<damo22>
hi i just discovered A13 stuff and the linux-sunxi project, i have a question regarding the kernel: how come one cannot examine the sources for the android release to work out the specs?
<asiekierka>
there are tons of A10 netbooks still on aliexpress
<asiekierka>
$90-$130
<asiekierka>
the problem is, do ALL of them run Linux properly?
<asiekierka>
me and a friend got Debian running on a RK2818 recently
<asiekierka>
using the Android kernel, tho
<asiekierka>
essentially we'd let the Android kernel boot, then we'd chroot from the boot partition
<damo22>
asiekierka: i guessed this would be a chroot
<damo22>
but what is wrong with that?
<asiekierka>
nothing, really
<asiekierka>
stops us from running Arch or any distro requiring a recent kernel, though
<buZz>
asiekierka: afaik _all_ A10 devices can boot berryboot and get proper linux
<damo22>
afaik if you are referring to linus' kernel you are supposed to refer to the kernel as linux and the rest of the system as GNU if its "proper linux", so the combination can be unmistaken as GNU/Linux. and the whole Android system (OS + Kernel) is just Android afaik. is this correct?
<asiekierka>
pretty much
<asiekierka>
Android is just called Android, perhaps because everyone knows what the term means
<rm>
asiekierka, the link you posted also says VIA 8850 even in the title
<rm>
I'd be wary to buy from them
<asiekierka>
not on the pics, though
<orly_owl>
windows key >_<
<asiekierka>
i believe the problem is they pick the CPUs at "random"
<asiekierka>
many Chinese companies are low on details
<drachensun>
I just looked the sun5i branch did it right too but only in the sun5i_csi/csi0 driver code, didn't seem to change the bug in sun4i_csi
<drachensun>
a10s has it corrected in sun4i_csi
<techn_>
strange.. I cant find any commits related to csi on that branch :/
<techn_>
so 10s branch should be same as sunxi-3.0 branch :/
<Turl>
drachensun: my camera has always worked on sun4i kernel
<drachensun>
you can look at the code around like 858 though and see its different
<techn_>
.. for csi part
<drachensun>
turl: does your camera use that sun4i_csi0 driver?
<Turl>
yeah
<Turl>
both of them iirc
<drachensun>
what program did you test it with?
<Turl>
android camera app
<drachensun>
one anomaly I had was cheese would snatch about 3 frames then crash, fswebcam and everything else I tried crashed right away
<drachensun>
and only my opensuse image would even run cheese
<drachensun>
I haven't tried it in Android
<drachensun>
that import branch is having build errors for me, but of course I'm trying to build it with a regular A10 defconfig, is that expected?
<techn_>
I suggest to import what you need there.. Its 3.0.8 kernel
<drachensun>
ugh ok
<techn_>
and has quite a lot old stuff.. but there are some gems
<drachensun>
I was hoping to just run a test with it and see how the camera ran, I guess I can move the camera parts over and see what happens
<drachensun>
so is the A10s code newer than the sun5i import code?
rz2k has quit []
<mnemoc>
drachensun: yes
<mnemoc>
we have 3 sources of sun5i code. the base we use from the history in Qware's release a year ago. pieces from the gpl-violating a13 SDK tsvetan got in... april? and pieces from the gpl-violating A10s SDK tsvetan got last month
<mnemoc>
the import branch was created from the A13 SDK
<mnemoc>
which is not simple as several parts are destructive for sun4i
<mnemoc>
also there are parts of the A10s SDK destructive for A13 :|
* mnemoc
gone again
<drachensun>
wow
<drachensun>
ok thanks
valhalla has quit [Remote host closed the connection]
hp__ has quit [Ping timeout: 276 seconds]
<drachensun>
alright correction, the a10s code does not have my fix either, it does setup the field value but in another function, the GPL header files adds just enough lines to line it up with another similar function
<drachensun>
since that other function has the field set, I think a calling program only crashes if it calls v4l to capture a certain way
hg_5_ has joined #arm-netbook
hg_5__ has joined #arm-netbook
hg_5 has quit [Read error: Connection reset by peer]
<penguin42>
lkcl__: What are the two mini connectors on the right hand side? One is USB I guess, the other hdmi?
<jinzo>
but still, a ballpark assesment would be interesting
<penguin42>
jinzo: Thing is that's a very versatile, it could be sold into a lot of markets
<servili007>
Turl: Okay, I assumed as much, any suggestions on where to get started?
<jinzo>
I heard that the BOM for a basic iMX6 system is around 100USD
<Turl>
servili007: clone that one and start modifying for your device :)
<servili007>
Turl: Fair enough
<L84Supper>
jinzo: it could be around the price of the wandboards, less if there is interest
<jinzo>
don't they ship the dual core versions only?
<jinzo>
yes indeed.
<L84Supper>
I'm still waiting for an i.mx6 to test, the a10 is just too slow now to work with
<jinzo>
73for dual core - not so bad
ganbold___ has joined #arm-netbook
<Turl>
mripard: stefanro: ping
<jinzo>
hm, the iMX6 doesen't come with the GPU bundled like A10s for example?
<L84Supper>
I've been considering a superior form factor for cpu cards that EOMA, if there is enough interest I'd build with the i.mx6
<penguin42>
L84Supper: Well if you don't care about having the case there is that standard someone was working on U?? (can't remember the acronym)
<L84Supper>
jinzo: the GPU is on-chip
ganbold__ has quit [Ping timeout: 248 seconds]
<jinzo>
then it's the same Vivante GC2000 + GC300 everywhere - suprisingly lack of info on the wandaboard page
<L84Supper>
penguin42: It would have a case
<L84Supper>
I just stopped supporting EOMA
<penguin42>
L84Supper: Ah ok, what shape would you go for; I mean it's ok not to support it if you don't want, but hey if it's the right shape for you
<jinzo>
unfortunately most of the efforts have quite a long product turnaround - even nowadays it's hard to get a board with A10 actually shipped
<L84Supper>
not sure what the problem was with the cubieboards
<penguin42>
L84Supper: They made a few hundred, their is another batch due to ship any day
<penguin42>
they got indiegogo funding for the next batch
<L84Supper>
EOMA is just poor management and lack of tech knowledge
<jinzo>
L84Supper, actually I think that most of the project out there are a bit too entusiastic (from RPi to Wandaboard)
<L84Supper>
no point in making cards if you can't purchase silicon
<jinzo>
a lot of hype and then big problems with delivery.
<jelly-home>
wasn't the point of making cards in creating a standardized form factor
<L84Supper>
I don't see that many ARM netbooks selling
<penguin42>
L84Supper: I'm not sure it's necessarily a problem to take a year to bring a project like this up, the problem is that the reality as long as you get your level of optimism right is to realise it's going to take a while and you have to design for the chips that are going to be available then not when you start
<jinzo>
I know it's hard ... but still.
<L84Supper>
might be easier to just mod the chromebooks
<jelly-home>
(which means reusable components around it, which means profit)
<penguin42>
L84Supper: Yeh it would be nice if you could buy the boards - although I guess the olimex (?) boards are pretty much there
<L84Supper>
penguin42: it's a few weeks, we been designing mainboards since the 8086
<penguin42>
L84Supper: Fair enough, but that's ok if you are in the business
<L84Supper>
penguin42: we just need a supply of parts, we looked at the i.mx6 a year ago, and it's been nothing but problems with Freescale
<penguin42>
L84Supper: Nod
<penguin42>
L84Supper: Unfortunately true of a lot of ARM stuff
<L84Supper>
poor management, design problems
* penguin42
isn't aware of any A9 or better parts you can just buy
<L84Supper>
yeah, that's one of the problems
tinti has quit [Remote host closed the connection]
<L84Supper>
the i.mx6 was supposed to be available last June
<penguin42>
chips always take longer than expected; especially since it's like 3 months turn around when you screw up
<L84Supper>
depends on the supplier
<L84Supper>
some just have better engineers and management
<penguin42>
L84Supper: It's unusual to be much less, especially given it takes you to find out why/how you screwed up
<penguin42>
(and convince someone to form out another $1M+ to have another crack)
<L84Supper>
the problem is how many boards will you sell?
<penguin42>
L84Supper: The cubieboards problem (ditto Pi) was that they underestimated that by lots
<L84Supper>
some people dream that millions of people are holding off buying PC, notebooks and tablets until there are FSF endorsed versions
<jinzo>
L84Supper, on the other hand - with FSF endorsed version you have guaranteed quite some early adopters that will buy the product even if it's more expensive than the competition and less capable
The-Compiler has left #arm-netbook ["WeeChat 0.4.0-dev"]
<jinzo>
I'm not thinking that'll make you rich - or that it's even enough to break even
<penguin42>
jinzo: Problem is when you're talking $10M+ for a chip design I don't think there are enough there
<jinzo>
but it's something
<jinzo>
depends what the chip does
<jinzo>
if you can use it from mobile phones to networking gear
<penguin42>
jinzo: You can't
<L84Supper>
I'm not even talking about a chip, just boards
<jinzo>
yeah I know
<L84Supper>
hardly anyone cares about FSF or open hardware
<penguin42>
L84Supper: I'm talking chips; I think for boards there is enough interest to pay - the cubieboard and Pi are showing that
<jinzo>
L84Supper, alas - is it a secret for what company you work for?
<penguin42>
it makes it a lot easier to keep work and play separate
<hno>
There is quite a bunch of people and organisations that is caring for open hardware today, and for perfectly economically sensible reasons.
<L84Supper>
most people just car about a status symbol at just about any price, and most of the open source community just go for lowest cost
* penguin42
admits to being incurably tight
<L84Supper>
how many people really cared about the Pi having a binary blob for a bootloader?
<hno>
I do not agree that open source community equals lowest cost.
<jinzo>
L84Supper, but it was not marketed as an open source or. even open device
<L84Supper>
hardly anyone, they could have kept it closed and still sold 100k's
<jinzo>
it was marketed as a cheap PC
<jinzo>
and hyped
<jinzo>
and even with all their failures, they built a big community
<L84Supper>
jinzo: in the early days it was hyped as open hardware
<penguin42>
L84Supper: I think you could get 10k+ sales on the back of something being open; the community is very large these days - scarily so
<jinzo>
we have cubieboard
<penguin42>
L84Supper: Sure, that's not millions, but I bet you could get 10k - but it really has to be fully open
* hno
would not have cared about the rPI GPU bootloader if they had placed a small SPI flash on the board for it.
<jinzo>
with roughly 3k orders
<L84Supper>
cubie and wandboards now
<L84Supper>
whats left to fill any voids right now?
<jinzo>
an actually shipping product I can order and get?
<penguin42>
L84Supper: There is the olimex stuff
<jinzo>
not order and wait unknown time to get it :D
<L84Supper>
the cubies and "in the mail" :)
<penguin42>
L84Supper: They seem to have delivered some
<jinzo>
but regarding niche, as it was discussed - maybe something aimed at Internet of Things - simmilar to 8devices carambola
<hno>
olimex boards is readily awailable. Only A13 yet, but easy to get with only some days deliery time.
<jinzo>
hno, indeed - but probably the only one from the Allwinner family
<jinzo>
the mk802's too - but there stuff is a bit more complicated :P
<hno>
Olimex is also working on an A10 board. But still early prototype state.
<penguin42>
L84Supper: And the guys from the 1st batch of cubieboards seem to be asking questions etc on their forums
<jinzo>
and having problems, they are also working on the TI whatiscalled-soc-that-is-in-beaglebone board
<hno>
And Cubieboard is maturing. There likely will be some available in january.
<penguin42>
difficulty with the A10 is that the tablets with them are so damn cheap; there is a UK offer for a tablet for #45 on a half price sale (very small batch available on that low price though)
<penguin42>
and a year ago it was an ok spec, these days it's way under the bar
<hno>
Yes, the A10 is dated for tablet usage today.
<L84Supper>
I just don't see there being any real market for an a10 on a card, nobody really wanted it in China 6 months ago, we asked around and they wanted "faster than a10 and windows compatible"
<hno>
that's why they are launching the A31 now, and soon A20 as well.
<penguin42>
L84Supper: The cubieboards guys raised $96k of a 50k indiegogo campaign in a few days, so there is demand
<jinzo>
hno, indeed - but that's like with most hardware products: "announcing a quad core gadget for 99USD, breaking stuff, available in a year or so ofcourse"
<hno>
A31 products are available for preorder now. Expected to ship in a couple of weeks.
<penguin42>
L84Supper: And there are loads of higher end boards now, the oli stuff, the cubox and others
<hno>
but not looking at A31 until there is some indication of a usable SDK release.
<hno>
A10 and A13 is quite sufficient for my uses right now.
<jinzo>
we'll see how that pans out indeed.
* penguin42
likes the look of the mirabox but it's apparently not shipping yet
<L84Supper>
penguin42: lots of people consider their time to be of very little value as well
<penguin42>
L84Supper: Sure, but at the 100k mark it's getting to be a useful amount
<hno>
and will also dive into openrisc in January for some peopel that really really NEED open hardware stuff.
<L84Supper>
penguin42: how many boards were promised in exchange for the $96K
<hno>
L84Supper, which $96K?
<L84Supper>
whats the margin on those boards? few $
<jinzo>
L84Supper, I think that 2k are being made - but the info is somewhere on the forum/ml
<hno>
The indiegogo cubieboard capaign ended up at about 1500 boards.
<hno>
they will produce a bit more.
<hno>
there is 1K + 1K PCBs produced, but not all of the extra 500 will be soldered I think.
<penguin42>
that seems to be reasonably in the order of normal small scale commercial production
<hno>
Yes. Would personally have preferred the indiegogo campaign had a bit more margin so that they actually could get in shape for producing 1K batches after this. But will likely require a second funding campaign.
<jinzo>
indeed - but then again it probably wouldn't sell that good/raise that much money.
<jinzo>
maybe the A10 will be cheaper now that there's a successor
<L84Supper>
isn't somebody here working on an i.mx6 laptop?
<penguin42>
jinzo: There is also the possibility it will disappear
<L84Supper>
they made the imx5 netbook
<hno>
The A10 is likely to stay some years more. But sure, in some years it will be gone.
<L84Supper>
maybe it's just a better idea to find a easily purchased arm soc and supply the design files
<penguin42>
L84Supper: I think the problem though is that there aren't many
<hno>
L84Supper?
<hno>
A10 & A13 can easily be purchased via Olimex
<hno>
and possible to pull some strings and purchase in China as well without too much effort.
<L84Supper>
if the i.mx6 starts to float around on the Chinese trading sites then that could be it, just like the arm9 2440 and the sam9
<L84Supper>
the way to get cheap boards is just supply the files to the board makers in China
<L84Supper>
they like to make 10K boards at $2 margin each
<penguin42>
L84Supper: Nice thing about that MX6 bunniethingy link that started this covnersation
<L84Supper>
penguin42: it's nice to have the open files, but it might be too complex a board
<L84Supper>
maybe if i cut it down to something that could be more general purpose
<hno>
L84Supper, mainly someone supplying an open hardware design they can use.
<L84Supper>
they could make copies of the panda right now
<L84Supper>
maybe I'll talk to them next week
<hno>
and some reasonable supply of the required components.
<L84Supper>
they have a Chinese website as well
<L84Supper>
just found 10K of the omap 4430's
<L84Supper>
they could also just easily respin the beaglebone
<jelly-home>
sounds seeedy
<L84Supper>
parts are sold all over China
<L84Supper>
we buy parts from manufacturers that make boards, we get to ride on their megaorders
<damo22>
i had a quick look at the olimex a13-olinuxino linux-sunxi kernel config, apparently they are using completely open drivers *except* the wifi addon uses a proprietary firmware... why was this chosen?
<Gumboot_>
Maybe it's cheap?
<hno>
L84Supper, I don't think they will accept building Pandaboard without someone from Pandaboard approving it.
<hno>
damo22, there is not very many wifi usb modules that do not require firmware.
<damo22>
well maybe someone could make an open addon wifi thing for this and it could be approved by the FSF?
<hno>
Sure, it's a standard USB connection.
<hno>
But A13 is not FSF endorsed at the moment due to it containing both MALI and Cedarx which will likely lure users into using proprietary pieces.
<damo22>
:(
<damo22>
unfortunate
<hno>
lima project will help the first. And if someone reverse-engineers how to use the CedarX hardware decoders then things may change.
<hno>
but it't not something that will happen tomorrow or even next year.
<damo22>
yeah
<damo22>
is it possible to run a VGA on this board without using mali at all?
<hno>
Yes ofcourse. Plain framebuffer is all open.
<hno>
even HDMI.
<hno>
well, not on the A13. But on A10.
<damo22>
then i can still run trisquel on it for example without wifi or mali or cedarx
<libv>
it's just a steaming pile of ...
<hno>
libv?
* libv
goes back to playing with rgb565 textures on lima
<libv>
the disp driver is a steaming pile of ...
<libv>
but it at least is "free" software
<hno>
yes.
<damo22>
im not interested in gfx on it anyway i want to run it headless
<libv>
free for all who have a strong stomach ;)
<hno>
Only those who dare diving into the code need one. The others just hope it works most of the time.
<damo22>
i dont mind code
<Turl>
libv: free doesn't mean good, indeed :)
<damo22>
i reverse engineered a small quirk for a usb sound card by bus snooping
<damo22>
mind you, i would never make a decent living doing this kind of stuff
<hno>
damo22, there is code and there is code...
<hno>
Allwinner code has a too long history from times before Linux.
<damo22>
how do you mean?
* damo22
curious
<hno>
Allwinner drivers were originally written for another OS, mainly by hardware engineers. And then glued onto Linux.
<hno>
and patched up a bit for A10.
<damo22>
oh ok
<Triffid_Hunter>
hno: argh I know just enough to realise what a disaster that spells
<damo22>
but since the code leak there is work being done to utilize the specs and write the drivers from scratch?
<hno>
What leak?
<damo22>
as in, the specs inferred by the old stuff?
<hno>
Allwinner published the A10 sources.
<damo22>
um ok, i was told differently
<hno>
the only truly leaked stuff in our tree is the A13 core support.
<damo22>
oh right
<hno>
which is almost the same as A10.
lkcl_ has joined #arm-netbook
<damo22>
is the leaked stuff for A13 useful for inferring ways to improve the A10 stuff?
<hno>
Not really. But some drivers from A13 SDK release has bug fixes that has been merged, but those are not leaked.
<damo22>
ok
lkcl__ has quit [Ping timeout: 264 seconds]
<hno>
Just to be clear. The leaked stuff is only the .c files in arch/arm/mach-sun5i/
<hno>
the rest, including headers in same directory is all published.
<damo22>
okay thanks
gimli has joined #arm-netbook
<damo22>
i heard from someone that 3.0.8 is the best version for A13 support at the moment, which git branch of linux-sunxi is that?
<damo22>
or in particular for the olimex board
<hno>
3.0.8 is the version used by Allwinner. You can find this in Allwinner SDK releases.
<hno>
Not sure I would call it "the best".
<techn_>
damo22: If that so.. why come they are not participating to get a13 support par with a10 :/
<techn_>
I think we got the sources to get it there
<hno>
techn_? who they?
<techn_>
hno: <damo22> i heard from someone that 3.0.8 is the best version for A13 support at the moment, which git branch of linux-sunxi is that?
<damo22>
=hno
<damo22>
?
<techn_>
oh.. sorry.. someone :)
<hno>
damo22, be warned that the Allwinner SDK releases for A13 are NOT GPL compliant.
<damo22>
oh
<damo22>
are the A10 ones fully compliant?
<hno>
as arch/arm/mach-sun5i/ is a binary blob.
<hno>
yes
<Turl>
hno: do you happen to know if I need to set any special config option to get uart output on 3.7? I upgraded my uboot to sunxi-current and used the 3-option bootm with "uimage - dtb" addresses
<damo22>
cool
<hno>
at least the ones I have seen so far.
<Turl>
but I am not getting any output
<hno>
Turl, not sure. Haven't had time to play with 3.7 yet. Been ill and then doing hand-off to other people at my current customer.
<damo22>
who makes a development board based on A10? eg, a nice one with lots of usb current draw capability?
<hno>
damo22, the cubieboard with external power supply can supply a good current on it's usb ports. And I would expect Meles to as well.
<damo22>
do any of them support open hardware? as in being able to use it commercially?
<hno>
What have open hardware with being able to use commercially?
<damo22>
good question. i am not a lawyer
<damo22>
but olimex readme puts these two ideas in the same sentence
<damo22>
so i assumed...
* buZz
gets a cubieboard next week \o/
<hno>
open hardware is about having access to board design and not being locked to a manufacturer.
<buZz>
finally!
<buZz>
w00t
<hno>
and allowed to make modifications to the same.
<buZz>
s/gets a/receives his/
<hno>
buZz, Cubie?
<damo22>
hno: you are a wealth of knowledge!
<buZz>
hno: yah
<buZz>
rich package, so i expect it to be the 1GB version
<hno>
damo22, just been around here for some time.
<buZz>
finally some proper brains to put in a lapdock ;)
<hno>
buZz, there is only 1GB versions produced.
<buZz>
ok :)
<hno>
the 512MB version was produced in a very small quantity in the first batch only.
<buZz>
on #cubieboard there are ppl with 512 still
<hno>
or second actually. But first "production" batch.
<buZz>
yeah, ok
<buZz>
so 1GB it is :)
<buZz>
but if i discovered that odroid board just a couple hourse earlier, things would have been different ;)
<hno>
U2?
<damo22>
hno: i wish to buy a board that has the following requirements: arm cpu, 2 or 3 500mA usb 1.1 (or 2) ports, open source GPL drivers, trying to think of others....
<damo22>
and allows me to sell the hardware with software in a commercial project
<Turl>
why would any vendor oppose to such use?
<Turl>
I mean, you will be buying them large qty of boards right? :)
<damo22>
well hopefully, if my idea works
<damo22>
so far it sounds like the A10 is a better choice at the moment
<hno>
Cubieboard, Olimex boards (both A13 and IMX233, and some others.
<damo22>
ideally i would like to choose a board that will eventually be approved by FSF
<damo22>
because i am all for freedom
<damo22>
even though i still want to make money
<hno>
None of the boards available today are likely to get FSF approval during their commercial lifetime.
<damo22>
:(
<damo22>
sad day
<hno>
but still your product might, depending on what your application is.
<damo22>
really?
<hno>
Yes. An A10 board without hdmi/lcd/vga/video out could probably pass.
<hno>
maybe...
<damo22>
id like to share my idea with you guys because you have shared so much of your knowledge of the boards
<damo22>
but perhaps another day when i am closer to having it working
<hno>
whenever you feel ready to discuss it.
<damo22>
:)
<damo22>
1 clue: decentralizing servers for the average common user
<damo22>
opposite of cloud computing
<damo22>
someone told me "everything on A13 is working under linux"
<damo22>
must be wrong
<rm>
sooo you mean like FreedomBox then
<Turl>
damo22: linux is a kernel, so you kind of could say that I guess
<damo22>
hno: considering it is a more powerful device
<Turl>
hno: any thoughts why is it dying after [ 0.430000] turn off boot console earlycon0?
<hno>
damo22, because with it having MALI and CedarX which both requires userspace proprietary libraries users would be too tempted to use said proprietary software.
<hno>
even if you do not include those libraries in the software shipped with the device.
<damo22>
hno: in that case i will use A10
<hno>
A10 has the exact same problem wrt FSF.
<hno>
The only practical difference between A10 and A13 is that A10 is much more complex to design a board for, and A10 have HDMI and SATA which quite many wants to have.
<hno>
oh, and that A13 is crippled to only handle 512MB RAM.
<damo22>
thanks for that info
<hno>
The kernel GPL issue of A13 is only if you use the SDK supplied by Allwinner.
<damo22>
i understand now thanks
<damo22>
sata2 is a nice feature of A10
<hno>
Yes, very nce.
tinti_ has joined #arm-netbook
<hno>
Actually I am not 100% sure A13 really is limited to 512MB. Can probably do 1GB with the right RAM chips like the ones used on Cubieboard. But specifications say 512.
<damo22>
is there anything lacking from the current boards based on A10 that could be fixed by inventing a new board?
<damo22>
like major ignored features?
<damo22>
sorry i just have no clue about hardware
<hno>
there is some pins missing on cubieboard, and some poorly chosed control pins..
<hno>
Olimex A10 board exposes a bit more, but is much bigger and costs a bit more.
<hno>
plus not available yet.
<damo22>
do any of the A10 boards have tactile buttons like a keyboard input?
<damo22>
for example i know the A13 olimex has 5 buttons
<hno>
cubieboard have none. But trivial to add such buttons, you only need buttons and some resistors.
<damo22>
it would be gpio based?
<hno>
There is also a full keyboard controller in the A10, but Cubieboard do not expose those pins.
<damo22>
interesting
<hno>
the "android" keys are not GPIO based, its based on a low frequency ADC designed for the purpose.
<damo22>
wow ok
<hno>
You can also to GPIO based keys, but you then need to poll the GPIO pins frequendly.
<penguin42>
hno: ADC? With resistor values for each key?
<hno>
Yes. Or actually a resistor chain with buttons between each.
<hno>
same thing.
<penguin42>
I guess if you've only got 3-5 buttons then gpio is overkill
<damo22>
rm: freedombox looks like a good project
<hno>
yes, and too many wires. A resistor chain is just one trace.
<lkcl_>
damo22: freedombox is a software project, you're aware of that?
<hno>
or two, if including the GND trace needed for the buttons.
<damo22>
lkcl_: yes, i would like to create a project that is both hardware and software where the software is GPL and the hardware + software combination has the potential to be approved by FSF
<penguin42>
lkcl_: Hmm so your board is totally silent?
<lkcl_>
damo22: that's one of my goals. it's actually really really hard to do, because the processors that are FSF-Endorseable are... well... they're shit, basically.
<lkcl_>
embedded ones, anyway
<lkcl_>
penguin42: yep.
<penguin42>
lkcl_: Hmm, have you checked all the volts are right, and can you see clocks on anything?
<hno>
lkcl_, so no sign of life at all now? But it worked with livesuit once?
<lkcl_>
ok i have a FT232, its LEDs go off when the power's on
<hno>
?
<lkcl_>
i need to get my digital oscilloscope out.....
<lkcl_>
hno: maybe. wits-tech were a bit quiet as to answering questions
<lkcl_>
hno: what happened to arch/arm/cpu/armv7/sunxi/gpio.c
<hno>
well, they are chineese. no answes other that everything is allright.
<hno>
lkcl_, what about it?
<lkcl_>
it seems to have moved to sunxi_gpio.c
<lkcl_>
but i've got undefined reference to `gpio_set_value'
<hno>
Odd.
<lkcl_>
yeh
<hno>
undefined reference from where?
<hno>
and using what config?
<lkcl_>
link phase
<lkcl_>
cubieboard (slightly modified)
<lkcl_>
not much
<lkcl_>
i want to set PI19 as a) an output b) set it to 1
<lkcl_>
that's for pull-up of USB0-DRV (PI19 has been tasked as USB0-DRV)
<Turl>
lkcl_: I built hno's sunxi-current cubieboard config an hour or so ago, built fine
<lkcl_>
Turl: are you using gpio_set_value?
<hno>
cmd_gpio are.
<lkcl_>
hmmm... cmd_gpio - tag not found
<Turl>
I didn't touch any of the code, just make cubieboard_config; make
<lkcl_>
ahh it's u-boot-spl which is failing to build
<hno>
You need GPIO settings in SPL?
<lkcl_>
well.... probably not.
<lkcl_>
but i put a set_gpio call into board.c :)
<lkcl_>
in gpio_init
hg_5__ has quit [Ping timeout: 260 seconds]
<hno>
lkcl_, there is a CONFIG_SPL_GPIO_SUPPORT define to enable GPIO support in SPL.
<lkcl_>
ack
<hno>
or you can guard the code with an #ifndef
<lkcl_>
ah?
<hno>
#ifndef CONFIG_SPL_BUILD
<hno>
around the parts you don't want to build in SPL.
<lkcl_>
actually it'd be good to have it in the SPL
<hno>
why?
<damo22>
hno: i noticed you have a few cool github projects :), can you describe in a nutshell the main difference between linux-allwinner and linux-sunxi, or why you decided to fork this, i am quite curious.
<hno>
linux-allwinner is the old name.
<lkcl_>
because if the USB0-DRV is enabled i can test a known pin
<lkcl_>
i don't even know if this thing's reading from the SDcard yet
<penguin42>
lkcl_: Should you see USB clocks, maybe SD card clocks?
<lkcl_>
i need to get my digital scope don't i.... *sigh*
<hno>
lkcl_, if you get FEL mode working then you can play with all GPIO pins without any sofotware loaded.
<hno>
I would start there.
<damo22>
hno: but it says "forked from linux-sunxi/linux-sunxi" does that not imply that the project allwinner was made after sunxi? sorry I'm confused with terminology
<hno>
damo22, no, because what is today linux-sunxi/linux-sunxi was originally amery/linux-allwinner
<hno>
which I forked from.
<hno>
just as linux-allwinner/uboot-sunxi was originally hno/uboot-allwinner
<damo22>
thanks for the clarification
<hno>
lkcl_, do FEL mode work?
<hno>
lkcl_, to the level that the device shows up in lsusb if you jumper it for fel boot.
<hno>
lkcl_, if it does then you can use fel-gpio from sunxi-tools to play with GPIO settings.
<damo22>
hno: would it be clearer if it said "linux-allwinner: forked into linux-sunxi" ?
<hno>
damo22, there is very good reasons why we renamed the repositories and forked back to our personal.
<damo22>
ohhh ok
<damo22>
sorry
<hno>
this was to update all forked repositories to point to the linux-sunxi repositiories instead of our personal ones.