bsdfox\ has quit [Read error: Connection timed out]
bsdfox\ has joined #linux-sunxi
ZetaNeta has joined #linux-sunxi
ZetaNeta has quit [Ping timeout: 246 seconds]
geecko has quit [Ping timeout: 272 seconds]
bsdfox\ has quit [Read error: Connection timed out]
bsdfox\ has joined #linux-sunxi
mndbzr has joined #linux-sunxi
<mndbzr>
hi
bsdfox\ has quit [Read error: Connection timed out]
bsdfox\ has joined #linux-sunxi
mndbzr has left #linux-sunxi [#linux-sunxi]
mndbzr has joined #linux-sunxi
apo has joined #linux-sunxi
<apo>
Hi
<apo>
Just got myself a cubietruck and configured the kernel successfully, for the most part, but was unable to find any options regarding the security stuff - hardware crypto (AES, etc) and RNG
<apo>
is there anything for that available, or will I have to wait?
<apo>
(Cubietruck uses the A20)
torindel has quit [Ping timeout: 264 seconds]
torindel has joined #linux-sunxi
tinti has quit [Quit: Leaving]
<mndbzr>
it seems there no one here
<mndbzr>
q apo
Sonicadvance1 has joined #linux-sunxi
<apo>
mndbzr: I count over a hundred... you mean nobody who actually knows anything about the boards? Or just not here right now?
bsdfox\ has quit [Ping timeout: 245 seconds]
<mndbzr>
right now I think
<apo>
that's fine, I can wait
<mndbzr>
first time here
<apo>
I'm about to go to bed anyway ;)
<mndbzr>
can you give me a little hand
<apo>
I don't know.
<mndbzr>
so you can go to bed with a clean consciuos
Sonicadvance1 has quit [Ping timeout: 246 seconds]
<apo>
montjoie[home] seems to know something about the hardware crypto. :D
<Nyuutwo>
I have disasemblered lcd.ko from Android
<Nyuutwo>
and I have LCD_panel_init in which is called lp079x01_init
Sonicadvance1 has joined #linux-sunxi
<Nyuutwo>
and also TC358768_init TC358768_666_init but I know that they don't run beacuse they use printk
bsdfox\ has joined #linux-sunxi
<Turl>
apo: there's no driver available for the hw crypto acceleration nor the PRNG so far
<apo>
Turl: aw, okay. Thanks, though.
<Turl>
Montjoie is working on one, you may want to check the status with him
<apo>
Neat
<Turl>
apo: meanwhile you can keep using the software-only solutions
<apo>
Turl: of course
<Turl>
:)
<apo>
(I'm using them for my encrypted ~ atm. It's sufficient for a box that only does IRC and Jabber ;))
<apo>
It's much better at IRC and Jabber than the fitpc 1.0 that it replaced, though. And compiles very much faster...
torindel has quit [Ping timeout: 272 seconds]
torindel has joined #linux-sunxi
torindel has quit [Ping timeout: 260 seconds]
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
benn-on-cubietru has joined #linux-sunxi
<benn-on-cubietru>
I am trying to use cubietruck as my PC(lubuntu 12.10 desktop). After moving rootfs to SSD, I feel good with it
* apo
still has lubuntu on the NAND, but only used it to bootstrap gentoo onto a SATA disk
<Turl>
hey benn-on-cubietru :)
<benn-on-cubietru>
:P
<benn-on-cubietru>
Turl, my SSD can get 106MB/s seq read, and 148MB/s seq write
<benn-on-cubietru>
on cubietruck
<Turl>
benn-on-cubietru: nice
<Turl>
benn-on-cubietru: on performance governor?
<apo>
SSD feels like a bit overkill, but I guess you don't exactly have to get a Samsung 840 Pro ;)
<apo>
(And I can get 100MB/s sequential writes with a normal SATA disk, but those aren't interesting)
<apo>
funny that the writing's faster
<benn-on-cubietru>
High IOPS can improve the user experience much
<apo>
don't need high IOPS, just dump everything into RAM!
<benn-on-cubietru>
I agree, we have 2GB ram now :D
* apo
has a laptop with SSD, boots from grub to chrome in 6 seconds
<Turl>
benn-on-cubietru: talking about ram, did you play with mbus? :)
<benn-on-cubietru>
Turl: no
<Turl>
benn-on-cubietru: we can get more memory bandwidth with a higher mbus
<Turl>
benn-on-cubietru: but we don't know if it is stable, or why allwinner set it to 300mhz only
<benn-on-cubietru>
apo: github/cubieboard/linux-sunxi (branch cubie/sunxi-3.4), wifi should work
<benn-on-cubietru>
Turl: if we want to change mbus 400mhz, we need to change the core vcc to 1.3v
<apo>
Then I'll try that, thanks again
<benn-on-cubietru>
I will upgrade the .config I am using currently for lubuntu to github.com/cubieboard/cubie_configs repo
<benn-on-cubietru>
apo: BTW, cb1,2 and ct share the same kernel now
<apo>
mm
<benn-on-cubietru>
apo: what is your problem on the bcmdhd wifi driver? compile or modprobe or others ?
<apo>
benn-on-cubietru: Can't compile it, due to the unknown field errors I mentioned above
<benn-on-cubietru>
apo: I am compiling the kernel with your kernel config
Gerwin_J has quit [Quit: Gerwin_J]
<apo>
I'm still cloning the git repo :P
<Turl>
benn-on-cubietru: so we need to undervolt core vcc?
<Turl>
benn-on-cubietru: uboot configures it to 1.4V if I am reading this right
<benn-on-cubietru>
wow! If it's 1.4v, it think we should be able to run mbus 400mhz
<benn-on-cubietru>
ALLWINNER tell me that they set it to 1.25V at bootloader. When benchmark, It wil be tuned to 1.3v.
<apo>
cheaters
<benn-on-cubietru>
apo: :)
<Turl>
benn-on-cubietru: I will confirm with hno when he is around
<Turl>
but axp152 driver says Vcore is dcdc2
<Turl>
and uboot set it to 1.4V
<benn-on-cubietru>
actually, mbus can work on 400mhz. but for PAD, power comsumption is very important
<Turl>
Vcpu is 1.25V hmm
<Turl>
benn-on-cubietru: maybe they meant Vcpu?
<Turl>
cpufreq controls vcore (dcdc2)
<benn-on-cubietru>
Turl: I confirmed it just now. dcdc2 is for CPU, the core vol we just refer to is dcdc3
<benn-on-cubietru>
Turl: for cubies, we configure it to 1.3v. with that, mbus 400mhz should work for cubies
<benn-on-cubietru>
dcdc3 is for system
<Turl>
dcdc3 is Vddr on axp152
<Turl>
(A10S)
<Turl>
let me check what is on 209
<benn-on-cubietru>
apo: I compile the kernel with your kernel config successfully
<apo>
benn-on-cubietru: cool
<apo>
benn-on-cubietru: So when I get up tomorrow I'll probably have working wifi :D
<apo>
... not that I really need it, since the board is stationary, but I don't like having broken stuff
<benn-on-cubietru>
apo: sorry but failed with modules target
<apo>
benn-on-cubietru: oh
<benn-on-cubietru>
apo: and I know the reason. wait a minute
* apo
waits
<Turl>
benn-on-cubietru: hmm, dcdc3 on axp20 seems to be ddr too
<benn-on-cubietru>
apo: you need to enable CONFIG_WEXT_PRIV
<apo>
I remember fiddling with that
<apo>
(still checking out git, though.)
<apo>
benn-on-cubietru: hm, but that's only selected by other things, I can't change it myself
<apo>
benn-on-cubietru: What should I change to enable it?
<benn-on-cubietru>
apo: a very simple to fix it is, enabling any wireless drivers, e.g. USB ZD1201, or RNDIS USB wireless driver will caused CONFIG_WEXT_PRIV selected
<apo>
benn-on-cubietru: Hm, shouldn't the BCMDHD enable it, if it needs it?
<benn-on-cubietru>
apo: the bcmdhd drvier of cubie is just making the modules work. Merging to linux-sunxi still need lots of work :<
<apo>
:D
<apo>
benn-on-cubietru: Can I just cheat and set WEXT_PRIV manually?
<apo>
(Will that get overwritten the next time I use menuconfig?)
<benn-on-cubietru>
apo: yes
<benn-on-cubietru>
apo: but I don't think it's a good way, better way is just select some more wireless driver. And the user may want them also actually
<apo>
I'm the user
<apo>
I like small kernels :P
<benn-on-cubietru>
apo: :D
<apo>
my raspberry pi kernel fits on a floppy
Black_Horseman has quit [Remote host closed the connection]
<apo>
Iiii... guess I'll add one driver. Bah. :P
<benn-on-cubietru>
that's cracy.I haven't seen floppy for many years!
<apo>
I haven't either
<Turl>
benn-on-cubietru: if we make a patch to run mbus on 400mhz can you test on many cubieboards to make sure it's stable? :)
<apo>
but if I find one, I know what to put on it.
<apo>
... if I also find a drive
<Turl>
floppy drives make good instruments
<benn-on-cubietru>
Turl: yes, but how many?
<Turl>
benn-on-cubietru: I don't know, but you can certainly try on more than us :P
<benn-on-cubietru>
memtester ?
<Turl>
yes, but also general system test
<benn-on-cubietru>
Turl: run memtester and glmarks the same time
<Turl>
yes, that would be great
<Turl>
also test mmc, I think ssvb said it became unstable, but I think it was higher mbus
<Turl>
booting full ubuntu from card should be enough I think :)
<benn-on-cubietru>
Turl: nice! so that I don't need to flash so many boards
<benn-on-cubietru>
Turl: with card image, I can test more than 100 pieces
<apo>
Turl: They do
<Turl>
benn-on-cubietru: :D
<benn-on-cubietru>
If the test is simple enough, we can test more than 500 pieces
<Turl>
benn-on-cubietru: :)
<Turl>
benn-on-cubietru: I will let you know then when we have patch ready to test
<benn-on-cubietru>
Turl: ok
<Turl>
benn-on-cubietru: thanks :)
<Turl>
time to sleep, good night :)
<benn-on-cubietru>
Turl: good night !!:D
<benn-on-cubietru>
apo: I think CONFIG_SUSPEND should be enabled
<apo>
benn-on-cubietru: Why? My IRC box is on 24/7/365
<benn-on-cubietru>
I got mmc compiling error, I enable CONFIG_SUSPEND, and try again for the whole compiling
<apo>
huh
<benn-on-cubietru>
apo: the error refer to SUSPEND subsystem, i think
<apo>
that did not happen for me
<apo>
Did I even enable MMC...
<apo>
Nope! :D
<apo>
I should get into kernel dev't
<benn-on-cubietru>
apo: your config work now. I add CONFIG_SUSPEND, CONFIG_USB_ZD1201 with menuconfig. Do you want me send it back to you ?
<apo>
benn-on-cubietru: Not if that's all you did :) I added another broadcom module instead, and the MMC fix is easy, too
<apo>
thanks again for thehelp
<apo>
(Yes, I did check if it enables WEXT_PRIV)
<benn-on-cubietru>
apo: ok, BTW, the cubie branch kernel also require gpio_sunxi
arokux2 has quit [Remote host closed the connection]
enrico_ has joined #linux-sunxi
<focus>
buZz: having 8 cores or even 4 cores is useless if the DRAM cycle time is 10MHz
<focus>
most applications will not see improvement in performance
<focus>
allwinner engineers and arm world living in delusion
<focus>
the only things that see improvement and anything that can spend a lot of time computing with local cache
<focus>
not many applications then!
<focus>
most applications need to multi-task and that means constant change of context which means ddr cycles at around 10MHz
<focus>
its way more powerful to have 2 x dual cores each with its own ram and talking through ethernet - at least that means the total MHz rating goes up to 20MHz
<buZz>
focus: i like delusions
<focus>
try it with apache - a quad core v 2 x dual core linked by ethernet - one serving images, one serving web pages - it will service twice as many requests
<focus>
buZz: u sad tosser!
<buZz>
:D
<focus>
still there is hope - since A20 is so cheap, i plan to make 2 x dual core boards and 4 x dual core boards and parallel up other bits of hardware to up the through put rates and make it appear as a single computer with shared ram / or ethernet
<focus>
such server boards will outperform the best and still clock in at under $100 BOM costs
<focus>
you just get one ethernet connection and that is all that is needed for headless server - don't need to understand what is behind that ethernet socket!
<focus>
</rant>
<focus>
:D
<mnemoc>
focus: you might want to talk with tom about it, afaik he is interested in doing something similar and also oshw
notmart has joined #linux-sunxi
<mnemoc>
bbl
<hipboi>
focus, that's exactly what we are doing now
<focus>
mnemoc: excellent!
<focus>
hipboi: - need any help?
<hipboi>
focus, you can help to review our work :)
<buZz>
i would like to see EOMA68 getting more ground
<hipboi>
they provide the same but cheaper modules
<EGM>
Hi guys!
<arokux>
hi EGM !
<arokux>
we did get grip on the hackberry 10 USB problem, but your original problem was different, as said in the e-mail
<JohnDoe_71Rus>
How much interesting things
<EGM>
arokux: I just noticed that you asked me in mailing list to do bisect. But IIRC, I could not build 3.4.43-r1 because there's no support for A20 yet (sun7i_defconfig is missing)
<arokux>
EGM: no problem, can you identify some other working commit?
<arokux>
thanks hipboi
<arokux>
hipboi: do you know somebody willing to donate a basic A31 board, so I can add mainline and U-Boot USB support to it? :)
<EGM>
arokux: I believe that r2 was the commit that added A20 support, but there's already broken.
<arokux>
EGM: I see... that is bad. your hardware is cubieboard2?
<EGM>
arokux: Yes, Cubieboard2 with A20 and 1GB ram
<EGM>
I do not have the A10 based cubieboard.
<arokux>
EGM: I have cubietruck (it is also A20 based) now, so I can try to reproduce your problem
<arokux>
EGM: it is wrong to apply it in general, but hackberry seems to have some strange (incorrect) PCB design
<rings_IIV>
Hi Guys, This is Robert from AW-SOM, to be clear, the sodimm modules are my design but the coreboards are from WITS. We offer them as a courtesy and I added there footprint to upverters for use.
<EGM>
arokux: Thanks for the info, but I won't be using Hackberry board anymore probably.
<arokux>
EGM: you do not have the first problem (usb devices won't go away) with Hackberry?
<arokux>
hi rings_IIV
<EGM>
arokux: We are using it for some video capturing, and single core does not have enough CPU power to encode that in real time, so I need A20 based board for that.
<rings_IIV>
Hello
<arokux>
rings_IIV: can you confirm you'll be releasing some A31+base board?
<EGM>
arokux: Yes, I believe that 3.4.43-r1 is working as expected on Hackberry.
<EGM>
I think that I wrote that to mailing list.
<rings_IIV>
If there is enough people I would finish the design
<arokux>
EGM: ok.. so this is cubieboard2 (or maybe A20) related. good to know. I'll continue investigation.
<JohnDoe_71Rus>
rings_IIV: For what can applicable sodimm modules format? sodimm mostly for memory
<arokux>
rings_IIV: I'm asking just because you seem to have lots of links to the resources created by sunxi.org. it would be nice if you can donate some hardware to sunxi.org :) for example I'd need an A31-board to work on USB support. I've got other boards donated from other manufacturers.
<rings_IIV>
Definitely, This was discussed in the past
<arokux>
JohnDoe_71Rus: the SODIMM connector is just reused, so they do not need to invent a new connector and can buy ones that are ready available.
<rings_IIV>
Yes, my preferred connector brand is by Tyco
<arokux>
rings_IIV: the other point is the documentation of the chips. we have the source code of the kernel, but this is sometimes not enough. since you collaborate with WITS (which is close to AW) maybe you can ask them for some docs that sunxi.org needs to improve its kernels?
<Turl>
oliv3r: yes, dcdc3 I think? which turned out to be vddr
<arokux>
jemk: how design ware fit together with STMicroelectronics? STMicroelectronics have used the IP too?
<Turl>
oliv3r: I was thinking of something like #ifdef 400M_MBUS #define VOLT, FREQ... #else ... #endif
<jemk>
arokux: yes, stm used it and wrote a linux driver for it
<Turl>
then use 400M_MBUS on boards we get confirmation that it's ok
<oliv3r>
Turl: btw, delete from works just fine, but i can't drop it; if i use sqlite3 cli, i can drop fine, while the webinterface gets polled, so for some reason i'm not allowed to drop a table
<jemk>
arokux: and allwinner copied some functions from their source and changed many things
<jemk>
but many things are copied 1:1...
<arokux>
jemk: yes, this is their style.
<Turl>
oliv3r: what about delete'ing and using 'create table if not exists'? :)
<oliv3r>
i dunno if i wanna introduce such a huge codechange :p
<oliv3r>
well actually that might work
<oliv3r>
i'm gonna try delete *; drop table;
<oliv3r>
i first thought that maybe the webserver kept polling that specific table; but doesn't seem to be true either
rellla3 has joined #linux-sunxi
<oliv3r>
Turl: yeah but then there's boards where it's okay and stable, but due to power usage, we want to have it lower anyway
<oliv3r>
Turl: personally I feel strongly on haveing it on 400MHz on stable boards, powerusage .. pff who cares :)
<Turl>
oliv3r: :)
<oliv3r>
but fuck, even delete; drop; doesn't work :S
<arokux>
jemk: what is the current state of ethernet for cubietruck? does it work with sunxi-3.4 at all?
<arokux>
oliv3r: ^
kz1_ has joined #linux-sunxi
<oliv3r>
erm, same as 3 weeks ago? ;p
<oliv3r>
i know someone was interested in starting on mainline work for gmac
<oliv3r>
wens i think
<jemk>
arokux: it works with benns version from github.com/cubieboard
<arokux>
oliv3r: I just do not know the status from 3 weeks ago :)
Guest83880 has quit [Ping timeout: 272 seconds]
<arokux>
jemk: is somebody working to add the support to sunxi-3.4?
<oliv3r>
arokux: gmac only works with lichee kernel
<arokux>
oliv3r: will EMAC still work on cubietruck?
<oliv3r>
i think that's the lichee-3.4 version copied to cubieboard-3.4; i don't think anybody moved it to sunxi-3.4
<oliv3r>
arokux: yes, but the RGMII PHY on the board doesn't like to talk to the EMAC which doesn't talk RMII, only MII
<oliv3r>
the PHY only talks RGMII and RMII
kz1 has quit [Ping timeout: 245 seconds]
kz1_ is now known as kz1
JohnDoe_71Rus has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org]
<arokux>
oliv3r: so what it means in the end result? no ethernet on cubietruck with sunxi-3.4?
<oliv3r>
rgr
<oliv3r>
only if you use 3.3 or 3.4 lichee kernels
<oliv3r>
for now
<oliv3r>
but gmac from liche-3.4 should be just a copy/paste away
<oliv3r>
libv: i guess I understand what you mean, maintaining a single page with all possible drivers would be huge; but i'm worried about the maintanance things, who will maintain all those boards etc
<oliv3r>
and the 3.4 kernel is quite limited on supported drivers, since only allwinerized drivers would owrk
<libv>
oliv3r: if i had gotten a new board, the first thing i would want to see is a page which lists all i need to know
<libv>
i do not wish to trawl through a possibly endless list of obscure pages
<libv>
also, what other boards will use this chip?
<libv>
if there are other boards, then create an rtl8211 specific page, and link to that
<oliv3r>
libv: some nice templates work wonders for that :D
<oliv3r>
libv: i like how linuxtv somehwat does it
<libv>
oliv3r: i think that you are the only one who works templates
<oliv3r>
heh, i wish i could, i know nothing of them!
<oliv3r>
but I see how they can be very usefull
<oliv3r>
if I knew more i probably would make some
rellla3 is now known as rellla
<oliv3r>
Turl: ok every function that access the database to be dropped is for sure 'finalized()';
arete74 has quit [Ping timeout: 246 seconds]
<slapin>
ah, cool, he just pulled my changes
* slapin
is proud of contribution
<slapin>
I think this meta is worth adding to wiki
<slapin>
oliv3r: ^
Black_Horseman has joined #linux-sunxi
<oliv3r>
slapin: yeah probably too; if you could add it somewhere sensible, that'd be great
<slapin>
oliv3r: where you think it would be feasible?
<slapin>
oliv3r: I think some setup page could be nice too
<oliv3r>
well there's FirstSteps
* slapin
is long term OE fag
<slapin>
oliv3r: please show the place as my mind is somewhere now
<slapin>
enrico_: I think we should fine 'upstream' repo and propogate thet to koen's scripts so 'next' branch have up to date one
<slapin>
enrico_: so to habe 2 less steps
<slapin>
s/have/have
<slapin>
damn
<slapin>
s/habe/have
<enrico_>
from "metas" activity the good candidates for upstream are naguirre and vonfritz, i try to keep mine in a working and tested state but they are more active then me
<oliv3r>
i'm sure if we want a central linux-sunxi repo, that can be arranged and have that behave as 'upstream'
<oliv3r>
for allwinner anyway
<oliv3r>
mnemoc: ^
<enrico_>
now that you say it, it would be better to have a fork in linux-sunxi and setup the maintainers there
<slapin>
enrico_: a good thing about naguirre's repo is that everything is out of the box, just add it to scripts, configure, build, just as per OE docs, no extra steps
<enrico_>
yes, i tested it and recently vonfritz one too, you even get a working sd card image that works perfectly
<enrico_>
i don't know who manages linux-sunxi github, but i'd create a fork there and someone should maintain it (merging pull requests etc...), and it becomes the official meta
<slapin>
mnemoc: ^
<enrico_>
about the wiki page: i'd just remove method 1, the second one looks good to me
<oliv3r>
enrico_: ok, go for it :)
<libv>
naguirre should have his repo at the sunxi github
<enrico_>
for now leave the naguirre repo, then it will be changed with the official one
<apo>
hm, sync's dead, waiting for IO...
Skaag has joined #linux-sunxi
wingrime has joined #linux-sunxi
apo has quit [Read error: Connection reset by peer]
<mnemoc>
if the person maintaining it wants to move the repo to the orga, I can gladly add him to the team
<mnemoc>
and there are quite a bunch of admins who can do it too ;-)
<mnemoc>
i don't have the monopoly
arete74 has joined #linux-sunxi
<mnemoc>
it's better that they move the repo and then fork for development than doing it the other way
<slapin>
mnemoc: could you write to github.com/naguirre so he knows the opportunity?
<mnemoc>
i think i've talked with him before in... #olimex?.... and he wasn't interested
<slapin>
mnemoc: or I could become formal lazy maintainer merging pull requests for the official repo if he won't agree
<mnemoc>
it makes more sense to have the "stable" in the orga, and dev trees for wip development
<slapin>
mnemoc: it is just the matter of tags and branches
<mnemoc>
to keep the users from getting an experimental branch by... accident
<slapin>
mnemoc: I can setup CI stuff for testing
<mnemoc>
ask him if he want to move it to the orga, then I add him to an oe team so he can move and fork
<slapin>
mnemoc: OE IS development stuff, so people are prepared
<slapin>
mnemoc: who me?
<mnemoc>
sure
* slapin
scratches his head
<mnemoc>
if he doesn't want to maintain a linux-sunxi/meta-sunxi you would ;-)
<mnemoc>
so, you contact him
<slapin>
mnemoc: you make lazy person do a thing, do you know consequences?
<mnemoc>
=)
ganbold_ has joined #linux-sunxi
ganbold__ has quit [Remote host closed the connection]
<slapin>
mnemoc: written that mail, will wait for answer
deasy has joined #linux-sunxi
<mnemoc>
thanks!
Sonicadvance1 has quit [Remote host closed the connection]
<mnemoc>
arokux: all my toys are still on a box. but i hope to resume sunxi programming this weekend on a CT
<arokux>
mnemoc: no cb2?
<mnemoc>
nope. my former-former employer stole it
<arokux>
mnemoc: ?! what kind of employers you have....
<mnemoc>
i have 2 cb1, 2 ct, 1 bdd kit, 1 itead kit, and a10s, a20 and a13 olinuxinos
<mnemoc>
and one radxa rock
<mnemoc>
and an old mele a1000
captainigloo has joined #linux-sunxi
<mnemoc>
naguirre!
<captainigloo>
hi
<captainigloo>
yes it's me :)
<mnemoc>
:)
<mnemoc>
so you'll move your meta-allwinner into linux-sunxi/meta-sunxi and join the community? ;-)
<captainigloo>
yes i do :)
<mnemoc>
\o/
<mnemoc>
1m, let me add you to the team
<captainigloo>
hehe thanks :)
eebrah has quit [Read error: Connection reset by peer]
<mnemoc>
captainigloo: permissions granted, now you can move it
<mnemoc>
:)
<captainigloo>
ok !
<captainigloo>
thanks a lot
<mnemoc>
yw ;-)
<mnemoc>
slapin: thanks you too ;-)
<captainigloo>
slapin: it was in my todo for a while
<captainigloo>
;)
<mnemoc>
captainigloo: please ping me after moving it, to grant control over the new repo to the OE team
<mnemoc>
so you can keep commiting and admining it
geecko has joined #linux-sunxi
<captainigloo>
i plan to do that tonight
eebrah has joined #linux-sunxi
<captainigloo>
btw what's the best way for the move ? i push to the the repo and remove mine ?
eebrah is now known as Guest8720
<mnemoc>
captainigloo: no, tell github to move it
<mnemoc>
yours will get a redirect
<captainigloo>
ok perfect
<mnemoc>
then you fork the new one
<mnemoc>
for your developement
<mnemoc>
so naguirre/meta-allwinner is internally redirected to linux-sunxi/meta-sunxi, and naguirre/meta-sunxi a development fork for linux-sunxi/meta-sunxi
<captainigloo>
and i would like tochange the name for meta-sunxi ?
<captainigloo>
yep ok perfect
<mnemoc>
it's better to avoid trademark issues
<mnemoc>
and allow you to keep both names
ykchavan has quit [Ping timeout: 264 seconds]
<arokux>
what is this meta good for? explain to a noob please :)
<mnemoc>
arokux: openembedded overlays
<captainigloo>
arokux: it's a onverlay for openembedded/yocto
<arokux>
this sounds even more complicated
<mnemoc>
captainigloo: btw, we can also host bin packages if you like
<arokux>
what can I do with it? :)
<captainigloo>
you can byild your own distribution
<mnemoc>
custom-linux-distros in a corporate friendly way :p
<captainigloo>
mnemoc: what for ?
<arokux>
what is the difference if I go for the debian/ubuntu/arch/you-name-it rootfs?
<mnemoc>
captainigloo: allow people to just install/update kernel packages maintained by you
<captainigloo>
mnemoc: hum ok
<mnemoc>
captainigloo: like apt or yum repos
<mnemoc>
captainigloo: but only if you want
<captainigloo>
mnemoc: hum yes i see
<captainigloo>
in fact i also manage a distribution
<captainigloo>
and we plan to have binary packages (ipk)
<captainigloo>
but we allready have a server for that
<captainigloo>
i'm using cubieboard and mele a2000 for home automation server and interface
<captainigloo>
it works pretty well
<captainigloo>
and since we got opengl working, the touchscreen interface is just perfect ;)
<captainigloo>
it's a video running calaos on the mele a2000
<mnemoc>
captainigloo: nice :)
<arokux>
Unfortunately, this video is not available in Germany because it may contain music for which GEMA has not granted the respective music rights.
Black_Horseman has quit [Quit: Αποχώρησε]
<captainigloo>
arggg my girlfriend was listening to Adele at the same time ...
<mnemoc>
<3 GEMA <3
<captainigloo>
youtube filters are really good !
<arokux>
captainigloo: maybe you upload it somewhere else?
<slapin>
elotouch is cool thing, but it is soo huuuge...
bsdfox\ has joined #linux-sunxi
<captainigloo>
yep it's for integration into walls
<slapin>
captainigloo: cool thing for teaching environment, might buy a few to torture students in more hi-tech way
<mnemoc>
poor students having to use EFL...
<mnemoc>
(/me loves efl btw)
<slapin>
well, my students are not poor, mine are unfortunate ones, and they still use chalk and black board, or marker and whiteboard...
<arokux>
so what is the difference if I go for the debian/ubuntu/arch/you-name-it rootfs? (instead of oe/yocto overlays)
<slapin>
arokux: you feel closer to corporations
<captainigloo>
arokux: depends on what you want to do
<mnemoc>
arokux: on one you adopt an existing enviorment, with oe (or similars) you decide each component
<captainigloo>
id you want to build your own distro, openembedded is good
<mnemoc>
until they start imposing systemd on you :\
<slapin>
arokux: actually this is something you make your custom firmware in VERY flexible, hard way, you will feel elite and slightly sad
<slapin>
arokux: hint: developer stuff, hardcore and dangerous, and you can feel frustration at times
<slapin>
arokux: but unique and unreplacible if you know what you're doing
<slapin>
mnemoc: Aah! lets start systemd vs whatever flame!
eebrah_ has quit [Read error: Connection reset by peer]
<slapin>
actually I already submitted to systemd in angstrom, it works as you learn to prepare it, but I can't live with it on desktop
<captainigloo>
we are using systemd with calaos
<captainigloo>
it's just perfect
<captainigloo>
i love it :)
<slapin>
actually, systemd is not evil, it is just buggy overfeatured undermaintained can of warms
<slapin>
*worms
<captainigloo>
:P
<slapin>
we here have some specific tasks here and had to maintain our own fork of systemd to suit our needs as the politics of the project is too ambitious for our taste
* arokux
communicates with systemd devs to fix a very long hang on reboot/shutdown
<slapin>
arokux: they refused to make logging modular, this is our primary issue
<slapin>
arokux: we had to do this
<slapin>
arokux: damn journald
<arokux>
slapin: what do you teach?
eebrah_ has joined #linux-sunxi
<slapin>
arokux: general workflow with embedded Linux using OE and qemu
<slapin>
arokux: a prepare a few girls as educated testers
<arokux>
slapin: so it is not university course?
<slapin>
arokux: no, it is site-based local excuse for low salary for some employees
<slapin>
arokux: you pick up students, teach them and then force them work for food
<arokux>
slapin: where can one enroll? :)
<captainigloo>
mnemoc: ok i'm transferring the repo
<captainigloo>
it ask me for a new owner
<captainigloo>
it's linux-sunxi (the organisation) ?
apo_ has joined #linux-sunxi
<slapin>
arokux: well, if you want to work with traffic tracking systems (with glonass, etc.) till you die of exhaust, then please come visit Saint Petersburg, Russia, call me and we can talk, but the salary will be low even for Russian standards
<arokux>
depends on traffic :D
<slapin>
arokux: public busses, tramways, trolleybuses, various other vehicles
apo_ has quit [Ping timeout: 260 seconds]
apo has joined #linux-sunxi
<slapin>
captainigloo: linux-sunxi
<mnemoc>
captainigloo: yes, linux-sunxi orga
<Tsvetan>
slapin why IT salary is low in Russia? here IT sector paychecks are x3 than in any other sector, maybe just banking and finance take more
<captainigloo>
it seems i can't
<mnemoc>
captainigloo: you are team member. you can.
<slapin>
arokux: the good side - you don't pay for education, the bad side - you get salary of 7000-12000 rubles without chance of increasing it, and agree to work for 1 year after graduation
<captainigloo>
mnemoc: naguirre for the username ?
<mnemoc>
captainigloo: yes. and you should have received a notification email from github
<captainigloo>
no i don't :(
<mnemoc>
check spam :p
<mnemoc>
doh. forgot to save the team :\
<mnemoc>
captainigloo: try again
<arokux>
slapin: this is not bad. imagine you get nothing in some lousy university and then ask yourself how to use your "knowledge" after graduation.
<slapin>
Tsvetan: it is not generally low in Russia - it is greedy organizations exploiting untrained cheap labour
<captainigloo>
i didnt't get any email from github in the last 2 hours
<slapin>
arokux: the minimal prace here with flat rent is 25K roubles
<arokux>
slapin: I know, I come from a neighbor country
<captainigloo>
ah
<captainigloo>
i try again :)
<slapin>
arokux: so you can only live with parents' support or you should be local punk with own flat (or free rent :))
<Tsvetan>
slapin this means the demand for IT graduates is low, here the companies fight and steal employees :)
<Tsvetan>
one company even made private own ´university´ to educate free of charge people who want to work in IT industry later
<arokux>
they are not yet IT graduates
<slapin>
arokux: and knowing the prices, I think you can't be either if toy agree for such "job"
<arokux>
slapin: but how is that different if you go to university? you still have to pay 25k rent and get nothing.
<arokux>
with IT specialty after your education/work the possibilities are much higher then after university
<slapin>
Tsvetan: a problem here is that there is no real industry, only corps supporting militarys and various other government orgs, business wants candidates with experience, so usual student is unable to get any normal salary, so he gets to whatever job he can to gain experience
<slapin>
arokux: in Russia in IT education means nothing, but experience means everything
<Tsvetan>
so education is not good like here
<arokux>
slapin: so you see it is better to go to your university
<arokux>
Tsvetan: it is good, but not practically oriented.
<slapin>
arokux: I hate myself breedeng hordes of slaves
<Tsvetan>
this is what I mean - obsolete programs, probably still learn Fortran77
<slapin>
arokux: girls intended to know what embedded linux is fail to find SysRq on keyboard, so the level.
<arokux>
what could they do otherwise? they are happy to have you and will get lots of chances!
<slapin>
arokux: not obsolete, they know MS Excel quite well, but they don't know what Fortran is. Optimized education, that is
<arokux>
Tsvetan: education is strong in math, algos, but not up-to-date technologies. maybe not fortran77, but pascal :)
<slapin>
arokux: not strong in anything anymore :(
<slapin>
arokux: thay fail at probablity theory even
<arokux>
you are too pessimistic, in europe computer science students are much much weaker in math/algos, but have access to technology
<mnemoc>
captainigloo: so?
<captainigloo>
mnemoc: i'm waiting for the email
<slapin>
arokux: well, math/algos work best when actually used in applications, but they don't really associate it with anything they see or can touch
<mnemoc>
captainigloo: just try to move it
<slapin>
arokux: they know some formulas, but don't know the global picture, so they don't know what to do with them
<captainigloo>
ah ok, i misread
<captainigloo>
ok it's moved to linux-sunxi/meta-allwinner
<slapin>
arokux: they are not educated to think creatibely, like engineer, they don't have any creative freedom in the head, and I have to teach them to think independently and have their own opinion and be dependable at problem resolution
<captainigloo>
there was no way to change the name
<mnemoc>
captainigloo: into your own, for development
<mnemoc>
captainigloo: you are admin of the project, and obviusly can commit too
<captainigloo>
awesome
<arokux>
slapin: how old are they, are they after the university already?
<mnemoc>
captainigloo: old checkouts will be redirected internally by github
<slapin>
Tsvetan: the IT education was limited to what general consumer have to cope with every day, they have some abstract education about algoruthm theory, but they just forget everything as get out of the building, and nobody cares really to test them
* mnemoc
looks at his algo. books in the shelf and wonders if he remembers anything from there
<slapin>
arokux: ~24 y.o.
<arokux>
slapin: I think what you describe is normal :)
<slapin>
arokux: it is too good I have my damn trackers to forget all about this at times
<slapin>
mnemoc: all you have to remember is method of self-education, then you can learn new things and do your job. But these people fail at this and I have to fix this up. about 1/3 just leave the second day, frustrated. As I'm not going to use magic, that is.
<slapin>
strangely, girls are more stubborn at this, like they feel humiliated by the situation, and have great success rates, they really fast learn technically complicated things, probably just to prove something, but lose interest soon, prefering simpler work as rootfs/system testers.
<slapin>
aha
<slapin>
so we need to update README telling it is OFFICIAL now, and then fix wiki page, again :)
andreasw has joined #linux-sunxi
<enrico_>
uhm i don't know if it can be called official, to me official means supported by allwinner
<techn__>
captainigloo: thanks! :)
<techn__>
Does OE have support for pre-build repository yet?
<captainigloo>
hey, np :)
<captainigloo>
whatt you mean by pre-build repository ?
<techn__>
so that I wouldn't have to build everything again everytime
<mnemoc>
enrico_: but now it's meta-sunxi, not meta-allwinner. and it's officially supported by the sunxi.org community now ;-)
rz2k has joined #linux-sunxi
<enrico_>
mnemoc: right :D
<captainigloo>
ah it's up to the distro to do such things and provides binary packages
<captainigloo>
IMO
<slapin>
techn__: if you build first, then you can just build what you need in additive manner, as long as you don't update metadata
<slapin>
techn__: there are also image builders from binary packages
<slapin>
techn__: whateber you prefer
<techn__>
yeah.. so situation is same as 2 years ago
<mnemoc>
enrico_: should I add you to the OE team too?
<mnemoc>
captainigloo: ---^
<captainigloo>
i'm not against :)
<enrico_>
sure, thanks
<mnemoc>
ebutera?
<enrico_>
yes
<mnemoc>
done
<slapin>
techn__: you can build repository of packages in install in debianish way with opkg, so have homebrew binary distro.
<mnemoc>
i offered we can host an ipkg/opkg repo in packages.linux-sunxi.org....
<slapin>
techn__: I just don't understand your goal, as this is not primary OE target
<techn__>
slapin: but still I have to build it from beginning..
<techn__>
I need to build even toolchain to build it ?
<techn__>
and for every board?
<slapin>
techn__: you can as somebody to build it for you :) then it will look as it is really binary distro
Fusing has joined #linux-sunxi
<slapin>
techn__: you better read docs as here YMMV and lots of things can be shortcut
<enrico_>
wiki updated
<slapin>
techn__: angstrom, for example, provide binary packages
<slapin>
techn__: but you might not like some options
EGM has quit [Ping timeout: 264 seconds]
FR^2 has quit [Quit: Connection reset by peer]
<slapin>
enrico_: you did half-assed job :), as you didn't fix commands, and better leave development repositories links somwhere at bottom of the page
<captainigloo>
techn__: not for every board, as most of them have the same arch, so it's build once per arch
<enrico_>
slapin: ops, more than half :D i'm fixing it
<arokux2>
wingrime: you can try what ever kernel you have after sunxi-3.4.43 I think
<hramrach>
arokux2: for me devices disconnect normally on CT on the EHCI port (1 or 2, not 0)
<arokux2>
hramrach: ok. this was on CB2
<hramrach>
I don't use port 0 dues to wiring problems and it's device mode by default
<hramrach>
that should not make any difference and I did not notice any issue there
<hramrach>
but can't test easily atm
<wingrime>
mnemoc: thats for apply
<wingrime>
mnemoc: thanks
<arokux2>
hramrach: it would be very kind of you! please reply with your findings to that thread
<wingrime>
mnemoc: I simply afraid send big patches
<hramrach>
arokux2: actually, I use only USB1 devices, probably
<hramrach>
that might make more of a difference
<arokux2>
hramrach: what do you mean by USB1?
<hramrach>
full-speed, not high-speed
<hramrach>
but I used the USB Etherenet thing and that should be high-speed and it disconnects all right
<arokux2>
that guy says your image (which has some other kernel) works, but ours sunxi-3.4 won't disconnect the devices. are you testing with ours sunxi-3.4 now?
<arokux2>
hramrach: also, maybe you haven't just payed attention if it disconnects or not? :)
<hramrach>
actually, it's broken. the usb ethernet does not disconnect
<hramrach>
it got reset on reconnect all right, though
<arokux2>
hramrach: so you are able to reproduce it?
<arokux2>
hramrach: "it" the issue reported by EGM
<hramrach>
yes, only with high-speed/usb2/ehci device
<hramrach>
a serial port goes away on disconnect which is presumably because it's full-speed/ubb1.1/ohci
<hramrach>
yes, it's dangling off one of the ohci buses
<arokux2>
hramrach: what?
<arokux2>
hramrach: "dangling off one of the ohci buses" ?
<hramrach>
when you see the lsusb output - thete is bus number for each device
lump has joined #linux-sunxi
<hramrach>
the 1.1 and 2 hubs have each its bus number and so does every device
<hramrach>
or you can look at this in some nice graphical device manager or /sys
<deasy>
:)
<arokux2>
hramrach: so you just said that OHCI device is disconnecting fine == "dangles off"? :)
<hramrach>
arokux2: no, I mean that the the USB to serial converter is a loose exterior part of the OHCI bus - as you could see in something like the kde device manager
hramrach has quit [Remote host closed the connection]
<popolon>
I will try to register to it then
<arokux2>
popolon: you do not need to register
<popolon>
oh no, google groups ?
<arokux2>
popolon: ?
<popolon>
I mean subscribe/add to a newsreader...
<arokux2>
popolon: this is just our mailing list in a nice to read form
<deasy>
back :)
hramrach has joined #linux-sunxi
<popolon>
ah ok
<popolon>
comp.hardware.netbook.arm.sunxi <= this remember me newsgroup hierarchies
<arokux2>
popolon: please try to reproduce the issue once you have time
<deasy>
hramrach, hmm dunno if i have this issue
<hramrach>
that bug could have been there for ages without anybody noticing. I mean, I don't even have usbutils on cb1, and there is not problem with devices breaking because of that
<deasy>
it's cubian ?
<hramrach>
what's cubian?
<deasy>
debian for cubieboard
<popolon>
I've a little time now, I can try, but where is it's kernel ?
<hramrach>
cubian is a debian image for cubieboard created by somebody
<popolon>
ah he uses cubian
<deasy>
no i have not say it
<deasy>
it was a question
<deasy>
as i use cubian
<hramrach>
I run debian on cubieboard. Probably since before there was any cubian :p
<deasy>
lot of change in 3.12, in a good way for desktop :)
<popolon>
with the 3.4 kernel I didn't seen this kind of problem
<popolon>
but I don't have usb wifi key
<arokux2>
popolon: you should try with the latest sunxi-3.4 kernel
<popolon>
but used lot of usb devices with 0 problems
<popolon>
from nighty ?
<hramrach>
popolon: it should manifest with any high-speed device
<popolon>
nighty build ?
<hramrach>
whatever you have
<hramrach>
if it does not that's good data point too
<arokux2>
popolon: there could be no noticeable issues. you would need to issue lsusb after plugging off the device
<hramrach>
eg, thumb drive or usb ethernet should work for testing
<popolon>
ok
<arokux2>
popolon: just issue lsusb before plugging and after unplugging
<popolon>
do you think wacom tablet would be a good test ?
<popolon>
ah ok
<hramrach>
depends on the tablet
<popolon>
but I never had problem with lususb
<popolon>
lsusb and usb devices
<deasy>
i try with the last cubian
<popolon>
I pluged/unpluged a lot
<hramrach>
it's not a problem per se
<popolon>
tablet, keyboard, mouse
techn_ has joined #linux-sunxi
<hramrach>
it's just untidy
<hramrach>
the high-speed devices don't go away when unplugged
<popolon>
will high speed
<popolon>
(meaning usb2.0
<hramrach>
yes, usb2
<popolon>
I've card reader perhaps, that's sur is usb2
<hramrach>
the tablet might too
<hramrach>
you can see with lsusb
<popolon>
no problem with lsusb so far
<hramrach>
note the bus number next to device and next to a root hub
<popolon>
I use it a lot
<popolon>
but will retry
<popolon>
ok
<deasy>
time for dl the last cubian and write it on my sd
<hramrach>
popolon: also hte issue only happens with devices sonnected directly to port 1 or port 2 of cubieboard.
<popolon>
I use both
<hramrach>
you cannot connect keyboard, mouse and tablet to those.
<libv>
i stated 60% over the performance of the X11 binary, due to fire-and-forget job interleaving and proper integration with the display engine
<libv>
i seem to be at 6% (or more) ahead atm, and proper job interleaving adds about 50% in performance, getting to just under 60
shineworld has quit [Quit: Leaving]
jinzo has quit [Quit: Leaving]
jinzo has joined #linux-sunxi
jinzo has joined #linux-sunxi
jinzo has quit [Changing host]
<Turl>
deasy: yeah, all supposedly have trustzone
<Turl>
I've never seen anyone using it though
<Turl>
and trustzone won't accelerate your crypto operations, I dunno why do people always think that
<deasy>
Turl, thank you, yes a people ask me it
<arokux2>
Turl: it won't accelerate nothing at all?
<pfdm>
Turl: Maybe because of this on the wiki : Allwinner SoC include a TrustZone cryptographic (crypto) engine co-processor that optimizes cryptographic operations with a 100x factor or even more comparing to a plain software implementation run on the CPU.
<Turl>
arokux2: trustzone it's a thing to run a 'second OS' on a secure world
<arokux2>
Turl: I see
<deasy>
"TrustZone cryptographic engine and security accelerator co-processor that supports both decryption and encryption of AES, DES, and 3DES, as well as SHA-1, "
<Turl>
arokux2: the security system (SS) should be able to do hardware crypto though
<Turl>
deasy: I think the person who wrote that wiki page just mixed up SS and trustzone
<eagles0513875>
hey Turl
<Turl>
hi eagles0513875
<eagles0513875>
hows it going Turl
<deasy>
Turl, thank for informations :)
<arokux2>
Turl: isn't that page just a dump from aw's user manual?
<arokux2>
"security system" should be interesting
<Turl>
would you rather see the page deleted or sanitized?
<Turl>
if I clean it of inaccuracies it's not going to be any sunxi-relevant really
<arokux2>
Turl: I'm not sure the effort it worth it. somebody who will write the drivers will clean it