<ssvb>
jernej: very cool, the HDMI driver was the only missing thing before we can have a full fledged firmware
<ssvb>
miasma: you can read the boot rom code and review it
<miasma>
ssvb: you mean i can read the code from the chip
<ssvb>
yes
jernej has quit [Ping timeout: 265 seconds]
<miasma>
ok that's cool
<ssvb>
it is not protected at all
<miasma>
but I guess in theory there could be two bootroms, one for inspection and one for booting, if the vendor wants to do some nasty things
<ssvb>
and you can dump it, for example, using the sunxi-fel tool
<miasma>
wasn't there also some source available
<ssvb>
yes, the datasheet also mentions something about the secure ROM
<miasma>
probably for secure booting
<ssvb>
well, I have no idea how it works
<ssvb>
I guess that some bit in the eFUSE may probably switch it into the secure boot mode
<ssvb>
but then again, I have no idea what would happen and it does not seem to be well documented in general
vagrantc has quit [Quit: leaving]
BenG83 has joined #linux-sunxi
mzki has quit [Ping timeout: 260 seconds]
jernej has joined #linux-sunxi
jernej has quit [Ping timeout: 244 seconds]
popolon has quit [Quit: WeeChat 1.4]
yann-kaelig has joined #linux-sunxi
BenG83 has quit [Ping timeout: 240 seconds]
<wens>
tkaiser: can you push the fex file onto sunxi-boards?
willmore has quit [Ping timeout: 260 seconds]
<wens>
KotCzarny: gpu is disabled? :p
willmore has joined #linux-sunxi
<wens>
mripard: afaik the usb pins that exist on gpio headers don't have any regulator / vbus tied to them
<wens>
the user needs to get 5v from the power pin
Colani1210 has joined #linux-sunxi
Colani1200 has quit [Ping timeout: 265 seconds]
cnxsoft has joined #linux-sunxi
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
dh1tw has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
Gerwin_J has joined #linux-sunxi
stk is now known as swiftk
ninolein_ has quit [Ping timeout: 240 seconds]
ninolein has joined #linux-sunxi
bugzc has joined #linux-sunxi
<MoeIcenowy>
jernej: Can your H3 HDMI U-Boot patches support unstandard HDMI resolutions?
<MoeIcenowy>
for example 1024x600
<MoeIcenowy>
jernej: and detect H3 by CONFIG_MACH_SUN8I is not enough
<MoeIcenowy>
some SUN8Is are also using DE1 -- A23,A33
dpg_ has quit [Ping timeout: 246 seconds]
<wens>
great... something wrong with secure/non-secure setup on the a80
<wens>
i am getting the secure copy of the gic cpu interface registers :(
iamfrankenstein has quit [Ping timeout: 256 seconds]
dpg_ has joined #linux-sunxi
pg12 has quit [Ping timeout: 240 seconds]
pg12 has joined #linux-sunxi
IgorPec has joined #linux-sunxi
[7] has quit [Ping timeout: 260 seconds]
TheSeven has joined #linux-sunxi
JohnDoe_71Rus has joined #linux-sunxi
camh has quit [Ping timeout: 246 seconds]
yann-kaelig has quit [Quit: Leaving]
camh has joined #linux-sunxi
reinforce has joined #linux-sunxi
jernej has joined #linux-sunxi
<jernej>
MoeIcenowy: ifdefs and defines are not yet fixed, just enough to work for H3. Actually I didn'tknow how to properly organize them and I wanted to ask here first
<jernej>
MoeIcenowy: Resolution (timings) are based ob EDID and yes, it can be whatever is stored there, also nonstandard
<MoeIcenowy>
jernej: hooray
<MoeIcenowy>
the sh*t BSP driver cannot deal with non-standard timings
<jernej>
MoeIcenowy: But it works only for H3 and others SoCs with same PHY. I think that A64 (and H5) is a bit different, according to BSP U-Boot video driver code
<MoeIcenowy>
but I won't use BSP...
iamfrankenstein has joined #linux-sunxi
<jernej>
but it should be easily portable
<MoeIcenowy>
jernej: I do have a H3 board
<MoeIcenowy>
an opi one
<MoeIcenowy>
jernej: in sun8i line, A23 and A33 is using disp1
<MoeIcenowy>
A83t, H3, H2+ is using disp2
<jernej>
hm, proposed H3 mainline driver doesn't support arbitrary resolutions...
<jernej>
yet
<MoeIcenowy>
jernej: yes
<MoeIcenowy>
I tested it, and it nearly burned my 1024x600 LCD
<MoeIcenowy>
it's a LCD with a simple control chip, which can only work under its only EDID timing
IgorPec has quit [Ping timeout: 260 seconds]
<MoeIcenowy>
Or, on Raspberry Pi, specify the timing in config.txt
<jernej>
Well, you can try my u-boot driver and see it it works. For now, result will be greenish output, but should be easily readable
<jernej>
I would be interested if it works for you
<MoeIcenowy>
Yes... but I think A64 may be around it
<MoeIcenowy>
someone has reported A64 HDMI is just H3 HDMI + CEC function
<jernej>
Code for setting timings was tested on 720p, 1080p and 1280x1024 resolutions
<jernej>
that was me
<jernej>
and i was a bit wrong, it seems
<jernej>
anyway, driver could be copied from U-Boot BSP and just add BSP functions, at least for BSP
<MoeIcenowy>
let me clone your source code
<MoeIcenowy>
and the post office told me to get my package from Xunlong
<MoeIcenowy>
any have got a H5 here?
<MoeIcenowy>
anyone *
petr has quit [Ping timeout: 260 seconds]
<ssvb>
MoeIcenowy: maybe I'll receive a DHL package from Xunlong on Thursday, IgorPec provided me with a tracking number
petr has joined #linux-sunxi
DullTube has joined #linux-sunxi
jernej has quit [Ping timeout: 252 seconds]
<MoeIcenowy>
I should make the decision in last Friday...
<MoeIcenowy>
thus I can get it last Saturday
<MoeIcenowy>
The shipment from Shenzhen to Canton costs only ~10hrs
berkutta has joined #linux-sunxi
Putti has joined #linux-sunxi
<KotCzarny>
wens: either that, or mainline manages things more efficiently (ie. switches more clocks off when unused), because box is in the same physical config
IgorPec has joined #linux-sunxi
paulk-collins has joined #linux-sunxi
yann has quit [Ping timeout: 265 seconds]
<wens>
KotCzarny: the gpu clock is disabled
<KotCzarny>
wens, what i meant was 'mainline might be more power efficient/smart'
<wens>
it just shuts down everything that's not claimed
<KotCzarny>
and that could be another significiant selling point for mainline vs legacy
iamfrankenstein has quit [Ping timeout: 245 seconds]
<MoeIcenowy>
yesterday I have heard that bpi still do not focus on mainline
<MoeIcenowy>
and I got my opipc2 now.
<KotCzarny>
cool, did it boot with h3 image?
<ganbold>
it is H5 so maybe not
<KotCzarny>
allwinner does lazy things when it comes to 'new' socs
<MoeIcenowy>
First thing I should say is that
<MoeIcenowy>
the SPI Flash on board do not boot onw
<MoeIcenowy>
now *
<MoeIcenowy>
So easyly it enters "Bus 001 Device 007: ID 1f3a:efe8 Onda (unverified) V972 tablet in flashing mode"
<apritzel>
IgorPec: Allwinner shot themselves in the foot by not properly abstracting the PMIC interface
<apritzel>
wens: yeah, 1718
<MoeIcenowy>
apritzel: is H64 1689?
<ssvb>
MoeIcenowy: the SPI support is still in a separate branch, but now that we have an actual board, we can test it and push into the sunxi-tools master branch
<MoeIcenowy>
I think your Jide Remix Mini is H64
<MoeIcenowy>
but we have now no bootloader for it.
<MoeIcenowy>
except BSP boot0
<wens>
apritzel: what happened with the PMIC?
<apritzel>
wens: the H5 is not supposed to be paired with the AXP803
<MoeIcenowy>
I'm trying to run a allwinner testing system on the board
<MoeIcenowy>
also called "dragonboard" in lichee
<apritzel>
wens: so it has the SY8106A like many H3 boards
<apritzel>
wens: but AW managed to rely on the AXP in every part of their A64 software chain - from kernel to U-Boot to ATF to SCP
<apritzel>
wens: so not surprisingly Pine64/A64 images do not work on the H5, though the differences are really minimal
Putti has joined #linux-sunxi
<ssvb>
KotCzarny: having bigger SPI flash is always fine, but probably Xunlong is a bit conservative with the advertised specs because they want to reserve the right to reduce it in the future
<KotCzarny>
ssvb: so it's early bird's special? or just dev samples
<ssvb>
MoeIcenowy: you can pretend that H5 is the same as A64 and just add the new SoC ID to the sunxi-tools source code in the switch statements
<apritzel>
MoeIcenowy: you could try to use my pine64-firmware.img from http://andrep.de/a64, I guess that should work
<MoeIcenowy>
apritzel: why do you think so :-)
<apritzel>
MoeIcenowy: I haven't tried it lately on the SPI flash, but it used to work there
<MoeIcenowy>
I got a H5, not a A64
<apritzel>
MoeIcenowy: because I don't _rely_ on the AXP ;-)
<apritzel>
and the whole stack is written with a family idea in mind
<ssvb>
apritzel: we would probably need to add the new SoC ID support in the pins muxing setup code
<apritzel>
ssvb: in what part? sunxi-fel?
<ssvb>
apritzel: both in sunxi-tools and in U-Boot
<ssvb>
but if you just define H5 as A64, then it probably should already work
<apritzel>
ssvb: or if you just throw some of the existing A64 (upstream) stuff on it ...
<ssvb>
why would Allwinner define the new SoC ID, maybe something is different after all?
<apritzel>
ssvb: do we really check the SoC ID in U-Boot at runtime and make pinmux setup dependent on it?
<ssvb>
we define it in defconfig
<MoeIcenowy>
how about the pinctrl?
<MoeIcenowy>
I read H5's datasheet
<MoeIcenowy>
the pinctrl is more like H3
<MoeIcenowy>
at least for UART0
<apritzel>
ssvb: two more USB controllers, Mali 450, not normally AXP
<apritzel>
ssvb: that's what we know so far
<apritzel>
MoeIcenowy: the A64 is very close the H3 already
<MoeIcenowy>
ssvb: how to build uart0-helloworld?
<NiteHawk>
MoeIcenowy: try "make binfiles"
<MoeIcenowy>
ssvb: there's also a problem on H5: if I burn something in, I cannot enter FEL
<ssvb>
apritzel: btw, the code is ready to support runtime detection, it can be just hooked instead of the IS_ENABLED macro - http://git.denx.de/?p=u-boot.git;a=blob;f=drivers/mtd/spi/sunxi_spi_spl.c;h=67c7edd92d0c6b1d7b20d072ab50085ce9bcc5b8;hb=HEAD#l163
<MoeIcenowy>
not on H5
<MoeIcenowy>
but on Orange Pi PC2
Putti has quit [Ping timeout: 252 seconds]
<apritzel>
MoeIcenowy: FEL entering if SPI is broken: yeah, that's nasty, it came to me lately
<ssvb>
apritzel: the SPI0 pin function code is different on A64 and H3
<MoeIcenowy>
how can I break it :-(
<apritzel>
MoeIcenowy: you have to use the FEL stub on SD card
<MoeIcenowy>
oh yes
<ssvb>
MoeIcenowy: btw, is the HOLD pin pulled up or connected to 3.3V directly?
<MoeIcenowy>
I forgot this
<MoeIcenowy>
I do not know
<ssvb>
because if it is pulled up, then it is possible to just connect this pin to the ground to prevent booting from SPI flash
<ssvb>
but if it is connected to 3.3V directly, then this is not a very good idea :-)
<MoeIcenowy>
ssvb: thanks I built uart0-helloworld
<MoeIcenowy>
will test it now
Wizzup has quit [Ping timeout: 250 seconds]
<apritzel>
ssvb: according to the schematic it's 3.3V directly
<apritzel>
also SPI_CE is pulled up, so not connected to SPI_CS0
<apritzel>
WP is also 3.3V
<ssvb>
hmm, is the SPI CS pin not connected to the SoC?
<MoeIcenowy>
ssvb: uart0-helloworld can return back to FEL?
<apritzel>
ssvb: oh wait, _CE _is_ connected to CS0
<ssvb>
MoeIcenowy: yes
<MoeIcenowy>
ok I think I can flash it
<ssvb>
apritzel: ok, makes sense
<apritzel>
ssvb: the schematic is a bit confusing there
<apritzel>
it's both connected and pulled up
<ssvb>
MoeIcenowy: try it on the SD card first
<MoeIcenowy>
ok
<ssvb>
MoeIcenowy: if H5 is similar to H3, then the SPI pins muxing code needs to be updated in sunxi-fel appropriately
<apritzel>
wow, according to the data sheet there is no PortB on the H5
<MoeIcenowy>
apritzel: also on H3.
<MoeIcenowy>
maybe they're pin-compatible
<apritzel>
MoeIcenowy: so for Linux try to go the h3 compatible strings
<MoeIcenowy>
I will try a H3 32-bit SPL on it
<ssvb>
MoeIcenowy: just checked the datasheet, the SPI0 pin function is 3 on H5 (sames as H3), while on A64 it is 4
<MoeIcenowy>
yes
<apritzel>
so any Pine64/A64 stuff will not work out of the box
<MoeIcenowy>
seems that the pins are really compatible
<MoeIcenowy>
H3 and H5
<wens>
all the pins? and all the functions?
<wens>
otherwise you are still going to need a new pinctrl driver :|
<MoeIcenowy>
I have temporarily no thoughts to check it
Worf has joined #linux-sunxi
<MoeIcenowy>
wens: H5 datasheet described a "IO Disable" status
<MoeIcenowy>
on all pins at function 7
<ssvb>
apritzel: I wonder what is the boot priority on H5, it would be not very nice if SPI happens to have the highest priority and there is no FEL button :-)
<wens>
MoeIcenowy: all the newer ones (iirc A80+) have that
<wens>
MoeIcenowy: but we don't care about that
<MoeIcenowy>
PA checked pass
<MoeIcenowy>
some functions are added for H5.
<MoeIcenowy>
they are backward compatible like A10 to A20
<apritzel>
ssvb: I guess someone will find out (one or the other way)
<apritzel>
ssvb: my board is scheduled for delivery tomorrow
<wens>
ssvb: the first brickable allwinner device :p
<MoeIcenowy>
I think H64 is the first brickable
<ssvb>
is it?
<MoeIcenowy>
According to apritzel's experiments on Banana Pi M64 and Jide Remix Mini
<MoeIcenowy>
H64 seems to make SDC2 the first
<apritzel>
MoeIcenowy: but it has a proper FEL button
<apritzel>
MoeIcenowy: and I can access the eMMC from a FEL loaded U-Boot
<apritzel>
and that's different between BPi-M64 (A64, SDC0 first) and Remix Mini (H64, SDC2 first)
<wens>
semi-working PSCI patches for A80 posted
gianMOD has joined #linux-sunxi
<wens>
apritzel: any idea how the gic returns the secure copy of registers when the system should be in non-secure?
<MoeIcenowy>
will mainline U-Boot SPL check the SoC ID?
<apritzel>
MoeIcenowy: I don't think so ...
<oliv3r>
ssvb: bah. even @ 408 MHz with only 65 degrees it crashes
<oliv3r>
so gonna lower it to 384 and see if it can remain stable for 2 days, and if not, lower the freq (with 1.4v) to exclude that factor
<apritzel>
wens: which registers specifically?
<wens>
apritzel: GICC_BPR
<apritzel>
wens: IGROUPR for instance is RAZ/WI in non-secure
dpg_ has quit [Ping timeout: 245 seconds]
<apritzel>
GICC_BPR is banked between secure and non-secure
<apritzel>
so if you set it up, you have to do it on both sides
<apritzel>
wens: (but Linux doesn't care about this one, IIRC)
Putti has joined #linux-sunxi
f0xx has joined #linux-sunxi
<ssvb>
a funny thing is that sun8i means cortex-a7 and sun50i means cortex-a53 in the allwinner's naming scheme
<ssvb>
and now we have the old allwinner's naming convention, the new allwinner's naming convention, and the mainline linux naming convention
<ssvb>
all of these are different
<oliv3r>
annoying
<oliv3r>
what are the sunNi names for teh A80, A83T and A64 in mripards version? they are nameless ...
<oliv3r>
ssvb: the allwinner 2013 namingscheme i was aware off; but i think we pretty much all ignore that
<apritzel>
ssvb: how are they different in mainline Linux?
<wens>
oliv3r: they are sun9i, sun8i-a83t, and sun50i-a64
<oliv3r>
ah, but ont in that linked README :)
<ssvb>
apritzel: it's a mix of the old allwinner's scheme and the new allwinner's scheme, plus some improvisation
OtakuNekoP has joined #linux-sunxi
<ssvb>
now I wonder what would be the name for h5?
<apritzel>
ssvb: but it's not contradicting, AFAICT
<ssvb>
sun8i or sun50i?
<wens>
the kernel version drops the trailing "wXpY"
<apritzel>
new SoCs go with the new Allwinner scheme
<wens>
ssvb: sun50i-h5?
<mripard>
ssvb: what did I "invent" exactly?
<apritzel>
old SoCs just stick with the existing scheme
<ssvb>
mripard: read the backlog
<apritzel>
mripard: I think nothing ;-)
<mripard>
I did.
<jski>
I am trying to test if an ip or domain name resolves using ping or arping. Is there a way of making it timeout every 1 second saying no reply if it fails?
<apritzel>
mripard: ssvb: seems just like a misunderstanding to me ...
<ssvb>
mripard: do you realize that sun8i means cortex-a7 in the new allwinner's naming scheme, rather than the soc family?
<wens>
jski: ping -w X?
<ssvb>
and pretending that it is the soc family is a really silly idea
<jski>
does not work. maybe just using nc to see if a port is open might work
<jski>
then I can get Connection refused
<apritzel>
jski: -c 1 ?
<mripard>
ssvb: coming up with an entirely different naming scheme isn't very smart either.
<mripard>
ssvb: but I'd be delighted to have patches instead of complaints and reproaches
gianMOD has quit [Read error: Connection reset by peer]
gianMOD_ has joined #linux-sunxi
<jski>
problem is I am trying to check if dns is working with a valid nameserver
<oliv3r>
i'm not sure if mripard invented it; i think he mearly documented it? We have had the community naming for a while, so i'd call it the linux-sunxi naming scheme :)
<beeble>
jski: why not use dig or nslookup?
<jski>
I think if I just try connect to the nameservers port first
<wens>
iirc there was already a compromise as apritzel wanted to drop the sunXi part completely?
<jski>
not build in to this minimal rootfs
<ssvb>
mripard: who cares about making you delighted?
<oliv3r>
boys!
<oliv3r>
mripard: i've tried to cc you in a u-boot mail, but i misstyped your e-mail. sorry!
<oliv3r>
mripard: you should have gotten it via dev@linux-sunxi.org however
popolon has joined #linux-sunxi
<oliv3r>
mripard: it was a pachset I was sitting on for almost a year; but I just saw you sent an eeprom uclass patchset which much intregued me
Worf has quit [Quit: Konversation terminated!]
dh1tw has joined #linux-sunxi
tkaiser has joined #linux-sunxi
tkaiser has quit [Ping timeout: 268 seconds]
f0xx has quit [Ping timeout: 248 seconds]
gianMOD_ has quit [Remote host closed the connection]
f0xx has joined #linux-sunxi
gianMOD has joined #linux-sunxi
<NiteHawk>
ssvb: would you consider it reasonably safe if FEL "scratch" code (at sram_info->scratch_addr) uses up to 0x400 bytes (256 32-bit words)? afaict the combined size of scratch code (growing 'upwards') and FEL stack (growing 'downwards') should be 0x1000 for all the memory layouts we currently use
<NiteHawk>
tkaiser: is that bin2fex output? The character encoding of (chinese) comments looks broken
<KotCzarny>
according to description its from some lubuntu image
<tkaiser>
NiteHawk: Nope, that's serial console doing 'cat /boot/sys_config.fex'. I tried to include comments
<NiteHawk>
okay, i just wondered if fexc might need some more polishing :)
<KotCzarny>
;extremity_freq = 1344000000
<KotCzarny>
cutie pie
<KotCzarny>
and either something got broken, or things near [usbc1] are corrupted
<KotCzarny>
unless its just utf weirdness
<KotCzarny>
nope. apparently serial transfer got corrupted
<tkaiser>
KotCzarny: True, that's broken. But the fex itself is broken too (dvfs settings are wrong)
<KotCzarny>
you might dump again, to see if its repeatable and some content encoding interferes
<KotCzarny>
or just some serial noise
<oliv3r>
mripard: 2 patchsets: "Do not force master mode on unaffected PHY's" and "[PATCH v3] Retrieve MAC address from EEPROM"
<tkaiser>
I'll send a new PR, just booting the image again
<NiteHawk>
tkaiser: assuming the boot partition is reasonably small, maybe 'raw' image it and try to extract the .fex directly (e.g. by loop mounting)?
<NiteHawk>
(and btw: yes, sunxi-boards seems to be in a sad state regarding issues/PRs)
<MoeIcenowy>
The only thing I made my Orange Pi PC2 to run is, uart0-helloworld...
<KotCzarny>
MoeIcenowy: :)
<MoeIcenowy>
So It's a development board -- a board in development
<KotCzarny>
those android images didnt run?
Putti has joined #linux-sunxi
<MoeIcenowy>
I didn't download them
<MoeIcenowy>
only the ones on Baidu Pan is right
<MoeIcenowy>
but Baidu Pan will restrict your speed without using their nearly-malware client
<MoeIcenowy>
I get only 20KB/s speed
<MoeIcenowy>
I now do not dare to declare I'm in China :-(
<KotCzarny>
run it inside some vm?
<miasma>
is it due to the firewall or why connections to china are so slow
<MoeIcenowy>
The file on the Google Driver link is called "sun8iw7p1_android_dolphin-p1_uart0._PCPlus.rar", but on BaiduPan is "OrangePiPC2.rar"
<miasma>
steal sources -> develop your own patches on a live system -> once it works (kind of), wrap it up and upload
<MoeIcenowy>
even if I use the others' image to derive, I will still automatically, and won't use interactive bash
<miasma>
tkaiser: but didn't xunlong's steven (?) design some lemaker board?
<MoeIcenowy>
so not professional the guys is :-)
<MoeIcenowy>
miasma: I think the answer is no...
<miasma>
ok
<KotCzarny>
wasnt it about original banana?
<miasma>
yea, about some banana pi board
<KotCzarny>
not some, original banana baby
<MoeIcenowy>
but it's said that the one behind Xunlong is the designer of banana
<MoeIcenowy>
[citation needed]
<tkaiser>
NiteHawk: Thanks for heads up, script.bin is different compared to sys_config.fex
<miasma>
so is steven a techie or some pr/marketing guy? seems weird he's personally answering on aliexpress
<MoeIcenowy>
miasma: give me an example :-)
<KotCzarny>
afair someone on this channel said he done ODM work for original banana
<KotCzarny>
whatever it is
<MoeIcenowy>
maybe he is only using too many Chinglish like me
<miasma>
MoeIcenowy: i asked him about the opi pc board i got since the sd slot was broken when it arrived. he said 'hello friend' and then told he forwards the message to the engineers
<miasma>
and they told me to try other sd cards, which didn't help at all
<MoeIcenowy>
I think steven used "we" not "I"
gianMOD has joined #linux-sunxi
<miasma>
ah yes
<miasma>
"Hello friend, Thanks for your info,we will check with our engineer and give you reply. Thanks"
mhlavink has joined #linux-sunxi
<KotCzarny>
miasma, could be that his inbox being handled by secretary ;)
<MoeIcenowy>
good problem
<MoeIcenowy>
jernej: forgot to try your driver... sorry
<MoeIcenowy>
too excited for H5
<MoeIcenowy>
do I need any extra settings to enable the U-Boot con?
<MoeIcenowy>
but opi0 is nothing exciting except the Allwinner Wi-Fi chip
<MoeIcenowy>
It's now the time to mainlinize the board...
<tkaiser>
MoeIcenowy: From a practical point of view the PoE option is amazing, either add zero ohm resistors and combine it with a central 6V PSU or use a higher voltage and buck converters in between. Total costs for an OPi Zero node with passive PoE (and soldered SPI flash ;) ) = less than $10
<MoeIcenowy>
PoE?
<MoeIcenowy>
I do not know IoT well...
<IgorPec>
Power over ethernet
<MoeIcenowy>
oh!
<NiteHawk>
tkaiser: it would still be nice to preserve the .fex comments (assuming they might provide some useful insights), maybe the two could be diffed/merged somehow?
<MoeIcenowy>
a good IoT station...
<MoeIcenowy>
although the first usage of it thought by me is... Wi-Fi AP...
<MoeIcenowy>
(nearly all my networks have no wired LAN
<zoobab>
requests.exceptions.ConnectionError: HTTPConnectionPool(host='pan.baidu.com', port=80): Max retries exceeded with url: /s/1midMu96 (Caused by NewConnectionError('<requests.packages.urllib3.connection.HTTPConnection object at 0x7f5a424430d0>: Failed to establish a new connection: [Errno 110] Connection timed out',))
<tkaiser>
miasma: I don't know what's wrong with passive PoE. But sure you can add active PoE components for $50 per port to a $7 board as well ;)
OtakuNekoP has joined #linux-sunxi
yann-kaelig has quit [Client Quit]
yann-kaelig has joined #linux-sunxi
<KotCzarny>
tkaiser: or connect 200eur hd to it ;)
Andy-D__ has joined #linux-sunxi
<miasma>
tkaiser: it might work just fine, but what if the board breaks and you can't find a new one. you might need extra components to filter and collect the dc
Andy-D_ has quit [Ping timeout: 260 seconds]
<tkaiser>
miasma: Which board should break?
<miasma>
e.g. the OPi zero
<tkaiser>
IgorPec: Strange, on Armbian dhd driver seems to 'win': http://sprunge.us/gHME
<tkaiser>
miasma: Buy another one? Buy 10? And in case your replacement board doesn't have these PoE option add $2 for a passive PoE splitter?
cnxsoft has quit [Ping timeout: 245 seconds]
<zoobab>
homemade POE with 7805 and beer cap as heatsink, been there
<miasma>
i'd love to have POE, but that would basically mean building a new house. none of the existing ones have cat5/6 cables
<tkaiser>
IgorPec: I also made another change to the fex file. So all that's now to do is to add firmware files, adjust paths and figure out why dhd module 'wins'?
DullTube has quit [Quit: Leaving]
gianMOD has quit [Remote host closed the connection]
<zoobab>
"Allwinner XR819 is a completely new WiFi module"
gianMOD has quit [Remote host closed the connection]
reinforce has joined #linux-sunxi
orly_owl_ is now known as orly_owl
orly_owl has quit [Changing host]
orly_owl has joined #linux-sunxi
tkaiser has joined #linux-sunxi
Putti has joined #linux-sunxi
<jski>
I have a microsd card which I am setting up for the cubietruck. I dd it clean, dd bootloader, create partitions, mkfs, tar x the bootfs and rootfs then unmount and it contains the old boot files not the new ones
<jski>
so I remount it go into the directory and delete them so now it is empty tar x it and again the same thing
<jski>
:X
gianMOD has joined #linux-sunxi
<jski>
yes I sync also :P
<miasma>
does your dmesg show any problems
<jski>
yes interesting.. I created it as ext2 and it was fat..
<jski>
FAT-fs (sdb1): Volume was not properly unmounted. Some data may be corrupt. Please run fsck.
<miasma>
is it authentic sd or some cheap chinese counterfeit product
<jski>
samsung evo u1
<miasma>
what's the block size in fdisk
<jski>
ah not what I created so I think it is not allowing me to change the partition
<miasma>
some guy in arch-linux arm channel had some issues cause the card had 4k blocks instead of 512b or so
<miasma>
did your system say that kernel is still using the old partition table
apritzel has joined #linux-sunxi
<jski>
yes
<jski>
how do I see block size ?
<jski>
/dev/sdb1 2048 206847 204800 100M b W95 FAT32
<jski>
204800 sectors
<miasma>
with p command it mentions it few times
Andy-D__ has quit [Ping timeout: 245 seconds]
<jski>
Units: sectors of 1 * 512 = 512 bytes
<miasma>
ok, so it's a normal sd
<jski>
fix it with a hammer?
netlynx has joined #linux-sunxi
<miasma>
well if the kernel is still using the old partition table, you might need to unplug the card before creating the new fs, if nothing else helps
<miasma>
you could list the partition after each command to see which one overwrites it
gianMOD has quit [Remote host closed the connection]
<jski>
haha if fdisk o then w and exit then fisk and p it shows both old partitions
<miasma>
ok that's weird. does it have some readonly switch?
gianMOD has joined #linux-sunxi
<jernej>
MoeIcenowy: You must add CONFIG_VIDEO=y to the U-Boot defconfig in order to test this
<miasma>
jski: might be a broken card
<jski>
I think so
<jski>
thing is I think I have 3 of these
<jski>
this must of occured from flashing a certain image
<jski>
found a good one so will hit the other with hammer
<jemk>
jernej: you are setting the divider in ccm->hdmi_clk_cfg. does that really work? i had the impression this register wouldn't do anything on my h3, even the enable/disable didn't change anything.
<jernej>
jemk: It works, why wouldn't it? There is no limitations stated in user manual.
<jernej>
anyway, you can test my code
<jemk>
just write some arbitrary divider there, or even disable it, it didn't change anything on my board, display still worked
<jernej>
hm... 1080p resolution needs 148.5 MHz, which is pll3 at 297 MHz with hdmi_clk divider /2
<jernej>
and it works...
<jemk>
set it to /3 and it will still work
<jemk>
at least that is what happens here
<jernej>
let me test
IgorPec has joined #linux-sunxi
<jernej>
true...
<jemk>
thats why i concluded h3 hdmi would use pll3 directly in the wiki
<jernej>
so only tcon needs exact value? That would make sense
camh has quit [Ping timeout: 260 seconds]
apritzel has quit [Ping timeout: 248 seconds]
<jemk>
h3 has a divider in the phy
<jernej>
But I guess that phy doesn't need exact divider, from my experience
gianMOD has quit [Read error: No route to host]
<jemk>
it seems to somewhat work with wrong divider, but the image gets distorted and the refresh rate is wrong
<jernej>
1280x1024 is reported working
<jemk>
(i tried to modify the mainline driver to support arbitrary resolutions)
<jernej>
but I'm not sure about refresh rate
<jemk>
you can however just set the exact divider in 0x10030 bits 3:0 instead of their fixed switch
<jernej>
yes, I noticed that
<jernej>
but what about other values in that function?
<jernej>
why are they similar, but different?
<jemk>
all the other different parameters aren't linked to the divider, but to the resulting pixel clock
<jemk>
with higher clocks some parameters need to be adjusted to get clean signals
orly_owl has quit [Quit: leaving]
<jemk>
that's only guessed however, i neither have documentation
orly_owl has joined #linux-sunxi
iamfrankenstein has joined #linux-sunxi
<jernej>
BTW: did you noticed that in sun8iw11 hdmi blob there is another magic number for PHY, which seems to unscrumble register addresses?
<jernej>
doesn't work on H3
camh has joined #linux-sunxi
plm has joined #linux-sunxi
<jemk>
oh, didn't know that yet. would have been nice if it would work
<jernej>
jemk: do you think it is better to wait a bit, and try to understand those registers better, before trying to merge driver to mainline U-Boot?
<jernej>
register 0x10014, value 0x42494E47
<jemk>
i don't know if we ever will understand the phy registers
<plm>
Hi all
<jernej>
they used it for writing to continous CEC buffer
<plm>
The linux-sunxi exists becouse allwinner to not back changed code to community?
<jemk>
jernej: hm, "BING", the readlock was "TREE" and "WATR"
<plm>
Let me explain: I would like to use orangepi products, but are there very poor support of allwinners processor in the community.
<tkaiser>
plm: Poor support? Really?
<jernej>
jemk: yeah, I noticed that. So it seems that PHY block is completely custom made
<jernej>
although I'm still optimist that they just modified some other IP block
<tkaiser>
BTW: OPi Zero with 'IoT settings' (DRAM downlocked as hell, USB disabled, only Ethernet active) idles below 500 mW (yeah mW, that's below 100 mA)
<plm>
tkaiser: the support for the community for the allwinners products is very good?
<jemk>
jernej: i think its some glue block, the phy might be bought, i doubt they did it themself, but i havent found it elswhere yet
Gerwin_J has quit [Quit: Gerwin_J]
<tkaiser>
plm: I would call it excellent. But support in vendor/manufacturer forums pretty much sucks.
<miasma>
plm: depends on the metric.
<jernej>
jemk: and register addresses may be also scrambled, which make searching much more harder
<plm>
So, I have two questions: if is excellent Why this afirmation in the sunxi site: "Allwinner does not actively participate in or support this community. In fact, it is violating the GPLv2 license in several ways and has so far not shown willingness to resolve this"?
<plm>
tkaiser: miasma ^
<jernej>
jemk: do you know anything about TV? It seems similar, if not the same, as in other SoCs
<miasma>
plm, allwinner is one part of the supply chain. they design the SoCs. if you need support from the SoC vendor, you should probably buy commercial support from them
<plm>
Wait, but if I use a allwinner processor and I want to use linux on it, I need support, right?
<plm>
miasma: ^ from where will be that support?
<plm>
allwinner or sunxi?
<miasma>
i can't tell what kind of support you might need
<plm>
drive support on linux, kernel, etc, etc
<plm>
*drivers
<jemk>
jernej: I didn't look at the tv encoder, but i would expect it to be the same, why should they add new stuff for same function
<plm>
linux full support
<miasma>
plm: you can take a look at e.g. armbian and if it solves your problem, use their forums. if not, maybe the allwinner's kernel is better
<miasma>
or some of the official opi images
<plm>
miasma: And I read so much on the internet that allwinners has poor support, like this:
<jernej>
jemk: well, they did change hdmi... But it should be better than before
<jemk>
jernej: and it doesn't look like the phy is scrambled
<jemk>
it has 4K now
<jernej>
how can you tell that it isn't scrambled?
<jernej>
well, why do you think so
<jemk>
the registers have a somewhat meaningful order
<jemk>
the controller didn't have it, took me quite a headache to figure out what it is
<tkaiser>
plm: Just read this: https://hackaday.com/2016/11/07/orange-pi-releases-two-boards/ -- the whole blog post as well as 90 percent of comments is BS. While those guys do Allwinner bashing we have OPi Zero up and running with full hardware support. Mainline kernel (up to 4.9) is already running on it. It will take maybe a day or two until Wi-Fi driver is also ported to mainline.
<tkaiser>
plm: Ah, nice. You were already referring to this BS
<miasma>
plm: with allwinner stuff, you can't assume that everything works 100% and everything is supported for free. you could try if it works for you and if you run into problems, then people can try to help you
<plm>
look at the end of post: "When you buy a single board, though, you’re buying into a community dedicated to improving Linux support on the board. From what I’ve seen, that support probably won’t be coming but I will be happy to be proven wrong."
<tkaiser>
plm: Look not at BS blog posts but better at reality. I run mainline kernel on Orange Pi products since Dec 2015.
<jemk>
jernej: now that its known its a synopsys controller it looks quite obvious, but it wasn't easy at all to figure that out
<miasma>
plm: yes, you probably won't find 10 million tutorials for turning gpio led on and off for each allwinner board :S
<jernej>
jemk: I know, I still wonder, how did you figure this out.
mzki has joined #linux-sunxi
alexvf has quit [Quit: Page closed]
<jernej>
jemk: do you know any possible explanation, why Allwinner paired same HDMI controller with different phys?
<plm>
miasma: tkaiser I just starting with allwinner (using for now orangepi), so tell me if I talking something wrong. But I don't need '0 milion tutorials about turn on/off a led. But just the main line kernel and are there alldrivers for all hardwares feature on linux mainline, and/or patches where I can works everything on allwinners.
<tkaiser>
plm: And that components are routed as they're routed on OPi PC 2 has a simple reason: Efficiency, all connectors placed identically as on Orange Pi PC or PC Plus and still some room to add Wi-Fi and eMMC later when a 'Orange Pi PC 2 Plus' will be released (my assumptions)
<jemk>
jernej: by drawing a lot of graphs with binary addresses that would make sense to be next to each other, and a lot of luck
<jemk>
jernej: i think they only had the a83t phy for 28nm from a80, but h3/a64 aren't 28nm
<miasma>
tkaiser: so h2+ is basically very much like h3? would it be safe to add h5 and h2+ to the mainlining status matrix
leviathanch_ has joined #linux-sunxi
<tkaiser>
miasma: At the moment it seems H2+ is just a H3 with two features disabled: Gbit Ethernet MAC and 4K HDMI output. I wouldn't differentiate between both.
<ssvb>
plm: about that hackaday blog, the dude just wants to be a drama queen
dpg_ has joined #linux-sunxi
<ssvb>
plm: basically he is just saying that he is even incapable of finding a usable sd card image for the board, but just picks some random crap and blogs about it
leviathanch has quit [Ping timeout: 252 seconds]
<plm>
I would like allwinners becouse is cheaper than other and has excellent performance, right both afirmations?
<ssvb>
plm: also very new boards tend to have a bit rocky start, so it's unfair to judge them at this point
<plm>
Actually I do not use orangepi(allwinners) yet, but I would like to start with it
<miasma>
plm: rpi zero is only $5 and you can build a cluster of rpis :P a cluster of 10 might be able beat opi pc
<jernej>
jemk: but h3 and a64 doesn't have same phy.
paulk-collins has joined #linux-sunxi
<ssvb>
yes, the Raspberry Pi people use a fruity strategy by keeping everything secret and only announcing their products when they have everything ready
<ssvb>
this prevents them from having negative early feedback
<jemk>
jernej: they have, the u-boot code is wrong, look at the kernel code
<plm>
yes, that strategy is real
<plm>
oragepi not, release new hardware and drivers still in development
<jemk>
jernej: also, a80/a83t have 1.8v for hdmi, which is what synopsys phys want. h3 and a64 have 3.3v for hdmi, which is easier because they have it on the board anyways, but synopsys only offers 1.8v and 2.5v (at least publicly)
<plm>
anyone know about the orange pi camera?
<plm>
in teh orangepi site dont have this product, but are there to buy on others sites, like as alibaba
<jernej>
jemk: do you know for any source code of those blobs? I quickly looked at disassembly of blob from longsleep's repository and it seems to be identical to that from u-boot
<ssvb>
plm: orange pi boards are officially sold at aliexpress, and they are mostly *hardware* people, so you can't expect good *software* support from them
<tkaiser>
plm: Sure, but here you're amongst developers. Better check out Armbian forum. Community members fixed the camera driver, Armbian includes this fix already and he can even make use of hardware accelerated video encoding.
<jemk>
jernej: noticed there are only blobs right now too, but i was pretty sure i had found it somewhere
<ssvb>
but it's perfectly fine as long as they design decent hardware and don't charge premium price for it
<plm>
Not, I mean: why orange pi camera is not in the orangepi.org site together all other products (boards) to buy and see the sepcifications?
<jernej>
jemk: there is another (wishful) explanation: both phys are present and one is chosen by magic numbers :)
<tkaiser>
plm: Really, we don't sell hardware here ;) From time to time some items disappear from Xunlong's aliexpress shop just to be added later. But that doesn't matter since you can also use other cameras with cheap H3 boards. Again check Armbian H2/H3 forum and search for OV5640 for example
<jemk>
jernej: switching between differnet digital parts, ok, but switching between phys? i doubt that, would be way to expensive
<jernej>
tkaiser: If you have zero at hand, can you check wifi SDIO id? It may be that it is the same as stated in dhd driver, which means they just buy same core and for some reason wrote new driver & firmware
seo has joined #linux-sunxi
seo has left #linux-sunxi ["Leaving"]
<tkaiser>
BTW: Finished with OPi Zero consumption measurements. When using the new Wi-Fi OPi Zero idles at 520 mW (with active powermanagement). So for the price of an absolutely featureless RPi Zero you get a hardware that consumes less and is as powerful as overpriced RPi 2
<jernej>
jemk: sorry, you are right, mine aarch64 reading skills are still bad. But why they would have non working hdmi driver in u-boot? Doesn't really make sense. I need to test it some day
lemonzest has joined #linux-sunxi
<jernej>
tkaiser: I checked my linux kernel build folder and I found following in dhd folder: MODULE_ALIAS("sdio:c00v*d*");
<jernej>
tkaiser: but anyway, other drivers, like 8189fs works, except if alphabetical order has something to do with it
<tkaiser>
jernej: The driver needs some polishing anyway. Always loads at 2nd try only. Anyway... OPi Zero will be my wired IoT toy from now on so I won't look into Wi-Fi any more :)
<jernej>
tkaiser: if you have time, you can remove line "{ SDIO_DEVICE_CLASS(SDIO_CLASS_NONE)}" from drivers/net/wireless/bcmdhd/bcmsdh_sdmmc_linux.c
<tkaiser>
jernej: No time, especially not now. But good to know
<jernej>
tkaiser: As you wish. I can test it, but I don't have Armbian build environment set up, so it will take some time... Maybe we could ask IgorPec?
tkaiser has quit [Ping timeout: 256 seconds]
yann has quit [Ping timeout: 260 seconds]
<jernej>
jemk: Same "mistake" is also made in H5 sdk. I also find content of the hdmi makefile to be interested (http://pastebin.com/DfYt5g2Y)
<jemk>
what a mess...
iamfrankenstein has quit [Ping timeout: 245 seconds]
lemonzest has quit [Quit: Leaving]
HeavyMetal has quit [Ping timeout: 256 seconds]
HeavyMetal has joined #linux-sunxi
HeavyMetal has joined #linux-sunxi
HeavyMetal has quit [Changing host]
|Jeroen| has joined #linux-sunxi
jernej has quit [Ping timeout: 240 seconds]
<plm>
Why the allwinners SoCs are cheaper than others?
<Keziolio>
they save on mainline linux support
<NiteHawk>
push-into-market-and-forget
<vpeter>
They count on #linux-sunxi for a sw support :)
<KotCzarny>
and they make cheap, good enough socs that are usable for specific needs
<plm>
ssvb: "they save on mainline linux support" and "push-into-market-and-forget"
<ssvb>
btw, some years ago I had an argument with a pandaboard guy in the pandaboard irc channel
dpg_ has quit [Ping timeout: 260 seconds]
<plm>
ssvb: pandaboard use omap5
<ssvb>
he was extremely unhappy about odroid being much cheaper and beating them on market
<ssvb>
and talking some bullshit about the asian sweatshops, child labour, etc.
<ssvb>
and at the same time bragging that pandaboard sponsored open source conferences, etc.
<plm>
ssvb: I know, but odroid has better support than orange pi, it use sansumg
<ssvb>
so I just told this guy that I was not sure if I was really happy about some fraction of *my* money paid for my board being spent on open source conferences sponsorship :-)
<ssvb>
and he went ballistic after that, calling me an enemy of open source, etc. :-)
<ssvb>
it's always good to have a choice
<Keziolio>
he was right tbh
<Keziolio>
allwinners are almost unusable
<ssvb>
Keziolio: can you elaborate about them being almost unusable?
<Keziolio>
those are smartphone chips, a gpu with a cpu attached
<ssvb>
which Allwinner SoC are you talking about?
<Keziolio>
H3 in my case
<ssvb>
Allwinner chips are not used in smartphones
<Keziolio>
and with mainline kernel you have a bugged headless server at most
<KotCzarny>
define bugged
<Keziolio>
sure it works "as advertised" but you can't say the support is good
<ssvb>
huh, that's a rather bold claim
<ssvb>
H3 works just fine as a headless server with the mainline kernel
<Keziolio>
KotCzarny: the emac driver is out of tree, it works but it has some glitches, i.e. you have to disconnect and reconnect the cable sometimes to get it to work
<KotCzarny>
keziolio: you must be out of the loop for some time
<Keziolio>
i don't think so, unless you got it mainline in the last rc
<KotCzarny>
mainline isnt the lates
<KotCzarny>
t
<Keziolio>
lol
<ssvb>
and for graphics and multimedia you could always use the legacy kernel
<Keziolio>
KotCzarny: what is the "latest"?
<KotCzarny>
and if you have a bugreport, please ring wens
<KotCzarny>
and please, if you find bugs with it, make a bug report instead of just complaining in the air
iamfrankenstein has joined #linux-sunxi
<Keziolio>
yes KotCzarny i'm probably running a kernel tree from wens
<ssvb>
Keziolio: or buy some beer and peanuts with the saved money, and then bullshit in blogs and forums ;-)
<Keziolio>
k
apritzel has joined #linux-sunxi
<plm>
well, if sunxi community are working to put in mainline kernel all features of allwinners, that is fine
dpg_ has joined #linux-sunxi
<apritzel>
ssvb: MoeIcenowy: I compared the H3 and H5 datasheets
<apritzel>
indeed the H5 is pin compatible to the H3
<apritzel>
I checked every single one of them ;-)
<apritzel>
few exceptions, though:
<apritzel>
the H5 has a separate VCC_PC pin (G15), this is connected to VCC PA on the H3
<apritzel>
this is probably to be able to drive SDMMC2 with 1.8V support HS400
<plm>
ok if sunxi don't have support from allwinner, bt the questin is: the documentation about allwinners SoCs is good, very good, excellent do to a excellent opensource (linux..) support?
<apritzel>
plm: "enough" ;-)
<plm>
apritzel: ok, enough is enough :)
<ssvb>
the documentation is not very good (written in somewhat broken English), but it surely exists
<apritzel>
actually bad, but we managed to get something out of it anyway
<plm>
all right, so, that is enough
<ssvb>
I don't know why some dudes are persistently posting bullshit about having no documentation at all
<KotCzarny>
hardware: okayish, vendor support: bad, community support: excellent
<KotCzarny>
;)
tkaiser has joined #linux-sunxi
<ssvb>
and some dudes are even posting bullshit about the allegedly "stolen" documentation
<apritzel>
also (most?) boards are bascially unbrickable, and firmware can easily be replaced
<KotCzarny>
ssvb: maybe there is a need for /debunk_myths in linux-sunxi wiki?
<apritzel>
I think the firmware argument is pretty important, other SoC families suffer from being stuck with bad vendor firmware
<ssvb>
the anti-allwinner hysteria is really ridiculous and attracts all kinds of crazy people who are readily willing to spread rumors
<apritzel>
yeah, clueless people ...
<KotCzarny>
oh, and put it in the topic if you write such page
<KotCzarny>
:)
tkaiser has quit [Ping timeout: 240 seconds]
<apritzel>
ssvb: so with the H5 pin compatibility to H3, I wonder if that stretches into the software world, so if one can boot an H3 image on it ...
<apritzel>
(32bit of course only, but still)
<ssvb>
we'll see when we get the boards
<ssvb>
mine is also supposed to arrive tomorrow
<montjoie>
Keziolio: from where did you get emac driver ?
<ssvb>
DHL is just a little bit inconvenient because they usually want to deliver parcels to my door (it means that I need to be at home), and they also can take care of the customs/VAT formalities for a fee
<ssvb>
just picking up parcels at the post office is much more convenient :-)
<Keziolio>
montjoie: lol
<Keziolio>
montjoie: it was yours
<montjoie>
v5 ?
<Keziolio>
yes
<montjoie>
so you said to have some disconnect ? anything in dmesg ?
<Keziolio>
i'll try to reproduce
<Keziolio>
ok, booting with the ethernet cable attached does not work
<KotCzarny>
keziolio: did you try booting with different device at the other end?
<Keziolio>
i don't think i have emac support if it's a recent thing
<Keziolio>
in the uboot*
<KotCzarny>
remember, devices sometimes are refusing to work together
<Keziolio>
if I attach the cable after boot it works
<KotCzarny>
so try different switch etc
<KotCzarny>
also, it reminds me of pine64, where it also didnt boot with cable attached :>
<KotCzarny>
though in your case its connection not working and boot is ok?
<ssvb>
KotCzarny: "you are holding it wrong"? ;-)
<Keziolio>
maybe it's just a distro problem idk
<Keziolio>
boot is ok
<KotCzarny>
ssvb: happens with bad cables ;)
<montjoie>
Keziolio: your 3 last mac digit asaid that you have an old uboot
<Keziolio>
ok
<ssvb>
KotCzarny: yes, bad ethernet cables are a reality, I have a few of such cables in a garbage box
<KotCzarny>
sometimes its about the length of the cable too
<KotCzarny>
my bpi-r1 dislikes short cables
<KotCzarny>
and by short i mean <1m
arete74 has quit [Ping timeout: 260 seconds]
<montjoie>
Keziolio: after boot, do you have some packet loss ? ethtool -S eth0 after some time could be usefull
BenG83 has joined #linux-sunxi
<Keziolio>
i'll recompile u-boot, it whines about ethernet indeed
<Keziolio>
to make it to work i have to ip link set eth0 down, remove cable, attach cable, set up
<Keziolio>
i have "rx_payload_error: 44" that slowly rises
<Keziolio>
regarding ethtool
<Keziolio>
but the connection works
netlynx has quit [Quit: Ex-Chat]
leviathanch_ has quit [Remote host closed the connection]
f0xx has quit [Ping timeout: 258 seconds]
<Keziolio>
ok fixed
<Keziolio>
when this is coming to mainline linux?
<miasma>
plm: odroids don't have better support than allwinner chips. e.g. odroid c1 can't even mount root with mainline linux
<miasma>
the mainline doesn't support emmc/sd and probably not even ethernet
<vagrantc>
the exynos odroids have pretty good support
<miasma>
yea, but they're pretty obsolete now afaik
<vagrantc>
but the amlogic ones are still pretty poor
<miasma>
i have odroid-x and c1
<miasma>
with c1 I'm stuck with some old kernel
<vagrantc>
the odroid-XU4 is hardly obsolete...
<KotCzarny>
<keziolio: what fixed what?
<miasma>
yea true, but they had several older exynos models
<Keziolio>
updating u-boot to master fixed that
<Keziolio>
fixed the "boot without internet" problem*
<KotCzarny>
keziolio, then its not emac driver problem but user installation problem
<KotCzarny>
now some 'sorry' for the author of the driver wouldnt be bad ;)
<KotCzarny>
s/for/to/
<Keziolio>
lol i've never said anything to the author of the driver
jernej has joined #linux-sunxi
<Keziolio>
sorry montjoie i'll pay you a beer
<ssvb>
vagrantc: there is a firmware blob with unclear license, which is *required* for the odroid boards to boot, see http://git.denx.de/?p=u-boot.git;a=blob;f=doc/README.odroid;h=c088ec4cb08f76b91c6efce089b316416c0c6dc8;hb=HEAD
<ssvb>
vagrantc: is this what you call good support?
<ssvb>
vagrantc: how are you building and distributing this blob in debian?
florianH has quit [Quit: Connection closed for inactivity]
<vagrantc>
ssvb: no, the blob is outside debian
<ssvb>
how does it work in practice?
<vagrantc>
ssvb: but it runs mainline u-boot and linux ... the blobs are unfortunate, yes.
mzki has quit [Ping timeout: 240 seconds]
<montjoie>
Keziolio: I wait for an answer on v4 for sending v5
<ssvb>
vagrantc: are you sure that the mainline u-boot works on odroids (more specifically odroid-x)? the last time I checked it was severely broken
<ssvb>
vagrantc: but the u-boot from hardkernel works, I guess that's what most people use
<vagrantc>
ssvb: i run three odroid-xu4 and one odroid-u3 using mainline u-boot (well, slightly patche in debian to support distro_bootcmd)
<ssvb>
okay, maybe I'll check it again
<vagrantc>
the blob installation process is annoying
mzki has joined #linux-sunxi
<miasma>
ssvb: i supose some guys got burned when opi pc arrived, they bought the board and realized it didn't work out of the box with distros and new kernels. so now they're spreading BS. just a guess
<vagrantc>
overall, thanks to the linux-sunxi community, i've found the allwinner stuff to work better out of the box usually, except for the newest stuff
tkaiser has joined #linux-sunxi
<miasma>
ssvb: i bought opi pc when it was new, got it for $15 which was cheaper than now. i waited over a year and now it works :) it's magic
<ssvb>
vagrantc: well, this ugly odroid blob makes allwinner significantly better than exynos, imho :-)
<vagrantc>
ssvb: i hear ya
<ssvb>
miasma: except that it worked within a few months after the release, and it was documented in the linux-sunxi wiki
<miasma>
vagrantc: what defconfig should one use for odroid-x with mainline u-boot?
<vagrantc>
miasma: no idea with odroid-x
<miasma>
ssvb: that's highly subjective. i'd say there are several levels of support the board can have. e.g. some people don't want to patch kernels or use latest rcs. in that case the board is still missing features like audio and ethernet
tkaiser has quit [Ping timeout: 250 seconds]
<vagrantc>
miasma: doesn't look like odroid-x is supported in mainline u-boot
<ssvb>
miasma: you already could cherry pick enough patches to get a usable kernel at that time, also the things like the DRAM reliability and U-Boot have been taken care of
<ssvb>
when was the Orange Pi PC board released?
<miasma>
sure. but it depends on your needs if the kernel is recent enough
<miasma>
e.g. i might want to use overlayfs and btrfs which might be broken with older kernels
<ssvb>
yes, there were some moderately expensive H3 based board before, but the key to success is the low price because this way the board attracts a lot of developers
yann has joined #linux-sunxi
<ssvb>
now I'm expecting the same with the Orange Pi PC 2, and maybe it will become usable a lot faster
<ssvb>
the mainline kernel is developed at a slow pace though, and it takes a lot of time for the patches to land there
<miasma>
the competition is tougher now :) rpi3's cpu power is pretty decent
<ssvb>
rpi3 is still blob infested, it is more expensive, and the i/o performance is a joke
<miasma>
but they sell ssd shields
<ssvb>
:-)
<miasma>
you can install 300 MB/s mSATA SSD there
<ssvb>
are you sure? what kind of performance is expected from it?
<miasma>
but you need a gigabit ethernet dongle if you want it to work faster as a web server
<ssvb>
20-40 MB/s while also serving data over the ethernet?
<miasma>
possibly 20 MB/s as a (t)ftp server
IgorPec has quit [Ping timeout: 245 seconds]
<miasma>
but 10..15 MB/s might be a more realistic expectation
<miasma>
ssvb: with good usb-sata adapters you can get 37 MB/s and there was some benchmark with the realtek gigabit usb3 dongle. i think the rpi3 got around 350-400 Mbps. but when using both, they need to share the bandwidth :)
<miasma>
but it's still quite weak. i expect 35 MB/s from opi one as a ftp server :)
my123_ has joined #linux-sunxi
my123_ has joined #linux-sunxi
my123_ has quit [Changing host]
nashpa has quit [Ping timeout: 268 seconds]
nashpa has joined #linux-sunxi
my123 has quit [Ping timeout: 250 seconds]
<KotCzarny>
hehe, and that ssd shield costs ~50-80usd? ;)
<miasma>
i found one rpi3 store which charges 75,90€ for rpi3 + 4,10€ shipping
<miasma>
it's not even a starter set, only the board
IgorPec2 has quit [Ping timeout: 268 seconds]
<KotCzarny>
you know you can just buy marvel board at that point?
<jernej>
jemk: fun fact - if you check H3 driver code, you'll notice that phy constants are slightly different in comparison to previosly know BSP code
<KotCzarny>
oh, wow, almost recent uboot
<KotCzarny>
:)
<KotCzarny>
i wonder what will happen when they finally reach uboot with linux-sunxi contributions