* TheSeven
can't create a user page either, ran into this a few hours ago
<libv>
plaes: you are an "autoconfirmed user", so let me change protection to "only autoconfirmed users"
<oliv3r>
mripard: i hope it's safe to assume that you set the address in *iomap and then the data pointed to the pointer will be treated as const, so in effect, shouldn't void __iomem * always be const?
<oliv3r>
mripard: to make it more confusing to me, it's a *const *? which is different to const * obviously not sure how?
<libv>
plaes: try now
KBme has joined #linux-sunxi
<plaes>
nope :)
<plaes>
"
<plaes>
You do not have permission to edit pages in the User namespace.
<libv>
hrm. that's probably some global setting that mnemoc should know about
21WABU2IG has joined #linux-sunxi
<libv>
let me set you to "people"
<libv>
try now
<plaes>
\o/
<libv>
ok, so people need to join the "people" group before they can edit their own pages. makes sens.
<plaes>
meanwhile, I discovered that betemcu's UART has RXD and TXD mislabeled..
<libv>
ah, so swap them around, and all is well?
BluesBoy- has joined #linux-sunxi
<plaes>
yes
<plaes>
though it always takes few times to get them right anyway..
<libv>
cool, document it clearly :)
<libv>
also, windows sucks :p
21WABU2IG has quit [Ping timeout: 264 seconds]
BluesBoy has quit [Ping timeout: 252 seconds]
BluesBoy- is now known as BluesBoy
zeRez has joined #linux-sunxi
Taduro_ has joined #linux-sunxi
<mnemoc>
people are verified humans
<mnemoc>
the easiest way to not annoy normal people when they edit in spamish fashion
<libv>
:)
ganbold_ has joined #linux-sunxi
<TheSeven>
can someone put me into the people group then?
eebrah has joined #linux-sunxi
Black_Horseman has quit [Quit: Zwi se logou mou!!!]
<mnemoc>
libv can =)
<libv>
TheSeven: i didn't see any recent contributions by you
<libv>
hrm.
<TheSeven>
I just registered hours ago, didn't edit anything yet
<libv>
why is it always people from a certain part of the world who come with whiny questions that basically state "i am too stupid and/or too lazy to actually do what i am paid for, please help me"?
<TheSeven>
actually I ran into this when I wanted to prepare some info on livesuit etc. in my user namespace that I feel need to mature before placing them into the publuc part
<TheSeven>
public*
<libv>
TheSeven: oh, i overlooked you, i hadn't immediately spotted your account name in recent contributions
<libv>
TheSeven: fixed now
\\Mr_C\\ has quit [Quit: .]
<TheSeven>
libv: thanks :)
<notmart>
hi all, i have just quick query: weird problem while testing opengles an an a20 based eoma68
<notmart>
all gles applications fail with error 0x3005
<notmart>
..except the small test app provided in sunxi-mali
<notmart>
but then there isthe very weird thing: after i ran that test application, all other openglex applications magically start to work
<notmart>
so seems that application does something to change the state of the driver and make opengl work after it ran..
<notmart>
am i missing something obvious? :)
<libv>
notmart: modprobe mali?
<notmart>
hmm, mali module seems to be loaded beforehand
<notmart>
(and /dev/mali and /dev/ump exist and have right permissions)
<libv>
notmart: well, the code is there, there is supposedly nothing special in there
<libv>
it could be that there is something in fbturbo though
<notmart>
i just built current fbturbo from master don't know if there are issues with a particular revision..
<libv>
what kernel driver are you using?
<libv>
r3p0 or r3p2?
<notmart>
r3p0
<notmart>
it's i think the one that is included in the linux sunxi kernel tree on gitorious?
<libv>
github you mean
<notmart>
yeah, sorry :)
<notmart>
github
<libv>
ok
<libv>
strange, i have never heard of that before
<libv>
how did you install the mali binaries?
<libv>
by letting the sunxi-mali makefiles do their work?
<notmart>
i did make config VERSION=r3p0 ABI=armhf EGL_TYPE=x11
<libv>
did you reboot since you ran the sunxi-mali test?
<notmart>
on each boot i have to run that test to make other opengl stuff work
<libv>
to ascertain that it really is the running of the test that fixes it?
<libv>
ok
<libv>
strange, as that code does nothing special
<notmart>
yeah, looking at it, is not really doing much else than creating a context..
<notmart>
will do other tests..
<libv>
notmart: comment out the egl config samples bit
<notmart>
o the test aplpication?
<libv>
it's not clued to do so, but this is immediately different from all other egl apps
<kivutar>
hello, what's the reference repo and branch for u-boot-sunxi for the cb2?
<notmart>
hm, have to redo an image with gcc and all, will take a bit, btw is the eglChooseConfig() line that i have to remove right?
<libv>
no, EGL_SAMPLES, 4
<notmart>
ok
<libv>
notmart: but if it is that much work, leave it
<libv>
as it is not a sensible change
<notmart>
nah, i'll do it anyways ;)
<libv>
i would blame some weird interaction with X and fbturbo on your mer image
<libv>
not between just those two, but between those two and the x11 binaries
tomboy65 is now known as tomboy64
<notmart>
worse case scenario i put something derived from that test autostarting on x before the rest.. just a bit ugh.. :)
<TheSeven>
dammit, tell me some android core component that allwinner has NOT messed with!
Taduro_ has quit [Read error: Connection reset by peer]
<notmart>
libv: btw the test app still works also after commenting it
<libv>
yeah, it should
<notmart>
what i see, is that it draws the triangle both on the x11 display and on the framebuffer at the same time, maybe has something to do with that..
<libv>
whut?
<libv>
it should not be doing that at all.
<notmart>
ah, it does that just one time
<libv>
?
<notmart>
just the first time i launch it, subsequent starts it just pops up on x11
<libv>
so yes, there is an issue with the three components mentioned before
<libv>
ssvb: any ideas on whether this is tied to your scaler work?
<libv>
err, layer work
<oliv3r>
mripard: if a driver starts to do 'platform_data_add/add_data' it's basically accessing lower lvl functions that are usually wrapped around in high level frameworks right?
JohnDoe_71Rus has joined #linux-sunxi
shineworld has joined #linux-sunxi
montjoie[home] has quit [Ping timeout: 252 seconds]
KBme has quit [Ping timeout: 252 seconds]
montjoie[home] has joined #linux-sunxi
KBme has joined #linux-sunxi
<oliv3r>
mripard: ok i think i figured it out somewhat :p but I do have a new Q :) http://lxr.free-electrons.com/source/drivers/ata/ahci_platform.c?a=arm#L127 gets me mmio remapped, great, a few lines lower (151) that mmio address is passed to the init call to setup the controller (sunxi_ahci_init). After that, the AHCI controller in the soc does things again, but it doesn't know the register address etc, so how does it know how to communicate?
<oliv3r>
hmm, come to think of it,all those vars are stored in hpriv of course, but are those register offsets 'default' and expected all on the same address?
<mripard>
I think it's passed to ata_host_activate
<mripard>
that knows how to behave
<mripard>
but yeah, those things are actually standard
<mripard>
it's the point of *hci
<oliv3r>
yeah i imagine
<oliv3r>
i just find it 'hard to belive' all manufactures actually follow the standard and have those registers on the exact right place
<oliv3r>
:p
<oliv3r>
mripard: just an fyi; if you haven't guessed; i'm trying to 'copy' the sdhci-pltfrm to ahci_platform
ganbold_ has quit [Remote host closed the connection]
<oliv3r>
jusst a little weird it's put in 'private_data'
<mripard>
well, not all the manufacturers do this
<mripard>
antdt they are not ahci compliant
ganbold_ has joined #linux-sunxi
<JohnDoe_71Rus>
There are those who are familiar with android?
<oliv3r>
mripard: if the controller is fully compatible, just register layout is different, some remapping would be required I suppose?
<TheSeven>
JohnDoe_71Rus: familiar with which area of it?
<JohnDoe_71Rus>
Need information on how to configure usb wifi via GUI. Where specified adapter.
<oliv3r>
if its recgonized by the kernel; normal WIFI setup should work
<JohnDoe_71Rus>
no, then check "wifi on" they try turn other device
<TheSeven>
you mean an additional, second wifi adapter?
<plaes>
yeah, android UI doesn't support multiple wifi cards
<plaes>
TheSeven: what if you remove the first driver
<plaes>
JohnDoe_71Rus: ^^
<JohnDoe_71Rus>
no. the image has setup one wifi adapter. but i don't have it. i get kernel with my adapter suport.
reallife has joined #linux-sunxi
<reallife>
ping ssvb
<reallife>
hi
printallthething has quit [Ping timeout: 246 seconds]
<plaes>
JohnDoe_71Rus: by looking at your logs I have no idea
<ssvb>
if the memory clock speed is too high, mali usually becomes unstable first (while the rest of the system seems be be able to cope with it)
KBme has quit [Ping timeout: 252 seconds]
<notmart>
ssvb: dram_clk: 384
<ssvb>
this looks ok
<ssvb>
nothing extreme at least
guizamboni has quit [Quit: Leaving]
FR^2 has joined #linux-sunxi
n01 has quit [Ping timeout: 240 seconds]
<libv>
notmart: what X version is this?
<notmart>
1.13.0
rz2k has quit [Read error: Connection reset by peer]
<zumbi>
bfree_: are you around?
<bfree_>
hi zumbi
booss has quit [Ping timeout: 245 seconds]
<zumbi>
bfree_: hey.. I am picking up on the debian work
<zumbi>
bfree_: I recently ordered a cubietruck, finally, I'd like to work on the debian enablement
<zumbi>
bfree_: do you happen to know latest status?
<zumbi>
or where can I find your work (or others if any)
<bfree_>
zumbi: packaging wise little has changed (libv has some gaphics related stuff for ubuntu) ... but I did earlier put up http://dl.linux-sunxi.org/users/niall/debian.preview/incoming/ which is a lightly tested much improved sunxi-tools idea (see README.txt about u-boot)
<kenny>
So, I had cubian working and I decided to grab the latest nightly u-boot-sunxi and dd it on the SD card. Now the doesn't boot ("Starting Kernel ..." then u-boot again). Is there an easy way to get this working again (other than dd-ing back the old u-boot)
<Turl>
oliv3r: did you break your mail client or does your name have two 'l'?
<bfree_>
zumbi: your big build-farm full of stuff would make building u-boot and kernel packages much nicer ;) compile times are far too slow (kernel and u-boot at least).
<zumbi>
bfree_: those can be cross built
<zumbi>
bfree_: I got my server on-line, now but I am not doing any public service
<zumbi>
bfree_: thanks, I need to do some reading to get all the sunxisms as sunxi-tools and cedarx stuff
<zumbi>
bfree_: also, what were you saying on graphics? isn't mali and sunxifb doing well on cubies?
<bfree_>
zumbi: vis-a-vis real debian enablement, u-boot probably wants upstreaming for that (doubt external patches will be too welcome) and the recent kernel stuff might see a more usable (but no video) kernel in debian for 3.13 or 3.14 perhaps? I must admit I don't know how d-i and arm work together and don't reagrd that as too interesting until the kernel is in better shape
<WarheadsSE>
who needs build farm...
<zumbi>
WarheadsSE: bfree_ and I were planning one to do automated builds of packages
<bfree_>
zumbi: mainline kernel certainly doesn't play well with graphics. I'm not sure exactly how much libv's mali, libump and xerver-xorg-video-fbturbo will do on 3.4
<WarheadsSE>
zumbi: for a deb derivative?
<bfree_>
well more zumbi teased with mention of one, so I'm teasing back reminding him about it :-p
<zumbi>
WarheadsSE: aye
<WarheadsSE>
mm
* WarheadsSE
has Arch build farm
<WarheadsSE>
helps a lot ..
<zumbi>
bfree_: video-fbturbo?
booss has joined #linux-sunxi
<zumbi>
aren't you using xerver-xorg-video-sunxifb? that one worked quite well on odroid
<zumbi>
WarheadsSE: well, I got some OBS setup.. bfree_ we can use OBS upstream for some of this
<WarheadsSE>
yeah, OBS can do deb
<WarheadsSE>
they only recently began Arch
<zumbi>
nod
<WarheadsSE>
we had to build ours from scratch :P
<zumbi>
re-inventing wheels?
<ssvb>
zumbi: it was renamed, precisely because it also works on odroid an rpi
<ssvb>
*and
<zumbi>
uhm.. rpi.. nice
<zumbi>
but rpi is not mali based, is it?
<bfree_>
libump dsc from there can (could) build on debian sid last I tried ... I got a fugly hackish mesa working but failed when I tried to pick out the patches onto the latest debian sid version (and for fbturbo afair you need both that and to rebuild the xserver dev package from ubuntu)
bfree_ is now known as bfree
<tomboy64>
ssvb: please excuse the interruption; in this walkthrough http://linux-sunxi.org/CedarX/libve, am i supposed to clone you libhybris (cedarx branch) into system/lib/[libhybris...]?
<ssvb>
zumbi: no 3d support for rpi at the moment, but it might be probably implemented
<zumbi>
what about using a xserver-xorg-video-wayland :)
<zumbi>
bfree: are you using mesa on sunxi? (with lima patches?)
<ssvb>
zumbi: basically fbturbo for rpi only solves the horrible window moving and scrolling performance, so you don't get the horror showcased on http://www.youtube.com/watch?v=0UkUal_hHx8 video :)
KBme has joined #linux-sunxi
popolon has quit [Ping timeout: 245 seconds]
<zumbi>
bfree: not using mali-400-drivers instead?
<bfree>
zumbi: I'm basically only using them headless for now anyway, often debian kernel with an nbd-root ;)
<zumbi>
ssvb: right, on X11 you mean
<WarheadsSE>
zumbi: at the time, we needed a build system for ALARM, and OBS had not and didn't want a clue
<zumbi>
bfree: ok - this is going to be fun
<WarheadsSE>
So the dev team made one.
<WarheadsSE>
takign the archisms into account
<bfree>
graphics output doesn't mean much to a make deb-pkg (current task just to try a hans sunxi-next) or compiling u-boot ;)
<zumbi>
WarheadsSE: what is ALARM?
<ssvb>
tomboy64: the /system directory is from android (that's where libhybris finds the android binary blobs)
<WarheadsSE>
Arch Linux ARM
* WarheadsSE
poof
<ssvb>
tomboy64: the libhybris repository just contains a wrapper, which allows to load these android cedar blobs
<zumbi>
WarheadsSE: i see, thanks
Black_Horseman has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
Black_Horseman has quit [Changing host]
<tomboy64>
ssvb: i know; i was just a bit confused with the build-instructions. seems it worked out fine, i have the libvecore.so :)
<parabyte>
zumbi, the real magic is getting the touch screen working
Gerwin_J has joined #linux-sunxi
parabyte has quit [Quit: Leaving]
<zumbi>
is it multi touch?
notmart has quit [Quit: notmart terminated!]
tzafrir has quit [Ping timeout: 252 seconds]
zeRez has joined #linux-sunxi
shineworld has joined #linux-sunxi
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
ZetaNeta has quit [Read error: Operation timed out]
FR^2 has quit [Quit: und weg...]
shineworld has quit [Ping timeout: 240 seconds]
<techn__>
Turl: any eta for pll3/7?
zeRez has quit [Quit: zeRez]
mosfet_ has joined #linux-sunxi
<Turl>
techn__: none planned as nobody has been demonstrating an interest in disp on mainline so far
<mosfet_>
Hi, I'm looking for an example of using fel-gpio with pio to toggle IO for a sanity check. I'm in the process of trying to bring up an A31 coreboard.
<Turl>
techn__: if you're going to work on that I can get them implemented soonish though
<Turl>
(or anybody else)
AreaScout has quit []
<techn__>
Turl: I started experiments.. no hurry in that sense, I can manage with my own hacks
n01 has quit [Ping timeout: 250 seconds]
SiCo has joined #linux-sunxi
arete74_ has quit [Ping timeout: 248 seconds]
<SiCo>
Hi
arete74 has joined #linux-sunxi
kivutar has quit [Quit: Ex-Chat]
<SiCo>
somebody know how i can create my own Cubieboard 2 (A20) kernel and install it into the nand?
<SiCo>
I have created a kernel using sunxi git kernel source
booss has quit [Ping timeout: 250 seconds]
<SiCo>
now how i can make the nand installation?
<SiCo>
i have uImage and output folder from modules compilation