<MoeIcenowy>
Oh it seems that Allwinner is prolonging A20's life
<MoeIcenowy>
(as well as A33
<MoeIcenowy>
And I'm doubting what's the T series
<tkaiser>
R40 is called: 'A20 Upgrade Edition: Dual-core upgrade to Quad-core(CPU), 55nm upgrade to 40nm(Craft), lower power consumption, smaller package.'
putti_ has quit [Ping timeout: 240 seconds]
dearfibonacci has quit [Ping timeout: 250 seconds]
<tkaiser>
Different package: BGA411 vs. BGA 468
dearfibonacci has joined #linux-sunxi
perr has joined #linux-sunxi
libv has joined #linux-sunxi
perr has quit [Client Quit]
libv_ has joined #linux-sunxi
keh has quit [Quit: = ""]
libv has quit [Ping timeout: 276 seconds]
libv_ has quit [Ping timeout: 265 seconds]
libv has joined #linux-sunxi
<tkaiser>
T-series mentiones 'Cellular Connectivity': T2 looks like A20, T3 like R40 (SATA but with AXP221s as companion) and T8 like H8/A83T
libv has quit [Ping timeout: 258 seconds]
libv has joined #linux-sunxi
libv has quit [Ping timeout: 240 seconds]
<tkaiser>
And V series is for automotive use (dash cams and 'smart rear-view mirror'), V40 is just another R40/T3 and V66 just another T8/H8/A83T/R58 (one SoC, 5 names)
<tkaiser>
But H series receives only Android 4.4 support (H5 gets 5.1) while A series gets Android 6.0, so the main differentiation are a few bits that enable market segmentation or what is this all about?
<KotCzarny>
laziness
<KotCzarny>
diff. android versions require different minimal kernels
<tkaiser>
Allwinner actively prevents running Android 5.1 or above on H8 (which is otherwise exactly the same as an A83T) but they 'allow' 5.1 on A83T. They implement different chip IDs that get checked when booting with their old u-boot 2011.09
yann has quit [Ping timeout: 244 seconds]
<tkaiser>
So now we have A20, A33/R16, A83T/H8/T8/V66/R58/H8vr, A64, H3, H5 (more or less an improved A64 without PMIC support) and R40/T3/V40 (the A20 successor)
jstein_ has joined #linux-sunxi
<tkaiser>
So exactly one new SoC that is an upgrade of an older one
jstein_ is now known as jstein
<tkaiser>
Regarding R40 they claim: Open Sources: Supports our own lightweight Linux OS called Tina, which specialized designed for smart hardware.
iamfrankenstein has joined #linux-sunxi
<tkaiser>
And on the product page a photoshopped C.H.I.P with R40 on it can be seen. That's all... weird.
<plaes>
:D
<plaes>
link?
<KotCzarny>
their PR staff must be paid by content amount then
<MiLeon>
armbian works good, out of the box vdpau and desktop
<KotCzarny>
wonder what m2+ ultra will be
<tkaiser>
And it seems they use their usual Terminus USB hub since they claim '3 x USB 2.0 host'
<tkaiser>
Ah, ok. It's just their usual 'we don't give a shit what we write'. On another page they write instead: BPI-M2 Ultra have two USB 2.0 interface on board
<speakman>
Hi folks. How do you configure stdout on U-boot SPL? Looks like it's hardcoded somewhere. I don't want "vga" in "stdout" but that seem to be the default.
libv has joined #linux-sunxi
ganbold has quit [Ping timeout: 250 seconds]
mosajjal has joined #linux-sunxi
powerpan_ has joined #linux-sunxi
<powerpan_>
apritzel: so you like soc or not ?
libv_ has joined #linux-sunxi
libv has quit [Ping timeout: 276 seconds]
<apritzel>
powerpan_: which one? the A64?
<powerpan_>
yes
<speakman>
Or; do any of you use Falcon Mode?
<powerpan_>
apritzel: I like A64 comes with emmc.
<apritzel>
powerpan_: well, the A64 is cheap - and you get what you pay for ;-)
<apritzel>
eMMC has some potential, it's probably the fastest interface on the chip
<apritzel>
powerpan_: I was just confused because you spoke of eMMC and Pine64 yesterday
<powerpan_>
apritzel: sure, emmc is the fastest.
<powerpan_>
I have to way to make emmc onboard.
<powerpan_>
so, I just wanna make the full image like Android img
<apritzel>
so it's not the normal Pine64 we all have?
libv has joined #linux-sunxi
perr has quit [Quit: Leaving]
<powerpan_>
well, some other A64 boards comes with emmc
<apritzel>
powerpan_: sure, but this is not Pine64, then ;-)
<apritzel>
that's why I spoke of the BananaPi-M64 yesterday
libv_ has quit [Ping timeout: 244 seconds]
<apritzel>
which has eMMC - though a pretty slow one :-(
IgorPec has joined #linux-sunxi
<tkaiser>
apritzel: That's why I wait for Orange Pi 3 Plus (not even announced yet ;) -- 3 real USB host ports and fast(er) eMMC
<apritzel>
tkaiser: what would be the SoC on this?
<tkaiser>
H5 -- Xunlong's naming scheme is somewhat weird but I would assume an OPi 3 Plus will be like 3 with eMMC
<powerpan_>
apritzel: so, is that easy to port a usb wifi+bt dongle to pine64 ?
<tkaiser>
On OPi 3 PCB is enough space to add eMMC later (as they did with most of their most recent Orange Pis, add eMMC to the board and 'Plus' to the name)
<apritzel>
powerpan_: if there is a Linux driver for that: yes
<powerpan_>
lots of linux supported wifi dongle
<apritzel>
tkaiser: so H5 is A64 with Mali450 plus more USB host ports, right?
<jelle>
is it on sale?
<powerpan_>
but linux could work, dont' know if android ok or not.
<tkaiser>
apritzel: Check cnx-software.com. Jean-Luc asked Allwinner's marketing and they also mentioned improved video engine
<tkaiser>
jelle: What is on sale?
<jelle>
H5
* jelle
starts reading cnx :p
<tkaiser>
jelle: Nope, but Xunlong leaked pictures of OPi 3 and OPi PC 2 where H5 can be seen and Allwinner recently updated their website and now a 2 page product brief can be found there
<jelle>
haa
mosajjal has joined #linux-sunxi
<jelle>
tkaiser: ah ok, thanks
<apritzel>
tkaiser: I already poll cnx-software every hour or so, but still missed those new entries ;-)
<MoeIcenowy>
apritzel: it seems that H5 came with no AXP
<apritzel>
MoeIcenowy: I wonder what that means
<jelle>
btw, should I put something on the wiki now that I start working on the zet622x touchscreen driver?
<MoeIcenowy>
apritzel: It means that H5 boards will have no regulator chip standard
<MoeIcenowy>
different boards may choose different chips
<tkaiser>
apritzel: MoeIcenowy: That means 'OTT box', no battery support and situation just with H3: board specific
<MoeIcenowy>
tkaiser: yes
<MoeIcenowy>
A64 can be used in OTT boxes, however, H5 can *only* be used in OTT boxes
<tkaiser>
MoeIcenowy: It's not only about chip used but also on engineering skills. BPi M2+ uses the same SY8113B voltage regulator like on OPi One/Lite or NanoPi M1/NEO. But 'Team BPi' forgot to implement voltage switching and so BPi M2+ always remains at 1.3V
libv has quit [Ping timeout: 244 seconds]
<MoeIcenowy>
tkaiser: ;-)
<MoeIcenowy>
and I have talked with one of my friend about SYxxxx series regulators
<MoeIcenowy>
he said that "It's only cheap, but may be not so stable"
<MoeIcenowy>
sometimes the DCDC in SYxxxx may even be directly connected, which may hurt other ICs
<powerpan_>
power supply is very important
MiLeon has quit [Quit: Leaving]
libv has joined #linux-sunxi
apritzel has quit [Read error: Connection reset by peer]
dearfibonacci has quit [Ping timeout: 252 seconds]
<tkaiser>
MoeIcenowy: Enjoy the marketing chitchat on the processor pages. Here A83T is called an 'Octa-core In Car Entertainment Navigation System Processor' -- so maybe Transportation?
<tkaiser>
But if the date there (2015-07-23) means that the device was available back then... then it's A83T and not the relabeled T8 ;)
<MoeIcenowy>
I'm checking whether R40 is pin compatible with A20...
<MoeIcenowy>
R40 seems to have Keypads and SATAs, which are missing on other series of SoCs
<MoeIcenowy>
tkaiser: that device may use T2
<MoeIcenowy>
not T8
ganbold has joined #linux-sunxi
Leepty has joined #linux-sunxi
<MoeIcenowy>
the full T series is In-Car usage.
<tkaiser>
MoeIcenowy: Maybe, but it's mentioned on T8 page. But on R40 page a photoshopped C.H.I.P is mentioned so...
<tkaiser>
R40 and A20 don't have the same package so I doubt they're pin-compatible. And then there's still A20E found in A64 BSP ;)
<MoeIcenowy>
R40 have a 28 pins more than A20.
<MoeIcenowy>
A20E may be different package of R40, which cut down the 28 pins
<MoeIcenowy>
To be honest, I'm interested in the so-called Tina
ganbold has quit [Client Quit]
ganbold has joined #linux-sunxi
<MoeIcenowy>
I think Allwinner is trying to create more products, after the shrink of the tablet market
mosajjal has joined #linux-sunxi
ganbold has quit [Quit: Leaving]
ganbold has joined #linux-sunxi
<wens>
doesn't help us if we have to figure all the new products out :/
<MoeIcenowy>
yes
<MoeIcenowy>
but some devices may become too difficult to hack
<MoeIcenowy>
maybe now only Axx and Hx is worth hacking
<tkaiser>
MoeIcenowy: R40 is said to be accompanied by Tina and some 'Open Source' claims so I would believe this would be the easiest new SoC?
<KotCzarny>
open source with blobs?
<KotCzarny>
;)
<tkaiser>
wens: Using mainline u-boot and kernel do CT+ and BPi M3+ behave differently (H8 vs. A83T)?
<wens>
tkaiser: i don't recall any differences
<MoeIcenowy>
oh I forgot the R series
<MoeIcenowy>
sorry
<MoeIcenowy>
at least my R8 device is easy to hack now ;-)
<tkaiser>
Since R8 is A13, R16 is A33, R58 is A83T and R18 (now disappeared) was A64 ;)
paulk-collins has quit [Quit: Leaving]
<tkaiser>
wens: I tested through an H8 image on A83T board and found also no differences. Only 'boot0' had to be exchanged since otherwise Allwinner's u-boot refrained from booting with 'ic check fail,IC must=H8,but id=0x00000001': http://forum.linksprite.com/index.php?/topic/4516-lm-sensors-no-worky/#entry12084
<tkaiser>
So I still assume that there are different business units doing all their own thing and trying to position the very same SoC under different names with loads of marketing chitchat
andor2007 has quit [Quit: Ex-Chat]
<MoeIcenowy>
I remembered the mainline DT for the R16 evb included sun8i-a33.dtsi ;-)
andor2007 has joined #linux-sunxi
andor2007 has quit [Read error: Connection reset by peer]
<wens>
wha?
<MoeIcenowy>
sun8i-r16-parrot.dts
<wens>
oh right
<wens>
tkaiser: sounds possible, if not somewhat naive on their part :p
<MoeIcenowy>
I think maybe {A,H}64 are also different only in BROM
JohnDoe_71Rus has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org]
Net147 has quit [Quit: Quit]
atsampson has quit [Quit: leaving]
Net147 has joined #linux-sunxi
atsampson has joined #linux-sunxi
apritzel1 has quit [Ping timeout: 244 seconds]
<wens>
i started working on a83t psci last night
<tkaiser>
wens: Naive or not. I would believe the only kind of people interested in these SoCs being identical or not are hackers like you :) Normal customers take V66 + marketing chitchat + BSP or A83T + different chitchat + different BSP and are happy
<wens>
unfortunately i did the a80 mcpm smp stuff probably a year ago, and i forgot everything
<wens>
i had to retrace all the code, and the a83t is slightly different from the a80 with regards to control registers :(
<wens>
tkaiser: true
<MoeIcenowy>
tkaiser: yes
<MoeIcenowy>
the BSPs are optimized for usages
<MoeIcenowy>
but not the chips
<FergusL>
I don't remember who answered to my USB MIDI question on Armbian for Orange Pi PC yesterday
<FergusL>
but I'm doing more tests with USB support on the image that I build
<FergusL>
built*
Net147 has quit [Quit: Quit]
<wens>
I remember doing USB MIDI w/ an Akai EWI USB controller and a virtual instrument a few years ago
Net147 has joined #linux-sunxi
<FergusL>
wens: what SoC was it?
dearfibonacci has quit [Quit: Leaving]
<wens>
FergusL: a20
<FergusL>
wens: interesting! do you remember you had anything specific to do?
<wens>
don't remember
<FergusL>
I cannot get USB MIDI devices to be detected, even though I enabled everything in the kernel
<FergusL>
either as built-in or modules, I tried loading the modules in a random order but it still didn't work
<FergusL>
but I realised even a mouse isn't detected
<FergusL>
I noted 2 out of the 4 ports on the Orange Pi PC don't even seem to power the usb devices ><
<tkaiser>
FergusL: Do you use Armbian?
<KotCzarny>
if you have spare usb card try unmodified armbian
<KotCzarny>
s/usb card/sd card/
<FergusL>
I first tried Armbian Debian then Ubuntu and then I compiled my own image, modified the kernel (USB_AUDIO and MIDI lines weren't set in unmodified armbian)
<FergusL>
tkaiser: yes
<FergusL>
the MIDI sequencer works well but the USB part seems to be what's missing, I'm going to check with a simple usb key
cnxsoft has quit [Quit: cnxsoft]
<tkaiser>
FergusL: Then please do an 'sudo armbianmonitor -u' and post support URL. The symptoms sound as if you use fex file for OPi 2 with PC.
<tkaiser>
FergusL: I second wens' suggestion. Settings look ok but the board obviously crashed on first boot when resizing partitions (SD card or PSU related in 101 percent of cases)
<tkaiser>
FergusL: If a wrong fex will be used then USB peripherals might get power but won't negotiate a connection (I made this mistake with NanoPi NEO: thought USB power is to weak to power disks but disabled usb3 and so every disk I tried refused to spin up -- power ok but no USB data negotiation). But your settings are ok
florianH has joined #linux-sunxi
<FergusL>
I think I'm still using the OrangePi power adapter haha, I think I haven't come across another 5V plug with such a slim barrel
<FergusL>
I might try with the usb cable + good USB power adapter
<montjoie>
apritzel: when adding sun504i, you could add a patch for adding N: sun504i to ARM/Allwinner sunXi SoC support in MAINTAINERS
<FergusL>
Huuu I feel stupid, I think I actually own an OPi 2... not an OPi PC
<tkaiser>
FergusL: Then it's easy: cd /boot && ln -sf bin/orangepi2.bin script.bin && reboot
<KotCzarny>
FergusL: search ebay/aliexpress for 4.0/1.7mm 5v
<KotCzarny>
plenty of those, dont have to be usb brick
ninolein has quit [Remote host closed the connection]
<tkaiser>
FergusL: Without correct fex settings the internal USB hub won't be powered so USB behaviour is clear now
<KotCzarny>
you can even just grab the plain cable and use ANY power brick (for example meanwell rd-65a)
<KotCzarny>
(or rs-50-5)
premoboss has joined #linux-sunxi
ninolein has joined #linux-sunxi
<KotCzarny>
and opi2 is rated for 5V/2A, so your current brick might be ok
<FergusL>
I can read "Orange Pi 2" and V1.0 right below it on the topside of the board
<tkaiser>
FergusL: Do what I told you and USB problem is resolved. It can't work with settings for PC on the 2
<tkaiser>
KotCzarny: Please don't tell people about these 'power ratings' since they're crap :)
<KotCzarny>
if you want to make some order from crazy chinese naming schemes
<KotCzarny>
tkaiser: good 2A power brick is ok usually, emphasis on good
ikmaak has quit [Ping timeout: 240 seconds]
<KotCzarny>
tkaiser, opi2 is bigger than opipcand has wifi and internal usb hub
<KotCzarny>
also lacks a mic
<tkaiser>
KotCzarny: And the power ratings in this huge table are still BS :)
<KotCzarny>
yup, but who am i to fix them
<KotCzarny>
even board rated for 3A might do well with 1A power source if done right (ie. use case doesnt require much work)
<tkaiser>
Add 3 2.5" USB disks to any SBC and you know why Xunlong ships a 3A PSU with OPi Plus 2E. Since more people will do that with this device. Apart from that power requirements of all H3 based Orange Pi are pretty much the same
<tkaiser>
It's only about the count of peripherals and their consumption.
<KotCzarny>
i do not use usb for disks, and even then i would power them from separate connector
<tkaiser>
KotCzarny: But people do and that's the reason why in your table 3A appear for OPi Plus 2E
<KotCzarny>
but people who do that should have enough brain to know that usb takes power, lots of
<KotCzarny>
in similar case in pc world one choses PSU suitable for their system (strong gpu vs embedded)
<KotCzarny>
maybe we should just add a footnote 'powe req for board WITHOUT peripherials'
<KotCzarny>
*power
<MoeIcenowy>
montjoie: what's 504i?!
<MoeIcenowy>
maybe you mean 50i?
<KotCzarny>
done.
<KotCzarny>
also footnotes should be on top i guess, so people would read the warnings and notes first
<montjoie>
MoeIcenowy: yes:)
<tkaiser>
KotCzarny: Using cpuburn-a7 running on a H3 with 1200 MHz consumption increases by 4W compared to idle. No normal workload is that heavy and therefore consumption increase under full CPU load is less. So based on this PSU requirements for all H3 boards are 5V/1A without taking peripherals into account. So simply delete the column since it's misleading anyway ;)
Andy-D has quit [Remote host closed the connection]
reinforce has quit [Quit: Leaving.]
<KotCzarny>
you should always have some margin
<KotCzarny>
also, say what you want, bpi-r1 works very nice as a home server/router/shell
<KotCzarny>
and i've rebooted 3 months ago because i had to do some room renovation
GeneralStupid has joined #linux-sunxi
<GeneralStupid>
Hi, want to use libcec with my orangepi pc. i read about a patch... do i need to compile a new kernel?
<KotCzarny>
if patch is for the kernel, then probably answer is 'yes'
<FergusL>
KotCzarny: My OPi 2 has mic and wifi
<GeneralStupid>
KotCzarny: i'll give it a tr
<GeneralStupid>
y
<FergusL>
then maybe it's not an Opi2 after all? :D
<FergusL>
let's try swapping the fex first
<KotCzarny>
FergusL: opipc doesnt have wifi, so you misread. but mic . hmm
<FergusL>
sorry yes I misread/misanswered
<FergusL>
YES! it's working now! fantastic. Thanks a lot tkaiser
<FergusL>
so... maybe I should go back to the standard Armbian OPi2 image and MIDI is actually working?
<foxx_>
<tkaiser> KotCzarny: Using cpuburn-a7 running on a H3 with 1200 MHz consumption increases by 4W compared to idle
<FergusL>
by default when building Armbian, is the kernel config the one used in standard armbian?
<KotCzarny>
check if its working now first
<foxx_>
how did you measure?
<tkaiser>
foxx_: I used a Banana Pro and power all sorts of SBC through Banana's USB ports, then subtracting Banana's idle consumption et voilˆ.
<FergusL>
KotCzarny: it is working, midi usb device detected in lsusb AND in "aconnect -il"
<tkaiser>
FergusL: Then please give stock Armbian for OPi 2 a try and in case you encounter problems please report back in Armbian forums so we could improve Armbian if there's something missing for MIDI
<jemk>
tkaiser: it has to be run with lima-memtester, otherwise it won't show much difference
<jemk>
maybe i should prepare some fel lima-memtester as ssvb did
<tkaiser>
jemk: Maybe it's easier since lima-memtester can be run on a normal OS image and we can adjust DRAM clockspeed from userspace (I learned that just recently unfortunately when doing consumption tests/improvements)
<KotCzarny>
hmm, 'Advanced Banana Pi Case - Tactical Black' actually looks nice
<foxx_>
tkaiser: that just sounds a bit crazy for me, since as far as i understood, cpuburn loads vector alus along with cpu cores, is there any part that could increase the consumption so significantly?
<jemk>
tkaiser: adjusting dram clockspeed from userspace will likely hide the effect of this patch
<tkaiser>
foxx_: Ask ssvb please :)
<tkaiser>
jemk: With H3 legacy kernel we can do something like this: echo 408000 >/sys/devices/platform/sunxi-ddrfreq/devfreq/sunxi-ddrfreq/userspace/set_freq
<tkaiser>
jemk: So it's enough to use a kernel with disabled DRAM throttling bevaviour, a script.bin allowing values up to 800 MHz and then running lima-memtester and iterating through clockspeeds in 24 MHz steps.
<tkaiser>
jemk: A, ok, now I understand. It's about calibration...
<ssvb>
tkaiser: by reclocking the dram on a running system, you are testing the Allwinner's dram code rather than the dram code from the mainline U-Boot
<tkaiser>
Ok, understood
<jemk>
tkaiser: i'm just searching wheter this driver does some calibration too
<tkaiser>
Then we need a FEL lima-memtester package again testing through different clockspeeds
<tkaiser>
jemk: Ah, ok. Well, I fear I lack the time to test through. If you can wait 3 weeks I'll give it a try.
yann has quit [Ping timeout: 240 seconds]
paulk-collins has joined #linux-sunxi
<jemk>
on my opi plus it didn't make a difference, but it always worked well up to 768mhz
<jemk>
but since others have boards crashing even at 672 it might change something there
<tkaiser>
Maybe they can test? For example plaes? IIRC he was one of those crashing at 672?
<ssvb>
jemk: but if I understand it correctly, we still have non-deterministic calibration even with your PLL lock fix
powerpan_ has quit [Ping timeout: 264 seconds]
JohnDoe_71Rus has joined #linux-sunxi
<jemk>
ssvb: it looks pretty stable to me
<KotCzarny>
when you make the limamemtester i can give it a go
<ssvb>
jemk: on both cold and warm reboot?
<KotCzarny>
i have a opipc that crashed on 672
<jemk>
ssvb: yes, so far (but its not related to zq)
<jemk>
it mainly fixed the period measurements done to get the dqs signals 90° phase shifted to dq
<jemk>
if we measure that with too low freq (too early) it won't be correct once the pll reaches the final freq
GeneralStupid has left #linux-sunxi [#linux-sunxi]
aalm has quit [Ping timeout: 265 seconds]
<jemk>
i don't know if this is really relevant, but it was another small difference to boot0
<ssvb>
do you mean the dqs gating delay configuration?
<jemk>
not the gate delay, but the delay between dq and dqs
Mr__Anderson has joined #linux-sunxi
<jemk>
dqs has to be shifted to cross the middle of the data eyes
tkaiser has quit [Ping timeout: 240 seconds]
MiLeon has joined #linux-sunxi
yann has joined #linux-sunxi
yann has quit [Ping timeout: 244 seconds]
premoboss has quit [Quit: Sto andando via]
jernej has joined #linux-sunxi
yann has joined #linux-sunxi
jstein has quit [Remote host closed the connection]
GeneralStupid has joined #linux-sunxi
<GeneralStupid>
Ok iam back again.
<GeneralStupid>
I need to compile the armbian orangepi pc kernel.
<KotCzarny>
and?
<GeneralStupid>
where do i find there standard config
<GeneralStupid>
because iam already using their kernel
<GeneralStupid>
gcc: error: unrecognized command line option '--min_array_alignment=4'
<GeneralStupid>
hmmm
<KotCzarny>
upgrade gcc?
<KotCzarny>
or drop that option
<KotCzarny>
or if you are lazy like me just compile on-device
<GeneralStupid>
i compile on the device...
<KotCzarny>
it only takes ~40 minutes
* KotCzarny
remembers good ol' amd-k2@300mhz times
tkaiser has joined #linux-sunxi
<GeneralStupid>
i used netbsd on an amd geode for a long time
apritzel has quit [Read error: Connection reset by peer]
<buZz>
:D
<KotCzarny>
also, you can use crosscompiles on x86 via distcc too
<buZz>
i used to run gentoo on a 486sx25
<buZz>
now THAT was an exercise in patience :D
<buZz>
>8 hours for kernel compile
* KotCzarny
used ssh on amiga500 once
<buZz>
KotCzarny: did they ever fix distcc to analyze version numbers of glibc/gcc ?
<buZz>
back when we had a ~30 node distcc group, we had to blacklist everyone with 'wrong versions'
<KotCzarny>
dont know, all my distcc hosts are homogenous in regard to os/gcc
<buZz>
mhm
<KotCzarny>
ie. armbian ;)
matthias_bgg has quit [Quit: Leaving]
<buZz>
devuan :)
<buZz>
armbian subscribe to poetty
<buZz>
+s
<GeneralStupid>
yes ... you can :) But i dont. i just compiled the whole userland and kernel on that thing
<GeneralStupid>
then i switched to debian
<KotCzarny>
buzz, yeah, unfortunatelly :/ still, on mine i have systemd removed (Still)
<GeneralStupid>
buZz: iam already using devuan. But i also use armbians kernel
<buZz>
devuan <3
<buZz>
GeneralStupid: you are forgiven
iamfrankenstein1 has joined #linux-sunxi
<buZz>
:)
<KotCzarny>
hmm, armbuan ;)
reinforce has joined #linux-sunxi
<KotCzarny>
armvuan, better
<buZz>
devuan already has arm support
<GeneralStupid>
its no problem, you can mix armbian with devuan. already read about that
<KotCzarny>
buzz, but does it have sunxi support?
iamfrankenstein has quit [Ping timeout: 260 seconds]
iamfrankenstein1 is now known as iamfrankenstein
<KotCzarny>
GeneralStupid: do you just switch repos or there is more trickery required?
<GeneralStupid>
KotCzarny: just saying. took devuan, boot armbian kernel. and it works
<buZz>
not sure, but i THINK i saw cubieboard builds of devuan
<GeneralStupid>
normally you just switch repo and thats it
<GeneralStupid>
is min_array_alignment important?
<tkaiser>
GeneralStupid: Do you try to compile legacy kernel? 3.4?
<GeneralStupid>
3.4.
<tkaiser>
On Devuan?
<GeneralStupid>
yes
<foxx_>
my parents still take care of mine first celeron@333mhz based pc, e.g. wipe the dust etc.
yann has quit [Ping timeout: 276 seconds]
<tkaiser>
GeneralStupid: Well, that's funny :)
<KotCzarny>
foxx, does it do work for them or is just a decoration
<GeneralStupid>
tkaiser: works on yours?
<GeneralStupid>
tkaiser: which gcc do you have installed?
apritzel has joined #linux-sunxi
<tkaiser>
GeneralStupid: I never compile on ARM boards since I lack the time, kernel compile on my i7 is ~2 minutes. BTW: How do you handle our legacy kernel patches?
<GeneralStupid>
tkaiser: i use the kernel source from armbian
<KotCzarny>
but you want to apply a patch/recompile, so you will have to (or miss the point)
<tkaiser>
GeneralStupid: Outdated, members of the 'systemd hater's club' might not realize that kernel updates come also via apt-get upgrade ;)
<KotCzarny>
it would still work unpatched, though
<KotCzarny>
tkaiser: i dont use armbian's kernel packages, compile my own
<tkaiser>
KotCzarny: I don't really care. Just FYI: Armbian is a build system and not a kernel. And when there's a kernel patch that enables more cpufreq operating points then 'maybe' there's also an update to fex/DT stuff and 'maybe' also an u-boot patch to make things work smoothly
<GeneralStupid>
that still does not solve my gcc problems :D
<tkaiser>
Armbian users get all this stuff via apt-get, others I don't know/care (most probably not)
<tkaiser>
GeneralStupid: There are no GCC problems, use Armbian's build system on a Xenial amd64 host. Armbian will download the necessary toolchains for you if you messed up your system
<GeneralStupid>
tkaiser: can you introduce me how to quick and dirty build a kernel using armbians build system?
<tkaiser>
GeneralStupid: Documentation is made for reading. Everything explained there. Just read through the chapter. It saves you time.
<GeneralStupid>
ok
<tkaiser>
KotCzarny: Yes, that's a feature and not a bug!
Netlynx has joined #linux-sunxi
Netlynx has quit [Changing host]
Netlynx has joined #linux-sunxi
<FergusL>
Is there consensus on quality and support of orange pi h3 versus nano pi h3 ?
<KotCzarny>
which orange pi
<KotCzarny>
there are thousands of them
<huawei>
is the A20 audio playback availables in mainline kernel?
mzki has quit [Quit: leaving]
<KotCzarny>
yes
<tkaiser>
FergusL: Oranges show better thermal behaviour and consume less power (at least OPi Lite vs. NanoPi NEO -- havent's tested NanoPi M1 yet)
<tkaiser>
FergusL: NanoPi M1 is just like an Orange Pi PC with worse thermal/consumption behaviour. NanoPi NEO is something completely different. That's the hardware side.
<FergusL>
Friendly arm has always seemed "better" to me, cleaner, less shady and imprecise
<tkaiser>
FergusL: Regarding software/support and matching peripherals (so called Matrix) stuff FriendlyARM is simply awesome. Especially if you compare to other $fruitpi vendors from China
<FergusL>
I remember being very sceptical about orange pi
<FergusL>
Ha yes, that's how I felt
<tkaiser>
FergusL: Oranges are nice hardware but vendor's software/support obviously suck. But IMO that doesn't matter that much since we have linux-sunxi community and good OS images in the meantime.
putti_ has joined #linux-sunxi
<KotCzarny>
tkaiser, not true, it matters to promote good guys
Mr__Anderson has quit [Remote host closed the connection]
tkaiser has quit [Read error: Connection reset by peer]
MiLeon has quit [Quit: Leaving]
MiLeon has joined #linux-sunxi
tkaiser has joined #linux-sunxi
MiLeon has quit [Quit: Leaving]
apritzel has quit [Ping timeout: 244 seconds]
MiLeon has joined #linux-sunxi
OverCR has quit [Quit: Leaving.]
|Jeroen| has quit [Quit: dada]
ricardocrudo has quit [Remote host closed the connection]
<FergusL>
What KotCzarny said! Purchasing allwinner based devices is already supporting not so good guys
<KotCzarny>
there are worse evils
<KotCzarny>
broadcom, qualcomm
<KotCzarny>
those are top of the sh*tpies
<FergusL>
So if I could at least support good makers doing the boards in the most open manner, I would do it
<FergusL>
I really like the new form factor of these boards, like the opi one and lite
<KotCzarny>
and xunlong (orange) isnt so bad itself, boards are made with sense, and not cheap clones
<FergusL>
Air will seemingly be even more bare bones and small
<FergusL>
Kotczarny: yes embedding audio processing and synthesis in possibly small devices
<FergusL>
So the RPI form factor for example is itself too big in width and height for some applications
z0x79 has joined #linux-sunxi
<z0x79>
hey there
<KotCzarny>
electronic flute?
<z0x79>
i have a big gap in understanding some stuff concerning NAND. I have an A33 unknown device that has Android 4.4.2 installed, no ADB and no recovery to boot from. I have managed to boot from sdcard and try to access the NAND but the devices will not get loaded.
<z0x79>
so here goes the first question: by booting from sdcard, should I be able to see /dev/nand or at least /dev/block/nand?
<KotCzarny>
though page is from ancient a10/a20 times
<KotCzarny>
bu maybe you will get some hints
<z0x79>
do i miss the part where it says if i can access NAND from a booted sdcard?
<FergusL>
Haha that might happen! But for such sizes a micro controller instead of full Linux might be enough
<KotCzarny>
electronic flute might be just a synth organ with input connected to fingers and presses modulated via some air pressure sensor
Nacho____ has joined #linux-sunxi
<ssvb>
z0x79: of course you can access the NAND hardware after booting from the SD card, but in order to do this in a meaningful way you still need NAND drivers too
Nacho has quit [Ping timeout: 244 seconds]
<z0x79>
ok, good news. next question: i was trying out different images that all loaded nand.ko. but the device never came up
<z0x79>
so would i need a driver for the SoC or for the NAND?
<z0x79>
as i understand i need a driver for the nand controller in my A33
<z0x79>
correct?
<z0x79>
so the NAND should be completely irrelevant as there is the controller abstraction layer talking to the kernel driver?
<ssvb>
are you trying to recover your broken Android system or something?
<z0x79>
exactly
<z0x79>
i have no ADB, no recovery
<KotCzarny>
z0x79: isnt there usb flasher ? (phoenix tools)
<z0x79>
i thought the flasher just flashes my device
<z0x79>
i dont have a stock rom
<z0x79>
it is an unknown device
<z0x79>
i basically need a backup of the device without using ADB or a recovery
<z0x79>
so i thought booting to sd and dd-ing nand would be a good idea
<KotCzarny>
or just run the device from microsd and forget about nand
<z0x79>
yeah not an option
<z0x79>
i need a dump and the data on it
<KotCzarny>
are you fbi? :>
<z0x79>
well
<z0x79>
not really but indeed i need to do some forensic stuff
<ssvb>
allwinner uses its own partitioning scheme for nand and also its own software layer, which emulates a regular block device on top of the mtd hardware
<ssvb>
moreover, the data format of all of this stuff is not set is stone, but has been changing
<z0x79>
that means i need an allwinner nand.ko?
<z0x79>
or just a tweek to nand.ko source to make it work?
<ssvb>
what kind of kernel are you using now?
<z0x79>
give me a sec
<ssvb>
btw, allwinner's nand driver can potentially corrupt data if you try to use it on a wrong device
<KotCzarny>
ssvb, on read too?
<ssvb>
iirc this happened to some people, who tried to use the nand driver from the a20 bsp kernel with some older devices (a10 or maybe a20 devices using an older bsp)
<ssvb>
backward/forward compatibility seems to be an alien concept to allwinner
<z0x79>
it is a 3.4.39
<z0x79>
android 4.4.2 KVT49L
<z0x79>
i heard that too, concerning nand damaged by wrong version of a driver
<KotCzarny>
hmm, does that android still boot?
<KotCzarny>
if yes, you might check if the debug root bug is still on
<FergusL>
Is your device a TV box?
<z0x79>
no a tablet PC
<z0x79>
KotCzarny: i do not have a shell, i guess i need a shell for this to work
<KotCzarny>
shell == terminal app
agraf has quit [Ping timeout: 258 seconds]
<KotCzarny>
just to run simple echo something > /proc/somewhere
Netlynx has quit [Quit: Ex-Chat]
MiLeon has quit [Quit: Leaving]
<z0x79>
i cannot install APKs
<z0x79>
;)
<z0x79>
custom app store with signed apks
<KotCzarny>
maybe there is some terminal app preinstalled?
aalm has joined #linux-sunxi
<z0x79>
there is a "Shell" but i cannot launch it, so doing anything on the device is not really possible
<z0x79>
the file browser is even scoped down to have access only to the sdcard
<tkaiser>
ssvb: Last task priot to vacation: I did some DRAM related testings regarding performance, consumption and temperatures with NanoPi NEO (first H3 board with single bank DRAM configuration in the wild). Maybe that's the reason why Olimex experienced overheating issues with their H3 prototypes: On NEO DRAM clockspeed influences both temperature and consumption way more than on OPi Lite for example:
<ssvb>
tkaiser: hmm, interesting, so the NanoPi NEO board uses a single x16 ddr3 chip?
<ssvb>
such setup is usually bad for graphics and also driving a 1080p monitor may cause an additional severe performance drop
<ssvb>
that's a good reason to avoid this NanoPi NEO board
lemonzest has quit [Quit: Leaving]
<GeneralStupid>
ssvb: i dont think nano pi is designed for driving full HD monitors...
iamfrankenstein has quit [Quit: iamfrankenstein]
Mr__Anderson1 has quit [Quit: Leaving.]
<tkaiser>
ssvb: There's neither HDMI nor composite video on this board. It's meant as a headless IoT node or server
iamfrankenstein has joined #linux-sunxi
reinforce has quit [Quit: Leaving.]
<tkaiser>
But I found the consumption and temperature differences surprising, especially when comparing to an Orange Pi Lite with identical settings
<ssvb>
well, there might be some other reason for the power consumption difference
<ssvb>
but yes, for a purely IoT device even the 16-bit dram bus width is likely good enough, while this saves them some pennies and a bit of space on the PCB
<tkaiser>
ssvb: I mean only consumption difference when adjusting DRAM clockspeed between 132 and 672 MHz (470 mW on NEO, 305 on Lite). NEO has higher idle consumption anyway...
<tkaiser>
And since temperature increase reported by the SoC is also way higher... I thought it must be memory controller related. But anyway: Time to go to bed :)
<jemk>
do these low clockspeeds really work? ddr3 is only specified down to 300mhz.
<jemk>
the controller does turn off some clocks when idle, maybe that safes more power being idle with high clock then being active all the time with low clock
ricardocrudo has quit [Remote host closed the connection]
jernej has quit [Ping timeout: 258 seconds]
ganbold has quit [Read error: Connection reset by peer]