<steev>
and then after it's done compiling, i want spl/u-boot-spl.bin and u-boot.bin
<steev>
that spl/u-boot-spl.bin does not seem to work for me here at all, it just loads the uboot from nand, however if i use the u-boot-sunxi-with-spl.bin it will at least attempt to use the u-boot that i've compiled
<steev>
then... what are the ACTUAL instructions to make an sdcard
<steev>
because what's in the wiki is obviously wrong, and i'm tired of trying the same things over and over to have it fail when i can move to a different device and have it up and running in a few seconds, and being told to fix then wiki when i don't have the knowledge in what is actually wrong
<steev>
isn't particularly helpful
<Turl>
steev: TL;DR version is configure crosscompiler, make Cubieboard2; make; dd if=u-boot-sunxi-with-spl-blah-blah of=yoursdcard skip=8
<Turl>
steev: good then, so it must be something on your side (compiler, luck?) or some regression since commit b4bca5e0
<steev>
Turl: actually
<steev>
using mine works too
<tm512>
so I'm preparing a rootfs that has busybox, musl libc, and pkgsrc for package management
<steev>
the instructions say to use spl/sunxi-spl.bin - this fails for me - using u-boot-sunxi-with-spl.bin works just like yours does
<steev>
now to turn debug_ll back on to see why it's not booting, but i'm going to guess it's complaining about the machid even though it says Using machid 0xf35 from environment
<Turl>
steev: were you dd'ing just the SPL?
<Turl>
you need to dd uboot too (u-boot.img)
<steev>
Turl: no, the instructions say to do spl/sunxi-spl.bin bs=1024 seek=8 then u-boot.bin bs=1024 seek=32
<steev>
when i do that, it throws that error that you see
<Turl>
steev: u-boot.IMG for a not ancient release
<steev>
when i use u-boot-sunxi-with-spl.bin bs=1024 seek=8 and then u-boot.bin bs=1024 seek=32 it works
<steev>
ykchavan: i know, i ignore that part, and clean before starting
<oliv3r>
arokux: i really feel something about leaving them at usb0 and usb1, the logical ordering of the host ports; i know usb0 is sometimes otg
<steev>
my issue was that it was saying to use u-boot.bin and it should have been u-boot.img with newer u-boot
<ykchavan>
ok
<steev>
i'm actually up and running, not just trying to anymore :)
<ykchavan>
:-)
<ykchavan>
I removed 'seek=544' from it to really clean.
n01 has joined #linux-sunxi
<steev>
ykchavan: i have no idea
<steev>
whenever he's around, he's around
<ykchavan>
lol
massi has joined #linux-sunxi
Black_Horseman has quit [Quit: Zwi se logou mou!!!]
Black_Horseman has joined #linux-sunxi
Black_Horseman has quit [Changing host]
Black_Horseman has joined #linux-sunxi
gnufan has joined #linux-sunxi
mcbrick has quit [Quit: Leaving]
soldoKyn has joined #linux-sunxi
Tsvetan has quit [Read error: Connection reset by peer]
eebrah|FTS has quit [Read error: Connection reset by peer]
<oliv3r>
steev: while backreading; you may have enabled FALCON_MODE and with a wrong gcc version, your SPL ends up being more then 24kb; if it's too big, it fails to load the rest
<steev>
oliv3r: i did absolutely nothing but clone the repo and run the command given. the issue was that the instructions said to use u-boot.bin and not u-boot.img
<oliv3r>
steev: the 'then' is wrong, it should be u-boot.img; a recent change
<oliv3r>
i should backread better before commenting :)
<steev>
it's been fixed :)
<steev>
and i'm up and booting on my own built stuffs
<steev>
still need to test something
<steev>
in fact, i'll go do that now
<oliv3r>
steev: but yeah, u-boot_with_spl is even better; as that includes everything
<steev>
oliv3r: yeah, i realize that now :)
<oliv3r>
i'm done backreading yay :)
<oliv3r>
i should learn to first backread; then comment
<steev>
and yay, it actually works with my hdmi->dvi cable to the SOYO monitor (which normally is a huge pita)
<oliv3r>
not comment as i backread ;)
<steev>
which, very few other machines here actually do
<oliv3r>
hdmi -> dvi is a pita?
<tgaz>
i'm struggling (or was struggling before going to work) with making gadget serial work on cb2. my host doesn't pick up any device. haven't dug deep yet, but has anyone made this journey and could share some tips on getting it up?
<oliv3r>
tgaz: not me :)
<Montjoie>
Why I cannot see the boot on HDMI display wih the cubieboard2 ? Does I need to activate/change something ?
rellla has joined #linux-sunxi
<oliv3r>
Montjoie: there's no HDMI or display driver in u-boot
<oliv3r>
Montjoie: once the kernel initializes those +fbdev you can see the rest of the bootlog
<Montjoie>
so I need the framebuffer
<steev>
oliv3r: that monitor's edid lies
<tgaz>
Montjoie: yes, the sunxi framebuffer and the sunxi HDMI driver (AFAIK)
<mnemoc>
maybe it's a good idea to keep a sunxi-3.10 backporting sunxi mainlined code
<mnemoc>
a dts based sunxi-3.10 "stable" branch
atiti has quit [Ping timeout: 245 seconds]
<arokux>
mnemoc, patches welcome :p
* arokux
is happy to say this phrase!
<focus_it>
hi, i got linaro distro running off sata disk for eoma A20 - its booting ok and network ok, but when try to do apt-get to install packages, it shows -11 system error - log of that here: http://www.gplsquared.com/eoma_boot/linaro_apt_problems.txt
<arokux>
oliv3r, aliases ehci1 and ehci2 are really ids, they don't do anything more. bindings actually have ehci0 and ehci1
<focus_it>
arokux: hmm.. so what you are saying is that although interface is working, then networking stuff needs to be set up manually?
<focus_it>
oops! sory arokux - just notice cable unplugged -wait a mo..
<arokux>
focus_it, :D
<arokux>
focus_it, try to ping google.com then
<focus_it>
arokux: ping google.com is still unknown host, dmesg shows link up, and then eth0: no IPv6 routers present
ykchavan has quit [Ping timeout: 245 seconds]
<arokux>
focus_it, i'm not ubuntu guy, but here is what I think: ubuntu installer will probably configure network for you. but if you are using a rootfs then you'd need to do it by yourself... but again, i'm not ubuntu guy.
<focus_it>
arokux: that be good enough - i work on it - thanks
<arokux>
focus_it, yeah, just read that page, it should be fairly easy
<focus_it>
ping 8.8.8.8 works
<arokux>
focus_it, you miss dns servers in /etc/resolv.conf, but do not add the manually, use learn how to use dhcp from that doc
<focus_it>
ok
<focus_it>
arokux: YEEEEHA!! edited the /etc/network/interfaces file and now goole comes up!!
<arokux>
focus_it, document what you've done, so that you do not need to fish for this info once again later
<focus_it>
everything is being 100% documented for lkcl so that he can launch and have some infrastructure at time of launch
<arokux>
focus_it, why do you do this favor to him? :p first I thought because EOMA68-A20 is opensource hardware, but it is not. now i'm puzzled.
<focus_it>
arokux: first it was open source, then for some reason he went to wits and that closed it, but his plans are to release everything once has cash flow
orsic has quit [Remote host closed the connection]
<focus_it>
arokux: if there are other open projects i help as much as i can. i stopped believing in closed projects.
<focus_it>
the more open, the more help
<arokux>
focus_it, I see. olimex is completely open *now* (as far as I know)
<focus_it>
really? i thought the pcb files were (gerbers and cad files) were closed? (schematics and software open)
<focus_it>
arokux: my idea is that kicad has come of age, and we can cut and paste any cpu board and design and make a product. proprietary pcb packages kill that idea.
<arokux>
focus_it, can you look into it and tell which state of openness does it have? :)
orsic has joined #linux-sunxi
<focus_it>
arokux: i think these are eagle files - sort of free - but to do anything decent requires a largish payment.
<focus_it>
kills projects with one stone if engineers can't freely copy paste pcb designs
<arokux>
focus_it, for opensource Eagle is 120 EUR or so
<focus_it>
arokux: i used to be expert in eagle cad - the fees are higher for the best package and if 5 engineer need it here, we must spend thousand of dollars
<focus_it>
only the high end good for 20cm x 20cm boards and lots of layers for our products
<arokux>
<Tsvetan> Eagle non for profit licensee is EUR 100
<arokux>
Tsvetan is an engineer from Olimex
<arokux>
but I actually fail to find such low price at the website
<focus_it>
too much red tape and if they change condition all projects toast!!
<focus_it>
kicad is free regardles of size and layers - only one thing lacking is decent autoroute
<arokux>
focus_it, aren't there converter from eagle to kicad?
<focus_it>
yes
<focus_it>
but i manually route all pcbs because it is a lot better for small to medium size project
hipboi_ has quit [Quit: Leaving]
<soldoKyn>
ssvb: hi! here I am again! I'm trying to build panda3d engine on the olinuxino a20 with official olinuxino debian, latest 3.4 kernel and the mali drivers correctly installed (and also tweaked :-) ).
<soldoKyn>
I've got some error regarding things not defined. For example GL_COMPRESSED_RGB_S3TC_DXT1_EXT is not defined in sunxi/mali/include/GLES2/gl2ext.h but is correctly defined in my x86 fedora 19.
<soldoKyn>
Is something that I can fix (for example with a brute copy from my x86 fedora 19 to the olinuxino debian) or there's a particular reason for this issue like lack of functionality by the mali driver or something else? thanks!
<arokux>
focus_it, so are able to install packages now?
atiti has joined #linux-sunxi
<focus_it>
arokux: yes!!! fantastic!!!! thank you very much :)
<focus_it>
all to be documented
<arokux>
focus_it, cool
gnufan has quit [Ping timeout: 248 seconds]
gnufan has joined #linux-sunxi
<oliv3r>
focus_it: so you gonna export from eagle to kicad, the olimex projects :) :)
<arokux>
oliv3r, don't think so :) we were just discussing the state of openness of olimes designs
<focus_it>
oliv3r: always problems with schematics, pcb and so on - better just to start with kicad and just get on with building up the stuff for that
<focus_it>
also errors and bugs get copied
<arokux>
focus_it, I wonder how much more effort one needs to invest if designing with KiCAD vs. Eagle.
<focus_it>
pcmcia socket for example - is not 100 correct
<focus_it>
all kicad needs is a good self contained autorouter
<focus_it>
other than that, it can do far more than any other packages i know of when it comes to pure pcb work
<focus_it>
because its all in text files, i can for example open up the data into spread sheet, save it back and i get perfect arrangement of pins and numbering and so on
Olaff has quit [Quit: Cassos]
<focus_it>
this kind of work impossible if the data is coded into binaries
<arokux>
focus_it, eagle files seem to be in plain text too.
<focus_it>
because everything in text files, easier to build tools aroun d it
<arokux>
focus_it, true
<focus_it>
ooo.. i did not know that arokux
<focus_it>
still i get worried about the crap nature of proprietary - pull everyone into one project with promises of freedom and then kill it by taking out one company
<focus_it>
if the irrevocably commit something to public domain, then it be better - like fedora and rhat
<focus_it>
mysql and oracle aquisition for example
<focus_it>
open office and oracle again
<arokux>
focus_it, no, I was wrong Eagle is binary actually.
<focus_it>
mono and .net
<focus_it>
good to know arokux
<focus_it>
i was tempted to go mono and .net - and then it gets dumped by microsoooft - luckily i went with gambas and never looked back
<arokux>
focus_it, I'm just asking because sometime using a commercial tool can save you lots of time. you say this is not the case with kicad vs eagle, which is nice.
<arokux>
focus_it, I'd go Python, but I do not know your requirements.
<focus_it>
in kicad v eagle, i believe kicad wins for small and medium projects out right against commercial offerings.
<focus_it>
Somthing the complexity of a cubieboard is easy enough to do in kicad and not suffer any consequences
<focus_it>
there are a ton of peripheral projects under way for example within CERN to uprade all aspects of kicad for their internal projects
<focus_it>
so i think kicad will keep on improving as it is essential for a lot of pcb projects
<focus_it>
universities for example
<focus_it>
arokux: python good, i know little bits of it and tried to make gui, but gambas is lightening fast for gui work where there are a lot of screens.
orsic has left #linux-sunxi [#linux-sunxi]
gnufan has quit [Ping timeout: 260 seconds]
<rellla>
f..... cross compile!
<oliv3r>
rellla: cross compile ie easy
<oliv3r>
use openbricks
<oliv3r>
the buildsystem used to build geexbox :)
<oliv3r>
i know they are activly building a10 images
<arokux>
oliv3r, for which hardware?
deasy has joined #linux-sunxi
<oliv3r>
arokux: cubieboard, 1 i know of, i have it on my SD card
<oliv3r>
they may have advanced to cb2 now
<arokux>
oliv3r, yocto, buildroot, openbricks - how many others?! the list makes me sick
eebrah has quit [Ping timeout: 264 seconds]
eebrah has joined #linux-sunxi
eebrah is now known as Guest26812
panda84kde has joined #linux-sunxi
<ssvb>
soldoKyn: the gles2 headers from sunxi-mali is more like a question for libv (I'm using the mesa headers in gentoo linux)
<ssvb>
soldoKyn: also it might be a problem with panda3d. Why does it want GL_COMPRESSED_RGB_S3TC_DXT1_EXT so much?
<oliv3r>
arokux: openbricks focuses on multimedia distribution creation
<oliv3r>
arokux: embedded*
<oliv3r>
arokux: so think of it like buildroot, on steroids? :)
<oliv3r>
ssvb: that's right, you are a gentoo user too :)
<ssvb>
oliv3r: and steev has just joined :)
<libv>
soldoKyn: why do you need this defined?
<libv>
soldoKyn: don't you think that this might be an extension that is simply not supported by the mali?
<oliv3r>
ssvb: what do you run on your a10 hardware? gentoo too?
<oliv3r>
ssvb: i tried to do the embedded build thing from within gentoo, but never got beyond that point
<ssvb>
oliv3r: what is embedded build?
<oliv3r>
deboostrap!
<ssvb>
oliv3r: all my armv7 devices are sharing the same nfs root, if I need more horsepower to build something, I just use 2x Cortex-A15 from Chromebook
<oliv3r>
i never wrote the rest of it; cause i never got it to work
<ssvb>
oliv3r: why would you want to build debian from the gentoo system running natively on arm?
<oliv3r>
gntoo crosscompile a deb root
<oliv3r>
so that I can run and compile on my server :) as i odn't have apixel
FDCX_ has joined #linux-sunxi
<oliv3r>
typing -1
\\Mr_C\\ has quit []
<arokux>
oliv3r, you'd better work on mainlining :p
<oliv3r>
i just sent a patch :p
Black_Horseman has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
<arokux>
oliv3r, which wasn't even tested by you
<arokux>
"It is possible to use the pre-compiled linaro toolchain, but why would you?" --> facepalm
\\Mr_C\\ has joined #linux-sunxi
<oliv3r>
arokux: huh? i sent a documentation patch
<arokux>
oliv3r, i'm just asking myself why....
<arokux>
oliv3r, why not use linaro, ppl done their work for you. so that you pick the compiler and actually create something new instead of repeating the work done
rellla2 has joined #linux-sunxi
rellla has quit [Ping timeout: 246 seconds]
<oliv3r>
'repeating'? i use gentoo, i have 'crossdev' and can compile my own compiler very easily; why would I want to isntall the linaro compiler?
<oliv3r>
then, most distro's should supply binary packages for various compilers
<mnemoc>
stop using gentoo, problem solved :p
<arokux>
oliv3r, to save time? :)
<oliv3r>
let your ditro's package manager handle the package
<oliv3r>
now if your distro does not, or the normal copmiler doesn't seem to work; linaro seems like a safe choice
<oliv3r>
mnemoc: but i have no problem :(
<oliv3r>
arokux: i have a fast pc :)
<oliv3r>
arokux: takes < 5 minutes
<oliv3r>
and it's not like I have to install a new compiler every day
<mnemoc>
true
<oliv3r>
but i have compilers installed for arm, avr, x86_64 and msp430
<arokux>
oliv3r, < 5 min to compile, but how much time to find out what to do?
<oliv3r>
arokux: i've been using gentoo for 10+ years now; i've gotten to be quite good at that :)
<oliv3r>
arokux: and it's also nice to know how your system works and what needs to be done to fix things :)
<oliv3r>
arokux: besides, aren't you an arch user?
<arokux>
oliv3r, yes, I am. and yes it is *very* nice to know how to fix things. and no, arch does not provide toolchains as standard packages. that is why i'm using linaro straight away. a year ago i've tried crosstool-ng, but you'd need to know to much for it to work.
<oliv3r>
arokux: so how is arch any different from gentoo?
<oliv3r>
other then that gentoo provides 'crossdev'
<arokux>
oliv3r, pacman -S <package name> -- that is all I need to know.
<oliv3r>
crossdev -t arm-pc-linux-gnueabi-
<oliv3r>
wait
<oliv3r>
done :)
<arokux>
hf?
<arokux>
etc..
<arokux>
i.e. what versions of gcc, glibc will it use
<oliv3r>
mripard: crap i cc-ed maxime-ripard instead maxime.ripard :( sorry
<oliv3r>
arokux: genuabi implied hardfloat, by default it installs the latest stable versions; you can override (if libc etc is build at all via the -s (stage) switch)
<oliv3r>
arokux: if you really want an softfloat toolchain, you can set that up aswell; if you want elibc, you can set that up too
Black_Horseman has quit [Quit: Zwi se logou mou!!!]
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
<arokux>
oliv3r, ok, gentoo is your personal preference, which is fine. me personally I just fail to see its advantages.
<oliv3r>
arokux: it's the same as arch
<oliv3r>
arch vs gentoo; who cares, one prefers the other :)
<oliv3r>
they both have their minor advantages/disadvantages
ganbold_ has joined #linux-sunxi
JohnDoe_71Rus has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org]
<oliv3r>
arokux: but to argue one distro over the other; when they are SO similar, is really silly :)
<oliv3r>
right, time to merge some dtv-tables
arokux1 has joined #linux-sunxi
<oliv3r>
wb :)
arokux has quit [Ping timeout: 245 seconds]
<soldoKyn>
ssvb: libv: yes, maybe it's something not supported by mali...I'll try the reason why panda3d need that defines and if it's a mandatory requirement! thanks for the suggestion!
<arokux1>
oliv3r, needed a reboot
<libv>
soldoKyn: could it be that panda3d is not openGLESv1/2 but full-on openGL?
<oliv3r>
arokux1: arch eh :p
<arokux1>
oliv3r, do you deploy a new kernel without reboot?
shineworld has joined #linux-sunxi
<oliv3r>
arokux1: of course not; i only tease you :)
<arokux1>
I'm just not satisfied with the speed of mainlining, that's all :p
Tsvetan has joined #linux-sunxi
Cubear has quit [Quit: Leaving]
<oliv3r>
arokux1: getting a patch accepted can take weeks, if not months
<oliv3r>
arokux1: especially when you are waiting for potential comments :)
<arokux1>
oliv3r, I fail to see patches :p
<oliv3r>
arokux1: i did mail them to linux-sunxi, but linux-sunxi is slow sometimes
<oliv3r>
i should configure sunxi mailserver
<oliv3r>
but I have to do the dtv-scan-tables first
<arokux1>
oliv3r, you mean ahci?
<mripard>
arokux1: if you're not happy about the speed of mainlining, you know that you can do something about it, right?
<mripard>
so far, only Turl and I have been working on this on a regular basis
<oliv3r>
mripard: i've sent the patch to the wrong ML
<oliv3r>
mripard: erm e-mail; i used maxime-ripard instead of maxime.ripard :(
<panda84kde>
libv: no, it' not. It's buildable for both OpenGL ES 1.x and OpenGL ES 2.0, and doesn't require plain OpenGL. I've been able to run it on a Pandaboard and on a i.MX6 (Vivante) with both OpenGL ES and OpenGL ES 2 backend
<mripard>
oliv3r: where did you send it to?
<arokux1>
mripard, I didn't mean to criticize you!
<oliv3r>
mripard: erm allt he maintainers of get_maintainers for Documentation/devicetree/bindings/interrupts/sunxi/sun4i-a10.txt
<oliv3r>
mripard: so the lkml, the docs@vger, devicetree@vger +the names listed there
<mripard>
oliv3r: hmmm, ok, I'm not subscribed to lkml
<oliv3r>
mripard: i'm pretty sure i sent it to linux-sunxi aswell; but haven't seen it there
<mripard>
ah
<mripard>
devicetree
<mripard>
when such thing is happening, you usually do a resend
<arokux1>
mripard, I also do my best to add something to mainline (usb host) maybe it seems to be a little but I needed lots of time, as I was a complete noob and still is.
<mripard>
you can also use --cc-cmd in git send-email to avoid this :)
<libv>
panda84kde: ok
<oliv3r>
mripard: looking up cc-cmd
geecko has quit [Quit: Quitte]
<mripard>
arokux1: yeah, I know, and it's great that you got results on this
<mripard>
but look at how much time it took you, when you mostly had to write some glue to a working driver
<mripard>
most of the time, we have to remake the driver from scratch
<oliv3r>
mripard: is it ok to send a [RFC] instead of a patch for pre-first version? when you know the patch will not be accepted in its current form?
<mripard>
it depends
<oliv3r>
mripard: well the AHCI patch has some iffy things, and I'd ask some input from the scsi/ahci guys first
<mripard>
usually, RFC are for stuff that are subject to controversy, and you want to gather everyone's opinion
<oliv3r>
ok then it's a patch
<libv>
i don't think we even install a gl2ext.h, no?
<libv>
oh, we do
<libv>
perhaps we should pull in the latest version, and be done
<libv>
copy that to your /usr/include/GLES2/ directory
<libv>
i will push that out with other pending sunxi-mali changes
popolon has joined #linux-sunxi
<oliv3r>
mripard: how are things anyway?
<soldoKyn>
libv: yes, that should be the problem...telling panda3d to build against openGLES and not against openGLES2 the error disappear!
<mripard>
oliv3r: early, I'm in new orleans this week
<libv>
soldoKyn: opengles2 would be preferred though
<mripard>
and it's great so far :)
<libv>
soldoKyn: the hw is gles2 and emulates gles1
<mripard>
oliv3r: wow, you got your first NAK :)
<oliv3r>
mripard: lol allready? I haven't even checked yet
<oliv3r>
mripard: he doesn't want your stuff there either!
<oliv3r>
mripard: how did you get it passed him the first time ;)
<oliv3r>
mripard: new orleans looked like it used to be a really nice place
<panda84kde>
libv: what do you mean by "emulates gles1"? Do you mean that gles1 calls get software-translated to something equivalent in gles2 which in turn gets executed?
<arokux1>
oliv3r, to which ml have you sent your patch?
<oliv3r>
mripard: good food, good music
<oliv3r>
arokux1: i responded to mripard earlier :) he asked the same
<mripard>
oliv3r: yeah, I ate gator before going to a jazz club yesterday evening
<mripard>
pretty unusual :)
<arokux1>
oliv3r, found it.
<oliv3r>
mripard: gator nuggets are pretty good; my father lives in FLorida, so I (used to) visit him very regularly; florida is known for its gators :)
<libv>
panda84kde: yes
<arokux1>
oliv3r, you know why we fail to see you patch on linux-sunxi?
<libv>
panda84kde: check out my ioquake3 tree on github
<arokux1>
oliv3r, because you've sent it to: linux-sunxi@google-groups.com, instead of linux-sunxi@googlegroups.com
<mripard>
oliv3r: but yeah, it's a great city
<oliv3r>
arokux1: bah; soon it will be linux-sunxi.org :)
<mripard>
one of the few in the US where being french is a real advantage :)
<oliv3r>
mripard: haha, yeah a lot of french spoken people there strangly
<mripard>
and I saw stefan there
<oliv3r>
whats' the event you are visiting?
<oliv3r>
and how's the city recovering? After the flooding and storms?
<mripard>
and Ian Campbell, the guy that you replied to over SATA recently
<oliv3r>
I know they got in dutch dike workers to get their water problems sorted at some point
<mripard>
I'm at Linux Plumbers
<oliv3r>
is that the same Ian Campbell @citrix.com?
<oliv3r>
as he used a different e-mail on sunxi
<mripard>
well, I've mostly stayed downtown, but it's seems to be mostly over
<oliv3r>
that's pretty cool, are you going to the kernel mini summit in edinburgh too?
<mripard>
downtown has no traces of the flood anymore
<mripard>
and yes, that's the guy from citrix
<oliv3r>
that's good; it looked pretty brutal not too long ago
<mripard>
and yes, I'm going to the ARM kernel summit
<oliv3r>
busy man
<oliv3r>
:)
<oliv3r>
all payed for by free-electrons?
<arokux1>
do you know who Matt Porter (mdp) is?
<mripard>
the arm kernel summit yes
<oliv3r>
Matt Porter is the guy who said was gonna do DMA if i remember correctly
<mripard>
plumbers, partly only
<oliv3r>
he's been in here though
<oliv3r>
mripard: even so, having to pay a part for it is probably worth it
<mripard>
oliv3r: yep, that's why I'm here :)
<arokux1>
oliv3r, yes, I'm asking because of DMA... any idea if he is working on it?
<panda84kde>
libv: mmm... thank you so much. Really useful information. That (emulating GLES1 over GLES2) could well be the case for some other embedded chipsets out there, right?
<oliv3r>
arokux1: said to be the case yeah
* panda84kde
start suspecting about poor GLES1 performances on Vivante
<oliv3r>
arokux1: did you send your patch to the ML's allready?
<arokux1>
oliv3r, not yet. it depend on the patches not yet sent to any ML. I've decided to sent two patches that add USB clocks and see the reaction
<mripard>
arokux1: that's something I've began to work on
<mripard>
the easiest path to get DMA is to do SPI, and then DMA, I have a SPI driver that should be not that far from working
<mripard>
next step is dmaengine and spi using DMA
<libv>
panda84kde: it is the case for all _current_ mobile GPUs
<libv>
panda84kde: descend into code/egl/
<arokux1>
mripard, I was just thinking what to do next. storage is important, I think. so needs to be done first. how do you think?
<libv>
it includes gl command capturing, gles2 and limare code (the latter 2 are not complete, but the visual defects are only limited - limare is better than gles2 though)
<oliv3r>
arokux1: test my ahci patch and we hav estorage :p
<arokux1>
oliv3r, test it first by yourself, report it working and i'll test it by myself on the same evening :p
* panda84kde
cannot thank libv enough
<oliv3r>
arokux1: haha, i have no idea what the status is, my hardware is packed away atm
<oliv3r>
arokux1: it may not even work :)
<oliv3r>
though it should in theory
<oliv3r>
and i need to cherry-pick turls stuff
<arokux1>
oliv3r, from what i've learned there is a long way from theory to "works"
<oliv3r>
mripard: anyway, make sure you have a damned good time t here in new orleans :) it's deffinatly a place well worth visiting
<oliv3r>
arokux1: we'll see :)
<oliv3r>
gonna fix the dts now so it can actually be tested :)
ykchavan has joined #linux-sunxi
deasy has joined #linux-sunxi
<arokux1>
wens, ping
<arokux1>
Tsvetan, hi. there seems to be no license for opensource for Eagle, only for hobbyists, it prohibits any commercial usage. or have I missed something?
<oliv3r>
arokux1: maybe Tsvetan only says that you can use it cheapily as hobbiest, and he bought a commercial license himself?
<arokux1>
<Tsvetan> Eagle non for profit licensee is EUR 100
<arokux1>
the license for hobbyists is 160 EUR
<arokux1>
oliv3r, that seems to be the case.
<arokux1>
oliv3r, do you know if AW_AXP18 or AW_AXP19 are actually used by some hardware?
<oliv3r>
well there's also a a free version
<oliv3r>
arokux1: nope
<arokux1>
oliv3r, sun[457]i use AW_AXP20
boss has joined #linux-sunxi
<arokux1>
oliv3r, nope - you do not know? or they are not used and can be dropped?
<boss>
I have cuibe board v1, I need to transfer 400MB/s data using parallel port to memory, how can I do it, I found no parallel port
<boss>
Even serial port also OK
<tgaz>
arokux1: aren't those separate PM ICs that could be combined with any SoC?
<buZz>
3200Mbit serial IO, good luck ;)
<atsampson>
boss: can you describe your problem in more detail? (what are you actually trying to build?)
<boss>
@ Buzz, would you mean SATA
<arokux1>
tgaz, no idea, that is why i'm asking. do you know some hardware that uses AXP18 or 19?
<tgaz>
arokux1: no. they are all PM ICs. but I don't know if they are simply older versions, or specialized for different applications.
tinti has joined #linux-sunxi
<boss>
I need to transfer 48 bit data to memory, at 100MB/s
<boss>
which means 400 MB/s
shineworld has quit [Quit: Leaving]
<tgaz>
boss: so you basically want to circumvent the CPU, and just stream into RAM from external?
<boss>
I have device wich generates the data 48 bit at 100 MHz
<boss>
yes
<boss>
tgaz, you are rite
<boss>
I need to stream to RAM
<boss>
from external
<tgaz>
looking at the A20 SoC (on cubieboard2), it doesn't have any slave parallel port, and the serial port is rated at max 150 Mbps.
<tgaz>
i would assume the A10 on cubieboard1 isn't better
<Tsvetan>
arokux we have full licensee of Eagle
<tgaz>
but right. SATA is rated higher. ;)
<boss>
what about the GPIO?
<tgaz>
but there are 4 SPI channels...
<tgaz>
boss: at those speeds, i would *guess* that's too slow. you'll want something that can use DMA
<boss>
yes
<boss>
or any memory map I/O
<boss>
I/O mapped transfer is slow
<boss>
Is it possible to make memory map I/O
<tgaz>
theoretically... if the 4 SPI channels can use 4 separate DMA channels, you could reach 600 Mbps, which is still not nearly fast enough.
woprr_ has quit [Quit: Ex-Chat]
<boss>
yes
<tgaz>
hmm... SATA at 3 Gbps is still barely 400 MBps. I give up. :(
<boss>
But interfacing to SATA port will be complex, I guess
<tgaz>
not as straight forward as spi. but i've never looked at the protocol.
<mripard>
arokux1: you can work on the PMICs if you'd like
<mripard>
that would really be useful
<mripard>
and it's not really complicated, just quite complex
<boss>
If any chip found from parallel to SATA, then its OK, but SATA protocol is not so easy I guess
<arokux1>
mripard, AXP209? I could add very basic support, the one needed for a headless server as I do not have tablet with battery etc.
<mripard>
arokux1: yep
<mripard>
well, there's quite a lot things there already
<mripard>
interrupts, regulators, gpios, etc.
<arokux1>
mripard, good.
<arokux1>
mripard, btw, with working USB you could have your rootfs on a USB stick
<arokux1>
from chinese code for AXP: "Reverse engineered partly from Platformx drivers"
<oliv3r>
boss: 48 bit at 100 MHz would be 4.800 megabit per second, so you need about half a gigabit
<oliv3r>
boss: which is probably even above the memory speed itself for some boards
<oliv3r>
boss: I doubt that the allwinner chips are fast enough for that
<wens>
arokux1: pong
<arokux1>
wens, how's your work on AXP?
<wens>
arokux1: i don't remember saying i was going to work on AXP...
<arokux1>
wens, no? but you said you'll clean it up? you even edited its wiki page :p
<boss>
@oliv3r, how much faster its possible? Cuibe have 480Mhz Ram with 32 bit
<wens>
arokux1: i said i'd clean up the wiki page
<arokux1>
wens, ah.. I've thought the sunxi-3.4
<arokux1>
wens, are you working on smth for mainline?
<boss>
Is it possible to use the SATA
<boss>
using simplier commands
<boss>
with reducing the bits
<n01>
boss: what is giving you 400MB/s of data?
<boss>
ADC
<oliv3r>
boss: sata on the a10 will not satisfy that throughput ii'm sure
<boss>
external ADC
<n01>
wow ... hell of ADC ... 48bit???
<oliv3r>
boss: unless you directly talk to the ram or some rambus device, i don't think anything can talk that fast
<boss>
no
<oliv3r>
48 100 MHz adc, wow; do you have a chip number so i can look up the datasheet?
<boss>
10 bit
<arokux1>
mripard, how dangerous for a board could it be to mess up with AXP?
<boss>
4 channel
<oliv3r>
arokux1: if you configure it wrongly, it can fry the chip :)
<oliv3r>
arokux1: e.g. max voltage to 1.8v device
<arokux1>
oliv3r, ok. is it configured already somehow by the mainline?
<wens>
arokux1: not currently
<n01>
how fast are ADCs in cb?
<boss>
The 10 bits are coming out after some interval
<boss>
with clock delay
<mnemoc>
n01: it's on the A10 user manual
<n01>
so you have 4 channel 12 bit @ 100MHz each
<boss>
Yes, You can think
<tgaz>
boss: it seems rather advanced/pro. how come you don't have an fpga in there and dual port ram?
<tgaz>
boss: i think using some programmable logic is the standard way of hitting those speeds/timings.
<boss>
yes
<boss>
SO, SATA is much speedy
<tgaz>
and you seem to have pretty high demands, so i was just wondering how come you're not doing what others are doing. ;)
<boss>
in A10
<boss>
But I am worried about protocol
<boss>
I guess SATA protocol is not simplier
<boss>
@tgaz, the USB is 480Mbps
<n01>
boss: it is really weird gather data in output from ADC with SATA
<tgaz>
yes, but you said you needed 8x that
<tgaz>
n01: but creative ;)
<boss>
:P
<tgaz>
another question is what the overhead of the SATA protocol is
<tgaz>
and how good the drivers are etc
<boss>
I am reading the protocol
<tgaz>
s/are/will be/
<n01>
boss: can't you take an external ADC->USB?
<boss>
Till now not sure
<tgaz>
boss: how are you going to trigger the ADCs?
<tgaz>
ie handle the timing there
<n01>
I guess continuous conversion
<boss>
ADC circuit is seperate
<boss>
It have seperate clock
<tgaz>
ok
<n01>
btw what kind of signals are with bw 50MHz?
<n01>
just corious
<atiti>
hm
<atiti>
does anyone know if its possible to set some overscan parameters through kernel cmd line ?
<atiti>
i have a pretty crappy monitor where the overscan is unbearable
<boss>
The data that transfer over 50Mhz carrier
<boss>
the signal carrier freuency is unknown
<boss>
So, I cann't filter
mdp has quit [Ping timeout: 245 seconds]
<oliv3r>
hometime :)
<n01>
oliv3r: c'mon .. it's early
<tgaz>
more time to do awesome sunxi stuff at home
<tgaz>
i just bought the cb2 the other week, and my wife started complaining about my absence yesterday. :/
<n01>
tgaz: I can relate *_(
mdp has joined #linux-sunxi
<atsampson>
boss: is it really unknown to +/- 50MHz? you can't mix down to a more convenient lower frequency (or even better use a tuneable external oscillator)?
<atsampson>
are you planning on doing software downconversion on the A10? (it's probably not fast enough for that...)
<boss>
yes
<boss>
What about the camera port
<n01>
???
<boss>
SATA will be complex to handle I think
<boss>
using the camera port as data receiver
<boss>
I think Camera port receives 24x54 MBPS
<boss>
sorry
<boss>
24x54 mbps
<boss>
and what about the CSI0
<boss>
CSI1 all signals are documented
<boss>
but CSI 0 is not
<boss>
is it possible to use the CSI0/CSI1 together in same cycle
<boss>
like CSI 0 in raising edge and CSI1 in falling edge
ykchavan has quit [Quit: Leaving]
<tgaz>
ah, i missed that one
<tgaz>
boss: have you found any numbers on how fast it can go?
<tgaz>
in bps
<boss>
its 54mhz
<boss>
24 bit
<boss>
parallel
<tgaz>
the csi0? i only see 16 bits in the a20 user manual
<tgaz>
in the csi0 block diagram
<boss>
I am using A10
<boss>
not A20
<tgaz>
it's different?
<boss>
A10 have CSI0 8bit
<boss>
CSI1 24 bit
<boss>
is it possible to use the CSI0/CSI1 together in same cycle
<boss>
[20:48] <boss> like CSI 0 in raising edge and CSI1 in falling edge
<boss>
is it possible to use the CSI0/CSI1 together in same cycle
<boss>
[20:48] <boss> like CSI 0 in raising edge and CSI1 in falling edge
<tgaz>
but isn't csi1 serial?
<boss>
no
<boss>
I think parallel
<tgaz>
yep, looking at pinouts, there definitely many pins
<boss>
and multiplex
<tgaz>
ah, the block diagram had the CS Data lines at the bottom. didn't see them
<boss>
Also it have LCD1 port
<boss>
are the port can be used as Input port?
<tgaz>
csi1 specifically states it has pass-through.
<arokux2>
deasy, that'd be great, will you really do it?
<deasy>
yes
<deasy>
if it's in english :p
<arokux2>
deasy, there is english data sheet, but for AXP202
<arokux2>
deasy, but very little is different
<deasy>
the chinese DS is in english lulz
<deasy>
no no just some parts
<deasy>
bad bad bad!
hramrach has joined #linux-sunxi
<popolon>
I can help for chinese translation :)
<popolon>
(if needed)
<arokux2>
popolon, thank, not yet :)
<wingrime>
popolon: axp209 only cn manual
<buZz>
deasy: hey schatje
<arokux2>
hno, do you know whether AXP18 and AXP19 are used? I want to drop them.
<popolon>
ok
<popolon>
come back in about one hour
<wingrime>
arokux2: I will NACK any patch of it
popolon has quit [Quit: Quitte]
<arokux2>
wingrime, give me examples of their usage then. NACK without a reason is not a NACK :p
<deasy>
REG 84H i don't know how to do the divide
<deasy>
check in the manual
<wingrime>
mnemoc: ^
<mripard>
wingrime: ?
<arokux2>
deasy, I also do not know how to do this. leave if for now
<arokux2>
mripard, do you know if AXP18 or AXP19 are actually used by known hardware?
<wingrime>
mripard: you have any movement with SDIO, or still working on SMP ?
boss has joined #linux-sunxi
<mripard>
arokux2: I don't think so
<mripard>
wingrime: leviathanch is working on the MMC
<mripard>
and SMP is being done by someone else as well.
<wingrime>
mripard: what are you currently working on?
<oliv3r>
boss: while backreading; i'm thinking more that your 48bit dac doesn't need to process 480 MB/s; the chip probably runs at 100 MHz, but in those 100 MHz it samples all 4 adc's in turn, not simultaniously
<mripard>
wingrime: HS timer and SPI
<wingrime>
mripard: nice
<mripard>
and I have some patches for the A31 SMP support that I still have to test
<atsampson>
arokux2: there's some nice Chinglish in that page -- "lineal" should be "linear", "duck regulator" should be "buck regulator"
<arokux2>
atsampson, yes, this is machine translation. please improve it, thanks.
boss has quit [Ping timeout: 264 seconds]
<atsampson>
arokux2: fixed
<arokux2>
atsampson, thanks!
<oliv3r>
Turl: ? i haven't done anything yet except install; what are you talkin' about exactly?
<Turl>
oliv3r: ?
<oliv3r>
Turl: aliasfile
<Turl>
oliv3r: ah yeah
<Turl>
oliv3r: there you can add aliases to email addresses
<oliv3r>
Turl: but i haven't set anything up yet :)
<Turl>
oliv3r: so you git send-email .... --to maxime --cc linux-arm
<oliv3r>
OHhhhhh
<arokux2>
Turl, do you know if AXP18 or AXP19 are actually used by known hardware?
<oliv3r>
i thought you where talking postfix
<Turl>
arokux2: I think A10S have AXP192 or sth
<steev>
ssvb: what? just joined what?
<Turl>
and A13 have some other AXP
<arokux2>
Turl, are they actually supported by sunxi-3.4?
<Turl>
arokux2: or maybe it's 152 I don't know yet
<Turl>
arokux2: yeah they are
<ssvb>
steev: the community of allwinner hardware users, also using gentoo on it :)
<steev>
ssvb: oh, yeah :)
<arokux2>
Turl, what are their defconfigs?
<Turl>
arokux2: sun5i
<steev>
i have the ly-f1, and i also have a mid970, but those both still run android
<oliv3r>
arokux2: i strongly agree that it doesn't make much sense to drop axp* from axp209 just because in 3.4
<arokux2>
Turl, defconfig for sun5i uses axp209
<arokux2>
oliv3r, please, not an opinion but arguments.
<Turl>
AXP152 should be enabled for sun5i then, I believe hansg created it so it could be used simultaneously with 209
<oliv3r>
arokux2: in 3.4 it's pure reference material.
<arokux2>
oliv3r, reference is original code drop.
<oliv3r>
arokux2: no.
<oliv3r>
arokux2: go read the discussion on the ML we had about wingrime dropping dead code from the disp driver
<oliv3r>
it will get nacked
<arokux2>
oliv3r, this is because you did not know what this code does. my case is different, if there is no know hardware with AXP18x and AXP19x, then they should be dropped, same as it was with sun3i.
<oliv3r>
arokux2: no, we knew what the code possibly did
<arokux2>
oliv3r, *possibly*.
<oliv3r>
sun3i was dropped because it was getting in the way and causing bitrot as nobody has hardware
<oliv3r>
look at any of the IP cores we have, most is very probably from the sun3i and likly compatible with sun3i
<oliv3r>
but now we have double drivers, and we can't unify sun3i -> sunxi
<oliv3r>
because we can't test the cod
<oliv3r>
so you either merge it in and 'guess' if its right
<oliv3r>
or you leave it as a duplicate
<oliv3r>
IF someone is interested in the work and obtains hardware for it, it's quite likly most drivers will work 'as is'
<arokux2>
oliv3r, there is still no hardware with sun3i, so the decision was right.
<oliv3r>
as it's all one big family
<oliv3r>
there's a LOT of sun3i hardware
<oliv3r>
but none of the devs have it
<oliv3r>
a lot of cheap tablets run sun3i
<steev>
if none of the devs have it, then it doesn't exist
<oliv3r>
tablets == e-readers
<oliv3r>
in any case, it was getting in the way and was causing duplication
<arokux2>
oliv3r, it wasn't asked about on ml, it wasn't asked about here. so...?
<oliv3r>
both on the eoma and on the 'motherboard'
popolon has joined #linux-sunxi
<oliv3r>
arokux2: now if your going to bring the axp209 to mainline; you don't have to add the other chips drivers till we have hardware for it and can test it obviuosly
<oliv3r>
but say someone gets the axp192 and wants to merge it into the axp209 driver in mainline, it's good to have a reference point in 3.4
<wingrime>
oliv3r: I will NACK axp remove patches
<arokux2>
oliv3r, I assumed that. I just wanted to get familiar with the code by cleaning it. sunxi-3.4 is developed and improved. the reference is original code.
<oliv3r>
wingrime: the idea is not horribly bad; but our 3.4 kernel is also reference for when moving code to mainline imo
<atsampson>
the EOMA-68 A10 board has an AXP209; the "KDE tablet" EOMA design also has an AXP209
<oliv3r>
arokux2: but people will hardly look at the reference
<oliv3r>
arokux2: people will look at 3.4
<oliv3r>
atsampson: i thought so yeah
<arokux2>
oliv3r, if they'd like to add AXP18x what else can they do? of course they will look.
<wingrime>
oliv3r: better move used hw to mainline
<oliv3r>
arokux2: they will look at stage/3.4 probably mostly
<wingrime>
oliv3r: but axp work, and we should not broke it
<oliv3r>
axp152 for 2 Euro
<arokux2>
I want to kill AXP18 and AXP19
<arokux2>
Turl, ^
<oliv3r>
wingrime: yeah but when taking it to mainline, sure only supoprt what you have and is known to work; besides, 20 bux says that there should be 1 axp driver that supports all their chips without too much effort
<arokux2>
wingrime, by removing *unused* code you cannot break things.
<wingrime>
oliv3r: thats imposible do in one driver
<oliv3r>
arokux2: as someone just told you axp19 = axp192 which is used on a13 hardware a lot
<wingrime>
arokux2: that not dead code
<oliv3r>
arokux2: so you are breaking sun5i platforms
<oliv3r>
arokux2: and again; ti's reference code for those who port it to mainline :)
<oliv3r>
wingrime: it's dead to arokux2 :)
<wingrime>
oliv3r: there is more strange dead code in disp in bak_ folder
<wingrime>
oliv3r: that do SPI and i2c by soft bit-bang
<arokux2>
A10S-AXP152
<arokux2>
not 192 ;)
<wingrime>
arokux2: better check our fex repo
<wingrime>
arokux2: thats have axp name
<oliv3r>
wingrime: isn't that the 'backup' folder?
<oliv3r>
wingrime: ohh backlight will go away in mainline
mcbrick has quit [Ping timeout: 260 seconds]
<oliv3r>
wingrime: backlight is handled via dts and pwm driver
<wingrime>
arokux2: something around
<oliv3r>
wingrime: so I think it's safe to just ignore it in mainline
<wingrime>
oliv3r: hdmi have some stranges I noticed
<wingrime>
oliv3r: it ignore (according hsang) pll3 settings and have own pll regs
<wingrime>
oliv3r: thats realy strage
<wingrime>
Turl: ping
<Turl>
wingrime: pong
iamfrankenstein has joined #linux-sunxi
<wingrime>
Turl: what do you think about HDMI IP stranges
ZetaNeta has joined #linux-sunxi
<arokux2>
wingrime, grep shows there is nothing about AXP18x or AXP19x in sunxi-boards.
notmart has quit [Quit: notmart terminated!]
<wingrime>
arokux2: there is way for choose
<wingrime>
arokux2: must be
<arokux2>
wingrime, i'm saying no board is using them.
<Turl>
wingrime: what about it?
ganbold_ has quit [Remote host closed the connection]
<Turl>
arokux2: they're chosen by i2c address
<oliv3r>
arokux2: sunxi-boards is VERY limited
<Turl>
twi_something property
<atsampson>
arokux2: there are so many Allwinner designs out there that I think that's highly unlikely -- sunxi-boards doesn't even cover all the devices I've got ;)
<wingrime>
Turl: ...
<oliv3r>
atsampson: so why haven't you submitted them ;)
<wingrime>
Turl: hdmi have own pll settings reg
<Turl>
wingrime: really?
<wingrime>
Turl: accoring hasng note
<wingrime>
Turl: it ignore pll3 settings
<arokux2>
atsampson, yes, submit them so that we know if AXP18x and AXP19x are used :p
<wingrime>
Turl: also you can see that nearly correct
<wingrime>
Turl: in 20 manual
<Turl>
wingrime: pll3 & pll7 should be them?
<wingrime>
Turl: yes
<wingrime>
Turl: whait I give you link
iamfrankenstein has quit [Ping timeout: 268 seconds]
<wingrime>
Turl: page 471 in a20 man
<oliv3r>
Turl: !! ian cambel tested ahci
<oliv3r>
Turl: while he didn't have a power source
<wingrime>
oliv3r: please take a look at page 471 in a20 man
<wingrime>
oliv3r: thats have fully documented pll
<wingrime>
oliv3r: with all but fields
<wingrime>
*bit
<Turl>
wingrime: 1 bit CKIN_SEL
<Turl>
wingrime: documentation really awesome and complete :(
<mripard>
Turl: btw, why didn't you submitted yet your clock patches?
<mripard>
the one used by arokux2 and wingrime ?
<mripard>
is there some obvious drawback ?
<Turl>
mripard: it has become a huge pile of patches
<atsampson>
oliv3r: 'cause I'm a bit reluctant to submit patches for .fex files where I don't actually have the physical device...
<wingrime>
mripard: we can test hdmi engine patches separately , it have test-colorbars testmode
<Turl>
mripard: I can try to send the basics today if you're in real need of getting them in ASAP
<atsampson>
oliv3r: I went searching last week for config files for a10/a13/etc. that had Allwinner's comments in, having spotted they were there in the most recent chips
<wingrime>
oliv3r: do you looked at bitfields for pll?
<atsampson>
oliv3r: but it appears that some vendors include the .fex in their Android images...
<wingrime>
Turl: how we will deal in more one strange pll in HDMI regs, thats f***g awesome s**t
<oliv3r>
atsampson: i thought you said you had those devices :p
<atsampson>
no, just the config files
<Turl>
wingrime: disp IP can register as clock provider I think
<Turl>
there is work to make clk unregister work so you can have it as module
<wingrime>
Turl: not sure thats acceptable for mainline
<oliv3r>
Turl: i kinda need your patches too
<wingrime>
Turl: but looks pll3/7 ignored
ykchavan has quit [Quit: Leaving]
<Turl>
oliv3r: I'll start pushing them then
<oliv3r>
well pll7
<Turl>
oliv3r: I still haven't heard from mturquette on my composite patch I sent some time ago
<Turl>
wingrime: you can do easy test, get sunxi device working with screen then zero pll3/7 register
<wingrime>
Turl: nice idea
<arokux2>
U-Boot code is an utter shit.
<wingrime>
Turl: clock settings must be configured for current video mode
<oliv3r>
arokux2: whut
<arokux2>
oliv3r, ?
<oliv3r>
i wrote some u-boot code
<oliv3r>
what is shit
<arokux2>
oliv3r, u-boot code. its anarchy.
<arokux2>
oliv3r, everybody hacks there stuff in
<oliv3r>
arokux2: what are you looking at that makes you say this?
<arokux2>
oliv3r, git grep "define BIT(x)"
<oliv3r>
ah yeah each arch does it themselves
<oliv3r>
either a) they haven't merged the latest kernel libs yet
<oliv3r>
or they belive each arch needs their own code?
<oliv3r>
or nobody put in the work yet :)
<arokux2>
oliv3r, no, nobody is guarding u-boot code as kernel maintainers do.
<drachensun>
anyway know a good way to test 3d acceleration without X?
<drachensun>
I'm trying es2_info but I'm not really sure if it will work outside of X or if I'm calling it right
<techn_>
drachensun: if you have X running, you must have "export DISPLAY=:0.0" set in environment
<techn_>
or was it .0 or 0
<drachensun>
no X, I'm working with ubuntu touch
<drachensun>
I think I figured it out
<drachensun>
es2gears_screen seems to try and run
<libv>
drachensun: sunxi-mali/test should work
<libv>
drachensun: as should the limare tests, the egl ones
<drachensun>
libv: I'm unfortunately working with a powervr on the A31, so I assume they don't apply, or are they generic?
<libv>
ah, no the framebuffer stuff is not generic
<wingrime>
libv: code in bak_lcd is used?
<wingrime>
libv: it was ever builded?
<libv>
wingrime: feel free to not update that code, but do not simply delete it for now
<arokux2>
drachensun, have you got some interesting documentation from Allwinner? :)
<wingrime>
libv: that code do, SPI using bit-bang....
<oliv3r>
arokux2: of cours ethere are; there's reviews and everyhting; but much MUCH smaller team
<wingrime>
libv: also you can explain it to arokux2 same thing, he want drop some axp's
Black_Horseman has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
<drachensun>
arokux2: I don't think I have anything thats not generally available
<wingrime>
libv: also, HDMI HW have own hardware testcolorbars, so It can be tested with out FE and BE , thats nice for mainline
<arokux2>
wingrime, the difference between you and me is that I know what my code does. you do not know.
<drachensun>
wingrime: The axp22 seems to work with all he battery monitor software I had tried without any modification
<arokux2>
drachensun, do you know about some future SoCs by allwinner?
<drachensun>
nothing more than anyone else
atiti has quit [Ping timeout: 260 seconds]
<wingrime>
arokux2: you not writed axp
<arokux2>
wingrime, I want to kill axp18 and axp19 and I do *not* want to kill some code I have no idea of.
<drachensun>
I was hoping to go meet them at the show in January and try to build a more personal relationship, but the way the current crowd funder is going I'll likely be out of the tablet business by then
<arokux2>
drachensun, because of this powervr shit everybody is switching to RK for quadcore.
<steev>
which powervr shit?
<steev>
s/which/what/
<arokux2>
powervr gpu in A31
<drachensun>
arokux2: I dont know, all my vendors offering anything decent were in the process of switching to A31 from what I could see
<steev>
oh
<wingrime>
drachensun: powervr internals so bad?
<steev>
they aren't great
<drachensun>
wingrime: I haven't had to mess with it much yet, under nemo everything worked really easily
<drachensun>
under ubuntu its been a lot more of a challenge
<arokux2>
drachensun, i'm talking about more or less open boards like radxa, olimex.
<drachensun>
but thats the big name everyone wants to hear
<wingrime>
drachensun: nemo?
<drachensun>
wingrime: nemo-mobile based on mer
<drachensun>
arokux2: unfortunately there is a lot more to designing a whole tablet than an dev board
<wingrime>
drachensun: oh, never saw something like than
<drachensun>
so the market hasn't shown enough demand to consider a bottom up new design
<drachensun>
thought ideally that is what I would be offereing
<wingrime>
drachensun: how much registers are you indentifed from powervr?
<drachensun>
offering
<wingrime>
drachensun: or, commands
<arokux2>
drachensun, so you are really designing a tablet by yourself? guy from Olimex told us you buy it in china
<drachensun>
wingrime: I'm not trying to reverse engineer it, I'm trying to get it working with libhybris and the Android libraries
<wingrime>
drachensun: sad
<drachensun>
arokux2: No, I'm buying from a vendor and reflashing them
<drachensun>
arokux2: My original plan was to use the funding from the first device to offer a custom device, but it wasn't successful enough to fund anything close to that
geecko has joined #linux-sunxi
<drachensun>
I was hoping a quad core device would bring more interest
<drachensun>
it has, our numbers on contributors/visitors are way better than before
<drachensun>
but the press has essential abandoned us this time
<drachensun>
last time we had way more articles
<arokux2>
drachensun, yes, press is totally silent about you.
<drachensun>
so I'm hoping a name like ubuntu touch will get their attention
<arokux2>
drachensun, but there are just so many tablets out there, you know
<drachensun>
as far as I know I
<arokux2>
drachensun, and so few Linux geeks...
<drachensun>
I'm the only place you can buy a tablet that boots Linux without spending a lot of your own hours doing it
<drachensun>
yeah, that is the bigger thing
<drachensun>
I think the commercial market is where the real demand is
<drachensun>
I had two big contracts ready before the A10 ceased to be offered, I'm hoping to switch them over to the A31
<drachensun>
but without big enough orders I just can't get it
<drachensun>
well
<drachensun>
I can't get it from the only vendor I know who shares their source
<drachensun>
and never leaves you hanging with some touch screen chip that will never work under Linux
<drachensun>
oh well, I gotta get back to it, thanks for the tips about gl testing
<wingrime>
drachensun: thay want nice hw, sw that will work nice together
<brain__>
hey everyone
<wingrime>
drachensun: something with cool features , that china get give then
<wingrime>
drachensun: not linux
<wingrime>
drachensun: hw-sw complex
<drachensun>
wingrime: Not everyone wants to spend 20 hours with their new toy before it actually boots Linux like they bought it for, I think I offer a pretty good value. Also another reseller of my same model is in the EU they cost 20% more than my crowdfunded prices and are still very popular
woprr has joined #linux-sunxi
<wingrime>
drachensun: do you found some aw-based notebooks?
<drachensun>
and all they offer is Android
<brain__>
On my A10 CM9 build, I get sound only through headset. were should I be looking for the routing? I always hear of android audio routing but do not know were to look :(
<drachensun>
wingrime: the netbook market is dying, most suppliers aren't upgrading their netbook designs to the quad core chips, they are just forgetting about it.
<brain__>
Google was no help, only got tons of posts of audio routing being fixed..never were to look for the issue.
<wingrime>
drachensun: I taking about you, market still exsist when demand present
<wingrime>
drachensun: geeks want cheap, netbooks with linux,
<brain__>
yep :)
<drachensun>
wingrime: They way I see it, they have some really good, battery/keyboard/case attachments, put that with my tablet and you have a netbook and if you upgrade later your keyboard will work with that too
<wingrime>
drachensun: not everyone want android
<drachensun>
and if you just want a tablet someday, you have that too
<drachensun>
better than a netbook
<drachensun>
I agree, but see my current product appeals to people who just want to experiment with Linux, or just want Linux sometimes as it easily dual boots
<wingrime>
drachensun: yes, but it need hw connectors for kbd + internal batt, not usb
<wingrime>
drachensun: cables looks ugly
<drachensun>
they are bluetooth now
<brain__>
I have A10 tablet with serial console mod, cheap case with keyboard and a small usb hub for a tiny mouse and wifi dongle when needed. its great
<wingrime>
drachensun: internal bat for 18 hours for example?
<drachensun>
a linux netbook, while strongly favored by the linux crowd, wont have a larger market appeal, and the market is small already
<wingrime>
brain__: yes, but same , and ready to connect
<drachensun>
I'll never be able to compete with the chromebooks on price either
<drachensun>
wingrime: I dont think I have seen one with that kind of battery life, in a netbook or any other form with an Allwinner chip
<wingrime>
drachensun: I not like chrome-os
<drachensun>
yeah, but has someone figured out how to put Linux on those yet?
<drachensun>
seems like it shouldn't be too hard
<wingrime>
drachensun: like add you data belows to nsa
<brain__>
the biggest drain on allwinner is the wireless chipset they chose.
<wingrime>
*all
<brain__>
on tablets that is
<wingrime>
err
<wingrime>
belongs
<drachensun>
wingrime: Yeah, a big reason I like an OS you can build yourself
<wingrime>
drachensun: no internal data, only payed 3g, thay are crazy....
<drachensun>
wingrime: I hadn't looked at the latest one... yeah that sucks
<brain__>
my tablet last longer running Linaro images with external dongle and keyboard than it does in android hands down.
<oliv3r>
drachensun: people aren't looking forward to a31/powerVR
<drachensun>
they pay people too much at google, their staff dont get why you wouldn't want a $100 a month wireless plan just to do basic stuff
<Turl>
drachensun: chromebooks can run ubuntu can't they?
<wingrime>
drachensun: main concept of google - PC/notebook/tablet -- only consume device
<wingrime>
drachensun: it can be chaep, becose you pay for service
<wingrime>
err
<wingrime>
cheap
<drachensun>
yeah, I gotcha
jemk has quit [Remote host closed the connection]
<drachensun>
yeah, I can't compete in things they go after, they have a lower cost and can survive on much smaller margins
<drachensun>
I have to pick the hardware that seems competitive and has enough margin, I'm well below them but can make a living
<Turl>
drachensun: 3G is cheapish on the states
<wingrime>
drachensun: tablets actualy consume device itself, as you can't create something new it
<wingrime>
drachensun: google have enoght money for make tablets cost 200-300$
<brain__>
give me something openpandora-ish with no gaming controls and more pc like.....WANT
<wingrime>
drachensun: they only want more and more people use thay services, and see ads
<drachensun>
yup
<drachensun>
they can break even
<drachensun>
take customers from apple
<drachensun>
and be perfectly happy
<oliv3r>
a31 is conciderd a flop by AW management
<wingrime>
drachensun: so tablet only "thin client" , or old term - terminal
<oliv3r>
even a bad decision
<wingrime>
drachensun: I think you can only improve your situation with extending dialog with manufactures
<oliv3r>
drachensun: as for a31 powervr drivers, don't count on them; aw has no intention ofbuying them
<drachensun>
the drivers they have now work fine
<brain__>
are they open?
<drachensun>
I have them kicking ass on 3d in mer
<wingrime>
drachensun: like special case, logos, tux everywhere.....
<oliv3r>
i guess it's just like the ubuntu-edge; didn't make it either, maybe if they promissed true source
<drachensun>
wingrime: I don't have any pull with them, sales=pull=progress on the issues that are important to us
<drachensun>
oliv3r: That was crazy, they want 10 months
<drachensun>
no real specs
<drachensun>
no clear design
<drachensun>
high price
<drachensun>
from a company that has never made hardware
<drachensun>
and they still got that many people interested
<wingrime>
drachensun: thats impossible to make direct order from vendor?
<drachensun>
I think its a good they had
<drachensun>
I plan to offer the exact same thing they specified if I have the funding to put someone on it full time
<drachensun>
on MTK processors :-)
<drachensun>
good idea I mean
<drachensun>
wingrime: I'm not sure what you mean, I am working directly with a vendor
<drachensun>
wingrime: And they make lots of stuff, netbooks like you mention, GPS, and tablets
<drachensun>
wingrime: And they can do whatever, if have the volume
<wingrime>
drachensun: you still not anought volume?
<wingrime>
*have not enought
<wingrime>
thats bad
<drachensun>
wingrime: no, with the first project, I barely paid my bills for 10 months while family volunteer labor shipped orders and I worked on the quad cores
<drachensun>
drachensun: This project, if it gets its goals, will represent enough to have 1 or 2 full time engineers just pushing out upgrades
<drachensun>
oops
<drachensun>
anyway and hopefully that will lead a 'virtous cycle'
<drachensun>
where the new features push sales
<oliv3r>
drachensun: you DID offer pretty good value; but powerVR, they know they'll never get support for that, other then some cognation like powervr
<drachensun>
and allow us to offer a variety of things
<wingrime>
drachensun: hard to say something , atleast you add tons of stickers in package?
<drachensun>
lol
<drachensun>
and tote bags
<drachensun>
wingrime: My marketing guy does want to add stickers this time
<drachensun>
and t shirts
<wingrime>
lol
<wingrime>
drachensun: one/two sticker will be good
<wingrime>
drachensun: bags no, useless
<wingrime>
drachensun: t-shirts.... no I think
<drachensun>
they only cost $1 in volume, with our logo
<drachensun>
can you believe that?
<drachensun>
the bags I mean
<wingrime>
wow
<drachensun>
its going to cost more to ship them, than to make them
<arokux2>
drachensun, you'll have more luck designing something like dev boards
<drachensun>
na, become what the 3rd, 4th guy in that market?
<drachensun>
I've been a PCB design guy
<wingrime>
but $1 can be many in mass production
<drachensun>
I'm not very interested in going back to that :-)
<arokux2>
drachensun, there are just too many tablets out there, what number are you at this market?
<arokux2>
:)
<drachensun>
with linux?
<drachensun>
0
<drachensun>
with android? who knows, lots
<drachensun>
but thats not what I'm selling
<drachensun>
I had a guy making them terminals for data entry
<drachensun>
portable
<drachensun>
cheaper than a PC
<drachensun>
linux, so all his software already worked
<steev>
a rugged tablet would be nice
<drachensun>
get was set to buy 100 a month or so
<drachensun>
so man, if we can just get the new tablet out there, I think we will be able to keep it going
<arokux2>
drachensun, ppl just buy a tablet go to our wiki and flash an image with linux to SD card. the only problem with our wiki it has no support for A31
<drachensun>
steev: yeah, there has been a lot of request for that
<wingrime>
drachensun: I still think , service, package, branding important, but there many tablets on market...
<steev>
drachensun: there's a reason for a lot of requests :P
<drachensun>
arokux2: tell that to a vendor who wants to buy hundreds or thousands, thats not their business, they dont want to be the OS maintenance/hacking business they just want it to work so they can be in their business
<wingrime>
drachensun: you better,
<drachensun>
steev: Its also a fairly hard problem, again without volume
<wingrime>
drachensun: do 18 hours netbook with aw
<arokux2>
drachensun, if you have a vendor that wants to buy a tablet that boots linux this is another story. but you do not have one.
<drachensun>
well I do
<drachensun>
and my A10 supply got cut off
<drachensun>
and none of them have the volume for the new MOQs
<arokux2>
drachensun, how comes? Allwinner isn't manufacturing A10 anymore?!
<drachensun>
but they will buy again, once we have stock
<wingrime>
drachensun: a20 nice supported now
<drachensun>
my suppliers stopped offering the model we were selling, basically all the A10 designs are getting A20s
FDCX_ has quit [Ping timeout: 264 seconds]
<drachensun>
wingrime: All the sample samples I have tried have wonderful new touchscreen controllers
<drachensun>
with driver source I can't get
<wingrime>
drachensun: are you saw a20 sdk?
<drachensun>
yeah, as far as I can tell, its not available in there
<wingrime>
drachensun: It have some new ts drivers that we not have in sunxi
<drachensun>
yup
<arokux2>
drachensun, cann't you just buy a driver?!
<drachensun>
lol I tried
<arokux2>
:)
<drachensun>
I got some code that didn't compile :-)
<drachensun>
I didn't really buy it
<drachensun>
I pressed my supplier, they got what they thought was it
<wingrime>
drachensun: once I was in situation that I have no drivers for ts
<drachensun>
so basically, I've just decided, I'm going to work with a vendor who gives me source, always, straight from what they use
<arokux2>
if you would have orders, then find the driver for ts by all means and you are done!
<drachensun>
and if they have a high MOQ, then I have to figure out a way to reach it
<arokux2>
a driver for ts isn't a driver for a gpu after all...
<wingrime>
drachensun: I readly RE-it, i2c ts drivers not difficult
<wingrime>
*realy
<drachensun>
hmmm
<oliv3r>
drachensun: but more may have pledged if it would have been guaranteed open by marky mark
<wingrime>
drachensun: firstly try read i2c regs
<arokux2>
oliv3r, drachensun cannot publish something he doesn't own
<drachensun>
yup
<drachensun>
ok, I really gotta go now, later all
<arokux2>
drachensun, pay wingrime to RE a driver for you ;)
<arokux2>
drachensun, anyway. good luck!
<oliv3r>
later D
<oliv3r>
and you guys talked faster then i could backraedy
<wingrime>
arokux2: useless without real hw
<arokux2>
wingrime, he could send hw to you
<wingrime>
drachensun: good luck
<wingrime>
arokux2: no want
<wingrime>
arokux2: but I can help free
<arokux2>
wingrime, why? you are REing here for free.
<wingrime>
arokux2: yes, payed work add some additinal things that make it less fun
<arokux2>
wingrime, how do you make your living if I dare to ask?