<oliv3r>
and that's why you should run opensource software
<oliv3r>
go go openwrt
bamvor has quit [Remote host closed the connection]
<focus>
keebler: droooooollll!!!! how much are they planning to sell the module at and the mboard - hint will be fine - i'll order 2 for my company
EGM has joined #linux-sunxi
<EGM>
Hi guys, anyone of you encountered problems with USB devices (kernel does not detect disconnection of USB devices in some cases)?
massi_ has joined #linux-sunxi
balage has joined #linux-sunxi
<balage>
arokux1, hi, are you there? i was offline for some time but had time to test sunxi-3.4 on my hackberry A10 for usb issues. I though I would share my findings and also look for some news on usb frontier :)
mcbrick has quit [Ping timeout: 272 seconds]
deasy has quit [Ping timeout: 240 seconds]
<balage>
arokux1, so in short. i managed to have usb buses seen by dmesg and lsusb, however none of them actually works. I mean when I plug an usb device (kb, usb stick, stick over passive usb hub) nothing happens, neither in dmesg nor in lsusb output. so I experienced the same with sunxi-nighly and Fedora19 kernel from Hans de Goede.
<balage>
arokux1, the most I could achieve (not 100% reproducible) is to have some outputs in dmesg saying 'ehci_irq: port change detect' but nothing else.. so I am stuck with 3.0 at the moment.
soul has joined #linux-sunxi
soul is now known as Guest721
Soru has quit [Ping timeout: 272 seconds]
<kriegerod>
balage: maybe power issues?
<balage>
kriegerod, then why does it work with same fex file, same hardware plugged in but on 3.0?
<kriegerod>
3.0 what?
<kriegerod>
kernel?
<balage>
kriegerod, sunxi-3.0 (3.0.76 at the moment)
<EGM>
I have problems with USB on sunxi 3.4.x (3.4.43 and 3.4.61), but not on 3.3.0.
popolon has joined #linux-sunxi
<mnemoc>
tried on stage-3.4 ?
<balage>
EGM, other people having usb issues too, at least on Hackberry (http://goo.gl/LCfGJg)
<EGM>
balage: Oh, thanks for the link. I noticed it on Cubieboard, but it's somehow USB device specific...
<mnemoc>
cb1 or cb2?
<EGM>
cb2
BluesBoy has joined #linux-sunxi
<balage>
mnemoc, no I haven't, last time someone told me it is not necessary as it doesn't have usb changes over sunxi-3.4, but I could try in the afternoon..
<EGM>
Haven't tried stage-3.4
<mnemoc>
balage: just for the sake of testing. also there might be some a20/usb related changes on the ML which haven't been noticed by the lazy maintainer wannabe
<balage>
mnemoc, okay I'll try stage-3.4 in the after work and report here
<mnemoc>
thanks
<EGM>
mnemoc: Thanks for the prebuilts link!
<mnemoc>
EGM: they aren't really "nightly"... they only update when the corresponding branch updates
<mnemoc>
and tried every 3h
<EGM>
BTW, what does "nuclear" in version mean?
<mnemoc>
nuclear is the nickname of aw's devkit
<mnemoc>
usually implies android
<mnemoc>
like crane for a1
<mnemoc>
a10
<EGM>
Oh, ok.
<mnemoc>
the prebuilts are for each defconfig, not all of them are really usuful
<EGM>
Looks like my USB issues are bit different from what is described on miniand forums. But it seems related, because it works in older version than 3.4
\\Mr_C\\ has quit []
<mnemoc>
EGM: can you try with the latest -rN tag with N > 0 ?
<mnemoc>
added the rev... but filenames are kind of.... long :|
<balage>
arokux1, tried on kernel (without usb debug enabled in sw_hci_sunxi.h +37) plugged a passive usb2.0 hub in, then tried to plug a simple stick into its ports one-by-one
<balage>
arokux1, then I realized it's a non debug kernel, so booted a debug kernel up and it did not occur again
<oliv3r>
mnemoc: we should add references to the nightly kernels on the wiki
<oliv3r>
i see if i have time (read, feel like) doin gthat later
<oliv3r>
i have an emergency project i have to take care of in the next 7.5 days
<oliv3r>
i'm getting a new power meter, which can be read from, and i wanna monitor it :)
<Tsvetan>
I dont know if we need A23 board, no Ethernet, no SATA, no HD ...
<oliv3r>
Tsvetan: it's the replacement of the A13; e.g. dualcore a13
<oliv3r>
personally, it's more for tablets
<oliv3r>
but if its (much) cheaper then a20, it's a viable replacement for robotics
<mnemoc>
as a13, but there is an olinuxino with it
<oliv3r>
it's NOT pincompatible; it's the same size as a10s, but less pins
<mnemoc>
ok
<balage>
arokux1, though i noticed that /proc/ccmu changed a bit, newertheless which debug option I used: http://sprunge.us/UDPR
<arokux1>
balage: you should get additional output, even for non-debug kernel.
<arokux1>
balage: that is strange, debug option shouldn't change anything in that file.
<balage>
arokux1, I recompiled it without debug mode and it hasn't changed, ccmu output was the same, the only difference for my prior .configs was to disable rtl8192_sw and use rtlwifi :)
Guest721 has quit [Read error: Connection reset by peer]
<arokux1>
balage: the point is, with "sw_hci_sunxi.h +37" set to 1 you should see a whole lot of the output, in particular: "sw_hci_sunxi.c +765"
<EGM>
mnemoc: 3.4.43-r2 still has issues with USB :(
<mnemoc>
uh
<mnemoc>
EGM: and -r1 ?
<mnemoc>
uhm... that might not have A20 support
<arokux1>
EGM: which issues with USB? usb host or usb otg?
<balage>
arokux1, i see but nothing was displayed (side question, is it still necessary to power cycle the device between kernel changes?)
<arokux1>
balage: hm.. if you replace the kernel you must reboot. turning off and on is not necessary.
Soru has joined #linux-sunxi
<balage>
arokux1, just asking because older kernels did not 'boot' correctly, though I have updated u-boot also recently. ok, all is clear.
<arokux1>
balage: so did you reboot after "sw_hci_sunxi.h +37" set to 1
<EGM>
arokux1: Host, seems like kernel does not register disconnection of some devices.
<EGM>
For example, when you unplug wifi dongle (rtl8188 based), it still shows up in lsusb, iwconfig and such.
<arokux1>
EGM: ah, you were the one who sent an e-mabil about this to the ML?
<arokux1>
e-mail*
<EGM>
Yeah, I thought it got lost.
<balage>
arokux1, so what do you suggest, try it again with debug kernel (sw_hci_sunxi.h +37 set to 1) and plug in different thing? as it hasn't worked so far :( maybe with stage-3.4?
<EGM>
I am not familiar with mailing lists, to be honest...
<mnemoc>
can anyone suggest a better gmane group name to ask for a rename? we aren't a subporject of "netbook"....
<EGM>
I believe that that hub related error is in fact related to USB hub integrated in one deviced I tried, so it's probably not issue of CB nor sunxi
<arokux1>
balage: I suggest you to get the output of those debug lines. as said, you do not need to put to kernel to the debug mode (though I'm not sure what do you mean by this), you just need to put 1 in sw_hci_sunxi.h, rebuild your kernel and reboot with it. just for making sure you are booting the correct kernel you could add a random printk at sw_hci_sunxi_init in sw_hci_sunxi.c
<arokux1>
EGM: so is everything works as expected with another hub?
<EGM>
arokux1: Basically yes, but I believe that that hub works only one USB port, not in the other one.
<arokux1>
EGM: is your hub externally powered?
<EGM>
It can be externally powered, but it worked without external power.
vicenteH has joined #linux-sunxi
<arokux1>
EGM: it worth to try with external power to rule out these kind of issues.
JohnDoe_71Rus has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org]
<balage>
arokux1, yes I meant 'debug kernel' when sw_hci_sunxi.h +37 has been set to 1... I am a bit lost between my tests with different kernels, I'll do a clean test tonight. thx
<arokux1>
EGM: will USB devices disappear on disconnect if you have them disconnected directly from the board?
<arokux1>
balage: ok. just make sure you get those debug lines, only then we can move forward.
<EGM>
arokux1: No, I have wifi dongle connected right in the board and it does not disappear on disconnect.
<EGM>
arokux1: No, I have wifi dongle connected right in the board and it does not disappear on disconnect.
<EGM>
Ooops, sorry
<EGM>
Oh cra
<EGM>
Wow, when I disconnect the hub from board, it does not disappear either...
<EGM>
<3>hub 3-0:1.0: port 1 disabled by hub (EMI?), re-enabling... [ 1071.750523] hub 3-0:1.0: port 1 disabled by hub (EMI?), re-enabling... <6>usb 3-1: USB disconnect, device number 6 <6>usb 3-1.1: USB disconnect, device number 7 <6>usb 3-1: new high-speed USB device number 8 using sw-ehci <6>hub 3-1:1.0: USB hub found <6>hub 3-1:1.0: 5 ports detected <6>usb 3-1.1: new high-speed USB device number 9 using sw-ehci <6>generic-usb 0003:21
<arokux1>
EGM: what does the "uname -a" outputs right now?
<EGM>
Sorry for formatting, probably webchat caused that...
<EGM>
Linux colposcope 3.4.43 #1 SMP PREEMPT Mon Oct 14 10:53:56 CEST 2013 armv7l armv7l armv7l GNU/Linux
<EGM>
That's 3.4.43-r2
<mnemoc>
Tsvetan: btw, http://dl.linux-sunxi.org/nightly/linux-sunxi/ includes pre-builts (updated each 3h) of the different sunxi kernel branches. including the experimental (dts-based) 3.10 and mainline. and the .dtb files for your olinuxinos ;-)
<arokux1>
EGM: I see, I do not have any ideas right now, I'll try to reproduce your problem tonight with my boards.
<EGM>
arokux1: I can provide you my config, if you want. But I believe that it behaves same way using defconfig for sun7i
<mnemoc>
Tsvetan: they are built out of the defconfigs, but intended to be used by real users, so improvements to defconfigs are welcomed
<EGM>
arokux1: I will try to test it under stage/sunxi-3.4 today and post results...
<mnemoc>
if you detect relevant outstanding patches in the ML, please let me know
<arokux1>
EGM: yes, I think config shouldn't matter, but maybe it is a good idea to test it too. So just cat .config | curl -F 'sprunge=<-' http://sprunge.us
<EGM>
Should it be FEX related? Should I try fex file from 3.3.0 image that works as expected?
cubie has joined #linux-sunxi
<EGM>
arokux1: Ok, I just booted stage-sunxi-3.4 and it's same :(
<arokux1>
EGM: I do not know why it should be FEX related.
<mnemoc>
EGM: 3.3.0 is a AW's GPL violating and SOC-specific kernel, sunxi-3.* are integrated and community maintained. reviewing the changes done by AW and integrating them properly into the unified takes time and resources. any help is welcomed
<arokux1>
EGM: can you please try 3.4.43-r1?
<mnemoc>
EGM: what you are facing implies that we are missing something AW did in their 3.3 for a20/usb
<mnemoc>
arokux1: I think he did already...
<arokux1>
mnemoc: what works for EGM import/lichee-3.3/a20-dev or import/lichee-a20-3.3 ?
<arokux1>
mnemoc: and what is the difference between them?
<mnemoc>
import/lichee-3.3/a20-dev is based on a git drop
<mnemoc>
the other is a source drop
<mnemoc>
git drop has history ;-)
<arokux1>
EGM: which one works for you?
bamvor has joined #linux-sunxi
<mnemoc>
but it's from march. devices are usually shipped with a newer snapshot.... diffs need to be found
cubie has quit [Quit: Leaving]
<mnemoc>
a20-dev vs latest 3.3 from aw
<arokux1>
mnemoc: diffs between what? do we have other kernels shipped with new devices?
<mnemoc>
arokux1: 3.3 kernel in their 2.0 SDK and the -dev we got
<mnemoc>
to see what changed after march
<arokux1>
mnemoc: I see.
<EGM>
arokux1: I will try 3.4.43-r1 in a while.
<arokux1>
EGM: what exact kernel works for you as expected?
<mnemoc>
http://sprunge.us/cjdg .... interesting diff of the builddirs of sun7i and the others
<mnemoc>
^---- from the 3h-ly "nightly" builds
pirea has joined #linux-sunxi
<EGM>
Unfortunately, using that image, I have problems with wifi support... :\
<arokux1>
EGM: ok, I'll try to see if we have missed something during unification tonight. but please first try 3.4.43-r1 once you have time.
<arokux1>
EGM: thank you for helping us to track down this problem.
<arokux1>
EGM: what are you wifi problems? in-kernel rtlwifi and vendor's drivers work just fine, although the latter demonstrates superior performance.
<pirea>
arokux1 i will make nighly builds :)
<pirea>
where i can upload my nightly builds ? :)
<arokux1>
pirea: what nightly builds do you mean?
<pirea>
i will compile every night kernel from git on my cubieboard
<mnemoc>
pirea: we do nightly kernel builds out of all branches, with their defconfigs. help improving those defconfigs is appreciated
<arokux1>
pirea: mnemoc has already set up the each-3rd hour builds.
<arokux1>
mnemoc: do you think that link deserves the place on the main page of the wiki? maybe it should be a link to some other page, that explains all the branches...
<mnemoc>
feel free do improve the wiki as you find better :)
<mnemoc>
it's a wiki ;-)
<mnemoc>
next step will be to do the same for every board in u-boot-sunxi
<arokux1>
mnemoc: yes, I do it with pleasure, however it is better to ask before changing the main page
<mnemoc>
arokux1: :) +1
<arokux1>
mnemoc: how about building hardware packs with our sunxi-bsp?
<arokux1>
mnemoc: for some distros...
<mnemoc>
arokux1: the sunxi-bsp needs to be adapted
<mnemoc>
u-boot and the blobs have changed a lot
<mnemoc>
also we have projects that's don't cross compile yet
<mnemoc>
arokux1: so reviving the nightly hwpacks is also in the TODO, but implies more work
<mnemoc>
we need different branches on the bsp (mimicing what we have in the kernel)
<arokux1>
mnemoc: I see, hwpack automatic build could be cool I think. we may have a buildroot image to minimize it size and ask people to test it, if bisecting etc
<mnemoc>
a lot to todo, and too little time :(
<mnemoc>
err
<mnemoc>
a lot to do, but too little time :(
Sonicadvance1 has quit [Ping timeout: 272 seconds]
mcbrick has quit [Quit: Leaving]
Sonic1 has joined #linux-sunxi
Sonic1 is now known as Sonicadvance1
atiti has joined #linux-sunxi
jemk has quit [Remote host closed the connection]
balage has left #linux-sunxi [#linux-sunxi]
Sonicadvance1_ has joined #linux-sunxi
<EGM>
arokux1: Well, using 3.4.x kernel, 8192cu works fine. But under that 3.3.0 kernel, iwconfig shows two interfaces for one wifi dongle, and it does not work well (I can't connect to cubieboard configured as AP, using same configuration as in newer kernel)...
Sonicadvance1 has quit [Ping timeout: 246 seconds]
<popolon>
I would like to put my cubieboard2 in an old laptop wheremotherboard is near from hs
<popolon>
:)
<vgrade>
libv: if you think the chromebook keyboards suck then don't go near one of these
<popolon>
=> longer batterytime, lighter, less noise
<chazy->
vgrade: thanks, I'll give it a go!
<vgrade>
popolon: seen the Pi hacks with the atrix dock
<vgrade>
chazy-: maintainer mentions he's running nfs so looks ok
eebrah|PC has joined #linux-sunxi
<popolon>
interesting :)
<popolon>
but I alread have a laptop
<popolon>
with a big screen
<popolon>
and big keyboard
wingrime has joined #linux-sunxi
<popolon>
If only the pcmcia A20 modules would work on the pcmcia included ports
<popolon>
I could include 2 :)
<mnemoc>
gzamboni: did you do something to "enable" this pin?
<chazy->
vgrade: hmm, I already had those patches in my tree and I heard from ojn as well that he's using nfs, but I somehow can't get this thing to work.
<chazy->
I guess I can load an initrd and poke around
Soru has quit [Read error: Connection reset by peer]
<gzamboni>
mnemoc no, in fact i had an extension board and with the ext board it didnt work, dunno why because it was nc to anything. maybe its a gnd issue.
<Turl>
chazy-: I know ojn boots a cubietech board via the network
<Turl>
maybe he has an idea on what's wrong
<gzamboni>
Turl i made my i2c device work, it was set to spi, it works both using spi and i2c, but you have to set it thrue other pins
<chazy->
Turl: yeah, may wait for his wisdom and move onto something else. thanks for the input.
<Turl>
gzamboni: :) great
<Turl>
chazy-: yw
<Turl>
chazy-: you could try ethernet on uboot, I know for a fact that tftp works
<chazy->
Turl: which u-boot do you use? (wondering if it turns on some clock or configures the PHY layer that Linux needs…)
<wingrime>
Turl: a13 have emac?
<Turl>
chazy-: I don't recall the exact version, it's pretty dated now