<akaizen>
Anyone know of an A20 HDMI stick with RTL8188EUS?
\\Mr_C\\ has quit []
hipboi has joined #linux-sunxi
\\Mr_C\\ has joined #linux-sunxi
bamvor has joined #linux-sunxi
ykchavan has quit [Read error: No route to host]
ykchavan has joined #linux-sunxi
rz2k has quit []
[7] has quit [Disconnected by services]
TheSeven has joined #linux-sunxi
robb83 has joined #linux-sunxi
ykchavan has quit [Ping timeout: 264 seconds]
igor- has quit [Ping timeout: 240 seconds]
hramrach has quit [Ping timeout: 240 seconds]
fredy has quit [Ping timeout: 245 seconds]
hramrach has joined #linux-sunxi
fredy has joined #linux-sunxi
lkcl_ has quit [Ping timeout: 245 seconds]
Tsvetan has joined #linux-sunxi
lkcl_ has joined #linux-sunxi
shineworld has joined #linux-sunxi
iamfrankenstein has quit [Ping timeout: 246 seconds]
PiyushVerma has quit [Remote host closed the connection]
techn__ has joined #linux-sunxi
techn_ has quit [Ping timeout: 260 seconds]
massi has joined #linux-sunxi
n01 has joined #linux-sunxi
<n01>
'moning
<oliv3r>
lo
<mnemoc>
moin
<oliv3r>
how's things
<oliv3r>
mnemoc: time to make a cedarX page on linux-sunxi iki :D
<mnemoc>
iki???
<n01>
last day on my current job \O/
<Tsvetan>
oliv3r does cedarX works now?
<mnemoc>
n01: congrats! assuming you have a signed replacement already
<oliv3r>
Tsvetan: works is a great statement, but they've gotten it too work; needs a lot of polishing and work, but yes, jpeg, mpeg2 and h264 can be decoded with opensource now :)
<n01>
yup ... more embedded less networking (hopefully)
<oliv3r>
boycottg00gle: i have an olimex a20 too, but i know my default script.bin says i shoudl use a LCD :p
<oliv3r>
Turl: that's the eff statement that counts worldwide?
<ynezz>
"This FAQ is intended for non-lawyers who want some general information about how U.S. laws "
<oliv3r>
that was kinda my point :p
<oliv3r>
though U.S. laws are probably the most strict
<ynezz>
well, I think, that it's more likely that 3rd world war begins, then that some Chinese company is going to sue you
<oliv3r>
what about european countries
<ynezz>
how can they sue someone if they itself doesn't follow laws?
wingrime has joined #linux-sunxi
notmart_ has joined #linux-sunxi
<oliv3r>
how does europe not follow laws?
<ynezz>
someone has take you to the court for violating some law, right?
<oliv3r>
yeah
<ynezz>
and who that could be in this case?
<ynezz>
Allwinner?
<oliv3r>
Allwinner EUrope :)
<mnemoc>
allwinner would sue you in zuhai
<oliv3r>
if someone is pissed off enough; they can find a way to sue you!
<ynezz>
why would they be pissed? sales increase?
<ynezz>
it might be the case, if they can't produce enough :p
notmart has quit [Ping timeout: 264 seconds]
<oliv3r>
heh, if 'increased sales' would be seomthing they where after, they would have released the source :p
<ynezz>
they can't probably
<ynezz>
as I see it, this RE is for them ideal clean room case also, they didn't provided any IP to 3rd party
vicenteH` has joined #linux-sunxi
<ynezz>
Why do they release incomplete docs for their SOCs? just because they want to make it harder? I don't believe in that
vicenteH has quit [Remote host closed the connection]
<boycottg00gle>
oliv3r: my script.bin is ok
<oliv3r>
ynezz: they probably can't; but also, they could have been a bit more helpfull ;)
<oliv3r>
ynezz: I don' tthink they care enough really
<ynezz>
they're chinese = they care just about money
ykchavan has quit [Ping timeout: 264 seconds]
panda84kde has joined #linux-sunxi
<boycottg00gle>
hdmi state always is HDMI_State_Wait_Hpd
mouchon has joined #linux-sunxi
<boycottg00gle>
maybe hdmi_core.c:104 isn't good enough?
<mouchon>
hello
<oliv3r>
mouchon: gooday!
sud0x3 has quit [Ping timeout: 240 seconds]
<oliv3r>
boycottg00gle: did you try forcing EDID?
<wingrime>
oliv3r: have you saw anyowhere dts file example?
<wingrime>
oliv3r: I hope dts are usefull....
<mouchon>
just a noob question i take the last u-boot-sunxi from git repo and compiled it for my A20 olimex board but when i put it on sd card, the board still boot on the nand what can be wrong ?
<oliv3r>
wingrime: dts for?
<wingrime>
oliv3r: audio cedar
<oliv3r>
mouchon: which SD card slot are you using?
<mouchon>
the micro SD one
<oliv3r>
wingrime: your working on alsa driver?
<oliv3r>
mouchon: that's the right one :)
<oliv3r>
mouchon: sec
<wingrime>
oliv3r: no I looking at blobs with IDA
<wingrime>
oliv3r: I about cedarA
<boycottg00gle>
oliv3r: yep
<oliv3r>
wingrime: but cedarA doesn't use devicetree?
<oliv3r>
wingrime: youu Dolby DTS!
<wingrime>
oliv3r: cedarA is ACE
<boycottg00gle>
echo f1c16004 > /sys/class/misc/sunxi-dbgreg/rw/read returns 0x80000000 - hdmi should be on no?
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
<wingrime>
oliv3r: I just found dts and ac3 hw blobs on a20 sdk
Black_Horseman has quit [Remote host closed the connection]
<mouchon>
oliv3r for info i followed the wiki page of the u-boot-sunxi but may be i miss something. Should i use the first set of dd instruction or the second one ?
<mouchon>
oliv3r: 2 sec before i made the update the card was working ;-)
<mouchon>
oliv3r thank will try it
HeHoPMaJIeH has joined #linux-sunxi
<mouchon>
oliv3r it works ;-)
<Tsvetan>
wingrime: there is A20-SDK 2.0
<shineworld>
wow ... is on line ?
<Tsvetan>
I will speak with rz2k to add it to the torrent
<shineworld>
good to know !
* atsampson
applauds that CedarX work -- nice job :)
<Tsvetan>
also on the same cloud space there were some other sources, which may be interesting as were left over from wits employee
<Tsvetan>
but they are 6G
<Tsvetan>
and I have no idea where / how to keep
<mouchon>
oliver3r this is what i did for compiling my own: git clone https://github.com/linux-sunxi/u-boot-sunxi.git; cd u-boot-sunxi; A20-OLinuXino_MICRO CROSS_COMPILE=arm-linux-gnueabihf-
boycottg00gle has quit [Remote host closed the connection]
<Tsvetan>
just the cheese in the mousetrap :)
<stulluk>
is there a
<mouchon>
olive3r: yeah you are right :-)
<stulluk>
s
<stulluk>
in nanda there is no script.orig either
<Tsvetan>
wingrime: I already blogged about it, good work :)
<oliv3r>
stulluk: there's a backup, i forgot ts name
<Tsvetan>
wingrime you may better check SDK 2.0 it may have more new stuff
<stulluk>
and the funny point is, even I change the resolution settings in sys_config.fex, and convert to script.bin by using fex2binit doesn't take effect
<oliv3r>
Tsvetan: audio decoder is really old though; newer chips all do it in software :(
<oliv3r>
stulluk: that's very strange, your not booting form SD by accident?
<stulluk>
oliv3r, I am running on NAND flash with Android OS
<wingrime>
Tsvetan: I wait torrent
notmart has joined #linux-sunxi
notmart has joined #linux-sunxi
<stulluk>
wingrime +1 :)
<stulluk>
Recompiling everything from start... we are pretty sure that script.bin in /dev/block/nanda IS the real HW config file, right?
<stulluk>
HeHoPMaJIeH, do you get this IRQ problems only on TS, or do you get this on EMAC also? Or EMAC work fine in 3.4 ?
<HeHoPMaJIeH>
EMAC works fine
<stulluk>
I see.. In old SDK, EMAC also have problem, and I feel it is related with IRQ settings...
<stulluk>
When you enable it first time, it works fine, but after 20-30mins, it suddenly give IRQ errors, and need to disable -re enable the interface...
<mouchon>
oliv3r: which tool chain are you using ? i see binary size difference between your version and mine
<HeHoPMaJIeH>
stulluk, with which version of SDK ?
<stulluk>
Old SDK, I mean the torrent file on the Olimex's website
hipboi has quit [Quit: Leaving]
<HeHoPMaJIeH>
stulluk, hm that is strange i haven't noticed any problems with EMAC with old SDK
dragonn has joined #linux-sunxi
<stulluk>
HeHoPMaJIeH: I feel you have tested this on Linux build, not android, right?
deasy has joined #linux-sunxi
<HeHoPMaJIeH>
stulluk, yes
<stulluk>
I wish I could have used linux on this device, it would be almost perfect...
dragonn has quit [Killed (idoru (Spam is off topic on freenode.))]
<HeHoPMaJIeH>
mnemoc, now i see that we have linux-3.4 sent by allwinner
<stulluk>
HeHoPMaJIeH, you mean, in the SDK 2.0, kernel version is 3.4 ?
<Tsvetan>
:-)
<HeHoPMaJIeH>
stulluk, right
<HeHoPMaJIeH>
:)
<HeHoPMaJIeH>
now i'm trying to build it
<stulluk>
great, lets see if all I/O is working fine by default...
<stulluk>
HeHoPMaJIeH: in old SDK, u-boot didn't allow us to stop in console, did you notice that? ( It might be related with "fastboot" setting ?
<wingrime>
oliv3r: so, a10 man about realy have AE and CE in sram
<oliv3r>
stulluk: that's controlled by boot.env
iamfrankenstein has joined #linux-sunxi
<oliv3r>
wingrime: so Audio Engine and Camera ENgine?
<stulluk>
yes, but for android build, I couldn't find where is the boot.env ?
<wingrime>
oliv3r: no
<oliv3r>
wingrime: so maybe jpeg falls under CE but since it's so similar with mpeg fell under video engine
<wingrime>
oliv3r: I looks like more one IP
<oliv3r>
stulluk: could be compiled into the binary
<oliv3r>
wingrime: aye
<stulluk>
oliv3r: which binary
<stulluk>
?
<wingrime>
oliv3r: ACE have AE and CE modules,
<oliv3r>
stulluk: u-boot
<wingrime>
oliv3r: CE have PNG and TSCC
<wingrime>
oliv3r: and maybe more
<oliv3r>
wingrime: could be for camera purpouses
<wingrime>
oliv3r: DTS and AC3 looks ike in AE moduels
<stulluk>
Ok, I get confused now... In old SDK, there is already u-boot source, but there is no uEnv.txt or boot.env either?
<stulluk>
You mean in the source file?
<oliv3r>
stulluk: it's defined in the source
<oliv3r>
stulluk: yep
<stulluk>
Let me find it again
<stulluk>
Ok find it
<stulluk>
in u-boot source
<oliv3r>
the a20 sdk is only interesting to find stuff to port to 3.4
<stulluk>
include/configs/sun7i.h
<oliv3r>
the SDK entirly is only barely usefull
<stulluk>
in this header file, #define CONFIG_EXTRA_ENV_SETTINGS
<stulluk>
bootdelay=3\0" \
<stulluk>
"boot_fastboot=fastboot\0"
<stulluk>
So, bootdelay was set to 3
<stulluk>
but it never waits
<stulluk>
So I am suspicious about "fastboot"
<oliv3r>
mripard: '<our-device>' did you mean that literally? (Working in your comments right now
<oliv3r>
stulluk: there's several places that can override that
<oliv3r>
stulluk: boot.env partition; boot.env file, boot.scr file possibly too
notmart_ has joined #linux-sunxi
<mripard>
oliv3r: yes
<mripard>
or if you have any other suggestion
<mripard>
my point was just don't put the soc.0/deadbabe.eeprom in the doc
notmart has quit [Ping timeout: 264 seconds]
shineworld has quit [Remote host closed the connection]
<stulluk>
oliv3r: let me try to find all 3 places you mentioned
<oliv3r>
mripard: yeah i agree, but i copied that from a nother driver i think :)
notmart has joined #linux-sunxi
<oliv3r>
mripard: and while I like having the supported chips in the KConfig (as a user, its always helpfull to press '?' to know which chip is or is not supported, i'll remove it too :)
<oliv3r>
mripard: as for the ugly and fixme comments, those are kinda intended for Gregkh when he fixes them up, which I mentioned in the mail itself, any better idea to make it easy for gregkh to spot what to fix later?
notmart_ has quit [Ping timeout: 276 seconds]
<oliv3r>
mripard: and while i fixed the sun7i-sid name, i don't understand why we add a20 there, except for a10; does not compute :)
iamfrankenstein has quit [Ping timeout: 276 seconds]
<mripard>
oliv3r: for the comments, I'm not against the comments per se
<mripard>
I'm against meaningless commetns
<mripard>
so if you want to put these comments, sum up the discussion we had with greg
<oliv3r>
mripard: i didn't want to put long comments in, but can be a little more descriptive then :)
<mripard>
ie. what's there to fix, why you shouldn't be doing that, etc.
<oliv3r>
ok
<mripard>
otherwise, in 6 months, we'll just know that there's something to fix, and no clue about what it is.
<oliv3r>
mripard: ;)
<oliv3r>
mripard: i'll fix it and push a nother version; but silence is good, right? :)
<mripard>
ask n01 if silence is good...
<oliv3r>
:p
<oliv3r>
so the sun7o-a20- change, is it because it was 'forgotten' for sun4i-a10- and we don't want to fix it?
<mripard>
yep
<oliv3r>
so if someone sends a patch to do the rename; that'll be good?
<mripard>
no
<oliv3r>
why not? it's not that anybody is activly using it :)
<mripard>
maybe, but it's still a stable ABI.
gnufan has joined #linux-sunxi
<oliv3r>
but it looks ugly now :(
FunkyPenguin has quit [Quit: No Ping reply in 180 seconds.]
FunkyPenguin has joined #linux-sunxi
<n01>
-.- I'm still waiting ... and I don't know why ... sigh
<oliv3r>
n01: yeah, waiting is annoying :(
jukivili has joined #linux-sunxi
<wingrime>
oliv3r: I checked fscc codec, its look like zlib based as PNG
<wingrime>
oliv3r: *tssc
<mripard>
n01: if wim doesn't show a sign of life by then, as soon as 3.12-rc1 is released, I'll send a pull request to merge your driver through arm-soc.
<n01>
thank you
<oliv3r>
wingrime: explains why it's paired with png
<wingrime>
oliv3r: looks like it can do zlib's deflate
<oliv3r>
3.12-rc1? that's quite some more weeks is it not?
<oliv3r>
wingrime: zo offload zlib :D
<mripard>
oliv3r: a bit more than 2 weeks
<wingrime>
oliv3r: no any code example
atiti has joined #linux-sunxi
<wingrime>
oliv3r: also, ac3 very simular with dts
<oliv3r>
mripard: and a10s i completly forgot about ;)
<oliv3r>
mripard: but 3.11 hasn't even been released yet has it?
<oliv3r>
wingrime: i'm not supprised
<mripard>
oliv3r: a10s?
<oliv3r>
mripard: i mean, 3.11 is at rc7
<oliv3r>
mripard: you mentioned why I forgot about sun5i-a10s.dtsi :)
<mripard>
yes, and?
<wingrime>
oliv3r: ac3 use M-IDCT and IQ , jpeg like
<oliv3r>
so how can 3.12-rc1 be out in 2 weeks!
<mripard>
3.11 will be released sunday/monday depneding on your timezone
<oliv3r>
wingrime: yeah i wouldn't be supprised if the idct is a shared block on the die
<mripard>
and then, 2 weeks of merge window
<oliv3r>
wingrime: though it's quite possible they have it all split; without vhdl hard to say ;)
<wingrime>
oliv3r: no MDCT and DCT are differnct formula
<oliv3r>
nove: you provided help and input :) you helped, you are a hero :)
<oliv3r>
wingrime: your video is linked on the wiki page as embedded video
<oliv3r>
hipboi: wingrime, jemk, nove and paoulo
<wingrime>
nove: but paullo doing strange ....
<oliv3r>
andoma: oh that's right :p you would want this progress report too
<wingrime>
oliv3r: also, was ininb guy that start this before us
<oliv3r>
ianB
<oliv3r>
he also did a tracer didn't he; but did you use any of his work?
<oliv3r>
i'll add his name anyway
<wingrime>
oliv3r: yes , but we used new valgrind traces by nove
<nove>
oliv3r, ianB was the trigger
<oliv3r>
won't hurt having his name there :)
<wingrime>
oliv3r: he now hava no a10 hw , as he writed in email
<oliv3r>
wingrime: ianb or paulo?
<oliv3r>
rellla: ping!
<oliv3r>
rellla has a20 hw, so he should test this stuff
<wingrime>
oliv3r: paulo recompiled vp8 , but I will be glad if he add vp8 to jemk's code
<nove>
oliv3r, i did try its tracer, but has it require to implement dessamblley arm instructions, to get the address, data, i just went to find if there was better options, and then found tha valgrind worked
<oliv3r>
how much work will vp8 be to figure out?
<wingrime>
oliv3r: no idea
<oliv3r>
wingrime: vp8 uses the same HW engine doesn't it?
<oliv3r>
vp8 possibly next step, since webm/google (youtube) will probably use it extensilvy
<nove>
oliv3r, wingrime, vp8 is more simples, compared to h264
naobsd has joined #linux-sunxi
<nove>
the register dance i mean
<stulluk>
oliv3r, I have done it... I could be able to enter uboot console finally, but it is just useless now..
<oliv3r>
stulluk: :p
<stulluk>
First, I had to modify sun7i.h config file inside the uboot source and added #define CONFIG_ZERO_BOOTDELAY_CHECK
<stulluk>
Then it allowed me to get key during boot, so I entered uboot console,
<stulluk>
but it is terrible
<stulluk>
no tftp, no bdinfo....most of the commands are missing
<stulluk>
So, here is the good question now, how can I port uboot-sunxi to my SDK ?
<wingrime>
oliv3r: vp8 use same h264 engine but different regs
<oliv3r>
stulluk: if you want to boot from NAND; you can't
<oliv3r>
stulluk: if you boot from mmc; why bother, just checkout sunxi-u-boot, compile, install and be happy :)
<stulluk>
oliv3r, you mean current uboot-sunxi doesn't allow to be running on the NAND ?
<oliv3r>
stulluk: exactly
<stulluk>
OMG
<stulluk>
Then, I have to find way to add tftp bootm...etc command to my current uboot
<HeHoPMaJIeH>
3.4 from allwinner fails to boot
<oliv3r>
well, strickly speaking, the sdk u-boot doesn't do that either
<oliv3r>
HeHoPMaJIeH: allwinner has 3.4?!
<stulluk>
OMG
<oliv3r>
stulluk: boot0/boot1 boot the board, u-boot gets chainloaded somewhere at the end
<HeHoPMaJIeH>
yes
<stulluk>
3.4 fails to boot for linux ?
<oliv3r>
HeHoPMaJIeH: since when?! sdk kernel is 3.3 isnt' it
<oliv3r>
HeHoPMaJIeH: or is allwinner shipping linux-sunxi kernel now in their sdk
<HeHoPMaJIeH>
they are bothin new sdk
<stulluk>
oliv3r, yes, old SDK was 3.3, but 1 month ago, AW put a news on their website and stated that they will provide SDK 2.0
<HeHoPMaJIeH>
no it is not linux-sunxi
<oliv3r>
HeHoPMaJIeH: interesting
<HeHoPMaJIeH>
i must kill this super/ultra/turbo power management
<stulluk>
HeHoPMaJIeH: it fails on uboot or kernel ?
<HeHoPMaJIeH>
in kernel i think aw_pm is a problem
<HeHoPMaJIeH>
i mean super standby or whatever is called
<stulluk>
HeHoPMaJIeH: I feel it is a sys_config.fex issue
<wingrime>
oliv3r: actualy paullo612 works still not used, but will be very useful late
<HeHoPMaJIeH>
i think so
<nove>
wingrime, jemk, next week i will have some time to work in trace viewer, if you have suggestion that would make this tool more useful for yourselfs, feel free to ask
<wingrime>
nove: I maded some stupid python parser that make table used regs for each frame
<wingrime>
nove: I want see something wireshark allike
<nove>
wingrime, i will look at it
<wingrime>
nove: you can expand any frame by '+'
<wingrime>
and see all regs, also you can expand beyond bits
<wingrime>
oliv3r: don't add legality problems
<wingrime>
oliv3r: thats all clean-room
<wingrime>
oliv3r: jemk have not touched blob!
<oliv3r>
wingrime: yeah, but we have to make it clear, that joe user can't just download this code, and be like 'ohh goodie'
<nove>
wingrime, the problem with wireshark alike, is that the RE will finished before that tool is complete
<oliv3r>
wingrime: even worse, what if tsvetsan thinks 'oh cool, we need this' and starts shipping it with his roms (I'm not saying he would, just an example)
<stulluk>
HeHoPMaJIeH: Old-SDK even doesn't compile without AXP :)
<stulluk>
Unbelievable...
<HeHoPMaJIeH>
stulluk, i do not try to remove axp :)
<stulluk>
How to disable it without removing ?
<oliv3r>
there are HDMI chips that don't even have an axp :)
shineworld has joined #linux-sunxi
ZetaNeta has quit [Ping timeout: 264 seconds]
<HeHoPMaJIeH>
it is not a sys_config issue
<stulluk>
hmm, then, something wrong with the kernel?
<wingrime>
nove: I just want see something in binary code
<wingrime>
nove: I want any reg in bin form
<nove>
wingrime, this binary 1010101010,, understood
<oliv3r>
mnemoc: guess we'll have to see what AW did in their 3.3 -> 3.4 move
<hramrach>
well, at least the YT video desc works for me
<hramrach>
mplayer \o/
<wingrime>
jemk: firstly I saw in a20 blob new name for mpeg huffman regs
<wingrime>
jemk: thats interesting becose it means we can use sram regs not only for read but write
<hramrach>
looks cool
<wingrime>
jemk: sorry
<hramrach>
should try myself I guess
<wingrime>
jemk: not only write but read
<jemk>
wingrime: possible, i didn't try that
<hramrach>
where do you expect the disp scaler code to go?
<hramrach>
vdpau? mplayer? xv?
<wingrime>
jemk: sramPortRWOffset_reg - e0
Soru has quit [Read error: Connection reset by peer]
<wingrime>
hramrach: this so early, not it can be in vdpau lib
notmart has joined #linux-sunxi
<wingrime>
jemk: sramPortRWdata_reg e4
<hramrach>
vdpau api has not support for scaler?
<wingrime>
hramrach: no, disp can
<wingrime>
hramrach: we currently simple use disp api in vpdau lib
<jemk>
hramrach: vdpau has many things that aren't implemented in the poc, but disp handling has to move to some driver
<wingrime>
jemk: also, you should add something like "print_once"
Soru has joined #linux-sunxi
notmart_ has quit [Ping timeout: 240 seconds]
<wingrime>
jemk: debug can seriosly slow down play back
<wingrime>
jemk: also, as I writed , reigster name suggest that we can write
<wingrime>
*read
<hramrach>
so mplayer output driver or xv I guess
<jemk>
hramrach: dri or something like that
<hramrach>
you still have to add an userspace part that knows about the kernel buffers
<jemk>
wingrime: you can test it, but where should we need read sram?
<wingrime>
hramrach: problem, is that stuff depends on manilne process
<wingrime>
jemk: mostly debug
<hramrach>
regardless of teh kernel interface like AW disp/drm/we
<Turl>
wingrime: cubiebook has no spam protection I think
<wingrime>
Turl: yeax I noticed
<wingrime>
hramrach: unforunly we have no any disp support in mainline
<jemk>
hramrach: sure, but handling direct hardware access in userspace is a bad idea
notmart has quit [Read error: Operation timed out]
notmart has joined #linux-sunxi
<hramrach>
pff, I am root on all devices anyway. And dri is still direct hardware access. You just choose to give some users the permission or not. Not that disp is that much different, just has much more security holes in the code.
sanka has joined #linux-sunxi
<wingrime>
hramrach: thats will not be in mainline ever
<jemk>
hramrach: i tried to add some colorkeying to the output to put the video into the window instead of overlay, but that clashes with sunxifb
<hramrach>
overlay is fine for first iteration
sanka has quit [Client Quit]
<wingrime>
jemk: about ACE device , I looked in blobs, I can say that can do 1) ac3 and dts (looks simular coding ) 2) PNG decoding (zlib's deflate )
<hramrach>
but for something 'finished' the keying is required so you can get some windows over the video :S
<jemk>
hramrach: if you look at what vdpau doc you'll see that output handling is far from anything finish
<hramrach>
can the scaler output into a buffer which sunxifb then composites on its own?
<hramrach>
this disp layer is a mystery to me
<jemk>
wingrime: nice, but i will stay with h264 for some more time, still needs interlaced support. but ac3 and dts would be good
<jemk>
hramrach: me too, but it works
<hramrach>
to some extent
<jemk>
hramrach: vdpau way would be videosurface(YUV)--videomixer->outputsurface(rgb)->window
<hramrach>
that's the only possible way for X, basically
<jemk>
hramrach: and videomixer should be able to do many sorts of scaling/filtering and mix additional outputsurfaces
<hramrach>
for fb output you can just spam the video over everything
<lkcl_>
hno: ping, bug in running u-boot spl compile (see arm-netbook mailing list)
<wingrime>
jemk: I think ac3 may be simple, I have no any way to run that blobs
<wingrime>
jemk: in png case we have no any code for it
panda84kde has quit [Quit: Konversation terminated!]
sud0x3 has quit [Ping timeout: 248 seconds]
<techn_>
ssvb: should kernel be able to handle/fixup everykind SIGBUS?
sud0x3 has joined #linux-sunxi
<ssvb>
jemk: with the colorkey you still use overlay, but the layers stacking and configuration rules are a bit weird
<ssvb>
jemk: how exactly did it clash?
<jemk>
ssvb: opening xv window -> weired things happen. ok, was intentional, but anyways
<deasy>
great new about the opensource cedarx driver...thank you !
<ssvb>
jemk: I can probably remove the use of colorkey from XV
<ssvb>
techn_: in theory the kernel could fixup it, but that's a bad idea (the buggy neon code needs to be fixed instead)
<jemk>
ssvb: that wouldn't be the right way i think, handling the disp layers in vdpau only leads to problems
<techn_>
I have problem that my memory allocator is capable only 4 byte allocs
<techn_>
4-byte aligned
<wingrime>
jemk: why not use same color as sunxi-fb ?
<wingrime>
*for color key
notmart has quit [Read error: Operation timed out]
<ssvb>
jemk: yes, but but my understanding was that you are doing a quick and dirty hack now
massi has quit [Remote host closed the connection]
<techn_>
ssvb: so should gcc add alignment checking code automaticly? or I?
notmart has joined #linux-sunxi
_enrico_ has quit [Quit: Bye]
ZetaNeta has joined #linux-sunxi
<ssvb>
jemk: as for the colorkey, sunxifb needs a neon scaling/colorspace conversion fallback anyway, and it could handle the case when something is overlapping the video (albeit slower and with tearing)
<ssvb>
jemk: so the colorkey cold become a nice, but optional feature
<ssvb>
*could
<jemk>
ssvb: kind of quick and dirty, yes, but then you simply can say it works as long as you dont use xv same time
<ssvb>
jemk: yes, imho that's good enough
<ssvb>
techn_: what kind of neon code is failing for you?
<ssvb>
techn_: if you are not explicitly using assembly or intrinsics, then gcc should be generating valid code which does not fail
igor- has quit [Ping timeout: 248 seconds]
<techn_>
ssvb: thanks for clarifying that.. I'm new with thesekind problems
<ssvb>
techn_: if you are using assembly or intrinsics, then they can be just changed to the unaligned variants
<techn_>
usually compiler generates valid code :p
<techn_>
ssvb: yes.. compiler generates code that expects 8-byte alignment.. when I remove that alinging expectation it works
<techn_>
but wait couple moments.. I'll send example code
robb83 has quit [Ping timeout: 264 seconds]
<techn_>
hmm
<techn_>
no space left on device.. btw.. how I can make bootable sdcard for cubie2?
<techn_>
"Allwinner Video Decode Gets Reverse-Engineered" :)
ykchavan has joined #linux-sunxi
<techn_>
wingrime: jemk: nice work :)
Seppoz has joined #linux-sunxi
<WarheadsSE>
:)
<WarheadsSE>
i take it more from olimex than from moronix, but hey
<WarheadsSE>
still, good work guys :)
notmart has joined #linux-sunxi
notmart_ has quit [Ping timeout: 256 seconds]
notmart_ has joined #linux-sunxi
notmart has quit [Ping timeout: 264 seconds]
n01 has joined #linux-sunxi
<wingrime>
jemk: can't find sd-rot buffer in h264 reg list , but I think it present , also for mpegs there any flag that enable /disable sdrot buffer ? I think It can be MACC_H264_VLD_ADDR or _MPEG_VLD_ADDR's "Flag'
<wingrime>
jemk: there must be route selector
<jemk>
wingrime: the flags in vld_addr have other functions (don't know exatly which, but input memory related)
<wingrime>
jemk: But I think some unknown registers for H264 defenetly LUMA and CROMA
<wingrime>
jemk: for sd_rot
jemk_ has joined #linux-sunxi
<jemk_>
wingrime: whats with 0x244
notmart has joined #linux-sunxi
igor- has joined #linux-sunxi
jemk has quit [Disconnected by services]
jemk_ is now known as jemk
<wingrime>
jemk: sourry not noticed
<jemk>
wingrime: but i don't know if it works, gets never set
<n01>
oliv3r: does mainline kernel boot on A20?
<wingrime>
jemk: you can easy try , just set it too, and change disp layer offset
notmart_ has quit [Ping timeout: 264 seconds]
notmart_ has joined #linux-sunxi
notmart has quit [Read error: Connection reset by peer]
igor- has quit [Ping timeout: 245 seconds]
<wingrime>
Tsvetan: so I finaly downloaded sdk
_enrico_ has joined #linux-sunxi
igor- has joined #linux-sunxi
<oliv3r>
n01: yep; but onyl with mripards patches; it's not merged just yet
<oliv3r>
n01: i expect it to be so for 3.11 or 3.12-rc1
<oliv3r>
lkcl_: no clue, but we only use it for reference if anything; we don't run it as we have working u-boot for a20 merged allready
<wingrime>
oliv3r: libdemux_rmvb - ffmpeg included
vinifr has quit [Quit: Saindo]
<wingrime>
oliv3r: but rmvb looks only soft decoder
<oliv3r>
in ace blob?
<oliv3r>
looks like more GPL violations
<oliv3r>
wingrime: btw, i mailed phoronix about your progress ;)
<wingrime>
oliv3r: no, simple blob from cedar blobs
<oliv3r>
i see
<wingrime>
oliv3r: wtf, mix ffmpeg and rmvb looks crazy for layer
<oliv3r>
the blobs are crazy shit
<oliv3r>
they hacked some nasty stuff together for cedarX
<wingrime>
mnemoc: ,oliv3r new sdk have git inclued
<wingrime>
oliv3r: I see .git for boot and linux repos
<oliv3r>
do a git log ;)
<mnemoc>
they are usually symlinks to a missing .repo
<wingrime>
mnemoc: lichee-v2.0 have git
<oliv3r>
see if you can see history; if so; that'll make mnemoc's job much easier, as he can just continue from the previous hash
<wingrime>
oliv3r: I have wait, tar in tar
<mnemoc>
unfortunatelly my job has been web oriented for quite a while :(
<oliv3r>
mnemoc:
<oliv3r>
mnemoc: :(
<wingrime>
oliv3r: some gerber files
<oliv3r>
wingrime: not bad
<wingrime>
oliv3r: some reference board
<oliv3r>
probably the evaulation board
<oliv3r>
EVB-2.0
<techn_>
oo.. hdmi register guide
<wingrime>
oliv3r: bill of components and gereber files
<wingrime>
oliv3r: also looks likes altium project
hramrach has quit [Ping timeout: 240 seconds]
<oliv3r>
techn_: ohh that's a20 manual; yeah we had that
<oliv3r>
wingrime: not supprising
hramrach has joined #linux-sunxi
paulk-aldrin has joined #linux-sunxi
<oliv3r>
hey paulk-collins-aldrin
<rah>
what does the PhoenixCard software do?
<oliv3r>
rah its related to mmc if i remember it right
<rah>
can I just dd an image intended for PhoenixCard use, onto a card?
<oliv3r>
rah: e.g. create a bootable and then flashable SD card
<rah>
oliv3r: it is, yes, it's for "burning" images to SD cards
<oliv3r>
i could have remembered it completly wrong!
<oliv3r>
i think it's the other way around
<oliv3r>
you put the image on an SDcard
<oliv3r>
insert sdcard
<oliv3r>
and sdcard copies itself to nand
<rah>
isn't that what I said?
<oliv3r>
idea behind it is probably, make 10 sd cards, flash 10 boards, swap sd card to 10 new boards, flash
<oliv3r>
you said burning TO sd cards ;p
<rah>
right
<rah>
putting an image on an SD card
<rah>
"burning" is the terminology the manufacturer uses
atiti has joined #linux-sunxi
<oliv3r>
well you burn the nand, from the sd card :)
<oliv3r>
you burn the sdcard, from the PC
<wingrime>
oliv3r: thats not EVB
<oliv3r>
ohhh, you NAMED the image 'burning'
<wingrime>
oliv3r: its rubix
<oliv3r>
wingrime: rubix?
<rah>
I didn't name it burning
<oliv3r>
a 'burning'-image
<oliv3r>
e.g. a 'burn-to-nand'-image
<rah>
the manufacturer uses the word "burning" to refer to copying an image to an SD card
<oliv3r>
rah doesn't matter; potato tomato :)
<oliv3r>
rah i'd expect the manufactures using an SD card to burn an image onto the device's nand
<rah>
and strictly speaking you don't burn an image to NAND either as it isn't a PROM
<oliv3r>
the sd card would be their tool, their 'torch' :)
igor- has quit [Ping timeout: 245 seconds]
<oliv3r>
rah :p did we ever burn stuff?
<rah>
early writable ROMs were "blown" or "burned", in that the internal diodes were actually burned with an overload
<wingrime>
oliv3r: rubux v 1.1
<oliv3r>
rah they'd melt, but technically, did they really 'burn'? :)
<oliv3r>
rah: i know you still 'burn' or 'blow' efuses via the same technique
<oliv3r>
rah but even those can actually be reset
<rah>
mmm
<rah>
anyway
<oliv3r>
rah as I said, potatoes tomatoes :)
<oliv3r>
since i can't even do that on IRC properly :)
<rah>
nobody here knows exactly what this PhoenixCard software does to the SD cards?
<oliv3r>
what are you after exactly?
<rah>
I want to know if I can replicate the process in Linux
<oliv3r>
rah i'd imagine it puts a nand + mmc enabled SPL on it, and then copies some image to the internal nand
<rah>
possibly even if dd'ing will suffice
<n01>
oliv3r: and for linux-sunxi I don't see sun7i_defconfig
<oliv3r>
n01: i think that was added not too long ago
<oliv3r>
rah: in fel mode should be possible eventually; but i guess once we have a nand driver with the SPL also that should be possible from sd
<wingrime>
oliv3r: do you saw de_bsp
<wingrime>
folder
<wingrime>
oliv3r: bsp - are board support package for win ce
<n01>
uhm, not in sunxi-3.4 branch
<rah>
oliv3r: I'm looking for knowledge rather than guesses
<rah>
oliv3r: one can imagine any number of things the software might do with an SD card
<oliv3r>
wingrime: yeha i heard they where maybe going that route
<rah>
oliv3r: I'm interested in knowing what it actually does rather than speculating
<oliv3r>
wingrime: or did in the past, but we know of noactual code
<oliv3r>
rah nobody knows :)
<oliv3r>
n01: stage/sunxi-3.4?
<wingrime>
oliv3r: but wince.bmp in nanda ))
<wingrime>
oliv3r: so wince , andoird , melis
<oliv3r>
yeah :p
<oliv3r>
and melis is 'sun3i' old stuff
<oliv3r>
this sdk looks 'bigger' then the ones before
<n01>
oliv3r: yep
<n01>
found it
<n01>
gee...too many branches
<oliv3r>
n01: :p
<wingrime>
oliv3r: linche are twice
<oliv3r>
wingrime: what?
<wingrime>
oliv3r: there two linchee.tar
<oliv3r>
ohh lychee
<n01>
oliv3r: sorry to bother you tonight ... do we have anywhere a repo with the full set of patches for mainline applied? I would like to compile the latest mainline to work on it
<oliv3r>
n01: your not bothering me at all :)
<oliv3r>
n01: you should always work from vanilla of course
<oliv3r>
i find that almost unworkable because of the missing patches
<oliv3r>
i use mripards a20-clocks branch now
<oliv3r>
but i think turl made a 'next' branch that's most mainline work in it?
<oliv3r>
Turl: ^
<n01>
yeah on the wiki it says that the best pick for development is sunxi-nexi branch in mripard repo
<wingrime>
oliv3r: I find why cedarx for sun7i claimed 4k ))
<oliv3r>
hmm, turl should have an even more uptodate repo; but i think he put it up at sunxi
<wingrime>
oliv3r: it overclocked
<oliv3r>
wingrime: dual core dual videocore?
<wingrime>
oliv3r: no, its overclocked
<oliv3r>
wingrime: heh, CPU underclocked; cedarX overclocked; nice one: )
<n01>
Turl: I summon you :)
<wingrime>
oliv3r: a10 - 180 Mhz
<wingrime>
oliv3r: here userland driver can choose 400 Mhz
<oliv3r>
wow, that's quite hot
<oliv3r>
more then double the speed
<oliv3r>
well if it can run it safely
<oliv3r>
i'm sure a10 can OC it too
<wingrime>
oliv3r: on a13 it underclocked 160
<wingrime>
oliv3r: what a marketing logic
<oliv3r>
maybe power?
<oliv3r>
a13 is tablet SoC
<wingrime>
oliv3r: no , marketing
<oliv3r>
a10 settopbox
<oliv3r>
yeah but a13 came after a10
<oliv3r>
'look at this new a13, it's slower!'
<oliv3r>
'runs the same video speed!!! WTFPWN!
<wingrime>
oliv3r: no you not get it right
<wingrime>
oliv3r: it market placement
<oliv3r>
unless you mean power wise 'a13 uses less power even decoding hd content'
<oliv3r>
well yeah, but if you can make it look good, while underclock it
<oliv3r>
i think it's for power management
<wingrime>
oliv3r: unless a13 will interfere with a10
<oliv3r>
a13 is cheaper, more power efficient, but a little slower
<wingrime>
oliv3r: so you can't sell a10 more
<oliv3r>
doubt that, they both have special market
<oliv3r>
look what happend, a13 pushed a10 out of tablet market
<wingrime>
oliv3r: a13 mostly same
pitillo has quit [Changing host]
pitillo has joined #linux-sunxi
<oliv3r>
nobody needed sata etc a nyway, so a13 was good
<oliv3r>
and a10 still offers 1gb (possibly 2) of ram
<wingrime>
oliv3r: yes, but looks cedar is same
<oliv3r>
a13 is limited at 512
<oliv3r>
wingrime: on a20, taht's kinda awesome
<wingrime>
oliv3r: I defenetly try overclock it
<wingrime>
oliv3r: wow, we have sunxi-temperature
<wingrime>
oliv3r: sunxi-temperature.c
<oliv3r>
wingrime: yeah, should come from touchscreen
<oliv3r>
but doesn't work
<oliv3r>
wait, can you sprunge that file somewhere?
<wingrime>
oliv3r: it will make cedarx slower on a13
<oliv3r>
so sun5i is stock clock; sun4i is slightly higher clocked
<wingrime>
I just checked with 1920 video
<wingrime>
it realy noticeable slower
<oliv3r>
but 'is it enough'
<mouchon>
oliv3r: on which system does you compile u-boot ?. I try with a fresh ubuntu 12.04 install on vm and alos on the olimex board itself , i have changed the sd card i use, but no luck the sd doesn't boot
<oliv3r>
if its enough; its fine i suppose
<oliv3r>
mouchon: my server :) phenom II 4x i think
<wingrime>
jemk: ping
<oliv3r>
mouchon: techn_ has the same issue hno and him are investigating
<oliv3r>
wingrime: it has to as it might not run at all
<oliv3r>
wingrime: so for whatever reason, it was decided to keep cedar at default clock
<mouchon>
i will install a vm with gentoo to check this
<Turl>
oliv3r: maybe higher clock crashes A13 cedar
<_enrico_>
can someone with the new a20 sdk send me the drivers/media/video/sun7i_tvd directory (lichee 3.4)? i'd prefer not to download 6GB to just have a look at that :D
<wingrime>
oliv3r: we have all signs , that cedar x same for a10-a13-a20 but ,clocked different way
<hno>
techn__, strange indeed.
<hno>
techn__, how large is your spl/u-boot-spl.bin?
<oliv3r>
wingrime: yep
<oliv3r>
mouchon: i highly doubt it's the compiler ;)
<oliv3r>
mouchon: and gentoo is hard :p and takes AGES
<jemk>
wingrime: pong
<mouchon>
yes i know is hard , i managed many server with it :-)
<jemk>
oliv3r: temp seems to work
<wingrime>
jemk: set clock to 320
<oliv3r>
jemk: you trying the temp driver?
<jemk>
wingrime: why?
<wingrime>
jemk: smoother
<oliv3r>
jemk: can you try it on sun4i and sun5i too?
<jemk>
oliv3r: no, uboot
<wingrime>
jemk: 1920 playes notable smother
<techn__>
hno: 24644 bytes
<jemk>
wingrime: can't be, as it already decodes much faster than needed
<oliv3r>
hno: mouchon seems to have the same prob
<mouchon>
oliv3r:other thing , if sunxi need space i can see if we can host it in the ftp.belnet.be as i am one of the systeme admin
<hno>
techn__ & mouchon, have you ever sucessfully booted on the sd cards you are using? There is some SD cards that is not detected by BROM.
<wingrime>
jemk: realy no, I saw a20 sources
<wingrime>
jemk: it overclocked
<techn__>
hno: that sd card used to boot with a10 device
<wingrime>
jemk: to 500
<mouchon>
hno when i put the version of u-boot-sunxi-with-spl.bin that oliv3r provide to me, the sd card boot
<hno>
techn__, and now it doesn't boot on the same a10 device?
<jemk>
wingrime: why should we overclock when it works well with normal clock?
<techn__>
hno: I'll check that :)
<wingrime>
jemk: notable better
<hno>
mouchon, ok.
<wingrime>
jemk: looks cedar are same for a10-a13-a20-a31, but runs on differnt speed
<i-rinat>
jemk: Hi, just found your vdpau driver for cedarx. Here's another vdpau driver: https://github.com/i-rinat/libvdpau-va-gl Maybe you'll find something useful there.
<jemk>
wingrime: so all in acceptable range
<mouchon>
simple noob kernel question, how do you debug driver ? using printk where needed ?
<oliv3r>
wingrime: what happens if you enable that call :)
dwilkins has quit [Ping timeout: 256 seconds]
<oliv3r>
hno: yeah i compiled that for mouch to test and that one works
<wingrime>
oliv3r: I need setup new kernel for a13 test\
<wingrime>
jemk: we can use max possible clock in cubieboard allike batt-less systems
<oliv3r>
wingrime: in the far future,w hen we have mainline driver, i would let the userspace control which clock to run the VPU at
<wingrime>
jemk: it must be controlled using cpufreq driver
<wingrime>
oliv3r: ^
<oliv3r>
cpufreq driver, if that's even possible, yeah
<oliv3r>
:)
<wingrime>
oliv3r: why not let it to cpufreq
<jemk>
i-rinat: thanks, looks good, will look at it.
<oliv3r>
wingrime: cpufreq will run it at whatever governor you choose; which will require 'load monitoring' etc i dont' know if its that simple
<wingrime>
oliv3r: yeax
<oliv3r>
jemk: did overlooked; did you try the temp sensor on a10, a13 and a20?
<jemk>
wingrime: clock mostly depends on video format and size, i would ack oliv3r and let userspace decide
<jemk>
olive3r: only a10, others i dont have
<oliv3r>
wingrime: but yeah, sure looks like a decent framework to plug into
<oliv3r>
jemk: that's really good as that confirms that it even works on a10 still
<wingrime>
oliv3r: ppl4 using cpufreq
<wingrime>
*pll
notmart_ has quit [Quit: notmart terminated!]
<wingrime>
oliv3r: divider by userspace
<oliv3r>
yeah
<wingrime>
jemk: we also have cedar personal divider
<wingrime>
jemk: I think let cpufreq decide about pll4
<wingrime>
jemk: divider by userspace
<techn__>
hno: with mini-x spl it tries to boot from sd card
<oliv3r>
wingrime: but for cpufreq to work, you'll need to device a way to monitor VPU load :)
<techn__>
hno: minix spl size: 23916bytes
<wingrime>
oliv3r: we can measure time
<wingrime>
oliv3r: between frame decoding launch and IRQ
<oliv3r>
ok
<oliv3r>
sounds reasonable
<wingrime>
jemk: how you decide thats time are aceptable?
<techn__>
hno: so 24kB limit seems there
<wingrime>
jemk: with 320 clock I still not saw mplayer *slow pc* warning
<wingrime>
jemk: but with 240 saw....
<jemk>
wingrime: never seen this here, how i decided? 1/framerate >> time per frame
<wingrime>
jemk: if that time only between trig and ioctrl return?
<mouchon>
ls
<jemk>
wingrime: was whole h264_decode function, can do trig->irq too
<wingrime>
jemk: thats not fps, don't forget audio, etc
<jemk>
wingrime: yeah, sure, but audio doesn't depend on ve clock
<wingrime>
jemk: but it depends on witch is 'aceptable'
<wingrime>
if less < 30 not
<wingrime>
jemk: also, I forget mnemoc add you to sunxi group, you can move repo to it any time you want
Black_Horseman has joined #linux-sunxi
Black_Horseman has quit [Changing host]
Black_Horseman has joined #linux-sunxi
ZetaNeta has joined #linux-sunxi
i-rinat has quit [Quit: Leaving.]
<hno>
techn__, it's odd. I loaded max sized SPL in my tests. But I'll check again.
<techn__>
hno: any option I could drop from build?
<hno>
techn__, CONFIG_SPL_OS_BOOT unless you want to play with falcon boot mode.
<wingrime>
jemk: I pulled your driver, but I have strange lag on my 800x480 screen
<wingrime>
jemk: after update
<jemk>
wingrime: allways or only when resizing?
<wingrime>
jemk: always
<jemk>
wingrime: thats bad. possibly those x calls each frame are a bad idea.
<wingrime>
jemk: no, its looks like you simply incorretly calc some sizes
<jemk>
wingrime: what sort of "lag"? incorrect sizes normaly dont lead to lags
rz2k has joined #linux-sunxi
<wingrime>
jemk: wait I send pics
dwilkins has joined #linux-sunxi
<techn__>
hno: I dropped ext2, ext4, fat_write and networking.. now trying with smaler spl :)