<libv>
after all the nand wiki work, i now get to work the cubietruck wiki page...
<libv>
the neverending story
<spv>
libv: do we now have nand support on A20?
pwhalen has joined #linux-sunxi
<libv>
spv: can you explain how you came to that conclusion?
<spv>
you mentioned some nand wiki work, and I considered the possibility that the "work" involved stating in the wiki that there was a potential driver.
<libv>
spv: or are you just so desperate for that to happen that you believe that mentioning 'nand" and an a20 device in the same sentence could mean that you finally get your problem solved?
<libv>
spv: where in that sentence do i make a link?
<libv>
spv: i needed nand for my second attempt at getting a working a13 board with uart, as the uart on that device was multiplexed with the sdcard
<libv>
spv: i ended up asking a load of questions and fixing an issue left and right, and spent a lot of time documenting this
<spv>
Apologies for taking your time then.
<libv>
i am now bringing up a cubietruck, so i can do a tiny bit of kms work on it first, and then do a bit of lima work, so i have something to talk about at fosdem
<libv>
no cubietruck pictures... nice.
pwhalen has quit [Ping timeout: 245 seconds]
TheSeven has quit [Disconnected by services]
[7] has joined #linux-sunxi
pwhalen has joined #linux-sunxi
spv has quit [Quit: Ex-Chat]
<libv>
hrm, cubietech.com refuses to load for me
<Turl>
libv: try cubieboard.org
<Turl>
libv: I bet mnemoc has some pics on his G+ too
<mrnuke>
libv: sure, I'll do that as soon as you man up, get a proper browser, and stop complining about bloatware working like crap ;)
t3st3r has quit [Remote host closed the connection]
t3st3r has joined #linux-sunxi
<libv>
mrnuke: idiot.
<mrnuke>
it's spelled ID 10 T ;)
<Turl>
do I need to start babysitting people here? :)
<mrnuke>
Turl: I'm good
hero100 has joined #linux-sunxi
<libv>
and i have better things to do
* mrnuke
sighs. Some people just don't understand other people have differring opinions. It's the former tring to correct the latter that results in the flame war
<libv>
...
<Turl>
philosophy--; sunxi++
<libv>
holy_war_baiting--
<mrnuke>
[OT] BTW, I think all msjor browsers suck
<libv>
mrnuke: no-one cares, now stop it.
<mrnuke>
kinda hard to ignore omeone when they keep yelling at you
<Turl>
I'd rather see you two collaborate to get proper disp support on coreboot & linux
<mrnuke>
Turl: sure. as soon as the atmosphere cools down
* Turl
prepares liquid nytrogen in case it's needed
* mrnuke
hands Turl a dewar of liquid helium
<libv>
the A20-Cubietruck page looks a tiny bit better already
<libv>
but this whole sunxi status thing is quite messy
<Turl>
libv: I thought we didn't have one
<libv>
and i have a distinct feeling that a lot of that might be outdated
<Turl>
although on a second thought it makes sense - different pins than cb1/2
<Turl>
libv: can't read it now, this DSL I'm using sucks *badly*
<libv>
imho cb1/cb2 need separate pages as well
<libv>
i am very certain that most information in the cubieboard pages, and the pages linked from there are duplicated somewhere else on our site already, or refer to cubieboard.org anyway
<libv>
Turl: about coreboot... i have contributed to coreboot from 2007 til 2010
<libv>
imho a very very important project for x86
<libv>
but i do not see the point for arm
<libv>
i did via stuff on it, did the first fully native C vga console code for a commercial motherboard (some amd/unichrome based asus) and significantly restructured flashrom
<libv>
oh, and ended up plucking all the bits and pieces off the interweb and writing up a x86 vendor bios extraction tool
<libv>
as for the vga console, google and intel recently put their heads together to provide native display support for some chromebooks. 3-4 years on.
<mrnuke>
sure there is a point for ARM. There are a lot of things coreboot has to offer that other firmwares just can't
<libv>
mrnuke: such as?
<mrnuke>
from a developer point of view: modular source tree that is easy to work with
<mrnuke>
from a security standpoint: crypto-verified boot
<libv>
is u-boot such a mess then?
<libv>
how hard is encryption to add to u-boot?
<mrnuke>
uboot is a firmware and bootloader mashed together
<mrnuke>
it's designed to do too many things, and not do any of them well
<mrnuke>
google actually wrote their own bootloader so that they'd stop using uboot
<mrnuke>
it's not hard to add crypto, just that it would be a clusterfuck, given the current structure
<mrnuke>
(BTW, Allwinner can't have a verified firmware path)
<montjoie[home]>
and the security system on A20 could do some key management for boot
hero100 has quit [Ping timeout: 252 seconds]
Gerwin_J has joined #linux-sunxi
Gerwin_J has quit [Ping timeout: 252 seconds]
<Turl>
montjoie[home]: hi, I got someone asking me about DES/3DES SS linux support on an email the other day
<Turl>
montjoie[home]: did they talk to you?
<montjoie[home]>
no but I will add it
<montjoie[home]>
I wait for comment on my patch before working more on it:)
<montjoie[home]>
adding DEs/3DES should be simple
<montjoie[home]>
and I have a idea that perhaps could boost performance, preload key in security id, perhaps it is more fast than load it from memory
<Turl>
montjoie[home]: don't you need to write key to register once?
<libv>
hrm sun7i defconfig is rather... inclusive?
Quarx has joined #linux-sunxi
<Turl>
libv: I've heard some people saying it missed stuff a couple of times
<Turl>
to each their own :p
<Turl>
good night
<libv>
400+MB of modules?
<Turl>
that's just probably debug symbols on :)
<Turl>
get them stripped
<libv>
no, there is a lot more in sun7i than in sun4/5i
<Turl>
yeah but not 400M worth :p
<libv>
it's finally done building.
<libv>
110mb net/wireless, 73mb media,
<montjoie[home]>
Turl sadly no, I must write all stuff for each request (aka for each 512b)
<montjoie[home]>
512bytes for dm-crypt
_newleaves_ has joined #linux-sunxi
<_newleaves_>
test
ggcock has joined #linux-sunxi
hero100 has joined #linux-sunxi
<_newleaves_>
hi anybody ?
hero100 has quit [Remote host closed the connection]
Black_Horseman has quit [Remote host closed the connection]
kurfust has joined #linux-sunxi
kurfust has quit [Client Quit]
<libv>
nice
<libv>
libreoffice-impress doesn't take that long to load on cubietruck
<libv>
but that's probably because it is still cached
<libv>
still, should run smooth enough for fosdem
Fusing has joined #linux-sunxi
FreezingCold has quit [Ping timeout: 272 seconds]
maz_ has joined #linux-sunxi
techn_ has joined #linux-sunxi
rz2k has quit []
ggcock has quit [Read error: Connection reset by peer]
ggcock has joined #linux-sunxi
bfree_ has joined #linux-sunxi
bfree__ has quit [Ping timeout: 276 seconds]
Gerwin_J has joined #linux-sunxi
wingrime has joined #linux-sunxi
mart has joined #linux-sunxi
hipboi has joined #linux-sunxi
mart has quit [Client Quit]
notmart has joined #linux-sunxi
Gerwin_J has quit [Ping timeout: 252 seconds]
bsdfox has quit [Ping timeout: 260 seconds]
_newleaves_ has quit [Ping timeout: 252 seconds]
_newleaves_ has joined #linux-sunxi
bsdfox has joined #linux-sunxi
paulk-collins has joined #linux-sunxi
bsdfox has quit [Ping timeout: 272 seconds]
<ggcock>
hello, i just read the pack.sh in cubieboard. The bootfs.fex is create by "fsbuild bootfs.ini split_xxxx.fex". How is the split_xxxx.fex comes from and what does it mean?
notmart has quit [Quit: notmart terminated!]
hipboi has quit [Read error: Connection timed out]
<oliv3r>
libv: hmm i thought i uploaded some ct pics at some point
<andhe_>
would implementing support for PS2 in A20/Cubietruck on top of mainline be the most useless task ever? does anyone have a suggestion for something else that's really simple that I can use to get started hacking?
bsdfox has quit [Ping timeout: 245 seconds]
<oliv3r>
libv: on my CT with SSD, the whole thing boots and loads impress extrmply fast
<oliv3r>
nikrou: probably bootloader, or the debian is a framebufferless OS; hard to say without debug :)
Vespine has quit [Remote host closed the connection]
<nikrou>
I take same kernel, same bootloader, ...
deasy has joined #linux-sunxi
cajg has quit [Ping timeout: 272 seconds]
<sgo11>
hi, how to check if my ubuntu or kernel has battery module? or how to check battery status in command line? thanks.
<plaes>
sgo11: it should be somewhere under /sys/class/power_supply
<sgo11>
plaes, yeah, I have dir "/sys/class/power_supply/battery/" "status" file has content "Full". Does this mean the battery module is working? I should unplug the power to see if status will change or not. thanks.
<sgo11>
unplug the power will make "status" file content becomes "Discharing". thanks a lot.
cajg has joined #linux-sunxi
penghb has quit [Ping timeout: 245 seconds]
<sgo11>
s/Discharing/Discharging/
Night-Shade has quit [Ping timeout: 252 seconds]
dlan_ has joined #linux-sunxi
dlan_ has quit [Changing host]
dlan_ has joined #linux-sunxi
Gerwin_J has joined #linux-sunxi
penghb has joined #linux-sunxi
Gerwin_J has quit [Ping timeout: 252 seconds]
_newleaves_ has quit [Quit: Leaving]
<nikrou>
I successfully but my tablet with debian from external sdcard.
<nikrou>
Now, how can I restore my android system on it ? Didn't find that kind of link on the wiki. :-(
<nikrou>
thanks in advance
<hramrach>
nikrou: to restore andriod to factory state you need to download the factory image from your device manufacturer
<nikrou>
I've got all nand partition.
<hramrach>
you copied it from nand earlier?
<nikrou>
hramrach, I try with Livesuit but didn't succeed. Oliv3r suggest me to boot from sdcard to restore it
<nikrou>
yes
<hramrach>
then you can presumably copy it back
<nikrou>
But I think I need to do something else because when I do it now booting on android and copy it from shell (adb shell) it doesn't work
<hramrach>
how did you copy it?
<nikrou>
cat nandf_saved > /dev/block/nandf
<nikrou>
for all nands
wens has quit [Read error: Connection reset by peer]
<hani>
nikrou : try the new phoenixSuit it should do the trick
wens has joined #linux-sunxi
<hani>
nikrou : just mount those partition and copy the contents or dd to those bloks
<nikrou>
but I'm on linux :-)
<nikrou>
so no phoenixSuit !
<nikrou>
With a debian on my tablet I'm in my garden !!!
<nikrou>
But I only need to unable wifi on startup because I haven't got a micro usb female port to access it with my keyboard !
<nikrou>
s/unable/enable/
<hani>
nikrou: Ok then boot from sd then create the android partition table and restore from your back up
<hani>
I'm assuming that u have changed ur partitions when u installed debian
hani has quit [Quit: Page closed]
<nikrou>
@hani : no, I didn't modify something
<nikrou>
I had an internal sdcard and an external one
fredy has quit [Excess Flood]
<nikrou>
I boot from the external one
fredy has joined #linux-sunxi
earny has quit [Remote host closed the connection]
maz_ has quit [Ping timeout: 252 seconds]
<libv>
ssvb: how much slower than a10 is a20?
<libv>
is 312MHz all the mali400 has to give here?
<ssvb>
mali can clock up to 370mhz according to the datasheet
<ssvb>
and the practical overclocking limit is around 408mhz
<ssvb>
the 370mhz limit seems to be the same for a10 and a20
<libv>
ssvb: can you document the solution for your driver complaining about too small an fb?
<libv>
hramrach: thanks for the pics :)
<ssvb>
libv: I think I'll just submit a kernel patch to have everything handled by default
<ssvb>
libv: documenting it all over the place is not very effective, but surely can be also done
<libv>
ssvb: all over the place?
<libv>
ssvb: you just edited the binary drivers page
<libv>
i am not sure what i did wrong, but altering script.bin failed for me
<libv>
kernel command line worked
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
<libv>
ah, much better, 151fps for glmark2-es2
ggcock has quit [Quit: Leaving]
samul has quit [Ping timeout: 264 seconds]
<hramrach>
the pics are from a phone but better than no pics ;-)
<libv>
cpu usage does look rather good with fbturbo and double buffering
<libv>
hramrach: good enough indeed
dlan_ has quit [Quit: leaving]
sgo11 has quit [Quit: Leaving]
sgo11 has joined #linux-sunxi
sgo11 has quit [Client Quit]
<libv>
hrm, fbdev pan display got disabled?
<libv>
i don't recall that happening
<libv>
anyway, on a20, 1024x768, companion cube spinning: limare is 296fps, binary is 220. But most of that will be down to limare using both PPs while the binary probably doesn't
sgo12 has quit [Quit: Leaving]
bsdfox has quit [Ping timeout: 272 seconds]
bsdfox has joined #linux-sunxi
bsdfox has quit [Changing host]
bsdfox has joined #linux-sunxi
<Turl>
libv: phoronix doesnt need to know that ;)
<Turl>
libv: what blob are you using?
<libv>
r3p0, i need to reboot now, as i just triggered something with ump which used to work just fine
<Turl>
I think it's r3p2 which has dual pp support
<libv>
yup
<libv>
it's easy to change limare to single pp usage, so i'll give that a go soon
<oliv3r>
libv: i got an older ocz SSD connected to my CT, and it's WOOSH fast
<oliv3r>
and i also noticed that impress starts really fast, especially if I compare it to my laptop/desktop
<oliv3r>
as if arm is somehow faster to start it
<hramrach>
compared to anything else the board can connect to it's fast, yes
<oliv3r>
technically, my desktop should be huge amounts faster then the little a20
<oliv3r>
connect to?
<hramrach>
you have nand, mmc, USB, ethernet
<hramrach>
ethernet without interrupt mitigation ..
<hramrach>
faster impress .. interesting
<oliv3r>
i used gmac yeah
<hramrach>
maybe fewer junk services running which impress tries to register to
<hramrach>
also, did you try impress with the fbdev driver on your desktop?
<hramrach>
that's *FAST* because the graphics decelaration is not used
<Turl>
on my laptop it takes like 6s to start
<Turl>
no wonder it has a splash screen :P
<hramrach>
what are impress files named?
<hramrach>
odp?
<Turl>
.ppt/.pps? :p
<Turl>
odp yep
<hramrach>
take like 3s when in cache but a lot longer when it has to be fetched
<wens>
graphics deceleration?
<hramrach>
like when you are using a Radeon GPU
<hramrach>
the X server renders stuff on the GPU
Gerwin_J has joined #linux-sunxi
<hramrach>
which is slow because the work has to be submitted to the GPU, to do so the buffers are checked for errors that would lock it up
<wens>
one would normally think using the GPU should be faster :(
<Turl>
I used an AGP HD4something the other day on xubuntu
<hramrach>
also the bitmaps in the GPU are in special tiled format and when you need to read them back or perform an operation for which no accelerated version is available migrating them is *really* expensive
<Turl>
it was snappy
<Turl>
(yeah, AGP)
<hramrach>
wens: it's fast if you let it do its stuff but desktops don;t work like that
<hramrach>
Turl: fbdev would likely be even snappier :p
<Turl>
hramrach: on prescott hw? dunno :p
<hramrach>
on any HW
<Turl>
it was using compositing fwiw
<hramrach>
even APUs are faster when rendering in sw
<Turl>
yeah but you probably lose one core which is doin rendering
<Turl>
doing*
<hramrach>
which would be doing buffer checking otherwise ..
<Turl>
when your system has two and it's due to HT it's not so nice
Gerwin_J has quit [Ping timeout: 252 seconds]
<hramrach>
well, it is slightly less CPU expensive to render in hardware but way slower
<Turl>
I should do some benchmarks then
<Turl>
that box is full of surprises
<Turl>
it can boot xubuntu faster than my core2duo
<Turl>
way faster
<hramrach>
prescotts were good for their time
<Turl>
and it has 1/6th of the RAM
<hramrach>
and are still quite good if you do not pay your electricity bills
<Turl>
probably atom-range by now
<Turl>
otherwise a nice heater for your room
<hramrach>
if you count in the newest atoms then probably
<hramrach>
like the server atoms ;-)
<libv>
Turl: you as the resident main android hacker...
<libv>
Turl: do you use u-boot-sunxi at all?
<Turl>
even atoms a couple years ago were computationally on par I think
<Turl>
libv: lichee-dev branch or whatever it is :p
<libv>
Turl: where is this documented?
<hramrach>
computation is not that important for desktop
<hramrach>
it's pretty much all DMA once you get some baseline CPU performance which most CPUs do meet today
<hramrach>
libv: sunxi wiki?
<Turl>
libv: nowhere I suppose, that branch is just an allwinner dump with a patch to pas the mem info via atags corectly so sunxi-3.4 can have a try at booting
<hramrach>
there is something about cyanogen iirv
<Turl>
pass*
<hramrach>
iirc
<libv>
Turl: where is the code, where is the documentation for setting up a build?
<Turl>
libv: on u-boot-sunxi repo, normal uboot build other than you use the standard sun4i target and replace the uboot binary on nanda
<hramrach>
I don't have the cables to connect a LVDS lcd :s
<libv>
Turl: let me go back to the start: where/how are you using u-boot-sunxi lichee-dev, and please point me to the code, Makefiles or manifests
<Turl>
libv: on the 3.0 days we didn't use it for android, we just used nanda as is
<hramrach>
Also broke one already when my notebook fell from a table :/
<Nyuutwo>
hramrach: I just connected scope to PD7
t3st3r has quit [Ping timeout: 240 seconds]
<libv>
Turl: and nowadays?
<Turl>
if you want to use 3.4 you need to manually compile lichee-dev and drop the u-boot-bin on nanda before starting
<oliv3r>
hramrach: i did use only the basic fbdev
<Turl>
no makefiles or anything
<libv>
Turl: ok.
<libv>
Turl: so this is described nowhere?
<oliv3r>
about 8 seconds on my laptop btw, it's much les on my a20; i can test tomorrow
t3st3r has joined #linux-sunxi
<Turl>
libv: I don't think it is
<hramrach>
Nyuutwo: ask tsvetan. supposedly he got it working and reported the bug but nobody but him remembers it. Should be logged on whitequark, too
<Turl>
oliv3r: yeah but you have an ssd. It's a lot of disk spinning there
<libv>
Turl: so nobody apart from you ever does it?
<Turl>
libv: dunno :) it's an one time op in any case
maz_ has joined #linux-sunxi
<Turl>
libv: most people just use the aw sdk and flash whatever it generates with livesuit
<libv>
so... you know your way around, and the rest of the planet uses the aw sdk...
<libv>
to me, that states that this default boot commands change is an improvement
<Turl>
libv: you could keep both of us happy with #ifdef ANDROID_BLAH... #else .... #endif
<Turl>
and make a sun4i_android target with that flag
<libv>
makes sense, but i can only go test half of that
<Turl>
libv: also note that there's a lichee-a20 branch, your patches only seem to fix the sun4/5i one
<libv>
i know.
<libv>
i believe i have spread myself really thinly in this story already
<libv>
and have gone out of my way to document things on top of that
<oliv3r>
Turl: when as the last time you built android for allinner?
<oliv3r>
libv: i think android hasn't been built by members in a long long time; atsampson tried to bring it up, but gave up
<Turl>
oliv3r: 10.1 days I think
<libv>
fixing up lichee-a20-dev is as much my responsibility as it has been others, only very few of them bothered to document anything
<Turl>
oliv3r: [7] got it working I think
<oliv3r>
paulk-collins says he will do a replicant build for sunxi; which is prefect
<hramrach>
I gave up as well
<Turl>
he sent to patches to the ml iirc
<Turl>
to add vsync to disp and sth else
<Turl>
two
<Turl>
damn kbrd
<Turl>
bbiab
<hramrach>
android is just a bugpile anyway
<oliv3r>
i think most of sunxi is about plain linux, since the android patches on top would work for those users
<[7]>
oliv3r: I have a somewhat working android 4.2.2 on a20 (olinuxino)
<oliv3r>
as far as booting from nand goes (android or otherwise) we generally say to just use an mmc; nand relies on libnand and lots of crap
<hramrach>
I had splashscreen working on android and that's about as far as I got
<oliv3r>
(crap being lichee u-boot, nand-part etc)
<oliv3r>
i guess everybody is hoping for bboris and slapin to fix mtd :)
<oliv3r>
with mtd; all are troubles are from yesterday
<[7]>
(my android image is booting from nand as well, I haven't bothered looking into MMC yet)
<oliv3r>
but did you modify nanda at all?
<oliv3r>
i assume you use stock nanda?
<ccaione>
mripard: ping
<[7]>
I built most things from scratch
<[7]>
I used images for boot0/boot1, but recompiled everything starting from boot.axf
<[7]>
mostly to get recovery/fastboot buttons to work right
<oliv3r>
well i only use it to look at things to put into u-boot
<oliv3r>
how DOES the imx6 compare to suxni freedom wise?
<oliv3r>
ah it's a vivante
<lkcl>
oliv3r: the iMX6 is pretty damn good - etnaviv works. the only bit that's not reverse-engineered is the proprietary video engine.
<lkcl>
that's all!
<[7]>
one reason why I preferred to stick with boot1/boot.axf for my project here is that it already contains a charging screen when the device is powered off etc.
<oliv3r>
[7]: ah, boot.axf is the charging screen isn't it
<[7]>
boot.axf is most of boot1's functionality
<[7]>
it's basically its userspace
<[7]>
the boot1 binary is just a kernel-like execution environment for it
<Nyuutwo>
imx6 is neat, but there are some rough edges everywhere
<oliv3r>
Nyuutwo: so it's further then sunxi or just a bit behind
<Nyuutwo>
lkcl: you tried it or you just like concept
<lkcl>
Nyuutwo: i've worked with it on several occasions over the past 8 years and even made a design for debian cross-compiling infrastructure that used bitbake
<lkcl>
Nyuutwo: firmware-imx.inc
<lkcl>
DESCRIPTION = "Freescale IMX firmware such as for the VPU"
<lkcl>
SECTION = "base"
<lkcl>
LICENSE = "Proprietary"
<Nyuutwo>
neat, I just recently started using it and, it looks it has very seep learning curve
<lkcl>
answer: it's proprietary
<lkcl>
Nyuutwo: the combination of python and shell-script tends to do most people's heads in :)
<lkcl>
plus, as it covers such a vast number of systems as well as packages *and* targets, the recipes have been heavily optimised for code-reuse over the years.
<Nyuutwo>
lkcl: for me just some things like use bitbake ... | cat to get log what have been done
<lkcl>
so that the [very few!] people working with it stand a hope in hell of managing so many options
<Nyuutwo>
lkcl: you can just say crosscompiling and you will nail it
<lkcl>
Nyuutwo: i just let it run then hunt around in the work/ directory looking for any compile errors
<lkcl>
Nyuutwo: exactly! fucking awesome. and they've taken care of most of the horrendous errors that most people make
<lkcl>
in their package definitions that is
<Nyuutwo>
lkcl: but when you try making an image and it doesn't update code on which you work ...
<lkcl>
yeahh i know. but, what do you expect: you're building an *entire* OS? :)
<Nyuutwo>
yup
<Nyuutwo>
so I use | cat
<Nyuutwo>
but I don't think I found this on wiki
<lkcl>
oliv3r: answer no. iMX6 requires a proprietary firmware blob for the VPU, equivalent to libcedarx.
<lkcl>
so sunxi has libcedarx reverse-engineered and imx6's VPU is not
<lkcl>
and iMX6 has the 3D GPU reverse-engineered and sunxi does not. yet.
<lkcl>
not to the point where there's a full working opengl library, at least.
<lkcl>
Nyuutwo: anyway - i have to get on.
<mnemoc>
*g*
* nove
stays silent
popolon has joined #linux-sunxi
popolon has joined #linux-sunxi
netlynx has joined #linux-sunxi
kivutar has quit [Ping timeout: 260 seconds]
wingrime has quit [Ping timeout: 252 seconds]
wingrime has joined #linux-sunxi
rellla has joined #linux-sunxi
FreezingCold has joined #linux-sunxi
printallthething has joined #linux-sunxi
jemk has joined #linux-sunxi
<mrnuke>
oliv3r: but I got my freebe already. The next patch has to go to the ML
Night-Shade has quit [Quit: Client exiting]
Night-Shade has joined #linux-sunxi
<oliv3r>
mrnuke: ;)
Fusing has quit [Ping timeout: 245 seconds]
<oliv3r>
so the big difference between sunxi and freescale is, is that sunxi does it all in software (libcedar) whereas freescale all does it in firmware (with a blob)
<oliv3r>
nove: i really should install the tracer shouldn't I? :)
VargaD_ has joined #linux-sunxi
<oliv3r>
so i can take screenshots
pwhalen has quit [Ping timeout: 272 seconds]
<oliv3r>
cedarX is a chapter in itself in the presentation :)
<oliv3r>
nove: not sure if i'll do the 'ask question at any point'
<oliv3r>
we have a time limit, and questions can distract :p
<oliv3r>
gzamboni_: likly, but this is a31 stuff
<ssvb>
nove, jemk: are you coming to FOSDEM?
VargaD has quit [Ping timeout: 252 seconds]
<nove>
oliv3r: usally nobody asks questions from the talks that i saw, i only like this fact and then nobody asking
<nove>
oliv3r: is not important
<nove>
ssvb: i will be here in the irc chat watching the live feed, cheering
torindel_ has joined #linux-sunxi
<mnemoc>
live feed?
<nove>
by what i hear here, appear that will exist live feeds
<mnemoc>
:o
<jemk>
ssvb: no, sadly not. i have a job to do the whole saturday, and for a one day trip i'm a little bit too far away :(
<libv>
yes, FOSDEM organizers are working hard on providing every single devroom with a full recording setup
<mnemoc>
amazing
pwhalen has joined #linux-sunxi
<nove>
oliv3r: only the trace viewer and i can also give some traces for it, also if you want i could do a diagram of jpeg encoding for easy explaining
deasy has joined #linux-sunxi
<Nyuutwo>
oneliner to fix LVDS LCD
<libv>
Nyuutwo: :)
<libv>
Nyuutwo: in script.bin or in disp?
<Nyuutwo>
pll3 = 297
<Nyuutwo>
in script.bin [clock[
<libv>
...
<libv>
what was it before?
<Nyuutwo>
libv: there wasn't this line
<libv>
Nyuutwo: grmbl. bad disp code
<libv>
Nyuutwo: it should initialize this.
<libv>
Nyuutwo: my kms driver does this, and i think disp should've been doing this as well
<Nyuutwo>
so if I waited for after fosdem maybe we didn't find it out?
<libv>
it should just work
<Nyuutwo>
libv: and I also seen code in disp which "initializes pll"
<libv>
as i use pll3 per default
<libv>
i do not like clocking down the engines so i could get the dotclock slightly accurate
<libv>
i am already quite appalled at the limited accuracy of the dotclocks
<libv>
and am somewhat relieved that we will only end up using the second display channel when using composite
<libv>
yes, it's more powerhungry, i know, but this is known and can be fixed later on
netlynx has quit [Remote host closed the connection]
<libv>
but when you are stuck to hdmi or vga, you shouldn't care about this extra running clock
<hramrach>
libv: what is the status of chromebook kernel?
<hramrach>
is there any kernel that gives sane display support?
<libv>
hramrach: it didn't horribly crash all the time
<libv>
hramrach: beyond that, the xorg driver had the mali driver overwrite parts of the fb in the wrong pitch
<libv>
and the display doesn't come back from dpms
<hramrach>
well, the kernel which cames with chromeos somewhat works wrt display and somewhat fails wrt wifi
<libv>
it's been since september, so i remember few details
<hramrach>
I don't need to run mali on it. I just want the built-in display and preferably also HDMI working
* hramrach
considers USB WiFi
<libv>
i don't think i used either wifi or hdmi
<hramrach>
actually, I was running plain fbdev the whole time
<libv>
as it lived just 1.2m away from my wrt54g
<hramrach>
any idea about mainline display support?
<WarheadsSE>
turning off dpms is easy
<libv>
hramrach: nope, sorry
<libv>
hramrach: i only am in the capture/replay stage atm on chromebook
<libv>
so it is mighty important to have working binaries
<WarheadsSE>
im running armsoc at the moment but not seeing much better than fbturbo's asm optimizations
<hramrach>
ok, I try to build it and complain how it's totally broken afterwards
<libv>
i was going to blame ssvb for the kernel panic in the memory mapping code
<ssvb>
libv: oh, what happened?
<libv>
but this is something that either happened somewhere on 3.4 or on A20
FR^2 has joined #linux-sunxi
<mnemoc>
?
<libv>
sunxi-3.4 also crashes
<mnemoc>
:(
<libv>
although...
<libv>
it's always on the 2GB machine
* libv
throws together an sd-card for his a20 tablet
<hramrach>
wasn't there a fix for that?
<WarheadsSE>
hramrach: im jut using the eMMC directly or a 16GB SanDisk Cruzer Fit nub
<ssvb>
libv: maybe some peripheral is not getting a proper bus address and is corrupting memory
<libv>
hramrach: black screen is what happened on a 2GB machines, that's what ssvb his changes fix
nove has quit [Quit: nove]
<hramrach>
was considering something like fit but this thing is quite reusable and even more fitting
<libv>
ssvb: it's when mapping ump memory
<libv>
i will first try the tablet, and then enable some debug printing in the functions in the trace
<hramrach>
I mean I can put the uSD in any next gadget that needs storage
<ssvb>
libv: with the current "stable" sunxi-3.4 it is understandable, the black screen bug is there for UMP on 2GB systems
<ssvb>
libv: but I don't like the "fix" myself, the mali driver code does not make a clear distinction between bus and physical addresses and it is inconsistent for different type of memory
<libv>
ssvb: it is a null ptr dereference, and the ump attach call looks sane
<libv>
so something went horribly wrong along the way
<libv>
anyway, time for some digging
<mnemoc>
can someone compose a branch with the stuff to merge in stage/sunxi-3.4? :( .... my laptop died so I can't do normal `git am`s :(
<libv>
:(
<libv>
just the laptop died, or did the disk die too?
<mnemoc>
the SSD
<mnemoc>
but I have ssh from the phone, so I can pull, test, etc..., push
wolfy has joined #linux-sunxi
<mnemoc>
any recommendation for a RELIABLE ssd? :\
<hramrach>
mnemoc: intel
wolfy has left #linux-sunxi [#linux-sunxi]
<hramrach>
the reviews are favorable and they are the only company that gives 5y warranty
<hramrach>
had my share of A-DATA and OCZ SSDs die on me
<hramrach>
they are the cutting edge ..
<hramrach>
newest stuff but buggy
<nikrou>
Hi all. Is there an equivalent of lspci to know what is my hardware ?
<hramrach>
which part of hardware?
<hramrach>
lsusb + ls /sys
notmart has quit [Quit: notmart terminated!]
<nikrou>
I think my wifi driver was rt8192cu
<mnemoc>
hramrach: I've lost 2 ocz vertex2 in less than 3y :\
<Nyuutwo>
nikrou: cat /proc/cpuinfo?
<nikrou>
but I cannot see wlan0
<Nyuutwo>
if you want to know your cpu
<nikrou>
and the driver is loaded
<hramrach>
then lsusb
<Nyuutwo>
nikrou: lsusb
<nikrou>
I will try
<Nyuutwo>
probably not added VID:PID pair
<hramrach>
and try to figure out how to power on all USB buses
<nikrou>
I have an old lsmod run on it but I don't know if all modules are loaded on boot
<hramrach>
and if it's usb or sdio ..
<nikrou>
I have rtl8150. Is it a wifi driver ?
ssvb has quit [Ping timeout: 272 seconds]
fredy has quit [Excess Flood]
fredy has joined #linux-sunxi
ssvb has joined #linux-sunxi
FR^2 has quit [Remote host closed the connection]
<nikrou>
Ok lsusb is great. I've got id 0fba:0179 from Realtek ; so the driver is 8188eu. Just need to compile it for my kernel !
<hramrach>
or just build the current kernel which has the driver
<nikrou>
I have kernel 3.0.76
<nikrou>
I didn't compile it
<libv>
nikrou: sunxi-3.4 is 3.0.75, and has that id included
<hramrach>
well, 3.0 is no longer developed and may not have the driver at all
<libv>
err, 3.4.75
<hramrach>
3.4.75
<libv>
nikrou: what hardware is this?
<nikrou>
sun5
<nikrou>
A13
<libv>
nikrou: that's the SoC, what device is this?
<mnemoc>
hramrach: weirdly the serie 530 seems to be cheaper (and faster) than the 520... both having 5y of warranty
<foubarre>
wens: hi. Is there any way to follow your progress on OTG (other than asking here) and is there a way i can help (testing is the most efficient i can do i think)?