<libv>
he mentioned phoenixcard, and buying a similar tablet from aliexpress, and then decompiling the nand kernel module.
<libv>
yes
<libv>
these people call themselves researchers.
<libv>
they clearly did not know the magic art of googling.
<libv>
but that is actually a bit of a blessing, as booting a full linux with an allwinner own kernel could've resulted in the nand getting corrupted due to the different setup of the randomization
<libv>
(like we had with a20)
<libv>
this person also talked about how they thought about a logical analyzer to see how they access the nand.
<libv>
wtf.
<libv>
google. like normal people.
<libv>
this part of the talk would've been structured quite differently if they had read up on our wiki and/or contacted our community for more info.
<libv>
i got there as i heard even more people brag about CCC at fosdem
<libv>
i was first pretty dismissive as fosdem records 24 tracks versus 4 at CCC
<libv>
then i looked through the content and found this, and then got curious as to which soc would be used
<libv>
and then i got to roll my eyes at the "technical" skill of some of the speakers
<libv>
i think i will not bother with getting access, a hotel, and a trainride there at the end of this year
<libv>
and now i know that there is little point and a lot of hype.
<muvlon_>
there's a lot more to CCC than tech talks
<muvlon_>
but I agree you might be disappointed if that's what you attend it for
[7] has quit [Ping timeout: 258 seconds]
TheSeven has joined #linux-sunxi
<libv>
muvlon_: perhaps, but this is just plain bad googling.
<libv>
"hi, we are hackers, we even talked at CCC, but we are too dumb to google"
<libv>
this i find pretty annoying.
<libv>
and it's not as if these were some isolated kids somewhere
<libv>
these were people given a very rare tablet smuggled out of north korea
<libv>
my personality prevents me from stating "i am not angry, i am just extremely disappointed", as this sort of stupidity really does make me feel very angry.
jbrown has quit [Ping timeout: 255 seconds]
JohnDoe_71Rus has joined #linux-sunxi
jbrown has joined #linux-sunxi
IgorPec has joined #linux-sunxi
<KotCzarny>
libv: lol
<KotCzarny>
linux-sunxi.com needs more exposure, who is good at PR and convincing (preparing ready to consume informations) to plenty of tablet/software bloggers?
<KotCzarny>
or just at SEO
<muvlon_>
I didn't find it hard at all to come across linux-sunxi
<muvlon_>
in fact if i google any of the popular sunxi boards (bpi, cubieboard, etc.) linux-sunxi is always in the top
<KotCzarny>
muvlon: yeah, me neither, actually coming here saving me from buying a83t board which is still unsupported (2 years or more)
<muvlon_>
if you actually google for "Allwinner {SoC Number}", linux-sunxi is pretty much always the top result
<KotCzarny>
yes, googling 'allwinner' makes it 4th result (but google might be skewed by previous searches, so should be done on virgin browser)
<dan0_0>
Dual cursors and cursor trails/ black squares that don't render
terra854 has joined #linux-sunxi
jbrown has quit [Ping timeout: 255 seconds]
IgorPec has joined #linux-sunxi
jbrown has joined #linux-sunxi
<wens>
MoeIcenowy: exactly
ErwinH has joined #linux-sunxi
massi has joined #linux-sunxi
msevwork has joined #linux-sunxi
<jelle>
wens: hi
<jelle>
wens: the nanopi air, uses the same Ampak AP6121 as the bananpi m2+, for which I think you added support (in dts). I couldn't however got it to work with a build-in brcmfmac driver. Any tips?
<MoeIcenowy>
jelle: at least please enable the MMC controller
<MoeIcenowy>
thus at least a card can be recognized
florianH has joined #linux-sunxi
apritzel has joined #linux-sunxi
<jelle>
MoeIcenowy: where in /proc or /sys does it end up then?
<MoeIcenowy>
in dmesg there will be something like "new SDIO card"
<jelle>
ok
<MoeIcenowy>
and at /sys/bus/sdio/devices there's an item
<jelle>
I think I didn't see that yesterday
<jelle>
MoeIcenowy: for me this is unknown territory :)
<MoeIcenowy>
the message should be "mmcx: new high speed SDIO card at address xxxx"
<jelle>
ok
<MoeIcenowy>
for a card to be recognized, please ensure it's properly powered and enabled
<MoeIcenowy>
you can refer to sun7i-a20-cubietruck.dts
<jelle>
MoeIcenowy: looked at the datasheet and it looked a lot like how the card was wired in the bananapi m2+
<jelle>
MoeIcenowy: thanks, I'll try that later this week
apritzel has quit [Quit: Leaving.]
<MoeIcenowy>
thus you can just check sun8i-h3-bananapi-m2-plus.dts ;-)
<jelle>
MoeIcenowy: that's what I did
<MoeIcenowy>
and have you got the eMMC working?
<jelle>
MoeIcenowy: well no
<jelle>
also not sure how I can access eMMC or boot from it. Couldn't find much info about it
<MoeIcenowy>
refer to m2-plus again ;-)
<MoeIcenowy>
especially see &mmc2 part of it
<jelle>
MoeIcenowy: oh I copied it
<montjoie>
I hate bpim2+
<MoeIcenowy>
montjoie: ;-)
<montjoie>
gigabit that does 100megabit
<jelle>
lol
<MoeIcenowy>
montjoie: even with BSP/sun8i-emac?
arete74_ has quit [Read error: Connection reset by peer]
arete74 has joined #linux-sunxi
<montjoie>
with sun8i-emac the max TX is 180
<montjoie>
randomly get 100 or 180
<MoeIcenowy>
how about BSP?
<montjoie>
not tested
<MoeIcenowy>
or maybe you just got a broken board?
<montjoie>
perhaps
<MoeIcenowy>
P.S. how is sun8i-stmmac now?
<montjoie>
I need to publish last version
<montjoie>
waited to retest it on pine64 since it is the only real gigabit board I got
<montjoie>
but the napi perf patch got nuked by dmiller:)
<KotCzarny>
montjoie: why so?
<MoeIcenowy>
First get it work, then enhance the performance ;-)
<montjoie>
Because I follow NAPI doc from linux-fundation and in fact he dont like that
<KotCzarny>
o.O
<montjoie>
but I found a similar way to do that in intel igb intree
<montjoie>
just I need a minor change
<montjoie>
and it will be acceptable I think
f0xx has quit [Ping timeout: 252 seconds]
<wens>
montjoie: what other boards do you have? i just re-did a83t ccu support
<montjoie>
I got also bpim3
<montjoie>
wens: I vote for A83T support:)
<montjoie>
without it, I need to refound where I put your old tree source for it
<jonkerj>
backlog of last month suggests that mripard has submitted a patch which fixes the 'mmc clk timeout' bug in 4.10, but I am failing to find the commit
<jonkerj>
backlog sais wens accepted it. Can anyone point me to the patch/commit, so I can check/fix my local repositories to backport that patch?
<jonkerj>
backlog also suggested that this would be a DT-only fix, but I don't care as long as it works :-)
<MoeIcenowy>
I wonder whether it is worth to buy a BPi M3 now
<tuxillo>
morning
<MoeIcenowy>
jonkerj: the correct fix is a DT fix
jbrown has quit [Ping timeout: 255 seconds]
<MoeIcenowy>
but this is tenporary ugly hack
<jonkerj>
do you have a reference to that as well?
<jonkerj>
or do you mean it should be fixed in DT somewhere in the future?
<MoeIcenowy>
yes
<MoeIcenowy>
thus this patch wlil be reverted once DT is fixed
<jonkerj>
ah, it even mentions that in de commitlog. In 4.11 things will be better
<jonkerj>
tnx
<wens>
MoeIcenowy: i don't know, but i already have 3 A83t devices, might as well get them working
<MoeIcenowy>
P.S. the current U-Boot EFI cannot boot a linux kernel well in non-secure mode
<wens>
which SoC?
<MoeIcenowy>
A33 and H3 failed
<MoeIcenowy>
but A10 works
<MoeIcenowy>
(Seems that it's because A10 do not have SMP ;-)
yann-kaelig has joined #linux-sunxi
<wens>
that just means the EFI firmware lacks the proper secure setup stuff?
<wens>
proper drivers for the hardware
<MoeIcenowy>
seems that the EFI implementation of U-Boot do not properly alter the dtb
<MoeIcenowy>
(e.g. simplefb section is not inserted
<wens>
it probably doesn't call the dt fixup functions in u-boot
<MoeIcenowy>
so maybe PSCI node is not installed?
<MoeIcenowy>
P.S. If the kernel is booted from a GRUB, thus things become worse
<MoeIcenowy>
even UEFI info is not properly inserted into dt
<agraf>
MoeIcenowy: this is armv7?
<agraf>
MoeIcenowy: iirc the grub patches to boot linux via usb stub in efi grub on 32bit arm were not accepted into grub yet
<agraf>
MoeIcenowy: so you're basically stuck with an efi grub (so from u-boot's pov it's all efi) which then suddenly switches into traditional legacy bootz style boot mode
<agraf>
MoeIcenowy: u-boot's bootefi doesn't create a simplefb device tree node, because the efi way to pass the frame buffer on is efifb (efi gop)
<wens>
sounds like a big oops
jbrown has joined #linux-sunxi
matthias_bgg has joined #linux-sunxi
<agraf>
MoeIcenowy: and you don't want to have the fb described in both simplefb and efifb :)
matthias_bgg has quit [Remote host closed the connection]
<agraf>
MoeIcenowy: as for the other failures, I wanted to have a look this week
<agraf>
MoeIcenowy: so H3 also fails? that's good news
<agraf>
afaerber: ^
matthias_bgg has joined #linux-sunxi
<wens>
olimex laptop uses edp converter, wonder which one
<MoeIcenowy>
agraf: I did some code change to efi_gop, to pass by the console of cfb_console
<agraf>
MoeIcenowy: you mean skip it?
<MoeIcenowy>
not skip
<MoeIcenowy>
the original efi_gop code only support the LCD framework
<agraf>
yes
<agraf>
i didn't realize we have more frameworks in u-boot
<agraf>
MoeIcenowy: so efi gop currently supports dm video and lcd
<MoeIcenowy>
yes
<agraf>
MoeIcenowy: could we not just move cfb_console to dm_video?
<MoeIcenowy>
montjoie: OtakuNekoP got 740Mbps on his BPi M2+ with 3.4 kernel
f0xx has joined #linux-sunxi
<MoeIcenowy>
maybe you should test with the stock image
OtakuNekoP has joined #linux-sunxi
<montjoie>
I will try
BroderTuck has joined #linux-sunxi
<agraf>
MoeIcenowy: so which board uses that thing?
<MoeIcenowy>
"that thing"?
<BroderTuck>
wens: according to an earlier blogpost it seems they use ANX6345 for eDP on the laptop
<beeble>
wens: checked the schematics, its anx6345
<beeble>
so it used the parallel display interface. so no (r)gmii
<MoeIcenowy>
you are talking about Olimex Laptop?
<MoeIcenowy>
if it's really parallel display interface, it is something interesting ;-)
<wens>
beeble: you mean lvds/mipi? :p
<MoeIcenowy>
(if it's DSI, it's disaster
<wens>
dsi is a blob :/
<wens>
beeble: do you (or anyone over there) know how to do sigma-delta modulation on the allwinner socs?
<MoeIcenowy>
wens: The A33 user manual have some info about DSI
* wens
needs a crash course on clock frequency fuzzing
<beeble>
wens: i have no idea. have to ask around
<beeble>
wens: and i mean rgb. a64 has an rgb and a dsi interface
<beeble>
it uses the rgb
<wens>
beeble: right, i meant rgb, as opposed to lvds or dsi
OtakuNekoP_ has joined #linux-sunxi
<beeble>
no lvds on a64
<MoeIcenowy>
P.S. on A64 RGB is muxed with EMAC (*MII)
<MoeIcenowy>
so Pine64 uses DSI screen
<beeble>
MoeIcenowy: thats why i said no (r)gmii :)
cnxsoft has quit [Remote host closed the connection]
<wens>
oh!
cnxsoft has joined #linux-sunxi
<MoeIcenowy>
and as A64 do not have integrated PHY, I think this means the EMAC is wasted
<beeble>
yes
chomwitt1 has joined #linux-sunxi
<OtakuNekoP_>
montjoie: the 10s test data, [ 3] 0.0-10.0 sec 904960 KBytes 741222 Kbits/sec, need longer?
<montjoie>
OtakuNekoP_: could you test sun8i-emac/dwmac-sun8i on it ?
<OtakuNekoP_>
montjoie: just tell me how pls :)
<montjoie>
never compiled a kernel ?
<MoeIcenowy>
montjoie: I think he never compiled a mainline kernel.
<OtakuNekoP_>
I only compiled the BSP. :(
<KotCzarny>
then give him kernel and modules pack
<MoeIcenowy>
montjoie: which U-Boot config do you use for BPi M2+?
<montjoie>
good question
<MoeIcenowy>
wiki suggests orangepi_plus_defconfig
<montjoie>
it seems I used sun8i-h3-orangepi-pc-plus
OtakuNekoP has joined #linux-sunxi
<MoeIcenowy>
montjoie: is there any dt patches that enables sun8i-emac on bpi m2+?
<MoeIcenowy>
oh got it
<montjoie>
yes on the sun8i-emac branch
enrico__ has joined #linux-sunxi
<montjoie>
oh diff for uboot sun8i-h3-orangepi-pc-plus vs orangepi_plus_defconfig I see MACPWR=SUNXI_GPD(6)
<montjoie>
"immediate launch of compilation"
reev has joined #linux-sunxi
<reev>
Hi everyone i am using bpi m2+ and i have installed raspian lite on it and i am using it as access point through hostapd intially it was all good
<reev>
but then i noticed error Feb 7 17:45:59 bpi-iot-ros-ai kernel: [ 56.789985] CFGP2P-ERROR) wl_cfgp2p_do_ioctl : wl_cfgp2p_do_ioctl: IOCTL req 0x8b03 on p2p0 I/F. Ignoring
<reev>
in syslog and then my ssid went missing
<reev>
any idea about this
<reev>
also everytime i have to start hostapd i need to do this modprobe -v bcmdhd op_mode=2 to make it work
<reev>
then this error shows up
<beLiefs>
Can anyone tell me what I need to change besides pll_video timing and [disp_init] in script.bin to modify hdmi resolution?
<beLiefs>
Do I need to build uImage?
vicenteH has joined #linux-sunxi
<beLiefs>
H3 / OPI PC if it matters
<KotCzarny>
for h3 im using modified uboot to use 1280x1024 instead of 1280x720. and simplefb
<beLiefs>
that sounds about right
<beLiefs>
DNI from OPI forums, or something else?
<KotCzarny>
dont remember, will check
<beLiefs>
Was trying to combine that with loboris kernel tweaks, but I can't figure out where to begin reading.
<KotCzarny>
um, you can be better of to use armbian kernel patches for h3
<KotCzarny>
assuming you need legacy
<beLiefs>
I need a functional desktop, and I like the OPIPC a lot
<TheLinuxBug>
Armbian has been the best image (Linux) I have used for Linux desktop on PC2 yet. Didn't really spend much time on the server version, but I assume it works pretty stable as well.
<fALSO>
I didnt even knew it existed!
<TheLinuxBug>
All of Xunlong's images are pretty crappy
<fALSO>
I created a gentoo image
<fALSO>
with the kernel of the Debian xunlong version
<TheLinuxBug>
Thanks IgorPec ;D
<fALSO>
armbiang has 4.10.0
<fALSO>
AWESOMEEEEEEEE
<agraf>
MoeIcenowy: "that thing" means cfb_console
<apritzel>
fALSO: now that's cute, they have next week's kernel already ;-)
<fALSO>
im going to try it out today
<fALSO>
i think that there was a problem on the latest kernels
<fALSO>
that it didnt get the cpu temperature
<fALSO>
and could fry it
<fALSO>
maybe its fixed now =)
<IgorPec>
just dont't expect too much , it's unpolished first preview, automated build, ...
<Wizzup>
I was going to ask them to add SPI flash so we can some equivalent of a normal laptop 'BIOS' on there in case the eMMC dies, but someone else managed to suggest it, it seems
apritzel has quit [Quit: Leaving.]
<jelle>
2GB would be nice
<MoeIcenowy>
In fact I a bit want a Olimex laptop sample ;-)
<jelle>
:P
<jelle>
MoeIcenowy: with 2 GB memories
<MoeIcenowy>
even 1GiB is acceptable ;-)
<jelle>
MoeIcenowy: you'd have to fix the blacklight controller then it will be usable for battery life
<MoeIcenowy>
I think the backlight is PWM, right?
<jelle>
MoeIcenowy: well they didn't have any code for it now, so it was always 100% :p
<MoeIcenowy>
if they can provide me a sample I may advance it ;-)
<jelle>
MoeIcenowy: they have 50 samples so maybe you can apply for one
<ElBarto>
they don't have 50 yet
<ElBarto>
main board isn't assembled
<jelle>
ElBarto: uhh it says: " We have the hardware for 50 laptops ready (developer edition), but we do not want to ship before we take care for the software. "
<ElBarto>
but they have the rest for 50 kits yes
<jelle>
ohh I get it
<ElBarto>
I'm in contact with Tsevtan and that's what he told me today
<jelle>
oh I misunderstood the text ;-)
<ElBarto>
this weekend I understood like you when talking to him
<ElBarto>
they've just assembled 3 boards for FOSDEM
<jelle>
yup seem them :)
<jelle>
*seen
<ElBarto>
also A64-Olinuxino is supposed to be available at the end of february
<fALSO>
weird... the ORANGE PI PC2 images from armbian
<fALSO>
doesnt have a fat partition on the start
IgorPec has quit [Ping timeout: 260 seconds]
<fALSO>
probably its not needed
<KotCzarny>
its linux, change it if you prefer having separate /boot
<fALSO>
nah. i just tought it was required, like in RPI
<MoeIcenowy>
fALSO: with mainline kernel theortically it's not needed, as U-Boot can read ext4
<fALSO>
im at work... going to try it when i get to home :)
<fALSO>
moeicenowy, awsome
<fALSO>
but is there any "bootloader" on the beggining of the card ?
<KotCzarny>
yes
<KotCzarny>
uboot
<fALSO>
AHH OK
<MoeIcenowy>
but some people get used to have a VFAT partition for boot, for example me
<KotCzarny>
in the first meg of the device
<fALSO>
OK :)
<MoeIcenowy>
yes, u-boot-sunxi-with-spl.bin is written at 8K
<fALSO>
you guys are awsome, i just found out this channel yesterday
<fALSO>
before that I had little luck =)
<KotCzarny>
go to linux sunxi wiki to see what goes where (mainline uboot page)
<fALSO>
there isnt many info about the H5
<fALSO>
Linux orangepipc2 3.10.65 #3 SMP PREEMPT Tue Nov 15 09:46:50 CST 2016 aarch64 GNU/Linux
<fALSO>
love it =)
<KotCzarny>
h5 is just a mix of a64 and h3
<fALSO>
yes
<fALSO>
and über fast
<KotCzarny>
for fast you want a57 or a17 cores :P
<KotCzarny>
it's just 'faster than previous a7 cores'
<fALSO>
im waiting for a H5 with 2gb of ram
<Wizzup>
MoeIcenowy: I asked for a sample of that laptop, they are making some more in the next few weeks
<Wizzup>
I am also looking for screen upgrades already :p
<Wizzup>
ElBarto: yes, they are sending out some a64-olinuxino samples
<willmore>
fALSO how did you find this channel? There's been some recent talk about making it easier to find and your journey would be useful to know.
<fALSO>
on the wiki
<fALSO>
i was on #orangepi
<fALSO>
no one talks there
<willmore>
fALSO, how did you find the wiki?
<fALSO>
googling for a way to build the kernel for h5
<willmore>
Okay. That got your the wiki and the wiki got you here?
<fALSO>
yes
<willmore>
s/your/you
<willmore>
Cool, thanks.
<willmore>
Enjoy your board. I also have a PC2 and I'm quite happy with it. Are you familiar with the Armbian forums? Plenty of discussion there--especially for new users. #linux-sunxi is more kernel development focused.
<willmore>
BTW, never heard of #orangepi. ;)
<KotCzarny>
there is no point to go to vendornamed channels, there is only linux-sunxi (for dev) and armbian forums for os ;)
<fALSO>
yes, I already had a PC1
<fALSO>
and the other H3 with 512mb
<fALSO>
and I had amrbian on them
<fALSO>
But since i got the pc2, i've been always watchin the armbian site and never saw it there
<KotCzarny>
because its a new soc
<Ke>
KotCzarny: if it only was "There is only stock kernel and mainstream distros" ;o)
<fALSO>
and in the forums it seems that the devs didn like it that much
<fALSO>
becuse it was an "HACK" or something
<KotCzarny>
ke: one could dream ;)
<fALSO>
but ssems its OK now=)
<willmore>
KotCzarny, good, maybe that's why I never went there. ;)
<willmore>
fALSO, Orange Pi One?
<fALSO>
yap
<fALSO>
let me check
<MoeIcenowy>
#orangepi is a strange channel
<fALSO>
no one talks there
<fALSO>
forget about it
<fALSO>
i've been idling there for three days
<MoeIcenowy>
I may suggest the official orangepi to set up an IRC chan
<willmore>
Yeah, well, #armbian is similar. They prefer forums to IRC. And that seems to work for them.
<KotCzarny>
MoeIcenowy: or just suggest their dev to come HERE
<fALSO>
hate forums, theres a lot of "NAZI" moderators
<willmore>
MoeIcenowy, isn't #orangepi official? If not it should be ##orangepi, no?
<fALSO>
irc is better
<fALSO>
willmore, yes, I have a PC, ONE, AND THE PC2 now
<willmore>
fALSO, you need a Zero. :) LOL
<MoeIcenowy>
I think #orangepi is a random organized channel...
<fALSO>
nah, no use for me
<fALSO>
256mb of ram
<willmore>
fALSO, there is a 512MB version.
* willmore
has two
<MoeIcenowy>
I have a One, a Zero and a PC2, thus I have all SoCs yet used on Orange PIs ;-)
<fALSO>
lets hope for a new with: H5 and 2gb of ram
<fALSO>
lets hope for a new with: H5 and 2gb of ram - AAND SATA
<willmore>
Just keep in mind the wireless works poorly and is unlikely to ever work well. Best to ignore it.
<fALSO>
i've looked to the latest BANANA PI
<fALSO>
but its like 70€
<KotCzarny>
for sata grab uasp supporting dongle
<willmore>
fALSO, no native SATA on that chip, so SATA will be over a USB->SATA adapter. Don't expect great performance.
<fALSO>
ahh OK
<KotCzarny>
willmore: actually uasp is chugging quite fast
<fALSO>
its just usb :-(
<fALSO>
didnt knew
<KotCzarny>
40MB/s r/w
<KotCzarny>
and not that much cpu usage
<willmore>
KotCzarny, compared to USB3 or native SATA, I mean.
<willmore>
KotCzarny, great!
<KotCzarny>
we have no usb3 in sight ;)
<willmore>
That's up on my workbench to test.
<MoeIcenowy>
some benchmark on linux-sunxi.org shows that USB2 UASP have better random I/O speed than Allwinner's native SATA
<willmore>
KotCzarny, exactly.
<willmore>
MoeIcenowy, that's funny!
<willmore>
Make sure you only steal good IP, Allwinner!
<MoeIcenowy>
but I forgot where to see it
<MoeIcenowy>
and the sequentinal speed of native SATA is still better than UASP, of course
<willmore>
fALSO, there is hope of an H5 board that works with the NAS I/O board. If it had GigE and 2GB, there would be a lot of happy people.
<fALSO>
yes
<fALSO>
2gb would be awsome
<fALSO>
lets wait =)
<willmore>
If they made it the same size as the NAS board, that would make me happy, but I can understanding avoiding *yet another board outline*.
<KotCzarny>
there is opi plus 2e, quite nice (which has 2gb)
<fALSO>
but its h3
<KotCzarny>
so? doing any cpu work on thos slow boards?
<fALSO>
I use my H5 for php/mysql development
<fALSO>
some python development too
<fALSO>
works PRETTY well
<KotCzarny>
you should really get a marvel board then
<KotCzarny>
yeah, or any other that popped out recently
<fALSO>
lets wait until they build them =)
<MoeIcenowy>
I used to flash my Laptop with Coreboot on a Orange Pi One ;-)
<willmore>
KotCzarny, 2e looks okay, but at fALSO said, H5 is the base of what is being requested. :)
<MoeIcenowy>
(and of course the Coreboot binary is built on it
<fALSO>
YEs h5 compared with H3 is MUCCCCCCCCCCCCCCCH better
<MoeIcenowy>
for H5 and 2GiB DRAM, maybe we should wait for Orange Pi 3
<fALSO>
yes
<willmore>
I don't know about that many 'C's, but it is a bit better.
<KotCzarny>
willmore: but opi+2e is already in store and h5 is in early dev stage, and no 2gb device yet
<fALSO>
bnut probabvly the orange pi 3
<fALSO>
will have a H10 CPU
<fALSO>
LOL
<willmore>
KotCzarny, no hurry.
<MoeIcenowy>
is there H10?
<fALSO>
im just kidding =)
<willmore>
MoeIcenowy, if the 3 has a NAS hat capable header, then I like it. ;)
<MoeIcenowy>
no it won't have it
<MoeIcenowy>
but you can apply your own UASP
<willmore>
Or just the I/O expansion board from the Z.
<KotCzarny>
willmore, just print nice case that will snuggly keep 3 uasp dongles ;)
<willmore>
Yep, I have plenty of USB<>SATA UASP adapters.
<willmore>
KotCzarny, good point.
<willmore>
But that's a lot of wasted coiled up cable to fiddle with.
<fALSO>
the problem with that stuff
<KotCzarny>
then cut the cable and solder directly to port pads
<KotCzarny>
;)
<willmore>
Hmm, a NAS board with a full SATA+power connector on it.....
<fALSO>
its that you have to have an EXTRA power suplly
<fALSO>
just for the HDD
<fALSO>
i have a CHINESE ONE
<fALSO>
but I dont want to leave that connected to the power all day
<MoeIcenowy>
maybe they should make an advance version of Plus2 that uses UASP and H5 ;-)
<fALSO>
dont want to start a fire at home .-)
<willmore>
Well, 2.5" SATA is 3.3V and 5V, so no reason that couldn't be fed from a 5V supply with a small SMPS for the 3.3V. Like the existing NAS board does.
<KotCzarny>
MoeIcenowy: they will run out of names soon ;)
<MoeIcenowy>
Call it Plus 3 ;-)
yann-kaelig has joined #linux-sunxi
<willmore>
KotCzarny, don't tempt them!
<KotCzarny>
willmore: it was her, not me!
<KotCzarny>
;)
<willmore>
KotCzarny, you mentioned that they would run out of names. If they do, they'll just add something even more confusing. Don't tempt them!
<KotCzarny>
they will anyway, thats 100% sure
jbrown has quit [Ping timeout: 255 seconds]
* willmore
is trying to think of how a two drive direct drive connect NAS board would need to be laid out physically....
<KotCzarny>
willmore: btw. when using usb dongle you only need 5v
<KotCzarny>
and for my 12V drive i just use dual voltage psu
<fALSO>
that stuff that youre talking about
<fALSO>
UASP
<KotCzarny>
never needed 3.3v
<fALSO>
isnt that just for USB 3 ?
<KotCzarny>
falso: nope
<fALSO>
ahh OK
<fALSO>
thanks
hramrach has joined #linux-sunxi
<MoeIcenowy>
P.S. a 2GiB DRAM board with a Orange Pi Zero-compatible header seems impossible
<fALSO>
never heard about that, going to google moar
<willmore>
KotCzarny, power over USB is not reliable enough to make me comfortable putting a drive on it long term.
<MoeIcenowy>
as the OPi Zero header is designed to save some space on PCB
<KotCzarny>
falso: just read linux-sunxi wiki, it's there
<MoeIcenowy>
however, 2GiB DRAM willl cost a lot of PCD space
<MoeIcenowy>
PCB *
<willmore>
MoeIcenowy, and the lack of all those I/O connectors saves you some space. Also, (my opinion) you can make the board as big as the NAS board.
<willmore>
So, you're not stuck in the Zero outline.
f0xx has quit [Ping timeout: 258 seconds]
<willmore>
Again, this is just ideas inside my head leaking out.
ErwinH has quit [Read error: Connection reset by peer]
<willmore>
I should probably have that looked at.
Mr__Anderson has joined #linux-sunxi
my123_ has joined #linux-sunxi
my123_ has joined #linux-sunxi
my123_ has quit [Changing host]
nove has joined #linux-sunxi
jbrown has joined #linux-sunxi
my123 has quit [Ping timeout: 276 seconds]
cnxsoft has quit [Quit: cnxsoft]
<TheLinuxBug>
willmore: lol
<TheLinuxBug>
Was it ever decided if R40 had native SATA or not?
<KotCzarny>
yes, it has
<TheLinuxBug>
and if it was any more performant than A20?
<KotCzarny>
but same performance as a20
<TheLinuxBug>
figures
<TheLinuxBug>
so what, you gain 2 cores and thats it pretty much? ( well 1gb more ram too I guess)
<KotCzarny>
dont remember what else has changed
<TheLinuxBug>
well
<TheLinuxBug>
I would assume you also get actual gigabit on it
msevwork has quit [Quit: Leaving]
<TheLinuxBug>
instead of the 45M/sec limit cause its bound to 1 core on A20
<TheLinuxBug>
or is that the same as well?
<willmore>
Maybe more cores could mean less problems from CPU overhead.... Probably not.
<fALSO>
would require an extra cable for - POWER; right ?
arete74 has joined #linux-sunxi
<KotCzarny>
in ge3neral, 2.5" would not need extra cable, 3.5" will require
ErwinH has quit [Ping timeout: 240 seconds]
<fALSO>
AHHH
<fALSO>
2.5" disks
<fALSO>
OK
<fALSO>
i have a lot of EXTRA IDE DRIVES :-P
<fALSO>
3.5
jbrown has joined #linux-sunxi
ErwinH has joined #linux-sunxi
fkluknav has quit [Ping timeout: 258 seconds]
<nove>
one thing that tkaiser commented there, "1080p video on youtube in 1080p" can be very possible, but this is the same story, this is work that is only justified if there is a "community" that refuses to use "license issues" software
ErwinH has quit [Ping timeout: 255 seconds]
<KotCzarny>
if it would be that easy
vickycq- has joined #linux-sunxi
ErwinH has joined #linux-sunxi
ErwinH has quit [Ping timeout: 240 seconds]
anhn has quit [Ping timeout: 245 seconds]
fkluknav has joined #linux-sunxi
ErwinH has joined #linux-sunxi
arete74_ has joined #linux-sunxi
arete74 has quit [Read error: Connection reset by peer]
ErwinH has quit [Ping timeout: 240 seconds]
ErwinH has joined #linux-sunxi
anhn has joined #linux-sunxi
ErwinH has quit [Ping timeout: 240 seconds]
fkluknav has quit [Ping timeout: 240 seconds]
fkluknav has joined #linux-sunxi
ErwinH has joined #linux-sunxi
ErwinH has quit [Ping timeout: 260 seconds]
ErwinH has joined #linux-sunxi
ErwinH has quit [Ping timeout: 240 seconds]
arete74_ has quit [Read error: Connection reset by peer]
arete74 has joined #linux-sunxi
yann-kaelig has quit [Quit: Leaving]
ErwinH has joined #linux-sunxi
yann-kaelig has joined #linux-sunxi
scream has joined #linux-sunxi
OtakuNekoP has quit [Ping timeout: 260 seconds]
ErwinH has quit [Ping timeout: 240 seconds]
ErwinH has joined #linux-sunxi
feuera has joined #linux-sunxi
ErwinH has quit [Ping timeout: 248 seconds]
feuera has quit [Ping timeout: 260 seconds]
jernej has joined #linux-sunxi
ErwinH has joined #linux-sunxi
ErwinH has quit [Ping timeout: 248 seconds]
ErwinH has joined #linux-sunxi
<MoeIcenowy>
A64 DE2 is a disaster...
<MoeIcenowy>
I cannot still clock it up...
steev has quit [Read error: Connection reset by peer]
ErwinH has quit [Ping timeout: 264 seconds]
steev has joined #linux-sunxi
ErwinH has joined #linux-sunxi
<jernej>
MoeIcenowy: I'm also struggling with the same issue
<MoeIcenowy>
jernej: ah-oh
<jernej>
MoeIcenowy: I will take a look at /dev/mem on BSP kernel based image
fkluknav has quit [Ping timeout: 240 seconds]
<MoeIcenowy>
thanks ;-)
<MoeIcenowy>
I have already deleted my BSP image...
<jernej>
MoeIcenowy: Did you notice that DE is the only peripherial unit in this chain, which can be protected to require secure mode access
<jernej>
I disabled trustzone, but it didn't help
<MoeIcenowy>
it's de0-lcd1-hdmi
<MoeIcenowy>
lcd1 and hdmi works well here
ErwinH has quit [Ping timeout: 240 seconds]
<MoeIcenowy>
(I have a HDMI screen that uses a simple naive HDMI-to-RGB bridge
<jernej>
i know, TCON test pattern (blue screen) works as expected
<MoeIcenowy>
(so that if the pixel dotclock is wrong due to TCON issue, it will be seen on the LCD panel
<MoeIcenowy>
(just like the LCD panel is directly attached to TCON ;-)
<MoeIcenowy>
I also tested blue screen
jbrown has quit [Ping timeout: 255 seconds]
<jernej>
writing by mw.l also doesn't work, md.l always read all zeros
massi has quit [Remote host closed the connection]
<MoeIcenowy>
in addition, according to apritzel, TZPC do not work on A64 except in secure boot mode
<MoeIcenowy>
so you shouldn't face TrustZone problem ;-)
<MoeIcenowy>
(because you will brick your device to really enforce TZPC ;-)
<jernej>
ah, I remember this
<MoeIcenowy>
maybe some power is not supplied?
<ElBarto>
jernej: MoeIcenowy test pattern as generated by tcon directly ?
<MoeIcenowy>
(without DLDO1 be enabled, the HDMI block won't work, it's the reason of my initial "HDMI phy init timeout"
<jernej>
MoeIcenowy: Do you know what is "CCU security switch register" good for on A64?
<MoeIcenowy>
ElBarto: yes
<jernej>
ElBarto: Not exactly a pattern, but yes, test screen, deep blue color
<ElBarto>
MoeIcenowy: do you have more info on that ? I know that Jared who is working on A64 hdmi is having trouble right now
<ElBarto>
maybe that will help him
<MoeIcenowy>
who's jared?
<MoeIcenowy>
maybe he can join the talk here ;-)
<ElBarto>
jernej: as long as it output a correct hdmi signal this would be better
<ElBarto>
oh yeah I should add for FreeBSD :)
<ElBarto>
he comes here sometimes
<jernej>
ElBarto: Just set second bit in TCON global control register and you will get blue screen
<ElBarto>
ok cool
ErwinH has joined #linux-sunxi
<jernej>
ElBarto: This is described in A83T manual, but it works on A64 and H3 too
<MoeIcenowy>
jernej: it also do not work ;-)
<MoeIcenowy>
jernej: I think it's also described in A64/H3 manual.
<jernej>
MoeIcenowy: What doesn't work?
<ElBarto>
if it's in the manual I guess he have seen it, I'll check with him
Mr__Anderson has quit [Quit: Leaving.]
<MoeIcenowy>
"CCU security switch register"
<jernej>
Ok, but what it is supposed to do?
<MoeIcenowy>
prevent non-secure world from accessing some critical clocks
<jernej>
so, we can ignore all security locks on A64 for now?
<MoeIcenowy>
yes
<MoeIcenowy>
I tried to power all regulators up, and it do not work either...
<jernej>
this only leaves some misconfiguration
<MoeIcenowy>
misconfiguration?
<jernej>
I mean, we are missing some bit somewhere
<MoeIcenowy>
yes
<MoeIcenowy>
but what's missing...
<jernej>
I will check BSP kernel now...
ErwinH has quit [Ping timeout: 258 seconds]
jernej has quit [Quit: Konversation terminated!]
jernej has joined #linux-sunxi
popolon has quit [Quit: WeeChat 1.4]
<MoeIcenowy>
jernej: are you running Konversation on Pine64? ;-)
fkluknav has joined #linux-sunxi
jbrown has joined #linux-sunxi
<jernej>
no, on my main laptop
<jernej>
why are you asking?
enrico__ has quit [Quit: Bye]
<MoeIcenowy>
I saw you left and rejoined ;-)
<MoeIcenowy>
it's only a joke ;-)
<jernej>
ah, I had to reboot it, because I just upgraded kernel, you know, Arch
<MoeIcenowy>
at least it won't have heap of old kernels ;-)
ErwinH has joined #linux-sunxi
ErwinH has quit [Ping timeout: 245 seconds]
ErwinH has joined #linux-sunxi
ErwinH has quit [Ping timeout: 252 seconds]
fkluknav has quit [Ping timeout: 258 seconds]
ErwinH has joined #linux-sunxi
ErwinH has quit [Ping timeout: 240 seconds]
BenG83 has joined #linux-sunxi
<willmore>
nove, oh, Olimex had avoided talking about binary blobs. Now they did. Okay.
<MoeIcenowy>
the blob numbers in A64 finally become unbearable
<MoeIcenowy>
and some critical logics are covered by blob (DRAM init)
ErwinH has joined #linux-sunxi
<willmore>
MoeIcenowy, yeah, that's an upsetting one for me. Is there a complete list? DRAM, Mali, etc.?
<MoeIcenowy>
(Although this list is just generated using a for loop in bash with "file" command by me ;-)
<willmore>
Interesting. NAND? Why would that be a blob? The whole display pipeline, too? So confusing.
<willmore>
Thanks, MoeIcenowy.
ErwinH has quit [Ping timeout: 260 seconds]
<KotCzarny>
willmore: to obfuscate some unlicensed hardware?
<KotCzarny>
or 'here allwinner, you can have those third grade wafers, but you cant tell anyone you got it from us' ;)
yann-kaelig has quit [Quit: Leaving]
<MoeIcenowy>
maybe it's really unlicensed hardware, at least for dwmac and dwhdmi
<KotCzarny>
MoeIcenowy: more and more blocks are showing up as 'already existing, just obfuscated'
<TheLinuxBug>
Arg! need to figure out better dvfs settings for my S905x, everytime it runs with 1.5Ghz and I do anything intensive it just totally locks up :(
<KotCzarny>
then downclock a notch?
<TheLinuxBug>
yeah but next option is 1.2Ghz
<KotCzarny>
you wont notice difference between 1400 and 1500 etc
<KotCzarny>
oh?
<TheLinuxBug>
feel like I am losing a lot there
<TheLinuxBug>
yeah
<TheLinuxBug>
the scaling govenor they used only allows presets
<KotCzarny>
benchmark then, and compare if your fears are founded
<TheLinuxBug>
1200 then 1512
<TheLinuxBug>
and only allows interactive or performance
<TheLinuxBug>
no ondemand govenor etc
<beeble>
KotCzarny: i think they would have problems manufacturing at tsmc with unlicensed cores
<TheLinuxBug>
kernel probably needs recompiled
<KotCzarny>
beeble: unlciensed as in second grade and without support/publishing/pr rights
<TheLinuxBug>
KotCzarny: what is depressing is I added a bunch of heatsinks too and still overheats
IgorPec has joined #linux-sunxi
<TheLinuxBug>
was reading about it and someone said they were able to get it running at 1.5ghz with heatsinks with no issue, I can not replicate that
<KotCzarny>
TheLinuxBug: so, back to h5 ? ;)
<TheLinuxBug>
IgorPec: do you guys plan to make an official image for the Khadas Vim (S905x) I noticed they use Armbian as their default Ubuntu image anyways, however, they seem to have optimized it very poorly.
muvlon has joined #linux-sunxi
<TheLinuxBug>
KotCzarny: funny you say that as I just downloaded the most recently nightly to test ;p
<IgorPec>
thelinuxbug: we plan to add official support for Beelink GT, we got two boxes, the rest will probably remain unoffical way
<IgorPec>
in best case, we will try to collect configs within build system
<TheLinuxBug>
Well I ask cause it seems they literally are using a version of Armbian as their official Ubuntu image and it seems very poorly optimized from what I can tell, only have 2 cpu govenors avail: performance and interactive and seems to have set scaling options only, and running at full 1.5Ghz causes lock up, even with heatsink so I figure DVFS is in need of help too..
<MoeIcenowy>
beeble: thus... maybe it's a reason that Allwinner keeps to use 40nm process? ;-)
<MoeIcenowy>
(A83T seems to use a not obfuscated DW HDMI controller)
dave0x6d has joined #linux-sunxi
ErwinH has joined #linux-sunxi
<TheLinuxBug>
IgorPec: I know they have one of their images that are in use posted in the Armbian forum, just no official release it seems.
<KotCzarny>
TheLinuxBug: if they use armbian and kernel source is available, recompile, then submit changes to integrate maybe?
<TheLinuxBug>
yeah may have to do that
<TheLinuxBug>
at least to get better support for govenors like ondemand and better ability to control CPU frequency
<IgorPec>
yes, afaik balbes is using amlogic kernel sources, not the one from Odroid C2
arete74_ has joined #linux-sunxi
<IgorPec>
config is also a bit different but it's build using some older version of our build tools
<IgorPec>
the idea is to merge all this stuff, but i can't tell when and which boxes will be possible to build directly
<TheLinuxBug>
IgorPec: well I won't lie, my other thought was maybe if you did support it one day there might be an image with gfx accel support, but again maybe I can mess with that as well..
<TheLinuxBug>
;p
<TheLinuxBug>
but I know that is a dream hehe
<IgorPec>
acceleration should be possible to achieve, but i haven't dig into this amlogic much .. yet
ErwinH has quit [Ping timeout: 240 seconds]
<TheLinuxBug>
I was actually curious if the S905 and S905x were close enough that could actually try the odroid c2 image and possible have it work somewhat, but based on reading I am assuming there are some slight differences.
<fALSO>
Linux orangepipc2 4.10.0-sun50iw2 #6 SMP Tue Feb 7 18:11:11 CET 2017 aarch64 aarch64 aarch64 GNU/Linux
<fALSO>
nice
<fALSO>
going to do some tests to compare with the v3 kernel
<TheLinuxBug>
IgorPec: anyways, thanks for all your work and letting me bug you with stupid questions :D
f0xx has joined #linux-sunxi
<IgorPec>
Thelinuxbug: yes, there are some differencies, but you should reffer to bablbes for details. I am still learning them
<TheLinuxBug>
k
<IgorPec>
you are welcome. I would like to tell you more about, but I am also not there yet. much other things to do :)
<TheLinuxBug>
will have to go re-read and attempt my own kernel at some point
<TheLinuxBug>
see where I get to
<TheLinuxBug>
IgorPec: I bet ;p
<IgorPec>
yes, those huge topics at forum are a good start if you plan playing around
ErwinH has joined #linux-sunxi
ErwinH has quit [Ping timeout: 258 seconds]
Ntemis has quit [Remote host closed the connection]
Ntemis has joined #linux-sunxi
ErwinH has joined #linux-sunxi
<TheLinuxBug>
interesting, managed to have it freeze up at 1.2Ghz now
<TheLinuxBug>
which wasn't previously happening
<TheLinuxBug>
:(
ErwinH has quit [Ping timeout: 245 seconds]
terra854 has quit [Quit: Connection closed for inactivity]
komunista has quit [Quit: Leaving.]
ErwinH has joined #linux-sunxi
ErwinH has quit [Ping timeout: 260 seconds]
tllim has joined #linux-sunxi
<TheLinuxBug>
seems I will need a fan to do anythign useful
<TheLinuxBug>
that kinda stinks
Andy-D has joined #linux-sunxi
<tllim>
MoeIcenowy and jernej: I will be in Shenzhen on February 21-25. If Icenomy interested, I can bridge up Icenowy and Allwinner engineer. Allwinner mentioned that they are not able to releaes LibDRAM source but willing to help up. when Sunxi encounter issue on HDMI development. Please advise on your thought and I can make teh arrangement.
<tllim>
MoeIcenowy, if you have time on time on February 21-25, we can meet up either in Shenzhen or Guangzhou. I like to offfer you the PINE64 T-short and hoodie wich I pass to Sunxi folks during FOSDEM.
<TheLinuxBug>
AHAHAH
<TheLinuxBug>
yeah need something like that probably
<TheLinuxBug>
I am guessing there is a big difference between the 'Pro' and Basic versions of this board as far as heat, I got the Pro version with 16gb emmc and 2Gb ram and it is immediatly running like 66-68c just idle WITH heatsinks while others say theirs runs around 60c
<TheLinuxBug>
on the basic version
<TheLinuxBug>
so I keep hitting the shutdown point of 70c
<TheLinuxBug>
and the whole board freezes
<jernej>
tllim: thanks, if we talk about HDMI, then wishlist is HDMI PHY and DE2 documentation and GPL2 licensed code in full without blobs and how to enable DE2 on A64 (clocks, gates, resets, etc.)
ErwinH has quit [Ping timeout: 276 seconds]
vagrantc has joined #linux-sunxi
<tllim>
as mentioned, they may not provide the code. However, they are willing to engage in engineer discussion like Q and A, to resolve some outstanding issues that Sunxi facing
<jernej>
tllim: for example, DE2 is opened up, but lack of GPL2 statements in half of files prevent us to reuse it
<tllim>
If the file that already open up but lacking GPL statement, I may able to push for it.
<jernej>
well, this is not limited to DE2, many files lack GPL header
apritzel has joined #linux-sunxi
<tllim>
IPINE64 recently has released an A64 BSP with GPL statement install that release from Allwinner, please check whether your concern still valid.
<jernej>
I did, but I don't see any improvement, at least in DE2 part of the code
<jernej>
let me check again
<tllim>
please check, this release happens on last week before I travel to FOSDEM
ErwinH has joined #linux-sunxi
f0xx has quit [Ping timeout: 248 seconds]
<tllim>
if IceNoway interested on the offer, lets me know and I will make the ararngement to meet up with Allwinner. Allwinner BU2 is OK on such arrangement, eitehr email or meet in person.
<jernej>
tllim: It will take a while to download it again, but anyway, presence of the blobs means violation of GPL2, so the problem is still present
<plaes>
tllim: at FOSDEM dinner you also showed bunch of tablets with keyboard, is there ETA for those too?
ErwinH has quit [Ping timeout: 245 seconds]
<tllim>
plase, sorry that we are on table and not get chance to talk.
<tllim>
the tablet with keyboard targets for Q2 release. However, PINE64 will make some prototype and pass to Sunxi team if interested, need to get the touch screen issue and concern resolve first.
<tllim>
jernej, I try to push whatever I can push. Frankly speaking, I don't think Allwinner will release LibHDMI source, they have their concern. They have told me their conecrn but I don't want mentioned in IRC publicly. You are welcome to PM me
<vagrantc>
looks like that's in 4.10-rc7 ... so i'll just hae to wait
<vagrantc>
seems like eMMC is fine ...
<plaes>
yeah
<willmore>
KotCzarny, that was my guess as well. Ethernet is OTS. DE is from DesignWare, no? One the plus side, the video decode engine is likely theirs, right? But they still managed to mangle the userspace code for that.
anhn has quit [Ping timeout: 264 seconds]
<jernej>
MoeIcenowy: I dumped almost all CCU region, but unfortunatelly, if I copy values over, it doesn't help
ixnus has quit [Client Quit]
<willmore>
beeble, TSMC is the fab for Allwinner? I ask because I don't know.
<willmore>
KotCzarny, that makes it even less clear. :)
<beeble>
willmore: at least some newer ones are
<willmore>
beeble, okay. I would have guessed the 40nm chips would be made in China. 28nm parts like the octo-A7 parts I can see. But, as was mentioned, they don't seem to use obfuscated register versions of their IP. So, you might be right.
<willmore>
Oh, on a good note, my 128Mb NOR SPI chips finally arrived today! I have already soldered on into my early SPI-less OpiZ.
<willmore>
Time to update that to mainline if it isn't already and get working.
ErwinH has quit [Ping timeout: 255 seconds]
apritzel has quit [Ping timeout: 264 seconds]
<beeble>
willmore: i have to buy some hno3 and find a way to vent my office. then i can take look if there are any markings on the die
<KotCzarny>
Video Engine is the plain name used by Allwinner for the hardware block responsible for the task of decoding and encoding video formats.
<KotCzarny>
second paragraph
<KotCzarny>
deconding and encoding
<nove>
the video engine is only one block, but can simultaneous encode and decode (expect the ones older than A20, which can only do one task at a time)
<nove>
it is guessed to be a custom hardware design by chipbanks.com for allwinner
<nove>
a chipbanks.com email was as author for their kernel driver in the first sources released for A10
<KotCzarny>
so, is there some block that is designed by aw 100% ? or they glued all pieces from other sources?
<willmore>
beeble, go for it! ;)
Ntemis has quit [Read error: Connection reset by peer]
<willmore>
Don't forget to heat the NO3 until it's almost boiling. :)
<willmore>
nove, good info!
<willmore>
KotCzarny, that page says that--depending on the data sheet you read--that it's either the VPU, VE, or video Engine.
scream has quit [Remote host closed the connection]
<willmore>
So, I guess we can just refer to them all together by any of those names as I imagine any driver that touches one will control them all.
<KotCzarny>
willmore, it depends on the soc generation, imo they changed the name for that block
<willmore>
Yep, that's the confusion.
<nove>
willmore: it is what happens when they don't have a marketing name, but for their ISP the gotten a fancy name of hawkview
Mr__Anderson has joined #linux-sunxi
darkusas has quit [Ping timeout: 260 seconds]
Ntemis has joined #linux-sunxi
<willmore>
nove, understood. Having a half dozen different business units all fiddling with things--and with who knows what level of code sharing or 'apropriating' internally. And documentation for that matter.
ErwinH has joined #linux-sunxi
<fALSO>
anyone tried to build armbian for the pc 2 ?
reinforce has joined #linux-sunxi
<fALSO>
its fails right at the beggining with:
<fALSO>
[ error ] U-boot file not found [ u-boot-sunxi-with-spl.bin ]
[Awaxx] has quit [Quit: A man who say the truth is a free man, my friend.]
ErwinH has quit [Ping timeout: 240 seconds]
reinforce has quit [Quit: Leaving.]
ErwinH has joined #linux-sunxi
fl__0 has joined #linux-sunxi
<NiteHawk>
plaes: thanks for contributing your fixes to sunxi-tools. do you consider that patchset complete, or are you still working on it?
fl_0 has quit [Ping timeout: 258 seconds]
ErwinH has quit [Ping timeout: 264 seconds]
[Awaxx] has joined #linux-sunxi
jbrown has quit [Ping timeout: 255 seconds]
afaerber has quit [Quit: Leaving]
beLiefs has quit [Read error: Connection reset by peer]
cptG_ has joined #linux-sunxi
jbrown has joined #linux-sunxi
cptG has quit [Ping timeout: 258 seconds]
tuxillo has quit [Ping timeout: 240 seconds]
apritzel has joined #linux-sunxi
afaerber has joined #linux-sunxi
tuxillo has joined #linux-sunxi
ErwinH has joined #linux-sunxi
ErwinH has quit [Ping timeout: 240 seconds]
MXfive has quit [Quit: Sleep Quit.]
<willmore>
Anyone mind if I edit the Orange Pi 2E page to correct the "other" entry in the info table from 'IRDA' to "CIR"?
ErwinH has joined #linux-sunxi
<willmore>
Change made. Revert it if you don't agree. Pretty much nothing has IRDA these days, but lots of boards have Consumer IR receivers for remotes.
ErwinH has quit [Ping timeout: 256 seconds]
jstein_ has joined #linux-sunxi
jstein_ is now known as jstein
indy has quit [Ping timeout: 240 seconds]
muvlon has quit [Quit: Leaving]
indy has joined #linux-sunxi
<NiteHawk>
willmore: CIR is correct, that mixup with IrDA is present on quite a fel device pages. I once made an attempt to clarify a bit on that, see the note at http://linux-sunxi.org/LIRC
<NiteHawk>
s/fel/few/
nove has quit [Quit: nove]
<jernej>
apritzel: What is the proper way to put sunxi-spl.bin and u-boot.bin to sd card on mainline U-Boot? SPL can't find U-Boot on sd card...
<apritzel>
jernej: it has to be u-boot.img, not .bin
<apritzel>
jernej: and as always: 32KB behind the SPL
<apritzel>
(which is hardcoded for sunxi)
<jernej>
apritzel: ah, thanks, I missed that somehow
jstein has quit [Remote host closed the connection]
<apritzel>
jernej: are you after the H5?
<jernej>
maybe there should be a convenience step which creates u-boot-sunxi-with-spl.bin
<apritzel>
jernej: yes, I am on it
<jernej>
apritzel: no, A64
<apritzel>
it's a bit more complicated on 64-bit due to the ATF
<apritzel>
H5 and A64 share the same approach
<jernej>
I want to enable simplefb, but I and MoeIcenowy have issue how to enable DE2
jbrown has quit [Ping timeout: 255 seconds]
<jernej>
all clocks are enabled the same way as on BSP kernel, but any writting at 0x01000000 is ignored
<jernej>
apritzel: do you have any idea?
<jernej>
now I'm so desperate that I started looking in mbus code... Basically anything which have to do with DE in any way
<jernej>
it works on H5 without any issue
<apritzel>
what's at that address, exactly? 4MB is pretty big for some registers ...
<jernej>
0x01000000 is base address
<apritzel>
yes, but is it registers only?
<jernej>
actually, it is separated in many modules
<apritzel>
the manual says it's 4 MB big ...
<jernej>
at base address, there should be around 5 32 bit registers
<jernej>
then at 0x01100000 is MUX0 and at 0x01200000 is MUX1
<apritzel>
OK, so very sparse register space ...
<apritzel>
jernej: and writing to those register doesn't stick? can you read something sensible (reserved bit or so)?
<willmore>
NiteHawk, so I'm free to fix them as I run into them?
<jernej>
apritzel: writing doesn't stick, and initialized values are all zero, or at least it appears so
<apritzel>
smells like not enabled
<NiteHawk>
willmore: I'd say so. as you've already noticed, chances are slim you'll ever run into a sunxi device that has actual IrDA (as that would require a separate transceiver). the "native" IR0/IR1 on these SoCs should alway be standard 38kHz CIR
<apritzel>
I think this happens if the reset gate is not asserted
<jernej>
exactly, but I went over CCU many times, also copied BSP values to no avail
<apritzel>
jernej: and you said you dumped the CCU registers space on a BSP booted system as well?
<jernej>
apritzel: interesting enough, DE is the only part of video chain, which can be in theory secured
<jernej>
yes, I can upload it if you want
<willmore>
NiteHawk, okay, I'll clean things as I find them.
<willmore>
I will, of course, make sure that I'm right in doing so first.
<apritzel>
jernej: yeah, I see DE-TOP, DE-MIXER[01], DE-WB, DE-ROT, but I am pretty sure this doesn't work ;-)
* willmore
is getting his feet wet with SPI NOR flash.
<willmore>
So much reading.
<apritzel>
jernej: unless you brick your board as I did ...
<jernej>
apritzel: yeah, I dump also tz registers and they were all zeros on BSP
<apritzel>
jernej: are the base clocks as on the BSP? AHB1, AXI, MBUS ...
jbrown has joined #linux-sunxi
<jernej>
apritzel: not exactly, but I changed them according to dump (http://sprunge.us/FKXc) but it didn't help
<apritzel>
jernej: out of sheer despair: can you dump behind +0x320 with the BSP?
<jernej>
apritzel: if it is important, I'm working with your h5 branch and sun50i_h5_spl32_defconfig
<jernej>
ok, will do
<jernej>
sorry,
<jernej>
sun50i_spl32_defconfig
<apritzel>
jernej: do you load ATF?
<jernej>
no
<jernej>
I'm a bit new to aarch64
<apritzel>
argh, "git grep -i de" it not really useful ;-)
Mr__Anderson has quit [Quit: Leaving.]
<jernej>
Does ATF any important initialization?
<apritzel>
jernej: do you FEL boot? or SD card?
<jernej>
felboot
<apritzel>
jernej: you are just using U-Boot at this point, right?
<apritzel>
with FEL loading and executing ATF is easy
<jernej>
yes, only U-Boot
<apritzel>
jernej: there is nothing in the ATF I can think of should matter
<apritzel>
but I am just looking at Allwinner's original ATF code
<jernej>
in BSP U-Boot DE address space works
IgorPec has quit [Ping timeout: 256 seconds]
<apritzel>
without ATF you should be in 64-bit EL3 in U-Boot, which is secure
<apritzel>
the BSP U-Boot runs in non-secure SVC
<apritzel>
but I don't think this matters
<jernej>
well, previously I used 32 bit version, but there is no difference
<apritzel>
32-bit U-Boot should run in secure SVC
<jernej>
so, both ways secure
<apritzel>
yes, but it would be a bit backward if it works in non-secure, but is blocked in secure
<apritzel>
you can try to load ATF, that would U-Boot run in (non-secure) EL2
<apritzel>
clone the "allwinner" branch from my github, then: CROSS_COMPILE=aarch64-linux-gnu- make PLAT=sun50iw1p1 DEBUG=1 bl31
<jernej>
for that I can use your h5 branch (u-boot)?
<apritzel>
yes, that shouldn't matter
<apritzel>
then just add: "write 0x44000 bl31.bin reset64 0x44000" to your sunxi-fel cmd line
<apritzel>
with "spl u-boot-spl.bin load 0x4a000000 u-boot.bin" before
<jernej>
this is my fel line: sunxi-fel spl spl/sunxi-spl.bin write 0x4a000000 u-boot.bin write 0x44000 ../arm-trusted-firmware/build/sun50iw1p1/debug/bl31.bin reset64 0x44000
<jernej>
uh, should be u-boot 64 bit?
<apritzel>
jernej: can you try again? I think there are some stability issues. Do you have MoeIcenowy patch with the increased delay in?
<apritzel>
jernej: oh yes!
<apritzel>
copy the SPL somewhere
<apritzel>
CROSS_COMPILE=aarch64-linux-gnu- make pine64_plus_defconfig
chomwitt1 has quit [Ping timeout: 255 seconds]
<jernej>
apritzel: now it works, thanks, but unfortunatelly, situation regarding DE memory is the same
<apritzel>
too bad!
<apritzel>
jernej: did you copy the DE init sequence from the BSP as well?
<jernej>
apritzel: No, I wrote it based on datasheet and jfm DE2 mainline driver and it works on H3 and H5 without issues
<jernej>
BSP DE2 code is scattered beyond belief
<apritzel>
maybe some "external" enable bit sticks out somehow?
paulk-collins has quit [Remote host closed the connection]
<apritzel>
external as in not-CCU or not-DE MMIO
<jernej>
apritzel: I guess this is the only thing I can try now
<jernej>
apritzel: btw, what has mbus to do with DE? I saw something about that in the code
<apritzel>
jernej: good question ;-)
<apritzel>
jernej: AFAIK it's about bus arbitration
<jernej>
apritzel: so this is not the issue then
petr has quit [Ping timeout: 258 seconds]
<apritzel>
probably not
<apritzel>
my understanding is that it allows to prefer DE memory accesses over the CPU, for instance