Nacho has quit [Read error: Connection reset by peer]
Nacho has joined #linux-sunxi
aballier has joined #linux-sunxi
libv has joined #linux-sunxi
Putti has joined #linux-sunxi
Putti has quit [Client Quit]
ninolein has quit [Ping timeout: 240 seconds]
ninolein has joined #linux-sunxi
libv_ has joined #linux-sunxi
libv has quit [Ping timeout: 260 seconds]
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
Andy-D has quit [Ping timeout: 256 seconds]
kivutar has quit [Ping timeout: 252 seconds]
<willmore>
wens, half of a 'yeah!'
kivutar has joined #linux-sunxi
kivutar has quit [Ping timeout: 244 seconds]
berkutta has joined #linux-sunxi
JohnDoe_71Rus has joined #linux-sunxi
JohnDoe_71Rus has quit [Client Quit]
kivutar has joined #linux-sunxi
kivutar has quit [Ping timeout: 244 seconds]
kivutar has joined #linux-sunxi
victhor has quit [Ping timeout: 260 seconds]
pg12 has quit [Ping timeout: 246 seconds]
pg12 has joined #linux-sunxi
kivutar has quit [Ping timeout: 256 seconds]
kivutar has joined #linux-sunxi
IgorPec has joined #linux-sunxi
kivutar has quit [Excess Flood]
kivutar has joined #linux-sunxi
kivutar has quit [Ping timeout: 260 seconds]
kivutar has joined #linux-sunxi
kivutar has quit [Excess Flood]
kivutar has joined #linux-sunxi
Gerwin_J has joined #linux-sunxi
JohnDoe_71Rus has joined #linux-sunxi
kivutar has quit [Ping timeout: 245 seconds]
f0xx has joined #linux-sunxi
iamfrankenstein has joined #linux-sunxi
<wens>
looks like gmac in r40 is the new variant
TheSeven has quit [Ping timeout: 245 seconds]
TheSeven has joined #linux-sunxi
jernej has joined #linux-sunxi
reinforce has joined #linux-sunxi
sunxi_fan1 has quit [Ping timeout: 260 seconds]
lkcl has joined #linux-sunxi
xenoxaos has quit [Ping timeout: 240 seconds]
jernej has quit [Ping timeout: 260 seconds]
reinforce has quit [Quit: Leaving.]
xenoxaos has joined #linux-sunxi
reinforce has joined #linux-sunxi
<miasma>
wens: cool, so is the audio gonna be available in 4.10 or 4.11?
berkutta has quit [Ping timeout: 244 seconds]
<wens>
4.11, too late to get the dt bits in, and the mfd part for a23 isn't in either
<KotCzarny>
cant convince them that dt bits arent unsafe?
<miasma>
hm, 4.11 release will be around ~2017-04-23. opi pc is probably pretty obsolete already at that point. there's still some functionality left that needs patches, e.g. the usb otg ;-/
<mripard>
KotCzarny: it's not really the point. The policy is to have all patches in next for two weeks at least so that they get proper testing, make sure it compiles in all scenarios, and so on
<mripard>
4.10 merge window opens in a bit more than a week
philippe_fouquet has joined #linux-sunxi
<wens>
sigh, r40 dram code does not match what is setup by boot0 from the bpi image
terra854 has quit [Quit: Connection closed for inactivity]
florianH has joined #linux-sunxi
bugzc has quit [Ping timeout: 260 seconds]
terra854 has joined #linux-sunxi
fkluknav_ has joined #linux-sunxi
Macer_ has joined #linux-sunxi
Macer has quit [Read error: Connection reset by peer]
Macer_ is now known as Macer
ErwinH has joined #linux-sunxi
vbmithr has quit [Ping timeout: 260 seconds]
vbmithr has joined #linux-sunxi
codekipper has joined #linux-sunxi
pitillo has quit [Ping timeout: 260 seconds]
pitillo has joined #linux-sunxi
orly_owl has quit [Quit: leaving]
pulser has quit [Quit: No Ping reply in 180 seconds.]
bobryan has quit [Ping timeout: 244 seconds]
pulser has joined #linux-sunxi
speakman has quit [Remote host closed the connection]
speakman has joined #linux-sunxi
speakman has joined #linux-sunxi
speakman has quit [Changing host]
montjoie has quit [Ping timeout: 260 seconds]
montjoie has joined #linux-sunxi
bobryan has joined #linux-sunxi
silviop has joined #linux-sunxi
<silviop>
kernel 4.9-rc6 A13 tablet usb mouse not working , only mouse (usb keyboard ok) any ideas ?
<silviop>
I use an HUB
<silviop>
root@tablet0:~# lsusb -t
<silviop>
/: Bus 02.Port 1: Dev 1, Class=root_hub, Driver=musb-hdrc/1p, 480M
<silviop>
|__ Port 1: Dev 2, If 0, Class=Hub, Driver=hub/4p, 480M
<silviop>
|__ Port 1: Dev 3, If 0, Class=Human Interface Device, Driver=usbhid, 1.5M
<silviop>
|__ Port 1: Dev 3, If 1, Class=Human Interface Device, Driver=usbhid, 1.5M
<silviop>
|__ Port 2: Dev 6, If 0, Class=Human Interface Device, Driver=usbhid, 1.5M
<silviop>
|__ Port 4: Dev 5, If 0, Class=(Defined at Interface level), Driver=asix, 480M
<silviop>
/: Bus 01.Port 1: Dev 1, Class=root_hub, Driver=ehci-platform/1p, 480M
<silviop>
|__ Port 1: Dev 2, If 0, Class=Vendor Specific Class, Driver=r8188eu, 480M
<silviop>
.config semms ok
<silviop>
device-enumerator: scan all dirs
<silviop>
device-enumerator: scanning /sys/bus
<silviop>
device-enumerator: scanning /sys/class
<silviop>
unable to receive message
<silviop>
Bus 001 Device 002: ID 0bda:8179 Realtek Semiconductor Corp. RTL8188EUS 802.11n Wireless Network Adapter
<silviop>
Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
<silviop>
Bus 002 Device 005: ID 2001:1a00 D-Link Corp. DUB-E100 Fast Ethernet Adapter(rev.A) [ASIX AX88172]
<silviop>
Bus 002 Device 006: ID 046d:c050 Logitech, Inc. RX 250 Optical Mouse
<silviop>
Bus 002 Device 003: ID 0461:0010 Primax Electronics, Ltd HP PR1101U / Primax PMX-KPR1101U Keyboard
<silviop>
Bus 002 Device 002: ID 1a40:0101 Terminus Technology Inc. Hub
<silviop>
Bus 002 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
<silviop>
dmesg at mouse connection :
<silviop>
[ 346.686338] usb 2-1.2: new low-speed USB device number 6 using musb-hdrc
<silviop>
[ 346.840233] input: Logitech USB-PS/2 Optical Mouse as /devices/platform/soc@01c00000/1c13000.usb/musb-hdrc.1.auto/usb2/2-1/2-1.2/2-1.2:1.0/0003:046D:C050.0004/input/input7
<silviop>
[ 346.853953] hid-generic 0003:046D:C050.0004: input: USB HID v1.10 Mouse [Logitech USB-PS/2 Optical Mouse] on usb-musb-hdrc.1.auto-1.2/input0
<silviop>
mouse appears in /dev/input but not work
<jelle>
silviop: pastebins are really a great thing
<KotCzarny>
silviop: always use pastebin for anything bigger than 1-2 lines
<silviop>
ok ok
<silviop>
there is a way to debug why not work ?
<KotCzarny>
use another port
<NiteHawk>
especially as it seems your device was connected to the OTG port ("MUSB") there
f0xx has quit [Ping timeout: 240 seconds]
<silviop>
i swap port with keyboard
<silviop>
keyboard work mouse not
<silviop>
otg is set in host only mode in dts
<KotCzarny>
maybe your mouse needs some additional kernel module
<MoeIcenowy>
silviop: have you enabled evdev?
<KotCzarny>
or that ^
<MoeIcenowy>
CONFIG_INPUT_EVDEV=y or m
<KotCzarny>
MoeIcenowy: kb wouldnt work without evdev too
<MoeIcenowy>
KotCzarny: no
<MoeIcenowy>
for virtual console, evdev is not used
<MoeIcenowy>
but under X11 evdev is needed for keyboard
<MoeIcenowy>
and for gpm, the "legacy mouse" is needed
<MoeIcenowy>
CONFIG_INPUT_MOUSEDEV=y or m
<silviop>
under X11 keyboard work
<silviop>
i'm cheking .config
<MoeIcenowy>
can you ls /dev/input/
<silviop>
in input there is mouse0 and mouse1 (mouse 0 = touchscreen)
<silviop>
mousedev enabled
<MoeIcenowy>
how about event* ?
<silviop>
mouse1 disappear if i disconnect mouse
<MoeIcenowy>
you should have at least 4 event devices
<MoeIcenowy>
oh at least 5 for a q8 tablet
<MoeIcenowy>
one for the power key, one for volume key, one for touchscreen, one for external kbd, one for external mouse
<silviop>
5
<MoeIcenowy>
could you install evtest program and execute it?
<silviop>
Available devices:
<silviop>
/dev/input/event0:axp20x-pek
<silviop>
/dev/input/event1:ektf2127
<silviop>
/dev/input/event2:1c22800.lradc
<silviop>
/dev/input/event3:USB Keyboard
<silviop>
/dev/input/event4:USB Keyboard
<silviop>
Select the device event number [0-4]:
<silviop>
but there are 5 events
<silviop>
strange
<MoeIcenowy>
plz pastebin ;-)
<silviop>
no mouse as you see
<silviop>
two keyboards but no mouse
<silviop>
cat /proc/bus/input/devices
<silviop>
sorry 6 events
berkutta has joined #linux-sunxi
<silviop>
should i not compile kernel with legacy psaux support ?
kivutar has joined #linux-sunxi
zzeroo has joined #linux-sunxi
zzeroo has quit [Client Quit]
zzeroo has joined #linux-sunxi
f0xx has joined #linux-sunxi
LargePrime has quit [Ping timeout: 240 seconds]
Wizzup has quit [Remote host closed the connection]
Wizzup has joined #linux-sunxi
paulk-collins has joined #linux-sunxi
leviathanch has joined #linux-sunxi
LargePrime has joined #linux-sunxi
JohnDoe71rus has joined #linux-sunxi
JohnDoe_71Rus has quit [Read error: Connection reset by peer]
lordlod_ has quit [Ping timeout: 268 seconds]
lordlod has joined #linux-sunxi
meridion_ has joined #linux-sunxi
meridion has quit [Ping timeout: 260 seconds]
duracrisis has quit [Ping timeout: 260 seconds]
duracrisis has joined #linux-sunxi
Seppo has quit [Remote host closed the connection]
enrico__ has joined #linux-sunxi
<wens>
mripard: what do you mean by "the current approach doesn't work"?
<mripard>
wens: the current approach used by jean francois
<wens>
jemk: do you have any references for the h3/a64 dram controller?
<montjoie>
starting to really modify stmmac, not easy, all dma_ops need to be rewritten, some core file also
bobryan has quit [Ping timeout: 250 seconds]
iamfrankenstein1 has joined #linux-sunxi
marvs_ has joined #linux-sunxi
bobryan has joined #linux-sunxi
bobryan has quit [Changing host]
bobryan has joined #linux-sunxi
fire219_ has joined #linux-sunxi
lennyraposo1 has joined #linux-sunxi
iamfrankenstein has quit [*.net *.split]
lennyraposo has quit [*.net *.split]
marvs has quit [*.net *.split]
fdcx has quit [*.net *.split]
fire219 has quit [*.net *.split]
aliosa27 has quit [*.net *.split]
iamfrankenstein1 is now known as iamfrankenstein
<jemk>
wens: some regs are similar to the zynq documentation, everything else is mostly based on reproducing register dumps
<jemk>
wens: some small parts are from the bsp sources too, but there are no exactly fitting ones
fdcx has joined #linux-sunxi
aliosa27 has joined #linux-sunxi
<wens>
i got the dram code mostly working on the r40
<wens>
however i only get half the capacity
<wens>
the autodetection code gives row=15, page=4096
<jemk>
the rank and buswidth detection doesnt work on h5 anymore, maybe r40 is similar
<codekipper>
hi all
<wens>
:(
<wens>
codekipper: hi
<codekipper>
I'm struggling to netboot my pine64 with my own kernel
<wens>
jemk: otherwise i've got most of the registers matching boot0, except for some that aren't touched by either the released source code or u-boot spl
<codekipper>
however netboot works with the armbian kernel and dtb. On my setup it complains "sunxi-mmc 1c0f000.mmc: fatal err update clk timeout"
<jemk>
wens: but i havent seen any dualrank sunxi yet, and bw default is 32
<codekipper>
is there any patches/config setup that are doing the rounds.
<wens>
jemk: well the ram chip is supposed to be 16 rows, not 15
<wens>
also switching dx ddr_clk & hdr_clk to dynamic breaks it
leviathanch has quit [Ping timeout: 265 seconds]
<jemk>
wens: can you send me the register dumps? (boot0&spl) then i'll look at it later today
<jemk>
wens: i'll look at it more closely when i'm on pc again, but it looks like its more similar to h5, esp. the pir_qsgate wont work, try removing it
Andy-D has joined #linux-sunxi
fkluknav_ has quit [Remote host closed the connection]
fkluknav has joined #linux-sunxi
<KotCzarny>
maybe name building blocks with letters and create table for soc comparison? ie. a20 dram A, h3 dram B, pine64/h5 dram C
<KotCzarny>
:)
* MoeIcenowy
bought a 1024x600 Q8 tablet
<KotCzarny>
1GB ram? ;)
<Wizzup>
yay!
<MoeIcenowy>
512mb still
<Wizzup>
the one I have shows 1GB (on mainline)
JohnDoe71rus has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org]
JohnDoe_71Rus has joined #linux-sunxi
<silviop>
q8 and 1Gb ?
<silviop>
does someone try to use rtl8723bs_bt for bluetooth ?
<MoeIcenowy>
silviop: me
Nacho has quit [Remote host closed the connection]
<silviop>
work out of the box ?
<MoeIcenowy>
needs dt alter
<silviop>
to add serial port ?
<MoeIcenowy>
yes
<silviop>
do you have a link to serial port node ?
Nacho has joined #linux-sunxi
<MoeIcenowy>
see sun8i-a33-inet-d978-rev2.dts
<silviop>
ok i see i should copy four nodes
<silviop>
choose leds r_uart &uart1
Nacho has quit [Remote host closed the connection]
Nacho__ has joined #linux-sunxi
<MoeIcenowy>
silviop: leds is not needed
<MoeIcenowy>
onlu uart1 is needed
<silviop>
dw-apb-uart: probe of 1c28400.serial failed with error -22
<silviop>
leds is a typo i intend aliases
<MoeIcenowy>
please check your sun8i-a23-a33.dtsi's uart1_pins_a section
<MoeIcenowy>
if it have only two properties
<MoeIcenowy>
then you are using a not-new-enough version
kotky has joined #linux-sunxi
<silviop>
4.9-rc5
<silviop>
it has two pins
<silviop>
PG6 and PG7
<MoeIcenowy>
is there allwinner,pull and allwinner,drive in it?
<silviop>
No
<silviop>
can i copy uart0 ?
<silviop>
4.9-rc6 is as you describe
<silviop>
i try to backport
fkluknav has quit [Ping timeout: 248 seconds]
leviathanch has joined #linux-sunxi
tlwoerner__ has joined #linux-sunxi
tlwoerner_ has quit [Ping timeout: 244 seconds]
tlwoerner__ has quit [Ping timeout: 245 seconds]
tlwoerner__ has joined #linux-sunxi
<silviop>
[ 3.017287] 1c28400.serial: ttyS0 at MMIO 0x1c28400 (irq = 33, base_baud = 1500000) is a U6_16550A
<MoeIcenowy>
congrats ;-)
fkluknav has joined #linux-sunxi
<silviop>
./rtk_hciattach -n -s 115200 /dev/ttyS0 rtk_h5 give interesting output , but finish with Can't set line discipline: Invalid argument \n Can't initialize device: Invalid argument
meridion_ is now known as meridion
<silviop>
may some kernel .config
<silviop>
i check is all is in place
<MoeIcenowy>
have you enable uart bluetooth and h5 (3-wire) protocol in .config?
<MoeIcenowy>
CONFIG_BT_HCIUART_3WIRE=y
<silviop>
no i not enbale bluetooth in kernel!!! Stupid
<silviop>
too much things to check :-(
<MoeIcenowy>
hahahahahaha ;-)
<silviop>
if you enable a lot of things there are conflicts, if you not enable there is something not working is not so simple
tsuggs has quit [Remote host closed the connection]
VargaD has quit [Ping timeout: 264 seconds]
<silviop>
ok seems all working, i will dedicate to gmeter
<silviop>
gsensor
<miasma>
i'd love to run linux on my cherry trail tablet, but the digitizer (among other things) isn't supported by mainline :-/ maybe there are some alternatives (with free drivers) that can be installed on top via usb/i2c
<miasma>
it's pretty lame how even the audio isn't supported. there are some patches in github
<silviop>
ok not by mainline but from patch ?
<scelestic>
miasma: after a few tries i gave up on my cherry trial tablet, for whatever reason the device just hung
<silviop>
For my experience from launch to decent mainline support can pass 1-2 years on new tablet
<silviop>
but i'm based on arm
lkcl has joined #linux-sunxi
<miasma>
maybe allwinner tablets are better supported, but the specs looked promising for the x86 tablet. 4gb ram, 64gb storage, fullhd, 802.11n, bt 4.0, even usb3 iirc. around $200
<miasma>
like 8-9 months ago
tlwoerner__ is now known as tlwoerner
<MoeIcenowy>
miasma: digitizers usually do not matter
<MoeIcenowy>
but the touchscreen chip do matter
<silviop>
x86 as acpi , is simpler to support , arm not
tlwoerner has quit [Quit: Leaving]
tlwoerner has joined #linux-sunxi
<miasma>
MoeIcenowy: yea i was thinking about replacing it totally with something usb connected :)
<MoeIcenowy>
USB mouse ;-)
<miasma>
sure
<miasma>
it's just not as handy
VargaD has joined #linux-sunxi
<silviop>
does someone know mir3da i2c gsensor ?
<silviop>
It is'not in mainline , and i did'n found a patch
<MoeIcenowy>
to be honest, a GNOME can run on my Q8
<MoeIcenowy>
with a single-page epiphany
<MoeIcenowy>
but usually system with GNOME have two GNOME sessions running :-(
<MoeIcenowy>
(user one, gdm one
<MoeIcenowy>
when I'm doing the experiment, I used lightdm to login into GNOME, otherwise the second GNOME Shell will OOM
<KotCzarny>
o.O
<KotCzarny>
guys, remember windows95? running smoothly on 16MB of ram?
<MoeIcenowy>
windows 95 do not use as much Javascript as GNOME Shell ;-)
<KotCzarny>
and now we see the problem. bloat.
<MoeIcenowy>
We may expect freemangordon's hildon port
<KotCzarny>
yup, but they plan on porting to gtk3, eh
<mripard>
KotCzarny: the OS that didn't have IP nor USB support ?
<KotCzarny>
mripard, IP ? and usb landed there in last revision ;)
Putti has quit [Ping timeout: 265 seconds]
<mripard>
wasn't USB added in 98 ?
<KotCzarny>
nop. there was an update for 95 too
<KotCzarny>
win95 osr2.1
<MoeIcenowy>
low budget and touch friendly systems are really difficult to find
<mripard>
KotCzarny: my bad, but the argument is still valid though. It's a pretty weak argument to compare a system that was running on a hardware that was much simpler back then
<mripard>
there's probably some bloat in our current systems
<mripard>
but I'm pretty sure we'll say the same thing in 20 years :)
<KotCzarny>
yeah, im just amazed what we had then, what we considered top-tier hw
<KotCzarny>
and what our internet/os experience was
Putti has joined #linux-sunxi
chomwitt has joined #linux-sunxi
<ErwinH>
Did some tests with the OPi PC2 and the tx-delay and rx-delay aren't read properly from the DTB file. When I change the value's in the source code I can see changes in the throughput. (rx-delay at 31 and there isn't gigabit connection anymore)
DullTube has quit [Quit: Leaving]
Seppoz has joined #linux-sunxi
Putti has quit [Ping timeout: 250 seconds]
cnxsoft has quit [Remote host closed the connection]
yann-kaelig has joined #linux-sunxi
afaerber has joined #linux-sunxi
victhor has joined #linux-sunxi
Pepe has joined #linux-sunxi
lkcl has quit [Read error: Connection reset by peer]
leviathanch has quit [Remote host closed the connection]
lkcl has joined #linux-sunxi
leviathanch has joined #linux-sunxi
<jemk>
wens: closer look says: same as h5, the dqs gate training isn't used anymore. it wasn't even necessary in h3, because the gate is always open, but there it didn't hurt and is used for rank/buswidth detection. now something changed, maybe they removed the hardware feature, and it messes up the whole process
<jemk>
this will make the following if useless, we have to find out how to do this detection in h5/... but since most boards have 32bit and only 1 rank defaults work
<willmore>
ErwinH, I think tkaiser had looked at that. At least he had for the A64.
<MoeIcenowy>
maybe furthermore banana pis will have more ram ;-)
<MoeIcenowy>
someone from bpi said that it's laughed to be "8 banana pis 2 gb ram" ;-)
<willmore>
So, 256MB each?
fkluknav has joined #linux-sunxi
longsleep has quit [Remote host closed the connection]
libv_ is now known as libv
longsleep has joined #linux-sunxi
<MoeIcenowy>
willmore: 2gb each
<willmore>
Oh, 2Gb each?
* willmore
ducks
<willmore>
Sorry, MoeIcenowy, couldn't resist.
reinforce has quit [Quit: Leaving.]
pitillo has quit [Changing host]
pitillo has joined #linux-sunxi
lemonzest has joined #linux-sunxi
leviathanch has quit [Remote host closed the connection]
<pitillo>
hello, does some know about a good power bus usb to power around 7 devices?
<pitillo>
I've seen anker and key trademarks and I'm thinking in the anker, any opinion and/or sugestion?
<wens>
anker is good
<wens>
i have one that has 10 usb power ports
JohnDoe_71Rus has joined #linux-sunxi
<pitillo>
I've been reading good reviews and check that there is a 8 port option
<pitillo>
s/check/checked
leviathanch has joined #linux-sunxi
kotky has quit [Quit: Page closed]
<willmore>
Are these devices SBCs or something else, pitillo?
<MoeIcenowy>
usb hubs ;-)
netlynx has joined #linux-sunxi
netlynx has joined #linux-sunxi
netlynx has quit [Changing host]
<willmore>
I'll just say that powering an SBC with a USB connection is just asking for problems.
lordlod has quit [Ping timeout: 258 seconds]
<willmore>
That probably should be 'a SBC'.
lordlod has joined #linux-sunxi
<miasma>
pitillo: so do they also need to transfer data or just power the boards?
<ErwinH>
tkaiser: Your findings were right, the value isn't read correctly from the DTB. If you alter the source the delay is applied, but does not give much improvement. Max TX-speed I've seen so far is 58 MB/s, but have to recompile every time I change a value.
<tkaiser>
ErwinH: Well, lets wait for apritzel's findings. Based on tests with A64 I would leave RX-delay at 0 and just test through TX-delay from 0 to 7 and see what's going on. BTW: Maybe the problem is related to loading initrd.img.
nOOb__ has joined #linux-sunxi
lennyraposo has quit [Quit: Leaving.]
lennyraposo has joined #linux-sunxi
reinforce has joined #linux-sunxi
LargePrime has quit [Ping timeout: 246 seconds]
lennyraposo has quit [Client Quit]
nOOb__ has quit [Ping timeout: 258 seconds]
lennyraposo has joined #linux-sunxi
yeoldetoast has quit [Remote host closed the connection]
f0xx has quit [Ping timeout: 246 seconds]
Putti has joined #linux-sunxi
nOOb__ has joined #linux-sunxi
<tkaiser>
ErwinH: And please don't forget that the Xunlong OS images are currently limited to 1008 MHz cpufreq. Since iperf/iperf3 are CPU bound that limits numbers you get. Which OS image do you use for your tests?
<tkaiser>
ErwinH: I don't think throttling is an issue here but iperf3 process and IRQ processing running on the same core. Did you had a look in the github issue in the script's CheckPrerequisits function? This should prevent running both iperf task and IRQ processing on the same core.
Putti has quit [Quit: Leaving]
<tkaiser>
ErwinH: Also without running htop in another shell you never get the idea when everything is just bottlenecked by one CPU core being busy 100%.
<pitillo>
willmore: sorry for late answer. I want to connect some devices (cb/cb2/opipc/odroidxu4)
leviathanch has joined #linux-sunxi
popolon has joined #linux-sunxi
<pitillo>
some SBC... just to rack them and avoid a big amoubt of cables
<ErwinH>
Different core helps a lot!
<ErwinH>
74MB/s atm.
<tkaiser>
ErwinH: Well at least to check for the best TX/RX combination this seems mandatory, since otherwise result variations is way too high to draw any conclusions.
<tkaiser>
ErwinH: And please keep in mind to always check both directions. And please use iperf3 since this also shows retransmit errors. Maybe just using the script I used? ;)
<MoeIcenowy>
freemangordon: what panel do you used for your 1024x600 Q8?
orly_owl has joined #linux-sunxi
<ErwinH>
Big parts of the script you used :)
nove has joined #linux-sunxi
The_Loko has joined #linux-sunxi
<willmore>
pitillo, then please don't use usb for power. Most of those have proper barrel power connectors. Get a good quality 5V supply, some wire, and some of the right kind of jacks for the boards.
<willmore>
You might also want to put some fusing in those power runs, pitillo. You don't want a large power supply to put all its power through one little board because a screw dropped onto a GPIO header and hit the wrong pins.
<TheLinuxBug>
they muist have only had a small supply
<MoeIcenowy>
freemangordon: mine also works with it ;-(
<MoeIcenowy>
s/;-(/;-)
<MoeIcenowy>
oh I forgot to say that I bought a 1024x600 Q8 ;-)
philippe_fouquet has quit [Remote host closed the connection]
<freemangordon>
MoeIcenowy: good :)
<MoeIcenowy>
but also only 512MB ram...
<freemangordon>
MoeIcenowy: does android report 512 as well?
<MoeIcenowy>
I didn't care how much Android reports
<freemangordon>
why?
<freemangordon>
does it lie?
<MoeIcenowy>
it may lie
<tkaiser>
TheLinuxBug: They were already gone when Jean-Luc posted about. I had a look on monday and everything was sold out (both LeMaker devices, the useless one too)
fkluknav has quit [Ping timeout: 258 seconds]
<MoeIcenowy>
but yesterday we found out how to get the real value under Android
<MoeIcenowy>
use "busybox free"
<freemangordon>
hmm, now I remember there was a .ko colled something like "memorysizehack.ko" or somesuch
<KotCzarny>
o.O
<freemangordon>
MoeIcenowy: waht about /proc/meminfo?
<MoeIcenowy>
/proc/meminfo is easy to cheat
<MoeIcenowy>
freemangordon: if you can get the .ko, could you send it somewhere?
<KotCzarny>
open it and inspect drams?
<KotCzarny>
if they are nonames, most likely fake
<freemangordon>
MoeIcenowy: ok, gimme some time to boot the tablet
<MoeIcenowy>
thanks ;-)
<MoeIcenowy>
my tablet is getting its Linux rolling up
<freemangordon>
BTW, is sound supposet to work with your kernel?
<freemangordon>
*xuppsed
<freemangordon>
aaah
<freemangordon>
you got it :)
<MoeIcenowy>
headphone is
<MoeIcenowy>
if you want speaker
<MoeIcenowy>
manually enable its GPIO
<freemangordon>
ok
<MoeIcenowy>
I forgot the value
<MoeIcenowy>
(gpio id)
<MoeIcenowy>
oh it's PH9 (233)
<MoeIcenowy>
so please manually poke gpio233 to out mode and value 1
<MoeIcenowy>
and I found some headphones will lead to the audio mixer to behave incorrectly (it's a mysterious thing :-(
<freemangordon>
I am far from audio still, but anyway it is good to know
<pitillo>
willmore: yeah, I understand what you told but lot of home made cluster reviews I've seen are using power bus to power them. I have power supplies for all devices but they are eating lot of space at my desktop. I was looking for an easy way to sort them (trying to avoid soldering/electrical boards
<MoeIcenowy>
the audio driver is by Mylene.
<dgp>
pitillo: A higher voltage (48v or something) DC bus and then local lower voltage supplies seems like a good idea to me. The local supplies can do over current protection etc without killing all of the nodes
<pitillo>
the real question is if power bus provides a stable V/A for all busses, If they do, I can't see any problem if I attach an extra UPS to the power bus (keeping in mind the main question, if it provides stable V/A in all sockets)
<pitillo>
dgp: I think these anker power buses are ~50V which shouñd be enought for 8 usb ports (theorically)
<KotCzarny>
MoeIcenowy: headphones with additional ring?
<freemangordon>
MoeIcenowy: ok, it seems I saw this in dmesg "[82.298658] meminfo_proc_show: fake_dram script_get_item err." not a .ko
<KotCzarny>
or maybe just different resistance
<MoeIcenowy>
KotCzarny: ah-oh
<MoeIcenowy>
my old lichee do not have fake_dram things
<MoeIcenowy>
KotCzarny: I know nothing for electronics.
<MoeIcenowy>
I'm a CS student
fkluknav has joined #linux-sunxi
<KotCzarny>
MoeIcenowy: im just guessing. but you should know a bit how electricity/magnetism work
<KotCzarny>
apart from being useful, its fun
Pepe has joined #linux-sunxi
<KotCzarny>
fmg: maybe that fake started living as a testing module, then got snagged by some greedo
<KotCzarny>
offtopic, ld.gold eating ~5GB of ram, madness
nOOb__ has quit [Ping timeout: 260 seconds]
The_Loko has quit [Read error: Connection reset by peer]
<NiteHawk>
what the heck are you linking? :D
<KotCzarny>
droid. libwebchromium or something. told you, offtopic. ;)
The_Loko has joined #linux-sunxi
<KotCzarny>
sad part is that could be 10x smaller, just cant pinpoint source of '-g' in cmdlines. and grepping 20gigs isnt that much fun
The_Loko has quit [Max SendQ exceeded]
nOOb__ has joined #linux-sunxi
yann|work has quit [Ping timeout: 245 seconds]
<willmore>
pitillo, a single power supply with several leads off of it will be smaller than a bunch of random power bricks and a power strip. You'll need something more than a USB cable for that XU4 anyway.
<willmore>
dgp, I like that idea. Get a bunch of those little 2A variable buck mode DC/DC boards for $.40 from aliexpress and put one on each SBC then use a 12V or higher buss to feed them all.
* willmore
doesn't buy anker stuff anymore. One bad power brick and lots of bad reviews by people who actually do testing have changed my mind on their 'quality'.
<KotCzarny>
willmore: im using meanwell rd-50-5 for my opipc+opi+2e, and rd-65b for bpi-r1 + 2hdd
asmir has quit [Ping timeout: 244 seconds]
<KotCzarny>
so far it works beautifully. and nice thing is regulated voltage, bumping to 5.2V helps with crappy micro-usb cable
IgorPec has joined #linux-sunxi
<KotCzarny>
s/cable/power jack/
<willmore>
Meanwell? That sounds like a very questionable brand name choice. "Was it a good device?" "No, but they meant well." :)
<KotCzarny>
:)
<pitillo>
willmore: there are lot of examples of odroidxu4 and odroidc2 using these usb power supplies... I'll keep reading more about possible issues you told and I'll look for mlre alternatives to build something clean (I can't share any review from phone to show you an example). Thank you very much dor your comments and time
<KotCzarny>
they arent stellar, but good quality for the price
<willmore>
pitillo, no worries. I wish you the best of luck. I just don't want you to come up with some big solution that dissapoints you. I have been over at #odroid for quite a while and the #1 recommendation for any kind of system problem is "you're not powering it via USB, are you?"
<willmore>
#2 is uSD cards that suck, right tkaiser?
<willmore>
My SBC power supply is a little 50W open frame that I put in an old PC power supply case for protection. I have another that was actually an old PC power supply--I just put some bannana jacks on it and a small load resistor on the 12V(?) rail. Having 20A at 12V can cause problems from time to time, though. :)
<KotCzarny>
wouldnt it be better to just use dual voltage brick?
<willmore>
KotCzarny, I already had the power supply board from surplus, so...
<KotCzarny>
all those little voltage regulators lose efficiency
<KotCzarny>
nothing beats free stuff ;)
<willmore>
KotCzarny, yes they do. One multi-output device is likely to be more efficient than a half dozen little things.
<willmore>
Old PC power supplies are effectively free, too. They make great power supplies. Especially the older ones with high current on the 5V rail. Nowdays all the power is on the 12V rail.
<willmore>
Great if you want to arc weld...
<KotCzarny>
but pc bricks tend to be bad on efficiency too. especially if you just use one voltage
<willmore>
True.
<willmore>
But: Free
<KotCzarny>
running inefficiently 24/7 eats free part sooner or later
leviathanch has quit [Remote host closed the connection]
<willmore>
I use the PC power supply as a bench supply, so it's not on unless I'm working on something.
<KotCzarny>
i had a bunch of unused atx bricks, but playing with power meter quickly resulted in 'i need good, specialized power brick'
massi has quit [Remote host closed the connection]
<KotCzarny>
those monsters can eat 20W just connected to 230V.
<MoeIcenowy>
does anyone who uses a Silead GSLX68X-controlled touch screen met the situation that the touch is very inaccurate in some area near the edge?
<MoeIcenowy>
I met this situation on my two tablets
<tkaiser>
willmore: The 'USB problem' is mostly related to crappy cables (resistance), the crappy Micro-USB connector (adding contact to cable resistance and being rather limited regarding maximum current) and that a few boards that have a real power barrel and can be powered through (Micro) USB behave different.
leviathanch has joined #linux-sunxi
<KotCzarny>
tkaiser: you can simulate crappy power by using microusb adapters on a barrel equipped board
<scelestic>
KotCzarny: yeah atx psu's are well known for their energy inefficiencies, at least 10 years ago, made me switch to better ones early
cptG has joined #linux-sunxi
cptG_ has quit [Ping timeout: 258 seconds]
Putti has joined #linux-sunxi
enrico__ has quit [Quit: Bye]
<tkaiser>
KotCzarny: And you can do the other way around. I've a couple of adapters from 5.0/2.1 barrel to Micro USB and use them with el cheapo PSUs from Beelink (5.2V). No problems with undervoltage any more, only issue is that the board shouldn't exceed 10W consumption since tiny contacts (Micro USB still being crap for this purpose).
<tkaiser>
A Micro USB jack simply encourages users to do it wrong, that's the most important part of the problem. If someone knows what he's doing then he can deal with it. If not then 'smart chargers' are used that provide not more than 500 mA or cables that lead to 1V voltage drop and so on.
<KotCzarny>
hmm
<KotCzarny>
maybe add a little script that runs @boot that waits for idle, measures voltage, does some work, measures again, if it detects too noticeable drop, print warning?
libv_ has joined #linux-sunxi
libv has quit [Ping timeout: 256 seconds]
lemonzest has quit [Quit: Leaving]
<tkaiser>
KotCzarny: No PMIC, no ADC, no voltage measurement. This is only something that's possible with tablet SoCs (A series) + AXP PMIC.
<tkaiser>
Including cpuburn in u-boot would be enough. Simply deadlock the board in early boot stage. No 'strange' symptoms later, no corrupted filesystem, just a clear indication of 'something's wrong here'.
avph has quit [Ping timeout: 246 seconds]
jernej has joined #linux-sunxi
<KotCzarny>
true. :/
<KotCzarny>
still, for pine64 it would help, as that community is clueless as heck
<KotCzarny>
but your solution is good either, hehe
<tkaiser>
With A64 it might work if the board design allows it. But I didn't find any sysfs node with BSP kernel so maybe even if there's an ADC no readout is implemented. And true, Pine64 is 'worst case' by definition here.
avph has joined #linux-sunxi
Dodger78 has quit [Remote host closed the connection]
fALSO has joined #linux-sunxi
popolon has quit [Quit: WeeChat 1.4]
|Jeroen| has joined #linux-sunxi
<willmore>
It may seem like a strange question, but, from looking at the power diagrams of a few boards, what do these SBCs even use 5V for? Mostly, they just use it to feed other regulators.
<KotCzarny>
usb, sata
<willmore>
tkaiser, some kind of stability testing would be a nice feature for armbian.
<willmore>
KotCzarny, okay, A20 boards and those crazy ones with USB->SATA.
<willmore>
But, those will have a need for more current to the drives than micro-USB is wise for.
<KotCzarny>
and now you see the problem
<KotCzarny>
;)
<willmore>
Oh, I've always been against micro-USB for power. I've not always been happy with the alternatives--the little connectors the c1 and C2 use annoy me. The larger 2.1 and 2.5 inner diameter ones are much better.
<miasma>
willmore: well, usb assumes 5v dc. hdmi, ethernet,. ..
<willmore>
miasma, do they? I thought the ethernet chips used 2.5V.
<willmore>
HDMI I know very little about. Is that for the DCC functionality?
Gerwin_J has quit [Quit: Gerwin_J]
<KotCzarny>
hmm, either my 2303 just fried after running logcat, or win7 went crazy
<miasma>
willmore: doesn't ethernet have +/- 2.5 ?
<miasma>
which might be easier to achieve with 5V
<willmore>
miasma, I think that happens via the transformer.
<miasma>
ok
<willmore>
Don't quote me, though. I should go review a schematic.
<miasma>
i'm not sure if they always use transformers
<miasma>
e.g. if the phy is builtin
<miasma>
it's easy to do a voltage divider
<miasma>
willmore: anyway, since the boards have usb host ports, they need 5V at least. also the gpio pins have 5V out
<KotCzarny>
5v on gpios are usually direct from dc-in
<KotCzarny>
on most boards. but not on all probably
<KotCzarny>
hmm. might be that usb host on my laptop hung. or died.
<KotCzarny>
mouse stopped working too
<KotCzarny>
whoop dee frigging doo.
<miasma>
btw there's this problem with uart in sunxi boards that current is accidently fed back via the uart pins. have you tried if adding a resistor there helps? would the communication still work?
<miasma>
KotCzarny: if you didn't break it, resetting the host might help
<miasma>
polyfuses might also take some time to recover
<willmore>
Maybe it was a bad choice, but looking at the schematic for the Opi Z, there is no direct use for 5V. But it does have HDMI and SPDIF as consumers of the 5V rail.
<miasma>
people also expect the boards to work with any phone charger these days :)
<miasma>
smaller MCUs are often also 3.3V, but people buy bigger versions of boards with micro usb and onboard regulators
<willmore>
The ethernet only uses 3.3V and that's just for the LEDs.
<willmore>
A differential signal feeds the ethernet transformer, that's where the 5V P-P comes from.
<KotCzarny>
umkay. rebooted windoze fixed. (dont know how to reset usb host on win7)
Dodger78_ has joined #linux-sunxi
<miasma>
willmore: ok, good to know. i've just assumed that the transformer only isolates without changing the voltages
<willmore>
miasma, if it's a 1:1 then it won't change the voltages. But, a 2.5V differential signal is 5V P-P.
<miasma>
right
<tkaiser>
willmore: The 5V 'requirement' originates from USB and HDMI.
Ntemis has joined #linux-sunxi
<KotCzarny>
what voltage camera runs?
<tkaiser>
3.3V and 1.8V
<willmore>
tkaiser, yeah, that's what I'm seeing. 5V goes to the HDMI and obviously to the USB. KotCzarny camera gets 5V, but regulates it lower, I would bet.
<tkaiser>
Camera on OPi One and PC with the 'extension board' in between works flawlessly with DC-IN at 4.5V
<willmore>
Ethernet runs from a 2.5V supply--now looking at the Opi PC2.
<willmore>
tkaiser, doesn't that board house a regulator?
<willmore>
So, the board should be sane down to 3.3V+regulator drop. HDMI, USB, and camera may not work, but the *board* should be okay.
<KotCzarny>
i would toss 5V on gpio as an early requirement, rpi compatibility was a must then
<willmore>
So, I am confused why boards get flakey at 4.8V.
<KotCzarny>
willmore: voltage drops
<willmore>
Oh, good point, KotCzarny. For completeness sake.
<willmore>
KotCzarny, which ones?
<KotCzarny>
willmore: cpu+usb + crappy psu ?
<willmore>
KotCzarny, I'm thinking of 4.8V *at the board*. So, after all the crap.
<KotCzarny>
nah, crappy psus drop voltage even further when load rises
vagrantc has joined #linux-sunxi
<tkaiser>
willmore: 4.8V there are fine for everything except USB and HDMI. There's 4.75 and 4.8 IIRC (or 4.6V)
<willmore>
I think I need to hook the bench power supply up to some boards via GPIO, load them up and do some testing. What was the recommended stability testing method? Some version of some matrix math.
libv_ is now known as libv
<tkaiser>
willmore: The problem is really that voltage drops under load and short load peaks then do the mess.
<willmore>
tkaiser, so I need to put a scope on the power, too and set it to trigger on those drops.
<KotCzarny>
if you want to test load just run cpuburn-a7
<tkaiser>
willmore: Light load is 'sysbench' (useless for everything else, especially benchmarking), heavy load is cpuburn
<willmore>
tkaiser, there was a program that did matrix math--and which I'm blanking on--which did some checking and could tell if operations had failed.
<tkaiser>
Linpack, the NEON optimized variant. Uses OpenBLAS.
<willmore>
That's it.
<tkaiser>
willmore: But that's different things: 2 sorts of undervoltage ;)
<tkaiser>
DC-IN vs. VDD_CPU.
<willmore>
Good point. Another scope probe on the CPU VDD line, then, too.
<jernej>
NiteHawk: You can also update A10 user manual from Allwinner's github. This is the first datasheet where I saw TV out registers description. Old pdf on wiki doesn't include it.
jernej has quit [Quit: Konversation terminated!]
jernej has joined #linux-sunxi
quinward has quit [Quit: WeeChat 1.5]
<NiteHawk>
jernej: it's a wiki, feel free to upload those file and edit the page yourself?
<jernej>
well, you need special permissions for it?
<jernej>
let me make an account first :)
apritzel has joined #linux-sunxi
Putti has quit [Ping timeout: 244 seconds]
<NiteHawk>
jernej: afaik - no, editing rights should suffice
<NiteHawk>
ah, okay - i see you're already at it. nice
<jernej>
NiteHawk: I noticed. I already added, although A10 documentation section seems to have URLs instead of wiki links
<NiteHawk>
yes, some are pointing to the older (FTP, i think?) server where it's harder to upload files