kivutar has quit [Read error: Connection reset by peer]
enrico_ has joined #linux-sunxi
juanfont has left #linux-sunxi ["Saliendo"]
juanfont has joined #linux-sunxi
code-ninja has joined #linux-sunxi
<code-ninja>
hi, I just dual booted an A20 based Android tablet. But my touch doesn't work. I used Ubuntu 12.04 as my root fs. My tablet does boot and I have a desktop environment. I can even chroot into my SD card on my host system. Just the damn touch doesn't work. Any help?
<binaryferret>
First, if you can boot into Android and find out what touch screen driver it's running.
<binaryferret>
As you'll need this info to find out which driver to build with the kernel.
<binaryferret>
Hopefully it'll be part of the linux-sunxi tree, otherwise some hunting may be required.
popolon has joined #linux-sunxi
<binaryferret>
I don't know if this will help, or if it's the correct way to do it - but on the tablets I've used, the android would have a number of different touch screen modules loaded, so I had to adb shell in and start disabling the touch screen modules until the touchscreen stopped functioning in android.
Faisal has joined #linux-sunxi
<code-ninja>
hmm, tried that. Android uses the ft5x_ts.ko driver. the sunxi image has it already but it doesn't work. I tried gt8x too because it is supposed to work everywhere, no luck. Ok... can you tell me how to compile this: https://gitorious.org/gslx680-for-sunxi
<code-ninja>
by any chance?
<binaryferret>
what did dmesg say when you loaded the drivers? as that can help give an indication of the issue.
<binaryferret>
I've had an issue with an a13 where I loaded the module for the touchscreen and it still wasn't working. However, i checked dmesg and noticed that it had been assigned an input it's just the xorg conf wasn't picking it up. A little bit of tweaking and I got it working.
<code-ninja>
1 sec... I'll just dmesg and be back
<binaryferret>
I'm still fairly new to this level of linux myself, so I'm afraid it might be worth waiting for someone who's actually a lot more experianced to help. For gslx680 it provides a README file that all you need to do is pretty much follow.
<binaryferret>
generally modprobe <driver name> then dmesg will often let you know where it's falling over.
rz2k has joined #linux-sunxi
rellla2 has joined #linux-sunxi
kuldeepdhaka has quit [Quit: Leaving]
rellla has quit [Ping timeout: 252 seconds]
y0g1 has quit [Ping timeout: 272 seconds]
printallthething has joined #linux-sunxi
y0g1 has joined #linux-sunxi
bsdfox has quit [Ping timeout: 245 seconds]
y0g1 has quit [Ping timeout: 272 seconds]
bbrezillon has quit [Ping timeout: 245 seconds]
bsdfox has joined #linux-sunxi
y0g1 has joined #linux-sunxi
bsdfox has quit [Read error: Operation timed out]
bbrezillon has joined #linux-sunxi
bsdfox has joined #linux-sunxi
Seppoz has quit [Remote host closed the connection]
Seppoz has joined #linux-sunxi
sspiff has quit [Ping timeout: 245 seconds]
sspiff has joined #linux-sunxi
mkutsevol has left #linux-sunxi [#linux-sunxi]
<libv>
Turl: i have unblocked our intelligent friend
* libv
just went through recentchanges, for the first time in a week
<Turl>
libv: :)
<Turl>
libv: mediawiki has a timed ban mode btwq
<Turl>
btw*
<libv>
yes, but i had set it to indefinite back then
<libv>
k70e is your device, right?
<Turl>
libv: it's not
<Turl>
it's this dude's
<libv>
ah, right, sorry, mixing things up
<Turl>
I just moved the content to its own page :)
<Turl>
no worries
nove has joined #linux-sunxi
<nove>
"//A10's GPU has a bug, we can avoid it" at the openmax source transform_color_format.c
<libv>
gpu?
<nove>
what is interesting is that they are making the color transform in software
<nove>
they are converting the frames from the vpu into a gpu buffer??
<libv>
ah, ok, this is a new sdk and it hasn't been properly gitted yet
<libv>
mali has no such limitations
JohnDoe_71Rus has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org]
<nove>
doing the convertion in software could mean, that the vpu can't output other than that tile32x32
<nove>
or allwinner is also limited by what the binaries blobs allow
<libv>
i somehow feel they are using the word gpu wrongly...
<libv>
anyway, whatever you cedrus guys do, aim for dma-buf/dma-fence
<nove>
another funny thing, h264 encoding is hard coded to only baseline profile, are the other profile fake?
* libv
is returning to code, and starts off with facing the music for the second bit of apparently absolute bullshit he told at FOSDEM: dual monitor support
<Turl>
libv: Tsvetan was surprised the other day wrt that
<libv>
i couldn't find a way to switch hdmi or lcd0 to another lcdc
<libv>
yeah, there is a big box of olimex kit en route
<libv>
anyway, poking at script.bin right now
<wens>
Turl: you think we should do a patch to check for missing clock-output-names ?
<oliv3r>
libv: hdmi, is that forced on lcdc0 at all times? or can we do hdmi via lcdc1?
<libv>
oliv3r: i don't know yet/anymore
<Turl>
wens: hm?
ganbold_ has joined #linux-sunxi
<libv>
hrm, the howto there is for a vga/hdmi device...
* libv
powers down the tablet
<libv>
it smells as if hdmi _can_ be switched...
<libv>
vga does seem to need to be at lcdc0, which confirms the lcdc0 h/vsync being tied to vga
<wens>
Turl: well maxime mentioned that I should check the return code for getting clock-output-names in my gmac clock code
<Turl>
libv: maybe if they had connected the ldc1 h/vsync pins it'd work there? :)
<Turl>
wens: link?
<libv>
Turl: that was my initial suggestion, but it seems hdmi might be more flexible than i first thought
<libv>
Turl: yes, the hdmi connector is on its own pins
code-ninja has quit [Quit: Leaving]
<libv>
Turl: the module comes with its own full timing information, which i thought a bit strange
<libv>
but wait and see what the register dumps state
<Turl>
wens: ehm
<Turl>
wens: if you don't have the name, you're doomed, whether you print an error or not
<oliv3r>
libv: the hsync-vsync thing is obvioulsy a board design choice; so not a limitation of the chip luckly
<libv>
oliv3r: as i explained in my talk, yes
<oliv3r>
yeah
<oliv3r>
re-itterating!
<Turl>
wens: but yeah, it makes sense on the overall picture
<oliv3r>
it's all confusing now :p
<Turl>
wens: keep in mind the other clocks do not necesarily have to have output-names
<libv>
hrm, wonky cursor... on my kms driver.
<Turl>
as it previously was using node->name and it technically would be ok if you had to use an old dt
<oliv3r>
libv: hardware cursors, you basicallly upload an image to the video card and it uses that until you upload a new image right? So what about animations? Really bad for performance? or can you actually store animations too?
<libv>
oliv3r: like any animation, you just reload
<oliv3r>
libv: so only room for 1 image basically, not a list of
<oliv3r>
so using animation would be bad for performance; ok
<libv>
oliv3r: still better than redrawing on the actual framebuffer
<oliv3r>
absolutly :)
<mripard>
Turl: we're going to break the old DTs so hard the clocks aren't really an issue...
<oliv3r>
mripard: why are you going to break the old DT's?
<wens>
Turl: well we made them mandatory so we might as well check for them
cubear has quit [Quit: Leaving]
<mripard>
oliv3r: yep
<wens>
oliv3r: mripard submitted a bunch of patches renaming the old sun4i compatibles
<mripard>
I wasn't going to at first, but the DT maintainers said it was ok
<mripard>
I'm not really fond of it, but at least, we won't have to maintain legacy compatibles...
<wens>
mripard: I don't get the "use clock-names" part, and I've been through other DTs, didn't see anyone doing it
<mripard>
wens: hmmm, are you sure you looked for it properly ? If have 662 lines matching the clock-names property here
<ccaione>
mripard: what do you think of tglx reply?
<oliv3r>
mripard: rename sun4i to sun4i-a10?
<ccaione>
I'm not sure he got the point
<wens>
mripard: clock-names used in clock node, not the whole DT
<mripard>
wens: clock-output-names are used in clock nodes
<mripard>
clock-names can be used in any node that has a clocks property
Black_Horseman has quit [Ping timeout: 248 seconds]
<mripard>
ccaione: I'm not even sure I get what he said :P
<Turl>
mripard: only on sun4i* :P
<mripard>
Turl: and all the SoCs that use some sun4i IP
<mripard>
like, timers, for example :)
<mripard>
which they all do
<wens>
and that breaks almost everything :/
<Turl>
haha ok then :)
<mripard>
wens: well, the point the DT maintainer had is that, since we do a lot of development, we would have updated our DT anyway
<ccaione>
mripard: basically AFAICT he is complaining that my patch is valid only in the threaded case, that is true, but he is missing the central point that is that having a threaded handler is just the consequence of having an external level-triggered device whose IRQ regs are accessible via i2c
<wens>
hmm, using clock-names seems to get a lot, when all we need is the global parent name, and it doesn't guard against someone putting the clock names in the wrong order
t3st3r has quit [Ping timeout: 240 seconds]
<wens>
but the code should look cleaner
<mripard>
ccaione: yes, but you would be affected in the exact same way if you had a !threaded handler
<ccaione>
i'm not sure about that
<ccaione>
because it is the PMIC that is driving the IRQ line
<Turl>
wens: clk-names is nice so you can do stuff like devm_clk_get(NULL, "bus") and get the bus clock
<wens>
Turl: looking at your reply on of_clk_init
<Turl>
whatever its position may be
<wens>
I think it's not wise to use clock-names in a clock driver?
<mripard>
wens: it's a bit more flexible, if someone doesn't set the clock and their names in the same order, it's pretty much an user error
<mripard>
however, in your driver, you can very reliably use different policy for the two clocks
<mripard>
like changing the rate for one, but not the other, etc.
<mripard>
without having to rely on the exact order passed
<mripard>
and all you need is passing the clock name to clk_get, instead of NULL
<Turl>
mripard: the point is that clk_get is overkill on a clock driver
t3st3r has joined #linux-sunxi
<Turl>
wens: the muxing is done by setting a bit field right?
<wens>
Turl: yeah, and the behavior is actually tied into the gmac driver as well
<Turl>
you can just dictate the parent index # must match the bitfield # or something like that
<wens>
the gmac driver asks the clock to be muxed one way by asking for one or the other clock rate
<mripard>
Turl: ah, right.
<Turl>
mripard: you could have a of_clk_get_by_name or something though
<mripard>
wens: then fill free to ignore the clock-names thing :)
<mripard>
s/fill/feel/
<Turl>
that scans the clock-names until it finds the right one, takes it index and runs of_clk_get_parent_by_index on it
pwhalen has joined #linux-sunxi
<wens>
Turl: it still returns struct clk* right? kind of overkill IMO
<Turl>
wens: nah, it returns the parent name
<Turl>
that's not the right function name :P
<wens>
oh :p
<Turl>
of_clk_get_parent_name()
<Turl>
takes np and index iirc
<Turl>
anyway
kivutar has joined #linux-sunxi
<Turl>
that's overkill too for the issue at hand too :P
<Turl>
as well as using two toos
<sehraf>
is there a way to tell nand-part to use the remaining space for a partition?
<wens>
that's what I'm using ATM
<libv>
aha, mux_ctrl
<libv>
register in lcdc0 reg space
<Turl>
libv: great IP innit? :)
<wens>
of_clk_init
<libv>
:)
<libv>
aha, seems the tvecs can be switched too, but all from the first lcdc only.
<libv>
interesting stuff
* libv
runs off to implement it
<libv>
but first, one more script.bin and one more register dump to confirm
<wens>
sounds like the port controller in usbc0
mkutsevol has joined #linux-sunxi
Philippe_Fouquet has joined #linux-sunxi
<Philippe_Fouquet>
HI
<Philippe_Fouquet>
I want to use openssh on android
<Philippe_Fouquet>
all compil without error and run (use comand start-ssh)
<Philippe_Fouquet>
but went i want to conect it it crash with error : protoent* getprotobyname(char const*)(3) is not implemented on Android
<Philippe_Fouquet>
some body have ides
<Turl>
I think the error code is pretty self explanatory :)
<nove>
Philippe_Fouquet: that is a android specific question, you should ask #android-dev
Philippe_Fouquet has left #linux-sunxi ["Once you know what it is you want to be true, instinct is a very useful device for enabling you to know that it is"]
Philippe_Fouquet has joined #linux-sunxi
Philippe_Fouquet has quit [Client Quit]
Philippe_Fouquet has joined #linux-sunxi
Philippe_Fouquet has left #linux-sunxi [#linux-sunxi]
Philippe_Fouquet has joined #linux-sunxi
<Turl>
gzamboni: ping
<wens>
my board is stuck on reboot :(
<Turl>
wens: hung killing stuff?
<wens>
hung before actual restart
<wens>
maybe DT mismatch
<Turl>
do you have PSCI enabled?
<wens>
I do
<Turl>
or PCSI or however it's called
<Turl>
I've seen it happen, I think it was uboot mismatch or something
<Turl>
another thing I noticed is that the 2nd core failed to be brought up
<Turl>
or was brought up in some bogus state where it didn't do anything worthy
<Turl>
apps like htop hung on it except if run inside strace
<Turl>
odd stuff
<Turl>
haven't seen it in a while though
<wens>
anyway, should get a few wires to connect reset pin on my CT to my CB2..
<Turl>
there's a reset pin? :p
<wens>
there is :)
<wens>
it's a CPU reset
<wens>
the reset button on the board is actually for the AXP
notmart_ has joined #linux-sunxi
notmart has quit [Ping timeout: 245 seconds]
notmart has joined #linux-sunxi
notmart has joined #linux-sunxi
notmart has quit [Changing host]
notmart has quit [Remote host closed the connection]
notmart has joined #linux-sunxi
<ccaione>
mripard: probably you are right and the problem is still there also for !threaded
notmart_ has quit [Ping timeout: 250 seconds]
<ccaione>
still we have the problem that I cannot ACK the NMI controller if I have not ACKed ahead the device connected to NMI
Turl has quit [Ping timeout: 264 seconds]
rellla2 has quit [Remote host closed the connection]
<gzamboni>
Turl: pong
Turl has joined #linux-sunxi
<mripard>
ccaione: but in theory, tglx seems to agree with your patch, isn't he ?
<ccaione>
in theory yes, but to be honest I have no idea what does he mean with "use the proper functions"
<mripard>
ask him for clarifications then :)
<ccaione>
also "less convoluted" ... on a 10 lines patch????
<ccaione>
yeah, I'm trying to write something without making him angry :D
<mripard>
yeah, you don't want him to be angry
<ccaione>
hahaha
<mripard>
I learnt that with the last patches I sent him... :)
<gzamboni>
Turl: pong
<Turl>
gzamboni: hey
<Turl>
gzamboni: nevermind my earlier ping, I wanted you to check something on the mail server but it's all good now :)
<gzamboni>
if you guys want i can redirect the dev@linux-sunxi.org to several emails addresses
<gzamboni>
its currently redirecting only to you i think
shineworld has joined #linux-sunxi
shineworld has left #linux-sunxi [#linux-sunxi]
printallthething has quit [Ping timeout: 260 seconds]
printallthething has joined #linux-sunxi
e-ndy is now known as indy
<oliv3r>
gzamboni: isn't it fwd to linux-sunxi@gg.com
FreezingCold has joined #linux-sunxi
<gzamboni>
yes, youre right, its forwarded to linux-sunxi@googlegroups.com
<gzamboni>
and devel-tree to turl
<oliv3r>
mail; next week or in 2 weeks i really wanna spend time finishing the setup
<oliv3r>
with fosdem out of the way now :) i should have more time
<oliv3r>
and i have to check the schedule for the book; i probably need some time for that this weekend too
<gzamboni>
:) take your time i dont care to redirect them
<Turl>
gzamboni: we never got to use the devel-tree one, feel free to remove it if you want
<oliv3r>
gzamboni: yeah but it's the whole GG thing that has to go
cubear has joined #linux-sunxi
<gzamboni>
i can setup a list on my server if you want
<wens>
wow the ohci/ehci patches are at v9
rz2k has quit [Remote host closed the connection]
enrico_ has quit [Quit: Bye]
nove has quit [Quit: nove]
davi has joined #linux-sunxi
diego71 has quit [Ping timeout: 245 seconds]
diego71 has joined #linux-sunxi
* Montjoie
is annoyed, trying to buy another cb2, buying it on foreign country is cheaper than mine:(
<rm>
usual thing
<rm>
buying direct from China with free shipping is cheaper than from the greedy "local distributors" hiking the price by 2-3x
boycottg00gle has joined #linux-sunxi
<Montjoie>
2x is what I see
<Montjoie>
if some french people have some experience with foreign buying here, contact me please:)
FR^2 has quit [Quit: Connection reset by peer]
pwhalen has quit [Read error: Operation timed out]
pwhalen has joined #linux-sunxi
davi has quit [Ping timeout: 250 seconds]
<Turl>
mripard: ^ :)
<wens>
just saw an A31s based Android TV box, for around $120
boycottg00gle has quit [Remote host closed the connection]
<mripard>
ccaione: the alarm is also connected to the NMI, isn't it ?
<mripard>
that would be a good testcase for wether it's needed only for the threaded case or not
sehraf has quit [Ping timeout: 245 seconds]
GatoMiador has joined #linux-sunxi
popolon has quit [Quit: Quitte]
sehraf has joined #linux-sunxi
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
bbrezillon has quit [Quit: Ex-Chat]
<GatoMiador>
Hi, I am trying to install Linux on a Merrii M2 (http://www.merrii.com/en/detail.asp?id=157). It have an A20 SoC. I am having problems retrieving the file "boot1.header" from the Android installed and the command "./fel read 0x42400000 0x82d0 boot1.header" didn't work, so any hints where this file could be? I managed to copy "script.bin" from the Android.
diego_r has quit [Quit: Konversation terminated!]
bertrik has joined #linux-sunxi
bertrik has joined #linux-sunxi
bertrik has quit [Changing host]
rz2k has joined #linux-sunxi
mrnuke is now known as mrnuke_
mrnuke_ is now known as mrnuke
paulk-collins has joined #linux-sunxi
pwhalen has quit [Ping timeout: 245 seconds]
pwhalen has joined #linux-sunxi
montjoie_[home] has joined #linux-sunxi
montjoie_[home] has quit [Remote host closed the connection]
montjoie[home] has quit [Ping timeout: 276 seconds]
montjoie[home] has joined #linux-sunxi
y0g1 has quit [Ping timeout: 272 seconds]
y0g1 has joined #linux-sunxi
fredy has quit [Excess Flood]
fredy has joined #linux-sunxi
Philippe_Fouquet has quit [Ping timeout: 246 seconds]
<GatoMiador>
Nevermind about that. It should be an entry on this page http://linux-sunxi.org/Retrieving_device_information explaining one should use try using both methods for entering FEL mode. I tried only the first one and that didn't work for me.
od6 has joined #linux-sunxi
<od6>
hello
<od6>
i just received my cubieboard
<mnemoc>
.oO(where is ssvb?)o
<od6>
but i don't have an hdmi screen. how can i use it?
<od6>
so after connecting it a /dev/ttyUSB0 device will appear and i can connect to my cubieboard via the "screen" command, isn't it?
<plaes>
screen or minicom
<plaes>
or anything else
<mnemoc>
<3 microcom
hawi_ has joined #linux-sunxi
<od6>
ok thank you all
od6 has quit [Quit: Page closed]
<plaes>
btw, you could use Arduino too for that
<plaes>
oh well :)
<ccaione>
fuck arduino :P
<specing>
Fuckduino, now coming to a store near you
<mnemoc>
oshw!
<oliv3r>
while I'm a C guy, I don't appreciate how the arduino IDE uses C++ (without a sane reason)
<oliv3r>
the whole arduino movement should be appreciated
<oliv3r>
oshw, putting the device into users hands
<cubear>
I learned that ARM is so much more awesome than AVR (atmega/arduino)
<oliv3r>
letting people work on neat projects
<oliv3r>
it is not
<oliv3r>
AVR (8bit) is really really awesome
<cubear>
bleh
<oliv3r>
BUT, it's not reall yfair to compare it like that
<atsampson>
they're both pretty awesome, really
<oliv3r>
you can very much use them together rather then either/or
<cubear>
The only thing I like about atmel AVR is the IDE
<oliv3r>
what IDE?
<oliv3r>
avr-gcc is just a compiler
<cubear>
Atmel Studio
<oliv3r>
oh, i don't use propriatery crap IDE's :p
<oliv3r>
gvim is all you need tbh
<mripard>
tsss
<mripard>
heretic.
<oliv3r>
do not tell me your an emacs guy
<mnemoc>
oliv3r: hi, do you know if anyone has reported problems with CMA in stage 3.4? would like to merge the stage and jump to the latest stable
<mripard>
I'm afraid I am
<mnemoc>
*ouch*
<oliv3r>
mnemoc: i haven't been around a lot; but haven't seen anything said here really
<oliv3r>
BURN!
<ccaione>
omg, mripard, that was unexpected
<oliv3r>
i didn't see tha tone comming :p
<oliv3r>
emacs, vim; it's all OS goodness
<cubear>
vim yeah... the first time I used that editor I dodn't know how to exit it. ctrl+c didn't work, alt+f4 didn't work, q didn't work, so I ended up hitting the reset button to exit.
<oliv3r>
LOL
<cubear>
didn't*
<oliv3r>
yeah; vim's initial learning curve is very steep :p
<cubear>
seriously... :q
* atsampson
starts up vim and presses Ctrl-C
<cubear>
that's so different from anything I've seen before
<atsampson>
"Type :quit<Enter> to exit Vim"
<oliv3r>
i think there's vim-easy
<cubear>
":q"
<oliv3r>
atsampson: in the very very early days ..
<cubear>
I ended up using nano
<oliv3r>
hahaa
<oliv3r>
i sometimes even have to use vi
<oliv3r>
because it sits in busybox; and sometimes that's all you have
<cubear>
don't some old router firmware use vi too?
<oliv3r>
well, busybox can be compiled with support for vi
<oliv3r>
so it's available in a lot of embedded platforms
notmart has quit [Quit: notmart terminated!]
<cubear>
anyway some ARM microcontrollers are cheaper than the equivalent AVR that offer similar functionality
<cubear>
and they're 32bit
<mnemoc>
cortex-m3 are awesome too
<oliv3r>
how about power consumption?
<ccaione>
oliv3r: try cortex-m0
<cubear>
Yeah, I have a LPC1756 that I've never used
<oliv3r>
can arm micro's run for years on a single cr322 cell?
<ccaione>
_really_ low power
<oliv3r>
you can run an attiny85 for 2+ years on a single cell
<oliv3r>
even doing wireless communication
<oliv3r>
though I admit, the msp430's are in some scenario's even more power efficient
<oliv3r>
i'm not saying m0 isn't nice, it really is
<oliv3r>
but not every problem is a nail :p
<oliv3r>
and arm isn't the big hammer :)
<oliv3r>
i obviously really really like arm
<oliv3r>
with all the sunxi stuff that we do :)
<oliv3r>
but I also appreciate what AVR does :)
<cubear>
I started on AVR and managed to make a program for xmas LED strip - basically it controls the brightness on 4 PWM channels and plays a lot of different cool patterns.
<cubear>
In fact I ran out of flash space because I programmed in too many patterns
<cubear>
btw I have a question about allwinner
<cubear>
how's the encryption working? I hear there's some hardware acceleration available, but no driver yet
<mripard>
ccaione: the m-0+ draws even less
<cubear>
I'm planning to use cryptsetup to encrypt the sata HDD. Will that slow down the device?
<oliv3r>
cubear: so how's that a bad thing :p
<oliv3r>
cubear: for now; yeah; but montjoie[home] is working on the mainline encrpytion driver
<cubear>
What would I need to install the rootfs on an encrypted HDD?
<cubear>
I take it I'd have to use a nand boot partition with appropriate initramfs to decrypt the rootfs