do you have the actual stack trace from these crashes
<tkaiser>
Already shut down
<longsleep>
tkaiser: ah one of the errors is that the kernel is too old
<longsleep>
tkaiser: it needs backports for systemd to work correctly
<longsleep>
tkaiser: i added those to the pine64 tree
<KotCzarny>
or just remove systemd
<tkaiser>
longsleep: Ah, systemd madness :)
<KotCzarny>
and dont worry about system/kernel incompatibilities
<longsleep>
its one of the first things i backported to the pine64 tree
<KotCzarny>
it's really sad that you are planting that thing on your images
iamfrankenstein has quit [Ping timeout: 246 seconds]
iamfrankenstein1 is now known as iamfrankenstein
<longsleep>
well i stopped oposing it and started embracing it - closing eyes helps as in rl
<tkaiser>
KotCzarny: Armbian is not a distro but a build system. If Devuan people would want to support +40 SBC it could be done in hours
<tkaiser>
But maybe they're to busy shouting 'systemd is evil' to get work done ;)
<KotCzarny>
longsleep, in the same way sinovoip produces their images. closing eyes on any potential issue, but it boots! :)
<KotCzarny>
tkaiser: they are busy with getting it done
<beeble>
KotCzarny: why does it take so long if sysv init is so much better? *troll*
<KotCzarny>
beeble: the same way nvidia is fasther than amd *trollback*
<tkaiser>
longsleep: Yeah, I remember stumbling accross 'getrandom' already googling an hour before. Anyway: quite surprising to have a full X11 desktop on this device today. I had thought this is stuff for 2017
<longsleep>
tkaiser: right, but the kernel needs work if you want armbian support
<longsleep>
best would be if it just would boot the pine64 kernel tree :P
atsampson has quit [Ping timeout: 260 seconds]
<tkaiser>
Would be really nice. Anyway, now playing with different stuff: Solaris tunables to improve ZFS/iSCSI performance. Bye
<ssvb>
it's quite possible that they had multiple layers of contractors doing this work, and the management is just unaware that there may be some GPL license problems
<ssvb>
but once this issue gets their attention, their lawyers will be unleashed on us as a counterattack
<libv>
so now we let fear guide us?
<ssvb>
anyway, somebody just needs to review their SDK first to document all the blobs which are included there
<berenm>
when comparing to tina, it looks like to me they opened some of blobs at least
<berenm>
drivers/block/nand/lib/src
<libv>
just look at the list of a80 gpl violations
<libv>
and you're not far off.
<ssvb>
before accusing them of anything, first we need to know which parts of the SDK are actually included on the device
<ssvb>
I guess, they might have no need for any touchscreen drivers on this device?
<ssvb>
it would be silly to scream "GPL violators" and point at some blob, which they do not use
<ssvb>
and it's best to do all the communication in a polite and respectable manner
f2zubac has quit [Ping timeout: 268 seconds]
f2zubac has joined #linux-sunxi
yann-kaelig has quit [Quit: Leaving]
<libv>
heh, i do not know why i still let myself be fooled into caring about sunxi.
<libv>
whatever.
<ssvb>
anyway, my prediction is that this particular story will end up with some extremely cheap counterfeit Chinese clones of this Nintendo NES Classic being sold on aliexpress
<ssvb>
then Nintendo management will be unhappy about this and will start searching for a scapegoat
<tkaiser>
Can SLC NAND be cloned 1:1?
<libv>
ssvb: but we have devices out there which can do what the nes does, and much more
<libv>
the clones will just look like the nes
<libv>
which is a clear patent violation
<libv>
and has absolutely nothing to do with software, and thus us
<ssvb>
what I mean is that I don't care about this Nintendo NES Classic thing and wouldn't touch it with a ten-foot pole
<libv>
but whatever, see how nothing will change when no noise is made.
<miasma>
ssvb: you can probably already run nes games on some chinese $7 computers :P the plastic case and gamepads are just missing
<ssvb>
agraf: ^
<nove>
the chinese has better things, like NES-on-chip (real silicone that can emulate a nes)
<ssvb>
agraf: but you can always ask bbrezillon about the nand support if you are interested
<agraf>
ssvb: just asking because you know fel well
<ssvb>
well, fel allows you to load and execute U-Boot & Kernel
<ssvb>
if U-Boot or Kernel has a NAND driver, then you can access NAND
<miasma>
nove, aye but probably any sw emulation (with free emulators and stolen roms) would do. they're not that demanding. unless you want really precise emulation
<ssvb>
agraf: but I'm afraid that anyone providing any technical information which may assist reading, deciphering and/or cloning the Nintendo software in NAND may become a target for Nintendo lawyers
leviathanch has quit [Remote host closed the connection]
<ssvb>
and I don't understand why everyone has become so excited *now*
Ntemis has joined #linux-sunxi
<Ntemis>
tkaiser: around?
Mr__Anderson has quit [Remote host closed the connection]
<ssvb>
reading NAND has been always very interesting for extracting the script.bin file from Allwinner tablets, but nobody tried to do this
<Ntemis>
if only those had an hdmi out
<Ntemis>
and i dont know any that have
<ssvb>
do you mean tablets with HDMI out?
<Ntemis>
aha
<tkaiser>
ssvb: I also don't understand what's that special with this plastic box worth of $5. But what I still don't understand: Is it possible to clone NAND at the 'bit layer'?
<ssvb>
and why do you think that newer framebuffer mali blobs do not exist?
<ssvb>
you are just searching for them in a wrong place
<ssvb>
try to convince mripard that the CHIP board would benefit from running your lakka software and ask him to forward your request for mali framebuffer blobs to whoever may provide them :-)
<ssvb>
yes, it's an old a13 chip, but the same blobs may be also be usable on h3
btoneyo has joined #linux-sunxi
<ssvb>
or try to contact ARM directly via their support channels
<Ntemis>
hmm
<Ntemis>
worth a shot
<btoneyo>
need a help to get my history details in chat room (which server I joined)
mzki has joined #linux-sunxi
<Ntemis>
mripard: ping
nove has quit [Ping timeout: 244 seconds]
nove has joined #linux-sunxi
<nove>
if that is what you mean because is hard to understand what you have said
<nove>
btoneyo: you are in the channel #linux-sunxi of the irc network FreeNode (irc.freenode.net)
Ntemis has quit [Remote host closed the connection]
<phipli>
Hum, anyone playing with the NanoPi Neo? I quite like the form factor of the NanoPi Neo but the sunxi wiki page mentions they get hot - is it much of an issue?
KB3VGW has quit [Ping timeout: 258 seconds]
BigFellow has quit [Quit: BigFellow]
codekipper has quit [Ping timeout: 260 seconds]
<BurtyB>
phipli, I have a few - yes they get hot and slow down as is when under load. I've been using a ceramic heatsink and a 12cm fan blowing air over a them which fixed it for me
<phipli>
A few? Are you clustering them? or just putting one in every room?
<BurtyB>
phipli, just playing around with them atm